Production of procollagenase by cultured human keratinocytes.

Using a collagen film assay utilizing 14C-labeled type I collagen, we demonstrated that cultured human keratinocytes produced a procollagenase after treatment with the tumor-promoting phorbol ester, 12-O-tetradecanoylphorbol-13-acetate (TPA). Production of collagenase paralleled alterations in cellular morphology induced by TPA. When procollagenase was immunoprecipitated with antibody to human fibroblast collagenase and analyzed on sodium dodecyl sulfate-polyacrylamide gels, the zymogen was revealed as a 56- and 51-kDa doublet. The keratinocyte-derived collagenase was a neutral metalloprotease, required activation with trypsin for detection in the collagenase assay and produced the characteristic three-quarter and one-quarter length collagen cleavage products when incubated with type I collagen at 25 degrees C. The enzyme was inhibited by serum and cysteine and was largely unaffected by serine, thiol, and carboxyl protease inhibitors. Cycloheximide inhibited the TPA-induced production of collagenase, suggesting that the procollagenase was not stored preformed in the keratinocytes. Keratinocytes treated with a tumor-promoting analogue of TPA also produced collagenase, but cells treated with cytochalasin B, interleukin-1, or two non-tumor promoting phorbol esters did not. Keratinocyte-derived collagenase may play a role in wound healing and morphogenesis.

Using a collagen film assay utilizing 14C-labeled type I collagen, we demonstrated that cultured human keratinocytes produced a procollagenase after treatment with the tumor-promoting phorbol ester, 12-O-tetradecanoylphorbol-13-acetate (TPA). Production of collagenase paralleled alterations in cellular morphology induced by TPA. When procollagenase was immunoprecipitated with antibody to human fibroblast collagenase and analyzed on sodium dodecyl sulfate-polyacrylamide gels, the zymogen was revealed as a 56and 5 l-kDa doublet. The keratinocyte-derived collagenase was a neutral metalloprotease, required activation with trypsin for detection in the collagenase assay and produced the characteristic three-quarter and one-quarter length collagen cleavage products when incubated with type I collagen at 25 "C. The enzyme was inhibited by serum and cysteine and was largely unaffected by serine, thiol, and carboxyl protease inhibitors. Cycloheximide inhibited the TPA-induced production of collagenase, suggesting that the procollagenase was not stored preformed in the keratinocytes. Keratinocytes treated with a tumor-promoting analogue of TPA also produced collagenase, but cells treated with cytochalasin B, interleukin-1, or two non-tumor promoting phorbol esters did not. Keratinocyte-derived collagenase may play a role in wound healing and morphogenesis.
Vertebrate collagenase, a neutral metalloendoproteinase, cleaves native collagen into specific three-quarter and onequarter length fragments which are susceptible to further proteolysis (1). In the skin, increased or abnormal collagenase production may contribute to the pathogenesis of several dermatologic diseases including dystrophic epidermolysis bullosa (2), an inherited blistering disorder, and locally invasive basal cell carcinomas (3). Although dermal fibroblasts produce collagenase i n vitro (4) and are thought to represent the primary source of collagenase in the skin, the tailfin epidermis of the tadpole was the source of the first vertebrate collagenase described (5). Wound healing studies in humans (6) and animals (7,8) have suggested the presence of collagenase activity in wound edge epithelium. Eisen (9) also demon-* This work was supported in part by Grant AM 25871 and Research Career Development Award AM 00977 (to E. OK.) and by Grant AM 33625 and Research Career Development Award AM 01540 (to D. T. W.), all from the National Institute of Arthritis, Diabetes, Digestive and Kidney Diseases. 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.
$ Supported by National Institutes of Health Training Grant AM 07369.
ll Supported by Veterans's Administration research funds. strated increased collagenase activity in the epidermis from patients with epidermolysis bullosa. However, these studies were performed on cutaneous layers (enzymatically separated epidermis and dermis) and therefore could not establish the cellular source of the collagenase. Subsequently, identification of collagenase by indirect immunofluorescence using anticollagenase antibody demonstrated staining of the papillary dermis but not the epidermis (10). More recently, Woodley et al. (11) demonstrated type I and type IV collagenase activity in the conditioned medium from explants of human epithelium grown on nonviable pig dermis. The explants were grown at an air-fluid interface which promotes keratinocyte and suppresses fibroblast proliferation (12). Detection of collagenase production i n vitro is facilitated under serum-free culture conditions (4,13) and, in fibroblasts, can be augmented or induced by multiple agents including cytochalasin B (14), proteolytic enzymes (15), and interleukin-1 (16). Aggeler et al. (17) have shown that production of collagenase by rabbit synovial fibroblasts correlates with changes in cellular morphology characterized by cell rounding and loss of cell to cell contact. The phorbol ester tumor promoter TPA' is a potent inducer of these morphological changes and collagenase production in rabbit synovial fibroblasts (17) and endothelial cells (18). We therefore examined human keratinocyte monolayers for the production of collagenase in the presence or absence of TPA, cytochalasin B, interleukin-1, and extracellular matrix proteins. In this study, we found that human keratinocyte cultures in serum-free medium and in the absence of mesenchymal cells were capable of producing a procollagenase in response to TPA treatment.
Furthermore, this new keratinocyte-derived collagenase is similar to that produced by human dermal fibroblasts. Materials used to coat the culture dishes included type I collagen purified from rat tail tendons (20) and fibronectin, purified from human plasma according to Hayashi and Yamada (21). Cell Cultures-Human keratinocytes obtained from neonatal foreskins and adult skin (22) were grown to early confluency at 37 "C in 35-mm tissue culture dishes (Corning, Corning, NY) in serum-free MCDB 153 medium containing 0.1 mM Caz' (23) with the following supplements: hydrocortisone, 0.4 pg/ml; insulin, 5 pg/ml; phosphoethanolamine, 0.1 mM; ethanolamine, 0.1 mM; epidermal growth factor, 5 ng/ml; bovine pituitary extract, 150 pg/ml; histidine, 0.24 mM; isoleucine, 0.75 mM; methionine, 0.09 mM; phenylalanine, 0.09 mM; tryptophan, 0.045 mM; tyrosine, 0.075 mM. Cells were used between passages 3 and 7. At confluency, the cells were washed with Hank's balanced salt solution; the medium, as above but without hydrocortisone to avoid suppression of collagenase production (24), was replaced and various agents (see below) added to triplicate keratinocyte cultures. The medium was collected after 48-96 h of incubation, clarifed and stored at -20 "C until used in the collagenase assay. The following agents were dissolved in Me2S0 and added to the keratinocyte cultures to assess their effect on collagenase production: TPA, three phorbol analogues (4a-phorbol 12,13-didecanoate; phorbol 12,13-didecanoate, and 4c~-phorbol), cytochalasin B, and cycloheximide. The Me2S0 concentration in the media did not exceed 0.1% and control cultures were treated with 0.1% Me2S0. The interleukin-1 was diluted in culture medium and added to the cultures. Keratinocytes were also cultured on tissue culture plates coated with type I collagen or fibronectin as described (25) and collagenase production with and without TPA stimulation was assessed.
Human fibroblasts harvested from neonatal foreskins were maintained in Dulbecco's minimal essential medium plus 10% fetal bovine serum. For use in the immunoprecipitation experiment, the cells were subcultured into 35-mm tissue culture dishes and at early confluency washed with Hank's buffered salt solution and changed to serum-free medium.
Immunofluorescence Studies-Cultured keratinocytes grown on coverslips were incubated overnight with TPA, 10 ng/ml, washed with Hank's buffered salt solution, and incubated in fresh medium for 3 h with 1 p~ monensin to prevent secretion of the collagenase into the medium (26). The keratinocytes were fixed and extracted as described in Ref. 27, incubated with the mouse monoclonal antikeratin antibodies (mixed 1:1, undiluted) for 30 min, followed by rabbit anti-fibroblast collagenase antiserum (1:lOO). Control cells were incubated with normal rabbit serum (1:lOO). Cells were incubated sequentially with fluorescein-conjugated goat anti-mouse IgG and rhodamine-conjugated sheep anti-rabbit IgG.
Collagenase Assay-Conditioned medium was assayed directly for collagenase activity using a 14C-collagen fibril film method (28). Rat tail tendon collagen was purified using successive salt and phosphate precipitation (20) and acetylated with [1-"Clacetic anhydride (1 mCi) as described (28) to a specific activity of 3.7 X lo5 cpm/mg. Assays were performed at 37 "C for 90 min. Each sample was assayed in triplicate following activation with t-1-tosylamido-2-phenylethyl chloromethyl ketone-trypsin (10 pglml, 30 min at 25 "C) followed by a 5-fold molar excess of soybean trypsin inhibitor. Each assay included controls with unconditioned medium, trypsin (O.Ol%), and crude bacterial collagenase (5 mg/ml). Trypsin controls did not exceed 12% of the total available counts. The medium blank was subtracted from each experimental sample to determine the amount of "Clabeled collagen solubilized per sample.
In some experiments, organomercurials were used to activate the collagenase. Mersalyl (25 mM) and p-aminophenylmercuric acetate (10 mM) were dissolved in 0.05 M NaOH and added in a 1:lO dilution to each sample. In experiments to characterize the effect of protease inhibitors on the collagenase, PMSF, NEM, pepstatin, EDTA, cysteine, or normal human serum was added to trypsin-activated conditioned medium for 30 min at 25 "C prior to assay. PMSF, NEM, pepstatin, and cysteine were dissolved in 95% ETOH and EDTA was dissolved in distilled water. Samples were compared with appropriate solvent controls; ethanol alone (2.0% final concentration) added to conditioned medium was not inhibitory in the collagenase assay.
Electrophoresis and Fluorography of Solubilized Collagen-200 pl of keratinocyte-conditioned medium was activated as above and incubated with the "C-labeled collagen film for 0, 24, 37, 48, and 63 h at 25 "C. After incubation, the medium containing collagenase and the digested collagen was removed from the collagen film plate and added to 50 pl of 5 X concentrated sample buffer containing 5% SDS, 50% glycerol, 50 mM Tris-HC1, pH 7.4, 0.01% bromphenol blue, and dithiothreitol (60 mg/ml). 100 pl of unconditioned medium with 1 X concentrated sample buffer was added to each well of the collagen film plate and the plate was floated in a 56 "C water bath for 30 min to solubilize the remaining collagen film. The solubilized collagen was recombined with the appropriate sample, heated at 100 "C for 3 min, and 50 pl of each sample (2000 cpm) was analyzed on a 7.5% SDSpolyacrylamide resolving gel run according to Laemmli (29). Gels were lightly stained, impregnated with Fluoro-Hance and exposed to Kodak X-Omat XAR5 x-ray film for 5 days at -70 "C. Immunoprecipitation-TPA (20 ng/ml)-treated keratinocytes and fibroblasts in serum-free medium were labeled with ~(~~S l r n e t h i o n i n e (35 pCi/ml) for 24 h. Before the addition of the labeled methionine, cells were refed with 1 ml of MCDB 153 containing 0.03 mM unlabeled methionine (keratinocytes) or a 1:l solution of Ham's F-12 medium with Hank's buffered salt solution containing 0.015 mM unlabeled methionine (fibroblasts). The labeled medium was clarified and preabsorbed with pre-immune rabbit serum and Pansorbin as described by Stanley et al. (30). All immunoprecipitation steps were performed at 4 "C. 500 p1 of labeled preabsorbed conditioned medium was incubated overnight with 3 pl of rabbit human fibroblast collagenase antiserum or normal rabbit serum. Samples were centrifuged at 12,000 X g for 3 min and transferred to new 1.5-ml polypropylene centrifuge tubes. 50 pl of a 25% slurry of Protein A-Sepharose CL-4B was added to each sample and incubated with agitation for 2 h. Samples were washed three times with 1 ml of buffer containing 0.3% sodium deoxycholate, 0.3% Nonidet P-40, 0.5 M NaCI, 0.01 M Tris-HCl, pH 7.4, and 0.1% crystalline bovine serum albumin and three times with the same buffer without bovine serum albumin. Samples were then washed once with distilled water, transferred to new tubes, centrifuged, and the water was decanted from the Sepharose beads. 100 p1 of 1 x sample buffer (see above) was added to each sample which was heated for 4 min at 100 "C. Samples were analyzed on a 10% SDS-polyacrylamide gel and fluorography was performed as above.

RESULTS
Since human dermal fibroblasts are known to produce collagenase, it was necessary to establish that the keratinocyte cultures were free of contaminating fibroblasts and that the collagenase was produced by the keratinocytes. Keratinocytes were grown for at least three passages in low-calcium serumfree MCDB 153 medium, which promotes keratinocyte growth and selects against fibroblast growth (24). Fibroblast proliferation was not noted in subsequent subcultures of these keratinocyte monolayers grown for 14 days in medium which supports fibroblast growth (Dulbecco's minimal essential medium plus 10% fetal calf serum). The cellular source of the collagenase detected in the keratinocyte cultures was verified further by the immunofluorescent studies of TPA-and monensin-treated keratinocytes. Particulate intracellular collagenase staining was present in keratin-containing epithelial cells (Fig. l), confirming the synthesis of collagenase by keratinocytes.
Treatment of pure cultures of neonatal and adult keratinocytes with TPA produced cell rounding and loss of cell to cell contact (Fig. 2) and these morphologic alterations correlated with production of collagenase. No collagenase activity was found in conditioned medium from control cultures treated with Me2S0 alone. Mixing experiments using medium from both control and stimulated cells indicated that the lack of enzyme activity in the control medium was not due to the presence of significant amounts of collagenase inhibitor. In stimulated cultures, the morphologic changes and collagenase activity at 48 h occurred after as little as 30 min exposure to TPA. The amount of collagenase activity detected in medium from triplicate cultures of TPA-treated keratinocytes is shown

Collagenase and Human
Keratinocytes 837 P

FIG. 1. Demonstration of intracellular collagenase and keratin in human keratinocytes by indirect immunofluorescence.
Cells were incubated at 37 "C with 10 ng/ml of TPA for 24 h followed by 1 p~ of monensin for 3 h, fixed, and permeabilized as described under "Experimental Procedures." Cells were treated sequentially with mouse monoclonal anti-keratin antibodies or control by hybridoma supernatant and fluoresceinated goat anti-mouse IgG to demonstrate keratin filaments. They were then treated with rabbit anticollagenase antiserum or normal rabbit serum and rhodamine-conjugated sheep anti-rabbit IgG to demonstrate intracellular collagenase. Cells were examined using epifluorescence optics and photographed using Kodak Tri-X film at an ASA of 1600. in Fig. 3A. Maximal collagenase activity (0.11 pg of 14C-labeled collagen solubilized per lo5 cells/min) was seen at 48 h in response to 10 ng/ml of TPA (1.6 X lo-' M). Collagenase activity was detectable 18 h after TPA exposure (Fig. 3B) and FIG. 2. Morphologic changes produced by TPA. Confluent neonatal human keratinocytes were incubated in the presence or absence of TPA for 48 h. a, untreated cells are well spread and in contact with adjacent cells; b, cells treated with 1 ng/ml of TPA showed slightly prominent borders but no marked morphologic change and no detectable collagenase activity; c, cells treated with 10 ng/ml of TPA showed marked retraction and cell rounding. Greater than 85% of the adherent cells in these cultures excluded trypan blue. Collagenase activity was detected at this concentration (see Fig. 3A).
increased little after 24 h. Cells treated with cycloheximide (3 pg/ml) and TPA (100 ng/ml) produced no detectable collagenase activity when conditioned medium was assayed after 48 h, suggesting that the collagenase produced in response to TPA was newly synthesized.
The keratinocytes secreted a latent enzyme which required activation with trypsin for detection in the collagenase assay and was optimally activated by trypsin at concentrations of 5-10 pg/ml of serum-free conditioned medium. Incubation of the conditioned medium with the organomercurial, mersalyl (2.5 mM), resulted in partial activation of the procollagenase: 26.0% +-3.5 (mean +-S.E., n = 5) of the activity produced by activation by trypsin (10 pg/ml). Similarily, p-aminophenylmercuric acetate (1 mM) activated 18.7% k 0.9 ( n = 5) of the procollagenase activated by trypsin.  3. A, stimulation of collagenase activity by TPA. TPA was added in concentrations shown to triplicate 35mm culture plates of confluent human keratinocytes containing 2 ml of medium for 6 h after which time the medium was replaced with medium without TPA. The conditioned medium was collected after 48 h of culture. A collagenase film assay was used to quantitate collagenase activity. The type I "C-labeled collagen had a specific activity of 3.7 X lo6 cpm/mg. The keratinocyte-derived collagenase was activated with trypsin as described under "Experimental Procedures" prior to assay. Maximum production of collagenase occurred after treatment with TPA at 10 ng/ml: 0.11 pg of 14C-labeled collagen solubilized per lo5 cells/min. Data are expressed as the mean of triplicate plates f S.E. B, time course of TPA-stimulated collagenase production. TPA (10 ng/ml) was added to 35-mm culture plates of confluent keratinocytes and conditioned medium from triplicate plates was harvested at the indicated times (closed circles). Control cultures (open circles) were treated with Me2S0 alone (0.05%) and the conditioned medium harvested at the times shown. Data are expressed as the mean of triplicate plates f S.E.

TABLE I
Effect of interleukin-1 on produetion of collagenase activity by keratinocytes and fibroblasts Human keratinocytes (passage 3) and human dermal fibroblasts (passage 11) were cultured as described under "Experimental Procedures." At confluence recombinant interleukin-1 or TPA was added in the concentrations shown to triplicate cultures for 48 h. The conditioned medium was activated with trypsin (5 pg/ml, 30 min, 25 "C) and collagenase activity assessed with the collagen film assay (see "Experimental Procedures"). The specific activity of the 14Clabeled type I collagen was 7.5 X IO5 cpm/mg. tumor-promoting properties, also induced both the morphologic changes and procollagenase production: 0.07 gg of I4Ccollagen was solubilized per lo5 cells/min by activated conditioned medium from these cultures. No collagenase activity nor morphologic changes were seen in keratinocytes treated with 3 X lo-* M of 4a-phorbol 12,13-didecanoate or 4aphorbol, two TPA analogues without tumor-promoting potential. No collagenase activity was detected in medium from keratinocytes treated with 1, 5, or 10 units/ml of interleukin-1; while fibroblasts treated in parallel in the same experiment with 5 units/ml of interleukin-1 produced substantial collagenase activity (Table I).
No collagenase production was detected in cells treated with cytochalasin B (1, 5, or 10 gg/ ml), an agent shown to induce collagenase production in rabbit synovial fibroblasts (14,17). Keratinocytes grown on type I collagen or fibronectin did not produce detectable collagenase activity, nor did the presence of these extracellular matrix proteins augment or inhibit the keratinocyte response to TPA (data not shown).
The progressive accumulation of the cleavage products of the '*C-collagen resulting from incubation of the labeled collagen with keratinocyte-conditioned medium is shown in Fig.   4A. These cleavage products, the N-terminal three-quarter fragment (TCA), and a C-terminal one-quarter fragment (TCB), are specific for vertebrate collagenase and reflect enzyme cleavage at the 775-776 Gly-Ile residues on the a1 chain and the Gly-Leu residues on the a2 chain of type I collagen (31). Immunoprecipitation of "S-labeled TPA-treated cell supernatants with rabbit antiserum to human dermal fibroblast collagenase indicated antigenic similarity of keratinocyte-derived collagenase with the fibroblast enzyme. Fluorograms of SDS-polyacrylamide gels demonstrated 56and 51-kDa protein bands in the conditioned medium from both the dermal fibroblasts and the keratinocytes (Fig. 4B). These bands correspond in molecular weight with the inactive zymogens produced by human dermal fibroblasts (32).
Experiments with protease inhibitors were designed t o characterize the keratinocyte-derived collagenase. Its activity was totally inhibited by EDTA (Table 11). Cysteine, which is thought to inhibit collagenase activity by chelation of an intrinsic metal ion of the enzyme (33), decreased the keratinocyte-derived collagenase activity by 93%. Human serum, which contains a*-macroglobulin, a potent collagenase inhibitor (34), reduced collagenase activity by 80%. Little effect on collagenase activity was produced with addition of serine (PMSF) or carboxyl (pepstatin) protease inhibitors. NEM, a thiol protease inhibitor, reproducibly inhibited collagenase activity by approximately 30%, but this effect was independent of the concentration of the inhibitor and therefore was probably nonspecific. A similar effect with NEM was reported in studies of endothelial cell collagenase (18). These data characterize the keratinocyte-derived enzyme as a metalloprotease similar to other vertebrate collagenases (31).

DISCUSSION
In this study we have demonstrated that cultured human keratinocytes treated with the phorbol ester, TPA, produce a procollagenase similar to that produced by dermal fibroblasts. Conditioned medium from the cultures produced characteristic three-quarter and one-quarter length fragments after incubation with type I collagen. Immunoprecipitation of the zymogen with antibody raised against dermal fibroblast collagenase revealed a 56-and 51-kDa doublet similar to the  4. A, collagen digestion by keratinocyte-derived collagenase. Cultured human keratinocytes were treated with TPA (10 ng/ml) for 48 h and the medium was activated with trypsin (10 pg/ml, 30 min, 25 "C) and assayed. After incubation for 0 (Lane I ) , 24 (Lane 2), 37 (Lane 31, 48 (Lane 4), or 63 (Lane 5 ) h at 25 "C, the digested or undigested collagen was solubilized with electrophoresis sample buffer as described under "Experimental Procedures." Samples were heated (100 "C for 3 min) and 50 p1 of each sample (2000 cpm) was electrophoresed on a 7.5% SDS-polyacrylamide gel and detected by fluorography. The native trimers (y), dimers (p), and monomers (a1, az) of type I collagen are shown in Lane 1. The progressive degradation of the collagen yielded the characteristic three-quarter size TCA and one-quarter size TCB fragments of the two a chains. In addition, the three-quarter size cleavage products of the dimers can be seen. B, Immunoprecipitation of procollagenase from the conditioned medium of TPA-treated keratinocytes and fibroblasts. Cells were treated with medium containing 10 ng/ml of TPA for 24 h, then labeled with 35 pCi of [35S]methionine for 24 h. Secreted proteins present in the conditioned culture medium from human fibroblast and keratinocyte cultures were immunoprecipitated with pre-immune rabbit serum or rabbit anti-collagenase antiserum (see "Experimental Procedures"), electrophoresed on a 10% SDS-polyacrylamide gel, and fluorographed. Lanes 2 and 4 show the control immunoprecipitations using pre-immune rabbit serum with conditioned medium from fibroblasts and keratinocytes, respectively. Lane 3, fibroblast-conditioned medium with anti-collagenase antiserum. proenzyme produced by human fibroblasts in serum-free medium (32). The enzyme, a metalloprotease, was secreted as an inactive zymogen in serum-free culture. Production of collagenase by the TPA-treated keratinocytes was dependent on new protein synthesis, as evidenced by the inhibitory effect of cycloheximide. This finding and the profile of response of the keratinocyte-derived collagenase to protease inhibitors parallels results obtained with collagenase produced by fibroblasts in vitro (6,35). Activation of the procollagenase by trypsin and, to a lesser extent, by organomercurials is also a feature shared by human fibroblast collagenase (36). In contrast to fibroblasts (14,37,38), keratinocyte collagenase production was not stimulated by cytochalasin B, interleukin-1, or type I collagen. These findings suggest that the intracellular signals resulting in production of collagenase in epidermal cells may differ from those in the fibroblasts. Alternatively, increased production of an inhibitor may mask collagenase activity. Although the regulation of production and secretion of collagenase may differ in fibroblasts and keratinocytes, the limited characteristics of the enzyme described here do not indicate any functional or biochemical differences between

Effect of protease inhibitors on the collagenase activity produced by TPA-treated human keratinocytes
Protease inhibitors were dissolved in 95% ethanol and added to the activated conditioned keratinocyte medium for 30 min at 25 "C at the concentrations shown. A control, activated conditioned medium containing 2% ethanol, was assayed in conjunction with each sample. The collagenase assay was performed at 37 "C for 90 min as described under "Experimental Procedures." The activity is expressed as the mean percentage f S.E. of control conditioned medium containing 2.0% ethanol, which did not alter collagenase activity. Number of samples tested.
the fibroblast and keratinocyte enzymes. The discovery of keratinocyte-derived collagenase supports the studies by Eisen (6) and Woodley et al. (11) in which collagenolytic activity was detected in explant cultures of wound edge epithelium and keratinocytes grown on a nonviable dermal substrate, respectively. It was not possible to exclude the presence of contaminating fibroblasts producing the observed collagenase activity in the study by Eisen, whereas Woodley et al. excluded this possibility more rigorously. Several methods were used to verify that the collagenase detected in our cultures was produced by the keratinocytes. First, all keratinocytes used in these experiments had been passaged at least three times in serum-free, low-calcium MCDB 153 medium which does not support fibroblast growth (23); keratinocyte cultures refed with medium which supported fibroblast growth did not reveal fibroblasts in subsequent subcultures. Second, no collagenase activity was detected in low passage keratinocyte cultures treated with interleukin-1, a potent stimulator of collagenase production in fibroblasts, while fibroblast cultures treated in parallel in the same experiment produced substantial collagenase activity. Lastly, and most importantly, we demonstrated intracellular collagenase in TPA-treated keratinocytes by indirect immunofluorescence.
The production of collagenase by various cell types including rabbit synovial and human dermal fibroblasts (13,14), hepatocytes (39), and endothelial cells (18) has been augmented in vitro by multiple agents including cytochalasin B (14), phorbol esters (17), phagocytosis (40), or the calcium ionophore A23187 (41). Aggeler et al. (42) demonstrated that treatment of rabbit synovial fibroblasts with such agents caused a marked alteration in gene expression resulting in induction of procollagenase and that such induction paralleled alterations in cell shape produced by some of these agents. We found that treatment of human keratinocytes with TPA produced similar morphologic changes, such as loss of cellular adhesion and cytoplasmic vacuolization, and that these effects occurred in our TPA-treated keratinocyte cultures in parallel with the production of collagenase. TPA is structurally similar to diacylglycerol and is thought to act at least in part through activation of protein kinase C (43). Protein kinase C has a central role in signal transduction for biologically active substances leading to activation of cellular functions and prolif-eration. Thus, the effects of TPA, including collagenase production, presumably reflect physiologic cellular responses.
One possible model for collagenase production by keratinocytes in vivo is wound healing. Epidermal cells from wounds show enhanced proliferation (44) and marked phenotypic changes including the formation of peripheral cytoplasmic actin filaments, increased numbers of gap junctions, retraction of tonofilaments, and loss of desmosomes (45). These phenotypic alterations are thought to enhance cellular motility, which is needed for re-epithelialization of the wound. The production of collagenase may be stimulated in turn by these phenotypic alterations. In wounds in which the basement membrane has been disrupted, epidermal cells migrate directly over the extracellular matrix of the dermis, whose major constituent is type I collagen, and a provisional matrix of fibronectin, fibrin, and type V collagen (46, 47). Cultured keratinocytes have been reported to synthesize and secrete certain matrix proteins including fibronectin (27), which may contribute to the motility of the cells (27, 48), and laminin and type IV collagen (49). Similarly, ARL-6, an epithelial cell line derived from rat liver, has been shown to synthesize type I, 111, and IV collagen, laminin, and fibronectin (50). Thus, epithelial cells may well have an active role in the synthesis and degradation of the extracellular matrix during the process of wound healing.
Previously, the presumptive role of epidermal cells in collagenolysis has been the production of several cytokines which stimulate collagenase production by dermal fibroblasts (51). Our studies suggest that the epidermal cells may play a primary role in collagen degradation through production of collagenase. Production of excessive or unregulated collagenase by keratinocytes could contribute to the pathogenesis of the dermolytic bullous disease, dystrophic epidermolysis bullosa, and other skin disorders characterized by dermal-epidermal separation. Furthermore, keratinocyte-derived collagenase may play a role in the mechanism of connective tissue invasion by cutaneous tumors.