Functional Re-expression of laminin-5 in laminin-gamma2-deficient human keratinocytes modifies cell morphology, motility, and adhesion.

Herlitz junctional epidermolysis bullosa (H-JEB) is characterized by a reduced adherence of keratinocytes consequent to deficient expression of the extracellular adhesive ligand laminin-5. To complement the genetic defect causing H-JEB, we transferred an eukaryotic cassette expressing the cDNA for the γ2 chain of laminin-5 into H-JEB keratinocytes in which the expression of the polypeptide is hampered by a homozygous mutation generating a premature termination codon. Transfection using adenovirus-polylysin-transferrin-DNA complexes resulted in a transient synthesis of the recombinant laminin γ2 chain that associated with the endogenous α3 and β3 chains to form laminin-5 molecules readily deposited on the tissue culture substrate. Furthermore, retroviral-mediated transduction of the γ2 cDNA yielded persistent expression and polarized secretion of laminin-5. The protein incorporated into the basement membrane produced by the revertant cells inoculated subcutaneously in nude mice. In these transfectants, re-expression of laminin-5 induced changes in cell morphology and reorganization of focal adhesions that assumed the shape and distribution of the counterparts detected in normal keratinocytes. These observations correlated with an enhanced cell-substrate adhesion and a reduced motility of the transfected cells. Our results demonstrate that a restored expression of laminin-5 induces a phenotypic reversion of genetically altered H-JEB keratinocytes and open new perspectives to the analysis of the mechanisms regulating adhesion of epithelial cells.


Herlitz junctional epidermolysis bullosa (H-JEB) is
characterized by a reduced adherence of keratinocytes consequent to deficient expression of the extracellular adhesive ligand laminin-5. To complement the genetic defect causing H-JEB, we transferred an eukaryotic cassette expressing the cDNA for the ␥2 chain of laminin-5 into H-JEB keratinocytes in which the expression of the polypeptide is hampered by a homozygous mutation generating a premature termination codon. Transfection using adenovirus-polylysin-transferrin-DNA complexes resulted in a transient synthesis of the recombinant laminin ␥2 chain that associated with the endogenous ␣3 and ␤3 chains to form laminin-5 molecules readily deposited on the tissue culture substrate. Furthermore, retroviral-mediated transduction of the ␥2 cDNA yielded persistent expression and polarized secretion of laminin-5. The protein incorporated into the basement membrane produced by the revertant cells inoculated subcutaneously in nude mice. In these transfectants, re-expression of laminin-5 induced changes in cell morphology and reorganization of focal adhesions that assumed the shape and distribution of the counterparts detected in normal keratinocytes. These observations correlated with an enhanced cell-substrate adhesion and a reduced motility of the transfected cells. Our results demonstrate that a restored expression of laminin- 5

induces a phenotypic reversion of genetically altered H-JEB keratinocytes and open new perspectives to the analysis of the mechanisms regulating adhesion of epithelial cells.
Laminins are components of the extracellular matrix that contribute to the architecture of the basement membranes and mediate cell adhesion, growth, migration, and differentiation (1,2). These multifunctional glycoproteins are heterotrimers composed of individual ␣, ␤, or ␥ polypeptide chains expressed at various stages of development in focalized tissue locations (3). The tissue-specific localization of variants suggests that each laminin isoform fulfills definite functions (4).
Laminin-5 is found in the basement membrane zone of the skin, trachea, alimentary tract, and other specialized squamous epithelia with secretory or protective functions (5,6). This heterotrimeric protein is synthesized and secreted by the basal epithelial cells (5). In the skin it colocalizes with the anchoring filaments of the lamina lucida, the thread-like structures that bridge the hemidesmosomes to the lamina densa of the dermal-epidermal junction (7,8). Two predominant forms of laminin 5 with molecular masses of 400 and 440 kDa are found in the extracellular matrix of cultured keratinocytes and provide a specific substrate for the adhesion of proliferating and migrating cells (7). The 400-kDa form consists of an ␣3 (165 kDa), ␤3 (140 kDa), and ␥2 (105 kDa) chain, each encoded by a distinct gene, that are covalently linked and folded into a rod with globular domains at each end (9 -11). This species is produced by extracellular cleavage of the N-terminal domain of the 155-kDa ␥2 chain comprised in the 440-kDa heterotrimer. The 440-kDa form derives from a cell-associated 460-kDa precursor as a result of a processing event that generates the ␣3 chain from a 200-kDa polypeptide (12). A new laminin-5 variant prominently expressed in mucosal epithelia and harboring an ␣3B chain isoform of 300 kDa was also recently described (13). In this novel ␣3B chain, two alternating EGF-like domains and two globular domains substitute the short N-terminal domain of the ␣3A isoform (13). In the skin, the large majority of laminin-5 is thought to resemble the fully processed form of the protein.
An indication of the crucial role of laminin-5 in cell adhesion was first provided by the observation that in cell cultures, the mAb 1 BM 165 induces rounding and detachment of adherent keratinocytes without affecting fibroblasts (7). It was also demonstrated that treatment of skin biopsies with antibody BM 165 causes extensive de-epithelialization at the level of anchoring filaments (7). The importance of laminin-5 in maintaining the cohesion of the integuments was definitely proved by the identification of mutations in the genes coding for this basement membrane component in patients affected by the Herlitz syndrome (H-JEB) (Ref. 14 and references therein). This severe form of junctional epidermolysis bullosa is characterized by the detachment of the squamous epithelia from the underlying mesenchyme (15).
Human H-JEB keratinocytes presenting an abnormal expression of laminin-5 may constitute an interesting tool to dissect the contribution of this extracellular molecule to cell adhesion and migration. However, cultured H-JEB keratino-cytes display a reduced growth rate and an altered adhesion to the tissue culture substrate that hampers propagation. Using SV40 infective particles, we have recently immortalized a human keratinocyte cell line LSV5 that grows in culture continuously but retains differentiation potential and the phenotypic characteristics of the parental H-JEB keratinocytes, noticeably a reduced adhesion and enhanced motility in vitro (16). Because these keratinocytes derive from a H-JEB patient with a homozygous non-sense mutation in the gene encoding the laminin ␥2 chain that hampers production of laminin-5 (17), we surmised that transduction of LSV5 cells with a eukaryotic expression cassette containing the laminin ␥2 chain cDNA and directing the synthesis of the full-length polypeptide could restore the defective expression of laminin-5.
We report that the recombinant laminin ␥2 cDNA recovers synthesis and secretion of laminin-5 in H-JEB keratinocytes. We also provide evidence that the recombinant laminin-5 molecules are biologically active and induce reversion of the H-JEB phenotype.
Plasmids and Retroviruses-Plasmid pC␥2 contains the full-length open reading frame of the cDNA encoding the laminin ␥2 chain inserted in the expression vector pcDNA3 (InVitrogen) downstream of the human cytomegalovirus (CMV) immediate-early enhancer-promoter ( Fig.  1). Fig. 1 also depicts the amphotropic replication-defective retroviral vector pTG␥2, in which the laminin ␥2 cDNA is cloned downstream the mouse sarcoma virus long terminal repeat (LTR) of plasmid pTG5192, a generous gift of M. Methali. Plasmid pTG5192 is a construct derived from plasmid pLXSN (20), where the 3Ј LTR of moloney murine leukemia virus was replaced by the myeloproliferative sarcoma virus LTR. Plasmids pCMV-␤-gal (InVitrogen) and pLXSP (20), contain a Escherichia coli ␤-galactosidase gene driven by the CMV promoter and the mouse sarcoma virus LTR, respectively. Plasmids were amplified in E. coli XL1blue, and purified either on CsCl-ethidium bromide equilibrium density gradients or using a QIAGEN plasmid purification kit.
Transient Transfection Assays-50 -80% confluent keratinocyte monolayers (2.5-4.5 ϫ 10 5 cells) grown in 3-mm Petri dishes were transfected using either lipofection or adenovirus-transferrin-polylysine-mediated gene delivery. Lipofection was performed using Lipofectamine (Life Technologies, Inc.). 2 g of plasmid DNA and 10 l of Lipofectamine were separately diluted in 100 l of serum-reduced Opti-MEM I medium. The two solutions were combined, incubated 30 min at room temperature, diluted with 800 l of serum-reduced Opti-MEM I medium and added to the cells. After 5 h at 37°C, cells were rinsed, fed with KSM medium, and supplemented with 10% fetal calf serum (Life Technologies, Inc.). Transfection efficiency was evaluated 42 h later using an in situ ␤-galactosidase-staining method (21). Receptor-mediated endocytosis associated with inactivated adenovirus particles was performed using a adenovirus-polylysine-transferrin conjugate (AdpL-TfpL) (22,23). To prepare the AdpL-TfpL-reporter DNA complex, 8 l (3 ϫ 10 11 particles/ml) of biotinylated adenovirus particles (strain dl1014) were added to 25.3 l of HBS buffer (10 mM HEPES, pH 7.3, 150 mM NaCl), then diluted with a solution made of 133 ng of streptavidin-polylysin in 32 l of HBS, and incubated 30 min at room temperature. After the addition of 2 g (50 l) of plasmid DNA in HBS, the mixture was kept for 30 min at room temperature. 2 g of trans-ferrin-polylysine in 50 l of HBS were then added, and after 30 min of incubation at room temperature, the complex was shed 4 h onto cell monolayers at 37°C. Cell cultures were then fed with fresh medium.
Retroviral Infection-Infectious recombinant retrovirus was generated by transfecting the recombinant plasmid DNAs into the ecotropic packaging cell line psi-cre (18). The supernatants obtained from these cell cultures were used to infect the amphotropic packaging cell line psi-crip (18). Infection of H-JEB keratinocytes was then performed using fresh supernatants collected from confluent cultures of the amphotropic cell lines and filtered on 0.45-m filters (Gelman Sciences). Stable transduced cells were selected in the presence of 400 g/ml G418 (Sigma).
Cell Migration Assays-Cell motility was evaluated using a phagokinetic track assay (24, 25) following a procedure detailed previously (16) .
Evaluation of Cell-Substrate Binding Strength-The binding strength of H-JEB cells to the plastic tissue culture substrate was quantified using a modification of the assay devised by McClay (26). Cells suspended in KSM were seeded in 96-microtiter well plates at a final concentration of 10 5 cells/well and kept 19 h at 37°C in a 5% carbon dioxide humidified atmosphere. Plates were then centrifuged for 20 min at 500 ϫ g at 4°C in a bench centrifuge (Jouan, Inc., Winchester). Adhering cells were fixed with PBS 1% glutaraldehyde, stained with 0.1% crystal violet. Control cultures in non-centrifuged microtiter plates were also fixed and stained. Plates were then scanned at 570 nm using an enzyme-linked immunoassay reader (Dynatech MR 7000).
Growth in Vivo-Cells were injected subcutaneously in 6 -8-week-old Swiss nude mice (nu/nu). Each animal was inoculated with 5.10 6 cells in 0.2 ml of KSM medium (16). Cysts that had developed at the inoculation site were removed 10 days after injection and used for immunohistological analysis. To detect possible appearance of tumors, injected mice were kept under observation for 6 months and inspected twice a week.
Reconstruction of H-JEB Epithelia-Cultured H-JEB cells (2 ϫ 10 5 cells/cm 2 ) were seeded on acetate cellulose filters (PIHA, Millipore, France). Filters were kept immersed 7 days in KSM medium to allow formation of a confluent monolayer. Cultures were then raised at the liquid-air interface for 7 days to induce the formation of a differentiated epidermis (27).
Immunofluorescence Procedures-Cells were grown in glass chamber slides (Nunc), fixed 20 min at room temperature in PBS pH 7.4 containing 1 mM CaCl 2 , 1 mM MgCl 2 , and 3% formaldehyde. Cells were then rinsed three times in PBS, 1 mM glycine and then permeabilized 2 min at room temperature with PBS 0,1% Triton X-100 and extensively washed with PBS before exposure to antibodies. Cysts and artificial epithelia were either immediately frozen in liquid nitrogen and stored at Ϫ80°C or fixed in 10% buffered formalin and embedded in paraffin for histological analysis. For immunofluorescence analysis, semi-thin sections of frozen samples were fixed as detailed above. After 30 min of incubation in PBS 1% bovine serum albumin, the samples were incubated 1 h at room temperature with the primary mouse or rabbit antibodies, rinsed with PBS, and incubated for 30 min with the appropriate rhodamine-conjugated rabbit anti-mouse (Dako) or FITC-conjugated swine anti-rabbit IgG (Dako) secondary antibodies. Actin filaments were visualized using Texas red phalloidin. Sections and cell monolayers were analyzed using a Zeiss Axiophot microscope.
Ultrastructural Analysis-Cells were seeded in 35-mm tissue culture dishes. At confluence, cultures were fixed and embedded (28). Ultrathin sections were examined with a JEOL EM 1200 transmission electron microscope.
Immunoprecipitation and Western Analysis-Analysis of the cell extracts was performed following standard methods (6,19). Radioimmunoprecipitation of the medium conditioned by cultured cells was also detailed elsewhere (16). For Western analysis, samples of medium (2 ml) were centrifuged for 10 min at 700 ϫ g. Protease inhibitors (1 mM 4-(2-aminoethyl)benzenesulfonil fluoride, 100 M of leupeptin, 1 M of pepstatin, 2 M aprotinin, and 10 mM EDTA, pH 8) were added to the supernatant before addition of 33% (w/v) ammonium sulfate and overnight precipitation at 4°C. After 1 h of centrifugation at 60,000 ϫ g at 4°C, the pellet was resuspended in 25 mM Tris, pH 7.5, 65 mM NaCl, 1 mM EDTA in the presence of the protease inhibitors. After extensive dialysis against 25 mM Tris-HCl, pH 7.5, 1 mM EDTA pH 8, 65 mM NaCl and concentration to a volume of 100 l using the AMICON (Danvers) Centricon® microconcentrator system, the protein content of each sample was determined by the Bradford assay (Bio-Rad). Samples were boiled in sample buffer (60 mM Tris, pH 6.8, 1% SDS, 10% glycerol, and 5% ␤-mercaptoethanol) separated by SDS-polyacrylamide gel electrophoresis (7.5% acrylamide gels) and processed after blotting on nylon filters (13). The antibody-antigen complex was revealed using the enhanced chemiluminescence (ECL) Western blotting kit (Amersham Corp.).

Delivery of the Laminin ␥2 Chain cDNA into Cultured H-JEB Keratinocytes by an Adenovirus-Polylysine-Transferrin-Re-
porter Complex-To verify that the genetic defect of H-JEB keratinocytes can be complemented by transfection of an ad hoc cDNA, a full-length cDNA encoding the human laminin ␥2 chain under the control of the cytomegalovirus early late promoter was inserted into the eukaryotic expression vector pcDNA3 to obtain plasmid pC␥2 (Fig. 1). Plasmid pC␥2 was then expressed in the H-JEB cell line LSV5 following transient transfection mediated by a AdpL-TfpL complex (22,23). This method was retained because we had established that 90 -95% of LSV5 cells, 75% of SVK14 cells, and 30% of normal human primary keratinocytes were successfully transduced in this way (Table I). Western analysis of total extracts of LSV5 cells transfected with plasmid pC␥2 revealed that the recombinant laminin ␥2 polypeptide was actively synthesized. The polypeptide migrated as a single band with an apparent molecular mass (155 kDa) identical to that of the endogenous ␥2 chain produced by normal human keratinocytes. Western analysis of the culture medium conditioned by the transfectants demonstrated that the recombinant laminin ␥2 chain was secreted and processed into the 105-kDa form. The ratio between the detectable amounts of the processed 105-kDa form and the unprocessed 155-kDa polypeptide appeared, however, lower than that found in medium conditioned by normal human keratinocytes (Fig. 2). Overexpressed 155-kDa recombinant laminin ␥2 accumulated intracellularly (not shown).
Immunofluorescence analysis using the mAb GB3, which binds to a conformational epitope in the laminin ␣3␤3␥2 heterotrimer (5,29), displayed a strong intracellular reactivity with the transfected LSV5 cells (Fig. 3). Immunoreactive trails were also detected in zones of the culture substrate encompassing the cells. Taken together, these observations indicated that in transfected H-JEB cells, expression of the recombinant laminin ␥2 chain results in the association of this polypeptide with the endogenous laminin ␣3 and ␤3 chains to form native laminin-5 molecules that are deposited on the culture substrate.
Retroviral-mediated Transduction of Laminin ␥2 Chain to H-JEB Keratinocytes-To achieve a persistent expression of the laminin ␥2 polypeptide in H-JEB keratinocytes, the fulllength cDNA for the human laminin ␥2 chain was introduced into LSV5 keratinocytes by retroviral-mediated gene transfer.
The cDNA was inserted into the retroviral vector pTG5192 carrying a selectable marker, the neomycin phosphotransferase gene cassette (PSVNEO), such that expression of the laminin ␥2 polypeptide was driven from the promotor-enhancer sequences contained within the viral LTR ( Fig. 1). High titer polyclonal viral producer cells were generated from the psi-crip amphotropic packaging cell line using standard methods based on selection in the presence of G418. Integrity of the laminin ␥2 retrovirus was verified by Northern analysis to detect the expected mRNA in viral producer cells (not shown) and by analysis of laminin ␥2 chain expression in transduced human keratinocyte LSV5 cells.
Expression of the transferred cDNA was first monitored by immunofluorescence using the pAb SE144, which recognizes the laminin ␥2 polypeptide (10), and mAb GB3, which reacts with the native laminin-5 heterotrimers (5). In the transduced cell cultures, at least 80% of the G418-resistant cells (LSV5-R) were reactive to the antibodies. Immunoreactivity of transfectants resulted in strong cytoplasmic fluorescence and staining of the extracellular matrix deposited on the culture substrate underneath the cells, which suggested synthesis and secretion of laminin-5 (not shown).
Cell lysates and samples of culture medium conditioned by the transduced cells were then analyzed by immunoblotting. As shown in Fig. 4A, pAb SE144 detected a specific migration band of 155 kDa in the cellular extracts and an additional 105-kDa band in the corresponding medium samples. These results imply that the recombinant ␥2 chain polypeptide is actively expressed and secreted by LSV5-R cells and then processed extracellularly (12). Subsequent immunoprecipitation analysis of medium conditioned by LSV5-R cells, using antibodies specific to each distinct chain of laminin-5 identified bands with an apparent molecular mass of 165, 155, 145, and 105 kDa (Fig. 4B), which correspond to the expected mass of the polypeptides that compose the 440-and 400-kDa extracellular forms of laminin-5 (12).

Polarized Secretion of Recombinant Laminin-5 by LSV5-R
Keratinocytes-In light of these observations, we investigated whether LSV5-R cells maintain the polarized secretion of laminin-5 observed in normal keratinocytes and whether the molecule integrates the basement membrane zone of the dermoepidermal junction. Cultures of LSV5-R cells were grown on cellulose acetate filters and exposed to air to obtain stratification into a multilayered epithelium (27). Immunofluorescence analysis of these artificial epithelia using mAb GB3 demonstrated that LSV5-R cells synthesize immunoreactive laminin-5 that is actively secreted in a polarized way and deposited on the culture substrate (Fig. 5A). As previously shown in similar experiments performed with normal human keratinocytes (27), laminin-5 accumulated on the inert surface and eventually diffused inside the filter.
The ability of LSV5-R cells to generate a basement membrane was then assessed in vivo. To this purpose, LSV5-R cells were injected subcutaneously in nude mice. In contact with the mouse mesenchyme, the original LSV5 cells display limited proliferation and form benign cysts presenting an outer layer of undifferentiated keratinocytes, internal layers of differentiating cells, and a core of dead corneocytes (16). These cysts are not immunoreactive with mAb GB3 (Fig. 5B). In contrast, with cysts obtained after injection of LSV5-R cells, immunofluorescence analysis using mAb GB3 revealed labeling of the basement membrane at the interface of grafted cells and the mouse mesenchyme (Fig. 5B), which indicates that laminin-5 produced by the retrovirus-transfected cells incorporates into complex supramolecular structures.

Re-expression of Laminin-5 Induces Changes in Cell
Morphology-LSV5 cells expanded in vitro in the presence of serum retain the morphology of the parental H-JEB keratinocytes (16). Their shape correlates with a loose attachment to the culture substrate, which is evidenced by the presence of wide zones of disconnection of the ventral plasma membrane from the plastic tissue culture support and numerous lamellipodia in the areas with prominent cell-substrate separation (Fig. 6A). In similar culture conditions, LSV5-R cells display a more flattened morphology maintained along with passages in culture. Inspection of electron microscope sections of cultured LSV5-R keratinocytes cut perpendicular to the culture surfaces revealed that the flat shape of the cells correlated with a close apposition of the ventral plasma membrane to the culture surface (Fig. 6B). The electron-dense material detected in the adhesion sites appeared more compact in LSV5-R than in parental LSV5 cells.
The number and topography of focal contacts in LSV5-R cells were then studied using the anti-vinculin mAb (hVIN1) and by interference reflection microscopy observation (30). Compared with LSV5 cells, LSV5-R keratinocytes display a modified morphology and distribution of adhesion devices. In LSV5 cells, vinculin spots co-distributed with numerous adhesion plaques that presented as tiny darkened interference reflection micros-  6) and SE85 (lanes 3 and 7) and mAb K140 (lanes 4 and 8) raised against the laminin ␥2, ␣3, and ␤3 chains, respectively. Lanes 1 and 5, control blank serum. The mass of molecular markers is indicated in kDa. Fractionated was performed on 7.5% SDS-polyacrylamide gel. Exposure time was 12 h. copy streaks at the peripheral edge of individual cells, associated with the endings of short microfilament bundles (Fig. 7B). These adhesion sites failed to enlarge and mature into focal adhesions. In spread LSV5-R cells, focal contacts were reduced in number and had assumed a spiky shape (Fig. 7D). They also appeared larger and denser than the counterparts seen in LSV5 keratinocytes and had acquired attached stress fibers as mature focal adhesions. As expected, talin displayed the same immunoreactivity pattern as vinculin (not shown).
Enhanced Binding of LSV5-R Keratinocytes to the Culture Substrate-Expression of recombinant laminin-5 appeared to enhance the adhesion of H-JEB cells, because monolayers of LSV5 cells spontaneously dissociated from the support, whereas a dispase treatment was required to detach confluent LSV5-R cells. We therefore measured the attachment strength of LSV5-R keratinocytes using a cell binding assay providing a quantitative measure of cell-substrate adhesion (31). Cells seeded in sealed compartments were allowed to adhere 19 h and were then submitted to a mechanical force to determine their resistance to detachment from the culture substrate (26,31). As shown in Fig. 8, at a relative centrifugal force of 500 ϫ g, 35% of LSV5 cells were dislodged against 18% in the case of LSV5-R keratinocytes. The binding strength of LSV5-R keratinocytes was estimated to be comparable with that of normal human keratinocytes (24% of detached cells) and sensibly enhanced compared with parental H-JEB keratinocytes (40% of detached cells). The results obtained from independent experiments suggested a marked strengthening in adhesion of LSV5-R keratinocytes compared with parental LSV5 cells.
Synthesis of Endogenous Laminin-5 Reduces Cell Motility-Keratinocyte motility is influenced by extracellular matrix (32). H-JEB keratinocytes seeded on collagen type IV display hypermotility compared with normal controls (16,25). To evaluate the migration capacity of LSV5-R keratinocytes, these cells were plated on gold particles coated with 15 g/ml collagen IV. Compared with LSV5 and parental H-JEB keratinocytes, which moved on the tissue culture support leaving tracks behind them, LSV5-R cells demonstrated reduced motility, comparable with that of wild type normal and SV40-transformed keratinocytes (Fig. 9). Migration indices were in the range of 12.8 and 13.6 for LSV5 and H-JEB keratinocytes, respectively, and 9 for LSV5-R cells, which approximately represents a 30% reduction of cell locomotion. No appreciable difference in migration rates was noted between LSV5 and control LSV5 cells transduced with a retrovirus expressing a ␤-gal reporter gene (not shown). In light of these results and consistent with the observation that laminin-5 inhibits motility of normal kerati- nocytes (25), we concluded that re-expression of endogenous laminin-5 influences migration of LSV5-R cells. DISCUSSION We have analyzed the effect of re-expression of the extracellular ligand laminin-5 in LSV5 keratinocytes that present a defective adhesion consequent to a genetic mutation hampering expression of the laminin ␥2 chain (16,17). This report provides the first evidence that delivery of DNA into keratino-cytes using AdpL-TfpL complexes results in high transfection efficiency and restoration of a genetically impaired function and that transduction with an exogenous ␥2 chain cDNA using a retroviral vector promotes the phenotypic reversion of the diseased keratinocytes.
Here we have shown that both in transient and long term expression experiments, the recipient cells produced a polypeptide identical in molecular mass to the 155-kDa species of laminin ␥2 chain synthesized by wild type keratinocytes. This polypeptide associated with the endogenous laminin ␣3 and ␤3 chains and formed a heterotrimeric ␣3␤3␥2 complex secreted extracellularly that was recognized by three distinct antibodies raised against each individual chain of laminin-5 and by mAb GB3. Because chain assembly into a complex preserving the native conformation of laminin-5 is required for binding with GB3 (5,29), GB3 recognition of the heterotrimer secreted by LSV5 cells in the culture medium and in the extracellular matrix suggests that the restored synthesis of ␥2 chain results in production of properly assembled laminin-5 molecules.
Expression vectors have been employed by other groups to study synthesis and assembly of recombinant laminin chains (33,34). However, the biological effect induced by the expression products of the delivered cDNAs could not be evaluated, either because the transfectants synthesized the corresponding wild type endogenous laminin chain (33) or because the cell types used for the transduction experiments do not constitutively produce the laminin species comprising the polypeptide encoded by the transfected DNA sequences (34). The novelty of our findings is that we demonstrate the effect of the restored synthesis of the laminin ␥2 chain on the morphology and behavior of cells re-expressing the adhesion ligand laminin-5.
Receptor-mediated endocytosis associated with adenovirus particles constitutes a method of choice to transduct epithelial cells (22,23). In this report we show that expression vectors linked AdpL-TfpL-DNA complexes are efficiently introduced in established keratinocyte cell lines but also in primary human keratinocytes. Transfection efficiencies were superior to any other procedure described thus far (35,36); therefore this method could constitute an important tool for examining the regulation of promoters and function of genes in epithelial cells as refractory to transfection as primary keratinocytes.
As others that used CMV vectors to express laminin-5 cDNAs (34), we observed that LSV5 cells transduced with ␥2 chain cDNA driven by the CMV promoter yielded an elevated production of the recombinant polypeptide that incorporated into laminin-5 molecules. However, consistent with the observation that the high level delivery inherent to this transfection method may have a certain cytotoxic effect (22), the recipient cells manifested a reduced growth rate and senescence that hampered the long term analysis of the transduced phenotype.
Lines of evidence indicated that in long term expression experiments, LSV5 keratinocytes transfected with viral vectors had reverted the diseased phenotype. Restored LSV5-R keratinocytes were isolated that re-expressed and deposited laminin-5 on the tissue culture substrate and the mesenchyme of nude mice grafted with the transduced cells as observed with human keratinocytes producing functional molecules of this adhesion ligand (16,27). LSV5-R keratinocytes also assumed an epithelial-like shape correlating with changes in the distribution of focal adhesions and a tighter adhesion of the basal plasma membrane to substrate. Laminin-5 has been identified as the major ligand for adhesion of keratinocytes via focal adhesions (37). It has been proposed that initial interaction of keratinocytes with laminin-5 results in cell attachment, spreading, and migration, whereas stable adhesion to the basement membrane of quiescent keratinocytes occurs via hemidesmosomes (38). In primary JEB keratinocytes and LSV5 cells, formation of focal adhesions appears hindered at a precursor stage in which microspikes on bundles of actin filaments are formed within the edge of the cells (16,28). Re-expression of laminin-5 induces disassembly of most of the minute focal contacts at the boundary of LSV5-R cells and enlargement of others that nucleate and associate stress fibers in a process similar to the maturation of focal adhesion (39). We have not yet determined which regulatory components of focal adhesion are involved in the process. However preliminary indications obtained using a panel of anti-phosphotyrosine antibodies argue for quantitative changes in the phosphorylation of focal adhesion components in these cells. 2 The changes in morphology of LSV5-R cells induced by reexpression of laminin-5 were concomitant with an enhanced adhesion of LSV5-R to the culture substrate, which suggests a more efficient functioning of the adhesion apparatus. Earlier biophysical studies on cultured cells suggested that optimal migration occurs at intermediate adhesion strength (40). According to these observations, re-expression of laminin-5 and consequent increase of focal adhesion stability may explain the decrease of locomotion of LSV5-R keratinocytes. Recent evidence indicates that migration of both normal and JEB keratinocytes is inhibited in vitro by the addition of exogenous laminin-5 to the culture medium that compensates the failure of JEB keratinocytes to deposit sufficient amounts of functional laminin-5 (25). Furthermore, these studies supported the idea that inhibition of migration induced by exogenous laminin-5 implicates integrin ␣3␤1 (41), which is the putative receptor of laminin-5 in focal adhesions (37,38). Our results indicate that laminin-5 may play a pivotal role in the coordination between cell adhesion and cell motility possibly via specific proteins in focal adhesion such as vinculin. In fact, in LSV5-R cells, the reorganization of vinculin in focal adhesions in response to the re-expression of laminin-5 correlates well with the reduced locomotion of these keratinocytes and also with the finding that expression of vinculin influences cell motility in vitro (42,43).
Laminin-5 is likely to play an essential role in the organization of hemidesmosomes, because in the skin of H-JEB patients without expression of this protein, hemidesmosomes are either rudimentary or absent (15,17). Consistent with this observation, binding of laminin-5 to integrin ␣6␤4 has been shown to constitute the crucial step in nucleation of hemidesmosome components (44). Accumulation of the extracellular matrix ligand at a threshold level appears also to be required for production and/or stabilization of hemidesmosomes in vitro (45). Ultrastructural observation of LSV5-R keratinocytes did not reveal formation of hemidesmosome-like adhesion structures. Because these cells synthesize all the known cellular components of hemidesmosomes (16), their inability to form defined hemidesmosomal structures may depend on the culture conditions we used but also on their restricted production of laminin-5 (Fig. 4). Furthermore, formation of hemidesmosomes may require the presence of extracellular 400-kDa species of laminin-5, which is poorly produced by LSV5-R cells. Interestingly, extracellular 400-kDa laminin-5 is abundantly produced by cancer epithelial cells that assemble mature hemidesmosomes in vitro (45)(46)(47), whereas colonic cancer epithelial cells, which derive from a tissue where hemidesmosomes are not detected, secrete exclusively the 440-kDa form of laminin-5 (48). 3 In summary, our results show that AdpL-TfpL complexes efficiently deliver recombinant cDNAs into keratinocytes. This method can therefore facilitate the analysis of assembly, stability, maturation, and secretion of laminin-5 using eukaryotic vectors expressing mutated cDNA constructs to complement genetically identified defects of laminin ␣3, ␤3, and ␥2 chains in H-JEB keratinocytes. We also demonstrate that retroviralmediated transfer of a vector expressing the laminin ␥2 chain cDNA restores stable synthesis of functional laminin-5 that induces changes in focal adhesions and influences morphology, motility, and adhesion of the transfected H-JEB keratinocytes. These results open new perspectives in the analysis of the adhesion mechanisms of epithelial cells. To this purpose, the establishment of keratinocyte cell lines and cultures of epithelial stem cells from JEB patients with mutations in the different hemidesmosome components and laminin-5 are in progress in our laboratory.