Laminin-5 Inhibits Human Keratinocyte Migration

EXPERIMENTAL CELL RESEARCH ARTICLE NO.

233, 330–339 (1997)

EX973586

Laminin-5 Inhibits Human Keratinocyte Migration Edel A. O’Toole,* M. Peter Marinkovich,†,‡ Warren K. Hoeffler,†,‡ Heinz Furthmayr,§ and David T. Woodley*,1 *Department of Dermatology, Northwestern University, Chicago, Illinois 60611; †Department of Dermatology and §Department of Pathology, Stanford University School of Medicine, Palo Alto, California 94305; and ‡Dermatology Service, VA Palo Alto Health Care System, Palo Alto, California 94304

Laminin-5 (previously known as kalinin, epiligrin, and nicein) is an adhesive protein localized to the anchoring filaments within the lamina lucida space of the basement membrane zone lying between the epidermis and dermis of human skin. Anchoring filaments are structures within the lamina lucida and lie immediately beneath the hemidesmosomes of the overlying basal keratinocytes apposed to the basement membrane zone. Human keratinocytes synthesize and deposit laminin-5. Laminin-5 is present at the wound edge during reepithelialization. In this study, we demonstrate that laminin-5, a powerful matrix attachment factor for keratinocytes, inhibits human keratinocyte migration. We found that the inhibitory effect of laminin-5 on keratinocyte motility can be reversed by blocking the a3 integrin receptor. Laminin-5 inhibits keratinocyte motility driven by a collagen matrix in a concentration-dependent fashion. Using antisense oligonucleotides to the a3 chain of laminin-5 and an antibody that inhibits the cell binding function of secreted laminin-5, we demonstrated that the endogenous laminin-5 secreted by the keratinocyte also inhibits the keratinocyte’s own migration on matrix. These findings explain the hypermotility that characterizes keratinocytes from patients who have forms of junctional epidermolysis bullosa associated with defects in one of the genes encoding for laminin-5 chains, resulting in low expression and/or functional inadequacy of laminin-5 in these patients. These studies also suggest that during reepithelialization of human skin wounds, the secreted laminin-5 stabilizes the migrating keratinocyte to establish the new basement membrane zone. q 1997 Academic Press

INTRODUCTION

The elements of wound healing include clot formation, inflammation, reepithelialization, angiogenesis, 1 To whom correspondence and reprint requests should be addressed at Department of Dermatology, Northwestern University Medical School, 4-711 Tarry Building, Chicago, IL 60611. Fax: (312) 908-1984.

330

0014-4827/97 $25.00 Copyright q 1997 by Academic Press All rights of reproduction in any form reserved.

AID

ECR 3586

/

6i21$$$261

fibroplasia, wound contraction, and connective tissue remodeling. During the course of reepithelialization, epidermal keratinocytes located at the cut edge of the wound begin to migrate across the wound bed and resurface it [1]. This process involves both cellular locomotion and proliferation [1, 2]. Keratinocyte migration is promoted by collagen I, collagen IV, and fibronectin, whereas laminin-1 inhibits keratinocyte locomotion [3, 4]. Laminin-5 has been shown to be an important adhesive ligand for cultured keratinocytes [5]. This laminin isoform (previously known as nicein, kalinin or epiligrin) is secreted by human keratinocytes and is an important component of anchoring filaments of the lamina lucida of the basement membrane zone [6–8]. It is a heterotrimeric molecule with three chains, a 200-kDa a3 chain, a 140-kDa b3 chain, and a 155-kDa g2 chain, which are products of different genes [9]. The a3 and g2 chains are processed to 165- and 105-kDa polypeptides, respectively [10]. Integrin receptors mediate cell–extracellular matrix and cell–cell interactions and initiate various signaling processes [11]. Basal keratinocytes are the least differentiated keratinocytes in the epidermis. Three major integrins, a2b1, a3b1, and a6b4, are expressed in the normal basal keratinocyte layer [12]. The a2b1 integrin mediates keratinocyte migration on collagens type I and IV [13], whereas keratinocyte motility on fibronectin is mediated by the a5b1 integrin receptor and is RGD-dependent [14]. The hemidesmosome is the predominant adhesion structure associated with cytoplasmic intermediate filaments and uses the a6b4 integrin as its adhesion receptor [15, 16]. Inherited defects of the b4 integrin, collagen XVII (a.k.a. ‘‘the bullous pemphigoid antigen 2’’), and laminin-5 cause blistering at the level of the epidermal–dermal junction in association with decreased hemidesmosome formation [17– 19]. The a3b1 integrin receptor is present in focal adhesions, actin-binding dynamic structures involved in adhesion, and migration [20]. Both the a6b4 and the a3b1 receptors have been shown to bind to laminin5. The a3b1 receptor may be involved in the initial attachment of keratinocytes to laminin-5 [15]. The

05-28-97 12:01:55

eca

LAMININ-5 INHIBITS KERATINOCYTE MIGRATION

a6b4 integrin then may mediate the subsequent stable adhesion of keratinocytes via the hemidesmosome [21]. In this study, we determined that laminin-5 inhibits keratinocyte migration. Its inhibitory effect appears to be at least as great as that of laminin 1, another major cruciate-shaped glycoprotein localized within the lamina lucida space [3]. We found that the inhibitory effect of laminin-5 on keratinocyte motility can be reversed by blocking the cell’s a3 integrin receptor. By inhibiting the expression of keratinocyte-derived endogenous laminin-5 with antisense oligonucleotides to the a3 chain of laminin-5 or an antibody that inhibits the secreted laminin-5, we demonstrated that the endogenous laminin-5 secreted by the keratinocyte also inhibits keratinocyte migration on matrix. METHODS Cell Culture Human keratinocytes obtained from surgical skin specimens and neonatal foreskins were cultured in low-calcium, serum-free MCDB 153 medium as described by Boyce and Ham [22] and as modified by O’Keefe and Chiu [23]. Cultures were passaged twice to remove any contaminating fibroblasts [24]. All experiments were performed at least three times, with three different keratinocyte donors.

331

cloning chamber which, when attached, completely filled the chamber area. After 2 h, the chamber was lifted, and the keratinocytes were allowed to migrate on the immobilized extracellular matrix. After 12 h, the keratinocytes were fixed with 0.1% formaldehyde in PBS and stained with 0.1% crystal violet and the total area (the original area of the attached cells plus the area filled by the newly migrated cells) of each group of cells was measured using computerassisted densitometry in mm2. The migration index (MI) is the total area minus the area of the cloning chamber. Each triplicate experiment was performed three times. In order to assess migration on endogenous laminin-5, confluent keratinocytes were lysed in 2.5% Triton X-100 in PBS containing 1 mM PMSF, followed by 2 M urea and 1 M NaCl, and then 8 M urea. The remaining substratum (which mainly consists of laminin5 secreted by the keratinocytes [7]) was washed extensively with icecold PBS and used as the extracellular matrix in identical cloning chamber fence assays. In vitro scratch assay. The effect of laminin-5 on keratinocyte migration was also assessed in an in vitro scratch assay as described by Cha and co-workers [31]. Keratinocytes were plated densely in tissue culture wells and after allowing the cells to attach to the substratum for 2 h, a standardized scratch was made through the confluent monolayer with the same sharp plastic device measuring 2 mm in width. The cells were then washed extensively with PBS to remove cellular debris. Laminin-5, laminin-1, or collagen I was added in fresh medium at a concentration of 15 mg/ml. The keratinocytes from the cut edge of the scratch were then allowed to migrate for 12 h and photographed. Second-Agent-Added Motility Assays

Materials Native rat type I collagen was purchased from Collaborative Research (Bedford, MA). Human plasma fibronectin was obtained from Gibco BRL (Gaithersburg, MD). Laminin-1 was prepared from the Englbreath–Holm–Swarm tumor as previously described [25]. Laminin-5 was purified from human squamous carcinoma cell medium as previously described [10]. Monoclonal antibody CD49f to the a6 integrin receptor (clone GOH3) was purchased from Immunotech (Westbrook, ME) [26]. Monoclonal antibodies to the a2 (clone PIE6), a3 (clone PIB5), and a5 (clone PID6) integrin receptors were purchased from Gibco BRL [27]. Monoclonal antibody LH7.2 to type VII collagen was purchased from Sigma (St Louis, MO) [28]. Keratinocyte Migration Assays Track assay. Keratinocyte migration was assessed by the method of Albrecht-Buehler [29] as modified by Woodley et al. [3]. Briefly, colloidal gold salts were immobilized on bovine serum albumincoated coverslips and covered with extracellular matrix components, in this case laminin-5 (5–100 mg/ml), laminin-1 (90 mg/ml), collagen I (15 mg/ml), collagen IV (30 mg/ml), and fibronectin (60 mg/ml). Second to fourth passage keratinocyte cultures were suspended and plated (1000 cells/cm2) on the coverslips and allowed to migrate for 18–20 h. The cells were fixed in 3% formaldehyde in PBS and examined under dark-field optics with a video camera attached to a computer equipped with a frame grabber. The computer analyzes 15 nonoverlapping fields under each experimental condition and determines the percentage area of each field consumed by cell migration tracks, a so-called migration index. Confirmation of a difference in migration as statistically significant requires rejection of the null hypothesis of no difference between mean migration indices obtained from replicate sets at the P Å 0.05 level with the Student t test [30]. Cloning chamber fence assay. In the cloning chamber assay, tissue culture wells were coated with collagen I, collagen IV, laminin1, laminin-5, or fibronectin. A cloning chamber was placed in the center of each well and 10,000 keratinocytes were plated within each

AID

ECR 3586

/

6i21$$$262

05-28-97 12:01:55

Addition of laminin-5 to cells migrating on collagen. To determine if laminin-5 inhibited the promigratory effect of a collagen I matrix in the colloidal gold track assay, keratinocytes were plated onto collagen I-coated coverslips, as described above, and allowed to attach for 1 h. Then, laminin-5 was added in concentrations from 0.5 to 100 mg/ml, the cells were allowed to continue migrating in the assay for 20 h as usual, and motility was analyzed by computerassisted image analysis as described above. Addition of anti-integrin antibodies to cells migrating on collagen in the presence of laminin-5. Laminin-5 is a putative ligand for the a3b1 and a6b4 receptors. In order to see if the inhibitory effect of exogenous laminin-5 on collagen I-driven motility in the colloidal gold track assay (vide supra and vide infra) could be blocked by nullifying this receptor, keratinocytes were plated onto suboptimal concentrations of collagen I (4 mg/ml) in the track assay and allowed to attach for 2 h, at which time we added antibodies to the a3 and a6 integrin subunits (either alone or in combination) simultaneously with 80 mg/ml of exogenous laminin-5. These antibodies do not affect cell attachment to collagen I in the concentrations used in this study (data not shown). Antibodies to the a2 and a5 integrin receptors were used as negative controls (laminin-5 is not a ligand for these receptor subunits). The cells were then allowed to migrate for 20 h and motility was analyzed as described above. Addition of agents that block endogenous laminin-5 to cells migrating on collagen. Keratinocytes synthesize and secrete laminin-5 [5]. To determine how the endogenous laminin-5 secreted by keratinocytes in the track assay affects motility, cells were plated onto suboptimal concentrations of collagen I-coated coverslips (4 mg/ml) which provided low MIs of between 9 and 11%. The cells were allowed to attach for 2 h, at which time BM165 antibody (an antibody that blocks the cell binding function of laminin-5 [32]) was added in concentrations from 0.5 to 10 mg/ml. Concentrations of BM165 antibody above 10 mg/ml are known to cause keratinocyte detachment [32]. The negative control was keratinocytes migrating on suboptimal amounts of collagen I in the presence of an irrelevant antibody, mouse monoclonal antibody to type VII collagen, LH7.2.

eca

332

O’TOOLE ET AL.

To examine the effect of the deposited laminin-5 in another way, the synthesis of laminin-5 was blocked by an 18-mer antisense oligonucleotide aligning with the signal peptide of the a3 chain mRNA sequence-5*. . .ATGGGATGGCTGAGGATC [33]. The antisense and sense oligonucleotides were synthesized and phosphorothiolated by the Biotechnology Department, Northwestern University. After the keratinocytes were allowed to attach to the substratum for 1.5 h, antisense and sense oligonucleotides were added to cultured keratinocytes to be used in the migration track assay and to the migration track assay dishes themselves at a concentration of 20 mM. Fresh oligonucleotide was added every 6 h for 18 h. Control cultures contained no added oligonucleotide or a sense oligonucleotide. The migration track assays were then analyzed as described above. Experimental and control cultured keratinocytes in the antisense experiments also were analyzed for their secretion and deposition of laminin-5. Briefly, cells were lysed with 2.5% Triton X-100 in PBS and the remaining substratum was washed extensively with sterile PBS. After removal of cellular debris, the keratinocyte matrix was extracted in 0.125 M Tris–HCl, pH 6.8, 2% SDS with 1 mM PMSF, with a plastic policeman. The extracts were vortexed for 1 min, sonicated for 30 s, and centrifuged at 100,000g and the protein concentration was determined according to Lowry [34]. Ten micrograms of protein per lane was subjected to SDS–PAGE and Western immunoblotting according to Towbin [35]. Purified laminin-5 was used as a positive control [10]. The nitrocellulose membranes were incubated with a polyclonal antibody to the g2 chain of laminin-5 (1:1000) for 2 h [36]. Normal rabbit serum at the same dilution served as a negative control. The blots were then incubated with secondary peroxidase-conjugated goat anti-rabbit IgG at a dilution of 1:2000 (Cappel/Organon Teknika, Durham, NC) for 1 h, and the membranes were developed using ECL (Amersham, Arlington, IL).

Keratinocytes plated on collagen I, laminin-1, or laminin-5 at a density of 12,500 cells/cm2 were evaluated for the protein expression of ezrin [37]. After an 18-h incubation on collagen I, laminin-1, or laminin-5, the cell layers were washed in cold PBS three times and then 0.8 ml of 0.125 M Tris–HCl, pH 6.8, 2% SDS (sample buffer) was added to the cell layers. The cell layers were scraped with a plastic policeman and the sample buffer was collected. The extracts were vortexed for 1 min, sonicated for 30 s, and centrifuged at 100,000g. The supernatants were collected and after protein determination, bromophenol blue dye, 5 mM DTT, 1 mM EDTA, and 1 mM PMSF were added to the extract. Thirty micrograms of protein per lane was subjected to SDS–PAGE and Western immunoblotting. The nitrocellulose membranes were incubated with a polyclonal antibody to ezrin, pAs 90.3, at a 1:1000 dilution for 2 h [38]. Normal rabbit serum at the same dilution served as a negative control. The blots were then incubated with secondary peroxidase-conjugated goat anti-rabbit IgG at a dilution of 1:2000 (Cappel/Organon Teknika) for 1 h and the membranes were developed using ECL (Amersham). Immunofluorescence Staining Keratinocytes were cultured in TissueTek chamber slides (Nunc, IL) on collagen I, laminin-1, and laminin-5 for 18 h. The keratinocytes were immersed in periodate–lysine–paraformaldehyde fixative for 10 min at room temperature [39]. The cells were washed several times in PBS to remove fixative and then permeabilized and blocked by incubating for 10 min at room temperature in PBS with 3% BSA, 1% saponin, and 0.05% Na azide. The cells were incubated with ezrin antibody, pAs 90.3, at a concentration of 1:200 in a humidified chamber for 2 h, then washed three times with PBS, 1% saponin, and counterstained with a FITC-conjugated goat antibody to rabbit IgG (1:100 dilution) for 1 h (Cappel/Organon Teknika).

ECR 3586

/

6i21$$$262

RESULTS

Track Assay

Western Immunoblots for Ezrin

AID

FIG. 1. Colloidal gold migration assay: laminin-5 inhibits keratinocyte migration. Coverslips were coated with colloidal gold and keratinocytes were plated on type I collagen 15 mg/ml (A), fibronectin 60 mg/ml (B), laminin-1 90 mg/ml (C), and laminin-5 100 mg/ml (D) and incubated for 18 h. Representative fields were photographed at 40X under dark-field optics. Bar, 150 mm.

05-28-97 12:01:55

Figure 1 is a representative experiment of human keratinocytes in the track assay in which keratinocytes are migrating on immobilized type I collagen, laminin1, laminin-5, and fibronectin. The figure shows representative microscopic fields of the keratinocytes migrating on these connective tissue matrices which are covering the colloidal gold. Human keratinocytes plated on immobilized gold alone, without added matrix molecules, made very small circular tracks that occupied less than 4% of the total field area after 18 h of migration (data not shown). The migration index is the percentage of the field taken up by the motility tracks. In the experiment shown in the figure, the MIs of the cells on no matrix were less than 3% (data not shown). As demonstrated in Figs. 1A and 1B, however, matrices of collagen or fibronectin dramatically promoted motility and produced MIs of approximately 30 and 12, respectively. In contrast to collagen or fibronectin, laminin-5, in various concentrations, had no promigratory effect on keratinocyte migration and produced MIs of 2.6%. As demonstrated in the representative experiment shown in Fig. 1D, regardless of the plating concentrations of laminin-5 used from 2.5 to 100 mg/ml, the keratinocytes did not migrate above the no matrix controls. The lack of keratinocyte motility on laminin-5 is very similar to that seen on a matrix of laminin-1 [3]. In the representative experiment shown in Fig. 1D, laminin-5 was used at 100 mg/ml. On

eca

LAMININ-5 INHIBITS KERATINOCYTE MIGRATION

TABLE 1 Summary of Results of Cloning Chamber ‘‘Fence’’ Assay Area of migration (mm2)

Matrix Collagen I Collagen IV Fibronectin Laminin-1 Laminin-5 No matrix

15.6 13.7 7.0 2.6 2.4 4.0

{ { { { { {

0.5 0.4 0.3 0.2 0.3 0.3

both laminin-1 and laminin-5, the keratinocytes made very small circular tracks but no linear migration tracks (see Figs. 1C and 1D). Cloning Chamber Fence Assay The cloning chamber ‘‘fence’’ assay results are summarized in Table 1. In this assay, the computer-assisted measurements in mm2 are the total area of the newly migrated cells after the cloning chamber ‘‘fence’’ is lifted and the cells have migrated for 12 h, a time point well before the cell doubling time. As shown in Table 1, in the cloning chamber fence assay, the migrating keratinocytes generate MIs of 15 { 0.6, 13.7 { 0.4, and 7.0 { 0.3 on matrices of type I collagen, type IV collagen, and fibronectin, respectively. In contrast, when the keratinocytes were provided with matrices of laminin-1 or laminin-5, MIs were 2.6 { 0.2 and 2.4 { 0.3, respectively, the same or slightly less than the triplicate MIs generated by the albumin and no matrix controls. Identical results were obtained using purified exogenous laminin-5 or endogenously secreted laminin-5 from keratinocyte cultures (data not shown).

333

scratch readily migrated into the scratched area, while very little migration from the cut edge occurred when matrices of laminin-5 or laminin-1 were used (Figs. 2A and 2B, respectively). Because the molecular size of laminin-1 and laminin-5 are different, the differences between these two laminin isoforms in their influence on cell motility may not be due entirely to their intrinsic biological properties, but could be due to the differences in the molarity of the two matrices used in the experiments. As shown previously [3, 13, 14, 30], human keratinocyte migration is dramatically supported by a matrix of type I collagen. As demonstrated in Fig. 3, when human keratinocytes were allowed to migrate on a matrix of type I collagen, but in the presence of increasing amounts of laminin-5 added 30 min after the cells were plated, laminin-5 inhibited the collagen-driven motility in a dose-dependent manner. This inhibition of colla-

In Vitro Scratch Assay In the track assay and the cloning chamber fence assay above, we used concentrations of type I collagen, type IV collagen, fibronectin, and laminin-1 that were optimal for their effects on motility based on previous studies [3, 13, 14, 30]. That is, these concentrations of matrix maximally promoted keratinocyte motility in the case of collagens and fibronectin and maximally inhibited keratinocyte migration in the case of laminin1. In the third independent motility assay, the in vitro scratch assay, we examined the effect of approximately equal amounts of laminin-5, laminin-1, and type I collagen on keratinocyte migration. In this assay, a uniform scratch was made on the bottom of confluent keratinocyte cultures and, after washing extensively with PBS, each matrix was added to the dishes at a concentration of 15 mg/ml. The keratinocytes were allowed to migrate for 12 h and then photographed. As shown in Fig. 2, a representative experiment, when a matrix of type I collagen was used, the cells at the cut edge of the

AID

ECR 3586

/

6i21$$$263

05-28-97 12:01:55

FIG. 2. Laminin-5 is more effective than laminin-1 in inhibiting in vitro wound closure. An in vitro wound was introduced in confluent cultures of human keratinocytes. The cells were incubated with 15 mg/ml of laminin-5 (A), laminin-1 (B), or collagen I (C), diluted in keratinocyte medium, for 12 h before photography under phase-contrast microscopy. These findings are representative of three different experiments. Scale bar, 100 mm.

eca

334

O’TOOLE ET AL.

FIG. 3. Laminin-5 inhibits the promigratory effect of collagen I in a dose-dependent manner. Keratinocytes were plated on a suboptimal concentration of collagen I. After attachment, laminin-5, in increasing concentrations, was added diluted in keratinocyte SFM and the cells were incubated for 18 h. Error bars, SE of three different experiments.

gen-driven motility also occurs with laminin-1, as reported previously [3], but not with albumin. In experiments of this type, the inhibitory effect was seen at very low concentrations of laminin-5 and when greater than 40 mg/ml of laminin-5 was used, keratinocyte migration was essentially negligible. Keratinocytes have two integrin receptors for laminin-5. In order to examine if the inhibitory effect of laminin-5 on collagen-driven motility could be blocked by nullifying these receptors, we added antibody to the a3 and a6 integrin subunits, alone and in combination, to keratinocytes migrating on suboptimal amounts of collagen I in the presence of laminin-5. As shown in Fig. 4, increasing concentrations of a3 integrin antibody added at the same time as an inhibitory concen-

tration of laminin-5 (80 mg/ml) to keratinocytes migrating on collagen I increased the MIs in the track assay from 2 to 6.3%. Identical experiments in which antibodies to the a6 subunit were added with exogenous laminin-5 demonstrated that the a6 subunit antibodies alone had some, but minimal, effect upon reversing the laminin-5-inhibition of collagen-driven motility. When, however, the a3 and a6 antibodies were added in combination at their highest concentrations, the inhibitory effect of laminin-5 was totally reversed (black and white dotted bar at 1:100 dilution), and the cells migrated to the same extent as the no laminin-5 added control (black bar). Antibodies to the a2 and a5 integrin receptors did not reverse the inhibitory effect of laminin-5 (data not shown). Therefore, with regard to nullifying collagen-driven motility by laminin-5, the exogenously added laminin-5 effect can be reversed by antibodies to the a3 and a6 integrin subunits which comprise integrin receptors for laminin-5 but not by antibodies to subunits of integrin receptors for collagen and fibronectin. The fact that the a3 and a6 antibodies in combination had a greater inhibitory effect on the cell’s response to the exogenously added laminin than either antibody alone suggests that both the a3b1 and a6b4 integrins may play a role and may act in concert. We sought to determine if endogenous laminin-5 produced by the keratinocyte played a role in keratinocyte motility. Our first approach to decreasing the expression of endogenous laminin-5 was to add BM165 antibody against the a3 chain of laminin-5 to the colloidal gold migration assay. This antibody is known to inhibit laminin-5 secreted by keratinocytes [32]. As shown in Fig. 5, blocking the endogenously secreted laminin-5 from keratinocytes with this antibody (open triangles) induced keratinocytes apposed to suboptimal amounts

FIG. 4. Effect of anti-a3 and anti-a6 integrin antibodies on the laminin-5-mediated inhibition of keratinocyte migration on collagen I. Human keratinocyte migration after 18 h of culture on suboptimal collagen I alone (black bar), with added laminin-5 (clear bar) and with added a3 and a6 integrin antibodies as described in the figure. Error bars, SE of three different experiments.

AID

ECR 3586

/

6i21$$$263

05-28-97 12:01:55

eca

LAMININ-5 INHIBITS KERATINOCYTE MIGRATION

FIG. 5. Effect of BM-165 antibody on keratinocyte migration. Keratinocytes were plated on suboptimal concentrations of collagen I. After the cells were allowed to attach for 1 h, BM-165 antibody was added in increasing concentrations (open triangle). LH7.2, a mouse monoclonal antibody to collagen VII, was used as a negative control (closed diamond). Error bars, SE of three different experiments.

of a type I collagen matrix to generate increased MIs in the migration track assay. Also, as shown in Fig. 5, the enhancement of cellular motility is dependent upon the concentration of antibody used and the amount of endogenous laminin-5 blocked. Our second approach to blocking endogenous laminin-5 was to treat cultured keratinocytes with antisense and sense oligonucleotides directed against the mRNA for laminin-5. The keratinocyte cultures were treated for 18 h with antisense and sense oligonucleotides and then these treated cells were used in the colloidal gold motility assay. As demonstrated in Fig. 6A, the keratinocytes treated with the antisense oligonucleotide (but not those with the sense oligonucleotide or no added oligonucleotide) demonstrated a hypermotile phenotype. The antisense-treated cells produced MIs of 15 { 0.9% (Fig. 6A) compared to the sense and negative control cells producing migration indices of 10.5 { 0.9% as shown in Figs. 6B and 6C. Therefore, blocking the endogenous expression of laminin-5 is associated with enhanced motility on a suboptimal type I collagen matrix. In order to be sure that the antisense experiments were indeed inhibiting the synthesis and secretion of laminin-5, we analyzed the underlying substratum using SDS–PAGE and Western blotting with antibodies to laminin-5. As shown in Fig. 6D, in the presence of the antisense oligonucleotide, laminin-5 secretion was down-regulated by approximately 75% (Fig. 6D, lane 4) compared with the control (no added oligonucleotide) and sense-treated cells (Fig. 6D, lanes 2 and 3). Ezrin is a member of the ERM family of proteins and is an important constitutive protein of epithelial cells involved in the formation of lamellipodia [40]. We examined the expression of ezrin by immunoflu-

AID

ECR 3586

/

6i21$$$263

05-28-97 12:01:55

335

orescence and Western blot analysis in keratinocytes plated on type I collagen, laminin-1, and laminin5. As shown in Fig. 7A, when the keratinocytes are juxtaposed to a collagen matrix and are in a motile mode, an apron of lamellipodia is formed, and it is polarized to one side of the cell. In contrast, when the keratinocytes are juxtaposed to laminin-1 or laminin-5 and in a nonmotile mode (Figs. 7B and 7C), the lamellipodia apron is lost and the microspikes are short, stunted, and nonpolarized. When human keratinocytes are plated and allowed to migrate on the three matrices and the cellular proteins are then extracted in sample buffer and subjected to SDS – PAGE and Western blot analysis with a polyclonal antibody to ezrin, one observes a significant down-regulation in the expression of ezrin in keratinocytes when they are plated on the two laminin isoforms compared to cells apposed to a type I collagen matrix which promotes motility, as shown in Fig. 7D (compare the ezrin band density in lane 1 with lanes 2 and 3). These experiments demonstrate that the expression of ezrin and the morphological expression of lamellipodia and microspikes are influenced by the motility state of the keratinocyte. Polarized, well-formed lamellipodia and high expression of ezrin are seen when cells are apposed to matrices that promote motility, while these elements are down-regulated on laminin-1 and laminin-5, which produce a nonmotile keratinocyte phenotype.

FIG. 6. Effect of antisense oligonucleotide to the a3 chain of laminin-5 on keratinocyte migration and keratinocyte secretion of laminin-5 on suboptimal amounts of collagen I. (A–C) Colloidal gold migration assay. Representative fields treated with antisense (A), sense (B), and no oligonucleotides (C) every 6 h for 18 h. Bar, 100 mm. (D) Keratinocyte matrix was extracted, as described under Methods, from keratinocytes treated with sense and antisense oligonucleotides and subjected to nonreducing SDS–PAGE, transferred to a nitrocellulose membrane, and then probed with a polyclonal antibody to the g2 chain of laminin-5. Lane 1, laminin-5 control. Lane 2, control (no oligonucleotide). Lane 3, sense oligonucleotide. Lane 4, antisense oligonucleotide. Protein bands were quantitated by densitometry. n Å OD 1 Area. Lane 3, n Å 10.5. Lane 4, n Å 3.2.

eca

336

O’TOOLE ET AL.

FIG. 7. Effect of laminin-5 on the expression of the lamellipodiaassociated protein ezrin. (A–C) Ezrin was detected by immunofluorescent localization using a polyclonal antibody to ezrin, followed by a secondary FITC-conjugated antibody. The cells were examined and photographed with a Zeiss epilluminating fluorescent microscope. In A, the cells are apposed to type I collagen and are highly motile. The arrow points to an apron of lamellipodia fringed with microspikes. In B, the cells are apposed to laminin-1 and are nonmotile. The arrow shows lack of an apron of lamellipodia. In C, the cells are apposed to laminin-5 and also are nonmotile. Again, the arrow points to attenuated lamellipodia compared with A. (D) Semiquantitative Western blot of 10 mg per lane of extracts from keratinocytes apposed to collagen I (lane 1), laminin-1 (lane 2), and laminin-5 (lane 3) and subjected to SDS–PAGE and Western blotting with a polyclonal antibody to ezrin, pAs 90.3 at a dilution of 1:1000. Protein bands were quantitated by densitometry. n Å OD 1 Area. Lane 1, n Å 8.5. Lane 2, n Å 4.9. Lane 3, n Å 4.8.

DISCUSSION

We have demonstrated that laminin-5 inhibits keratinocyte migration. This heterotrimeric protein, a component of anchoring filaments within the lamina lucida zone of human skin basement membrane, is known to have significant adhesive properties and is thought to be the main cell–matrix attachment factor for human keratinocytes [5]. Because human keratinocytes synthesize and deposit laminin-5, these cells provide their own matrix for substratum attachment. Therefore, it is not surprising that like laminin-1, laminin-5 has an inhibitory effect on cell motility because of its powerful attachment and anchoring functions. The finding that exogenous laminin-5 inhibits keratinocyte motility is in keeping with our previous observation that junctional epidermolysis bullosa keratinocytes are hypermotile compared to normal human keratinocytes [41]. Junctional epidermolysis bullosa (JEB) is a blistering disease that causes epidermal detachment through the lamina lucida of the skin dermal– epidermal junction associated with defective secretion of laminin-5 [6, 42, 43]. Mutations in the genes LAMA3, LAMB3, and LAMC2, which encode for the a3, b3, and g2 chains of laminin-5, respectively, have been identi-

AID

ECR 3586

/

6i21$$$263

05-28-97 12:01:55

fied in patients with JEB [33, 44–46]. JEB keratinocytes exhibit hypermotility on matrices of collagen and fibronectin [47, 48]. Nevertheless, JEB keratinocytes readily respond to exogenous laminin-5 which reverses their hypermotile phenotype to a normal level [47]. Restored expression of laminin-5 in genetically altered JEB keratinocytes also restores motility and adhesion to normal levels [49]. When laminin-5 is added to the medium of normal keratinocytes plated on collagen I, it also inhibits keratinocyte migration and negates the promigratory effect of collagen I. Conditioned medium from 804G matrix (rat laminin-5 in a soluble form) is sufficient to induce rapid spreading and hemidesmosome assembly in HaCaT cells (an immortalized keratinocyte cell line) [50]. A rat monoclonal antibody, termed CM6, recognizes the 150-kDa a chain of rat laminin-5 and binds the globular (G) domain of laminin-5, as determined by rotary shadowing. The G domain of laminin-5 is involved in induction of hemidesmosome assembly and is necessary for maintenance of the structural integrity of the formed hemidesmosome [51]. It is interesting that many cellular interactions of laminin-1, including attachment, are mediated by the G domain [52]. During reepithelialization, the marginal keratinocytes retract their tonofilaments, the hemidesmosomes disappear, and the gap junctions become more prominent [1]. In this context, the inhibitory effect of laminin-5 on cell motility and the induction of hemidesmosome formation by soluble laminin-5 are in accordance with the notion that laminin-5 plays an important role as a potent substratum adhesive factor for keratinocytes [5, 7, 8]. Ezrin, moesin, and radixin are constitutive proteins of many cell types that are involved in the formation of lamellipodia and microspikes [40] and have been shown to be difficult to modulate due to their high constitutive expression in cells. Ezrin and moesin have recently been shown to bind to actin [38, 53]. Following mechanical injury to endothelial cells, calcium-activated proteolysis of ezrin to a 55-kDa form occurs [54]. Moesin and radixin are up-regulated in experimental glomerulonephritis during mesangial cell repair after injury. This effect appears to be mediated by platelet-derived growth factor [55]. All these observations suggest that the ERM proteins are involved in cell motility. We found that keratinocytes, induced into a nonmotile state by apposition to laminin-1 or laminin-5, had decreased expression of ezrin and a morphological redistribution of lamellipodia and microspikes when compared to keratinocytes in a motile state induced by a collagen matrix. Taken together, these observations are consistent with the notion that ezrin plays an important role in lamellipodia expression and cell motility. As shown in Fig. 5, the inhibitory effect of exoge-

eca

LAMININ-5 INHIBITS KERATINOCYTE MIGRATION

nously added laminin-5 on keratinocyte motility was nullified by blocking the a3 integrin subunit of the cells with a functional blocking antibody. Another functional blocking antibody to the a6 integrin alone appeared to have less effect on the inhibition of migration by laminin-5. The presence of the two antibodies together, however, had a cumulative effect and completely reversed the inhibitory effect of laminin-5 on keratinocyte migration. The reversal of the inhibitory effect of laminin-5 by anti-a3 was not unexpected because human keratinocytes use a3b1 for interactions with laminin-5 via actin-containing focal adhesions [20]. Likewise, one can envision also that the a6b4 integrin could have an important role in cell motility or nonmotility since it is the integrin receptor localized to the hemidesmosome, and the anchorage of keratinocytes via the hemidesmosome provides substratum adherence and mechanical stability to the dermal–epidermal junction. The endogenous laminin-5 secreted by keratinocytes during the motility process also inhibits motility. Using antisense oligonucleotides aligning with the signal peptide of the a3 chain of laminin-5, we blocked synthesis and secretion of laminin-5. The consequence of blocking the expression of laminin-5 is that the keratinocytes developed a hypermotile phenotype reminescent of JEB cells which have mutations causing an underexpression of laminin-5 [47, 48]. Like the antisense experiments, blocking the cell binding function of the secreted laminin-5 with BM165 antibody also increased keratinocyte motility on suboptimal amounts of collagen I in a dose-dependent fashion. Aberrant expression of the g2 chain of laminin-5 is seen in cancer cells, particularly in budding neoplastic cells at the invasion front [56]. Migrating keratinocytes in regenerating epithelium secrete laminin-5 [57, 58]. In time-course experiments with the colloidal gold assay, we previously noted that approximately 10 to 12 h after attachment, the rate of increase in keratinocyte migration decreases, and the cells become stationary between 18 and 20 h [59]. The time course of the decrease in migration corresponds with the increased secretion of laminin-5 by keratinocytes 12 to 18 h after attachment [10]. As shown in this study, blocking the a3b1 integrin receptor enhances migration on collagen I and prolongs migration during the later time points of the track assay. This suggests that the interaction of the a3b1 integrin receptor with secreted endogenous laminin-5 is associated with a decreased rate of migration, probably due to enhanced cell–substratum attachment. Our data conflicts with a recent study showing that laminin-5 increases keratinocyte motility [60]. We used mAb K140 antibody affinity-purified laminin-5 from squamous cell carcinoma medium which contained no laminin-6 or laminin-7 [10]. Zhang et al. [60] used lami-

AID

ECR 3586

/

6i21$$$263

05-28-97 12:01:55

337

nin-5 from human amnion which has recently been demonstrated to contain laminin-6 and laminin-7 as well as laminin-5 [61]. It is possible that different laminin isoforms or laminin-5 from different organs may have different effects on cell motility. In addition, these investigators used a Boyden chamber assay to assess motility of keratinocytes, which measures attachment of the cells to the membrane facing the upper chamber as much as cellular motility, because if the cells do not attach well to the upper well membrane, they cannot migrate through it. Without some level of matrix attachment, cellular migration cannot ensue. Moreover, for the cells attached to the membrane to be counted as ‘‘migrated cells’’ on the other side of the membrane, they only need to squeeze through an 8-mm pore in a membrane approximately 10 mm thick. In contrast, in the colloidal gold assay keratinocytes can migrate on a collagen matrix for up to 500 mm in an 18-h period. When considering cell attachment to a substratum or a membrane, one needs to consider the percentage of the total cells attached and the quality or firmness of the attachment. These data are not provided in the Zhang and Kramer study for the Boyden chamber assay [60]. The fact that cells were observed in the presence of laminin-5 to migrate through the membrane to the other side could be explained by enhanced cell–substratum attachment over controls or enhanced invasion through the matrix and membrane. These authors, like ourselves, used an in vitro scratch assay. They did not use laminin-5 as a fresh ligand surface or as an additive in the medium in this assay. Recently, Gianelli et al. [62] have demonstrated that when laminin-5 matrix is treated with matrix metalloproteinase2 (Gelatinase A), it changes from being a stationary to a migratory substrate for mammary epithelial cells [62]. In general when keratinocytes are motile, there is increased expression of metalloproteinases. For example collagen I, hepatocyte growth factor, dibutryl cyclic AMP, and hypoxia all increase keratinocyte migration and cause increased expression of collagenase by keratinocytes [63–66]. It is possible that during in vivo reepithelialization, proteolysis of laminin-5 by abundant metalloproteinases [67, 68] reveals previously inaccessible domains of laminin-5 and switches it from being a potent inhibitor of migration to a migratory substratum. Our current ‘‘ratchet’’ model of human keratinocyte motility is that it is a sequence of events in which the adherent cell extends its lamellipodia, forms attachments (to deposited laminin-5 via the a3b1 integrin and perhaps collagen and other endogenously secreted matrices), forms focal adhesions, assembles and contracts its cytoskeleton, and finally, as it moves forward, disengages matrix adhesions by the expression of metalloproteinases and other proteinases that degrade matrix. During reepithelialization, we believe that se-

eca

338

O’TOOLE ET AL.

creted laminin-5 is the primary matrix attachment factor that stabilizes the migrating keratinocyte and ultimately is involved with the reestablishment of the new basement membrane zone. This work was presented as a paper at the Annual Meeting of the Society for Investigative Dermatology held in Washington, DC, May 1, 1996. The work was supported by Grants PO1 AR41045 and RO1 AR33625 from the National Institutes of Health, Bethesda, Maryland (Dr. Woodley) and a Dermatology Foundation Career Development Award and the Office of Research and Development, VA Palo Alto Health Care System (Dr. Marinkovich). Dr. O’Toole is a Howard Hughes Medical Institute Physician Postdoctoral Fellow. We are grateful to Dr. Jonathan Jones, Department of Cellular and Molecular Biology, Northwestern University, for his careful review of the manuscript and his helpful discussion.

REFERENCES 1. Woodley, D. T. (1996) in The Molecular and Cellular Biology of Wound Repair (Clark, R. A. F., Ed.), 2nd ed., Plenum, New York. 2. Sarret, Y., Woodley, D. T., Grigsby, K., Wynn, K., and O’Keefe, E. J. (1992) J. Invest. Dermatol. 98, 12–16. 3. Woodley, D. T., Bachmann, P., and O’Keefe, E. J. (1988) J. Cell. Physiol. 136, 140–146. 4. O’Keefe, E. J., Payne, R. E., Russell, N., and Woodley, D. T. (1985) J. Invest. Dermatol. 85, 125–130. 5. Rousselle, P., Lunstrum, G. P., Keene, D. R., and Burgeson, R. E. (1991) J. Cell. Biol. 114, 567–576. 6. Marinkovich, M. P., Verrando, P., Keene, D. R., Meneguzzi, G., Lunstrum, G. P., Ortonne, J. P., and Burgeson, R. E. (1993) Lab. Invest. 69, 295–299. 7. Carter, W. G., Ryan, M. C., and Gahr, P. J. (1991) Cell 65, 599– 610. 8. Verrando, P., Schofield, O., Ishida-Yamamoto, A., Aberdam, D., Partouche, O., Eady, R. A. J., and Ortonne, J. P. (1993) J. Invest. Dermatol. 101, 738–743. 9. Burgeson, R. E., Chiquet, M., Deutzmann, R., Ekblom, P., Engel, J., Kleinman, H., Martin, G. R., Meneguzzi, G., Paulsson, M., Sanes, J., Timpl, R., Tryggvason, K., Yamada, Y., and Yurchenco, P. D. (1994) Matrix Biol. 14, 209–211. 10. Marinkovich, M. P., Lunstrum, G. P., and Burgeson, R. E. (1992) J. Biol. Chem. 267, 17900–17906. 11. Hynes, R. O. (1992) Cell 69, 11–25. 12. Adams, J. C., and Watt, F. M. (1993) Development 117, 1183– 1198. 13. Kim, J. P., Chen, J. D., Wilke, M. S., Schall, T. J., and Woodley, D. T. (1994) Lab. Invest. 71, 401–408. 14. Kim, J. P., Chen, J. D., and Woodley, D. T. (1992) J. Cell. Physiol. 151, 443–450. 15. Carter, W. G., Kaur, P., Gil, S. G., Gahr, P. J., and Wayner, E. A. (1990) J. Cell. Biol. 111, 3141–3154. 16. Jones, J. C. R., Kurpakus, M. A., Cooper, H. M., and Quaranta, V. (1991) Cell. Regul. 2, 427–438. 17. Verrando, P., Blanchet-Bardon, C., Pisani, A., Thomas, L., Cambazard, F., Eady, R. A., Schofield, O., and Ortonne, J. P. (1991) Lab. Invest. 64, 85–92. 18. Vidal, F., Aberdam, D., Miquel, C., Christiano, A. M., Pulkkinen, L., Uitto, J., Ortonne, J. P., and Meneguzzi, G. (1995) Nature Genet. 10, 229–234. 19. McGrath, J. A., Gatalica, B., Christiano, A. M., Li, K., Owaribe,

AID

ECR 3586

/

6i21$$$264

05-28-97 12:01:55

20. 21. 22. 23. 24.

K., McMillan, J. R., Eady, R. A., and Uitto, J. (1995) Nature Genet. 11, 83–86. DiPersio, C. M., Shah, S., and Hynes, R. O. (1995) J. Cell Sci. 108, 2321–2336. Gil, S. G., Brown, T. A., Ryan, M. C., and Carter, W. G. (1994) J. Invest. Dermatol. 103, 31S–38S. Boyce, S. T., and Ham, R. G. (1983) J. Invest. Dermatol. 81, 33S–40S. O’Keefe, E. J., and Chiu, M. L. (1988) J. Invest. Dermatol. 90, 2–7. Petersen, M. J., Woodley, D. T., Stricklin, G. P., and O’Keefe, E. J. (1987) J. Biol. Chem. 262, 835–840.

25. Woodley, D. T., Rao, C. N., Hassell, J. R., Liotta, L. A., Martin, G. R., and Kleinman, H. K. (1983) Biochim. Biophys. Acta 761, 278–283. 26. Sonnenberg, A., Janssen, H., Hogervorst, F., Calafat, J., and Hilgers, J. (1987) J. Biol. Chem. 262, 10376–10383. 27. Wayner, E. A., Carter, W. G., Piotrowicz, R. S., and Kunicki, T. J. (1988) J. Cell Biol. 107, 1881–1891. 28. Heagerty, A. H. M., Kennedy, A. R., Leigh, I. M., Purkis, P., and Eady, R. A. J. (1986) Br. J. Dermatol. 115, 125–131. 29. Albrecht-Buehler, G. (1977) Cell 11, 395–404. 30. Chen, J. D., Kim, J. P., Zhang, K., Sarret, Y., Wynn, K. C., Kramer, R. H., and Woodley, D. T. (1993) Exp. Cell. Res. 209, 216–223. 31. Cha, D., O’Brien, P., O’Toole, E. A., Woodley, D. T., and Hudson, L. G. (1996) J. Invest. Dermatol. 106, 591–597. 32. Rousselle, P., and Aumailley, M. (1994) J. Cell. Biol. 125, 205– 214. 33. Vidal, F., Baudoin, C., Miquel, C., Galliano, M. F., Christiano, A. M., Uitto, J., Ortonne, J. P., and Meneguzzi, G. (1995) Genomics 30, 273–280. 34. Lowry, O. H., Rosebrough, N., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chem. 193, 265–275. 35. Towbin, H., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. Sci. USA 76, 4350–4354. 36. Matsui, C., Wang, C. K., Nelson, C. F., Bauer, E. A., and Hoeffler, W. K. (1995) J. Biol. Chem. 270, 23496–23503. 37. Gould, K. L., Bretscher, A., Esch, F. S., and Hunter, T. (1989) EMBO J. 8, 4133–4142. 38. Pestonjamasp, K., Amieva, M. R., Strassel, C. P., Nauseef, W. M., Furthmayr, H., and Luna, E. J. (1995) Mol. Biol. Cell. 6, 247–259. 39. McLean, I. W., and Nakane, P. K. (1974) J. Histochem. Cytochem. 22, 1077–1083. 40. Takeuchi, K., Sato, N., Kasahara, H., Funayama, N., Nagafuchi, A., Yonemura, S., and Tsukita, S. (1994) J. Cell. Biol. 125, 1371–1384. 41. Chen, J. D., Langhofer, M., Iwasaki, T., Kim, Y. H., Jones, J. C. R., Krueger, J. G., Carter, D. M., and Woodley, D. T. (1993a) J. Invest. Dermatol. 100, 517. 42. Fine, J. D., Bauer, E. A., Briggaman, R. A., Carter, D. M., Eady, R. A. J., Esterly, N. B., Holbrook, K. A., Hurwitz, S., Johnson, L., Lin, A., Pearson, R., and Sybert, V. P. (1991) J. Am. Acad. Dermatol. 24, 119–135. 43. Meneguzzi, G., Marinkovich, M. P., Aberdam, D., Pisani, A., Burgeson, R., and Ortonne, J. P. (1992) Exp. Dermatol. 1, 221– 229. 44. Pulkkinen, L., Christiano, A. M., Airenne, T., Haakana, H., Tryggvason, K., and Uitto, J. (1994) Nature Genet. 6, 293–297. 45. Pulkkinen, L., Christiano, A. M., Gerecke, D. R., Wagman,

eca

LAMININ-5 INHIBITS KERATINOCYTE MIGRATION D. W., Burgeson, R. E., Pittelkow, M. R., and Uitto, J. (1994) Genomics 24, 357–360. 46. Kivirikko, S., McGrath, J. A., Baudoin, C., Aberdam, D., Ciatti, S., Dunnill, M. G., McMillan, J. R., Eady, R. A., Ortonne, J. P., Meneguzzi, G., and Uitto, J. (1995) Hum. Mol. Genet. 4, 959– 962. 47. Chen, J. D., Marinkovich, M. P., Krueger, J., Lapiere, J. C., Hoeffler, W., Kim, Y., Langhofer, M., Jones, J., and Woodley, D. T. Submitted for publication. 48. Miquel, C., Gagnoux-Palios, L., Durand-Clement, M., Marinkovich, M. P., Ortonne, J. P., and Menegiuzzi, G. (1996) Exp. Cell Res. 224, 279–290. 49. Gagnoux-Palacios, L., Vailly, J., Durand-Clement, M., Wagner, E., Ortonne, J. P., and Meneguzzi, G. (1996) J. Biol. Chem. 271, 18437–18444. 50. Hormia, M., Falk-Marzillier, J., Plopper, G., Tamura, R. N., Jones, J. C., and Quaranta, V. (1995) J. Invest. Dermatol. 105, 557–561. 51. Baker, S. E., Hopkinson, S. B., Fitchmun, M., Andreason, G. L., Frasier, F., Plopper, G., Quaranta, V., and Jones, J. C. R. (1996) J. Cell Sci. 109, 2509–2520. 52. Skubitz, A. P., Letourneau, P. C., Wayner, E., and Furcht, L. T. (1991) J. Cell. Biol. 115, 1137–1148. 53. Algrain, M., Turenen, O., Vaheri, A., Louvard, D., and Arpin, M. (1993) J. Cell. Biol. 120, 129–139. 54. Shuster, C. B., and Herman, I. M. (1995) J. Cell. Biol. 128, 837– 848. 55. Hugo, C., Hugo, C., Pichler, R., Gordon, K., Schmidt, R., Amieva, M., Couser, W. G., Furthmayr, H., and Johnson, R. J. (1996) J. Clin. Invest. 97, 2499–2508.

56. Pyke, C., Rømer, J., Kallunki, P., Lund, L. R., Ralfkiaer, E., Danø, K., and Tryggvason, K. (1994) Am. J. Pathol. 145, 782– 791. 57. Larjava, H., Salo, T., Haapasalmi, K., Kramer, R. H., and Heino, J. (1993) J. Clin. Invest. 92, 1425–1435. 58. Ryan, M. C., Tizard, R., VanDevanter, D. R., and Carter, W. G. (1994) J. Biol. Chem. 269, 22779–22787. 59. Kim, J. P., Zhang, K., Kramer, R. H., Schall, T. J., and Woodley, D. T. (1992) J. Invest. Dermatol. 98, 764–770. 60. Zhang, K., and Kramer, R. H. (1996) Exp. Cell. Res. 227, 309– 322. 61. Champliaud, M. F., Lunstrum, G. P., Rousselle, P., Nishiyama, T., Keene, D. R., and Burgeson, R. E. (1996) J. Cell. Biol. 132, 1189–1198. 62. Gianelli, G., Falk-Marzillier, J., Stetler-Stevenson, W. G., and Quaranta, V. (1996) Mol. Biol. Cell. (Suppl.) 7, 58a. 63. Iwasaki, T., Chen, J. D., Kim, J. P., Wynn, K. C., and Woodley, D. T. (1994) J. Invest. Dermatol. 102, 891–897. 64. Petersen, M. J., Woodley, D. T., Stricklin, G. P., and O’Keefe, E. J. (1990) J. Invest. Dermatol. 94, 341–346. 65. Dunsmore, S. E., Rubin, J. S., Kovacs, S. O., Chedid, M., Parks, W. C., and Welgus, H. G. (1996) J. Biol. Chem. 271, 24576– 24582. 66. O’Toole, E. A., Marinkovich, M. P., Peavey, C. L., Amieva, M., Furthmayr, H., Mustoe, T. A., and Woodley, D. T. submitted for publication. 67. Inoue, M., Kratz, G., Haegerstrand, A., and Sta˚hle-Ba¨ckdahl, M. (1995) J. Invest. Dermatol. 104, 479–483. 68. Saarialho-Kere, U. K., Kovacs, S. O., Pentland, A. P., Olerud, J. E., Welgus, H. G., and Parks, W. C. (1993) J. Clin. Invest. 92, 2858–2866.

Received October 11, 1996 Revised version received March 31, 1997

AID

ECR 3586

/

6i21$$$264

05-28-97 12:01:55

339

eca