Biochemical characterization of the catalytic domain of human matrix metalloproteinase 19. Evidence for a role as a potent basement membrane degrading enzyme.

We have recently cloned MMP-19, a novel matrix metalloproteinase, which, due to unique structural features, was proposed to represent the first member of a new MMP subfamily (Pendás, A. M., Knäuper, V. , Puente, X. S., Llano, E., Mattei, M. G., Apte, S., Murphy, G., and López-Otin, C. (1997) J. Biol. Chem. 272, 4281-4286). A recombinant COOH-terminal deletion mutant of MMP-19 (proDelta(260-508)MMP-19), comprising the propeptide and the catalytic domain, was expressed in Escherichia coli, refolded, and purified. Interestingly, we found that proDelta(260-508)MMP-19 has the tendency to autoactivate, whereby the Lys(97)-Tyr(98) peptide bond is hydrolyzed, resulting in free catalytic domain. Mutation of two residues (Glu(88) --> Pro and Pro(90) --> Val) within the propeptide latency motif did not prevent autoactivation but the autolysis rate was somewhat reduced. Analysis of the substrate specificity revealed that the catalytic domain of MMP-19 was able to hydrolyze the general MMP substrate Mca-Pro-Leu-Gly-Dpa-Ala-Arg-NH(2) and, with higher efficiency, the stromelysin substrate Mca-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH(2). Kinetic analysis of the interactions of the catalytic domain of MMP-19 with the natural MMP inhibitors, the tissue inhibitors of metalloproteinases (TIMPs), showed strong inhibition using TIMP-2, TIMP-3, and TIMP-4, while TIMP-1 was less efficient. We also demonstrated that synthetic hydroxamic acid-based compounds efficiently inhibited the enzyme. The catalytic domain of MMP-19 was able to hydrolyze the basement membrane components type IV collagen, laminin, and nidogen, as well as the large tenascin-C isoform, fibronectin, and type I gelatin in vitro, suggesting that MMP-19 is a potent proteinase capable of hydrolyzing a broad range of extracellular matrix components. Neither the catalytic domain nor the full-length MMP-19 was able to degrade triple-helical collagen. Finally, and in contrast to studies with other MMPs, MMP-19 catalytic domain was not able to activate any of the latent MMPs tested in vitro.

The human matrix metalloproteinases (MMPs) 1 are a group of homologous zinc-dependent endopeptidases that degrade the different macromolecular components of the extracellular matrix. They have been implicated in the remodeling of connective tissues during such diverse processes as normal mammalian development and growth, wound healing, cartilage degradation during arthritis, and cancer metastasis (1)(2)(3). At present 18 members of the human MMP family have been cloned, and they have been classified into different subfamilies according to their substrate specificity and cellular location. This classification comprises the collagenases, gelatinases, stromelysins, and membrane-type MMPs (MT-MMPs). We have recently cloned a new member of the matrix metalloproteinase family, MMP-19, which showed the typical domain organization of soluble members of the MMP family, namely a signal sequence, a propeptide domain with the cysteine residue essential for maintaining latency, a catalytic domain with the typical zinc binding motif, a linker region, and a COOH-terminal fragment with sequence similarity to hemopexin (4). However, the enzyme lacks various structural features distinctive of the diverse MMP subfamilies, e.g. the fibronectin-like repeats of gelatinases or the Asp, Tyr, and Gly residues near the active site of collagenases, but possesses a unique insertion of five Glu residues within the linker region, an unusual latency motif in the propeptide domain ( . . . PRCGLEDP . . . ) and an additional Cys residue in the catalytic domain, when compared with other MMPs. In addition, the MMP-19 gene is the first MMP gene found to be located on chomosome locus 12q14 and initial data on its genomic organization has revealed a unique intron/exon distribution. MMP-19 may therefore represent the first member of a new subfamily of MMPs (4), whose role in vivo remains to be investigated. However, recently, MMP-19 mRNA was found to be constitutively expressed in arthritic (RA) and traumatic synovial membranes, which may imply the involvement of MMP-19 in this tissue during normal ECM remodeling processes (5). Northern blot analysis of polyadenylated RNA from various normal human tissues revealed strong expression of MMP-19 in placenta, ovary, lung, pancreas, spleen, and intestine, whereas expression in brain and leukocytes was undetectable (4). Since adult cells under non-pathological conditions do not frequently produce MMPs, it is possible that MMP-19 participates in normal ECM turnover or in activation of secreted and membrane-bound proteins such as growth factors and protein-ases (4). More recently, Sedlacek and co-workers (6) reported that enhanced anti-MMP-19 autoantibody titers appear to be frequent among patients suffering from RA, and that MMP-19 was detected using an anti-peptide antibody on the surface of lymphatic cells such as activated peripheral blood mononuclear cells, T H 1 lymphocytes, and Jurkat T lymphoma cells. Furthermore, a distinct expression of MMP-19 was observed, associated with the smooth muscle cells in the tunica media of synovial blood vessels of an RA patient, as well as in normal skin and uterine ligaments (7). In contrast, in capillaries of acutely inflamed RA synovium strong MMP-19 expression was also detected in the endothelial layer (8). The same authors report an elevated MMP-19 mRNA expression of smooth muscle cells in vitro after stimulation with 12-O-tetradecanoylphorbol-13acetate, epidermal growth factor, and basic fibroblast growth factor, whereby proliferating smooth muscle cells exhibited higher levels of MMP-19 mRNA than resting cells. MMP-19 protein and mRNA was detected in vitro in endothelial cells from various tissues, e.g. umbilical artery, skin, and fat tissue. These data support the hypothesis that MMP-19 participates in angiogenic processes and lymphocyte extravasation during arthritic diseases and therefore may be involved in the invasion of the inflamed synovial pannus into the joint space and thus in the destruction of joint tissues. Here we describe the expression, refolding, and enzymatic characterization of the catalytic domain of MMP-19.

EXPERIMENTAL PROCEDURES
Expression, Refolding, and Purification of Pro⌬ 260 -508 MMP-19 -An expression vector for pro⌬ 260 -508 MMP-19 was generated by PCR using the following primers: 5Ј-GGCGCCTGCAGACTACCTGTCACAATAT-GGGTACCTACAGAAGCC-3Ј (A) and 5Ј-CCGGAATTCTCAACTCT-TCTTGCCATAGAGAGCCTGGATCCCTGC-3Ј (B) using the full-length MMP-19 cDNA in pSP64 as a template, thereby introducing a PstI site at the 5Ј end and a stop codon following Ser 259 flanked by a unique EcoRI site at the 3Ј end of the PCR product. The 800-bp PCR product was cleaved with the restriction endonucleases PstI and EcoRI and ligated in frame into the pRSET B expression vector (Invitrogen), thereby adding an NH 2 -terminal His 6 tag to the protein. The correct sequence of the pro⌬ 260 -508 MMP-19 cDNA was confirmed by doublestrand dideoxy sequencing. The expression of pro⌬ 260 -508 MMP-19 as inclusion bodies in competent BL21(DE3)pLysS E. coli cells was carried out as described previously (4). The inclusion bodies were solubilized in 20 mM Tris/H 2 SO 4 , pH 8.0, 6 M urea, 5 mM ␤-mercaptoethanol, 0.01% NaN 3 at 37°C for 30 min. Refolding was achieved by dilution (1:250) into 20 mM Tris/H 2 SO 4 , pH 7.6, 100 mM Na 2 SO 4 , 5 mM CaSO 4 , 5 M ZnSO 4 , 5% glycerol, 0.03% Brij 35 (w/v), 0.01% NaN 3 at 4°C. The refolding mixture was filtered and applied to a nickel-nitrilotriacetic acid-agarose column (Qiagen) equilibrated in above buffer. Under these conditions, the protein bound via the NH 2 -terminal His 6 tag to the column and was washed with above buffer containing 5 mM imidazole. The protein was eluted with buffer supplemented with 100 mM imidazole. After SDS-PAGE analysis, fractions containing pro⌬ 260 -508 MMP-19 were combined and dialyzed against 20 mM Tris/HCl, pH 7.6, 200 mM NaCl, 5 mM CaCl 2 , 0.03% Brij 35 (w/v), 0.01% NaN 3 , followed by centrifugation to remove precipitated protein and concentration in an Amicon concentrator with a M r 10,000 cut-off cartridge. All buffers used for refolding and purification contain an EDTA-free protease inhibitor mixture at concentrations recommended by the supplier (Roche Molecular Biochemicals).
Mutagenesis of Pro⌬ 260 -508 MMP-19 -To alter two sequence motifs within pro⌬ 260 -508 MMP-19 site-directed mutagenesis was performed using the method of Ho et al. (9). The mutations were located in the propeptide (E88P/P90V) and in the catalytic domain (C166S).
A protein containing alterations in both, propeptide and catalytic domain, was produced, thereby altering E88P/P90V and C166S by ligating the PstI and NcoI fragment from pro⌬ 260 -508 MMP-19(E88P/ P90V) into the pRSET B vector containing the mutation for C166S in the catalytic domain, previously cleaved with the same restriction enzymes. Expression and refolding were performed as described above. All expression vectors were sequenced using the dideoxy chain termination method and confirmed the correct sequence for each construct.
Expression and Purification of Full-length MMP-19 -Human MMP-19 was purified from culture medium conditioned by NS0 mouse myeloma cells that had been transfected with MMP-19 cDNA, essentially as described previously (10). The expression vector, transfection method, and culture conditions were as described previously (11,12).
Activity Assays, Active Site Titration, and Kinetic Analysis of Inhibitor Binding-Enzymatic activity was determined after activation of wild type or mutated pro⌬ 260 -508 MMP-19 with 1 mM APMA for 30 min at 37°C (16). Routine assays were performed at 37°C using 1 mM enzyme and the synthetic quenched fluorescent peptide (7-methoxycoumarin-4-yl)acetyl-Pro-Leu-Ala-Nva-[3-(2, 4-dinitrophenyl)-L-2,3-diaminopropionyl]-Ala-Arg-NH 2 ( ex 328 nm, em 393 nm) as substrate at a concentration of 1 M in assay buffer (50 mM Tris/HCl, pH 7.6, 150 mM NaCl, 10 mM CaCl 2 , 0.05% (v/v) Brij 35, 0.01% NaN 3 ). The concentration of active enzyme was evaluated using the fluorescent assay by titration against a standard TIMP-2 solution of known concentration (12). Determination of the substrate specificity of the catalytic domain of MMP-19 was carried out using the same assay with different quenched fluorescent synthetic peptide substrates at concentrations of 1 M, which fulfilled the requirements of [S] Ͻ Ͻ K m . All peptide substrates were kindly provided by Dr. Graham Knight (University of Cambridge, Cambridge, United Kingdom). The pseudo-first-order rate constants (k) for the formation of the EI complex of 1 nM active wild type ⌬ 260 -508 MMP-19 with TIMP-1, -2, -3, and -4, and the ⌬ 128 -194 TIMP-2 mutants S2E, Y36G, and A70K were determined by analysis of the progress curves of McaPLANvaDpaARNH 2 hydrolysis (12). The dependence of k on TIMP concentration was evaluated using various amounts of TIMP (5-20 nM). In addition, the apparent K i app values for TIMP inhibition were determined using 1 nM enzyme incubated with a range of inhibitor concentrations for 24 h at 37°C to reach equilibrium before assayed as described above.
The apparent K i values for the inhibition of active wild type ⌬ 260 -508MMP-19 with the synthetic hydroxamic acid-based inhibitors CT-1746, Ro31-9790, and BB-94 were determined by incubation of 1 nM enzyme with a range of inhibitor concentrations for 24 h at 37°C before assayed.
pH Dependence of ⌬ 260 -508 MMP-19 Activity -The pH dependence of the activity of ⌬ 260 -508 MMP-19 was determined at 37°C, after complete activation with 1 mM APMA for 30 min, using 1 nM enzyme and 1 M synthetic fluorescent substrate McaPLANvaDpaARNH 2 in a buffer of 20 mM MES, 20 mM Tris, 20 mM CAPS, 150 mM NaCl, 10 mM CaCl 2 , 0.05% (v/v) Brij 35. The pH was adjusted with HCl in steps of 0.5 before every measurement, covering a range from pH 5 to pH 11.
Cleavage of Extracellular Matrix Components-A wide variety of extracellular matrix molecules such as collagen type I, gelatin type I, collagen type IV, laminin, nidogen, fibronectin, tenascin-C (small and large isoforms), fibrin, and fibrinogen were incubated in time-course experiments at 37°C (if not stated otherwise) with active ⌬ 260 -508MMP-19 prior to analysis by SDS-PAGE. Type I collagen was prepared from rat skin, as described previously (17). Tenascin-C small and large isoforms were obtained from baby hamster kidney cells transfected with tenascin-C cDNA constructs (18). Mouse laminin, human fibronectin and human fibrinogen were purchased from Sigma. Collagen type IV and nidogen were generous gifts from Klaus Kü hn and Rupert Timpl, respectively. Fibrin was generated from fibrinogen by clotting with thrombin prior to incubation with active ⌬ 260 -508 MMP-19 (19). Monoclonal antibodies to fibronectin domains were purchased from Life Technologies, Inc. Human recombinant stromelysin-1 (MMP-3) and gelatinase A (MMP-2), which were used as a comparison, were prepared as described previously and activated by trypsin and APMA, respectively (20,21). The enzyme/substrate ratio (w/w) used in these experiments was 1/10. Furthermore, 14 C-labeled gelatin and casein were used in an assay described by Cawston and Barrett (17) to quantify the specific catalytic activity of wild type ⌬ 260 -508 MMP-19 hydrolyzing these molecules. Gelatin and casein zymography was per-formed as described previously (22). NH 2 -terminal Sequence Determination-Proteins were purified by reverse phase high performance liquid chromatography using a C 18column followed by automated Edman degradation using a PE Biosystems 492 Procise protein sequencer operated according to manufacturer's instructions.

RESULTS AND DISCUSSION
Human MMP-19 is a novel member of the matrix metalloproteinase family, cloned recently from a human liver cDNA library (4). Consequently, biochemical analysis of the activation mechanism, substrate specificity, and inhibition profile of MMP-19 is of vital importance in order to understand its possible function in vivo and to design specific inhibitors as potential new therapeutic agents. Since purification of full-length MMP-19 expressed in mammalian and bacterial expression systems resulted in low yields, autoproteolytic activation, and partial fragmentation, we decided to use the COOH-terminal truncated form of MMP-19, pro⌬ 260 -508 MMP-19 (numbering starts at Met 1 , GenBank™/EBI accession number X92521), for this study and analyzed its enzymatic properties.
Refolding, Purification, and Activation of Pro⌬ 260 -508 MMP-19, and Assessment of Mutations within the Pro and Catalytic Domains-In order to further characterize the catalytic activity of MMP-19, we expressed and refolded the COOH-terminal truncated form of MMP-19, pro⌬ 260 -508 MMP-19, comprising the pro and catalytic domains, and analyzed the protein biochemically in detail. After solubilization and refolding, pro⌬ 260 -508 MMP-19 was purified using nickel-nitrilotriacetic acid-agarose and the eluted proenzyme displayed the expected mass of 30 kDa when analyzed by SDS-PAGE (Fig. 1A). Surprisingly, following dialysis to remove the imidazole, all the enzyme autoactivated, resulting in the generation of active enzyme with a mass of 20 kDa under reducing conditions (Fig.  1B). NH 2 -terminal amino acid sequence determination of the active catalytic domain confirmed that the Lys 97 -Tyr 98 peptide bond was hydrolyzed during autoproteolytic activation (Fig. 2).
Similar autoactivation was observed when we attempted to purify full-length MMP-19 from the culture medium of stable transfected NS0 cells (data not shown). In order to define the residues in the proenzyme that make proMMP-19 prone to autoactivation, we performed site-directed mutagenesis experiments by targeting the latency motif PRCGLE 88 DP 90 of proMMP-19 and the residue Cys 166 in the catalytic domain, which are structural elements different from other MMPs. The mutants pro⌬ 260 -508 MMP-19(C166S), pro⌬ 260 -508 MMP-19(E88P/P90V), and pro⌬ 260 -508 MMP-19(C166S/E88P/P90V) were expressed, solubilized, refolded, and purified in the same way as described for the wild type pro⌬ 260 -508 MMP-19. All mutants are of the same size as wild type when analyzed by SDS-PAGE and activate spontaneously during refolding and purification ( Fig. 1 and data not shown). However, the tendency to undergo activation is slowed down by approximately 50% for the two mutants pro⌬ 260 -508 MMP-19(E88P/P90V) and pro⌬ 260 -508 MMP-19(C166S/E88P/P90V), confirming previously published data, where the effects of single mutations within the latency motif of transin (rat stromelysin) upon activation were reported (23). In both of these pro⌬ 260 -508MMP-19 mutants, the alterations reestablish the sequence PRCGVPDV, conserved within the propeptide of MMPs. Thus, our results suggest that the tendency of pro⌬ 260 -508 MMP-19 and full-length MMP-19 to undergo autoactivation can be ascribed partially to its unique latency motif. Cys 166 within the catalytic domain seems not to influence activation and its function remains unclear. However, we observed low amounts of dimers of active recombinant wild type ⌬ 260 -508 MMP-19 by casein zymography and SDS-PAGE, whereas the mutants containing C166S did not form any dimers under the same conditions (data not shown). There is no in vivo evidence, however, that the residue Cys 166 facilitates dimerization or linkage to other proteins.
If not stated otherwise, we used the active wild type catalytic domain of MMP-19 (active ⌬ 260 -508 MMP-19) for further biochemical characterization.
Kinetic Analysis of Cleavage of Quenched Fluorescent Substrates and TIMP Binding-In order to assess the substrate specificity of active ⌬ 260 -508 MMP-19 further, seven synthetic quenched fluorescent peptide substrates were employed (Table  I). Only the general MMP substrate Mca-Pro-Leu-Gly-Dpa-Ala-Arg-NH 2 and the stromelysin substrate Mca-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH 2 were hydrolyzed efficiently, while other substrates were resistant to hydrolysis, thus confirming our preliminary data (4). The k cat /K m values obtained for these two substrates are in agreement with our earlier results. Therefore, the catalytic domain of MMP-19 has a substrate specificity similar to that for the stromelysin subfamily of MMPs since it preferably hydrolyzed Mca-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH 2 , which was designed as a substrate to study stromelysin activity. However, stromelysin-1 (MMP-3) hydrolyzes both substrates, Mca-Pro-Leu-Gly-Dpa-Ala-Arg-NH 2 and Mca-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH 2 , more efficiently than active ⌬ 260 -508 MMP-19 (20).
Furthermore, we followed ⌬ 260 -508 MMP-19 activity over a wide pH range (pH 5.0 -11.0) using a constant enzyme concentration of 1 nM and Mca-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH 2 as substrate in the quenched fluorescent assay (Fig. 3). The activity, measured in steps of 0.5, showed low k cat /K m values at pH 5.0 and a sudden increase toward pH 7.0, where the highest value was obtained, followed by a gradual decrease toward pH 10.5, where no activity was measurable. Thus, MMP-19 is active over a wide pH range and displays maximum activity at pH 7.0 in our assay. A comparison of the pH profiles of MMP-19 and MMP-3 reveals that the latter seems to prefer a more acidic environment, reaching maximum activity at pH 6.0 (24).
The apparent K i values for the inhibition of active ⌬ 260 -508 MMP-19 with the various TIMPs were determined using a constant enzyme concentration of 1 nM in the quenched fluorescent assay at 37°C with Mca-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH 2 as substrate (Table II). Under these conditions, only the K i app for TIMP-1 was determined with accuracy, since its value (57.6 nM) is well above the employed enzyme concentration and represents the weakest value ascertained for fulllength TIMPs. In contrast, the K i app for TIMP-2, TIMP-3, and TIMP-4 were all in the range of 4 -5 pM and are only rough estimates of the true values, because we were not able to determine them using enzyme concentrations below K i app due to the lack of assay sensitivity. However, the K i app values in the picomolar range indicate very strong enzyme-inhibitor interac-tions. The same restrictions apply to the K i app for the interactions of COOH-terminal truncated forms of enzyme and inhibitor, active ⌬ 260 -508 MMP-19 and ⌬ 128 -194 TIMP-2, which also result in values within the picomolar range and are, thus, not accurate. In addition, we investigated the inhibitory potential of mutants of ⌬ 128 -194 TIMP-2, containing single amino acid changes, with the aim to determine the sites in the NH 2terminal, inhibitory region of TIMP-2, which are important for interactions with MMP-19 (Table II). From the three mutants used in this study, Y36G, A70K and S2E, only A70K (745 pM) and S2E (61.57 nM) show a marked decrease in affinity for the enzyme. In particular, the mutation S2E results in a dramatic decline of the inhibitory potential when compared with the wild type ⌬ 128 -194 TIMP-2.
Furthermore, the association rate of enzyme and inhibitor complex formation were assessed by measuring the curvature in the progress curve of substrate hydrolysis and analyzed as described previously (Table II) (4), which is in agreement with the high homology between these two inhibitors (26). Interestingly, the interaction of the catalytic domain of MMP-19 with TIMP-2 was found to be about 4 times faster than the one observed with the COOH-terminal deletion mutant ⌬ 128 -194 TIMP-2 (4.18 ϫ 10 5 M Ϫ1 s Ϫ1 and 1.02 ϫ 10 5 M Ϫ1 s Ϫ1 , respectively), suggesting an influence of the COOH terminus of TIMP-2 on the association with active ⌬ 260 -508 MMP-19, which is in agreement with our previous data on the domain interactions involved in the binding of TIMPs to MMPs and suggestions from recent crystallographic analysis of the MT1-MMP/TIMP-2 complex (12,25,27). However, the small differences in K i app observed between TIMP-2 and ⌬ 128 -194 TIMP-2 (both picomolar) confirm that the NH 2 -terminal domain of TIMP-2 alone is able to form stable complexes with the enzyme and, thus, is sufficient for the inhibition of MMP-19 (Table II). These results are confirmed by data on the interactions of ⌬ 128 -194 TIMP-2 with MMP-3 (28).
The ⌬ 128 -194 TIMP-2 mutants Y36G and A70K do not show a significant change of association if compared with the wild type ⌬ 128 -194 TIMP-2. However, the mutation S2E slows down the association of active ⌬ 260 -508 MMP-19 and inhibitor with 1 order of magnitude (1.05 ϫ 10 4 M Ϫ1 s Ϫ1 ), and results in a 10,000fold increase in K i app (61.57 nM), therefore weakening the inhibitory properties considerably. In case of A70K, merely the K i app was increased about 50 times when compared with the wild type ⌬ 128 -194 TIMP-2, whereas the association rate constant was unchanged. From our data we can deduce that both mutations, A70K and S2E, result in increased enzyme-inhibitor complex dissociation, suggesting that the exchange of these amino acids compromise vital enzyme-inhibitor interactions. We therefore conclude that for the formation of the enzymeinhibitor complex, similar to other MMPs, ⌬ 260 -508 MMP-19 interacts with the so-called "ridge" region in TIMP-2 (Cys 1 -Cys 3 and Ser 68 -Cys 72 ) (15, 27, 29 -32).
Kinetic Analysis of Hydroxamate Inhibitor Binding-The involvement of MMPs in the breakdown and remodeling of the connective tissue under pathological conditions such as arthritis and cancer makes them attractive targets for the development of specific inhibitors for therapeutic intervention. In recent years, hydroxamic acid-based peptide inhibitors were developed, which react with 1:1 stoichiometry with MMPs, as revealed by x-ray crystallography (33,34). The apparent K i values for the inhibition of active ⌬ 260 -508 MMP-19 with the inhibitors BB-94 (British Biotech), CT-1746 (Celltech), and Ro31-9790 (Hoffmann-La Roche) were determined (Table III). Only the K i app for Ro30-9790 was determined with accuracy (1.22 nM), since its value is above the employed enzyme concentration, whereas the values obtained for the other two in-hibitors were both in the low picomolar range and therefore must be regarded as rough estimates of the real values. Interestingly, the K i app value for the inhibition of stromelysin-1 (MMP-3) with Ro31-9790 (119 nM) is about 100 times higher than our value determined for ⌬ 260 -508 MMP-19. In addition, the other two synthetic compounds tested seem also to inhibit ⌬ 260 -508 MMP-19 more efficiently than stromelysin-1, suggesting that the architectures of the active sites of these two enzymes, responsible for inhibitor interactions, are different (Table III).
Hydrolysis of Extracellular Matrix Components-A large variety of extracellular matrix components, purified from different connective tissues, were incubated with the catalytic domain of MMP-19 in time-course experiments prior to SDS-PAGE analysis to study its substrate specificity and to evaluate its possible function in vivo (Figs. 4 -8). These experiments were also performed using stromelysin-1 as a comparison, since active ⌬ 260 -508 MMP-19 exhibits similar specificity toward the synthetic fluorescent peptide substrate Mca-Pro-Leu-Ala-Nva-Dpa-Ala-Arg-NH 2 ( Table I).
As expected, collagen type I was resistant to hydrolysis by the catalytic domain of MMP-19 (Fig. 4A). However, full-length MMP-19 purified from the medium of stable transfected NS0 cells was also unable to cleave triple-helical collagen. Furthermore, catalytic domain and full-length MMP-19 exhibit similar activity versus gelatin (data not shown). Our data show that the influence of the COOH terminus on MMP-19 substrate specificity is negligible for the substrates studied, that the enzyme is non-collagenolytic, and thus the catalytic domain  represents a good model to study macromolecular substrate hydrolysis.
The catalytic domain of MMP-19 was able to degrade gelatin efficiently (Fig. 4B). However, at equivalent substrate/enzyme ratios (w/w), stromelysin-1 (MMP-3) and gelatinase A (MMP-2) display a considerably higher activity against this macromolecule, if analyzed by SDS-PAGE (Fig. 4B). Although ⌬ 260 -508 MMP-19 degraded gelatin, hydrolysis is not sufficient to generate a definable zone of lysis using gelatin zymography presumably due to the large mass of the generated fragments. Thus, it is not possible to use gelatin zymography to detect MMP-19 activity (Fig. 4C). In contrast, casein is hydrolyzed efficiently and ⌬ 260 -508 MMP-19 activity can therefore be monitored using casein zymograms (Fig. 4D). In comparison to ⌬ 260 -508 MMP-19, stromelysin-1 hydrolyzes gelatin and can therefore be analyzed by gelatin zymography. In addition, the specific activity of MMP-19 catalytic domain was assessed using 14 C-labeled gelatin and casein (17). In this assay, the en-zyme was able to hydrolyze gelatin with a specific activity of 78.45 units/mol and casein with 8491 units/mol. In comparison, the specific activities for the hydrolysis of gelatin by stromelysin-1 and gelatinase A are 2 to 4 orders of magnitude higher than observed for ⌬ 260 -508 MMP-19, respectively (20,21). On the other hand, specific activities of ⌬ 260 -508 MMP-19 and stromelysin-1 for the hydrolysis of casein are similar (20).
The two different isoforms of human tenascin-C were incubated with the MMP-19 catalytic domain (Fig. 5) (35). Analysis of the reaction products revealed that the small isoform was resistant (data not shown), as expected (18), while the large tenascin-C isoform was cleaved into two high molecular mass fragments displaying masses of 190 and 120 kDa. The cleavage pattern is identical to the one generated by stromelysin-1 under the same conditions, but MMP-19 seems to be more effective in processing the large isoform of tenascin-C. In addition, it was demonstrated that MMP-19 catalytic domain is able to hydrolyze the large isoform to the same sized products as gelatinase A (18).
Fibronectin was degraded by MMP-19 catalytic domain in a time-dependent manner (Fig. 6). Several fragments of 200, 160, 110, 100, 95, 80, 60, 45, 35, and 25 kDa were generated. Initially, fragments of 200, 160, 95, 80, 60, 35, and 25 kDa were produced after 2 h of incubation, and fragments of 160, 100, 60, 45, and 25 kDa were the final products (Fig. 6A). The cleavage products after 24 h of incubation were identical to those generated by stromelysin-1; however, active ⌬ 260 -508 MMP-19 was more efficient in cleaving fibronectin. The fibronectin degradation products generated by active ⌬ 260 -508 MMP-19 were characterized by immunoblotting using monoclonal antibodies against three different epitopes of the 220-kDa molecule (Fig.  6B). Three different bands, of 75, 64, and 60 kDa, were recognized by the antibody against the COOH-terminal heparinbinding domain (clone I). The larger two of these bands were also seen by the antibody raised against the central cell binding domain (clone II), but not by the antibody against the NH 2terminal gelatin-binding domain (clone III) of fibronectin, suggesting preferential processing within the NH 2 -terminal region of the molecule. Furthermore, clones II and III both detect bands of 110, 95, and 90 kDa. Interestingly, proteolytic fibronectin fragments have been shown to block fibronectin receptors, decrease cell attachment, and alter cell migration and therefore contribute to cell-regulatory processes (36). Fragments containing cell binding and heparin binding domains (75 and 64 kDa) may interfere with cell-proteoglycan interactions. In addition, fragments of fibronectin have been suggested to be involved in the induction of specific genes, such as proteases required for ECM remodeling (37). However, further studies are required to determine the possible function of fibronectin fragments generated by MMP-19 on cell attachment and cell migration.
Since MMP-19 was found in blood vessels in the synovium of RA patients, as well as in healthy skin and uterine ligaments, and therefore implicated in angiogenesis, degradation of basement membrane components was examined (Fig. 7) (7,8). In contrast to triple-helical collagen type I, collagen type IV, a major component of basement membranes consisting of two ␣ 1 (IV) chains and one ␣ 2 (IV) chain, was cleaved into two main high molecular mass fragments with molecular masses of 94 and 56 kDa in addition to some minor fragments (Fig. 7A). Stromelysin-1 generates two bands with masses of 42 and 94 kDa, which indicates a different specificity versus this molecule (Fig. 7A). Interestingly, MMP-19 was found to be co-expressed with collagen type IV in the tunica media of blood vessels of RA synovium (8). Laminin, a heterotrimer with an ␣-, ␤-, and ␥-chain (205, 215, and 400 kDa, respectively), was hydrolyzed by MMP-19 catalytic domain after 24 h to two major products with molecular masses of 64 and 140 kDa (Fig. 7B). The same cleavage products occur during hydrolysis with stromelysin-1 but not with the catalytic domain of MT1-MMP (38). However, under these conditions, the catalytic domain of MMP-19 seems to be more effective toward laminin than stromelysin-1. Nidogen, a basement membrane component with a mass of 158 kDa, was degraded by active ⌬ 260 -508 MMP-19 (Fig. 7C). The bands generated after a 24-h incubation had molecular mass values of 98, 72, and 60 kDa with a clear cleavage pattern, which was similar to the one generated by stromelysin-1. Like laminin, nidogen seemed to be degraded more efficiently by ⌬ 260 -508 MMP-19 than by stromelysin-1. Nidogen binds to laminin, type IV collagen, and proteoglycans and facilitates a build-up of the basement membrane's network (39). Thus, our results suggest that MMP-19 may be a potent player in the degradation of basement membrane components. We have demonstrated that type IV collagen, laminin, as well as nidogen are efficiently hydrolyzed by MMP-19, and, in conjunction with its expression in the tunica media, the enzyme may have a role during angiogenesis (7). To further investigate the ability of MMP-19 to hydrolyze ECM components, ⌬ 260 -508 MMP-19 was incubated with fibrinogen and fibrin, both molecules with a distinct role during new blood vessel formation (Fig. 7D). The cleavage pattern showed complete hydrolysis of the fibrinogen A␣ and B␤ chains after 1 h and loss of the ␥ chain after 5 h of incubation. After 24 h, fragments of sizes between 35 and 42 kDa seem to represent the final cleavage products. Fibrin was hydrolyzed by MMP-19 catalytic domain in a similar fashion to fibrinogen, resulting in three major cleavage products between 35 and 42 kDa and a double band at 72 kDa, which appeared to be derived from the ␥ dimers (19). In comparison, hydrolysis of fibrinogen and fibrin by stromelysin-1 for 24 h resulted in only two distinct fragments of approximately 42 and 37 kDa, whereas, in case of fibrin, a doublet at 42 kDa was obtained (Fig. 7D).
Activation of Other Pro-matrix Metalloproteinases-MMPs have been implicated in the activation and hydrolysis of secreted or membrane-bound proteinases (e.g. other MMPs) and precursors of growth factors (40,41). Thus, in order to determine the possible role of MMP-19 upon activation of other MMPs, we incubated various proMMPs (MMP-1, -2, -3, -8, -9, -13, and -14) with active ⌬ 260 -508 MMP-19. However, only human progelatinase B (MMP-9) was processed by the enzyme in a time-dependent fashion generating a final form of M r 82,000 after 24 h of incubation when analyzed by SDS-PAGE (Fig. 8).
The MMP-19 catalytic domain was not able to process any of the other MMPs tested (MMP-1, -2, -3, -13, and -14; data not shown). The processed form of human gelatinase B does not show hydrolytic activity against the synthetic quenched fluorescent substrate Mca-Pro-Leu-Gly-Dpa-Ala-Arg-NH 2 . NH 2terminal sequencing revealed that ⌬ 260 -508 MMP-19 cleaves the Lys 73 -Ala 74 bond upstream of the . . . PRCGVPD . . . sequence within the propeptide region of MMP-9, leaving its latency motif intact. This cleavage may induce conformational changes in the propeptide exposing the final activation site (Arg 87 -Phe 88 ) to be hydrolyzed by a second proteolysis, which can not be performed by MMP-19 (42).
Relatively high levels of MMP-19 expression were detected by Northern blot analysis in a wide variety of normal tissues (4). This pattern of expression is unusual for MMPs, which are not constantly produced by adult cells but are mostly induced during physiological conditions associated with extensive connective tissue remodeling, such as wound healing, uterine postpartum involution, or mammary gland involution. Therefore, MMP-19 may be involved in normal ECM remodeling processes. Interestingly, MT1-MMP (MMP-14) shows an expression pattern similar to that of MMP-19 and is regarded as major player in surface activation of other MMPs and ECM remodeling (43)(44)(45)(46). This work represents the first detailed characterization of the enzymatic properties of MMP-19; our studies have revealed the enzyme's tendency to autoactivate, a potent activity on a wide variety of ECM substrates, and evidence about TIMP interaction.