Prospects for treating osteoarthritis: enzyme–protein interactions regulating matrix metalloproteinase activity

Abstract

Primary osteoarthritis (OA) is a musculoskeletal disorder of unknown etiology. OA is characterized by an imbalance between anabolism and catabolism in, and altered homeostasis of articular cartilage. Matrix metalloproteinases (MMPs) and a disintegrin and metalloproteinase with thrombospondin motif are upregulated in OA joints. Extracellular matrix (ECM) proteins are critical for resistance to compressive forces and for maintaining the tensile properties of the tissue. Tissue inhibitor of metalloproteinases (TIMPs) is the endogenous inhibitor of MMPs, but in OA, TIMPs do not effectively neutralize MMP activity. Upregulation of MMP gene expression occurs in OA in a milieu of proinflammatory cytokines such as interleukin (IL)-1, IL-6 and tumor necrosis factor α. Presently, the medical therapy of OA includes mainly nonsteroidal anti-inflammatory drugs and corticosteroids which dampen pain and inflammation but appear to have little effect on restoring joint function. Experimental interventions to restore the imbalance between anabolism and catabolism include small molecule inhibitors of MMP subtypes or inhibitors of the interaction between IL-1 and its receptor. Although these agents have some positive effects on reducing MMP subtype activity they have little efficacy at the clinical level. MMP-9 is one MMP subtype implicated in the degradation of articular cartilage ECM proteins. MMP-9 was found in OA synovial fluid as a complex with neutrophil gelatinase-associated lipocalin (NGAL) which protected MMP-9 from autodegradation. Suppressing NGAL synthesis or promoting NGAL degradation may result in reducing the activity of MMP-9. We also propose initiating a search for enzyme–protein interactions to dampen other MMP subtype activity which could suppress ECM protein breakdown.

Introduction

Primary osteoarthritis (OA) is a musculoskeletal disorder characterized by an increasingly compromised capacity of synovial joint tissues to effectively undergo repair in the face of fundamentally altered cell profiles and metabolism. The aetiopathogenesis of OA is thought to arise through a convergence of many components including genetic (e.g. family history), mechanical (e.g. abnormal stresses) and other environmental factors (e.g. obesity, occupation) that can influence the onset and progression of the disorder.

Of note, the progression of OA appears to result from initial changes in the cytological appearance of cells located in the periarticular cartilage [Goldring and Goldring, 2007]. Such changes include signs of chondrocyte senescence and apoptosis. Chondrocytes also undergo proliferation with clonal clustering and hypermetabolism emerging early in OA as evinced by an elevated level of proteoglycan and collagen synthesis [Sandell and Aigner, 2001; Goldring and Goldring, 2007]. In this milieu, chondrocytes in the different zones of articular cartilage are differentially altered. Chondrocytes in the superficial layer exemplify perhaps the most destructive aspect of this early cellular response since these cells produce proteolytic enzymes responsible for the degradation of pericellular, extracellular matrix (ECM) components [Hollander et al. 1995]. While many of these histological changes are thought to occur during the early stages of OA when the disease would likely be indolent and asymptomatic [Lorenz and Richter, 2006], these cellular changes pave the way for the more painful, inflammatory, and chronic stages of OA which emerge later in the disease [Malemud and Schulte, 2008]. The more advanced phase of the OA disease process is characterized by robust inflammatory responses by chondrocytes exemplified, in part, by the production of reactive oxygen species. The OA process is also exacerbated by the accumulation of ECM fragments resulting from cartilage degradation in the vicinity of the chondrocytes, as well as molecular and biochemical signals received from the inflamed synovial membrane [Hedbom and Häuselmann, 2002]. In addition to the degradation of the ECM proteins and fibrillation of the articular cartilage surface, inflammatory responses in the synovial membrane, osteophyte development and sclerosis of the subchondral bone are also characteristically seen at this time in the OA process despite the fact that these changes in this tissue may also have occurred earlier [Mastbergen and Lafeber, 2011]. Finally, the synergistic effects of a weakening cartilage infrastructure and a continual abnormal mechanical stress placed on the joint accelerates the characteristic wearing away of the articular cartilage. This essentially characterizes the endstage of OA, at which point joint replacement surgery is often the only option [Matthews and Hunter, 2011]. The existing drug therapies for OA exist in the form of nonsteroidal anti-inflammatory cyclooxygenase 1 (COX-1) inhibitors and the cyclooxygenase 2 (COX-2) inhibitor (e.g. celecoxib), as well as the intra-articular administration of corticosteroids. However, these therapies are primarily designed to treat the pain and inflammatory responses which limit joint function and range of motion that are associated with OA. At present, there is no specific drug designed to inhibit the structural damage to articular cartilage or any disease-modifying OA drug (DMOAD) therapy [Matthews and Hunter, 2011; Le Graverand-Gastineau, 2010; Goldring et al. 2011; Kapoor et al. 2011; Gege et al. 2012; Kumagai et al. 2010; Wang et al. 2011a].

The central clinicopathologic features of the progression of OA, which are inflammation and articular cartilage destruction, are largely mediated by the imbalance between anabolic and catabolic cytokine signaling molecules which, in healthy cartilage, maintain ECM homeostasis [Goldring et al. 2011; Kapoor et al. 2011; Alcaraz et al. 2010; Schroeppel et al. 2011]. In OA, catabolic events predominate. These changes are characterized by the significant upregulation of several matrix metalloproteinase (MMP) subtypes and a disintegrin and metalloproteinase with thrombospondin motif (ADAMTS).

Major challenges from a clinical perspective include both the identification of biomarkers associated with these cellular and molecular changes and the association of these biomarkers with a reliable method of staging OA disease progression and response to therapy. Additionally, we have proposed that a deeper understanding of the signal transduction pathways implicated in OA [Malemud, 2004, 2007a, 2009b] and the attendant metabolic imbalances caused by dysfunctional intracellular signaling would also greatly assist in meeting these goals, as well as in the development of an effective DMOAD therapy.

In osteoarthritis MMPs are responsible for significant degradation of articular cartilage extracellular matrix proteins

Although the enzyme class ADAMTS has been implicated as playing a vital role in OA pathophysiology [Davidson et al. 2006; Struglics et al. 2006; Sandy, 2006], compelling evidence indicates that it is the activity of various MMP subtypes, especially MMP-2 (gelatinase A; 72 kDa gelatinase), MMP-3 (stromelysin-1), MMP-9 (gelatinase B; 92 kDa gelatinase), MMP-13 (collagenase 3) and MT1-MMP (membrane-bound MMP-14) that drive OA progression. Therefore, the inhibition of MMP activity must still be considered as a potential target for therapeutic intervention in OA.

A therapeutic intervention directed at MMP subtypes would essentially be designed to retard, suppress and even halt the progression of cartilage ECM protein degradation in OA [Malemud et al. 2003; Burrage et al. 2006; Malemud and Schulte, 2008]. In this regard, although viscosupplementation with various molecular weight sizes of hyaluronic acid was shown to only have variable degrees of efficacy in the therapy of knee OA [Strauss et al. 2009; Curran, 2010; Wang et al. 2011b], hyaluronic acid suppressed interleukin (IL)-1β-induced MMP activity by OA synovial tissue explants in vitro [Waddell et al. 2007]. This finding suggested the possibility that viscosupplementation with hyaluronic acid could provide clinical benefit by blocking MMP activity in the OA joint.

The focus of an interventional therapy capable of blocking MMP subtype activity could result in the restoration of the balance of cartilage metabolism to anabolic ECM protein synthesis by chondrocytes. In other words, we propose that pathophysiological damage to the articular cartilage in OA which is mediated, in large measure, by several MMP subtypes is potentially reversible. However, this could occur only if the ongoing degradation of ECM proteins was sufficiently inhibited, the vitality of the resident chondrocytes preserved and restoration of cartilage ECM stimulated and maintained.

There now is persuasive evidence that each of these MMP subtypes either acting alone or together with other MMPs can mediate the destruction of articular cartilage in several experimental animal models of OA and in human OA cartilage [Malemud et al. 2003; Malemud, 2009b]. In this regard, the finding that the combined enzyme activities of these MMP subtypes possess the capacity to degrade the ECM proteins [Brinckerhoff and Matrisian, 2002] which regulate biomechanical responses and other functions of articular cartilage was noteworthy. Thus, MMP-13 showed a substrate preference for degrading the fibrillar collagens, including cartilage type II collagen as well as having the capacity to degrade aggrecan, the hydrodynamically large aggregating proteoglycan of cartilage. MMP-3 was also shown to degrade other various cartilage proteoglycans and, in addition, MMP-3 degraded link protein, the glycoprotein responsible for the functional stabilization of the noncovalent interaction between aggrecan and hyaluronan. MMP-2 and MMP-9, although showing a substrate preference for degrading noncartilage collagens such as type IV collagen and type V collagen, was also shown to degrade denatured type II collagen that was initially cleaved by activated MMP-1 (i.e. collagenase 1). Importantly, MMP-2 and MMP-9 also degrade the aggrecan core protein [Freije et al. 2003; Malemud, 2006; Burrage et al. 2006]. Of note, MMP-3 and MMP-14 can activate the latent forms of pro-MMP-3 and pro-MMP-13 [John and Tuszynski, 2001; Ala-aho and Kahari, 2005] which contributes to the elevated levels of activated MMP-3 and MMP-13 in OA synovial fluid. Recently, Zhang and colleagues used mass spectrometry to demonstrate that MMP-13 could interact with several other ECM proteins besides type II collagen and aggrecan, including, fibronectin, type VI collagen, decorin, syndecan 4 and serglycin [Zhang et al. 2010]. Although the effect of these various protein interactions with MMP-13 remains to be further clarified, the consequences of these MMP-13–protein interactions may be to not only degrade these proteins but to preserve MMP-13 activity as well.

In support of the above are the findings from clinical studies which showed that the levels of these MMP subtypes were significantly elevated in the synovial fluid, sera and subchondral bone cells of patients with OA [Tchetverikov et al. 2005; Mahmoud et al. 2005; Kim et al. 2011; Hulejová et al. 2007; Koskinen et al. 2011]. Interestingly, the elevated levels of some MMP subtypes (e.g. MMP-9) in the synovial fluid of patients with OA were lowered in response to therapeutic management with sodium hyaluronate and corticosteroids, but not by corticosteroids alone [Shimizu et al. 2010].

When these results are coupled with other compelling experimental and clinical evidence that proinflammatory cytokines [e.g. IL-1β, IL-6 and tumor necrosis factor α (TNFα)] upregulate the gene expression of these MMP subtypes in human OA [Fernandes et al. 2002; Schwab et al. 2004; Aida et al. 2005; Malemud, 2010] the logical conclusion followed that an experimental strategy designed to directly inhibit the activity of these MMP subtypes would effectively limit the destruction of ECM proteins of OA cartilage. Although the development of potent experimental MMP inhibitors was achieved through medicinal chemistry strategies, their efficacy beyond preclinical in vitro cartilage explant studies [Piecha et al. 2010; Wang et al. 2011c] has been limited. As such, the efficacy of these experimental MMP inhibitors in vivo has been significantly compromised by dose- and duration-dependent musculoskeletal side effects [Li et al. 2011].

Similarly, employing strategies designed to neutralize, for example, the upregulation of MMP subtypes and the suppression of ECM protein synthesis brought about by IL-1β by employing an IL-1 receptor antagonist protein (IRAP) (i.e. anakinra), or the IL-1 receptor neutralizing monoclonal antibody, AMG108 [Malemud, 2010], has also been proven to be relatively unsuccessful in OA clinical trials [Chevalier et al. 2009; Cohen et al. 2011]. In interpreting the reasons for this result, one must consider the strong likelihood that other proinflammatory cytokines which are found in OA synovial fluid and capable of increasing MMP synthesis continue to drive MMP gene upregulation.

At the present time, there appears to be little if any enthusiasm by practitioners for employing systemic administration of any anti-proinflammatory cytokine monoclonal antibodies in the treatment of OA. Of note, an open-label clinical trial [ClinicalTrials.gov identifier NCT00686439] with the TNFα antagonist adalimumab, administered subcutaneously to patients with inflammatory OA of the knee, has been completed, although the study results remain unpublished.

IRAP, the IL-6 receptor antagonist tocilizumab, and several antagonists of TNFα were shown to successfully neutralize MMP gene upregulation caused by IL-1, IL-6, and TNFα respectively in patients with RA [Malemud, 2011b]. In this regard, a recent meta-analysis of seven clinical trials in which infliximab, adalimumab, and infliximab in combination with methotrexate were used to treat patients with RA early in their disease showed that combination therapy reduced radiographic progression of RA compared with methotrexate alone [Kuriya et al. 2010]. However, significant heterogeneity was also found in the extent to which these x-ray changes occurred compared with the clinical response of these patients to TNFα antagonists. Thus, an anti-TNFα or IL-6 receptor antagonist drug which has proven clinical efficacy in RA and reduces the level of TNFα and IL-6 in serum could potentially also neutralize the activity of TNFα or IL-6 which drive MMP gene upregulation in OA.

Although several novel paradigms have also been explored for future therapeutic interventions in OA, including gene therapy [Malemud, 2007b, 2011a], best exemplified by the experimental intra-articular administration of a tissue inhibitor of metalloproteinase (TIMP) gene construct [Malemud, 2007b], these experimental strategies have not reached the point where they can even be considered for testing in human OA trials. Taken together, the results of numerous studies have indicated that an entirely new strategy will likely have to be developed to therapeutically manage the progression of cartilage destruction in OA.

Limiting the activity of MMPs in osteoarthritis by tissue inhibitor of metalloproteinases

In articular cartilage and a variety of other tissues, the activation of pro-MMP subtypes is regulated by several mechanisms [Jackson et al. 2009; van den Berg, 2011]. However, the crucial endogenous mechanism that controls the activity of MMPs occurs via the binding of MMP subtypes to a family of homologous proteins called TIMPs (i.e. TIMP-1, -2, -3, -4) [Moore and Crocker, 2012]. The two-domain TIMPs are of relatively small size with diverse biochemical and physiological/biological functions. These include inhibition of active MMPs, pro-MMP activation, promotion of cell proliferation, extracellular matrix binding, inhibiting angiogenesis, and induction of apoptosis [Brew et al. 2000]. However, the MMP inhibitory capacity of TIMPs appears to be low in OA cartilage [Iannone and Lapadula, 2003; Burrage and Brinckerhoff, 2007] and attempts to regulate the overproduction of MMP subtypes in OA joints by employing exogenous TIMPs had little clinical efficacy [Clutterbuck et al. 2009].

Limiting the activity of MMPs in osteoarthritis by promoting MMP autodegradation

The autodegradation of MMP-9 and its regulation by neutrophil gelatinase-associated lipocalin

Another mechanism which preserves the activity of activated MMP types occurs when certain MMP subtypes form a complex with other proteins. The best understood example of this phenomenon is the interaction between pro-MMP-9 and neutrophil gelatinase-associated lipocalin (NGAL) [Yan et al. 2001].

MMP-9 is a regulated MMP which plays a central role in connective tissue remodeling and in the turnover of basement membrane [Masure et al. 1993] as well as in the regulation of the release of vascular endothelial cell growth factor from the ECM, the latter providing a critical signal that promotes angiogenesis, invasion, and cancer metastasis [Lochter et al. 1998].

MMP-9 has been immunolocalized to the edge of endochondral cartilage from 21-day-old rats where the degradation of type II collagen occurred [Lee et al. 1999]. This result indicated that intact cartilage type II collagen was also a target substrate for MMP-9. Moreover, a deficiency in MMP-9 [Praul et al. 2000] or reduced amounts of MMP-9 [Miao et al. 2004] were shown to contribute to abnormal growth plate development, indicating that MMP-9 also plays a fundamental role in skeletal long bone development [Malemud, 2006].

Of note, the synovial fluid from patients with knee OA was found to be enriched in an MMP-9/NGAL complex [Gupta et al. 2007]. An analysis by zymography showed that the high molecular weight complex 125–130 kDa band on the zymogram from OA synovial fluid was an MMP-9/NGAL complex that migrated on the gel to the same location as the MMP-9/NGAL complex purified from human neutrophils (Figure 1). Moreover, the complex of MMP-9 with NGAL preserved the activity of MMP-9 because MMP-9 was autodegraded when NGAL was depleted from the MMP-9/NGAL complex by immunoprecipitation. In support of this latter finding were additional results showing that MMP-9-mediated release of ECM proteoglycans by explanted human cartilage was significantly greater in the presence of NGAL.

Figure 1.

Zymography of the neutrophil gelatinase-associated lipocalin (NGAL)/matrix metalloproteinase 9 (MMP-9) complex from human neutrophils and the high molecular weight gelatinase activity recovered from osteoarthritis (OA) synovial fluid.

Synovial fluid samples from three patients with OA (lanes 1–3) and the NGAL/MMP-9 complex purified from human neutrophils (lane 4) were analyzed by gelatin zymography. The migration positions of MMP-2, MMP-9 and the NGAL/MMP-9 complex are shown. Precision Plus Protein Standards (Bio-Rad, Hercules, CA) were employed as molecular size markers. (Reproduced with permission from Gupta et al. [2007].)

NGAL recovered from the synovial fluid of patients with end-stage OA of the knee could be a product released from neutrophils, but could also be synthesized by articular chondrocytes because human chondrocytes derived from OA knee cartilage and treated in culture with IL-1β were shown to synthesize NGAL [Gupta et al. 2006]. As was also the case with OA synovial fluid, the NGAL produced by cultured OA chondrocytes protected MMP-9 from autocatalytic degradation. Of note, Karlsen and colleagues recently reported that NGAL synthesis by cultured epithelial cells was also increased when the cells were incubated with IL-17 and TNFα, the latter being dependent on the activity of IκBζ [Karlsen et al. 2010]. Thus, two of the proinflammatory cytokines implicated in the progression of human OA, namely IL-17 and TNFα [Malemud, 2010], were found to increase NGAL production in epithelial cells and it will be instructive to determine whether IL-17, TNFα or yet other proinflammatory cytokines modulate NGAL levels in human chondrocyte cultures.

The MMP-9/NGAL complex has also been shown to regulate the degradation of complex ECMs in other cell types besides chondrocytes. MMP-9 and NGAL are also synthesized and secreted into the culture medium by cells of the innate immune system, hepatocytes and renal tubular cells [Goetz et al. 2000; Schmidt-Ott et al. 2007]. The MMP-9/NGAL complex has also been implicated in angiogenesis and in metastasis in various forms of cancer, including breast cancer, kidney carcinoma and colorectal disease [Goetz et al. 2000; Fernández et al. 2005; Kubben et al. 2007; Provatopoulou et al. 2009; Du et al. 2011]. The MMP-9/NGAL complex can also be considered as a biomarker of disease activity in several inflammatory disorders, including those affecting the kidney, gastrointestinal system, and vasculature [Bolignano et al. 2008; Li and Chan, 2011; Cernaro et al. 2011]. Of note, NGAL by itself has been proposed to mediate inflammatory responses in these organ systems through its capacity to sequester polymorphonuclear leukocyte chemoattractants, including N-formyl tripeptides, leukotriene B₄ and platelet-activating factor [Goetz et al. 2000]. Thus, it will be pertinent to the further understanding of the role of activated synovium in the inflammatory response in OA to determine if synoviocytes from OA joints also produce NGAL.

The structure of neutrophil gelatinase-associated lipocalin

NGAL, also known as lcn2 or siderocalin, is constitutively expressed in myelocytes and stored in the neutrophil specific granules [Borregaard and Cowland, 2006; Chakraborty et al. 2012]. NGAL can exist as a monomer, as a disulfide-linked homodimer, or as a disulfide-linked heterodimer in a complex with MMP-9 [Kjeldsen et al. 1994, 2000]. Human NGAL consists of a single disulphide-bridged polypeptide chain of 178 amino-acid residues with a calculated molecular weight of 22 kDa. The molecular size of NGAL increased to 25 kDa when the molecular weight of the glycosylated residues was included in the calculation [Kjeldsen et al. 1993].

The binding of specific enzyme activity-preserving proteins has not yet been identified for any of the other MMP subtypes. However, autocatalytic degradation of MMP-2 [Bergmann et al. 1995] and MMP-3 [Qoronfieh et al. 1997] have been reported, suggesting perhaps that some type of MMP–protein binding structure may also preserve the activated forms of these MMP subtypes in addition to the preservation of activated MMP-9 by NGAL.

Could the structure of the siderocalins provide a working model for the discovery of other modulators of MMP activity?

NGAL belongs to the class of proteins called siderocalins. Siderocalins were originally characterized as antimicrobial proteins since they bind to bacterial ferric and apo-siderophores [Abergel et al. 2008], which prevent the delivery of host iron to the bacterium. Siderocalins also participate in the transport of small hydrophobic/lipophilic ligands [Yang and Moses, 2009]. A specific receptor for lcn2, also known as solute carrier family 22 member 17 (SLC22A17) has been identified [Miyamoto et al. 2011].

The crystalline structures of NGAL display a typical lipocalin fold, although an unusually large and atypical polar binding site was also seen [Goetz et al. 2000, 2002]. However, siderocalin differs slightly in its structure from other members of the lipocalin family because siderocalin contains a calyx that is larger, is shallower and lined with polar and positively charged residues [Goetz et al. 2000]. Additionally, the N-terminus of siderocalin is thought to possess a more highly ordered structure than the other lipocalins because siderocalin is composed of three bulges in the strands that form the β-barrel structure [Flower, 1996; Coles et al. 1999]. Of all of the known lipocalins, including siderocalin, Cys⁷⁶ and Cys¹⁷⁶ are the conserved residues which are also the sites where the disulphide bridges occur [Flower, 1996]. Of note, in addition to the Cys⁷⁶ and Cys¹⁷⁶ residues, siderocalin contains another cysteine residue located at Cys⁸⁷ where the homodimerization and heterodimerization with MMP-9 occurs. The typical binding modalities of the lipocalins are shown in Figure 2.

Figure 2.

Binding modalities of lipocalin.

The ligand binding pocket of lipocalin is shown in (a) based on the structure of the lipocalin L1 binding pocket [Flower, 1996]. This schematic was originally drawn by Rebecca Sweet (University of Georgia, Athens, GA). The drawing shows that there are two types of binding modes that the lipocalins use to interact with soluble macromolecules. The binding modes are the covalent association type shown in (b) and the covalently-linked association type shown in (c) which is used by siderocalin.

Presently there is no structural evidence that siderocalins form complexes with any of the MMP subtypes that are known to mediate articular cartilage ECM protein degradation in OA. Therefore it remains to be determined the extent to which the structure of the siderocalins can be successfully modeled to discover other enzyme–protein interactions that preserve MMP activity. If such interactions were to be found to exist going forward, it would be possible to develop experimental strategies to limit the influence of these enzyme–protein interactions on MMP activity with the long-term goal in OA of forcing MMP subtypes to undergo enhanced autodegradation.

In addition to facilitating the transport of iron into and out of the cell, NGAL also regulates iron responsive genes and in this way affects iron metabolism [Chakraborty et al. 2012; Correnti and Strong, 2012]. Thus, there could be potential metabolic side effects involving NGAL-like proteins that interact with MMP subtypes which should also be taken into consideration.

This is an opportune time to propose that strategies also be developed to inhibit the production of those proteins, such as NGAL, which protect MMP subtypes from autocatalysis. Some recently developed experimental techniques for employing silencing RNAs or micro-RNAs [Malemud, 2009a], for promoting protein sumoylation [Malemud and Pearlman, 2009], and for using nanoparticle delivery systems for drug administration [Wisler et al. 2011] may all be useful in this endeavor.

Finally, we envision that several novel approaches should be sought to alter the activity of MMP subtypes as opposed to their expression and synthesis. These approaches would include the discovery of a biochemical mechanism for limiting the enzyme–protein interactions that protect MMP subtypes from autodegradation or seeking biological agents or medicinal chemicals that could suppress the expression of genes which produce proteins that form complexes with these MMP subtypes.