Human neutrophil collagenase (HNC) has been purified from extracts of fresh and outdated buffy coats and from the exudates of phorbol myristate acetate-stimulated neutrophils. The HNC present in the starting material from such preparations can be either latent or active, or have an approximate molecular weight of 75 or 58 kDa, depending upon whether the extraction buffer contains protease inhibitors and/or antioxidants. The purification of these different forms of HNC is described and is made possible by taking appropriate precautions to stabilize the HNC. For example, a purification protocol is described that allows the purification to homogeneity of the active and PCMB-active latent 58 kDa forms of HNC in high yield with specific collagenase activities that greatly exceed that of trypsin-activated human fibroblast collagenase (HFC). The pattern of activation of the latent 58 and 75 kDa species by trypsin, organomercurials and oxidants has been investigated. HNC is shown to preferentially hydrolyze type I over types II and III collagens in solution. The specificity of HNC toward the hydrolysis of 60 octapeptides has been examined and compared with HFC. HNC is shown to be a glycoprotein that contains complex N-linked oligosaccharides.
The zymogens of matrix metalloproteinase 1 (MMP-1: tissue collagenase), MMP-2 (gelatinase/type IV collagenase) and MMP-3 (stromelysin) were purified from the culture medium of human rheumatoid synovial fibroblasts and the mechanisms of activation of each zymogen by proteinases and 4-aminophenylmercuric acetate (APMA) were studied by kinetic and sequence analyses. The treatment of proMMP-1 (M(r) = 52,000) with proteinases or APMA converted the zymogen to M(r) = 43,000, but it exhibited only 14-25% of the maximal activity. Incubation of a partially active MMP-1 with MMP-3 resulted in rapid, full activation by generating the 41,000-M(r) MMP-1 with Phe81 as the NH2-terminus. MMP-3 directly activated proMMP-1 by cleaving the Gln80-Phe81 bond, but this reaction was extremely slow, indicating that the Gln80-Phe81 bond is not readily available to MMP-3 in the native proMMP-1 molecule. ProMMP-2 (M(r) = 72,000) was activated only by APMA, but not by proteinases. The activation by APMA was rapid and generated an active MMP-2 of M(r) 68,000, but the enzymic activity declined rapidly after activation by autolysis. The NH2-terminal sequence analysis of active MMP-2 indicated that the Asn80-Tyr81 bond was cleaved upon APMA treatment. In contrast, proMMP-3 (M(r) = 57,000) was activated by a variety of proteinases with different specificities. The initial attacks of these proteinases are on a stretch of highly charged groups at the position 34-39 in the propeptide.(ABSTRACT TRUNCATED AT 250 WORDS)
Human melanoma cells secrete a 21 kDa protein which binds with 1:1 molar stoichiometry to the matrix metalloproteinase type IV collagenase proenzyme (70 kDa gelatinase) secreted by the same cells. We have purified this binding protein and determined its complete primary structure by directly sequencing overlapping peptide fragments which span the entire protein. We refer to this protein as CSC-21K based on the amino-terminal amino acids CSC and the apparent molecular weight of 21,000 daltons on gel electrophoresis. The amino acid sequence of CSC-21K demonstrates that this protein shares significant homology with human TIMP (tissue inhibitor of metalloproteinase), including conservation of the positions of the twelve cysteine residues and three of four tryptophan residues. The identification of CSC-21K now indicates that a family of TIMP-related proteins exists. Individual members of this family may possess selective affinities for different members of the matrix metalloproteinase family. Based on its sequence homology to TIMP and ability to inhibit type IV collagenolysis we propose the name TIMP-2 for this inhibitor. TIMP-2 produced by tumor cells can also be considered as an onco-suppressor gene product, because it could play an important role in regulating the metalloproteinases involved in tumor invasion and angiogenesis.
Recombinant human tissue inhibitor of metalloproteinases (TIMP) forms complexes with high-Mr active recombinant stromelysin that are stable over long periods under physiological conditions. TIMP-stromelysin complexes could be dissociated in the presence of EDTA at pH 3, releasing free TIMP and destroying stromelysin activity. The dissociated TIMP was apparently unmodified, in contrast with other known protein inhibitors of metalloproteinases and many classes of serine-proteinase inhibitor, which are slowly cleaved.
On purification, human fibroblast collagenase breaks down into two major forms (Mr22,000 and Mr 27,000) and one minor form (Mr 25,000). The most likely mechanism is autolysis, although the presence of contaminating enzymes cannot be excluded. From N-terminal sequencing studies, the 22,000-Mr fragment contains the active site; differential binding to concanavalin A shows the 25,000-Mr fragment is a glycosylated form of the 22,000-Mr fragment. These low-Mr forms can be separated by Zn2+-chelate chromatography. An activity profile of this column, combined with data from substrate gels, indicates no activity against collagen in the 22,000-Mr and 25,000-Mr forms, but rather, activity casein and gelatin. The 27,000-Mr form has no activity. The 22,000/25,000-Mr form can act as an activator for collagenase in a similar way to that reported for stromelysin. The activity of the 22,000/25,000-Mr form is not inhibited by the tissue inhibitor of metalloproteinases (TIMP). The 27,000-Mr C-terminal part of the collagenase molecule therefore appears to be important in maintaining the substrate-specificity of the enzyme, and also plays a role in the binding of TIMP.
A genetic approach to define the role of collagenase in physiological and pathological bone remodeling is to identify spontaneous mutations in the collagenase gene which alter enzymatic activity. Alternatively it is possible, though site-directed mutagenesis, to alter genes encoding critical amino acid sequences in the collagen substrate, in a manner analogous to the successful development of animal models for osteogenesis imperfecta. We have thus utilized this approach to alter the Col1a1 gene to encode amino acid substitutions in sequences around the known collagenase cleavage site (glycine-isoleucine at positions 775-776) in type I collagen, and transfect these genes into homozygous Mov-13 fibroblasts, in which the endogenous Col1a1 gene is inactive. Nonconservative substitutions of proline for isoleucine at the P1' site and double substitutions of proline for glutamine (P2) and alanine (P2') resulted in type I collagen resistant to hydrolysis by collagenase. Furthermore, in normal fibroblasts transfected with a mutant Col1a1 gene encoding collagenase resistance in which an additional methionine substitution at position 776 provided a marker for the mutant protein, mutant and wild type triple helical molecules were synthesized and secreted as heterotrimers. A single mutant alpha 1(I) chain did not prevent cleavage of the wild type alpha 1(I) chain but it is likely that the uncleaved alpha 1(I) chain would prevent dissociation of the triple helical fragments containing the other cleaved chains. Introduction of these genes into transgenic mice should result in abnormal phenotypes characterized by altered connective tissue remodeling.
We have examined the conditions for dissolution by live cells of an extracellular matrix composed of reconstituted type I collagen fibrils, using three different cell types which express varying constitutive or inducible levels of procollagenase and collagenase inhibitor. The two major conclusions from these studies were that (i) expression of collagenase is a necessary but not sufficient requirement for dissolution of the collagen fibrils and that (ii) activation of procollagenase is a rate-limiting step. Cells which secreted high levels of procollagenase dissolved collagen fibrils only to the extent that they were able to activate the enzyme. Cells which also expressed inhibitor failed to activate procollagenase in the culture medium and did not dissolve the collagen fibrils unless procollagenase-activation was assisted by exogenous proteinase activity. Cells that did not express inhibitor ultimately did activate procollagenase but the process was slow and incomplete. Introduction of exogenous proteinase activity either in the form of plasminogen, plasmin, or trypsin stimulated collagen breakdown by several fold. Analysis of the culture medium sampled from such cultures showed that the stimulating effect of exogenous proteinases could be ascribed to three separate, but synergistic events: elevated expression of procollagenase, conversion of procollagenase to active form and inactivation of collagenase inhibitor. Two lines of evidence suggested that the dissolution of collagen fibrils in these cultures was mediated by a collagenase-dependent pathway: (i) the rate of dissolution closely mirrored the level of expression of collagenase and (ii) the process was blocked by inhibitory collagenase-specific antibodies.
Bacterial collagenase from aerobic non-pathogenic Vibrio alginolyticus chemovar iophagus ("Achromobacter" collagenase, EC 3.4.24.08) is an inducible extracellular metallo-proteinase. Production of Vibrio collagenase is induced specifically by collagen or by its macromolecular fragments. On the cell surface is expressed a specific receptor recognizing collagen structure. The study of natural inducers led to synthetic peptides with inducing properties. Vibrio collagenase cleaves collagen helical chains preferentially at 3/4 from the N-terminal. Its specific activity on synthetic substrate, 180,000 ukat/mg, represents the highest value for known collagenases. Its specificity differs from that of Clostridium: The enzyme cleaves preferentially sequences with Gly or Ala in position P'1 and Pro in position P2 or P'2. Highly specific cleavages were obtained in beta-casein, prolactin, myosin, adenylate kinase and fibronectin. Autolysis yields partially degraded forms still active on native collagen and peptide substrate. The determination of the sequence of Vibrio collagenase is nearly achieved; the enzyme was not yet obtained in crystalline form. On basis of the already known sequence and structure of Hypoderma collagenase (EC 3.4.21.49), a hypothesis is advanced on the character of collagen binding site loops. Vibrio collagenase can be produced in kilogram quantities at low cost. It was found highly efficient in debridement of necrotic burns, ulcers and decubitus.
Molecular and biochemical studies have made substantial contributions to the understanding of metalloproteinases (MPs) and their natural inhibitor TIMP (tissue inhibitor of metalloproteinases) but knowledge of their specific roles in tissue breakdown in vivo is still meagre. A major problem is that there are few techniques available that can detect small amounts of these entities at the sites of resorption. One approach to this problem is to prepare specific polyclonal antisera for use in immunolocalization studies on cells and tissues ex vivo. Another is to develop model systems of rapid matrix destruction. Examples of these techniques are presented and discussed in relation to other studies. In many situations unique patterns of synthesis have been observed, consistent with specific roles for the individual MPs and biochemical data. Active collagenase can be localized to extracellular components in tissues where rapid destruction is taking place. Thus both approaches are proving invaluable in defining the roles of MPs and TIMP in normal and pathological situations.
Human mononuclear phagocytes have the capacity to participate directly in extracellular matrix turnover via the secretion of neutral proteinases. These neutral proteinases include the serine proteinases, elastase and cathepsin G and the metalloproteinases, interstitial collagenase, 92 kD type IV collagenase, 72 kD type IV collagenase and stromelysin. Mononuclear phagocytes also produce the counter-regulatory metalloproteinase inhibitor, TIMP (tissue inhibitor of metalloproteinases). We have studied the capacity of normal human mononuclear phagocytes and of the human monocytic tumor line U937 to elaborate proteinases and inhibitors. The serine proteinases, elastase and cathepsin G, are present only at the earliest stages of mononuclear phagocyte differentiation (U937 cells in the basal state, freshly isolated peripheral blood monocytes) and are stored within intracellular granules. As human mononuclear phagocytes differentiate (U937 cells exposed to phorbol esters, human monocytes cultured in vitro), the cellular content of these serine proteinases declines rapidly. Accompanying the acquisition of a more differentiated state, the ability for regulated secretion of the neutral metalloproteinases is attained. This capacity is acquired in a sequential manner, with secretion of the 92 kD type IV collagenase observed at earlier states of differentiation while release of stromelysin requires a fully differentiated and LPS (lipopolysaccharide)-stimulated alveolar macrophage. Interstitial collagenase and 72 kD type IV collagenase are synthesized at intermediate stages of differentiation. In comparison to human fibroblasts, human mononuclear phagocytes produce approximately 10-30% of the interstitial collagenase, 10% of the stromelysin and 1-2% of the 72 kD type IV collagenase on a per cell basis. Synthesis of the 92 kD type IV collagenase is restricted to the inflammatory cell (but also occurs in neutrophils and keratinocytes).
Both lysosomal cysteine-proteinases and collagenase appear to be necessary for the resorption of actively growing, immature woven bone, but their relative roles are not yet clearly elucidated. The present evidence indicates that, during bone resorption, the osteoclast first solubilizes the mineral by a secretion of acid and then removes the exposed demineralized collagen by the action of secreted lysosomal collagenolytic cysteine-proteinases. Collagenase in bone seems to be mainly a product of osteoblasts and related cells, not osteoclasts. Its role could be limited in the removal of any non-mineralized collagen layers which could be covering mineralized bone surfaces and which seem to prevent the activation of osteoclasts and thus their action; such a "shield" of unmineralized osteoid is well-established at the surface of actively growing woven bone, although not on the resorbing surfaces of mature lamellar bone. Moreover, some osteoblast-derived procollagenase is stored in the mineralized bone matrix from which it can be released by demineralization. It is therefore possible that it may also contribute to the degradation of demineralized bone collagen once it has been released and activated by lysosomal cysteine-proteinases under the osteoclast.
