A specific inhibitor of vertebrate collagenase produced by human skin fibroblasts.

Normal human skin fibroblasts which synthesize procollagenase were shown also to produce a specific inhibitor of the active form of this proenzyme. The inhibitor was derived from the flbroblasts themselves, as demonstrated by the following criteria: 1) 3H-labeled inhibitor was produced when flbroblasts were exposed to a 3H-amino-acid mixture in serum-free culture medium, 2) fetal calf serum, when processed through the basic purification steps employed for the inhibitor, displayed no collagenase inhibitory activity. Inhibitor was purified extensively from serum-free medium by a combination of cation exchange and gel filtration chromatography. Spectrophotometric scanning of polyacrylamide gels indicated that the inhibitor was greater than 95% pure. It had an apparent molecular weight of 31,000 as determined by sodium dodecyl sulfate-gel electrophoresis, and was remarkably heatstable, retaining more than 60% of its activity after 20 min at 90°C. Inhibition was stoichiometric, a 1:l molar ratio of inhibitor to enzyme being required for complete inhibition of collagenase activity. Attempts to demonstrate an enzyme-inhibitor complex were not successful; in fact, enzyme and inhibitor behaved independently of each other in a variety of chromatographic systems. Tight binding between active enzyme and inhibitor occurred only in the presence of the collagen substrate. It is therefore suggested that the inactive species is the ternary complex E.1.S. Human fibroblast inhibitor was effective against all vertebrate collagenases tested. Noncollagenolytic proteases and collagenases of nonvertebrate origin were not inhibited. Procollagenase, which does not bind to collagen, also failed to bind the inhibitor. However, inhibitor itself was capable of binding to collagen.

Normal human skin fibroblasts which synthesize procollagenase were shown also to produce a specific inhibitor of the active form of this proenzyme. The inhibitor was derived from the flbroblasts themselves, as demonstrated by the following criteria: 1) 3H-labeled inhibitor was produced when flbroblasts were exposed to a 3H-amino-acid mixture in serum-free culture medium, 2) fetal calf serum, when processed through the basic purification steps employed for the inhibitor, displayed no collagenase inhibitory activity. Inhibitor was purified extensively from serum-free medium by a combination of cation exchange and gel filtration chromatography. Spectrophotometric scanning of polyacrylamide gels indicated that the inhibitor was greater than 95% pure. It had an apparent molecular weight of 31,000 as determined by sodium dodecyl sulfate-gel electrophoresis, and was remarkably heatstable, retaining more than 60% of its activity after 20 min at 90°C. Inhibition was stoichiometric, a 1:l molar ratio of inhibitor to enzyme being required for complete inhibition of collagenase activity. Attempts to demonstrate an enzyme-inhibitor complex were not successful; in fact, enzyme and inhibitor behaved independently of each other in a variety of chromatographic systems. Tight binding between active enzyme and inhibitor occurred only in the presence of the collagen substrate.
It is therefore suggested that the inactive species is the ternary complex E.1.S. Human fibroblast inhibitor was effective against all vertebrate collagenases tested. Noncollagenolytic proteases and collagenases of nonvertebrate origin were not inhibited.
Procollagenase, which does not bind to collagen, also failed to bind the inhibitor. However, inhibitor itself was capable of binding to collagen.
The vertebrate collagenases form a class of enzymes capable of initiating the specific degradation of native collagen in the animal organism. Although specific collagenolytic enzymes have now been identified in numerous tissues from a wide variety of species, the nature of the processes which regulate vertebrate collagen degradation have not been clearly defined. The existence of naturally occurring inhibitors of collagenase has been recognized for some time, and it has been speculated * This work was supported by United States Public Health Service Grants AM 12129, HD 05291, AM 19537, TO-AM 07284, GMO-2016, and by a grant from the National Foundation-March of Dimes. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Recipient of Research Career Development Award 5-K04-AM00077 from the National Institutes of Health. that such inhibitors are involved in the regulation of collagenase activity in ho.
The major collagenase inhibitors studied thus far have been identified in serum, and it is unknown whether they can exist and function within the interstices of an organized connective tissue. ocz-Macroglobulin, the principal serum anti-collagenase, is a potent inhibitor of all vertebrate collagenases investigated to date (1,2). Yet its large molecular weight (780,000) and irreversible mechanism of inhibition (2) raise doubt concerning its actual in uiuo role in the regulation of collagen degradation at the tissue level. al-Antitrypsin, although a major serum protease inhibitor, is relatively ineffective as an anticollagenase (3). Furthermore, both an-macroglobulin and alantitrypsin are nonspecific antiproteases, since they are capable of inhibiting a broad range of proteolytic enzymes of varying specificity and chemical characteristics.
Recently, Woolley et al. have identified a PI-serum protein which appears to display the properties of a specific collagenase inhibitor (4, 5).
Tissue-derived collagenase inhibitors are more likely to be of functional physiologic significance in the regulation of connective tissue structure than inhibitors found principally in serum. Thus far, several such tissue-derived inhibitors have been reported, including an inhibitor extracted from rabbit tumor (6) and several small proteins of molecular weight less than 15,000 (7-12).
In this laboratory, cultures of normal human skin fibroblasts have been employed for the production, purification, and characterization of human procollagenase. This proenzyme can be converted to fully active enzyme by either incubation with trypsin, and resultant loss of a lO,OOO-dalton peptide, or via an autoactivation process without detectable change in molecular weight (13,14). In addition to synthesizing a collagenase zymogen, the human fibroblasts were found to produce an inhibitor of this enzyme simultaneously (15). The present report describes studies on the purification, properties, and mechanism of action of human skin fibroblast collagenase inhibitor.

MATERIALS AND METHODS
Reagents-Acrylamide and bisacrylamide were purchased from Eastman. Sodium dodecyl sulfate, 99% pure, was obtained from Gallard-Schlesinger. Tris base, bovine pancreatic trypsin, bovine serum albumin, ovalbumin, and soybean trypsin inhibitor were procured Co.) and put through several cycles of serumfree medium for 48 to 72 h, as described previously by Bauer et al. (15). The harvested serum-free medium was made 0.05 M with Tris-HCl, pH 7.5, and concentrated lo-fold by vacuum dialysis using a hollow fiber device (MDA Scientific). The resulting concentrated medium was stored at -2O'C.
Sources of Other Collagenoses-Rat skin and rat uterus collagenases were obtained from tissue culture medium and partially purified by use of techniques previously described (16,17). Pure collagenase from the hepatopancreas of the fiddler crab, Uca pugilator, was prepared by methods described by Eisen et al. (18). Bacterial collagenase from Clostridium histolyticum was purchased from Advanced Biofactors.
Activation of Procollagenase-The activation of procollagenase was accomplished proteolytically at 25°C by the addition of trypsin for 10 min. A 4-fold molar excess of SBTI was then added to prevent any further tryptic action. A range of trypsin concentrations, usually from 0.1 to 5.0 pg/5O-pl sample was used to ensure that maximal collagenase activity was achieved (15). Assay Procedures-Collagenase activity and inhibitor activity were measured at 37'C using native reconstituted ["'Clglycine-labeled guinea pig skin collagen as substrate (19). The collagen was allowed to gel at 37°C for at least 24 h to permit completion of the aggregation process to occur. Twenty-five or fifty microliters of a 0.4% collagen solution having a specific activity of approximately 22,000 cpm/mg was employed for each assay. Following incubation with enzyme, or inhibitor, or both, the reaction mixtures were centrifuged at 10,000 to 12,000 X g and the supernatant fraction counted in a liquid scintillation spectrometer.
Protein concentrations were determined spectrophotometrically by the method of Groves et al. (20). All dissolved samples were then incubated in boiling water for 10 min prior to application to the slab gels. Following electrophoresis, the protein bands were fixed in a solution of 12.5% 2-propanol and 2.5% acetic acid and stained with 0.005% Coomassie brilliant blue. Scanning of selected gel slices was done using a Gilford model 2400s spectrophotometer equipped with a gel scanning linear transport device.
Autoradiography-Autoradiography was performed as described by Bonner and Laskey (23) following application of the samples to SDS-polyacrylamide slab gels. After electrophoresis, the gels were soaked in 200 ml of MeSO for 30 min, and this procedure was then repeated four times using fresh MeSO. The gels were then submerged in 4 volumes of 20% (w/w) 2,5-diphenyloxazole in Mess0 (22.2% w/ w) for 3 h, rinsed in distilled water for 1 h, and then dried under vacuum.
The dried gels were exposed on x-ray films (RP Royal X-Omat; Eastman Kodak) at -70°C for approximately 24 h.

RESULTS
Human skin collagenase inhibitor could be highly purified from serum-free fibroblast culture medium. The use of cation exchange chromatography was the essential step in this purification process.
As can be seen in Fig. 1 inhibit,or elut,ed as a nearly symmetrical protein peak (Fig. 3).  FIG. 3. Sephadex G-100 chromatography. The pooled inhibitor fractions from AcA-44 (tubes 49 to 60) were lyophilized, the product was resuspended in 0.05 M Tris-HCl, pH 7.5, and then applied to a G-100 column equilibrated with this same buffer. Fractions collected were added to active collagenase and assayed on "C-labeled collagen fibrils at 37°C to determine inhibitory activity. M, 4. SDS electrophoresis of collagenase inhibitor. Highly purified inhibitor and standard proteins of known molecular weight were electrophoresed on an SDS-polyacrylamide slab gel: A, bovine serum albumin; B, ovalbumin; C, inhibitor from the main body of its peak from G-100 chromatography; D, same as C reduced with dithiothreitol; E, pepsin; F, trypsin.
reduction of the sample with dithiothreitol prior to electrophoresis (Slot D). The higher molecular weight of reduced inhibitor compared to the unreduced molecule suggested that in its native state the inhibitor was restrained conformationallv bv disulfide linkages. Scanning of Slot C in a snectrowhor " tometer equipped with a gel-scanning linear transport device indicated that the inhibitor was greater than 95% pure. Of note was that the purified human fibroblast collagenase inhibitor was remarkably heat-stable. Following incubation at 90°C for 20 min it retained up to 60% of its activity, while collagenase under the same conditions was denatured almost instantaneously.
Since serum contains known collagenase inhibitors, several experiments were performed to show that this particular inhibitor was a gene product of the fibroblasts themselves and was not derived from the fetal calf serum used during culture. First, 500 ml of fetal calf serum, a quantity equivalent to that used in the culture medium of 50 roller bottles, was processed through ammonium sulfate precipitation and phosphocellulose chromatography.
The column fractions were then assayed for inhibitor and none was found. Second, 3H-labeled fibroblast proteins were prepared by adding a 3H-amino-acid mixture (New England Nuclear, to serum-free culture medium. The 3H-proteins produced were partially purified through phosphocellulose and compared electrophoretically with highly purified inhibitor.
The slab gel was then applied to a photographic film. The autoradiographic results shown in Fig. 5 clearly establish that the protein band corresponding to the purified inhibitor had indeed been labeled during exposure of the cells to 3H-amino-acids.
Finally, when anti-whole bovine serum was reacted against crude, concentrated serumfree fibroblast medium by Ouchterlony gel diffusion, no reaction was noted (not shown). Therefore, we conclude that this inhibitor is a fibroblast cell product.
The specificity of human fibroblast inhibitor was investigated against a number of collagenases and noncollagenolytic proteases (Table I). Significant inhibition of human skin, rat skin, and rat uterus collagenases was seen, but little or no inhibition of either the clostridial or crustacean collagenase was noted. Furthermore, no inhibition was observed when trypsin or chymotrypsin were employed as enzyme sources. The inhibitor, therefore, appears to be specific for and effective against vertebrate collagenases only.
The relationship between collagenase and its inhibitor is complex. When either inhibitor was titrated with increasing enzyme (Fig. 6) or enzyme titrated with increasing inhibitor (Fig. 7), a sigmoidal relationship was seen. In Fig. 7, 2.0 pg of pure procollagenase, having a molecular weight of 57,500 (14), was trypsin-activated and then titrated with increasing amounts of highly purified inhibitor, apparent molecular weight 31,000. Complete inhibition of the fibroblast collagenase was attained at a molar ratio of inhibitor to enzyme of approximately 1:l. Similar calculations at 50% inhibition showed an inhibitor to enzyme ratio of 0.5:1.0, indicating that at 50% inhibition, all inhibitor present was bound to enzyme and none remained free in solution. Therefore, the enzymeinhibitor interaction arfpeared to be characterized by tight binding.
However, when a fully inhibited pure enzyme *inhibitor mixture was subjected to gel filtration or ion exchange chromatography, using the same conditions as described under "Materials and Methods" for purification of the inhibitor, no evidence of a higher molecular weight complex was seen. This suggested that to produce the tight binding characteristics evidenced by the above titration curves, all three proteins, enzyme, inhibitor, and the substrate collagen, were required to be present.
In order to assess the possibility that the inhibitor does indeed bind to collagen, a constant amount of inhibitor was incubated with increasing amounts of fibrillar collagen (Fig.  8) was a function only of the amount of active enzyme in the sample. Thus, the curve produced by an enzyme solution containing 57% active enzyme and 43% proenzyme was similar to the inhibition curve for the enzyme solution which was fully active but contained only 50% of the total enzyme protein used in the other samples. Therefore, proenzyme forms do not appear to compete with the active enzyme for inhibitor.

DISCUSSION
Human skin fibroblast collagenase inhibitor is a tissuederived protein which is specific for vertebrate collagenases. Noncollagenolytic proteases failed to be inhibited, as were two collagenases of non-vertebrate origin ( Table I). The inhibitor has been purified extensively from serum-free culture medium, and is greater than 95% pure, as assessed by spectrophotometric scanning of polyacrylamide gels. It has a molecular weight of approximately 31,000 and is remarkably heatstable, retaining more than 60% of its activity after 20 min at 90°C.
Of particular interest is the nature of the interaction of fibroblast inhibitor with human skin collagenase. Our results suggest that inhibition is stoichiometric, approximately a 1:l molar ratio being required for complete inhibition of collagenase activity. The results of enzyme-inhibitor titrations are indicative of a tight binding of enzyme to inhibitor (24); yet attempts to demonstrate the existence of a complex between inhibitor and collagenase have been unsuccessful. Inhibitor and enzyme appear to behave as independent entities in the presence of each other. Hence, it is reasonable to assume that the tight binding required by kinetics occurs only in the presence of the collagen substrate, and that the inactive species is the ternary complex E. 1. S. To our knowledge, this is the only example of a protease-antiprotease system in which a stable enzyme. inhibitor complex does not form in the absence of substrate. It is clear from the experiment depicted in Fig. 8 that the inhibitor itself is capable of binding to collagen, and it is possible that formation of the inactive complex is mediated by the precise, specific binding of both enzyme and inhibitor to the substrate. The ultimate fate of these enzyme. inhibitor.collagen complexes is unknown. It is interesting in this regard that both trypsin-activated and autoactivated enzyme compete effectively for inhibitor in the presence of the collagen substrate, but procollagenase, which does not bind to collagen (14), also fails to bind the inhibitor.
It should be noted that Reynolds and co-workers (25) have postulated a mechanism for the physiologic activation of vertebrate collagenases which involves the removal of an inhibitor from a stable, inactive enzyme. inhibitor complex. As discussed above, no evidence exists for such a complex between human skin fibroblast collagenase and the inhibitor described in these studies. The inhibitor produced by the fibroblasts appears to be destined to participate in the inactivation of collagenase, rather than its activation.
The results presented in this communication provide the first evidence for the existence of an inhibitor of a collagenolytic enzyme synthesized by the same cell type that produces the collagenase itself, and indicate that certain tissues may well have the ability to regulate their own collagenase activity. It is not known at this time whether the same cell can simultaneously produce both collagenase and its inhibitor, or whether different populations of cells exist within a culture, some synthesizing enzyme, others producing the inhibitor. In typical fibroblast cultures, collagenase levels appear to be considerably higher than levels of inhibitor, as activatable collagenase can always be observed in crude, serum-free culture medium, even though inhibitor is present. All fibroblast cultures examined to date have yielded detectable amounts of inhibitor, but different cell lines, all derived from normal human skin, vary markedly in the amount of inhibitor produced. In contrast, explants of human skin in culture have consistently failed to yield detectable levels of an inhibitor of the kind described here. The reason for the absence of inhibitor in this organized tissue is unknown, but may be related to the observation that the explants undergo massive collagen resorption in culture.' It is possible that under these conditions, production of inhibitor does not occur. Delineation of the pathways by which cells and tissues may regulate levels of collagenase inhibitor will be of great value in our attempts to understand the mechanisms whereby an organized connective tissue maintains its three-dimensional structure.