Alanine scanning mutagenesis and functional analysis of the fibronectin-like collagen-binding domain from human 92-kDa type IV collagenase.

The human 72-kDa (CLG4A) and 92-kDa (CLG4B) type IV collagenases contain a domain consisting of three contiguous copies of the fibronectin (FN)-derived type II homology unit (T2HU), T2HU-1, T2HU-2, and T2HU-3. To investigate the functional role of this domain, we have constructed plasmids expressing beta-galactosidase fusion proteins with one or more of the CLG4B-derived T2HU. The gelatin binding assays demonstrate that a single copy of T2HU-2 renders beta-galactosidase capable of binding gelatin. The three repeats, however, differ dramatically in their capacity to bind gelatin, with T2HU-1 and T2HU-3 having significantly less binding activity than T2HU-2. Using alanine scanning mutagenesis we have defined the amino acid residues (Arg307, Asp309, Asn319, Tyr320, Asp323) that are critical for gelatin binding of T2HU-2. The low gelatin binding of T2HU-1 compared to T2HU-2 was traced to the non-conserved residues Ala228-Ala and Leu253-Pro. The results suggest that the gelatin binding of the type IV collagenase proenzyme is mediated by the FN-like domain, although the presence of another gelatin-binding site cannot be excluded. The FN domain-mediated binding, however, is not a rate-limiting step in the hydrolysis of gelatin by the enzyme.     

Purification and characterization of human 72-kDa gelatinase (type IV collagenase). Use of immunolocalisation to demonstrate the non-coordinate regulation of the 72-kDa and 95-kDa gelatinases by human fibroblasts.

Human gingival fibroblast gelatinase (type IV collagenase) has been purified to homogeneity using a combination of ion exchange chromatography, gel filtration and affinity chromatography. The purified proenzyme electrophoresed under reducing conditions as a single band of 72 kDa which could be activated to a species of 65 kDa. Gelatinase was activated by organomercurials by a process apparently initiated by a conformational change and involving self-cleavage. It was not activated by trypsin or plasmin unlike the other family members, collagenase and stromelysin. Gelatinase otherwise exhibited properties typical of the metalloproteinases: it was inhibited by metal chelating agents and by the specific inhibitor TIMP (tissue inhibitor of metalloproteinases). Its major substrate was shown to be denatured collagen although it was also able to degrade native type IV and V collagens. A polyclonal antibody was raised in a sheep using the purified enzyme as antigen. The antiserum recognised and specifically inhibited the 72-kDa gelatinase but not a 95-kDa gelatinase from pig leukocytes. It was used in immunolocalisation studies on human fibroblasts to investigate the regulation of the production of the two Mr forms of gelatinase. These studies clearly demonstrate that human fibroblasts constitutively synthesize and secrete 72-kDa gelatinase but that 95-kDa gelatinase was inducible by agents such as cytokines. The significance of these results in relation to the likely in vivo r?le of gelatinases is discussed.     

Molecular cloning of murine 72-kDa type IV collagenase and its expression during mouse development.

We report the isolation of a cDNA clone providing the first and complete sequence of mouse 72-kDa type IV collagenase. The clone contains 2800 nucleotides with a 1986-nucleotide open reading frame coding for 662 amino acids. The amino acid sequence includes a 29-residue signal peptide, an 80-residue propeptide, and a 553-residue enzyme proper. The sequence identity between the mouse and human enzymes is 96% with all cysteine residues conserved. The carboxyl-terminal domain of the mouse enzyme contains two more residues than the human enzyme. Northern hybridization analysis revealed considerable expression of the enzyme gene in newborn mouse lung, heart, kidney, and psoas muscle tissues, whereas only weak or no signals were observed in liver, spleen, and brain. Expression of the gene was substantially reduced in the same tissues of 3-month-old mice. In situ hybridization analysis of 72-kDa type IV collagenase expression in 10-15-day-old mouse embryos showed that the gene was intensely expressed in mesenchymal cells. Brain and surface ectoderm were completely negative. The epithelial tissue component of developing organs was negative with the exception of salivary gland. Although the expression varied somewhat between different mesenchymal tissues, no temporal or spatial changes could be associated with the advancement of epithelial branching morphogenesis. These findings together with our previous data on the expression of 72-kDa type IV collagenase in human tumors indicate that this enzyme has some very specific roles both in the physiological and pathological degradation of extracellular matrix. Furthermore, it has become clear that the closely related 92-kDa type IV collagenase differs completely with respect to expression pattern as well as gene regulation. The mouse cDNA clones reported in this study may provide important tools unraveling the actual roles of these enzymes in vivo.     

Interleukin 4 inhibition of prostaglandin E2 synthesis blocks interstitial collagenase and 92-kDa type IV collagenase/gelatinase production by human monocytes.

Activation of human monocytes results in the production of interstitial collagenase through a prostaglandin E2 (PGE2)-cAMP-dependent pathway. Inasmuch as interleukin 4 (IL-4) has been shown to inhibit PGE2 synthesis by monocytes, we examined the effect of IL-4 on the production of human monocyte interstitial collagenase. Additionally, we also assessed the effect of IL-4 on the production of 92-kDa type IV collagenase/gelatinase and tissue inhibitor of metalloproteinase-1 (TIMP-1) by monocytes. The inhibition of PGE2 synthesis by IL-4 resulted in decreased interstitial collagenase protein and activity that could be restored by exogenous PGE2 or dibutyryl cyclic AMP (Bt2cAMP). IL-4 also suppressed ConA-stimulated 92-kDa type IV collagenase/gelatinase protein and zymogram enzyme activity that could be reversed by exogenous PGE2 or Bt2cAMP. Moreover, indomethacin suppressed the ConA-induced production of 92-kDa type IV collagenase/gelatinase. These data demonstrate that, like monocyte interstitial collagenase, the conA-inducible monocyte 92-kDa type IV collagenase/gelatinase is regulated through a PGE2-mediated cAMP-dependent pathway. In contrast to ConA stimulation, unstimulated monocytes released low levels of 92-kDa type IV collagenase/gelatinase that were not affected by IL-4, PGE2, or Bt2cAMP, indicating that basal production of this enzyme is PGE2-cAMP independent. IL-4 inhibition of both collagenases was not a result of increased TIMP expression since Western analysis of 28.5-kDa TIMP-1 revealed that IL-4 did not alter the increased TIMP-1 protein in response to ConA. These data indicate that IL-4 may function in natural host regulation of connective tissue damage by monocytes.     

92-kD type IV collagenase mediates invasion of human cytotrophoblasts

The specialized interaction between embryonic and maternal tissues is unique to mammalian development. This interaction begins with invasion of the uterus by the first differentiated embryonic cells, the trophoblasts, and culminates in formation of the placenta. The transient tumor-like behavior of cytotrophoblasts, which peaks early in pregnancy, is developmentally regulated. Likewise, in culture only early-gestation human cytotrophoblasts invade a basement membrane-like substrate. These invasive cells synthesize both metalloproteinases and urokinase-type plasminogen activator. Metalloproteinase inhibitors and a function-perturbing antibody specific for the 92-kD type IV collagen-degrading metalloproteinase completely inhibited cytotrophoblast invasion, whereas inhibitors of the plasminogen activator system had only a partial (20-40%) inhibitory effect. We conclude that the 92-kD type IV collagenase is critical for cytotrophoblast invasion.     

Human 92- and 72-kilodalton type IV collagenases are elastases.

Elastin is critical to the structural integrity of a variety of connective tissues. Only a select group of enzymes has thus far been identified capable of cleaving insoluble elastin. Recently, we observed that human alveolar macrophages secrete elastase activity that is largely inhibited by the tissue inhibitor of metalloproteinases (TIMP). This finding suggested that one or more of the metalloproteinases released by alveolar macrophages has elastase activity. Accordingly, we tested pure human interstitial collagenase, stromelysin, 92-kDa type IV collagenase, and 72-kDa type IV collagenase for elastolytic activity using kappa-elastin zymography and insoluble 3H-labeled elastin. The 92- and 72-kDa type IV collagenases were found to be elastolytic in both assay systems. A recombinant preparation of 92-kDa type IV collagenase with gelatinolytic activity was also found to be elastolytic. Organomercurial activation was essential to detect elastolytic activity of the native 92- and 72-kDa type IV collagenases and enhanced the elastase activity of the recombinant 92-kDa enzyme. On a molar basis the recombinant 92-kDa type IV collagenase was approximately 30% as active as human leukocyte elastase in solubilizing 3H-labeled elastin. Exogenously added TIMP in significant molar excess abolished the elastase activity of the 92- and 72-kDa type IV collagenases. Stromelysin and interstitial collagenase showed no significant elastolytic activity, although both were catalytically active against susceptible substrates. Conditioned media from cultures of human mononuclear phagocytes containing the 92-kDa enzyme produced a distinct zone of lysis in the kappa-elastin zymograms at this molecular mass. These results definitively extend the spectrum of human proteinases with elastolytic activity to metalloproteinases and suggest the enzymatic basis for elastase activity observed with certain cell types such as human alveolar macrophages.     

Differential susceptibility of type X collagen to cleavage by two mammalian interstitial collagenases and 72-kDa type IV collagenase

We have studied the degradation of type X collagen by human skin fibroblast and rat uterus interstitial collagenases and human 72-kDa type IV collagenase. The interstitial collagenases attacked the native type X helix at two loci, cleaving residues Gly92-Leu93 and Gly420-Ile421, both scissions involving Gly-X bonds of Gly-X-Y-Z-A sequences. However, the human and rat interstitial enzymes displayed an opposite and substantial selectivity for each of these potential sites, with the uterine enzyme catalyzing the Gly420-Ile421 cleavage almost 20-fold faster than the Gly92-Leu93 locus. Values for enzyme-substrate affinity were approximately 1 microM indistinguishable from the corresponding Km values against type I collagen. Interestingly, in attacking type X collagen, both enzymes manifested kinetic properties intermediate between those characterizing the degradation of native and denatured collagen substrates. Thus, energy dependence of reaction velocity revealed a value of EA of 45 kcal, typical of native interstitial collagen substrates. However, the substitution of D2O for H2O in solvent buffer failed to slow type X collagenolysis significantly (kH/kD = 1.1), in contrast to the 50-70% slowing (kH/kD = 2-3) observed with native interstitial collagens. Since this lack of deuterium isotope effect is characteristic of interstitial collagenase cleavage of denatured collagens, we investigated the capacity of another metalloproteinase with substantial gelatinolytic activity, 72-kDa type IV collagenase, to degrade type X collagen. The 72-kDa type IV collagenase cleaved type X collagen at both 25 and 37 degrees C, and at loci in close proximity to those attacked by the interstitial enzymes. No further cleavages were observed at either temperature with type IV collagenase, and although values for kcat were not determined (due to associated tissue inhibitor of metalloproteinases-2), catalytic rates appeared to be substantial in comparison to the interstitial enzymes. In contrast, type X collagen was completely resistant to proteolysis by stromelysin. Type X collagen thus appears to be highly unusual in its susceptibility to degradation by both interstitial collagenase and another member of the metalloproteinase gene family.     

SV40-transformed human lung fibroblasts secrete a 92-kDa type IV collagenase which is identical to that secreted by normal human macrophages

We have reported that SV40-transformed human lung fibroblasts secrete a 92-kDa metalloprotease which is not detectable in the parental cell line IMR-90. We now present the complete structure of this enzyme along with the evidence that it is identical to the 92-kDa metalloprotease secreted by normal human alveolar macrophages, phorbol ester-differentiated monocytic leukemia U937 cells, fibrosarcoma HT1080 cells, and cultured human keratinocytes. A similar, perhaps identical, enzyme can be released by polymorphonuclear cells. The preproenzyme is synthesized as a polypeptide of predicted Mr 78,426 containing a 19 amino-acid-long signal peptide and secreted as a single 92,000 glycosylated proenzyme. The purified proenzyme complexes noncovalently with the tissue inhibitor of metalloproteases (TIMP) and can be activated by organomercurials. Activation with phenylmercuric chloride results in removal of 73 amino acids from the NH2 terminus of the proenzyme, yielding an active form capable of digesting native types IV and V collagen. The in vitro substrate specificity of the enzyme using these substrates was indistinguishable from that of the 72-kDa type IV collagenase. The 92-kDa type IV collagenase consists of five domains; the amino-terminal and zinc-binding domains shared by all members of the secreted metalloprotease gene family, the collagen-binding fibronectin-like domain also present in the 72-kDa type IV collagenase, a carboxyl-terminal hemopexin-like domain shared by all known enzymes of this family with the exception of PUMP-1, and a unique 54-amino-acid-long proline-rich domain homologous to the alpha 2 chain of type V collagen.     

Domain structure of human 72-kDa gelatinase/type IV collagenase. Characterization of proteolytic activity and identification of the tissue inhibitor of metalloproteinase-2 (TIMP-2) binding regions.

The 72-kDa gelatinase/type IV collagenase, a metalloproteinase thought to play a role in metastasis and in angiogenesis, forms a noncovalent stoichiometric complex with the tissue inhibitor of metalloproteinase-2 (TIMP-2), a potent inhibitor of enzyme activity. To define the regions of the 72-kDa gelatinase responsible for TIMP-2 binding, a series of NH2- and COOH-terminal deletions of the enzyme were constructed using the polymerase chain reaction technique. The full-length and the truncated enzymes were expressed in a recombinant vaccinia virus mammalian cell expression system (Vac/T7). Two truncated enzymes ending at residues 425 (delta 426-631) and 454 (delta 455-631) were purified. Like the full-length recombinant 72-kDa gelatinase, both COOH-terminally truncated enzymes were activated with organomercurial and digested gelatin and native collagen type IV. In contrast to the full-length enzyme, delta 426-631 and delta 455-631 enzymes were less sensitive to TIMP-2 inhibition requiring 10 mol of TIMP-2/mol of enzyme to achieve maximal inhibition of enzymatic activity. The activated but not the latent forms of the delta 426-631 and delta 455-631 proteins formed a complex with TIMP-2 only when excess molar concentrations of inhibitor were used. We also expressed the 205-amino acid COOH-terminal fragment, delta 1-426, and found that it binds TIMP-2. In addition, a truncated version of the 72-kDa gelatinase lacking the NH2-terminal 78 amino acids (delta 1-78) of the proenzyme retained the ability to bind TIMP-2. These studies demonstrate that 72-kDa gelatinases lacking the COOH-terminal domain retain full enzymatic activity but acquire a reduced sensitivity to TIMP-2 inhibition. These data suggest that both the active site and the COOH-terminal tail of the 72-kDa gelatinase independently and cooperatively participate in TIMP-2 binding.     

Activation of the 92 kDa type IV collagenase by tissue kallikrein

Type IV collagenases are secreted as latent 92 and 72 kDa proenzymes which are then activated extracellularly. The mechanisms by which they are activated in vivo are not clear. We have studied the activation of porcine endothelial cell type IV collagenases by tissue and plasma kallikrein, and found that tissue kallikrein was a very efficient activator of the 92 kDa type IV collagenase. Enzyme cleavage was observed at concentrations of tissue kallikrein as low as 0.1 μg/ml. Plasma kallikrein had no effect. By comparison, plasmin, which has been proposed to be the physiological activator of interstitial collagenase and stromelysin, and elastase were much less effective, and high concentrations (plasmin at 100–200 μg/ml and elastase at 20 μg/ml) were required to cause only a limited cleavage which was not associated with an increase in activity, as observed by the gelatin-gel lysis assay. In addition tissue kallikrein was found by immunohistochemistry to be present in the extracellular matrix of the intima of porcine aortic vessel wall. These findings suggest that tissue kallikrein can be a potential activator of the 92 kDa type IV collagenase in vivo. © 1993 Wiley-Liss, Inc.     

Matrix metalloproteinase 9 (92-kDa gelatinase/type IV collagenase) from HT 1080 human fibrosarcoma cells. Purification and activation of the precursor and enzymic properties.

Matrix metalloproteinase 9 (MMP-9) has been purified as an inactive zymogen of M(r) 92,000 (proMMP-9) from the culture medium of HT 1080 human fibrosarcoma cells. The NH2-terminal sequence of proMMP-9 is Ala-Pro-Arg-Gln-Arg-Gln-Ser-Thr-Leu-Val-Leu-Phe-Pro, which is identical to that of the 92-kDa type IV collagenase/gelatinase. The zymogen can be activated by 4-aminophenylmercuric acetate, yielding an intermediate form of M(r) 83,000 and an active species of M(r) 67,000, the second of which has a new NH2 terminus of Met-Arg-Thr-Pro-Arg-(Cys)-Gly-Val-Pro-Asp-Leu-Gly-Arg-Phe-Gln-Thr- Phe-Glu. Immunoblot analyses demonstrate that this activation process is achieved by sequential processing of both NH2- and COOH-terminal peptides. TIMP-1 complexed with proMMP-9 inhibits the conversion of the intermediate form to the active species of M(r) 67,000. The proenzyme is fully activated by cathepsin G, trypsin, alpha-chymotrypsin, and MMP-3 (stromelysin 1) but not by plasmin, leukocyte elastase, plasma kallikrein, thrombin, or MMP-1 (tissue collagenase). During the activation by MMP-3, proMMP-9 is converted to an active species of M(r) 64,000 that lacks both NH2- and COOH-terminal peptides. In addition, HOCl partially activates the zymogen by reacting with an intermediate species of M(r) 83,000. The enzyme degrades type I gelatin rapidly and also cleaves native collagens including alpha 2 chain of type I collagen, collagen types III, IV, and V at undenaturing temperatures. These results indicate that MMP-9 has different activation mechanisms and substrate specificity from those of MMP-2 (72-kDa gelatinase/type IV collagenase).     

Interaction of 92-kDa type IV collagenase with the tissue inhibitor of metalloproteinases prevents dimerization, complex formation with interstitial collagenase, and activation of the proenzyme with stromelysin.

Secreted metalloproteases initiating proteolytic degradation of collagens and proteoglycans play a critical role in remodeling of the connective tissue. Activation of the secreted proenzymes and interaction with their specific inhibitors TIMP and TIMP-2 are responsible for regulation of enzyme activity in extracellular space. We have previously demonstrated that 92- and 72-kDa Type IV procollagenases, in contrast to interstitial collagenase (ClI), form specific complexes with TIMP and the related inhibitor TIMP-2, respectively. The physiologic significance of the proenzyme-inhibitor complex and the mechanism of activation of Type IV collagenases remained unclear. Here, we demonstrate that in the absence of TIMP, 92-kDa Type IV procollagenase (92T4Cl) can form a covalent homodimer and a novel complex with ClI. In the presence of TIMP, the formation of a 92T4Cl proenzyme complex with TIMP prevents dimerization, formation of the complex with ClI, and activation of the 92T4Cl proenzyme by stromelysin, a related metalloprotease. The proenzyme homodimer is unable to form a complex with TIMP. All TIMP-free forms of the proenzyme can be activated by stromelysin. The 92T4Cl-ClI complex can be activated to yield a complex active against both gelatin and fibrillar Type I collagen, suggesting a mechanism for cooperative action of two enzymes in reducing collagen fibrils to small peptides under physiologic conditions.     

Preferential inhibition of 72- and 92-kDa gelatinases by tissue inhibitor of metalloproteinases-2.

Transformed human fibroblasts secrete two structurally and functionally related inhibitors of matrix metalloproteinases, tissue inhibitor of metalloproteinases (TIMP) 1 and 2. In assays measuring the relative inhibitory capability of TIMP-1 and TIMP-2 against autoactivated 72-kDa gelatinase, which consists of two major active peptides and several inactive fragments, TIMP-2 was more effective than TIMP-1. The isolated 42.5-kDa active fragment that formed as a result of the autoactivation of 72-kDa gelatinase showed the greatest preference for TIMP-2; at half-maximal inhibition, TIMP-2 was greater than 10-fold more effective than TIMP-1. TIMP-2 was also greater than 2-fold more effective than TIMP-1 at inhibiting 72-kDa gelatinase-TIMP-2 complexes activated with 4-aminophenylmercuric acetate, and greater than 7-fold more effective than TIMP-1 at inhibiting 92-kDa gelatinase activated with 4-aminophenylmercuric acetate. Furthermore, these active gelatinases preferentially bound 125I-TIMP-2 when incubated with equal amounts of radiolabeled TIMP-1 and TIMP-2. The ratios of 125I-TIMP-2/125I-TIMP-1 binding to 92-kDa gelatinase, autoactivated 72-kDa gelatinase, and 42.5-kDa fragment were 4.4, 10, and 33, respectively. On the other hand, interstitial collagenase was inhibited by TIMP-1 greater than 2-fold more effectively than TIMP-2 in assays measuring cleavage of loose collagen fibrils.     

Studies on the ability of 65-kDa and 92-kDa tumor cell gelatinases to degrade type IV collagen

Two major gelatinolytic metalloproteinases (gelatinases) of 65 kDa and 92 kDa were purified from a tumor cell line. Analysis of collagen degradation showed that native full-length Engelbreth-Holm-Swarm (EHS) type IV collagen was not cleaved by the purified gelatinases under conditions where native pepsin-extracted human placental type IV and V collagen and heat-denatured collagens were markedly degraded. However, EHS type IV collagen degradation was noted at 37 degrees C, i.e., under conditions that would favor denaturation of the collagen molecule in solution. The pattern of degradation of human placental type IV and V collagen appeared similar for both gelatinases. Zymogram analysis of gelatinase activity in the absence of sodium dodecyl sulfate (SDS) (to eliminate possible SDS-mediated denaturation of type IV collagen) confirmed the inability of 65 and 92-kDa gelatinases to degrade native full-length EHS type IV collagen. Under the same conditions and in SDS-polyacrylamide gel electrophoresis zymograms the gelatinases degraded pepsin-predigested EHS type IV collagen and pepsin-extracted human placental type IV collagen. These data suggest that the 65- and 92-kDa tumor cell gelatinases are not true type IV collagenases. Their ability to degrade pepsin-solubilized, or denatured, type IV collagen suggests a specificity for telopeptide precleaved or conformationally altered forms of this molecule.     

Induction and stimulation of 92-kDa gelatinase/type IV collagenase production in osteosarcoma and fibrosarcoma cell lines by tumor necrosis factor alpha

Production of a 92-kDa gelatinase/type IV collagenase and tissue inhibitor of metalloproteinases (TIMP) was investigated with human sarcoma cell lines. Among the cytokines and growth factors examined, only human recombinant tumor necrosis factor alpha (TNF alpha) induced and stimulated the proteinase with concomitant increase in TIMP expression, but matrix metalloproteinase 2 (72-kDa gelatinase/type IV collagenase) expression was unchanged. These data suggest that gene expression of the two metalloproteinases is regulated in a different fashion and TNF alpha may be important to allow cancer cells to be more invasive and metastatic.