Current and emerging therapeutic strategies for preventing inflammation and aggrecanase-mediated cartilage destruction in arthritis

An equilibrium exists between metalloproteinases and their inhibitors to maintain a balance between anabolism and catabolism in articular cartilage. In arthritis, disequilibrium favors the catabolism of cartilage whereby protease activity outweighs their inhibition by tissue inhibitors of metalloproteinases (TIMPs). Although MMP and ADAMTS enzymes are responsible for the degradation of cartilage in arthritic disease, their roles in cartilage development and remodeling are crucial for joint formation and homeostasis. MMP-1 and -2 are localized in synovium and joint articular surfaces in human fetal limbs at 7 to 14 weeks’ gestation, suggesting roles for these proteases in the development and remodeling of synovial tissue and articular cartilage [5]. Studies using homozygous Mmp-9-null mice revealed its requirement for angiogenesis and ossification of the developing growth plate since these mice exhibited delayed apoptosis, ossification, and vascularization of hypertrophic chondrocytes, resulting in progressive growth-plate lengthening [6]. Furthermore, Mmp-13-null mice exhibit defects in growth-plate cartilage with expanded hypertrophic chondrocyte zones and increased trabecular bone as well as increased interstitial collagen accumulation, with combinatorial Mmp-9 and Mmp-13 knockout mice displaying an exacerbated phenotype, suggesting synergy between these two proteases in cartilage and bone formation [7],[8]. Importantly, mutations in MMP-9 and MMP-13 in humans cause genetic disorders in bone and cartilage growth and developmental phenotypes such as metaphyseal dysplasia and spondyloepimetaphyseal dysplasia, Missouri type [9],[10], which are disorders of abnormal growth and development of long bones and vertebrae. Mmp-14 (MT1-MMP)-deficient mice display severe skeletal abnormalities, including impaired vascularization of epiphyseal cartilage, leading to delayed ossification and hypertrophic zone lengthening, revealing a role for Mmp-14 in angiogenesis and bone growth [11]. Significantly, human mutations in MT1-MMP cause Winchester syndrome, which is associated with progressive osteolysis, osteoporosis, and joint erosions [12]. It has not yet been established whether ADAMTS4 or ADAMTS5 has a role in the development and growth of cartilage and bone, although their expression is upregulated in arthritic disease. Other `aggrecanases’ include ADAMTS1, ADAMTS9, and ADAMTS15, which may have roles during cartilage and bone development. Although Adamts1 mRNA is expressed in growth-plate and articular cartilage during normal mouse development and is upregulated in hypertrophic differentiation of growth-plate chondrocytes, it does not play a significant role in cartilage and bone development and growth [13] or in arthritis. Adamts9 mRNA is also expressed from 13.5 days post-coitus during mouse embryogenesis in the perichondrium, the proliferative zone in the growth plate and bone [14], but roles for ADAMTS9 have not yet been elucidated in cartilage and bone development or in arthritic disease. Furthermore, ADAMTS15 is expressed in chondrocytes and perichondrium of the synovial joints in the developing mouse embryo at 15.5 days post-coitus; however, its function in the joint during development or arthritis has not yet been elucidated [15]. Aggrecan degradation facilitated by MMP and ADAMTS enzymes is a process that occurs within normal and arthritic cartilage, signifying a role for these proteases in normal turnover as well as in arthritis [16], whereas structural changes in aggrecan occur during healthy aging [17].

MMP activity is upregulated in arthritic cartilage and synovial fluid [18],[19], which correlates with type II collagen cleavage [20]. The collagenases (MMP-1, MMP-8, and MMP-13) preferentially degrade type II collagen (collagen II) at Gly⁷⁷⁵↓⁷⁷⁶Leu causing loss of its trimeric structure, exposing it to further degradation [21]. MMP-2 and MMP-9 (the gelatinases) and MMP-3 (stromelysin), which degrade non-collagen matrix components of the joint, also promote further degradation of denatured collagen II after cleavage by collagenases [22]. MMPs also degrade aggrecan; MMP-3, first isolated from human articular cartilage, cleaves the Asn³⁴¹↓³⁴²Phe bond of aggrecan in its interglobular domain (IGD) [23],[24]. However, it was recently shown that MMP-generated aggrecan fragments are involved predominately in normal aggrecan turnover and that their preferred cleavage site is located C-terminal to the IGD and that these fragments may have a lesser role in aggrecan degradation in knee injuries and osteoarthritis (OA) in human cartilage [25].

In contrast to collagen II degradation by MMPs, aggrecan degradation by aggrecanases is an early and reversible event [26]. Furthermore, since aggrecan prevents collagen II degradation and therefore may impart overall cartilage protection [27],[28], inhibiting aggrecan degradation via the ADAMTS aggrecanases may be a crucial therapeutic strategy to prevent further collagen II degradation. ADAMTS4 and ADAMTS5/ADAMTS11 were first described in 1999 [29]-[31] and cleave aggrecan in its C-terminal chondroitin sulphate (CS)-rich domains (Figure 1) at the following sites: SELE¹⁵⁴⁵↓¹⁵⁴⁶GRGT, KEEE¹⁷¹⁴↓¹⁷¹⁵GLGS, TAQE¹⁸¹⁹↓¹⁸²⁰AGEG, and ISQE¹⁹¹⁹↓¹⁹²⁰LGQR; however, the most detrimental cleavage is thought to occur within its IGD at TEGE³⁷³↓³⁷⁴ARGS (Glu³⁷³↓³⁷⁴Ala bond), generating G1-NITEGE fragments that release the entire CS-rich region into the synovial fluid compromising joint function (reviewed in Fosang and colleagues [32]) (Figure 1).

In 2005, two independent landmark studies demonstrated that ADAMTS5 catalytic inactivation protected mice from experimentally induced OA and rheumatoid arthritis (RA) [33],[34] but that ADAMTS4-deficient or catalytically inactivated mice did not show this same protection. However, whether ADAMTS4 or ADAMTS5 is predominately responsible for the cleavage of aggrecan in arthritis in humans remains controversial. Although ADAMTS5 cleaves aggrecan extensively in human arthritic synovium and is abundant and widely distributed in human OA cartilage [35], other data have indicated that both ADAMTS4 and ADAMTS5 cooperate to mediate aggrecan degradation in human articular cartilage explants [36].

Since arthritis is a disease of the entire joint, ADAMTS4 and ADAMTS5 may have variable activity depending on their localization and which cytokines are present to stimulate their gene expression and activation profile. IL-1α and TNF-α induction of Adamts5 was found to occur predominately in synovium and the patellar but not in femoral head or tibial joint cartilage in ex vivo mouse joints, indicating that ADAMTS5 may not be the predominant aggrecanase in articular cartilage in arthritis but in fact may affect cartilage indirectly [37]. Furthermore, in bovine menisci, the gene expression of ADAMTS4 is preferentially upregulated by IL-1α in inner meniscal zones, whereas the gene expression of ADAMTS5 is preferentially upregulated by TNF-α in outer meniscal zones [38]. In human OA synovium, upregulation of ADAMTS4, unlike that of ADAMTS5, was IL-1- and TNF-α-dependent, again exemplifying the fact that ADAMTS5 may be constitutively expressed and more active in joint structures other than articular cartilage [3]. The differential roles of the aggrecanases may add complexity to potential treatments discussed below. However, evidence of their cooperative roles in cartilage degradation and common activation by inflammatory cytokines suggests that both ADAMTS4 and ADAMTS5 represent important therapeutic targets in arthritis.

There is evidence to suggest that statins have strong anti-inflammatory effects in RA. Simvastatin decreased articular macrophage infiltration and suppressed bone destruction in an RA rat model [56]. Furthermore, simvastatin inhibited the migration and invasion of fibroblast-like synoviocytes by preventing the activation of RhoA, a small GTP-binding protein known to activate NF-κB, therefore identifying a novel therapeutic agent for RA [57]. A prospective study in patients with RA demonstrated that 20 mg/day of simvastatin was a safe treatment that had anti-inflammatory effects with mild clinical improvements in measures such as swollen joint counts and Disease Activity Score in 28 Joints (DAS28) scores [58]. Disease activity also improved in RA patients undergoing a methotrexate regimen in combination with atorvastatin; evidence that inflammatory cytokines such as TNF-α were decreased systemically provided a potential mechanism to explain these observations [59]. A different trial of atorvastatin in patients with RA revealed a clinically apparent improvement in DAS28 scores as well as a decrease in C-reactive protein and erythrocyte sedimentation, which are typical markers used to diagnose RA [60].

However, conflicting evidence regarding the effects of statins has also arisen. Statins accelerated the effect of collagen type II-induced arthritis in mice [61]. Furthermore, statins may induce a pro-inflammatory response in peripheral blood mononuclear cells by activating IL-18 and caspase-1 [62]. Although rosuvastatin has been shown to reduce C-reactive protein in patients with RA, this effect did not correlate with an improvement in overall RA disease activity [63]. Therefore, statins may have both anti- and pro-inflammatory effects, depending on the form and progression of the disease, the type of statin prescribed, and whether the patient is undergoing a multiple-drug regimen. Further investigation into the effects of statins is required in RA to clarify whether they are clinically effective anti-inflammatory treatments in human trials and whether a corresponding reduction in aggrecanase activity in RA is apparent.

In OA chondrocyte cultures, atorvastatin produced a significant reduction in IL-1β and MMP-13 as well as an increase in aggrecan and ColIIa1 expression, and this is an indication that atorvastatin may have chondroprotective effects, as well as anti-inflammatory effects [64], which could be relevant in the treatment of OA. This was also demonstrated with mevastatin, which showed reduced inflammatory cell infiltration and IL-1β and matrix-degrading enzyme (MMP-3 and MMP-13) expression in a rabbit model of experimental OA [65]. Furthermore, in a rat model of mechanically induced knee OA, simvastatin produced anti-inflammatory and immunomodulatory effects via the inhibition of MMP-3, demonstrating a possible additional chondroprotective effect [66]. Therefore, several classes of statins not only may have anti-inflammatory effects but also may demonstrate chondroprotective effects in patients with OA or RA.

Further investigations regarding the effects of statins in patients with OA or RA are clearly required given the likelihood of high incidences of co-morbidities with hypercholesterolemia, cardiovascular disease, obesity, and arthritis. However, given their apparent effectiveness in reducing inflammation or cytokine activity or both, one might hypothesize that affected joints of co-morbid patients undergoing a statin regimen could be inadvertently protected from cartilage destruction (Figures 1 and 2) to varying extents.

TIMPs are specific endogenous MMP and ADAMTS inhibitors and also are essential for homeostasis of the joint and proper matrix turnover as described above. There are four mammalian TIMPs, designated TIMP-1 through −4; TIMP-1 was discovered in 1985. TIMP-3, unlike the other TIMPs, has a broad profile of inhibition that includes ADAMTS4 and ADAMTS5. TIMP-3 acts as a tumor suppressor and inhibitor of angiogenesis, and Timp-3 homozygous-null mice present with enhanced TNF signaling and serum IL-6 levels [77], indicating a key role for TIMP-3 in innate immunity. Furthermore, Timp-3 knockout mice present with an increased inflammatory response to antigen-induced arthritis and increased aggrecan and collagen II degradation with age [78],[79]. TIMP-3 may be a suitable therapeutic treatment for patients with arthritis to suppress not only innate inflammatory cytokines in arthritis but also ADAMTS4 and ADAMTS5. Their lack of selectivity precludes them as an adequate treatment option in their native form. However, truncated TIMP-3 (N-TIMP-3), lacking its C-terminal domain, is a potent endogenous inhibitor of ADAMTS4 and ADAMTS5 with inhibition also demonstrated toward MMP-1, MMP-2, and (to a lesser extent) MMP-3 [80]. Furthermore, the thrombospondin type-1 repeats of ADAMTS4 and ADAMTS5 promote binding with N-TIMP-3 [81], providing further evidence that N-TIMP-3 may effectively inhibit ADAMTS4 and ADAMTS5 with high affinity. Moreover, by engineering the reactive site through amino acid substitutions within the N-terminus of N-TIMP-3, some selectivity toward ADAMTS4 and ADAMTS5 has been gained without off-target MMP inhibition [82], suggesting that with further modifications TIMPs may be a powerful potential future therapeutic.

The need to develop novel and selective aggrecanase inhibitors has become increasingly essential to arthritis research. A series of (2R)-N⁴-hydroxy-2-(3-hydroxybenzyl)-N¹-[(1S,2R)-2-hydroxy-,3-dihydro-1H-inden-1-yl] butanediamide derivatives have previously been developed as potent and selective inhibitors of aggrecanase activity [83]. A 3-hydroxyl group on one of the inhibitors achieved selectivity through hydrogen bonding with a threonine in the S1 pocket of ADAMTS5; however, this threonine is conserved in ADAMTS4, reducing the selectivity of these derivatives. A valine residue replaces this threonine in active sites of most of the MMPs, thereby achieving some selectivity toward the aggrecanases [84]. Recently, a series of novel achiral N-hydroxyformamide inhibitors of ADAMTS4 and ADAMTS5, which are highly selective and potent in vitro, have emerged [85]. In addition, a new family of ADAMTS5 inhibitors such as non-hydroxamic inhibitors, which display a 1,2,4-triazole-3-thiol scaffold as a putative zinc-binding group, have reached a reasonable level of selectivity toward ADAMTS5 [86].

N-((8-Hydroxy-5-substituted-quinolin-7-yl)(phenyl)methyl)-2-phenyloxy/amino-acetamide inhibitors have been synthesized, and four of these inhibitors demonstrated greater ADAMTS5 potency and selectivity over ADAMTS4 and MMP-13 [87]. In addition, 5-((1H-Pyrazol-4-yl)methylene)-2-thioxothiazolidin-4-one inhibitors have been synthesized and have shown good selectivity of ADAMTS5 over ADAMTS4 [88]. Yet another series of compounds, α-glutamic acid scaffold based 4-(benzamido)-4-(1,3,4-oxadiazol-2-yl) butanoic acids, have been designed and synthesized to inhibit the activity of both ADAMTS4 and ADAMTS5 [89]. Furthermore, a series of 1-sulfonylaminocyclopropanecarboxylates and N-substituted sulfonylamino-alkanecarboxylates are potent ADAMTS5 inhibitors with good selectivity over MMPs such as MMP-1 [90]. More recently, potent and selective novel ADAMTS5 inhibitor scaffolds which lacked a zinc-binding motif and which contained a 1,3,5-triazine core were designed [91].

A recent study ameliorated OA progression in a spontaneous OA mouse model by using intra-articular injections of an anti-ADAMTS5 antibody (CRB0017), showing the first evidence of a biological agent (antibody-mediated targeting) used against ADAMTS5 to halt the progression of OA (Figures 1 and 2) and a proof of principle that inhibiting this enzyme might be a promising therapeutic [92]. The aggrecanase inhibitor AGG-523 (Wyeth Pharmaceuticals, now part of Pfizer Inc., New York, NY, USA), which inhibits both ADAMTS4 and ADAMTS5, has undergone clinical trials for the treatment of OA and may become a new OA drug in the near future. Although it has yet to be established which aggrecanase is predominant in human arthritis, ADAMTS4 inhibitors still may be beneficial, particularly if they inhibit ADAMTS5 concurrently (Figures 1 and 2 and Table 2).