Fibulin-1 acts as a cofactor for the matrix metalloprotease ADAMTS-1.

ADAMTS-1 is a metalloprotease that has been implicated in the inhibition of angiogenesis and is a mediator of proteolytic cleavage of the hyaluronan binding proteoglycans, aggrecan and versican. In an attempt to further understand the biological function of ADAMTS-1, a yeast two-hybrid screen was performed using the carboxyl-terminal region of ADAMTS-1 as bait. As a result, the extracellular matrix protein fibulin-1 was identified as a potential interacting molecule. Through a series of analyses that included ligand affinity chromatography, co-immunoprecipitation, pulldown assays, and enzyme-linked immunosorbent assay, the ability of these two proteins to interact was substantiated. Additional studies showed that ADAMTS-1 and fibulin-1 colocalized in vivo. Furthermore, fibulin-1 was found to enhance the capacity of ADAMTS-1 to cleave aggrecan, a proteoglycan known to bind to fibulin-1. We confirmed that fibulin-1 was not a proteolytic substrate for ADAMTS-1. Together, these findings indicate that fibulin-1 is a new regulator of ADAMTS-1-mediated proteoglycan proteolysis and thus may play an important role in proteoglycan turnover in tissues where there is overlapping expression.

A large percentage of mice that lack ADAMTS-1 die perinatally (45%), and those that survive display urinary track obstructions and kidney pathologies (13,14). In addition, most ADAMTS-1 null females are infertile because of abnormalities in the ovaries and endometrium (13). Evaluation of secondary follicles in the ovaries of ADAMTS-1 null mice revealed poor cumulus expansion and defective ovulation. These anomalies were attributed to incomplete versican proteolysis (15).
Thus, a concrete understanding of the array of ADAMTS-1 biological functions will likely rely on insights from its substrates and regulation of catalytic function.
The fibulins are a family of secreted glycoproteins that are components of elastic matrix fibers and basement membranes. Currently, there are six members of the fibulin family (16). The prototypic member, fibulin-1, contains the signature structural features of all fibulin family members, a series of epidermal growth factor-like repeats followed by a fibulin-type module at its carboxyl terminus (16).
Using a series of biochemical approaches, we have found that fibulin-1 binds to ADAMTS-1 and enhances its catalytic activities toward aggrecan. These findings, together with those showing that fibulin-1 can bind to aggrecan, led us to speculate that fibulin-1 acts to enhance ADAMTS-1-mediated proteolysis by facilitating a ternary complex formation.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screening-The Matchmaker Two-hybrid System 2 (Clontech Laboratories Inc. Palo Alto, CA) was used following the manufacturer's recommendations. A cDNA fragment encoding amino acids 827-951 of human ADAMTS-1 (GenBank TM accession number AF060152) was amplified by PCR using the primers 5Ј-CACCTACT-TCGTACATATGAAGAAG-3Ј and 5Ј-CTTAAACCAGTCGACTG-CATTCTG-3Ј and inserted into the vector pAS2-1 previously digested with NdeI and SalI restriction enzymes. The resulting construct encoded a GAL4BD-ADAMTS-1 827-951 chimeric protein. A library of prey proteins from human placenta was obtained from Clontech Laboratories Inc. Both bait and prey vectors were simultaneously introduced into Saccharomyces cerevisiae CG-1945 cells and double transformants selected on synthetic minimal medium lacking tryptophan, leucine, and histidine. Colonies were restreaked and tested for ␤-galactosidase activity using a filter assay. DNA from positive colonies were amplified by PCR using Matchmaker 5Ј-and 3Ј-AD LD-Insert Screening Amplimers and the resulting amplicons sequenced.
Pulldown Assay-Fibulin-1 or BSA (5 g each) were individually covalently linked to magnetic M450 tosyl-activated dynabeads (DYNAL, Brown Deer, WI) by overnight incubation in 100 mM phosphate buffer, pH 7.4, at 37°C. The beads were subsequently washed with 200 mM Tris, pH 8.5, with 0.1% BSA and incubated overnight at 4°C with ADAMTS-1 (2 g) in a volume of 400 l. Dynabeads were magnetically collected and washed with phosphate-buffered saline to remove unbound proteins. Bound proteins were released using sample buffer containing ␤-mercaptoethanol and separated on 10% SDS-polyacrylamide gels, transferred to Optitran nitrocellulose membrane (Schleicher and Schuell, Keene, NH), and immunoblotted with guinea pig anti-ADAMTS-1 polyclonal antibody (GPB).
Co-immunoprecipitation Analysis-Fragments of human full-term placenta were homogenized in cold lysis buffer (20 mM Tris, pH 8.0, 150 mM NaCl, 5 mM iodoacetamide, 0.5 mM CaCl 2 , 0.5 mM MgCl 2 , 0.02% NaN 3 , 1 mM PMSF, complete protease inhibitor mixture tablets following the manufacturer's direction (Roche Applied Science), 1% Triton X-100 using a Waring blender. The lysate was incubated with protein G-agarose beads (Roche Applied Science) overnight at 4°C to absorb resin-binding proteins. Aliquots of the lysates (1 ml) were then incubated with antibodies to fibulin-1 (mAb 3A11), ADAMTS-1 (GPB), or mouse IgG (Sigma) for 3 h at 4°C. Antibody complexes were precipitated for 1 h at 4°C using either protein G-agarose (Roche Applied Science) for mouse IgGs or protein A-agarose (Roche Applied Science) for GPB. The agarose-bound complexes were pelleted by centrifugation at 1,000 ϫ g and the pellets washed with 10 mM Tris, pH 8.0, 150 mM NaCl, and 0.5% Nonidet P-40. Bound proteins were released using sample buffer containing ␤-mercaptoethanol and separated on 10% SDSpolyacrylamide gels, transferred to Optitran nitrocellulose membrane (Schleicher and Schuell), and analyzed by immunoblotting.
Conditioned medium from fibulin-1C-expressing HT1080 cells cultured in the absence of serum was collected and precleared with protein G-agarose beads (Roche Applied Science) for 16 h at 4°C. Fibulin-1 monoclonal antibody (mAb 3A11) was added to 1-ml aliquots and incubated for 4 h at 4°C. Immunocomplexes were pelleted with protein G-agarose beads and the pellets washed extensively using 10 mM Tris, pH 8.0, 150 mM NaCl, and 0.5% Nonidet P-40 buffer. These fibulin-1immobilized immunocomplexes were incubated with ADAMTS-1 (87-kDa form) for 2 h at 4°C and unbound protein removed by washing with 10 mM Tris, pH 8.0, 150 mM NaCl, and 0.5% Nonidet P-40 buffer. After several washes, bound proteins were released using sample buffer containing ␤-mercaptoethanol, separated on 10% SDS-polyacrylamide gels, and transferred to Optitran nitrocellulose membrane (Schleicher and Schuell). The presence of ADAMTS-1 and fibulin-1 was evaluated by immunoblotting with GPB and mAb 3A11, respectively.
Analysis of ADAMTS-1, Fibulin-1, and Aggrecan Expression in Embryonic Kidneys-Kidneys harvested from E15.5 embryos were fixed in 2% paraformaldehyde overnight, embedded in paraffin, and sectioned at 5 m. Sections were deparaffinized and incubated with blocking buffer (BB, 5% goat serum and 2.5% BSA in PBS) for 2 h at room temperature. Sections were incubated with guinea pig anti-ADAMTS-1 (GP12) at 1:200 dilution in 0.5ϫ BB, rabbit anti-fibulin-1 (Rb1323) (24) at 10 g/ml in 0.5ϫ BB, rabbit anti-aggrecan G1 domain or TASELE neo-epitope at 1:200 dilution in 0.5ϫ BB for 2 h at room temperature. Sections were then incubated with anti-guinea pig IgG conjugated to biotin (Vector Labs, Burlingame, CA) at 1:250 dilution in 0.5ϫ BB for 1.5 h, followed by incubation with streptavidin conjugated to CY3 (Sigma) at 1:100 dilution in 0.5ϫ BB and anti-rabbit IgG conjugated to fluorescein isothiocyanate (Jackson ImmunoResearch, West Grove, PA) at 1:150 dilution in 0.5ϫ BB for 1 h at room temperature. Fluorescence immunohistochemistry was analyzed using a Bio-Rad MRC 1024ES laser confocal microscope.
Anti-ADAMTS-1 GP12 antibody was raised in guinea pig by BIO-SOURCE (Camarillo, CA) using purified ADAMTS-1 protein as immunizing antigen. Antibody specificity was tested against conditioned medium and cell lysates from T47D cells transfected with ADAMTS-1 or empty vector construct.
Proteoglycan Digestion-Rat aggrecan was a generous gift from Dr. John Sandy (Shriners Hospital for Children, Tampa, FL). Aggrecan is further purified to obtain the intact form and removed processed forms; however, 250-kDa fragments are frequent contaminants, and this degree of contamination varies from 10 to 50%. Thus, controls for each purification lot are always included. Aggrecan (7.5 g) either alone or in the presence of ADAMTS-1 (0.5-1 g) was incubated in 60 l of 50 mM Tris, pH 7.4, 10 mM CaCl 2 , 80 mM NaCl for 2 h at 37°C containing fibulin-1, fibronectin, vitronectin, fibrinogen, or BSA (1 g each). The reaction mixtures were deglycosylated in 50 mM sodium acetate, 50 mM Tris, 10 mM EDTA, pH 7.6 with 1.8 -15 milliunits of chondroitinase ABC (Sigma) at 37°C for 1 h prior to separation on 10% SDS-PAGE electrophoresis in reducing conditions with ␤-mercaptoethanol and immunoblotting. Rat aggrecan was also digested with ADAMTS-1 in the presence of varying amounts of fibulin-1 (0.35, 0.7, or 1.5 g) or laminin-1 (0.7, 1.5, or 3.0 g) for 2 h at 37°C (60-l reaction volume) followed by chondroitinase digestion as mentioned above. Digests were resolved on 10% SDS-PAGE or 4 -12% gradient polyacrylamide gels (Invitrogen). The extent of aggrecan cleavage was determined by immunoblotting using rabbit anti-G1 aggrecan antibody (a generous gift from Dr. John Sandy). ADAMTS-1 was detected in immunoblotting with GPB, laminin-1 was detected with anti-laminin (Sigma), and fibulin-1 was detected with mAb 3A11 antibodies.
RNA Isolation and Northern Analysis-Total RNA was isolated from several organs at E16.5, E17.5, and E18.5. Purification of mRNA was performed as previously described (25). Fractionation of mRNA (2 g/lane) was performed on agarose/formaldehyde gels, and subsequent transfer was done onto nylon membranes. Probes included a 951-bp cDNA fragment for ADAMTS-1 and a 360-bp fragment encoding the 3Ј-end of fibulin-1.
Generation of ADAMTS-1 Null Mice-Bacterial artificial chromosome (BAC) clone number 413009 carrying genomic fragments of adamts1 from mouse strain 129/SvJ was purchased from Genome Systems, Inc. (St. Louis, MI). An 8.4-kb EcoRI fragment extending from the 5Ј-untranslated region (UTR) to exon 2 and a 6.6-kb HindIII fragment spanning from intron 7 to the 3Ј-UTR were obtained after digestion of BAC clone number 413009 and were subsequently subcloned into pGEM11ZF(ϩ) and pGEM7ZF(ϩ) (Promega, Madison, WI), respectively. The targeting construct was generated by subcloning a 3.2-kb HindIII-XbaI fragment spanning the 5Ј-UTR to part of intron 1, a 3.6-kb BamHI-BamHI fragment encompassing exon 9 and the 3Ј-UTR and a 2.0-kb flox-PGK-Neo (XbaI) into pGEM11ZF(ϩ). The flox-PGK-Neo construct was a generous gift from Dr. K. Lyons. The resulting targeting construct was linearized with XhoI and HindIII to remove the pGEM11ZF(ϩ) plasmid prior to being introduced into 129/SvJ-derived embryonic stem cells by electroporation at the UCLA Molecular Genetics and Technology Center. Embryonic stem cells were then selected in growth medium containing G418 (300 g/ml), and homologous recombinants were identified by Southern blot analysis. Chimeric males were generated from two independent targeted embryonic stem cell clones injected into C57BL/6 blastocysts (UC Irvine Transgenic Mouse Facility, Irvine, CA). Male chimeras were crossbred with BALB/c females, and germ line transmission was obtained from both independent embryonic stem clones.
PCR Genotyping of ADAMTS-1 Mice-Genomic DNA was isolated from tails of embryos using a standard phenol:chloroform extraction protocol. Wild-type locus was amplified using primers that anneal to the 5Ј-end (5Ј-TGCCACACCTCACTGCTTAC-3Ј) and the 3Ј-end of exon 9 (5Ј-TTAATGGACACCCTGCTTCC-3Ј). Targeted locus was amplified using primers to the 3Ј-end of exon 9 as mentioned and the 5Ј-PGK-neo cassette sequence (5Ј-TATCGCCGCTCCCGATTC-3Ј). PCR was performed using HOTMASTER Taq polymerase (Eppendorf AG, Hamburg, Germany) with annealing temperature of 62°C for 35 cycles.

RESULTS
Identification of Fibulin-1D by Yeast Two-hybrid Screening as an ADAMTS-1-binding Protein-A yeast two-hybrid screen of a human placenta cDNA library was conducted to identify proteins capable of interacting with ADAMTS-1. A fragment of the carboxyl-terminal portion of human ADAMTS-1 containing the two last thrombospondin type I repeats (amino acid residues 827-951) was used as a bait to screen for potential binding proteins (Fig. 1). Among several positive clones identified, one clone, designated F3.3.2, contained a 1371-bp insert that corresponded to nucleotides 1439 -2810 of a human fibulin-1D cDNA (GenBank TM accession number AF126110). The sequence encoded by clone F3.3.2 corresponds to the carboxyl-end of fibulin-1D (amino acid residues 476 -703), including the two last epidermal growth factor-like elements and the carboxyl-terminal fibulin-type module (Fig. 1).
Evaluation of the ADAMTS-1-Fibulin-1 Binding Interaction-The interaction between ADAMTS-1 and fibulin-1 was further validated by several biochemical approaches. Tosyl-activated magnetic beads coupled to fibulin-1 were able to bind to purified ADAMTS-1 in solution, whereas BSA-coated magnetic beads did not ( Fig. 2A). Fibulin-1C immunoprecipitated from conditioned medium with mAb 3A11 was able to pull down ADAMTS-1 (Fig. 2B, lane 2). Anti-fibulin-1 mAb 3A11 only recovered trace amounts of ADAMTS-1 when in the absence of fibulin-1C (Fig. 2B, lane 5). In addition, the interaction was tested by enzyme-linked immunosorbent assay using purified proteins. As shown in Fig. 2C, fibulin-1 bound to immobilized ADAMTS-1. The reciprocal experiment was also performed, and ADAMTS-1 was found to bind to immobilized fibulin-1 (Fig. 2C). The binding affinity between these two proteins was also determined. Fibronectin was used as a positive control because it has been shown to interact with fibulin-1 (19,26). 96-well plates were coated with human fibronectin, ADAMTS-1, or BSA (negative control) at 3 g/ml and subsequently allowed to bind fibulin-1 at varying concentrations. Based on a fit of the data to a form of the binding isotherm (26), fibulin-1 bound to ADAMTS-1 with an apparent dissociation constant (K d ) of 1.0 M as compared with K d of 0.43 M for fibulin-1 binding to fibronectin (Fig. 2D). As previously shown for fibulin-1-fibronectin interactions (19), saturable binding was not achieved between fibulin-1 and immobilized ADAMTS-1. The basis for this may relate to the fact that fibulin-1 can self associate (19); thus, with increased binding of solution phase fibulin-1 to the immobilized ADAMTS-1, additional fibulin-1 binding sites are created.
The interaction between fibulin-1 and ADAMTS-1 was also tested using ADAMTS-1-Sepharose affinity chromatography of placental extracts. As shown in Fig. 3A, low ionic strength buffer released weakly interacting fibulin-1 (fractions 3-13). Subsequent exposure to higher pH and high ionic strength buffer released tightly bound fibulin-1 (fraction 22). No fibulin-1 binding was observed when the plain Sepharose column used to preclear the extract was exposed to the same elution protocol.
Finally, we performed co-immunoprecipitation analyses using placental extracts and fibulin-1 monoclonal antibodies. These experiments revealed that ADAMTS-1 could be co-immunoprecipitated with fibulin-1 (Fig. 3B, lane 3). Conversely, fibulin-1 was found to co-immunoprecipitate with ADAMTS-1 antibodies (Fig. 3B, lane 1, GPB). ADAMTS-1 or fibulin-1 were not present in immunoprecipitations using control mouse IgG (lane 2). Together these findings support the conclusion that fibulin-1 binds to ADAMTS-1. However, the biological significance of this interaction is limited to situations where both these proteins are localized. Thus the next step required a careful analysis of expression and localization.

ADAMTS-1 Colocalizes with Fibulin-1 in Vivo-
The relationship between fibulin-1 and ADAMTS-1 expression was first evaluated by Northern analysis. Interestingly, both transcripts displayed a similar  expression profile, at least in the organs evaluated from mouse embryos (Fig. 4A). Because the highest level of expression for both transcripts was noted in the kidney, we pursued immunohistochemical analysis in this organ using anti-fibulin-1 antibodies (24) and a newly developed ADAMTS-1 antibody (Fig. 4B). At day E15.5 of embryonic development, fibulin-1 localized to basement membrane of the developing nephron (Fig. 4C, panels d and g, arrows). This expression pattern correlated with the distribution of ADAMTS-1 protein, which was also detected in discrete patches within the basement membrane at this stage of development (panels e and h, arrows). It is important to emphasize that this colocalization was indeed patchy; in fact, ADAMTS-1 is not expressed in the adult kidney at that site. 4 The findings indicate that ADAMTS-1 is present in a relatively narrow window of time during the morphogenesis of the epithelial tubules in the developing kidney. This is consistent with previous findings by in situ hybridization (25).
Fibulin-1 Is Not Cleaved by ADAMTS-1-We next tested whether fibulin-1 was a substrate for ADAMTS-1. Immunoblot analysis using two fibulin-1 antibodies showed no cleavage product nor was the amount of full-length fibulin-1 diminished by incubation with ADAMTS-1 (Fig. 5, A and B). To address the possibility that digestion destroyed fibulin-1 antibody reactive epitopes, we incubated biotinylated fibulin-1 with ADAMTS-1 and detected the digests using avidinconjugated HRP (Vector Labs). No cleavage products were detected using this approach (Fig. 5C). Because fibulin-1 is an ECM protein, we speculated that conformational changes associated with ECM interaction could expose cleavage sites. We therefore allowed fibulin-1 to bind   to immobilized fibronectin and evaluated the ability of ADAMTS-1 to mediate fibulin-1 cleavage. As shown in Fig. 5D, fibronectin-immobilized fibulin-1 was also not cleaved by ADAMTS-1. Taken together, the findings indicate that fibulin-1 is not a substrate of ADAMTS-1.
Fibulin-1 Enhances the Catalytic Activity of ADAMTS-1-Because fibulin-1 was not cleaved by ADAMTS-1, we speculated that the interaction might modulate the function of one or another. Fibulin-1 has been shown to bind the C-type lectin domains of aggrecan and versican (20,22). These two proteoglycans are recognized substrates for ADAMTS-1 (9 -11). Thus, we investigated the possibility that fibulin-1 might modulate the proteolytic activity of ADAMTS-1 toward aggrecan. In the presence of fibulin-1, full-length aggrecan was completely cleaved to the 250-kDa fragment by ADAMTS-1 (Fig. 6A). In contrast, inclusion of other proteins (fibronectin, vitronectin, laminin-1, and BSA) resulted in incomplete digestion of the full-length aggrecan over the same time period (Fig. 6A). Fibronectin was also able to enhance ADAMTS-1 catalytic activity (ϳ2-fold over BSA); however, it was not as efficient as fibulin-1. The other ECM proteins tested failed to enhance ADAMTS-1 proteolytic activity. None of the ECM proteins tested (i.e. fibulin, laminin-1, fibronectin, and vitronectin) leads to an inhibition of ADAMTS-1 aggrecanase activity (Fig. 6A).
Next, we evaluated the effect of fibulin-1 on the kinetics of aggrecan proteolysis. In this figure, the degree of contamination with the 250-kDa form was ϳ50% (see "Experimental Procedures"), yet it is clear that addition of ADAMTS-1 processed the intact aggrecan to completion, as early as 30 min when in the presence of fibulin-1. In addition, the lower 65-kDa aggrecan species was also detected at 30 min, and levels increased progressively with time. By contrast, digestion was still incomplete after 120 min when laminin-1 was present (Fig. 6B). At the same time point, aggrecan cleavage product (65 kDa) with fibulin-1 was 4-fold greater than with laminin-1 (Fig. 6B).
The effect of fibulin-1 protein concentration on the cleavage reaction was also tested. These studies showed that the extent of aggrecan cleavage to release the 250-and 65-kDa fragments was proportional to the concentration of fibulin-1 in the reaction (Fig. 6C). The enhancement of ADAMTS-1 catalytic activity was specific to fibulin-1, as increasing concentrations of laminin-1 failed to augment ADAMTS-1 activity (Fig.  6C). Furthermore, the use of gradient gels revealed that presence of fibulin-1 significantly altered the mobility of aggrecan (Fig. 6C, compare  lanes 1 and 2, brackets). Thus, from a high molecular mass complex in the absence of fibulin-1, aggrecan resolves into a dual lower molecular mass doublet. This change was dependent on levels of fibulin-1 and did  not change after the ratio of both proteins reached equality (i.e. 1:1, aggrecan to fibulin). In contrast, the presence of laminin at 1:1 molar ratio did not result in the same mobility shift, although aggrecan appeared as a high molecular mass smear. The data indicate that fibulin-1 alters the conformation of aggrecan in a manner that is not affected by exposure to detergents. Although fibulin-1 significantly affected the mobility of aggrecan, processing of the proteoglycan required ADAMTS-1 (Fig. 6D).
ADAMTS-1 Is a Biologically Important Aggrecanase in the Developing Kidney-Our previous results demonstrated colocalization of fibulin and ADAMTS-1 in the developing kidney tubules, and thus we sought to determine whether this enzyme is relevant for the processing of aggrecan at similar sites. To answer this question we inactivated the ADAMTS-1 gene by homologous recombination in mouse (Fig. 7, A-C). Although the targeting vector was slightly different from those described by other investigators (13,14), the phenotype is very similar to previously described ADAMTS-1 null mice. Absence of ADAMTS-1 results in significant lethality at the intraembryonic/neonatal stage. Mendelian distribution of heterozygous crosses indicates that 40% of homozygous animals for the null mutation are not recovered (assessment done in Ͼ100 offspring of mixed background). The remaining homozygous animals display kidney pathologies and urinary track obstructions. In addition, ADAMTS-1 null mice have lower weight mass than littermate wild-type controls (wild-type ϭ 36.18 g; homozygous ϭ 28.83 g average from littermate males 4 -6 months of age). We evaluated expression of aggrecan and ADAMTS-1 in wild-type and ADAMTS-1 null littermates at E15.5 (Fig. 7D). A significant degree of colocalization was detected in the apical epithelial tubules as well as in the basement membrane (Fig. 7D, panels a-c), a site where we also found fibulin-1 (Fig. 4C). Expression of aggrecan was not altered in the absence of ADAMTS-1 (Fig. 7D, panels d-f). Interestingly, using a neoepitope antibody to recognize cleaved aggrecan (11), we were also able to detect colocalization with ADAMTS-1 (panels g-i). However, cleaved aggrecan was poorly detected in the absence of ADAMTS-1 (panels j-l). These results indicate that, at least in the kidney, ADAMTS-1 is a biologically relevant aggrecanase.

DISCUSSION
Using a yeast two-hybrid screen, we demonstrated that fibulin-1 binds to ADAMTS-1 through the last three epidermal growth factorlike repeats in fibulin-1 and thrombospondin type I repeats 2 and 3 in ADAMTS-1. Interaction of these proteins was further confirmed by several biochemical assays, including pulldown, solid phase binding, and co-immunoprecipitation, and it was also validated in vivo. It is likely that ADAMTS-1 interacts with both fibulin-1C and fibulin-1D variants, as the interacting region is common to both proteins. To gain insight into the biological relevance of this binding, we tested the effect of fibulin-1 on the catalytic activity of ADAMTS-1. We found that fibulin-1 enhanced the cleavage of aggrecan by ADAMTS-1. Interestingly, this proteoglycan binds to fibulin-1 with relatively high affinity through the C-lectin domains, and they are frequently co-expressed in vivo (20,22). Taken together, these findings indicate that the interaction of fibulin-1 with aggrecan may serve to bring these proteins together, resulting in more efficient proteolytic degradation by ADAMTS-1.
Yeast two-hybrid screens have been fruitful in the identification of extracellular protein interactions (27)(28)(29)(30)(31). For example, the use of catalytically inactive matrix metalloproteinases in yeast two-hybrid screens has revealed novel substrates and modulators for several enzymes (32). Catalytic activity requires proximity between the substrate and the enzyme. At times, this proximity is achieved by the interaction of sub-strates with non-catalytic regions of the enzyme. The hemopexin domain present in some matrix metalloproteases has been shown to fulfill this attribute (33). In fact, deletion of this region has a profound effect on the efficacy of the enzyme (33). In other cases, a third protein component fulfills this role (34).
The thrombospondin type I repeats in ADAMTS-1 have been shown to interact with heparin and are also likely to bind to other ECM proteins (35). Removal of this domain increases the solubility of the enzyme in tissue culture conditions and decreases its catalytic activity (36,37). In contrast, removal of the carboxyl-terminal region in ADAMTS-4 improves the activity of this molecule (38,39). Thus, it appears that the function of the carboxyl-terminal region might differ in different ADAMTS proteins. Recently, fibronectin was found to bind ADAMTS-4 via the carboxyl-terminal region. More importantly, this binding inhibits ADAMTS-4 proteolytic activity toward aggrecan (40).
Contrary to our expectations, fibulin-1 did not hinder the activity of ADAMTS-1 nor was it found to be a substrate for the enzyme. Unlike the interaction between fibronectin and ADAMTS-4, fibulin-1 significantly enhanced the catalytic activity of ADAMTS-1. Interestingly, we found that fibronectin could also increase ADAMTS-1-mediated proteolytic cleavage of aggrecan but not to the same degree as fibulin-1 (Fig.  6A). Cofactor activity has been observed for other ECM proteins. For example, procollagen C-proteinase enhancer increased the proteolytic processing of procollagen by bone morphogenetic protein-1, an activity that required binding of the cofactor to the substrate (34).
The similar in vivo expression patterns of ADAMTS-1 and fibulin-1 indicate that this interaction could be of physiological relevance. Fibulin-1 and ADAMTS-1 were found colocalized in the kidney basement membrane (Fig. 4C, arrows). These findings are consistent with studies that independently demonstrated expression of transcripts for fibulin-1 and ADAMTS-1 in the developing nephron and several epithelial structures (41,25). Interestingly, mice that lack fibulin or ADAMTS-1 die from kidney abnormalities (13,14,23).
Surprisingly, we did not detect ADAMTS-1 in developing cartilage, a prominent site for fibulin-1 and aggrecan. Thus, the participation of ADAMTS-1 in aggrecan proteolysis appears to be tissue specific. In fact, a recent study convincingly demonstrated that ADAMTS-1 does not participate in aggrecan turnover in the growth plate and articular cartilage, a site where expression of ADAMTS-4 and 5 predominates (42,43). In the kidney, however, the roles appear reversed, as expression of ADAMTS-4 and 5 is non-detectable. Supporting an aggrecanase activity for ADAMTS-1 in the kidney, we found diminished-to-undetectable levels of aggrecan processing in the ADAMTS-1 null mice in comparison to control (Fig. 7). These findings might bear significance to the kidney phenotype displayed by the ADAMTS-1 null mouse.
A genetic demonstration of the interaction between fibulin-1 and members of the ADAMTS family has been recently shown in Caenorhabditis elegans (47). These authors found that the ADAM proteinase, GON-1, and fibulin-1 have antagonistic roles in the regulation of gonad morphogenesis. Together, our findings and those on GON-1 suggest that fibulin has affinity for and is a relevant partner of multiple ADAMTS proteases.
Several ECM molecules are known to regulate the activity of matrix metalloproteases (34,33). Our findings offer the first demonstration of