Keratinocyte Collagenase-1 Expression Requires an Epidermal Growth Factor Receptor Autocrine Mechanism*

In response to cutaneous injury, expression of collagenase-1 is induced in keratinocytes via α2β1 contact with native type I collagen, and enzyme activity is essential for cell migration over this substratum. However, the cellular mechanism(s) mediating integrin signaling remain poorly understood. We demonstrate here that treatment of keratinocytes cultured on type I collagen with epidermal growth factor receptor (EGFR) blocking antibodies or a specific receptor antagonist inhibited cell migration across type I collagen and the matrix-directed stimulation of collagenase-1 production. Additionally, stimulation of collagenase-1 expression by hepatocyte growth factor, transforming growth factor-β1, and interferon-γ was blocked by EGFR inhibitors, suggesting a required EGFR autocrine signaling step for enzyme expression. Collagenase-1 mRNA was not detectable in keratinocytes isolated immediately from normal skin, but increased progressively following 2 h of contact with collagen. In contrast, EGFR mRNA was expressed at high steady-state levels in keratinocytes isolated immediately from intact skin but was absent following 2 h cell contact with collagen, suggesting down-regulation following receptor activation. Indeed, tyrosine phosphorylation of the EGFR was evident as early as 10 min following cell contact with collagen. Treatment of keratinocytes cultured on collagen with EGFR antagonist or heparin-binding (HB)-EGF neutralizing antibodies dramatically inhibited the sustained expression (6–24 h) of collagenase-1 mRNA, whereas initial induction by collagen alone (2 h) was unaffected. Finally, expression of collagenase-1 in ex vivo wounded skin and re-epithelialization of partial thickness porcine burn wounds was blocked following treatment with EGFR inhibitors. These results demonstrate that keratinocyte contact with type I collagen is sufficient to induce collagenase-1 expression, whereas sustained enzyme production requires autocrine EGFR activation by HB-EGF as an obligatory intermediate step, thereby maintaining collagenase-1-dependent migration during the re-epithelialization of epidermal wounds.

In response to cutaneous injury, expression of collagenase-1 is induced in keratinocytes via ␣ 2 ␤ 1 contact with native type I collagen, and enzyme activity is essential for cell migration over this substratum. However, the cellular mechanism(s) mediating integrin signaling remain poorly understood. We demonstrate here that treatment of keratinocytes cultured on type I collagen with epidermal growth factor receptor (EGFR) blocking antibodies or a specific receptor antagonist inhibited cell migration across type I collagen and the matrix-directed stimulation of collagenase-1 production. Additionally, stimulation of collagenase-1 expression by hepatocyte growth factor, transforming growth factor-␤1, and interferon-␥ was blocked by EGFR inhibitors, suggesting a required EGFR autocrine signaling step for enzyme expression. Collagenase-1 mRNA was not detectable in keratinocytes isolated immediately from normal skin, but increased progressively following 2 h of contact with collagen. In contrast, EGFR mRNA was expressed at high steady-state levels in keratinocytes isolated immediately from intact skin but was absent following 2 h cell contact with collagen, suggesting down-regulation following receptor activation. Indeed, tyrosine phosphorylation of the EGFR was evident as early as 10 min following cell contact with collagen. Treatment of keratinocytes cultured on collagen with EGFR antagonist or heparin-binding (HB)-EGF neutralizing antibodies dramatically inhibited the sustained expression (6 -24 h) of collagenase-1 mRNA, whereas initial induction by collagen alone (2 h) was unaffected. Finally, expression of collagenase-1 in ex vivo wounded skin and re-epithelialization of partial thickness porcine burn wounds was blocked following treatment with EGFR inhibitors. These results demonstrate that keratinocyte contact with type I collagen is sufficient to induce collagenase-1 expression, whereas sustained enzyme production requires autocrine EGFR activation by HB-EGF as an obligatory intermediate step, thereby maintaining collagenase-1-dependent migration during the re-epithelialization of epidermal wounds.
Efficient repair of a cutaneous wound requires a programmed series of spatially and temporally regulated events. Among these, effective proteolytic degradation of extracellular matrix (ECM) 1 macromolecules is thought to be necessary to remodel the damaged tissue, promote neovascularization, and facilitate efficient migration of cells during re-epithelialization (1). Matrix metalloproteinases (MMPs) constitute a family of zinc-dependent enzymes with the collective capacity to degrade virtually all ECM components (2). Although most MMPs can degrade many ECM proteins with overlapping substrate specificities, degradation of fibrillar type I collagen is initiated only by the catalytic activity of collagenases, a subgroup of the MMP gene family.
Previous studies from our group and others have shown that, in both normally healing wounds and chronic ulcers, basal keratinocytes at the leading edge of re-epithelialization invariantly express collagenase-1 (3)(4)(5)(6). Collagenase-1 expression is rapidly induced in wound-edge keratinocytes after injury, persists during the healing phase, and ceases following complete re-epithelialization (7,8). We demonstrated that induction of collagenase-1 by basal keratinocytes is mediated via ␣ 2 ␤ 1 interaction with native type I collagen (9,10), requires tyrosine kinase activity (11), and that the activity of this MMP is essential for cell migration over this matrix protein (10). Although much is known about the role of cell:matrix interactions regulating collagenase-1 expression by keratinocytes, the signaling mechanism(s) following collagen binding and integrin receptor occupancy remain poorly understood.
The epidermal growth factor receptor (EGFR; c-erbB1/ HER1) is a transmembrane cell-surface tyrosine kinase that, upon ligand binding, phosphorylates downstream effector molecules, leading to changes in cell function (12). EGFR-null mice demonstrate involvement of the receptor in a broad range of developmental processes, and these mice have pronounced defects in epithelial cell proliferation and differentiation (13)(14)(15). Activation of the EGFR by members of the EGF family of growth factors (i.e. EGF, TGF-␣, HB-EGF, and amphiregulin) is associated with multiple keratinocyte functions during wound repair, including cell proliferation, migration, and stimulation of ␣ 2 ␤ 1 integrin expression (16,17). Keratinocyte migration is essential for effective re-epithelialization, and expression of the EGFR and its ligands is up-regulated following injury in vivo (17,18). Indeed, exogenously administered EGFR ligands (EGF, TGF-␣) stimulate keratinocyte motility in vitro and re-epithelialization in vivo (19 -23), and wound healingspecific keratins (K6 and K16) contain upstream regulatory sequences responsive to EGFR activation (24). Furthermore, EGFR overexpression in cultured keratinocytes enhances ligand-mediated motility (25).
Autocrine signaling mechanisms often regulate cell function and behavior, and, in keratinocytes, autocrine activation of the EGFR can influence epithelial homeostasis and cutaneous repair. Indeed, recent evidence from Stoll et al. (51) demonstrates that heparin-binding ligands, namely HB-EGF and amphiregulin, mediate autocrine activation of the EGFR in a skin organ culture model, suggesting that these ligands play an important role in the amplification and transmission of the wound healing signal. Because collagenase-1 production is tyrosine kinase-dependent (11) and because keratinocytes express both the EGFR and its various ligands during wound repair, we reasoned that an autocrine loop mechanism may regulate keratinocyte collagenase-1 expression following ␣ 2 ␤ 1 integrin-mediated collagen binding. Here, we report that the initial induction of collagenase-1 by keratinocytes following contact with type I collagen requires only integrin receptor activation. However, autocrine activation of the EGFR by HB-EGF is required for the sustained expression of collagenase-1 by keratinocytes in vitro and during the full re-epithelialization process in vivo.

Isolation and Culture of Human Keratinocytes
Human keratinocytes were harvested from healthy adult skin from reduction mammoplasties or abdominoplasties as described (11,29). Briefly, the subcutaneous fat and deep dermis were removed, and the remaining tissue was incubated in 0.25% trypsin in PBS. After 16 h, the epidermis was separated from the dermis with forceps, and the keratinocytes were scraped into DMEM. The keratinocyte suspension was added to fresh DMEM supplemented with 5% fetal calf serum and 0.1% penicillin/streptomycin. Under these culture conditions, keratinocytes proliferate, migrate, differentiate, and cornify similarly to cells in vivo (29). A specified amount of keratinocyte suspension was then plated onto tissue culture dishes coated with 1 mg/ml type I collagen, which is necessary for induction of collagenase-1 and keratinocyte adhesion (5,9,11).

In Situ Hybridization
Collagenase-1 mRNA was detected in formalin-fixed tissue samples by hybridization with 35 S-labeled antisense RNA as described (30,31). Punch biopsies (2 mm) of human skin were obtained and grown as explant cultures in serum-containing DMEM for 4 days. Following treatment, the tissue was fixed in neutral buffered formalin for 24 h followed by washing in PBS and dehydration in graded ethanol. Sections of tissue were hybridized with 2.5 ϫ 10 4 cpm/l 35 S-labeled anti-sense or sense RNA overnight at 57°C. After hybridization the slides were washed under stringent conditions, including RNase A treatment, and were processed for autoradiography. After development of the photographic emulsion, slides were stained with hematoxylin-eosin. The specificity of the antisense RNA probe for collagenase-1 and the complete lack of reactivity by the sense probe have been demonstrated in previous studies (3,7).

Migration Assays
Colony Dispersion-Primary human keratinocytes (1 ϫ 10 4 cells) were plated within a siliconized cloning cylinder (6-mm inner diameter; BellCo Glass, Inc., Vineland, NJ) onto collagen-coated dishes. After a 24-h incubation period to allow the cells to attach and become confluent, the cloning cylinder was removed, and cell colonies were allowed to migrate for 96 h at 37°C in a 5% CO 2 humidified incubator. Keratinocytes were fixed in neutral buffered formalin, stained with 1.5% Coomassie Blue, and the area of the colony was determined by digitized scanning analysis. Migration is expressed as the increase in colony area relative to 0-h controls.
Colloidal Gold-Primary human keratinocytes were plated on chamber slides precoated with a mixture of 100 g/ml type I collagen and colloidal gold particles in serum-containing DMEM. Keratinocytes (ϳ330 cells) were added to each chamber, and 20 min later, nonadherent cells were removed and the medium was replaced. Twenty hours after plating, cultures were fixed in 1ϫ Histochoice tissue fixative (Amresco, Solon, OH), washed in PBS, and dehydrated through graded ethanol. Paths of cell migration (phagokinetic tracks) were identified as areas devoid of gold particles. A migration index was determined using image analysis software by measuring the area of the phagokinetic tracks associated with cells in randomly chosen fields under dark field illumination at 100ϫ magnification. For each experiment, all conditions were done in triplicate, and all experiments were repeated at least three times with keratinocytes from individual donors.

Enzyme-linked Immunosorbent Assay (ELISA)
The amount of collagenase-1 accumulated in keratinocyte conditioned medium was measured by indirect competitive ELISA (32). This ELISA is completely specific for collagenase-1, has nanogram sensitivity, and detects active and zymogen enzyme forms, as well as collagenase-1 bound to TIMP or bound to substrate. Results were obtained from triplicate determinations and were normalized to total cell protein as quantified by the BCA protein assay (Pierce) using bovine serum albumin as a standard.

Metabolic Labeling
Post-confluent keratinocytes plated on type I collagen were cultured for 24 h in the presence of serum-containing DMEM control or experimental solutions. The culture wells were then washed and replaced with methionine-free DMEM containing 5% dialyzed fetal calf serum (to remove free amino acids), 1 mM sodium pyruvate, 2 mM L-glutamine, 0.1 mM each of non-essential amino acids, 50 Ci/ml [ 35 S]methionine (ICN Radiochemicals, Irvine CA), and the identical concentrations of experimental reagents. Conditioned medium was collected 24 h later and analyzed by immunoprecipitation.

Immunoprecipitation and Western Immunoblotting
A specific polyclonal antiserum (33) was used to immunoprecipitate collagenase-1 from keratinocyte conditioned medium as described (34). To immunoprecipitate the EGFR, cell layers were washed with PBS and treated for 10 min at room temperature with cell lysis buffer (1.5% Triton X-100, 50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM CaCl 2 , 1 mM MnCl 2 , 1 mM MgCl 2 , 2 mM phenylmethylsulfonyl fluoride, 0.02 mg/ml aprotinin, and 0.01 mg/ml leupeptin). All samples were precleared with protein A-Sepharose (Zymed Laboratories Inc., San Francisco, CA), and supernatants were incubated with collagenase-1 or EGFR polyclonal (35) antibodies for 1 h at 37°C then overnight at 4°C. Immune complexes were precipitated with protein A-Sepharose and washed extensively. For visualization of collagenase-1, radiolabeled protein was resolved by polyacrylamide gel electrophoresis and visualized by fluorography as described previously (36). Immunoprecipitated EGFR was resolved by polyacrylamide gel electrophoresis and electrophoretically transferred to Immobilon-P polyvinylidene difluoride membrane (Millipore Corp., Bedford, MA) using a semidry blotting apparatus (Bio-Rad). Tyrosine phosphorylation of the EGFR was then visualized by incubating the membrane with anti-PY-20 (primary) and anti-mouse IgG-horseradish peroxidase (secondary) antibodies followed by detection using the ECL system (Amersham Corp., Arlington Heights, IL) according to the manufacturer's instructions. Total incorporated radioactivity (new protein synthesis) was determined from keratinocyte conditioned medium by trichloroacetic acid precipitation as described previously (37).

RNA Analysis
Collagenase-1, EGFR, EGF, TGF-␣, amphiregulin, and GAPDH mRNAs were detected by modification of previously described reverse transcription-polymerase chain reaction (RT-PCR) assays (9,38,39). HB-EGF mRNA was amplified using primers designed with Gene-Works software (Oxford Molecular Group Inc., Campbell, CA). The 5Ј-sense primer was complementary to bases 791-810 in exon 4, and the 3Ј-antisense primer was complementary to bases 1017-1036 in exon 6 of human HB-EGF (40). These primers amplify a fragment across three exons; thus, the 246-base pair cDNA produced from HB-EGF mRNA would be easily distinguished from contaminating DNA or preprocessed mRNA. In addition, all resultant cDNAs from each of the primer pairs contain restriction sites specific to the mRNA amplified and each product was subjected to digestion analysis to verify specificity. The primers used to amplify each mRNA, annealing temperatures, and resulting product sizes are listed in Table I.
Total RNA was isolated from cultured keratinocytes by phenol-chloroform extraction (41) and treated with RQ1 RNase-free DNase (Promega, Madison, WI) to remove any contaminating DNA as described (42). DNase-treated RNA was reverse transcribed with random hexamers using kit reagents and under the manufacturer's recommended conditions (GeneAmp RNA PCR kit, Perkin Elmer Cetus, Norwalk, CT). Signal strength for each RT-PCR cDNA product increased exponentially between 21 and 29 cycles using 10 ng of DNase-treated RNA and increased linearly between 1 and 25 ng of total RNA at 25 cycles. To amplify each mRNA, we used 10 ng of total RNA and 25 cycles. PCR products were separated through a 2% agarose gel and visualized by ethidium bromide staining. Specificity was determined by overnight transfer to Hybond N ϩ membrane (Amersham Corp.) followed by Southern hybridization with a radiolabeled product-specific oligonucleotide probe (EGF family members and EGFR) or radiolabeled cDNA probes (collagenase-1 and GAPDH). The probes used to detect EGF family members and EGFR transcripts following RT-PCR were developed to recognize only specific sequences that were amplified in the PCR reaction. For each oligonucleotide, a BLAST search was performed, and no sequence similarity to other known cDNAs was found. In addition, a parallel reaction was run without reverse transcriptase to assure that products were not generated by contaminating DNA. Oligonucleotide probes were labeled by terminal transferase (Boehringer Mannheim), and cDNA probes were labeled by random priming (Amersham Corp.) with [␣-32 P]dCTP (ICN Radiochemicals). Following hybridization, the membranes were washed and exposed to x-ray film for an appropriate duration.

Transient Transfection
Expression constructs contained the chloramphenicol acetyltransferase (CAT) reporter gene with a 2.2-kilobase pair fragment of the human collagenase-1 promoter, pCLCAT (provided by Dr. Steven Frisch, Burnham Institute, La Jolla, CA), and pSV-␤-galactosidase (Promega, Madison, WI). Primary keratinocytes were transfected at 50% confluence with 2 g/well of each construct using 10 l/well Lipo-fectAMINE (LifeTechnologies, Inc.). After 16 h, the medium was replaced, and the cells were incubated with control or experimental solutions for an additional 24 h and harvested. Detection of ␤-galactosidase and CAT activity was performed as described previously (37). Under the culture conditions used, we routinely achieve 85% transfection efficiency of keratinocytes (9, 11, 37).

In Vivo Studies
Partial thickness thermal burns were created on the dorsal skin of pigs as described (43). Briefly, two domestic male pigs, each weighing 35 pounds, were anesthetized with ketamine and xylazine, and anesthesia was maintained with halothane inhalation. The dorsal skin was chemically depilitated, and six partial thickness burns measuring 3 ϫ 3 cm were created by contact for 10 s with a solid brass block weighing 714 g heated previously in a water bath maintained at 70°C. The six burns were arranged in two rows of three burns on each side of the spine. Blister roofs were removed and two burns were treated topically with 3 ml of Silvadene ® cream (Marion Laboratories, Kansas City, MO) containing 20 g/ml tyrphostin 1478 (Calbiochem) (44). Two burns were treated with Silvadene ® cream alone, and two burns were left untreated. Each burn was individually covered with a 6 ϫ 6-cm 2 adhesive occlusive dressing (Steri-Drape™ 2, 3M, St. Paul, MN). Dressings were removed daily, the burns were retreated, and fresh dressings were applied. Five days after injury, the pigs were sacrificed and a fullthickness biopsy was taken diagonally across each burn, fixed in 10% buffered formalin, and embedded in paraffin. Sections were stained with hematoxylin and eosin. The extent of epithelial healing for each burn was calculated by measuring the distance between intact epithelial edges divided by the total length of the wound and was expressed as percentage of re-epithelialization. Epithelial healing for each of the three treatment groups was averaged and compared for statistical significance by analysis of variance and Tukey's HSD post-test.

Blockade of EGFR Signaling Inhibits Migration on Type I
Collagen-We used specific inhibitors of EGFR occupancy and tyrosine kinase activity to determine if signaling through this receptor was related to collagenase-1 induction and keratinocyte migration across type I collagen. Migration was assessed using colloidal gold and colony-dispersion motility assays as described (10). In the colloidal gold assay, keratinocytes were plated on chamber slides coated with a colloidal gold type I collagen mixture, and migration was quantified 20 h later. In the colony dispersion assay, cells were cultured within cloning cylinders for 24 h. After cell attachment to the matrix, the cloning cylinders were removed and migration was assessed at 96 h. As we have shown (10), keratinocytes migrated efficiently over type I collagen in both migration assays (Fig. 1, A and D).

Blockade of EGFR Function Inhibits Keratinocyte
Collagenase-1 Production-Because keratinocyte migration across type I collagen is dependent on both collagenase-1 activity and EGFR function, we determined if EGFR signaling was required for collagen-mediated collagenase-1 production. Human keratinocytes were plated on native type I collagen (all conditions shown except for "gelatin") or heat-denatured collagen (gelatin) and treated with inhibitors of EGFR signaling, and accumulation of collagenase-1 protein in the conditioned medium was quantified by ELISA. We used the IL-1 RA as a control in these experiments because signaling through this receptor is required for induction of collagenase-1 expression in fibroblasts (46,47). Treatment with PD153035 or EGFR Ab 528 inhibited collagen-mediated collagenase-1 production in a dose-dependent manner, reducing enzyme levels to those detected in cells plated on gelatin (negative control) ( Fig. 2A). In contrast, the IL-1 RA did not reduce collagenase-1 expression but had a slight stimulatory effect. Keratinocyte viability was not af-fected by treatment with either EGFR inhibitor as assessed by total protein synthesis (data not shown).
We also found that other known stimulators of collagenase-1 expression by keratinocytes required intermediate signaling through the EGFR. Collagen-mediated induction of collagenase-1 expression was augmented by transforming growth factor-␤1 (TGF-␤1), hepatocyte growth factor/scatter factor, phorbol ester, and interferon-␥ (IFN-␥). However, in the presence of PD153035 (Fig. 2, B and C), all such soluble factor-stimulated enzyme production was inhibited. In contrast, IL-1 RA had no effect on collagenase-1 stimulation by each of these factors (data not shown). Therefore, agents that stimulate collagenase-1 production by keratinocytes, whether matrix or soluble, appear to require an obligatory intermediate EGFR signaling step.
EGFR Blockade Inhibits Collagen-induced Collagenase-1 mRNA Levels and Promoter Activity in Keratinocytes-To gain an understanding of the molecular level at which EGFR blockade inhibits collagen-directed collagenase-1 production, total RNA was harvested from keratinocytes grown on collagen alone, or on collagen and treated with PD153035 or EGFR mAb 528 for 24 h. To detect collagenase-1 mRNA present in each sample, we used a semiquantitative RT-PCR assay (9). Only collagenase-1-specific products are detected by Southern hybridization or by ethidium bromide staining. Collagenase-1 mRNA was readily observed in keratinocytes cultured on collagen alone (Fig. 3A). In contrast, and paralleling the protein data, treatment of keratinocytes with PD153035 (500 nM) or EGFR mAb 528 (1.0 g/ml) inhibited collagenase-1 mRNA expression by 96% and 92%, respectively, indicating pretranslational regulation.
To determine if collagenase-1 transcription was affected following inhibition of EGFR signaling, keratinocytes on collagen were transfected with a CAT expression construct containing 2.2 kilobase pairs of the human collagenase-1 promoter. As shown in previous studies, changes in the activity of this promoter construct parallel changes in the activity of the endogenous collagenase-1 gene (48), and the activity of the promoter construct is about 5-fold greater in keratinocytes plated on native collagen than in cells on heat-denatured collagen (gelatin) (9). The normalized level of CAT activity detected in keratinocytes plated on collagen was reduced in cells treated with PD153035 or EGFR Ab 528 by 75% and 87.5%, respectively (Fig. 3B), indicating that inhibition of EGFR function blocks collagenase-1 production by a transcriptional mechanism.
Keratinocyte Contact with Type I Collagen Induces EGFR Phosphorylation-Becuase keratinocyte contact with collagen is the primary and requisite inductive event for collagenase-1 expression (9, 10) and because EGFR blockade inhibits enzyme production, we reasoned that EGFR activation should follow initial matrix binding. To assess this temporal relationship, we analyzed changes in keratinocyte steady-state mRNA levels of collagenase-1, EGF family members, and the EGFR at different time points after plating on collagen (0 -48 h) (Fig. 4A). As we reported previously (9), no collagenase-1 mRNA was detected in keratinocytes prior to matrix contact (Fig. 4A, C'ase-1, 0-h). Low levels of collagenase-1 mRNA were observed as early as 2 h after contact with collagen, and expression increased markedly and progressively over the next 8 h. Collagenase-1 mRNA dropped to lower levels between 24 and 48 h after plating, coincident with the keratinocytes reaching confluence. Expression of the mRNAs for EGF, TGF-␣, and amphiregulin paralleled that of matrix-stimulated collagenase-1 expression (Fig.  4A).
In contrast, mRNAs for HB-EGF and EGFR were constitutively expressed at high levels in keratinocytes prior to collagen binding (Fig. 4A). EGFR mRNA became undetectable following 2 h of contact with collagen (Fig. 4A, EGFR), but then increased progressively from 4 to 24 h, diminishing slightly at 48 h. HB-EGF mRNA was expressed at near-constant levels throughout the entire time course, except for a single drop at 4 h.
We next determined whether collagen binding induced EGFR activation. Primary keratinocytes were plated on collagen, and total cell lysates were harvested over the next 120 min (Fig. 4B). Phosphorylated species were immunoprecipitated with an anti-phosphotyrosine monoclonal antibody, and phosphorylated EGFR was detected by Western immunoblotting.

FIG. 2. Blockade of EGFR signaling inhibits matrix-and soluble factor-induced collagenase-1 production by keratinocytes.
Primary human keratinocytes were cultured on type I collagen (collagen alone) or heat-denatured collagen (gelatin) until confluent. A, cells on collagen were treated with PD153035 (100 or 500 nM), EGFR blocking antibody 528 (0.1 or 1.0 g/ml), or IL-1 RA (250 or 500 ng/ml). B and C, keratinocytes on collagen were treated with TGF-␤ (25 ng/ml), hepatocyte growth factor (25 ng/ml), phorbol ester (20 ng/ml), or IFN-␥ (1000 units/ml) in the presence or absence of PD153035 (500 nM). Collagenase-1 protein in the conditioned medium after 72 h was quantified by ELISA, and values were normalized to total protein content. Data shown are the means Ϯ S.D. of triplicate observations from the same cell preparation and are representative of up to four separate experiments.

FIG. 3. Blockade of EGFR signaling inhibits collagen-induced collagenase-1 mRNA levels and promoter activity in keratinocytes. Primary human keratinocytes were cultured on type I collagen.
A, keratinocytes were treated with vehicle control (collagen alone), PD153035 (100, 500 nM), or EGFR blocking Ab 528 (0.1 or 1.0 g/ml). Total RNA was harvested 24 h later and processed for collagenase-1 or GAPDH RT-PCR as described under "Experimental Procedures." A control sample was processed without reverse transcriptase (ϪRT) to assure RNA purity. B, keratinocytes cultured on type I collagen were transfected at ϳ60% confluence with a human collagenase-1 promoter construct containing the CAT reporter gene and a construct containing the ␤-galactosidase gene for normalization of transfection efficiency. After 16 h, keratinocytes were washed and treated with vehicle control (collagen alone), or with media containing PD153035 or EGFR blocking Ab 528 at the indicated concentrations. CAT activity was quantified by scintillation counting of non-acetylated and acetylated chloramphenicol and shown as a percent conversion to the acetylated forms normalized to ␤-galactosidase activity.
EGFR was not phosphorylated in freshly isolated keratinocytes from normal skin (Fig. 4B, 0 min). In contrast, marked phosphorylation of the EGFR was observed as early as 10 min following contact with collagen and persisted up to 120 min (Fig. 4B). Therefore, keratinocyte contact with native type I collagen rapidly induces phosphorylation of the EGFR. Furthermore, despite EGFR mRNA down-regulation at 2 h following matrix contact (Fig. 4A), the receptor remains present and functional on the cell surface as evidenced by continued phosphorylation (Fig. 4B). Figs. 1-3, inhibitors of EGFR function blocked collagen-mediated collagenase-1 expression as measured by enzyme protein and mRNA levels at 24 h. We next performed a time course of collagenase-1 expression in the presence of EGFR inhibitors to determine if the initial induction of collagenase-1 by collagen was mediated by EGFR signaling. Freshly isolated keratinocytes from normal skin were incubated with vehicle control or PD153035 (500 nM) for 2 h prior to plating on collagen. After plating, total RNA was isolated over a time course (0 -24 h), and collagenase-1 mRNA was assessed by RT-PCR. Surprisingly, the initial matrix-induced expression of collagenase-1 at 4 h was unaffected by inhibition of EGFR signaling (Fig. 5, A and B, ϩPD153035). In contrast, collagenase-1 mRNA was slightly diminished at 6 h and almost completely absent at 24 h in cells treated with PD153035 when compared with vehicle controls (Fig. 5, A and B). To ensure that PD153035 inhibited EGFR activation, keratinocytes were plated on collagen and treated with vehicle, 30 ng/ml EGF, or EGF ϩ PD153035 (500 nM). Stimulation of collagenase-1 expression by EGF was inhibited by PD153035 (data not shown). Therefore, the sustained expression of collagenase-1 mRNA requires EGFR signaling, whereas matrix contact alone is sufficient to induce enzyme expression.

EGFR Activity Mediates Sustained, but Not Early Collagendirected Keratinocyte Collagenase-1 Expression-In the experiments shown in
Heparin-binding Epidermal Growth Factor Is the Ligand That Mediates EGFR-dependent Sustained Collagenase-1 Ex- FIG. 5. Sustained keratinocyte collagenase-1 expression is mediated via EGFR signaling. Keratinocytes were isolated from normal human skin as described under "Experimental Procedures" and pretreated for 2 h with vehicle control or PD153035 (500 nM). After pretreatment some keratinocytes were collected for RNA isolation (0 h). The remaining keratinocytes were plated on collagen-coated dishes in the continued presence of vehicle control or PD153035 and total RNA was isolated over a time course (0 -24 h). A, collagenase-1 mRNA was amplified by RT-PCR analysis of DNase-treated RNA and visualized by Southern hybridization with a radiolabeled cDNA probe. Duplicate samples were processed without reverse transcriptase (ϪRT) to assure RNA purity. B, collagenase-1 mRNA hybridization bands were quantified by scanning digitized densitometry and values were plotted over the time course.
FIG. 4. Collagenase-1, EGF, TGF-␣, amphiregulin, HB-EGF, EGFR expression, and EGFR activation in keratinocytes cultured on type I collagen. Normal human skin was processed for keratinocyte isolation as described under "Experimental Procedures." As the trypsin-dispersed cell suspension was prepared, some cells were collected for RNA isolation (0 h). The remaining keratinocytes were plated on collagen-coated dishes. A, total RNA was isolated over a time course (0 -48 h) and the mRNAs for collagenase-1, EGF, TGF-␣, amphiregulin, HB-EGF, EGFR, and GAPDH were amplified by RT-PCR and analyzed by Southern hybridization with product-specific cDNA (c'ase-1 and GAPDH) or oligonucleotide probes (all others). Duplicate samples were processed without reverse transcriptase (ϪRT) to assure RNA purity. B, total proteins were isolated from keratinocytes grown on collagen (1 mg/ml) at the indicated time points and processed for immunoprecipitation of tyrosine-phosphorylated species followed by Western immunodetection of the EGFR as described under "Experimental Procedures." Results presented in A and B are representative of data obtained from at least three individual skin donors.
pression-Members of the EGF family, namely EGF and TGF-␣, up-regulate matrix metalloproteinase expression in several cell types, including keratinocytes (49,50). Furthermore, collagen-directed collagenase-1 expression in keratinocytes is enhanced by treatment with EGF (9, 10). Recent evidence has shown that autocrine production of HB-EGF is likely responsible for potentiation and transmission of the healing response during epidermal wound repair (51), yet effects of this cytokine on collagenase-1 expression have not been studied. Because HB-EGF was the only EGFR ligand constitutively expressed in keratinocytes from intact skin (Fig. 4A), we hypothesized that sustained collagen-mediated production of keratinocyte collagenase-1 is regulated via a HB-EGF/EGFR autocrine loop mechanism. We treated primary keratinocytes with several EGF family members to determine their effect on collagen-directed collagenase-1 expression (Fig. 6A). Collagen-ase-1 production, as assessed by immunoprecipitation of metabolically labeled proteins, was induced in keratinocytes cultured on collagen alone, and this expression was slightly enhanced in cells treated with EGF (Fig. 6A). Collagenase-1 production was substantially increased in keratinocytes treated with HB-EGF or TGF-␣, whereas cells treated with amphiregulin showed no stimulation above that induced by collagen. Thus, of the EGF family members tested, HB-EGF and TGF-␣ were the most potent modulators of matrix-mediated collagenase-1 production by keratinocytes, making them likely candidates for autocrine activation of the EGFR following contact with collagen.
To address this issue, keratinocytes were isolated from normal human skin and pre-incubated for 2 h with vehicle control or neutralizing antibodies to HB-EGF or TGF-␣ prior to plating on collagen. After plating, total RNA was harvested over the next 0 -24 h, and collagenase-1 mRNA was assessed by RT-PCR. The expression of collagenase-1 by keratinocytes was inhibited by treatment with the HB-EGF neutralizing antibody at both time points tested (Fig. 6B, 8 and 24 h). In contrast, collagenase-1 expression at 24 h was unaltered by TGF-␣ neutralizing antiserum (Fig. 6C). Therefore, the sustained production of keratinocyte collagenase-1 in the presence of collagen requires an intermediate HB-EGF/EGFR autocrine signaling step.

Blockade of EGFR Signaling Inhibits ex Vivo Epidermal Collagenase-1 Expression and Prevents Re-epithelialization of
Porcine Burn Wounds-We treated ex vivo wounded human skin explants with inhibitors of EGFR tyrosine kinase activity to determine the requirement, at the tissue level, of EGFR activation for keratinocyte collagenase-1 expression. Punch biopsies of normal human skin were cultured for 4 days with vehicle control or PD153035 (500 nM). As demonstrated by in situ hybridization of the control specimens, collagenase-1 mRNA was expressed by migrating keratinocytes at the leading edge of re-epithelialization (Fig. 7, A and A'). Collagenase-1 was also expressed by some dermal fibroblasts. In marked contrast to control injured skin, no expression of collagenase-1 mRNA was evident in wound-edge keratinocytes in specimens treated with PD153035, whereas fibroblast production was unaffected (Fig. 7, B and B').
Because EGFR signaling is required for sustained keratinocyte collagenase-1 expression both in vitro and ex vivo, and because the activity of this enzyme is essential for cell migration across type I collagen (10), we next determined if EGFR inhibition affected re-epithelialization of in vivo wounds. Partial thickness burn wounds were created in pigs and were treated with Silvadene ® cream containing tyrphostin 1478 (20 g/ml), a highly potent and specific EGFR tyrosine kinase inhibitor (52), vehicle control (Silvadene ® cream), or occlusive dressing only. All wounds were covered with occlusive dressing. In tyrphostin 1478-treated wounds, epithelial healing was significantly impaired when compared with burns treated with vehicle control or covered with occlusive dressing only (Fig. 8,  A-D). Specifically, wounds treated with tyrphostin 1478 had re-epithelialized only 22 Ϯ 16% (mean Ϯ standard error) of the surface of the burn, whereas wounds treated with vehicle control had closed an average of 90 Ϯ 5% of the burn area (p ϭ 0.008). Burns covered with occlusive dressing only had reepithelialized an average of 82% Ϯ 12% of the wound surface (p ϭ 0.016 versus 1478-treated wounds). There was no significant difference between epithelial healing of burns treated with vehicle control or covered with occlusive dressing (p ϭ 0.88).
To verify that tyrphostin 1478 blocked collagenase-1 expression, human keratinocytes plated on type I collagen were treated with PD153035 (100 -500 nM) or tyrphostin 1478 (0.1-FIG. 6. Sustained collagen-directed keratinocyte collagenase-1 expression is dependent on HB-EGF activity. Primary human keratinocytes were cultured on type I collagen until confluent. A, keratinocytes were treated for 24 h with vehicle control (collagen alone), HB-EGF (25 ng/ml), amphiregulin (25 ng/ml), EGF (30 ng/ml), or TGF-␣ (30 ng/ml). Proteins were metabolically labeled in the presence of each solution for an additional 24 h, and conditioned medium was analyzed by immunoprecipitation for collagenase-1. Results from one cell population are shown and represent data obtained from three separate skin donors. B, keratinocytes were cultured on collagen over a time course (0 -24 h) and treated with vehicle control or with a HB-EGF neutralizing antibody (10 g/ml). Total RNA was harvested at the indicated time points and processed for collagenase-1 or GAPDH RT-PCR as described under "Experimental Procedures." A control sample was processed without reverse transcriptase (ϪRT) to assure RNA sample purity. C, keratinocytes were cultured on collagen and treated with vehicle control or with a TGF-␣ neutralizing antibody at the indicated concentrations. Total RNA was harvested 24 h later and processed for collagenase-1 RT-PCR as described under "Experimental Procedures." A control sample was processed without reverse transcriptase (ϪRT) to assure RNA sample purity. Results presented in B and C are representative of data obtained from at least two separate skin donors.
1.0 M) for 72 h and collagenase-1 secreted into conditioned medium was quantified by ELISA (Fig. 8E). Both PD153035 and tyrphostin 1478 blocked collagen-induced collagenase-1 expression in a dose-dependent manner (Fig. 8E). Taken together, these findings support the hypothesis that inhibition of re-epithelialization in porcine burn wounds was due, at least in part, to blocking collagenase-1 production. DISCUSSION Activation of the EGFR in basal keratinocytes following injury provides several critical signals required for proper healing of the tissue defect. For example, the wound-associated keratins 6 and 16 are markedly up-regulated (53), and their promoters contain regulatory elements that are responsive to EGFR activation (24). Continued signaling through the EGFR promotes Bcl-X-L-mediated prevention of keratinocyte apoptosis, thereby potentiating re-epithelialization by maintaining survival of the migrating cells (54). We show here that the sustained expression of collagenase-1 in two systems that mimic epidermal healing (keratinocyte migration on type I collagen and ex vivo skin explants) requires an EGFR autocrine loop mechanism. In previous studies, we showed that collagenase-1 expression is induced at the onset of re-epithelialization as keratinocytes migrate onto dermis, and that collagenase-1 activity is required for cell migration over a type I collagen substratum (10,55). Thus, EGFR activation is central to yet another important phase of the wound repair process.
Our findings that both matrix-and soluble factor-induced collagenase-1 expression in keratinocytes require autocrine signaling through the EGFR (Fig. 2) draw interesting parallels to reports in human fibroblasts showing that all up-regulators of collagenase-1 production operate via an IL-1␣/IL-1␣ receptor autocrine loop. Fini and colleagues (46,47) have demonstrated that only IL-1␣ directly stimulates collagenase-1 gene transcription in fibroblasts. All other inducers, ranging from cytokines such as TNF-␣ to phorbol esters to agents causing cytoskeletal rearrangement, must first stimulate IL-1␣ gene transcription and subsequent release of IL-1␣ protein. This cytokine then binds to the cell-surface IL-1␣ receptor, resulting in triggering of collagenase-1 expression. Consequently, the addition of an IL-1␣ receptor antagonist to fibroblast cultures blocks collagenase-1 gene expression by any stimulating agent. Our findings suggest an analogous role for HB-EGF/EGFR in keratinocytes and offer the possibility that autocrine signaling pathways may represent a general biologic mechanism for inducing the expression of collagenase-1, and perhaps even other MMPs, in different cell types.
Our results implicate HB-EGF as the key ligand that mediates autocrine signaling through the EGFR, resulting in sustained keratinocyte production of collagenase-1. Evidence demonstrating multiple effects of HB-EGF on keratinocytes has accumulated through studies of its role in epidermal wound repair. This newer member of the EGF family is synthesized by multiple cell types, including vascular endothelium, smooth muscle, and keratinocytes (56). Exogenously added HB-EGF,  B and B'). The tissue was fixed, and sections were hybridized with a 35 S-labeled antisense RNA probe specific for collagenase-1 mRNA. In the control section (A and A'), prominent autoradiographic signal for collagenase-1 was detected in basal keratinocytes (arrows) at the migrating front of the epidermis (E). Collagenase-1 mRNA expression was also observed in several fibroblasts (arrowheads) within the dermis (D). In contrast, explants treated with PD153035 (B and B') showed no detectable collagenase-1 mRNA signal in keratinocytes at the wound margin (arrows), whereas expression was unchanged in dermal fibroblasts (arrowheads) when compared with controls (A and A'). TGF-␣, or EGF promotes autoinduction of HB-EGF mRNA, suggesting that this protein is an autocrine growth factor for these cells (56). The requirement for activation of the EGFR to sustain matrix-induced collagenase-1 production is an example of wound keratinocyte phenotype modulation by HB-EGF. Although HB-EGF mRNA is expressed in freshly isolated keratinocytes from normal skin, the membrane-bound form of this protein may not be processed and released in a soluble form until later in the wound healing response. In fact, other groups have shown that soluble HB-EGF does not appear in conditioned medium from excisional wounds or human skin explants until 24 h after injury (51,57). These findings correlate with our observations that complete inhibition of collagenase-1 expression by blocking EGFR or HB-EGF activity does not occur until 24 h following keratinocyte contact with collagen.
Our findings indicate that matrix-induced keratinocyte collagenase-1 expression involves two distinct pathways, an initial response not requiring EGFR activation and a sustained response obligatory to EGFR activation. Both of these responses are integrin-mediated, but early collagenase-1 expression is EGFR-independent. Our data demonstrate that ␣ 2 ␤ 1 integrin activation alone is sufficient to induce the early (0 -8 h) expression of collagenase-1 in keratinocytes following contact with type I collagen. However, the sustained production (Ն8 h) of this MMP requires signaling through the EGFR in addition to ␣ 2 ␤ 1 binding (Figs. 5 and 6B). Indeed, blocking ␣ 2 ␤ 1 activity blocks all matrix-induced collagenase-1 production (10), whereas only the sustained expression is inhibited when EGFR activity is blocked (this report) (Fig. 5). Stimulation by both integrin adhesion and cell binding of a soluble factor are required for a variety of cellular responses during tissue morphogenesis. Unique to our system, however, is that both integrin and EGFR signaling are required only after prolonged exposure to type I collagen (the inducing stimulus).
Similar to our findings, EGFR autophosphorylation is induced by contact of glomerular epithelial cells with type I collagen (58). Our data demonstrate that EGFR (and HB-EGF) mRNA is expressed in keratinocytes freshly isolated from intact skin and that the receptor is rapidly autophosphorylated following contact with type I collagen. These findings represent further examples of how keratinocytes in unwounded skin are "primed" to respond to injury. Indeed, keratinocytes in intact skin express endogenous ␣ 2 ␤ 1 integrin (59). Interestingly, although the EGFR is autophosphorylated rapidly following keratinocyte binding to collagen, collagenase-1 expression does not require EGFR signaling until later in the healing response. We speculate, however, that other early functions of the wound keratinocyte, such as non-enzymatic migration-related events or cell proliferation may require early EGFR signaling, although this remains to be determined.
In addition to our in vitro data, we also show a role for EGFR signaling in the re-epithelialization of porcine burn wounds in vivo. Blocking receptor activity with a specific inhibitor of the EGFR tyrosine kinase (tyrphostin 1478) inhibited keratinocyte collagenase-1 production and markedly delayed re-epithelialization when compared with normal controls. Similarly, treat-FIG. 8. Re-epithelialization of partial-thickness porcine burn wounds is inhibited by blocking EGFR activity. Partial thickness thermal burns were created on the dorsal skin of pigs and treated topically with Silvadene ® cream (A, vehicle control), Silvadene ® cream containing 20 g/ml tyrphostin 1478 (B) or occlusive dressing only (C). All wounds were covered with occlusive dressing. The two large arrowheads in each panel mark the original wounded edges of the epidermal burn. The extent of epithelial healing for each burn was assessed by measuring the distance between intact epithelial edges divided by the total length of the wound and expressed as percent re-epithelialization Ϯ S.E. D, epithelial healing for the three treatment groups was averaged and compared for statistical significance by analysis of variance and Tukey's Honest Statistical Difference post-test. E, primary human keratinocytes were cultured on type I collagen until confluent. Cells were treated with vehicle control (collagen alone), PD153035 (100 or 500 nM), or tyrphostin 1478 (0.1-1.0 M). Collagenase-1 protein in 72-h conditioned medium was quantified by ELISA. Data shown are the means Ϯ S.D. from triplicate observations from the same cell preparation and are representative of at least two separate skin donors. ment of ex vivo human skin punch biopsies with PD153035 inhibited collagenase-1 production in keratinocytes at the wound edge. Taken together, these data suggest that blocking EGFR activity inhibits keratinocyte migration across the dermal matrix and that this is due, at least in part, to inhibition of collagenase-1 expression. Although blocking EGFR signaling may inhibit other biological events necessary for the complete wound repair (e.g. cell proliferation), we suggest that EGFRdependent collagenase-1 expression is critical for sustained keratinocyte migration and normal re-epithelialization.
Our findings in this report add substantially to understanding the mechanisms that regulate collagenase-1 expression during wound repair. Previously, we found that collagenase-1 gene transcription is induced following injury when keratinocytes move off their underlying basement membrane and contact type I collagen of the dermal matrix (3,7,5). We identified the cell-surface recognition integrin as ␣ 2 ␤ 1 and showed that collagenase-1 activity was requisite for keratinocyte migration over a type I collagen substratum (10). Our present data indicate that intact skin is primed and ready to respond to injury with high endogenous levels of keratinocyte EGFR and HB-EGF mRNA. Upon keratinocyte ␣ 2 ␤ 1 binding to type I collagen, EGFR phosphorylation occurs within minutes. Collagenase-1 mRNA levels are induced within 2 h, however, by a mechanism that is EGFR-independent. Nevertheless, as collagenase-1 expression continues, the sustained high levels of enzyme production from Ն8 h following contact with type I collagen are dependent on an EGFR/HB-EGF autocrine signaling loop. Since re-epithelialization of even minor wounds takes days, this sustained collagenase-1 expression is likely essential for most keratinocyte migration and for complete re-epithelialization. Important unanswered questions raised by our findings include: 1) what signaling pathway does ␣ 2 ␤ 1 use to transmit the rapid initial, EGFR-independent induction of collagenase-1?; and 2) upon completion of re-epithelialization, does the cessation of collagenase-1 production involve the dismantling of the EGFR/HB-EGF autocrine loop? Studies to address these important questions are currently in progress.