TIMP-3 is a potent inhibitor of aggrecanase 1 (ADAM-TS4) and aggrecanase 2 (ADAM-TS5).

The proteoglycan aggrecan is an important major component of cartilage matrix that gives articular cartilage the ability to withstand compression. Increased breakdown of aggrecan is associated with the development of arthritis and is considered to be catalyzed by aggrecanases, members of the ADAM-TS family of metalloproteinases. Four endogenous tissue inhibitors of metalloproteinases (TIMPs) regulate the activities of functional matrix metalloproteinases (MMPs), enzymes that degrade most components of connective tissue, but no endogenous factors responsible for the regulation of aggrecanases have been found. We show here that the N-terminal inhibitory domain of TIMP-3, a member of the TIMP family that has functional properties distinct from other TIMPs, is a strong inhibitor of human aggrecanases 1 and 2, with K(i) values in the subnanomolar range. This truncated inhibitor, which lacks the C-terminal domain that is responsible for interactions with molecules other than active metalloproteinases, is produced at high yield by bacterial expression and folding from inclusion bodies. This provides a starting point for developing a biologically available aggrecanase inhibitor suitable for the treatment of arthritis.


The proteoglycan aggrecan is an important major component of cartilage matrix that gives articular cartilage the ability to withstand compression. Increased breakdown of aggrecan is associated with the development of arthritis and is considered to be catalyzed by aggrecanases, members of the ADAM-TS family of metalloproteinases. Four endogenous tissue inhibitors of metalloproteinases (TIMPs) regulate the activities of functional matrix metalloproteinases (MMPs), enzymes
that degrade most components of connective tissue, but no endogenous factors responsible for the regulation of aggrecanases have been found. We show here that the N-terminal inhibitory domain of TIMP-3, a member of the TIMP family that has functional properties distinct from other TIMPs, is a strong inhibitor of human aggrecanases 1 and 2, with K i values in the subnanomolar range. This truncated inhibitor, which lacks the C-terminal domain that is responsible for interactions with molecules other than active metalloproteinases, is produced at high yield by bacterial expression and folding from inclusion bodies. This provides a starting point for developing a biologically available aggrecanase inhibitor suitable for the treatment of arthritis.
Tissue inhibitors of matrix metalloproteinases (TIMPs) 1 are important regulators of matrix metalloproteinases (MMPs) that participate in the degradation of the extracellular matrix (1). To date, four isoforms of TIMP have been identified in humans that are designated TIMP-1, -2, -3, and -4 (2); these are homologous in sequence and have similar secondary and tertiary structures including six well conserved disulfide bonds. Structural and functional studies of TIMP-1 and TIMP-2 (3)(4)(5)(6) have shown that the full inhibitory activity of TIMPs resides in the N-terminal domain that is stabilized by three disulfide bonds. Inhibition studies with recombinant TIMPs have shown that each TIMP binds to MMPs with varying degrees of affinity, implicating that they have distinct functions in vivo (2,7).
TIMP-3 was originally discovered as a transformation-induced protein in chicken fibroblasts (8), which was later shown to have inhibitory activity against MMPs (9). In addition to its function as an inhibitor of MMPs, TIMP-3 has been reported to inhibit the shedding of cell surface-anchored molecules such as tumor necrosis factor-␣ receptor (10), L-selectin (11), interleukin 6 receptor (12), and syndecans-1 and -4 (13). The release of these molecules is thought to be catalyzed by membrane-bound ADAMs (a disintegrin and a metalloproteinase domain), multidomain proteins containing an N-terminal propeptide, a metalloproteinase, a disintegrin-like, a transmembrane, and a cytoplasmic domain. The primary structures of the metalloproteinase domains of the MMPs and the AD-AMs have little sequence similarity except near the catalytic Zn 2ϩbinding motif, HEXXHXXGXXH (14). Direct evidence for the apparently unique ability of TIMP-3 to inhibit a broad spectrum of metalloproteinases is provided by the demonstration of its inhibitory action against TACE or ADAM-17 (15) and ADAM-10 (16). These properties suggest that TIMP-3 has an important and distinct role in regulating ADAM activities in biological systems.
Recently a subgroup of eight ADAMs have recently been identified, designated ADAM-TS proteinases. The ADAM-TS group, unlike typical membrane-anchored ADAMs, lacks a transmembrane domain and a cytoplasmic domain at the C terminus. Instead, these metalloproteinases contain a varying number of thrombospondin type-1 domains (17). Among them, ADAM-TS4 and ADAM-TS5, purified from interleukin-1treated bovine articular cartilage, cleave cartilage aggrecan efficiently and thus are referred to as aggrecanase 1 and aggrecanase 2 (18,19). These enzymes are considered to play an important role in the breakdown of aggrecan in cartilage under physiological and pathological conditions (18 -22). Until this study, however, no endogenous inhibitor of the aggrecanases has been identified. Here we show that the N-terminal domain of TIMP-3 (N-TIMP-3), expressed at high yield in Escherichia coli and folded from inclusion bodies to give a functional metalloproteinase inhibitor, is a highly potent inhibitor of both aggrecanases. This truncated protein, or variants engineered to enhance the specificity for aggrecanases, may represent a route for the treatment of arthritis.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes were obtained from New England BioLabs. ABC Western blot analysis kits were purchased from Seikagaku Kogyo, Japan. The cDNA encoding human TIMP-3 was obtained from a human placenta cDNA library. Human recombinant TIMP-1 was expressed in CHO K-1 cells and purified from the conditioned medium (23), and human TIMP-2 was purified from human uterine cervical fibroblasts as described previously (24). Recombinant human TIMP-4, was expressed in E. coli. 2 Recombinant human MMP-1 lacking the C-terminal domain (MMP-1⌬C), MMP-2, and MMP-3 lacking the C-terminal domain (MMP-3⌬C) were prepared and activated as described previously (24 -26). Human recombinant ADAM-TS4 and * This work was supported by National Institutes of Health Grant AR 40994 and Wellcome Trust Grants Nos. 057508 and 061707. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Construction of the N-TIMP-3 Bacterial Expression
System-Human TIMP-3 cDNA encoding the N-terminal region of mature human TIMP-3, residues Cys 1 to Asn 121 , was amplified by PCR using a Vent PCR kit (New England BioLabs). The reactions were carried out for 25 cycles at 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min. An additional initiator methionine (bold) together with a NdeI restriction site at the 5Ј-end and a NotI site (underlined) at the 3Ј-end were introduced by specific primers, 5Ј-GTCATATGTGCACATGCTCG-3Ј (forward) and 5Ј-CGGCCGCGTTACAACCCAGGTG-3Ј (reverse). The PCR products were cloned into the pET42 vector (Novagen) using the NdeI and NotI sites to produce the coding sequence for truncated TIMP-3 (N-TIMP-3) with a His tag attached to the C terminus.
Expression in E. coli and Folding of N-TIMP-3-The plasmids containing wild-type N-TIMP-3 were transformed into BL21 (DE3) cells (Novagen), cultured in 3 liters of Luria-Bertani medium containing 50 g/ml kanamycin and induced with 1 mM isopropyl-␤-D-thiogalactopyranoside. After 3 h, the cells were harvested by centrifugation at 4,000 rpm for 20 min, washed twice with ice-cold wash buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 1 mM EDTA), and broken with a French press. Inclusion bodies were collected by centrifugation at 12,000 rpm for 20 min at 4°C, and the recombinant protein was extracted with 6 M guanidine hydrochloride, 50 mM Tris-HCl, pH 8.0, and 10 mM ␤-mercaptoethanol at room temperature with constant stirring for 2 h. The solution containing extracted protein was applied to a 5-ml Ni 2ϩ -NTAagarose column (Qiagen) and washed with 50 mM Tris-HCl, pH 8.0 containing 20 mM imidazole and 6 M guanidine hydrochloride. The column was subsequently washed with a solution composed of 4 volumes of 50 mM Tris-HCl, pH 8.0 containing 6 M guanidine hydrochloride and 6 volumes of isopropyl alcohol to remove bacterial endotoxins, and the protein was eluted with 50 mM Tris-HCl, pH 8.0 containing 60 mM imidazole and 6 M guanidine hydrochloride. The product was homogeneous on SDS-PAGE. The eluted protein was diluted to 20 g/ml with 50 mM Tris-HCl, pH 8.0 containing 20% glycerol and 6 M guanidine hydrochloride, treated with 20 mM cystamine with stirring for 16 h at 4°C. The solution was then dialyzed twice against 15 volumes of 50 mM Tris-HCl, pH 8.0 containing 20% glycerol, 150 mM NaCl, 10 mM CaCl 2 , 5 mM ␤-mercaptoethanol, and 1 mM 2-hydroxyethyl disulfide for 24 h at 4°C, twice against 20 mM Tris-HCl, pH 8.0 containing 20% glycerol for 8 h at 4°C and then centrifuged at 12,000 rpm for 30 min at 4°C. The supernatant was applied to a column (2.5 ϫ 5 cm) of carboxymethyl cellulose (Whatman CM-52) that had been equilibrated with 20 mM Tris-HCl, pH 8.0, and the bound protein was eluted using a linear gradient ranging from 0 to 1 M NaCl in the same buffer; all buffers contained 20% (v/v) glycerol. The concentration of N-TIMP-3 was determined by absorbance at 280 nm using the extinction coefficient (⑀ 280 ϭ 17,570 M Ϫ1 cm Ϫ1 ) of N-TIMP-3.
Inhibition Kinetic Studies-N-TIMP-3 was characterized as an inhibitor of MMP-1⌬C, MMP-2, MMP-3⌬ as described previously for N-TIMP-1 (4, 23). Inhibition constants (K i values) were calculated as described by Morrison and Walsh (30). For titration of N-TIMP-3, various concentrations of the inhibitor were incubated with MMP-1⌬C (100 nM) for 1 h at 37°C, and the residual activity was measured using 1 M Knight substrate.
Aggrecanase Assay-Activities of ADAM-TS4 and ADAM-TS5 were measured by incubating enzyme with purified bovine aggrecan (500 nM) in 100 l of 50 mM Tris-HCl buffer, pH 7.5, containing 0.1 M NaCl and 10 mM CaCl 2 , for 2 h at 37°C and terminating the reaction with 10 mM EDTA. The digestion products were then deglycosylated with chondroitinase ABC (0.1 units/10 g aggrecan) and then with keratinase (0.1 units/10 g aggrecan) and keratinase II (0.002 units/10 g aggrecan) for 2 h at 37°C in 0.1 M Tris-HCl, pH 6.5, containing 50 mM sodium acetate. The enzymatically treated products were analyzed by Western blotting using BC-3 antibody (27) or an antibody against the GELE 1480 neoepitope (28). To determine apparent inhibition constant K i(app) values of inhibitors against the aggrecanases, ADAM-TS4 or ADAM-TS5 (at a final concentration of 50 pM) was incubated with various concen-trations of the inhibitor in 44 l of the above buffer at room temperature for 30 min and then a solution of bovine aggrecan (5.5 l) was added. The reaction products were detected with anti-GELE 1480 antibody. The concentrations of ADAM-TS4 and ADAM-TS5 were confirmed by titration with recombinant N-TIMP-3.

RESULTS
Expression and Folding of N-TIMP-Approximately 100 mg of unfolded N-TIMP-3 was purified by Ni 2ϩ -NTA affinity chromatography of the inclusion bodies from 3 liters of bacterial culture. Separation of the soluble protein obtained after in vitro folding by ion exchange chromatography with CM-52 cellulose gave a single protein peak. SDS-PAGE analysis under reducing conditions showed a single band with an apparent molecular weight of 16 kDa following staining with Coomassie Blue R250, in agreement with a molecular mass calculated for N-TIMP-3 (including the His tag), of 15,230. A slightly faster electrophoretic mobility under nonreducing conditions indicates that the intrachain disulfide bonds are present (Fig. 1, inset). Assays for MMP-1 inhibition indicate that the specific activity of the later fractions in the protein peak was greater than that of the earlier fractions. Titration of MMP-1 with the later fractions showed full enzyme inhibition with a 1:1 stoichiometry, whereas the earlier fractions exhibited only partial inhibition (Fig. 1). Approximately 300 g of fully active N-TIMP-3 was obtained from 6 mg of the unfolded protein, corresponding to an overall yield of over 1.5 mg/liter of bacterial culture. N-TIMP-3 can be concentrated up to 0.2 mg/ml without precipitation. Only the fractions that gave total inhibition of MMP-1 with a 1:1 molecular stoichiometry were used for characterization of the recombinant N-TIMP-3.  (23) indicating distinct specificities that must be dependent on differences in the structures of the N-terminal domains of these two TIMPs. It is notable that the affinity of N-TIMP-3 for MMP-1⌬C was greater than that for MMP-2 and MMP-3⌬C.

Inhibition of MMPs by N-TIMP-3-N-TIMP
Inhibition of ADAM-TS4 and ADAM-TS5-The ability of N-TIMP-3 to inhibit ADAM-TS4 and ADAM-TS5 was examined first by detecting aggrecan cleavage at the classical aggrecanase site, the Glu 373 -Ala 374 bond, using BC-3 antibody. As shown in Fig. 2, A and B, the actions of ADAM-TS4 and ADAM-TS5 were completely inhibited by N-TIMP-3 at the concentration of 250 and 25 nM, respectively. By contrast, TIMP-1 or TIMP-2 even at a concentration of 1 M did not inhibit these enzymes (data not shown). However, the assay for aggrecanase activity, which detects the cleavage at the Glu 373 -Ala 374 bond,

Inhibition of ADAM-TS4 and ADAM-TS5 by TIMP-3 12502
was not suitable for determining inhibition constants of N-TIMP-3 because it required at least 1 nM aggrecanase. Therefore, a quantitative and sufficiently sensitive assay system was necessary. This was accomplished by using anti-GELE antibody that recognizes the fragment cleaved at the Glu 1480 -Gly 1481 bond, a site recently shown to be cleaved readily by aggrecanases (28). With 50 pM ADAM-TS4, the release of the GELE-fragment was linear for up to 2 h at 37°C (data not shown) and with 50 pM ADAM-TS5 for up to 4 h (Fig. 3). Thus, 50 pM aggrecanases was used to quantify the inhibitory activity of inhibitors. As shown in Fig. 2, C and D, N-TIMP-3 and a synthetic hydroxamate inhibitor BB-16 at a concentration of 250 nM completely blocked cleavage of aggrecan by ADAM-TS4 and ADAM-TS5, whereas TIMP-1 and -2 showed little or no inhibition. TIMP-4, however, at this relatively high concentration partially (about 35%) inhibited ADAM-TS4, but not ADAM-TS5. The inhibitory activity of varying concentrations of N-TIMP-3 was further measured against ADAM-TS4 and ADAM-TS5, and the apparent K i values (K i(app) ) were calculated to be 3.3 and 0.66 nM, respectively (Fig. 4, A and B). N-TIMP-3 formed a 1:1 molar stoichiometric complex with both aggrecanases (data not shown).

DISCUSSION
TIMPs expressed in connective tissues play important roles in the control of extracellular matrix metabolism, and the ability of TIMPs to inhibit the activities of MMPs in vitro has been well documented (2). TIMP-3, however, has several properties distinct from those of other TIMPs, which include its ability to bind tightly to the extracellular matrix (9,31), apoptotic effects on a number of cells (10,32), and inhibition of TACE (15) and ADAM 10 (16). This work provides additional important information, specifically that TIMP-3 is a potent endogenous inhibitor of aggrecanases, metalloproteinases that play a key role in the degradation of articular cartilage aggrecan. Inhibition of aggrecanases was not observed with TIMP-1 or TIMP-2. TIMP-4 is however a weak inhibitor of ADAM-TS4 but not ADAM-TS5.
The property of TIMP-3 binding to matrix components is considered to be important in the localized regulation of MMP activity, and its interaction with the polyanionic extracellular matrix is mainly due to the N-terminal domain of TIMP-3 (31). The inhibitor may be extracted from the tissue with sulfated compounds such as suramin and pentosan or enzymatic treatment with heparinase III or chondroitinase ABC (31), but its strong binding to the extracellular matrix has made it difficult to purify sufficient quantities of TIMP-3 for detailed characterization from tissues or even by expression of the recombinant protein in mammalian cells (33). In this study we have established a high yield bacterial expression and in vitro folding procedure for the N-terminal inhibitory domain of TIMP-3, which allows us to obtain about 5 mg of fully active TIMP-3 from a 3-liter bacterial culture.
Inhibition kinetic studies with N-TIMP-3 indicate that TIMP-3 is a relatively weak inhibitor of MMP-3 (stromelysin 1), whereas TIMP-3 inhibits MMP-1 (collagenase 1) and MMP-2 (gelatinase A) to a similar extent as TIMP-1. The most significant unique property of TIMP-3 that we have demonstrated here is that it is a potent endogenous inhibitor of   values of TIMP-3 for ADAM-TS4 and ADAM-TS5 will be at least 6-fold lower than the K i(app) values. These data suggest that a primary physiological function for TIMP-3 may be the inhibition of aggrecanases because its affinity toward ADAM-TS4 and ADAM-TS5 is much stronger than those for the three MMPs tested.
Aggrecanases are believed to play a crucial role both in the normal turnover of aggrecan in cartilage and in diseases such as osteoarthritis and rheumatoid arthritis (20 -22). On the other hand, TIMP-3 mRNA is expressed in cartilage and skeletal tissue during development of mouse embryo (34), suggesting that this inhibitor may play a role during skeletal tissue development. TIMP-3 is also expressed in normal bovine and human articular chondrocytes and in synoviocytes, and elevated TIMP-3 expression was found in human osteoarthritic synoviocytes (35). The expression of TIMP-3 in chondrocytes in culture is up-regulated by transforming growth factor-␤ (36), oncostatin M (37). Recent studies of Takizawa et al. (38) showed that the treatment of human rheumatoid synovial fibroblasts with an antiarthritic agent, calcium pentosan polysulfate, increases the synthesis of TIMP-3 protein without altering its mRNA levels, and this effect is further enhanced in cells treated with both calcium pentosan polysulfate and interleukin 1. The levels of other TIMPs and MMPs were not affected by this treatment. Because the degradation of aggrecan in cartilage occurs in the early stages of arthritis, the level of TIMP-3 in cartilage is likely to be an important factor in relation to the development of arthritis. Elevated TIMP-3 production therefore may be beneficial for protecting cartilage from degradation, because it can prevent not only the action of aggrecanases and MMPs in the cartilage but also the release of TNF-␣, one of the key inflammatory cytokines, from the synovial living cells by inhibiting TACE.
Whereas the general mechanism of inhibition of MMPs by TIMPs has been well characterized by crystallographic studies (6), it is not known which structural features confer on TIMP-3 its unique ability to inhibit metalloproteinases from families other than the MMPs. Both aggrecanases have little sequence similarity with MMPs, although the overall polypeptide folds may be predicted to be similar. The primary structures of the proteinase domains of ADAM-TS4 and ADAM-TS5 are about 48% identical to each other. They are members of the reprolysin family, but their sequence identities with adamalysins are only 20 -25% and with TACE and ADAM 10, only 10 -15%. Thus, it is not apparent how TIMP-3, but not TIMP-1, -2, or -4, can inhibit those additional groups of metalloproteinases that are only distantly related to MMPs. We are currently investigating key structural elements in TIMP-3, by mutagenesis and molecular modeling, that confer its ability to inhibit aggrecanases and other members of the ADAM family of metalloproteinases. The identification of such elements may provide us with strategies to engineer TIMPs that are more soluble than TIMP-3 and specifically inhibit aggrecanases.