HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 18, 16059-16066, May 3, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/18/16059    most recent M108607200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Westling, J. Articles by Sandy, J. D. Search for Related Content PubMed PubMed Citation Articles by Westling, J. Articles by Sandy, J. D. Social Bookmarking                      What's this?

ADAMTS4 Cleaves at the Aggrecanase Site (Glu³⁷³-Ala³⁷⁴) and Secondarily at the Matrix Metalloproteinase Site (Asn³⁴¹-Phe³⁴²) in the Aggrecan Interglobular Domain^*

Jennifer Westling^§, Amanda J. Fosang^¶, Karena Last^¶, Vivian P. Thompson, Kathy N. Tomkinson, Tracy Hebert, Thomas McDonagh, Lisa A. Collins-Racie, Edward R. LaVallie, Elisabeth A. Morris, and John D. Sandy^**

From the  Center for Skeletal Development and Pediatric Orthopedic Research, Shriners Hospital for Children, and the ^** Department of Pharmacology and Therapeutics, University of South Florida, Tampa, Florida 33612, the ^¶ Department of Paediatrics, Cell and Matrix Biology Research Unit, University of Melbourne, and the Murdoch Children's Research Institute, Royal Children's Hospital, Parkville 3052, Australia, and the  Genetics Institute/Wyeth Research, Cambridge, Massachusetts 02140

Received for publication, September 6, 2001, and in revised form, February 7, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Two major proteolytic cleavages, one at NITEGE³⁷³A³⁷⁴RGSVI and the other at VDIPEN³⁴¹F³⁴²FGVGG, have been shown to occur in vivo within the interglobular domain of aggrecan. The Glu³⁷³-Ala³⁷⁴ site is cleaved in vitro by aggrecanase-1 (ADAMTS4) and aggrecanase-2 (ADAMTS5), whereas the other site, at Asn³⁴¹-Phe³⁴², is efficiently cleaved by matrix metalloproteinases (MMPs) and by cathepsin B at low pH. Accordingly, the presence of the cleavage products globular domain 1 (G1)-NITEGE³⁷³ and G1-VDIPEN³⁴¹ in vivo has been widely interpreted as evidence for the specific involvement of ADAMTS enzymes and MMPs/cathepsin B, respectively, in aggrecan proteolysis in situ. We show here, in digests with native human aggrecan, that purified ADAMTS4 cleaves primarily at the Glu³⁷³-Ala³⁷⁴ site, but also, albeit slowly and secondarily, at the Asn³⁴¹-Phe³⁴² site. Cleavage at the Asn³⁴¹-Phe³⁴² site in these incubations was due to bona fide ADAMTS4 activity (and not a contaminating MMP) because the cleavage was inhibited by TIMP-3 (a potent inhibitor of ADAMTS4), but not by TIMP-1 and TIMP-2, at concentrations that totally blocked MMP-3-mediated cleavage at this site. Digestion of recombinant human G1-G2 (wild-type and cleavage site mutants) confirmed the dual activity of ADAMTS4 and supported the idea that the enzyme cleaves primarily at the Glu³⁷³-Ala³⁷⁴ site and secondarily generates G1-VDIPEN³⁴¹ by removal of the Phe³⁴²-Glu³⁷³ peptide from G1-NITEGE³⁷³. These results show that G1-VDIPEN³⁴¹ is a product of both MMP and ADAMTS4 activities and challenge the widely held assumption that this product represents a specific indicator of MMP- or cathepsin B-mediated aggrecan degradation.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Aggrecan is the major cartilage hyalectan (1), which, together with the collagen network, provides this tissue with its unique mechanical properties of compressibility and stiffness (2-4). Extraction of aggrecan in its native form (5) and subsequent structural analysis (6) have revealed that the molecular organization of aggrecan is perfectly suited to its central functional role in articular cartilage. The N-terminal region of aggrecan is composed of two globular domains (G1¹ and G2) separated by the interglobular domain (IGD). G1 interacts with hyaluronan and link protein, thereby keeping the aggrecan molecule anchored within the cartilage tissue. Further interactions with other matrix components such as tenascin-R and fibulin-1 and fibulin-2 (7, 8) may occur through a third globular domain (G3) at the extreme C terminus of the core protein. The extended core protein between G2 and G3 is composed of a short keratan sulfate-rich region followed by a longer chondroitin sulfate-substituted domain. The charge repulsion and hydration of the long negatively charged glycosaminoglycan (GAG) chains are thought to maintain the C-terminal portions of aggrecan in an extended conformation (9). The swelling pressure of the aggrecan-link protein complex with hyaluronan is restrained by the tension in the collagen network; and together, these components form a fiber-reinforced concentrated gel within the cartilage, which transmits forces across the articular joint.

In diseases characterized by cartilage degradation such as rheumatoid arthritis and osteoarthritis, increased aggrecan release from the cartilage occurs early (10, 11) and before the bulk of the collagen network is degraded (12). Proteolytic cleavage of aggrecan within the IGD separates the GAG-rich region from the hyaluronan-anchored G1 domain, resulting in GAG release from the cartilage matrix to the synovial fluid. Biomechanical tests on cartilage discs have shown that proteolysis within the IGD of aggrecan, and not cleavages near the C terminus, is primarily responsible for the loss of compressive resistance that accompanies interleukin-1-mediated degradation of the tissue (13). Identification of the proteinases responsible for this "destructive" cleavage of aggrecan has therefore been a major focus of experimentation in arthritis-related research.

In this regard, two major cleavage sites that occur in vivo have been identified in the IGD of human aggrecan. One is a matrix metalloproteinase (MMP)-sensitive site at VDIPEN³⁴¹F³⁴²FGVGG, which can be cleaved at neutral pH by any one of a range of MMPs, including MMP-1-3, -7-9, -13, -14, -19, and -20, and also by the combined endopeptidase and carboxypeptidase activity of cathepsin B at low pH (14). Proteolytic fragments of aggrecan resulting from cleavage at this site have been isolated from normal articular tissue (10, 15) and synovial fluids (16-18). The other site lies at NITEGE³⁷³A³⁷⁴RGSVI (19), and cleavage here is efficiently catalyzed by ADAMTS4 and ADAMTS5. Products derived from cleavage at this site represent the bulk of aggrecan fragments found in cartilage and synovial fluids from normal and arthritis patients (10, 20, 21), and we (18) and others (22) have proposed that the ADAMTS proteinases are primarily responsible for the degradative removal of aggrecan from cartilage.

Nonetheless, a review of the recent literature on this subject (Ref. 23 and references therein) shows that in many laboratories, there is a continuing debate over the importance of MMPs and ADAMTS enzymes in cartilage aggrecan degradation. Indeed, it has become widely accepted that immunodetection with specific anti-neoepitope antisera can provide reliable data on the relative role of MMPs or cathepsin B and ADAMTS enzymes in particular situations. Thus, the presence of products with anti-VDIPEN³⁴¹ or anti-F³⁴²FGVGG antibody reactivity is taken as evidence for MMP/cathepsin B-dependent cleavage, whereas fragments that react with anti-NITEGE³⁷³ or anti-A³⁷⁴RGSVI antibody are taken as proof for the involvement of ADAMTS enzymes.

In this study, we present data showing that incubation at neutral pH of mature human aggrecan with recombinant human ADAMTS4 (aggrecanase-1) results in a primary cleavage at the Glu³⁷³-Ala³⁷⁴ site and a secondary cleavage at the Asn³⁴¹-Phe³⁴² site. These results suggest that cleavage of aggrecan in vivo at Asn³⁴¹-Phe³⁴² may be due to MMPs, cathepsin B, or ADAMTS4 activity.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Materials-- MMP-3 and TIMP-1 (tissue inhibitor of metalloproteinase-1) were kindly provided by Merck. TIMP-2 was a generous gift from Dr. Gillian Murphy (University of East Anglia). Normal mature human aggrecan was from Dr. Peter Roughley (Shriners Hospital, Montreal, Canada). TIMP-3 was a generous gift from Dr. Keith Brew (Florida Atlantic University). Endo--galactosidase, keratanase II, and chondroitinase ABC (protease-free) were from Seikagaku America, Inc. 4-12% SDS-polyacrylamide gels were purchased from Invitrogen. Nitrocellulose was from Bio-Rad. Hyperfilm and ECL reagent were from Amersham Biosciences. The sources and characterization of antibodies have been previously described (17, 18, 24, 25) Briefly, anti-NITEGE antibody was from Dr. John Mort (Shriners Hospital, Montreal), and anti-VDIPEN antibody was supplied by Dr. John Mort or by one of us (A. J.F.). Anti-G1 antibody was a rabbit polyclonal antibody raised against bovine aggrecan G1, and AF-28 (anti-FFGVGG) was an anti-peptide monoclonal antibody. The purity of recombinant ADAMTS4 was established with a rabbit anti-human ADAMTS4 polyclonal antibody made against eight distinct synthetic peptides representing all domains of the protein. Poros HQ and Poros HS were obtained from Applied Biosystems (Foster City, CA). All other chemicals used were purchased from Sigma.

Cloning and Expression of Human ADAMTS4-- Human ADAMTS4 was isolated using a PCR strategy. Two sets of oligonucleotide primers were designed to amplify overlapping portions of the 5'- and 3'-halves of the gene. Of the seven human multiple-tissue cDNA libraries that were used as PCR templates, only the uterus cDNA library resulted in PCR products of the appropriate size (5'-amplimer of 1293 bp and 3'-amplimer of 1420 bp). PCR-amplified fragments were digested with EcoRI and BamHI (5'-product) or BamHI and NotI (3'-product), ligated into EcoRI- and NotI-digested COS expression vector pED6-dpc2, and transformed into ElectroMAX DH10B cells (Invitrogen). Cloned PCR fragments of ADAMTS4 were sequenced and found to have three silent changes as compared with the published sequence for ADAMTS4 (22). These changes were C to T at bp 60, A to G at bp 1724, and A to G at bp 2351 (numbering starts at position 1 for A of the ATG start codon for ADAMTS4). The 5'-primer set was 5'-AAATGGGCGAATTCCCACCATGTCCCAGACAGGCTCGCATCC-3' (this primer incorporated an 8-bp tail (AAATGGGC), an EcoRI site (GAATTC), and an optimized Kozak sequence (CCACC) upstream of the ATG start codon) and 5'-TAAGAGACAGTGCCCATAGCCATTGT-3'. The 3'-primer set was 5'-CTCCAAGCCATGCATCAGTTTGAATG-3' and 5'-GACTGACTGCGGCCGCATAGTGAGGTTATTTCCTGCCCGCC-3' (this primer incorporated an 8-bp tail (GACTGACT) and a NotI site (GCGGCCGC) downstream of the TAA stop codon for ADAMTS4).

The EcoRI-NotI fragment containing the intact ADAMTS4 coding sequence was subcloned into pHTop. This plasmid was derived from pED (26) by removing the majority of the adenovirus major late promoter and inserting six repeats of the tet operator (27). A CHO cell line stably expressing ADAMTS4 was obtained by transfecting pHTop/ ADAMTS4 into CHO/A2 cells and selecting clones in 0.05 µM methotrexate. The CHO/A2 cell line was derived from CHO DUKX B11 cells (28) by stably integrating a transcriptional activator, a fusion between the tet repressor and the herpesvirus VP16 transcription activation domain (27).

Purification of Human ADAMTS4-- The CHO cell-conditioned medium was harvested and diluted 3-fold with buffer A (20 mM Tris (pH 7.2), 5 mM CaCl₂, and 10 µM ZnCl₂) and applied to a 50 µ Poros HQ column. The column was washed with buffer B (20 mM Tris (pH 7.2), 50 mM NaCl, 5 mM CaCl₂, and 10 µM ZnCl₂), and the protein was eluted with a linear gradient of buffer B containing 50 mM to 1.0 M NaCl. The ADAMTS4-containing fraction was further purified by application to a 50 µ Poros HS column after a 10-fold dilution with buffer C (20 mM Tris (pH 6.8), 50 mM NaCl, 5 mM CaCl₂, and 10 µM ZnCl₂), and the column was washed with 10 column volumes. Protein was eluted from the column with a linear gradient of buffer C containing 50 mM to 1.0 M NaCl, and the calculated extinction coefficient at 280 nm was used for protein concentration determination as outlined by Gill and von Hippel (29).

Human Aggrecan Digests and Western Analysis-- To analyze the sites cleaved within human aggrecan, ADAMTS4 or MMP-3 at varying concentrations was incubated with normal mature human aggrecan at 37 °C in digestion buffer (20 mM Tris (pH 7.5), 100 mM NaCl, and 10 mM CaCl₂). Digests were then deglycosylated in 50 mM Tris (pH 7.6), 50 mM sodium acetate, and 10 mM EDTA with chondroitinase ABC (25 milliunits/100 µg of GAG, protease-free) for 1.5 h at 37 °C, followed by endo--galactosidase (0.5 milliunits/100 µg of GAG) and keratanase II (0.5 milliunits/100 µg of GAG) for 2 h at 37 °C. The samples were dried, and 5 µg of digested aggrecan (based on dry weight) was loaded per lane on 4-12% SDS-polyacrylamide gels and transferred to nitrocellulose for Western analysis with chemiluminescent detection as described (18).

Inhibitor Studies-- TIMP-1, TIMP-2, EDTA, and an inhibitor mixture (1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 5 mM N-ethylmaleimide, 1 µg/ml pepstatin A, and 250 nM TIMP-1) were incubated with the enzymes for 15 min at room temperature before the addition of human aggrecan substrate (200 nM, calculated based on a molecular mass of 1000 kDa) and further incubation for 16 h at 37 °C in digestion buffer. Specifically, ADAMTS4 (54 nM) or MMP-3 (44 nM) was incubated with TIMP-1 and TIMP-2 at 125, 250, or 750 nM. For incubations containing EDTA (10 mM) or the inhibitor mixture, the inhibitor was added to ADAMTS4 (109 nM). TIMP-3 (125 nM) was added to ADAMTS4 (100 nM, preincubated for 1 h at 37 °C), and the mixture was incubated for a further 2 h before the addition of the aggrecan substrate (200 nM). All digests were deglycosylated as described above and analyzed by Western analysis at 5 µg of digested aggrecan (based on dry weight) loaded per lane.

Digests of Recombinant G1-G2-- Recombinant G1 (rG1)-G2 substrates (wild-type, an aggrecanase site mutant (A³⁷⁴RGSV to N³⁷⁴VYSV), or an MMP site mutant (deletion of residues EN³⁴¹F³⁴²F)) were expressed and purified as described previously (30, 31). For digests, ADAMTS4 (381 nM) was incubated with the rG1-G2 substrate (1.4 µM) in digestion buffer at 37 °C for 16 h. The reactions were stopped by the addition of 20 mM EDTA. The samples were then dried, run on 4-20% SDS-polyacrylamide gels, and transferred to nitrocellulose for Western analysis.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

ADAMTS4 Generates Both G1-NITEGE and G1-VDIPEN-- Recombinant human ADAMTS4 was expressed and purified by gradient elution from sequential anion-exchange (Poros HQ) and cation-exchange (Poros HS) chromatography supports (see "Experimental Procedures" for details). The major silver-stained protein isolated by this protocol migrated at ~70 kDa and corresponded to the major anti-ADAMTS4 antibody-reactive species present (Fig. 1). Samples from a time course digestion of human aggrecan with this purified ADAMTS4 were taken for Western analysis with an antibody to the N-terminal G1 domain (Fig. 2, lanes 1-5). This showed, as expected, the gradual appearance of a 65-kDa product with the electrophoretic properties of G1-NITEGE³⁷³. In addition, an anti-VDIPEN antibody probe of this blot (Fig. 2, lanes 1-5, lower panel) revealed the gradual appearance of a minor product at ~60 kDa consistent with the migration behavior of G1-VDIPEN³⁴¹. It was then shown that in 24-h digests at a higher enzyme concentration, the aggrecan substrate was totally digested and that these two products were generated in approximately equal amounts (Fig. 2, lane 6). Furthermore, when this 24-h sample was probed with anti-NITEGE (Fig. 2, lane 7) or anti-VDIPEN (lane 8) antibody, the results clearly confirmed that under these digestion conditions, ADAMTS4 generated both the expected G1-NITEGE product and also a significant amount of the G1-VDIPEN product. In these samples, the anti-VDIPEN antiserum also reacted weakly with a species that comigrated with the G1-NITEGE fragment (Fig. 2, lane 8), although the exact nature of this species is unknown at present.

View larger version (42K): [in this window] [in a new window]   Fig. 1.   Characterization of purified recombinant ADAMTS4. Shown are the results from silver staining (200 ng of protein; lane 1) and Western blotting (50 ng of protein; lane 2) with anti-human ADAMTS4 antibody. Protein was separated on 10% NuPage SDS-polyacrylamide gels (Invitrogen). Migration positions and sizes of molecular mass standards are shown on the left.

View larger version (61K): [in this window] [in a new window]   Fig. 2.   Time course analysis of human aggrecan digested with ADAMTS4. Human aggrecan (200 nM) was incubated without ADAMTS4 (lane 1) or with ADAMTS4 (50 (lanes 2-5) and 109 (lanes 6-8) nM) for the time periods indicated. The digests were treated with chondroitinase ABC, keratanase II, and endo--galactosidase; run on 4-12% SDS-polyacrylamide gels (5 µg of aggrecan/lane); and then transferred to nitrocellulose for Western analysis. Anti-G1 (1:3000; lanes 1-6), anti-NITEGE (1:3000; lane 7), and anti-VDIPEN (1:3000; lane 8 and lanes 1-5 in the lower panel) antisera were used.

To further study the production of G1-VDIPEN by ADAMTS4, short-term (7-h) digests at a range of enzyme concentrations were probed with the general anti-G1 antiserum (Fig. 3). Aggrecan incubated with the lowest concentrations of ADAMTS4 (4-8 nM) showed some formation of G1-NITEGE, but no G1-VDIPEN formation was evident. As the enzyme concentration was increased (20-100 nM), there was clearly an increase in the loss of full-length aggrecan (Fig. 3, lane 8, black arrow) with a corresponding increase in both the G1-NITEGE and G1-VDIPEN products. At 100 nM enzyme for 7 h, the amount of G1-VDIPEN formed was approaching that formed by 109 nM enzyme in 24 h (Fig. 2, lane 6). Additionally, a minor product at ~130 kDa increased over time (Fig. 3, lane 8, white arrow); and because this species was anti-NITEGE antibody-reactive (data not shown), it appears to represent a non-reducible dimer of the G1-NITEGE product. At each enzyme concentration tested, there was always a greater yield of G1-NITEGE than G1-VDIPEN. However, at ~100 nM ADAMTS4, the amount of G1-NITEGE present appeared to decrease between 7 and 24 h of incubation (compare Fig. 3, lane 8, with Fig. 2, lane 6), whereas there was a concomitant increase in G1-VDIPEN. These results, together with Western analysis of these samples with the individual anti-neoepitope antisera (data not shown), suggested that G1-VDIPEN was formed by ADAMTS4-dependent cleavage of the initial product (G1- NITEGE) and that it was not formed directly from intact aggrecan in these incubations.

View larger version (47K): [in this window] [in a new window]   Fig. 3.   Effects of ADAMTS4 enzyme concentration on G1-VDIPEN production. Human aggrecan (200 nM) was incubated with ADAMTS4 at the concentrations shown for 7 h. Lane 1 shows substrate incubated without the addition of ADAMTS4. All digests were treated with chondroitinase ABC, keratanase II, and endo--galactosidase; run on 4-12% SDS-polyacrylamide gels (5 µg of aggrecan/lane); and then transferred to nitrocellulose for Western analysis. Anti-G1 antiserum (1:3000) was used.

The G1-VDIPEN Product Is Generated by ADAMTS4 and Not by a Contaminating Proteinase-- The ADAMTS4-dependent formation of G1-VDIPEN in a 16-h incubation with 109 nM enzyme was completely blocked by inclusion of 10 mM EDTA, but was not inhibited by the presence of a mixture of TIMP-1, 4-(2-aminoethyl)benzenesulfonyl fluoride, N-ethylmaleimide, and pepstatin (Fig. 4). These results suggest that the proteinase responsible is a TIMP-1-insensitive metalloenzyme (presumably ADAMTS4) and not cathepsin B or another contaminating enzyme from the serine, cysteine, or aspartic proteinase family.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Inhibitor studies on the ADAMTS4-dependent formation of G1-VDIPEN. ADAMTS4 (109 nM) was incubated with or without 10 mM EDTA or an inhibitor mixture containing 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride, 5 mM N-ethylmaleimide, 1 µg/ml pepstatin A, and 250 nM TIMP-1 in digestion buffer for 15 min at 25 °C, after which human aggrecan (200 nM) was added. The samples were then incubated at 37 °C for 16 h. The digests were run on 4-12% SDS-polyacrylamide gels (5 µg of aggrecan/lane) and then transferred to nitrocellulose for Western analysis. Anti-G1 antiserum (1:3000) was used. Lane 1, human aggrecan incubated without ADAMTS4 or inhibitors; lane 2, human aggrecan incubated with ADAMTS4 alone; lane 3, human aggrecan incubated with ADAMTS4 and EDTA; lane 4, human aggrecan incubated with ADAMTS4 and the inhibitor mixture.

To confirm this activity of ADAMTS4, we next tested the relative inhibitory effects of TIMP-1-3. TIMP-1 is a known stoichiometric inhibitor of MMP-1-3 and MMP-7-13, but does not appear to inhibit ADAMTS4-mediated degradation of aggrecan at concentrations up to 1 µM (32). We therefore used TIMP-1 to examine the possibility that the cleavage of aggrecan at the Asn³⁴¹-Phe³⁴² site was due to a contaminating MMP in the ADAMTS4 preparation. In control experiments, TIMP-1 at 125 and 250 nM completely blocked formation of G1-VDIPEN generated by 44 nM MMP-3 (Fig. 5, lanes 7 and 8). When human aggrecan was digested for 15 h with 54 nM ADAMTS4 (Fig. 5, lane 2), both the G1-NITEGE and G1-VDIPEN products were formed, as expected. However, preincubation of ADAMTS4 with 125, 250, or 750 nM TIMP-1, before the addition of the aggrecan substrate, failed to inhibit the subsequent formation of either G1-NITEGE or G1-VDIPEN (Fig. 5, lanes 3-5). The production of G1-VDIPEN in these incubations was therefore clearly not due to the activity of a TIMP-1-inhibitable MMP. Interestingly, TIMP-1 addition appeared to increase the activity of ADAMTS4 because here was an obvious increase in the amount of G1-NITEGE formed at 250 and 750 nM TIMP-1. Similarly, there was an increase in G1-VDIPEN, further suggesting that these two products are generated by the same ADAMTS4 activity.

View larger version (61K): [in this window] [in a new window]   Fig. 5.   Effect of TIMP-1 on the G1-VDIPEN-generating activity of ADAMTS4 and MMP-3. ADAMTS4 (54 nM) or MMP-3 (44 nM) was incubated with varying concentrations of TIMP-1 (0, 125, 250, or 750 nM) in digestion buffer for 15 min at 25 °C. Next, human aggrecan (200 nM) was added, and the samples were incubated at 37 °C for 15 h. All digests were treated with chondroitinase ABC, keratanase II, and endo--galactosidase; run on 4-12% SDS-polyacrylamide gels (5 µg of aggrecan/lane); and then transferred to nitrocellulose for Western analysis. Anti-G1 antiserum (1:3000) was used. Lane 1, aggrecan incubated without enzyme; lane 2, digest of aggrecan with ADAMTS4 incubated without TIMP-1; lanes 3-5, digests of aggrecan with ADAMTS4 incubated with TIMP-1 at 125, 250, and 750 nM, respectively; lane 6, digest of aggrecan with MMP-3 incubated without TIMP-1; lanes 7 and 8, digests of aggrecan with MMP-3 incubated with TIMP-1 at 125 and 250 nM, respectively.

The formation of G1-VDIPEN was next investigated with TIMP-2, which is a known stoichiometric inhibitor of MMP-1, -3, -7, -8, -10, -11, -13, and -14 and also does not inhibit ADAMTS4-mediated degradation of aggrecan at concentrations up to 1 µM (32). For this purpose, we repeated the experiments shown in Fig. 5, but with preincubation of ADAMTS4 with TIMP-2. The results (Fig. 6) were essentially identical to those obtained with TIMP-1, showing that the formation of G1-VDIPEN was not due to a TIMP-2-inhibitable MMP. This result provided further evidence that the production of G1-VDIPEN was mediated by ADAMTS4 and not by a TIMP-1- or TIMP-2-inhibitable MMP.

View larger version (71K): [in this window] [in a new window]   Fig. 6.   Effect of TIMP-2 on the G1-VDIPEN-generating activity of ADAMTS4 and MMP-3. ADAMTS4 (54 nM) or MMP-3 (44 nM) was incubated with varying concentrations of TIMP-2 (0, 125, 250, or 750 nM) in digestion buffer for 15 min at 25 °C. Next, human aggrecan (200 nM) was added, and the sample was incubated at 37 °C for 15 h. All digests were treated with chondroitinase ABC, keratanase II, and endo--galactosidase; run on 4-12% SDS-polyacrylamide gels (5 µg of aggrecan/lane); and then transferred to nitrocellulose for Western analysis. Anti-G1 antiserum (1:3000) was used. Lane 1, aggrecan incubated without enzyme; lane 2, digest of aggrecan with ADAMTS4 incubated without TIMP-2; lanes 3-5, digests of aggrecan with ADAMTS4 incubated with TIMP-2 at 125, 250, and 750 nM, respectively; lane 6, digest of aggrecan with MMP-3 incubated without TIMP-2; lanes 7 and 8, digests of aggrecan with MMP-3 incubated with TIMP-2 at 125 and 250 nM, respectively.

The formation of G1-VDIPEN was next investigated with TIMP-3, which is a known potent inhibitor of ADAMTS4 (IC₅₀ = 3-8 nM), but is much less effective against MMP-3 (K_i = 67 nM) (32, 33). As expected, when either ADAMTS4 or MMP-3 was incubated with aggrecan, the G1-VDIPEN product was generated, as indicated by Western analysis with anti-VDIPEN antibody (Fig. 7, lane 1 versus lanes 2 and 4). When TIMP-3 was added at 125 nM, G1-VDIPEN formation was blocked in the ADAMTS4 incubations (Fig. 7, lane 3), but was essentially unaffected in the MMP-3 digests (lane 5). Taken together, these results strongly support the conclusion that ADAMTS4 is the G1-VDIPEN-generating activity in these incubations.

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Effect of TIMP-3 on the G1-VDIPEN-generating activity of ADAMTS4 and MMP-3. ADAMTS4 (100 nM) that had been preincubated for 1 h at 37 °C or MMP-3 (100 nM) was incubated with or without TIMP-3 (125 nM) in digestion buffer for 2 h at 37 °C. Next, human aggrecan (200 nM) was added, and the sample was incubated at 37 °C for 15 h. All digests were treated with chondroitinase ABC, keratanase II, and endo--galactosidase; run on 4-12% SDS-polyacrylamide gels (5 µg of aggrecan/lane); and then transferred to nitrocellulose for Western analysis. Anti-VDIPEN antiserum (1:3000) was used. Lane 1, aggrecan incubated without enzyme; lane 2 and 3, digests of aggrecan with ADAMTS4 incubated without and with TIMP-3, respectively; lane 4 and 5, digests of aggrecan with MMP-3 incubated without and with TIMP-3, respectively.

ADAMTS4 Cleaves at Both Sites in Wild-type rG1-G2 and at Each Site in Mutant rG1-G2 Substrates-- To further examine the dual specificity of ADAMTS4, we investigated the capacity of the enzyme to cleave at both Glu³⁷³-Ala³⁷⁴ and Asn³⁴¹-Phe³⁴² in rG1-G2 substrates. First, the wild-type substrate containing both cleavage sites was digested with ADAMTS4, and the products were examined by Western analysis (Fig. 8A) with anti-G1, anti-NITEGE, anti-VDIPEN, and anti-FFGVGG (AF-28) antibodies. The rG1-G2 substrate incubated alone was detected with anti-G1 antibody as a major band at ~100 kDa and a minor unidentified band at ~66 kDa (Fig. 8A, lane 1). After digestion, the intact species was eliminated, and two new G1-reactive bands were detected at ~60 and 55 kDa (Fig. 8A, lane 2). The 60-kDa product was also detected with anti-NITEGE antibody (Fig. 8A, lane 3), and the 55-kDa product with anti-VDIPEN antibody (lane 4); however, no species reactive with anti-FFGVGG antibody were detected on these blots (lane 5). These results show that ADAMTS4 can cleave at both Glu³⁷³-Ala³⁷⁴ and Asn³⁴¹-Phe³⁴² in this recombinant substrate, much as seen in digests with native human aggrecan. Because no high molecular mass FFGVGG-reactive fragments were seen, it is likely that G1-NITEGE was formed first, and then G1-VDIPEN was generated by removal of the Phe³⁴²-Glu³⁷³ peptide from G1-NITEGE373, much as was observed with ADAMTS4 digests of native aggrecan substrate. The low molecular mass FFGVGG-reactive peptide Phe³⁴²-Glu³⁷³, expected as the third product at ~3 kDa (31) in this digest, was too small to be detected under the electrophoretic conditions used.

View larger version (69K): [in this window] [in a new window]   Fig. 8.   ADAMTS4 cleaves rG1-G2 primarily at NITEGE373A374RGSVI and secondarily at VDIPEN341F342FGVGG. A, wild-type rG1-G2 (1.4 µM) was incubated without (lane 1) or with (lanes 2-5) ADAMTS4 (381 nM) at 37 °C in digestion buffer for 16 h. The digests were run on 4-12% SDS-polyacrylamide gels (1 µg of rG1-G2/lane) and then transferred to nitrocellulose for Western analysis. Anti-G1 (1:3000; lanes 1 and 2), anti-NITEGE (1:3000; lane 3), anti-VDIPEN (1:100; lane 4) (30), and anti-FFGVGG (1:100; lane 5) (17) antisera were used. B, the same procedure as described for A was used, except that the rG1-G2 EN341F342F deletion mutant was the substrate. C, the same procedure as described for A was used, except that the rG1-G2 A374RGSV-to-N374VYSV mutant was the substrate.

Following the observation that ADAMTS4 can cleave at both sites in the wild-type substrate, we next investigated whether the enzyme can cleave at each site in a substrate without prior cleavage at the alternative site. Independent cleavage at each of these sites has been demonstrated for atrolysin, MMP-8, and MMP-14 (34-36). For this purpose, an rG1-G2 mutant in which the four amino acids spanning the Asn³⁴¹-Phe³⁴² cleavage site (EN³⁴¹F³⁴²F) had been deleted was incubated with ADAMTS4 and analyzed for cleavage products (Fig. 8B). In this case, the majority (although not all) of the 100-kDa substrate (Fig. 8B, lane 1) was digested by ADAMTS4 and converted to a single major G1-reactive product at 60 kDa (lane 2), which was identified as G1-NITEGE by reactivity with anti-NITEGE antibody (lane 3). No products reactive with anti-VDIPEN or anti-FFGVGG antibody were detected (Fig. 8B, lanes 4 and 5), consistent with the absence of the Asn³⁴¹-Phe³⁴² cleavage sequence.

Finally, another rG1-G2 mutant in which the aggrecanase cleavage site sequence NITEGEARGSVIL had been mutated to NITEGENVYSVIL was incubated with ADAMTS4 (Fig. 8C). In this case, the 100-kDa substrate (Fig. 8C, lane 1) was largely, but not completely cleaved to generate a new G1-bearing species at 55 kDa (lane 2). The 55-kDa product reacted strongly with anti-VDIPEN antibody, confirming its identity as G1-VDIPEN and suggesting that ADAMTS4 cleaves effectively at Asn³⁴¹-Phe³⁴² in this substrate, without cleaving the Glu³⁷³-Ala³⁷⁴ site first. A band at ~60 kDa in lane 2 was also present; however, its identity is unknown. It appears to be the unidentified 60-kDa band also found in the substrate alone (Fig. 8C, lane 1, arrowhead) because it did not react with any of the anti-neoepitope antibodies. The conclusion that ADAMTS4 cleaved at Asn³⁴¹-Phe³⁴² in this substrate was also supported by the generation of a 55-kDa species that reacted with anti-FFGVGG antibody (Fig. 8C, lane 5) and that therefore represents the corresponding F³⁴²FGVGG-G2 product. No products reactive with anti-NITEGE antibody were detected (Fig. 8C, lane 3), showing that the mutated sequence NITEGENVYSV does not represent a substrate for ADAMTS4 cleavage at Glu³⁷³-Ala³⁷⁴.

Thus, although ADAMTS4 appears to cleave primarily at Glu³⁷³-Ala³⁷⁴ and secondarily at Asn³⁴¹-Phe³⁴² in native aggrecan and wild-type rG1-G2, the present data show that with this mutant substrate (Fig. 8C), it is capable of cleaving at Asn³⁴¹-Phe³⁴² without prior cleavage at Glu³⁷³-Ala³⁷⁴. Therefore, G1-NITEGE is not the only possible substrate for ADAMTS4-dependent cleavage at Asn³⁴¹-Phe³⁴², although our detailed studies with human aggrecan (Figs. 2 and 3) suggest that it is very likely to be the substrate in vivo.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

This study provides evidence that ADAMTS4 (aggrecanase-1) has a less restricted cleavage specificity within the IGD of aggrecan than previously proposed (37). Although it has been widely accepted (22, 23, 30, 37-42) that ADAMTS4 (aggrecanase-1) cleaves only at the Glu³⁷³-Ala³⁷⁴ bond within the IGD, the results described here suggest that both of the major cleavages that are known to occur in vivo, at Glu³⁷³-Ala³⁷⁴ and Asn³⁴¹-Phe³⁴², can be catalyzed by ADAMTS4. Previously, cleavage at the Asn³⁴¹-Phe³⁴² site in the aggrecan IGD was considered a specific property of the MMP family, with the exception of a snake venom enzyme (30, 34) and cathepsin B, which cleaves Asn³⁴¹-Phe³⁴² under conditions of low pH (14). Earlier studies had suggested that selected MMPs could cleave at the aggrecanase Glu³⁷³-Ala³⁷⁴ bond (35, 36, 43), but there have been no previous studies to suggest the converse, that mammalian ADAMTS enzymes could cleave at Asn³⁴¹-Phe³⁴². Indeed cleavage at the Asn³⁴¹-Phe³⁴² bond in incubations with partially purified "aggrecanase" samples may have been interpreted as due to contaminating MMPs present in the enzyme preparation. The possibility that cleavage of the MMP-sensitive Asn³⁴¹-Phe³⁴² bond in the experiments described here was due to contamination of the ADAMTS4 preparation with an MMP-like activity was eliminated by the observation that although 10 mM EDTA and TIMP-3 totally blocked the cleavage, neither TIMP-1 nor TIMP-2 was inhibitory, even at 750 nM. Collectively, these results suggest that the distinction between MMP, cathepsin B, and ADAMTS activities based on neoepitope immunoreactivity of the products can no longer be considered definitive (see Table I for summary).

If, as suggested here, cleavage of the Asn³⁴¹-Phe³⁴² bond in native human aggrecan is catalyzed by ADAMTS4, the question arises as to why this cleavage activity has not previously been detected with recombinant preparations of ADAMTS4. The most likely explanation is that the early detailed studies on ADAMTS4 cleavage of bovine aggrecan (37, 44) did not examine this possibility directly. The routine monitor of cleavage activity in those studies was antibody BC-3, which detects the neoepitope at Ala³⁷⁴, but does not detect G1-bearing products such as G1-NITEGE and G1-VDIPES (bovine sequence). In the previous studies (37, 44), the ADAMTS4-dependent production of G1-NITEGE was demonstrated with anti-NITEGE antibody; however, the possibility that G1-VDIPES was also formed was apparently not addressed, probably due to the unavailability of an anti-neoepitope antiserum to the bovine product and the known poor reactivity of anti-VDIPEN antibody with the bovine product (45). Moreover, in those studies, it was concluded that ADAMTS4 does not cleave the Ser³⁴¹-Phe³⁴² site in bovine aggrecan because products with an N terminus of FFGVGG were not detected with antibody AF-28 (anti-FFGVGG). However, if ADAMTS4 cleaved at both sites in the bovine incubations, the short intervening AF-28-reactive peptide would have migrated off the gel under the electrophoretic conditions used. In other studies (46), aggrecan fragmentation was monitored by antibody BC-3 and an antibody to the chondroitin sulfate stubs on chondroitinase-treated aggrecan (monoclonal antibody 2035), neither of which would have detected the possible formation of G1-VDIPES. Most recently, a VDIPES-specific antiserum did not detect this product in an incubation of bovine aggrecan with 10 nM ADAMTS4 for 24 h (45). In this case, the extent of IGD cleavage may have been insufficient to generate enough G1-VDIPES product for detection. Detailed studies on the cleavage of native pig G1-G2 and recombinant human G1-G2 with aggrecanase purified from conditioned medium (30, 31) have also failed to detect this dual cleavage specificity. The most likely explanation for this is that, in general, studies with aggrecanase in this area have been optimized in terms of enzyme activity, sample loadings, and gel exposures to detect the products expected from cleavage at the Glu³⁷³-Ala³⁷⁴ bond.

Differences between the bovine substrates used in previous studies to assess ADAMTS4 activity and the native human aggrecan substrate used here may also explain differences in the observed cleavage specificity. For example, it is possible that the Ser³⁴¹-Phe³⁴² bond in bovine aggrecan is less susceptible to ADAMTS4-dependent cleavage than the equivalent Asn³⁴¹-Phe³⁴² bond in the human substrate. The ability of ADAMTS4 to cleave the Asn³⁴¹-Phe³⁴² bond does not, however, appear to be sensitive to the presence or absence of keratan sulfate substitution because DIPEN³⁴¹ fragments were generated from both native aggrecan and non-glycosylated rG1-G2 substrate. This is in contrast to the snake venom reprolysin atrolysin C, which failed to cleave non-glycosylated rG1-G2 at the Asn³⁴¹-Phe³⁴² site (30).

Substrate specificity studies have suggested that ADAMTS enzymes associate with a region of aggrecan at or near the MMP cleavage site; however, this association was presumed to be more in the nature of a substrate-docking site for the enzyme (31, 35). The docking site hypothesis is consistent with two observations. First, large aggrecan fragments with an F³⁴²FGVGG N terminus lacking the G1-VDIPEN³⁴¹ domain cannot be cleaved by ADAMTS enzymes at the Glu³⁷³-Ala³⁷⁴ bond (42, 47); and second, mutations on the N-terminal (but not C-terminal ) side of the Glu³⁷³-Ala³⁷⁴ cleavage site inhibit ADAMTS activity (48). If a region at or near the MMP site is indeed an enzyme-docking site, then the present data suggest that such docking may lead secondarily to cleavage at the Asn³⁴¹-Phe³⁴² site. Further mutagenesis studies and x-ray crystallographic data are needed to elucidate these complex ADAMTS4/aggrecan interactions.

Although the present data show that ADAMTS4 is capable of generating both G1-NITEGE and G1-VDIPEN and inhibitor studies strongly indicate that ADAMTS (and not MMP-8 or MMP-14) activity is responsible for the formation of G1-NITEGE in situ, it is unclear to what extent the formation of G1-VDIPEN in vivo results from ADAMTS4 or MMP activity. In this regard, analysis of human synovial fluids with antibody AF-28 (anti-FFGVGG) provided evidence for a degradative pathway in which cleavage at Asn³⁴¹-Phe³⁴² occurs independently of cleavage at Glu³⁷³-Ala³⁷⁴ (16). However, due to the qualitative nature of this study, the degree to which this pathway is responsible for the presence of abundant G1-VDIPEN in human articular cartilage is unclear. A comparison of the G1-VDIPEN-generating activity of equivalent concentrations of MMP-3 and ADAMTS4 (Fig. 7) suggests that the rate of product formation with MMP-3 is probably ~10-fold that seen with ADAMTS4. The data in this study therefore suggest the possibility that one pathway to G1-VDIPEN formation in articular cartilage is through ADAMTS-dependent cleavage of the IGD at both the Glu³⁷³-Ala³⁷⁴ and Asn³⁴¹-Phe³⁴² sites. In this regard, it seems likely that in tissue situations in which a high level of MMP activity is present, the formation of G1-VDIPEN will result primarily from MMP-dependent cleavage; on the other hand, in tissue environments in which ADAMTS4 activity is predominant and MMPs are either absent or inhibited, G1-VDIPEN may well be generated by ADAMTS4.

Immunolocalization studies with anti-NITEGE and anti-VDIPEN antibodies have shown a temporal and spatial separation of the immunoreactive products in developing and mature cartilages (39, 40, 49). If ADAMTS activity in situ is responsible for formation of both species, then our data would predict that abundant G1-NITEGE could be formed in matrix regions of limited ADAMTS activity, whereas G1-VDIPEN would appear only after extended periods of aggrecanolysis. Indeed, the early appearance of G1-NITEGE and the late appearance of G1-VDIPEN in the cartilage of two murine arthritis models (40) are entirely consistent with this explanation. Such a pathway involving only ADAMTS activity would also predict that in interleukin-1-mediated degradation of aggrecan in cartilage explants, both G1-NITEGE and G1-VDIPEN should be formed and that their formation should be blocked by an ADAMTS-specific (MMP-sparing) inhibitor. It will be interesting to examine these possibilities in more detail when specific ADAMTS inhibitors become available.

	ACKNOWLEDGEMENT

We thank Dr. Anna Plaas for insightful discussion during the development of this work.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ To whom correspondence should be addressed: Center for Skeletal Development and Pediatric Orthopedic Research, Shriners Hospital for Children, 12502 N. Pine Dr., Tampa, FL 33612. Tel.: 813-972-2250 (ext. 7728); Fax: 813-975-7127; E-mail: jwestling@shctampa.usf.edu.

Published, JBC Papers in Press, February 19, 2002, DOI 10.1074/jbc.M108607200

	ABBREVIATIONS

The abbreviations used are: G1, globular domain 1; rG1, recombinant globular domain 1; IGD, interglobular domain; GAG, glycosaminoglycan; MMP, matrix metalloproteinase; CHO, Chinese hamster ovary.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				J. L. Lauer-Fields, D. Minond, T. Sritharan, M. Kashiwagi, H. Nagase, and G. B. Fields Substrate Conformation Modulates Aggrecanase (ADAMTS-4) Affinity and Sequence Specificity: SUGGESTION OF A COMMON TOPOLOGICAL SPECIFICITY FOR FUNCTIONALLY DIVERSE PROTEASES J. Biol. Chem., January 5, 2007; 282(1): 142 - 150. [Abstract] [Full Text] [PDF]