2-Macroglobulin Is a Novel Substrate for ADAMTS-4 and ADAMTS-5 and Represents an Endogenous Inhibitor of These Enzymes *

Osteoarthritis is characterized by the loss of aggrecan and collagen from the cartilage extracellular matrix. The proteinases responsible for the breakdown of cartilage aggrecan include ADAMTS-4 (aggrecanase 1) and ADAMTS-5 (aggrecanase 2). Post-translational inhibition of ADAMTS-4/-5 activity may be important for maintaining normal homeostasis of aggrecan metabolism, and thus, any disruption to this inhibition could lead to accelerated aggrecan breakdown. To date TIMP-3 (tissue inhibitor of matrix metalloproteinases-3) is the only endogenous inhibitor of ADAMTS-4/-5 that has been identified. In the present studies we identify 2-macroglobulin ( 2M) as an additional endogenous inhibitor of ADAMTS-4 and ADAMTS-5. 2M inhibited the activity of both ADAMTS-4 and ADAMTS-5 in a concentration-dependent manner, demonstrating 1:1 stoichiometry with second-order rate constants on the order of 10 and 10 M 1 s , respectively. Inhibition of the aggrecanases was mediated by proteolysis of the bait region within 2M, resulting in physical entrapment of these proteinases. Both ADAMTS-4 and ADAMTS-5 cleaved 2M at Met / Gly, representing a novel proteinase cleavage site within 2M and a novel site of cleavage for ADAMTS-4 and ADAMTS-5. Finally, the use of the anti-neoepitope antibodies to detect aggrecanase-generated 2M-fragments in synovial fluid was investigated and found to be uninformative.

Loss of aggrecan from the cartilage extracellular matrix is an early and sustained feature of osteoarthritis (OA). 1 Aggrecan, the major proteoglycan in cartilage, consists of a protein backbone of 210 -250 kDa containing 3 globular domains referred to as G1 (located at the N terminus), followed by G2 and G3 (located at the C terminus) (1). Attached to the core protein between G2 and G3 are the glycosaminoglycans, chondroitin sulfate, and keratan sulfate. The chondroitin sulfate chains (100 -125 per monomer) are located in the C-terminal portion of the core protein, whereas the keratan sulfate chains  per monomer) are preferentially located toward the N terminus (2). Multiple aggrecan monomers interact with hyaluronic acid via their G1 domain to form aggregates of very high molecular weight. The negatively charged glycosaminoglycan chains are responsible for the extremely high osmotic swelling pressure of cartilage, which is counteracted by the resistance of type II collagen fibers (the other major macromolecule of cartilage). Thus, aggrecan provides cartilage with the ability to resist compressive forces, and its loss will have a severe effect upon the functionality of the cartilage.
Tight regulation of aggrecanase activity is critical for maintaining a fine balance between aggrecan anabolism and catabolism. In diseases such as OA the balance is disturbed in favor of catabolism, and this could be attributed to de novo synthesis of ADAMTS-4 (9,16) and/or post-translational activation of ADAMTS-4/-5 (17,18). Another control mechanism for aggrecan catabolism may involve endogenous inhibitors of the aggrecanases. Recently it has been shown that tissue inhibitor of matrix metalloproteinases-3 (TIMP-3) is a potent inhibitor with K i (app) values of 3.30 nM for ADAMTS-4 and 0.66 nM for ADAMTS-5 (19,20). TIMP-3 was found to have a greater affinity for ADAMTS-4 and ADAMTS-5 than for MMP-1, MMP-2, and MMP-3, suggesting that a primary physiological function for TIMP-3 is inhibition of the aggrecanases (19). Moreover, TIMP-3 has been detected in cartilage (21) and shown to have a high binding affinity for chondroitin sulfate polysaccharides, a binding affinity not found in the other 3 members of the family (TIMP-1, -2, -4) (22,23). Whether TIMP-3 is a physiologically relevant inhibitor of aggrecanase activity in cartilage has yet to be established. Recently, Mort et al. (24) show that aggrecan turnover involving MMP and aggrecanase activity are not under effective TIMP-3 control in an inflammatory arthritis model induced in TIMP-3 null mice, as both wild type and TIMP-3 Ϫ/Ϫ mice showed similar long term cartilage destruction (24). These data suggest that other inhibitors may be important in the regulation of aggrecanase activity. To date no other protein inhibitors of ADAMTS-4 and ADAMTS-5 have been identified. ␣ 2 -Macroglobulin (␣ 2 M), a general endoproteinase inhibitor, is a noncovalent tetramer of two 370-kDa disulfide-linked homodimers that circulates in blood at concentrations of ϳ2-4 mg/ml and is also found in the joint fluid at similar concentrations (25,26). ␣ 2 M is active against most endoproteinases. Each subunit of the ␣ 2 M molecule contains a region referred to as the "bait region," a short stretch of amino acids (beginning at Pro 667 and ending at Thr 705 ) that is very susceptible to proteolytic cleavage. When cleaved by an enzyme the macroglobulin changes shape in such a way as to trap the proteinase inside, resulting in inhibition of proteolytic activity by preventing substrate access to the active site of the proteinases through steric hindrance (27)(28)(29). The conformational changes initiated by bait region cleavage also cause activation of internal thiol esters formed from Cys 949 and Glx 952 in each subunit of ␣ 2 M (30). The activated thiol esters provide ␣ 2 M with a potential for covalent cross-linking of the activating proteinase through ⑀-lysyl (proteinase)-␥ glutamyl-(␣ 2 M) bonds and also for binding of other nucleophils present at activation (31,32). ␣ 2 M does not inhibit endoproteinases that are highly specific for one or a limited number of sequences, such as tissue kallikrein, urokinase, coagulation factor XIIa, and endoproteinase Lys-C (27). Whether ␣ 2 M can inhibit ADAMTS-4 and ADAMTS-5 activity by proteolysis of the bait region has not been investigated, although it has been shown that a related enzyme, ADAMTS-1, which can cleave purified aggrecan in vitro in the IGD at Glu 373 , forms an SDS-stable complex with ␣ 2 M (14,33). Analysis of the 39 amino acids representing the bait region reveals 5 glutamic acids, Glu 673 , Glu 678 , Glu 686 , Glu 701 , and Glu 702 , raising the possibility that ADAMTS-4/-5 may cleave ␣ 2 M. Therefore, the goals of the present studies were to determine 1) if ␣ 2 M can inhibit ADAMTS-4/-5 cleavage of aggrecan, 2) if inhibition is triggered by proteolysis of the bait region within the macroglobulin molecule, and 3) the aggrecanase site of cleavage within ␣ 2 M.
Neoepitope Antibodies-Polyclonal neoepitope antibodies were prepared (Quality Controlled Biochemicals, Hopkinton, MA) to the peptide sequences YESDVM and GRGHAR that are present on the C terminus and N terminus of the ␣ 2 M subunit, respectively, generated by cleavage at the Met 690 /Gly 691 bond. A polyclonal neoepitope antibody to the peptide sequence AGEGPSGI that is present on the N terminus of aggrecan fragments generated by cleavage at the Glu 1771 /Ala 1772 bond was prepared as described previously (10). Intact undigested aggrecan was not detected by the anti-AGEGPSGI antibody. Only the immunizing peptide for the antibody blocked binding to the antigen, whereas the spanning peptide (TQAPTAQEAGEGPSGI) did not block binding, indicating that the antibody does not recognize the same sequence when present within uncleaved aggrecan.
Inhibition of ADAMTS-4/-5 by ␣ 2 M-45 l of 2ϫ buffer (100 mM Tris-HCl, 20 mM CaCl 2 , and 200 mM NaCl, pH 7.5) were added to 25 l of buffer (5 mM Tris, 10 mM NaCl, pH 7.5) containing ADAMTS-4 or -5 followed by 25 l of buffer (5 mM Tris, 10 mM NaCl, pH 7.5) containing ␣ 2 M at a range of concentrations (concentrations based on the M r of the tetramer). After incubation of samples for 2 h at 37°C, 5 l of aggrecan in water were added, and the samples were incubated for 30 min at 37°C. The concentration of aggrecan was determined in the dimethylmethylene blue assay as previously described (36). Final concentrations in the reaction mixture were 25 nM ADAMTS-4 or ADAMTS-5, 500 nM aggrecan, and ␣ 2 M at 0, 5, 10, 15, 20, 27.5, 30, 40, 50, 75, 100, and 150 nM. Reactions were quenched by adding EDTA to a final concentration of 50 mM. The products were analyzed for cleavage of aggrecan at Glu 1771 by Western blot analysis using the 1772 AGEG neoepitope antibody. The densitometric response was found to be linear over the density ranges required for the blots as assessed by loading varying amounts of AGEG product and quantifying the immunoreactive bands by scanning densitometry using the software OneDscan. Active site concentrations were determined by fitting these data to the equation (37), where y is the measured 1772 AGEG product, y max is the amount of product in the absence of ␣ 2 M, k i is the second-order rate constant, t is time, and E o and I o are the initial concentrations of enzyme and ␣ 2 M, respectively. For this purpose the nonlinear curve-fitting program "the nonlinear regression software package GraFit" (Erithacus Software Ltd., Staines, UK) was used, and the second-order rate constants were fixed at 10 5 and 10 4 M Ϫ1 s Ϫ1 for ADAMTS-4 and -5, respectively.
Time-dependent inhibition of ADAMTS-4/-5 by ␣ 2 M-45 l of 2 ϫ buffer (100 mM Tris-HCl, 20 mM CaCl 2 , and 200 mM NaCl, pH 7.5) were added to 25 l of buffer containing ADAMTS-4 at 2 nM or -5 at a concentration of 20 nM followed by 25 l of buffer containing ␣ 2 M at 10 or 30 nM (ADAMTS-4 experiment) or ␣ 2 M at 100 or 300 nM (ADAMTS-5 experiment). After the incubation for various times at 37°C, aggrecan was added to a final concentration of 500 nM in a volume of 5 l, and samples were incubated for 30 min at 37°C. The reactions were quenched by adding EDTA to a final concentration of 50 mM, and the products were analyzed for cleavage of aggrecan at Glu 1771 by Western blot analysis using the 1772 AGEG neoepitope antibody. Control experiments showed no inhibition of ADAMTS-4 or -5 activity if ␣ 2 M and aggrecan were added at the same time without any preincubation. Thus, aggrecan appeared to be a completely effective inhibitor of ␣ 2 M action under these conditions.
Calculation of Second Order Rate Constants-Time-dependent inhibition data was analyzed by several different methods to estimate second-order rate constants for the reaction between ␣ 2 M and AD-AMTS-4 and -5. The time course of inactivation at each concentration of enzyme and ␣ 2 M was fit to a single exponential to yield observed inactivation rates of 0.109 and 0.330 min Ϫ1 for 0.5 nM ADAMTS-4 in the presence of 2.5 and 7.5 nM ␣ 2 M and 0.115 and 0.328 min Ϫ1 for 5 nM ADAMTS-5 in the presence of 25 and 75 nM ␣ 2 M, respectively. The linear dependence of inactivation rate on ␣ 2 M concentration suggested that at these concentrations it was proper to analyze this reaction as a simple second-order process. Dividing these observed rates by the ␣ 2 M concentration gave estimates for the second-order rate constants for ADAMTS-4 and -5, respectively. Inactivation data was also fit globally using the numerical integration software package, DYNAFIT (38) using the simple second-order model E ϩ I 3 EI to yield estimates for ADAMTS-4 and -5, respectively. Finally, because of the scarcity of data points during the 0 -10-min time range when most of the inactivation took place, the second-order rate constant was calculated directly (i.e. without curve fitting) from the extent of inactivation observed at 5 and/or 10 min using the equation Inactivation of ␣ 2 M by Methylamine-10 mg/ml ␣ 2 M was incubated with 100 mM sodium borate, pH 8.6, in the presence or absence of 400 mM methylamine for 2 h at 37°C (39). Subsequently, the untreated and methylamine-treated ␣ 2 M were tested for their ability to inhibit AD-AMTS-4 and ADAMTS-5.
Cleavage of ␣ 2 M by ADAMTS-4/-5-ADAMTS-4 and ADAMTS-5 at a concentration of 200 or 112 nM, respectively, were incubated with 250 nM ␣ 2 M in buffer containing 50 mM Tris-HCl, 100 mM NaCl, and 10 mM CaCl 2 , pH 7.5, for 4 h at 37°C. Subsequently, the products were analyzed for total protein by SDS-PAGE on 8% gels followed by staining with Coomassie Brilliant Blue R-250.
Cleavage of ␣ 2 M by Truncated ADAMTS-4 -250 nM ␣ 2 M was incubated with 200 mM mature truncated ADAMTS-4 (Phe 213 -Pro 431 ) in buffer containing 50 mM Tris, 100 mM NaCl, 10 mM CaCl 2 , pH 7.5, for 2 h at 37°C. After the digestions cleaved ␣ 2 M was detected by SDS-PAGE using an 8% gel followed by Coomassie staining. Truncated ADAMTS-4, lacking the C-terminal disintegrin-like and thrombospondin domains (Met 1 -Pro 431 ), was expressed in Drosophila cells. When truncated or full-length ADAMTS-4 are expressed in these cells, the prodomain is removed by cleavage at Arg 212 /Phe 213 by a furin-like activity found in insect cells, resulting in the production of active enzyme. No latent form of the enzyme was detected in the conditioned media.
Aggrecan Isolation-Nasal septa were removed from bovine noses obtained fresh from the slaughterhouse (Covance, Denver, PA). The cartilage was sliced, and aggrecan was extracted by stirring at 4°C for 48 h in 10 volumes of 4 M guanidinium-HCl in 0.05 M sodium acetate, pH 5.8, containing the protease inhibitors, 0.01 M EDTA, 0.1 M 6-aminohexanoic acid, 2 mM phenylmethanesulfonyl fluoride, and 0.05 M benzamidine HCl. Aggrecan monomers were isolated by equilibrium density gradient centrifugation in cesium chloride, and the bottom fraction of this gradient (d Ͼ 1.54 g/ml) containing aggrecan monomers was dialyzed at 4°C against water, lyophilized, and stored at Ϫ20°C.
Deglycosylation of Aggrecan Products-For analysis of fragments by Western blot aggrecan was enzymatically deglycosylated with chondroitinase ABC (0.1 unit/10 g of aggrecan), keratanase (0.1 unit/10 g of aggrecan), and keratanase II (0.002 unit/10 g of aggrecan) for 3 h at 37°C in buffer containing 50 mM sodium acetate, 0.1 M Tris/HCl, pH 6.5. After digestion the aggrecan was precipitated with 5 volumes of acetone and reconstituted in 30 l of Tris-glycine SDS sample buffer containing 2.5% ␤-mercaptoethanol and heated for 10 min at 95°C.
Western Blot Analysis-␣ 2 M or aggrecan product for each sample was loaded on a 4 -12% Tris-glycine gel and separated by SDS-PAGE under reducing conditions. The separated proteins were transferred to polyvinylidene difluoride membranes and immunolocated with a 1:1000 dilution of YESDVM 690 or 691 GRGHAR (for detecting ␣ 2 M) or 1772 AGEG (for detecting aggrecan) neoepitope antibody. Subsequently, the membranes were incubated with a 1:5000 dilution of goat anti-rabbit IgG alkaline phosphatase conjugate as the secondary antibody. Products were visualized by developing the blots in nitro blue tetrazolium/5bromo-4-chloro-3-indolyl phosphate color developing reagent. Overnight transfer resulted in complete transfer of both low and high molecular fragments, and the densitometric response was found to be linear over the density ranges required for the blots.
N-terminal Sequencing-␣ 2 M at a concentration of 1 M was incubated with ADAMTS-4 and ADAMTS-5 at a concentration of 200 or 448 nM, respectively for 4 h at 37°C. After the incubation, the products were separated by SDS-PAGE (8% acrylamide) and transferred to polyvinylidene difluoride membranes. The proteins were detected by staining with Coomassie Brilliant Blue R-250, and the bands of interest were excised, placed onto a polystyrene-treated glass filter, and sequenced in an Applied Biosystem 477A plus liquid sequencer with on-line 120A phenylthiohydantoin analysis.
Cleavage of ␣ 2 M Peptides-The following peptides were synthesized on an ABI solid phase peptide synthesizer using standard Fmoc (N-(9fluorenyl)methoxycarbonyl) protection chemistry: MGRG, VMGRGH, DVMGRGHA, SDVMGRGHAR, ESDVMGRGHAR, and YESD-VMGRGHAR. 10 M of each peptide (100 l of total volume) in buffer containing 100 mM Tris, 100 mM NaCl, 10 mM CaCl 2 , 0.05% Brij-35, pH 7.5, were incubated with ADAMTS-4 or ADAMTS-5 at a concentration of 50 nM. The digests were carried out for 18 h at 37°C and analyzed by reverse phase high pressure liquid chromatography followed by electrospray mass spectrometry to verify cleavage and determine the site of cleavage from the mass of the fragments if generated.
Detection of the GRGH Neoepitope in Osteoarthritic Synovial Fluid-Synovial fluids were collected from patients with radiographic and symptomatic knee OA. Synovial fluids were diluted 100-fold in buffer containing 50 mM Tris, 100 mM NaCl, 10 mM CaCl 2 , pH 7.5, and either analyzed for the presence of the GRGH neoepitope by Western blot analysis or incubated with ADAMTS-4 at a concentration of 50 nM for 1 h at 37°C. After the reactions cleavage of endogenous ␣ 2 M at Met 690 / Gly 691 was monitored by Western blot analysis using the polyclonal antibody against ␣ 2 M or the 691 GRGHAR neoepitope antibody.

Inhibition of ADAMTS-4 and ADAMTS-5 by ␣ 2 M-
We first examined the ability of ␣ 2 M to block ADAMTS-4 and AD-AMTS-5 cleavage of aggrecan. ADAMTS-4 or ADAMTS-5 at a concentration of 25 nM (determined by active site titration with TIMP-3) were preincubated with ␣ 2 M at varying concentrations for 2 h at 37°C. After the preincubation aggrecan was added to a final concentration of 500 nM, and the reactions were carried out for 30 min at 37°C. The samples were subsequently analyzed for aggrecan products generated by cleavage at the Glu 1771 /Ala 1772 bond by Western blot analysis using the 1772 AGEG neoepitope antibody. ␣ 2 M effectively inhibited both ADAMTS-4 and ADAMTS-5 cleavage of aggrecan in a concentration-dependent manner (Fig. 1, A and B), showing almost complete inhibition at 100 and 150 nM, respectively. The concentration of ADAMTS-4 and ADAMTS-5 determined by active site titration with ␣ 2 M (14 and 57 nM, respectively) are in alignment with the concentrations determined by TIMP-3 titration (25 nM).
Time-dependent Inhibition of ADAMTS-4 and ADAMTS-5 by ␣ 2 M-ADAMTS-4 at a concentration of 0.5 nM was preincubated with ␣ 2 M at a concentration of 2.5 and 7.5 nM for varying periods of time at 37°C. After the preincubation aggrecan was added to a final concentration of 500 nM, and the reactions were incubated for 30 min at 37°C. The products were analyzed by Western blot analysis using the 1772 AGEG neoepitope antibody. Inhibition of ADAMTS-4 ( Fig. 2) was concentration-and time-dependent. No inhibition of aggrecan cleavage at the Glu 1771 /Ala 1772 bond was observed when ␣ 2 M was not preincubated with ADAMTS-4, but after a 10-min preincubation ϳ100 and ϳ68% inhibition was achieved with ␣ 2 M at concentrations of 2.5 and 7.5 nM, respectively. Similar results were obtained when the inhibition of 5 nM ADAMTS-5 by 25 and 75 nM ␣ 2 M was examined (data not shown). A variety of approaches yielded estimates for the second-order rate constants for the reaction of ␣ 2 M with ADAMTS-4 and ADAMTS-5 to be on the order of 10 6 and 10 5 M Ϫ1 s Ϫ1 , respectively (see "Experimental Procedures").
ADAMTS-4 and ADAMTS-5 Cleave ␣ 2 M-␣ 2 M inhibits most proteinases by physical entrapment upon cleavage within the bait region by the enzyme. This process is disturbed by the treatment of ␣ 2 M with methylamine, which breaks several key thioester bonds important to the structural integrity of the macroglobulin molecule. After breakage of these bonds the protein undergoes conformational changes, resulting in a less exposed bait region, thus inactivating the "trapping" mechanism (40). Methylamine-treated ␣ 2 M failed to inhibit AD-AMTS-4 and ADAMTS-5 (data not shown), suggesting that the inhibition of these two proteinases by ␣ 2 M is triggered by proteolysis of the bait region and not through allosteric binding and/or steric hindrance.
To confirm that ␣ 2 M is cleaved by ADAMTS-4 and AD-AMTS-5, 250 nM ␣ 2 M was incubated with 200 nM ADAMTS-4 or 112 nM ADAMTS-5 for 2 h at 37°C, and the products were separated by SDS-PAGE on an 8% gel and analyzed for total protein by Coomassie staining. Intact ␣ 2 M in its monomeric form was detected at a molecular mass of ϳ180 kDa. In the presence of ADAMTS-4 most of the intact ␣ 2 M was converted to a lower molecular mass species of ϳ98 kDa (Fig. 3A). In the presence of ADAMTS-5 about 50% of intact ␣ 2 M was converted to the 98-kDa species (Fig. 3A). These levels of conversion are consistent with 1:1 stoichiometry between inhibitor and enzyme. To confirm that cleavage of ␣ 2 M by ADAMTS-4 and ADAMTS-5 was not due to contaminating proteinases present in our recombinant preparations, the study was repeated in the presence or absence of a mixture of protease inhibitors including E64 (cysteine proteases), phenylmethylsulfonyl fluoride (serine proteases), pepstatin A (aspartic proteases), and XS309 (matrix metalloproteinases) or in the presence of BB-16, which inhibits matrix metalloproteinases and ADAMTS-4 and -5. The mixture of protease inhibitors did not block conversion of intact ␣ 2 M by ADAMTS-4 (Fig. 3B) or ADAMTS-5 (data not shown) to the 98-kDa form. In contrast, BB-16, which is a potent aggrecanase inhibitor, completely blocked conversion. These results suggest that cleavage of ␣ 2 M in these studies is mediated by ADAMTS-4 and ADAMTS-5 and not a contaminating protease. The higher molecular weight bands observed in lanes 2 and 3 reacted with a polyclonal antibody that recognizes the catalytic domain of ADAMTS-4 (35), confirming that these bands represent a covalent complex between activated ␣ 2 M and the proteinase (data not shown).
Truncated ADAMTS-4 (Phe 213 -Pro 481 ) Cleaves ␣ 2 M-It has been shown that the thrombospondin motif of ADAMTS-4 is important for binding and cleavage of aggrecan. In fact, a truncated form of ADAMTS-4 lacking the disintegrin and thrombospondin motifs located within the C-terminal portion of the molecule does not readily cleave native aggrecan, even at concentrations as high as 200 nM, but maintains good activity against several peptide substrates (34). To determine whether the C-terminal regions of ADAMTS-4 are important for recognition and cleavage of ␣ 2 M, the ␣ 2 M protein was incubated with truncated ADAMTS-4 (Phe 213 -Pro 481 ) for 2 h at 37°C. In the presence of truncated ADAMTS-4 complete conversion of intact ␣ 2 M to the 98-kDa species was observed, suggesting that the C-terminal domains of ADAMTS-4 are not required for cleavage of the macroglobulin molecule (Fig. 4).

ADAMTS-4 and ADAMTS-5 Cleave ␣ 2 M within the Bait Region at Met 690 -
To determine the site of cleavage(s) within ␣ 2 M the protein was incubated in the presence or absence of ADAMTS-4 or ADAMTS-5 for 6 h at 37°C, and the products were analyzed by Coomassie stain. Analysis of the ADAMTS-4-␣ 2 M and the ADAMTS-5-␣ 2 M complex showed typical fragments of the ␣ 2 M subunits by SDS-PAGE (Fig. 5). N-terminal sequencing of the 98-kDa band, generated by both ADAMTS-4 and ADAMTS-5, revealed two N termini, 1 SVSGKPQYMVLV and 691 GRGHARLVHEEP, indicating that both ADAMTS-4 and ADAMTS-5 cleaved the subunit at the Met 690 /Gly 691 bond (Fig. 5). This site of cleavage is not generated by other proteases inhibited by ␣ 2 M, including collagenase, stromelysin, and trypsin and was found to be unique (41). Low levels of other N termini were detected, but the signal was too low to accurately identify (Ͻ0.2 pmol), indicating that ADAMTS-4 and ADAMTS-5 may cleave at additional sites within the bait region although much less efficiently than at the Met 690 /Gly 691 bond.
Several peptides spanning the cleavage site Met 690 /Gly 691 were synthesized (MGRG, VMGRGH, DVMGRGHA, SD- VMGRGHAR, ESDVMGRGHAR, and SDVMGRGHAR) and analyzed for their ability to be hydrolyzed by ADAMTS-4 and ADAMTS-5. Surprisingly, none of the ␣ 2 M peptides was cleaved by ADAMTS-4 or ADAMTS-5, suggesting that in addition to the site of cleavage the local conformation around the bait region is required for hydrolysis (data not shown). This inability to cleave a peptide spanning an enzymatic cleavage site was not observed with trypsin, which is known to cleave ␣ 2 M after Arg as all peptides containing an internal Arg were cleaved by trypsin.

Development of Neoepitope Antibodies That Recognize the New N Terminus ( 691 GRGHAR) and the New C Terminus (YESDVM 690 ) of ␣ 2 M Fragments Generated upon Cleavage by ADAMTS-4 and -5-
A polyclonal antibody to the new N terminus 691 GRGHAR generated by cleavage of the ␣ 2 M subunit at the Met 690 /Gly 691 bond by ADAMTS-4 and ADAMTS-5 was developed, as described under "Experimental Procedures." To test the ability of the antibody to detect ADAMTS-4/-5 generated ␣ 2 M fragments, 50 ng of ␣ 2 M was digested with 100 ng of ADAMTS-4 for varying periods of time at 37°C. After the incubations, products were analyzed for 691 GRGHAR macroglobulin fragments by SDS-PAGE using 4 -12% gels followed by Western blot analysis. ADAMTS-4 generated a ϳ98-kDa 691 GRGHAR-containing fragment that increased in band intensity over time (Fig. 6A). This cleavage was a result of aggrecanase activity and not due to a contaminating proteinase in the enzyme preparations as both EDTA and the ADAMTS-4/-5 inhibitor, SC81956, blocked the generation of GRGHAR neoepitopes by ADAMTS-4 (data not shown). To demonstrate specificity of the GRGHAR polyclonal antibody, the immunizing (GRGHAR) and spanning (YESDVMGRGHAR) peptides were analyzed for their ability to block antibody binding to AD-AMTS-4-generated 691 GRGHAR fragments. The immunizing peptide at a concentration of 1 M completely blocked binding to the 98-kDa 691 GRGHAR fragment, whereas the spanning peptide did not block, confirming specificity of the antibody (Fig. 6A).
Similarly, a polyclonal antibody to the new C terminus YES-DVM 690 , also generated by cleavage of the ␣ 2 M subunit at the Met 690 /Gly 691 bond, was produced. As was seen with the GRGHAR antibody analysis of products of ADAMTS-4 digestion of ␣ 2 M by Western blot demonstrated that ADAMTS-4 generated a ϳ98-kDa YESDVM 690 -containing fragment that increased in band intensity over time (Fig. 6B). Inclusion of inhibitors confirmed that generation of the YESDVM 690 neoepitope was the result of aggrecanase activity. Specificity of the YESDVM 690 polyclonal antibody was demonstrated by the ability of the immunizing (YESDVM) peptide to block and the inability of the spanning (YESDVMGRGHAR) peptide to block antibody binding to the 98-kDa YESDVM 690 fragment (Fig. 6B).
To demonstrate that cleavage of ␣ 2 M within the bait region at Met 690 is not generated by other metalloproteinases, representative matrix metalloproteinases including MMP-3, MMP-13, and MMP-25 were analyzed for cleavage of ␣ 2 M, and the products were analyzed by Western blot for all macroglobulin fragments generated using an anti-␣ 2 M polyclonal antibody and for macroglobulin fragments generated by specific cleavage at the Met 690 /Gly 691 bond using the YESDVM 690 and 691 GRGHAR neoepitope antibodies. ADAMTS-4, MMP-3, and MMP-13 cleaved intact ␣ 2 M, generating macroglobulin fragments with a molecular weight of ϳ98 kDa. MMP-25 under these conditions did not cleave ␣ 2 M. Unlike ADAMTS-4, neither MMP-3 nor MMP-13 generated macroglobulin fragments containing the 691 GRGHAR and YESDVM 690 neoepitopes, sug-gesting that these enzymes do not cleave the bait region of ␣ 2 M at Met 690 (data not shown).
The 691 GRGH Neoepitope Is Not Detected in OA Synovial Fluids-Fifteen synovial fluid samples from patients with radiographic and symptomatic knee OA were analyzed for the presence of ␣ 2 M using the polyclonal antibody against ␣ 2 M as well as ADAMTS-4/-5-generated fragments with the N terminus 691 GRGHAR, using the neoepitope antibody. OA synovial fluids contained mainly intact ␣ 2 M migrating at a molecular mass of 180 kDa (Fig. 7). A small amount of cleaved ␣ 2 M migrating at 98 kDa and a few very high and low molecular mass fragments were detected. The 691 GRGH neoepitope could not be detected in any of the fluid samples. To exclude that this was due to the inability of ADAMTS-4 to cleave ␣ 2 M in synovial fluid, the SF samples were incubated in the presence or absence of recombinant enzyme for 2 h at 37°C. This resulted in the generation of a 98-kDa fragment and higher molecular weight fragments representing an SDS-stable complex between ␣ 2 M and ADAMTS-4 (Fig. 7, A and B), demonstrating that ADAMTS-4 can cleave and, therefore, be inhibited by endogenous ␣ 2 M in synovial fluid. Both the 98-kDa ␣ 2 M fragment and the ␣ 2 M⅐ADAMTS-4 SDS-stable complexes were de- tected with the 691 GRGHAR neoepitope antibody, confirming cleavage of the ␣ 2 M subunit at the Met 690 /Gly 691 bond (Fig. 7,  A and B). Blocking of these bands by the immunizing peptide confirmed specificity of the 691 GRGHAR neopitope antibody in detecting ␣ 2 M fragments and complexes with the N terminus 691 GRGHAR (Fig. 7C). Because of the high levels of albumin in synovial fluid, both the polyclonal antibody against ␣ 2 M and the 691 GRGH neoepitope antibody appeared to react with albumin. However, further analysis of this nonspecific binding revealed that it was due to the secondary antibody.

DISCUSSION
Both ADAMTS-4 and ADAMTS-5 are inhibited by ␣ 2 M. Inhibition of a proteinase by ␣ 2 M is initiated by proteolysis of the bait region of the macroglobulin molecule, which in turn triggers large conformational changes that entrap the proteinase. In the current studies we demonstrate that ADAMTS-4 and ADAMTS-5 cleave ␣ 2 M primarily between amino acids Met 690 and Gly 691 . It is, therefore, concluded that in addition to serving as a potential endogenous inhibitor of ADAMTS-4 and ADAMTS-5, ␣ 2 M represents a substrate for both enzymes. The alignment of the aggrecanase cleavage sites known to date (Table I) indicates that glutamic acid at the P1 site and a small side chain at P1Ј are favored residues. Surprisingly, AD-AMTS-4 and ADAMTS-5 did not cleave (at detectable levels) at any of the five glutamic acid residues (Glu 673 , Glu 678 , Glu 686 , Glu 701 , and Glu 702 ) found within the bait region but, rather, cleaved preferentially at Met 690 . The susceptibility of the Met 690 /Gly 691 bond in ␣ 2 M cannot be readily explained, but the sequence around the aggrecanase cleavage site in ␣ 2 M is similar to those found in the core protein of proteoglycans at P1Ј, P2Ј, and P3Ј (GRG) ( Table I). These data, along with recently published findings (42,43) suggest that the aggrecanases are not exclusively glutamyl endopeptidases and that the presentation of the scissile bond along with the PЈ amino acids may be more important than recognition of amino acids that have been speculated to be upstream from the scissile bond in aggrecan (44). Interestingly, neither ADAMTS-4 nor ADAMTS-5 cleaved any of the ␣ 2 M peptides spanning the aggrecanase cleavage site. These findings suggest that a combination of the sequence around the site of cleavage and the local conformation of the bait region is important for proteolytic cleavage at Met 690 / Gly 691 . This also suggests that ADAMTS-4 and ADAMTS-5 may not be limited to cleavage of aggregating proteoglycans but may cleave other proteins, as recently reported by others (43,45). Importantly, the truncated form of ADAMTS-4, which is ineffective in cleaving native aggrecan, readily cleaves ␣ 2 M at the Met 690 /Gly 691 bond. This raises the possibility that processing of aggrecanase proteins to remove the C-terminal disintegrin-like and thrombospondin motifs may result in a change in substrate specificity of these proteinases. ␣ 2 M inhibits both ADAMTS-4 and ADAMTS-5 with secondorder rate constants on the order of 10 6 and 10 5 M Ϫ1 s Ϫ1 , respectively. These rates are intermediate between those reported for ␣ 2 M inhibition of human elastase (4.1 ϫ 10 7 M Ϫ1 s Ϫ1 ) and cathepsin G (3.7 ϫ 10 6 M Ϫ1 s Ϫ1 ), and those for kallikrein (4.8 ϫ 10 4 M Ϫ1 s Ϫ1 ), factor Xa (4.0 ϫ 10 4 M Ϫ1 s Ϫ1 ), and thrombin (4.9 ϫ 10 2 M Ϫ1 s Ϫ1 ) (37,46). ␣ 2 M is found at very high concentrations in the synovial fluid higher than 2 mg/ml, suggesting that the half-lives of free ADAMTS-4 and ADAMTS-5 under these conditions would be on the order of 0.1 and 1 s, respectively. It is interesting to speculate that after cleavage of aggrecan, ADAMTS-4 and ADAMTS-5 bound to aggrecan fragments diffuse into the synovial fluid where they are entrapped by ␣ 2 M. The size of ␣ 2 M (725,000 daltons) precludes the enzymes from diffusing back into the cartilage extracellular matrix. If this is indeed the case it would suggest that ␣ 2 M may play a key role in determining the half-life of ADAMTS-4 and ADAMTS-5 in the articular joint.
In addition to ␣ 2 M it was recently shown that TIMP-3 is a potent endogenous inhibitor of the aggrecanases, with K i (app)  values of 3.30 nM for ADAMTS-4 and 0.66 nM for ADAMTS-5 (19,20). TIMP-3 was found to have a greater affinity for AD-AMTS-4 and ADAMTS-5 than MMP-1, MMP-2, and MMP-3, suggesting that a primary physiological function for TIMP-3 may be the inhibition of the aggrecanases. Thus, both ␣ 2 M and TIMP-3 may be very important regulators of aggrecanase activity in the joint, and any factors causing an imbalance between these inhibitors and ADAMTS-4/ADAMTS-5 may result in the destruction of cartilage aggrecan seen in arthritic diseases.
OA is a slowly progressing disease, and it can take many years before clinical symptoms are observed. Because of the slow progression of the disease, evaluation of disease-modifying osteoarthritic drug efficacy in clinical trials will be very difficult and costly. Therefore, the need to develop molecular biomarkers that can accurately monitor the progression and perhaps the onset of disease is critical. To this end potential biomarkers of OA have been identified that include both nonmechanism and mechanism-based biomarkers (47). However, these biochemical markers have yet to be validated. In the current studies we have identified ␣ 2 M fragments containing the N-terminal 691 GRGHAR and the C-terminal YESDVM 690 generated by cleavage of the ␣ 2 M subunit by ADAMTS-4/-5 as potential markers of aggrecanase activity. Cleavage of ␣ 2 M within the bait region at Met 690 appears to be unique to the aggrecanases, as to date no other known proteinase has been shown to cleave at this site (48). However, the current methodology used failed to reveal these neoepitopes in OA synovial fluids. That we were unable to detect the ␣ 2 M⅐ADAMTS-4/-5 complex in OA synovial fluid suggests the complex is rapidly cleared through the vascular and lymph systems. It is well known that conformational change of ␣ 2 M leads to exposure of recognition sites that bind receptors such as the lipoprotein receptor-related protein on a variety of cells, including macrophages, resulting in clearance (49,50). More sensitive techniques will have to be developed for detection in either the synovial fluid or the blood.