INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Keratinocytes adhere to the basement membrane by hemidesmosomes that serve as anchoring sites for the intermediate filament system and play a critical role in stabilizing the association of the dermis with the epidermis. The transmembrane components of hemidesmosomes comprise the laminin-5 (LN-5) binding integrin 64 and the bullous pemphigoid antigen (BP)180. These proteins are connected via the hemidesmosomal proteins plectin and BP230 to the keratin intermediate filament system (reviewed by Jones et al., 1998; Borradori and Sonnenberg, 1999).

Based on their structural constituents, two subtypes of hemidesmosomes are distinguished. Type I hemidesmosomes contain 64, plectin, BP180, and BP230 (Green and Jones, 1996), whereas type II hemidesmosomes contain only 64 and plectin (Uematsu et al., 1994). Recently, the tetraspanin CD151 was identified as another component of both type I and II hemidesmosomes (Sterk et al., 2000). Type I or classical hemidesmosomes are present in basal keratinocytes of squamous and complex epithelia (Nievers et al., 1999). Type II hemidesmosomes are found in intestinal epithelial cells and some other cultured epithelial cell types (Uematsu et al., 1994; Orian-Rousseau et al., 1996; Fontao et al., 1997). The association of type II hemidesmosomes with intermediate filaments is less robust than that of type I hemidesmosomes, which may imply a more dynamic regulation of their assembly.

The stability of type I hemidesmosomes is illustrated by their continued presence during mitosis (Riddelle et al., 1992; Baker and Garrod, 1993), thereby ensuring that cells with a strong proliferative potential remain present in the basal compartment of the skin. However, during wound healing, hemidesmosomes are disassembled to allow keratinocytes to migrate on a newly deposited LN-5 matrix (Martin, 1997; Decline and Rousselle, 2001). Several growth factors have been implicated in the regulation of the disassembly of hemidesmosomes, including the epidermal and hepatocyte growth factors (Mainiero et al., 1996; Trusolino et al., 2001). As a result of interaction of these growth factors with their cognate receptors, the 4 subunit is tyrosine phosphorylated and recruits the signaling adaptor protein Shc (Mainiero et al., 1996; Mariotti et al., 2001; Trusolino et al., 2001). Conceivably, the phosphorylation of 4 on tyrosine residues may prevent its incorporation into hemidesmosomes. Studies by Rabinovitz et al. (1999), however, have revealed that EGF receptor-mediated disruption of hemidesmosomes depends on the ability of this receptor to activate protein kinase C and may involve the direct phosphorylation of the 4 cytoplasmic domain on serine residues. In addition, there is evidence suggesting that 64 activates phosphoinositide 3-OH (PI-3) kinase (Shaw et al., 1997; Shaw, 2001) and interacts with the actin cytoskeleton in filopodia and lamellipodia (Rabinovitz et al., 1997, 1999). This ability of 64 to activate PI-3-kinase signaling has been connected to the promotion by this integrin of the migration and invasion of carcinoma cells (Rabinovitz and Mercurio, 1997; Shaw et al., 1997; Gambaletta et al., 2000; Hintermann et al., 2001). Activation of PI-3 kinase by 64 may also contribute to adhesion and spreading of keratinocytes via binding of 31 to LN-5 (Nguyen et al., 2000). Finally, it has been shown that proteolytic processing of the LN-5-3 and -2 chains may determine whether this matrix protein supports stable adhesion or instead migration (Giannelli et al., 1997; Goldfinger et al., 1998). Taken together these data show that cell migration is regulated by both extrinsic (proteases) and intrinsic (signaling molecules) factors.

New insights into the dynamic properties of focal contacts have been provided by studying green fluorescent (GFP)-tagged integrin  or  subunits in live cells (Smilenov et al., 1999; Ballestrem et al., 2001; Laukaitis et al., 2001). To gain more information about the dynamics of hemidesmosomes, we expressed two different enhanced GFP (EGFP)-tagged 4 chimeras in 4-deficient PA-JEB keratinocytes (Schaapveld et al., 1998). In one, EGFP was fused to the carboxy-terminal end of the 4 cytoplasmic domain (4-EGFP), and in the other the extracellular domain of 4 was replaced by EGFP (EGFP-4), which made this chimera incapable of associating with the 6 subunit and thus of binding LN-5. However, independently of binding to ligand, it can still induce hemidesmosome formation through its interaction with the hemidesmosomal component plectin (Nievers et al., 1998, 2000).

We studied the formation of hemidesmosomes by these two chimeras by using time-lapse video microscopy. Our data indicate that hemidesmosomes containing either 4-EGFP or EGFP-4 are both dynamic structures that are assembled and redistributed during cell migration and division. Furthermore, we show that as a result of binding of 64 to plectin, the binding of 64 LN-5 is stabilized, which inhibits cell migration and reduces the dynamics of 64. We thus provide novel insights into the role of the 64 integrin in migration.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Cell Lines

Immortalized 4-deficient keratinocytes have been derived from a patient with pyloric atresia associated with junctional epidermolysis bullosa (PA-JEB; Niessen et al., 1996; Schaapveld et al., 1998). Full-length 4 cDNA and a cDNA encoding a chimeric protein consisting of the extracellular and transmembrane domains of the interleukin 2 receptor (IL2R) fused to the cytoplasmic domain of 4 were stably expressed in PA-JEB cells by retroviral infection to generate PA-JEB/4 (Sterk et al. 2001) and PA-JEB/IL2R-4 (Nievers et al., 1998), respectively. PA-JEB/4-EGFP, PA-JEB/EGFP-4, and PA-JEB/4^R1281W-EGFP cells were generated as described below. All cells were maintained in keratinocyte serum-free medium (SFM; Life Technologies-BRL; Rockville, MD) supplemented with 50 µg/ml bovine pituitary extract, 5 ng/ml epidermal growth factor, 100 U/ml penicillin, and 100 U/ml streptomycin. To stimulate hemidesmosome formation, cells were grown for 24 h in calcium-rich medium, consisting of HAM-F12 Nutrient Mixture (Life Technologies-BRL) and DMEM (Life Technologies-BRL) in a ratio of 1:3.

Plasmid Constructs and Generation of PA-JEB/4-EGFP and PA-JEB/EGFP-4 Keratinocytes

The chimeric 4-EGFP and EGFP-4 constructs were produced in pcDNA3. The 4-EGFP chimera consisted of the full-length human 4A sequence, a sequence encoding six glycine residues followed by EGFP (Invitrogen, Carlsbad, CA). The EGFP-4 chimera consisted of the amino acid signal sequence of the 4 subunit, EGFP, and transmembrane and cytoplasmic domains of human 4A integrin. Both constructs were generated by overlap PCR using the following sequences: 4 c-terminus/6*glycine, 5'-catgccgccgccgccgccgccagtttggaagaactgttggtc-3'; 6*glycine-EGFP, 5'-ggcggcggcggcggcggcatggtgagcaagggcgaggag-3'; 4 signal sequence-EGFP, 5'-ctcctcgcccttgctcaccatgcggtttgccaaggtcccaga-3'; EGFP-4 transmembrane, 5'-catgccatggtcttgtacagctcgtccatgcc-3'. The two PCR products were digested with EcoRI and cloned into the corresponding site of the pcDNA3 vector. The 4^R1281W-EGFP construct was generated by replacing a 5-kb EcoRI/EcoRV fragment from 4-EGFP cDNA with a corresponding fragment containing a CGG to TGG mutation in codon 1281 of 4 (Geerts et al., 1999). Subsequently, the EGFP-4, 4-EGFP, and 4^R1281W-EGFP cDNAs were released from pcDNA3 by digestion with EcoRI, and the resulting fragments were ligated into the corresponding site of the retroviral LZRS-IRES-zeo expression vector, a modified LZRS retroviral vector conferring resistance to zeocin (Kinsella and Nolan, 1996; van Leeuwen et al., 1997) to result in the LZRS-EGFP-4-IRES-zeo, LZRS-4-EGFP-IRES-zeo, and LZRS-4^R1281W-EGFP-IRES-zeo constructs, respectively. Correctness of the DNA constructs was verified by sequence analysis The retroviral constructs were transfected into the Phoenix packaging cell line (Kinsella and Nolan, 1996) by the calcium phosphate precipitation procedure, and after 2 days supernatants containing recombinant viruses were collected. Transduction of the recombinant viruses in PA-JEB cells was performed for 10 h at 37°C. PA-JEB cells expressing 4-EGFP, 4^R1281W-EGFP, or EGFP-4 were isolated by FACS, expanded, and analyzed.

Antibodies

The rat mAb GoH3 is a blocking antibody directed against the extracellular part of the integrin 6 subunit (Sonnenberg et al., 1987). The mouse mAb K20, anti-1, was purchased from Biomeda (Foster City, CA). The mouse mAb 450-11A, directed against the cytoplasmic domain of 4 was from PharMingen International (San Diego, CA). The rabbit polyclonal antibody against the LN-5 3 chain was a kind gift of Dr. R. Timpl (Max Planck Institut für Biochemie, Martinsried, Germany). The mouse mAbs 121, anti-plectin/HD1 (Hieda et al., 1992), and 233, anti-BP180 (Nishizawa et al., 1993), were generously provided by Dr. K. Owaribe (Nagoya University, Nagoya, Japan). The human mAbs 5E and 10D against BP230 (Hashimoto et al., 1993) were a kind gift of Dr. T. Hashimoto (Kurume University, Kurume, Fukuoka, Japan). The mouse mAb TB30, against the extracellular domain of the interleukin-2-receptor (IL2R), was purchased from the Central Laboratory of the Red Cross Blood Transfusion Service (Amsterdam, The Netherlands). Mouse mAb 3C12, anti-ezrin, was from Lab Vision Corp. (Fremont, CA), mouse mAb RV202, anti-vimentin, was kindly provided by Dr. F. Ramaekers (University of Limburg, Maastricht, The Netherlands), mouse mAbs against  and  tubulins were from Sigma Chemical (St. Louis, MO; clones B-5-1-2 and 2-28-33), mouse mAb KL-1, anti-keratin, was from Immunotech (Marseille, France), rabbit antiserum against human LN-5 (Marinkovich et al., 1992) was a kind gift of Dr. R. Burgeson (Cutanous Biology Research Center, Charlestown, MA). The mouse mAb P48, also known as 11B1.G4, was clustered as CD151 in the VI International Leukocyte Typing Workshop (Ashman et al., 1997). The mouse B34 mAb directed against GFP was purchased from BabCO (Richmond, CA).

The sheep anti-mouse horseradish peroxidase-coupled antibodies were purchased from Amersham Corp. (Arlington Heights, IL), Texas Red-conjugated goat anti-mouse rat or rabbit antibodies were obtained from Molecular Probes (Eugene, OR), Texas Red- conjugated donkey anti-human antibodies and Cy-5-conjugated goat anti-rabbit antibodies and were from Jackson ImmunoResearch Laboratories (West Grove, PA).

Time-lapse Observations and Fluorescence Recovery after Photobleaching Experiments

Time-lapse observations were made in a tissue culture device at 37°C and viewed under a Leica TCS-NT confocal microscope (Deerfield, IL) equipped with argon/krypton laser. The krypton/argon laser was used to excite the EGFP-tagged proteins at 488 nm, and emissions above 515 nm were collected. Images of 4-EGFP and EGFP-4 were collected every 2-15 min for periods up to 4 h. Phase-contrast images of cells were taken during time-lapse observations to obtain the corresponding cell shape image.

Fluorescence recovery after photobleaching (FRAP) experiments were performed by selecting a region of 4-EGFP or EGFP-4 hemidesmosomes located at the cell periphery, and oval-shaped regions were bleached using the krypton/argon laser for 1 s at 100% power, resulting in a bleached spot of 1 µm diameter. Images were collected after bleaching every 15 s for 10 min. The fluorescence intensity in the bleached region of the 4-EGFP or EGFP-4 hemidesmosome during 10 min of recovery was normalized to the fluorescence intensity measured in a nonbleached region. This procedure allowed us to account for the decreased fluorescence due to overall bleaching of the entire field as a result of image collection. Phase-contrast images of cells were taken during FRAP analysis to ensure that there was no significant change in cell shape and position during periods of observation. Imaging from live cells on our confocal system prohibits the collection of large numbers of images, so that reliable fitting of more than one component is not possible. In the inhibitor studies, antibodies (GoH3) were added at a concentration of 25 µg/ml 24 h before FRAP analysis.

Preparation of Laminin-5 Matrices

PA-JEB/4-EGFP and PA-JEB/EGFP-4 keratinocytes were grown to confluency in six-well tissue culture plates, washed three times with PBS, and incubated overnight at 4°C in PBS containing 20 mM EDTA and a cocktail of protease inhibitors (Sigma). After incubation the cells were removed by forceful pipetting, and the remaining matrices were dissolved in SDS sample buffer. For Western analysis a fraction (1/4) of the matrices in the well was loaded.

Immunofluorescence Microscopy

PA-JEB/4, PA-JEB/4-EGFP, PA-JEB/EGFP-4, and PA-JEB/IL2R-4 keratinocytes grown on glass coverslips were washed and fixed with 1% (wt/vol) formaldehyde for 10 min. Fixed cells were washed twice with PBS and permeabilized in 0.5% (vol/vol) Triton X-100 in PBS for 5 min. Cells were rinsed with PBS and incubated in 2% (wt/vol) BSA in PBS for 1 h, followed by incubation with the primary antibody for 1 h. After washing twice with PBS, Texas Red- and Cy-5-conjugated secondary antibodies directed against mouse and rabbit immunoglobulins, respectively, were applied for another 1 h. Actin was labeled with phalloidin Alexa 568 at a 1:40 dilution. After washing twice with PBS, coverslips were mounted on microscope slides with Mowiol (Longin et al., 1993) and viewed under a Leica TCS-NT confocal laser-scanning microscope. All steps were performed at room temperature.

Western Blotting Analysis

PA-JEB/4, PA-JEB/EGFP-4, and PA-JEB/4-EGFP keratinocytes were lysed in 1% (wt/vol) SDS, 10 mM Tris-HCl (pH 7.4), containing the proteinase inhibitors phenylmethylsulfonyl fluoride (1 mM), soybean trypsin inhibitor (10 µg/ml), and leupeptin (10 µg/ml). The protein extracts of 2 × 10⁵ cells were loaded per lane on 10% polyacrylamide gels, separated by electrophoresis, and transferred to Immobilon-PVDF membranes (Millipore, Bedford, MA). The membranes were stained with Coomassie blue to indicate the markers, destained (45% methanol, 5% acetic acid in demineralized water), and blocked by incubation in 5% nonfat dry milk in TBST-buffer (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.3% Tween-20) for 30 min at room temperature. Then, the membranes were incubated with the mouse primary antibodies B34 (anti-GFP) or 450-11A (anti-4), diluted 1:500 in 0.5% dry milk in TBST, for 1 h at room temperature. After washing three times with TBST, the membranes were incubated with secondary sheep anti-mouse Ig-coupled horseradish peroxidase (1:5000 dilution) for an additional hour at room temperature. Immunoreactive bands were visualized using enhanced chemiluminescence, according to the manufacturer's instructions (Amersham Corp.).

Measurement of Cell Motility

PA-JEB/EGFP-4, PA-JEB/4-EGFP, and PA-JEB/4^R1281W-EGF keratinocytes were grown to confluence on glass coverslips in keratinocyte-SFM. To assess the relative contribution of cell migration in the absence of proliferation, cells were treated with 10 µg/ml mitomycin C (Sigma Chemical) 2 h before wounding. A cell-free area was introduced by scraping the monolayer with a yellow pipette tip, followed by three washes with PBS to remove cell debris. Scratched areas were photographed at ×200 magnification. Cells were subsequently incubated at 37°C for 48 h in keratinocyte-SFM, and the wounded monolayers were photographed again. In the inhibitor studies, antibodies (GoH3) were added at a concentration of 25 µg/ml, 1 h before wounding.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

4-EGFP and EGFP-4 Induce the Assembly of Hemidesmosomes

In previous studies using keratinocytes transfected with an IL2R-4 chimera, it was shown that 64 can induce the formation of hemidesmosomes without binding to its ligand (Nievers et al., 1998, 2000). Under these circumstances hemidesmosome formation is driven by the cytoplasmic domain of the 4 subunit and is dependent on its association with plectin. To study ligand-dependent and -independent hemidesmosome formation in live cells, two different 4-EGFP chimeras were constructed. The 4-EGFP chimera consisted of EGFP fused to the carboxy-terminal end of the 4 cytoplasmic domain, whereas in the EGFP-4 chimera the extracellular domain of 4 was replaced by EGFP (Figure 1A). The chimeras were introduced by retroviral transduction into 4-deficient PA-JEB keratinocytes to create stable cell lines. Both EGFP chimeras were expressed, and their molecular masses were as expected (Figure 1B). Introduction of the EGFP chimeras did not affect the expression of the other members of the integrin family at the cell surface, because FACS analysis demonstrated that the levels of 2, 3, 5, and 1 subunits were unaffected (unpublished data).

View larger version (44K): [in this window] [in a new window]   Figure 1.   (A) Diagram of 4-EGFP and EGFP-4 constructs. In 4-EGFP the complete 4 subunit was fused to EGFP and in EGFP-4, the extracellular domain of the 4 subunit was replaced by EGFP. (B) Western blot analysis of 4-EGFP and EGFP-4 protein levels in PA-JEB/4-EGFP and PA-JEB/EGFP-4 keratinocytes. Whole cell lysates were resolved by SDS-PAGE and subjected to immunoblotting with monoclonal antibodies against 4 (450-11A) and GFP (B34). 4-EGFP and EGFP-4 are present as a single band of the expected molecular mass. (C). Western blot analysis of laminin-5 in matrices produced and deposited by PA-JEB/4-EGFP and PA-JEB/EGFP-4 keratinocytes. Matrices were solubilized, separated by SDS-PAGE under reducing conditions, and processed for Western blot analysis using anti-laminin-5 antiserum. The four major immunoreactive proteins of 190, 165, 155, and 145 kDa correspond to the unprocessed and processed forms of 3, and to the 2 and 3 chains of laminin-5, respectively.

Fluorescence microscopy of PA-JEB cells expressing the 4-EGFP or EGFP-4 chimera showed that both chimeras are able to induce the formation of hemidesmosome-like structures, the distribution pattern of which resembles that produced by the expression of 4 or IL2R-4 in PA-JEB cells, respectively (Figure 2A). Furthermore, the hemidesmosome-like structures contain plectin, BP180, and BP230. 4-EGFP was always colocalized with patches of LN-5 deposited underneath the cells, whereas sometimes EGFP-4 was not colocalized with LN-5 (Figure 2B). These results support the assumption that 4-EGFP when associated with 6 interacts with LN-5, whereas the EGFP-4 chimera does not.

View larger version (88K): [in this window] [in a new window]   Figure 2.   (A) Similar localization of 4-EGFP and 4, and of EGFP-4 and IL2R-4. Top: PA-JEB/4 and PA-JEB/4-EGFP keratinocytes in a mixed culture were fixed and incubated with antibodies against the cytoplasmic domain of 4 (450-11A), followed by Texas red-conjugated goat anti-mouse IgG. Both 4 and 4-EGFP hemidesmosomes occur in a cauliflower-like staining pattern, and there is no difference between 4 and 4-EGFP hemidesmosomes. Bottom: PA-JEB/IL2R-4 and PA-JEB/EGFP-4 keratinocytes in a mixed culture were fixed, and the IL2R-4 construct was labeled with antibodies directed against the IL2R part of the chimera (in the IL2R-4 chimera, the extracellular part of 4, is replaced by the extracellular part of the IL2 receptor), followed by Texas red-conjugated secondary antibodies. 4-EGFP and IL2R-4 that do not interact with LN-5 are mainly localized at the periphery of cells and the hemidesmosomes formed by them appeared to be more clustered than those formed by 4-EGFP or 4. No differences can be observed between EGFP-4 and IL2R-4 hemidesmosomes. Composite images were generated by superimposition of the green and red signals. Areas of overlap appear yellow in the images. (B) Colocalization of 4-EGFP and EGFP-4 with the hemidesmosomal components BP180, BP230, plectin, and LN-5. PA-JEB keratinocytes expressing 4-EGFP (left panel) or EGFP-4 (right panel) were fixed and immunostained for BP180, BP230, plectin, or LN-5. Composite images were generated by superimposition of the green and red signals. Areas of overlap appear yellow in the images. Both chimeras are colocalized with the hemidesmosomal components BP180, BP230, plectin, and LN-5. Only 4-EGFP is colocalized with LN-5. Bar, 10 µm.

Analysis of the matrices deposited by 4-EGFP- and EGFP-4-expressing cells by immunoblotting, using polyclonal anti-LN-5 antibodies, confirmed the presence of LN-5 in these matrices and furthermore showed that the 3 chain of a proportion of laminin-5 had been proteolytically processed. The ratio of unprocessed (190 kDa) and processed 3 (165 kDa) chain was slightly different between the two cell lines; more 3 chain being processed in PA-JEB/EGFP-4 than in PA-JEB/4-EGFP cells. There was no evidence for extracellular processing of the 2 chain (155 kDa). The band of 145 kDa corresponds to the 3 chain. Furthermore, the total content of laminin-5 in the matrices deposited by PA-JEB/EGFP-4 and PA-JEB/4-EGFP cells was comparable.

In summary, both EGFP chimeras are able to induce the assembly of hemidesmosomes, but although this assembly by EGFP-4 is entirely driven from within the cell (Nievers et al., 2000), that by 4-EGFP is also induced by an interaction of its extracellular domain with LN-5.

Dynamics of Hemidesmosome Formation

Time-lapse videomicroscopy was used to study the distribution of hemidesmosomes during the random movement of keratinocytes. When cultured in high Ca²⁺ medium, PA-JEB/4-EGFP and PA-JEB/EGFP-4 keratinocytes hardly migrate but move primarily in situ by continuously extending and retracting their membrane. Changes in the distribution pattern of hemidesmosomes are readily detected in these cells over a period of 3 h (Figure 3A). Existing hemidesmosomes in the central region of the cell disappear, and new hemidesmosomes are formed at the cell margins. Because of the extensions of the membrane of the cell in various directions, the distribution pattern of hemidesmosomes often has the appearance of a cauliflower. Cell retraction is accompanied by the formation of retraction fibers in which 4-EGFP, but not EGFP-4, is present. The 4-EGFP-positive retraction fibers originate from hemidesmosomes, probably because 4-EGFP when associated with 6 cannot be released from its ligand without effort. In migrating cells, retraction fibers are formed at the rear of the cell and occasionally are left behind as "footprints" (Figure 3B).

View larger version (56K): [in this window] [in a new window]   Figure 3.   Dynamics of hemidesmosome formation. (A) Selected fluorescence micrographs taken from a time-lapse recording at time points 0, 90, and 195 min, depicting the change in distribution of 4-EGFP and EGFP-4 hemidesmosomes in PA-JEB/4-EGFP and PA-JEB/EGFP-4 cells, respectively. The left and two middle panels show an overlay of transmission optics and fluorescence. To improve the contrast of the composite, the green fluorescence of the EGFP chimeras is shown in blue. Right panel: a composite of the fluorescence images at time points 0 (in green) and 195 min (pseudocolored in red), with areas of overlap represented in yellow. Note that after 195 min, new 4-EGFP hemidesmosomes (top panel, red signal) have assembled at the cell borders, whereas preexisting hemidesmosomes (top panel, green signal) in the central part of the cell have disappeared. Bar, 15 µm. (B) "Footprints" of PA-JEB/4-EGFP keratinocytes. PA-JEB/4-EGFP ker-atinocytes were fixed and immunostained for LN-5 (red). Composite images were generated by superimposition of the green and red signals. Areas of overlap appear yellow in the images. Migrating 4-EGFP keratinocytes leave footprints containing ripped off 4-EGFP membrane fragments still attached to the underlying matrix. Bar, 10 µm. (C) Time-lapse images of PA-JEB/4-EGFP demonstrating that hemidesmosomes convert into retraction fibers during cell movement. Left panels: fluorescent images; right panels: the corresponding transmission optics. Note that when the cell body moves forward, hemidesmosomes are gradually converted into retraction fibers (A-F). Some fusion of retraction fibers occurs and eventually the retraction fibers become smaller and are pulled back into the cell (arrows, D-G). The arrow in panel A points at the appearance of retraction fibers. The arrow in panel J points to membrane extensions of the neighboring cell. Note that the disassembly of hemidesmosomes and their conversion into retraction fibers lead to withdrawing of the cell in the left corner, thereby allowing the cell below it to migrate into the cleared space (F-J).

Time-lapse videomicroscopy also revealed that as the cell further retracts its membrane the 4-EGFP retraction fibers become thinner and longer, and new ones emerge (Figure 3C, A-F). Fusion of retraction fibers is also observed (Figure 3C, arrow, panels E and F). Eventually, the fibers become smaller and are pulled back inside the cell. In panels F-J, it can be seen that the empty space left behind by the migrating cell is quickly occupied by another cell that extends its membrane from the leading edge. No 4-EGFP can be detected in these membrane extensions, which suggests that 64 is not an essential component of newly formed filopodia or lamellipodia. Together these results show that both 4-EGFP and EGFP-4 hemidesmosomes are assembled and redistributed in a short period of time. Only 4 chimeras (4-EGFP) that interact with LN-5 are retained in retraction fibers.

Retraction Fibers Mediate the Final Bond of PA-JEB/4-EGFP Cells with LN-5 during Migration and Mitosis

Hemidesmosomes do not disassemble when keratinocytes divide (Baker et al., 1993). Nevertheless, during the mitotic process, cells have to undergo cellular rounding, which is accompanied by a complete reorganization of the cytoskeleton. To investigate whether the localization of hemidesmosomes changes during mitosis, we recorded the fluorescence of 4-EGFP- and EGFP-4-containing hemidesmosomes during spontaneous cell divisions. Figure 4A shows a series of images of a dividing PA-JEB/4-EGFP cell. The images were taken at 15-min intervals over a period of 90 min and after completion of cell division (+8 h). When mitosis begins, the cell starts to round up and to detach from the matrix. The hemidesmosomes located at the cell borders are converted into retraction fibers (Figure 4A, 30 min). When the rounding of the cell is completed, it is still attached to the LN-5 matrix via a network of 4-EGFP positive retraction fibers (45 min). These fibers keep the cell at its position during the mitotic process (60-75 min) and subsequently will facilitate the spreading of the daughter cells (Cramer and Mitchison, 1993; Mitchison and Cramer, 1996). After completion of cell division, the daughter cells spread on the LN-5 matrix that was already present beneath the dividing cell, and new hemidesmosomes are formed at exactly the same position at which the 4-EGFP retraction fibers arose from the matrix (Figure 4A, 8 h). Thus, as in migrating cells, in mitotic cells the 4-EGFP containing hemidesmosomes are converted into retraction fibers. The reverse reaction, the conversion of retraction fibers into 4-EGFP containing hemidesmosomes, also occurs and is most frequently seen after completion of cell division, when the rounded cells begin to spread.

View larger version (119K): [in this window] [in a new window]   Figure 4.   Hemidesmosomes of 4-EGFP and EGFP-4 keratinocytes disappear during mitosis. Time-lapse images of dividing PA-JEB/4-EGFP (A) and PA-JEB/EGFP-4 (B) cells. Shown are fluorescent images with the corresponding transmission optics in the right corner of every panel. Note that 4-EGFP hemidesmosomes convert into retraction fibers, whereas EGFP-4 hemidesmosomes do not but are reduced to small structures that no longer appear as hemidesmosomes. Retraction fibers not only form at the rear end during cell contraction, but also promote respreading of cells after mitosis, consequently the hemidesmosomal pattern before and after mitosis is similar.

EGFP-4 hemidesmosomes do not convert into retraction fibers but are redistributed during mitosis (Figure 4B). Their number at the sites where the cell is attached to the matrix decreases during the process of mitosis. Finally, when the cell is rounded, a spot-like staining pattern represents the only structure that remains of the EGFP-4 hemidesmosomes. After cell division, the EGFP-4 hemidesmosomes reappear in the daughter cells that start to spread. The hemidesmosomal pattern that is finally formed after spreading of the daughter cells is the same as before the onset of mitosis (Figure 4B, 8 h). The transmission images further demonstrate that spreading of the PA-JEB/EGFP-4 cells is associated with extension of the membranes and the formation of bleb-like structures (Figure 4B, 225-240 min). Together, these data demonstrate that during mitosis 4-EGFP hemidesmosomes are converted into retraction fibers, whereas those containing EGFP-4 are reduced to small structures that no longer resemble hemidesmosomes. After cell division is completed hemidesmosomes are reassembled.

4-EGFP Is Not Associated with the Intermediate Filament System in Retraction Fibers

Because retraction fibers that contain 4-EGFP appeared to originate from hemidesmosomes, we investigated which other components of hemidesmosomes are present in them. No BP180, BP230, or plectin could be detected in 4-EGFP-positive retraction fibers. Actually plectin, that mediates the linkage of 4 to the intermediate filament system, is already dissociated from 64 before the retraction fibers become visible (Figure 5A). This suggests that the integrin must be detached from the intermediate filament system before retraction fibers can be formed. Consistent with the fact that plectin is not present in retraction fibers, we could not detect the filament proteins, keratin or vimentin, or tubulin in these fibers either (Figure 5, B, C and D). On the other hand, the tetraspanin CD151 is colocalized with 4-EGFP along the retraction fibers and thus is the only other component that is present together with 64 in both hemidesmosomes and retraction fibers (Figure 5E).

View larger version (57K): [in this window] [in a new window]   Figure 5.   Components of retraction fibers. PA-JEB/4-EGFP were fixed and incubated with mAb 121 against plectin (A) or with mAb KL-1 against keratin (B), a mixture of mAbs B-5-1-2 and 2-28-33 against mouse  and  tubulins (C), mAb RV202 against vimentin (D), mAb P48 against CD151 (E), mAb K20 against 1 (F), phalloidin Alexa 568 to stain actin filaments (G), mAb 3C12 against ezrin (H) and rabbit polyclonal anti-LN-5 antibodies (H). Texas-red- and Cy-5-conjugated secondary antibodies against mouse and rabbit IgG were used to detect bound antibodies. Composite images were generated by superimposition of the green and red signals, with areas of overlap represented in yellow. Where three colors were used (H), composite images were also generated by superimposition of the red, green, and Cy-5 signals, with areas of overlap represented in white. Because 4-EGFP and keratin did not appear in the same confocal plane, (B) is presented as a maximum projection of XYZ stacks. 1, actin, and ezrin, are not part of hemidesmosomes, but appear at the base of retraction fibers, and although present in the same structure are not colocalized with 4-EGFP (F-H). The intermediate filament proteins, keratin and vimentin, and tubulin stain throughout the body of the cell and are not part of hemidesmosomes (B-D). The keratinocyte remains still attached to its LN-5 matrix by retraction fibers (H). The hemidesmosomal components 4-EGFP and CD151 are colocalized in hemidesmosomes and throughout the retraction fibers (E). Plectin is associated with 4-EGFP in hemidesmosomes but is dissociated during the conversion of these structures into retraction fibers (A). Bar, 10 µm.

The 1 integrins are not colocalized with 4 in hemidesmosomes but are present in the focal contacts surrounding them (Schaapveld et al., 1998). Because the retraction fibers of migrating fibroblasts are composed of actin and 1 (Chen, 1981), we investigated whether these components are also present in the retraction fibers of PA-JEB/4-EGFP cells. As shown in Figure 5, 1 is detected in retraction fibers, but it is not colocalized with 4-EGFP (Figure 5F). As expected, F-actin is also present in retraction fibers, and it was found to be more prominently present at their base (Figure 5G). No filamin, spectrin, talin, vinculin, or zyxin could be demonstrated in retraction fibers, which would imply that the 1 integrins are not connected to the actin cytoskeleton. However, ezrin, a protein that connects the cortical actin filaments with the plasma membrane, was detected in retraction fibers and, like F-actin, is more prominently localized at their base (Figure 5H). Ezrin is not present in hemidesmosomes. The 4-EGFP and ezrin-positive fibers are bound to LN-5 left behind after the cells have moved (unpublished data), confirming that they are in fact retraction fibers. Taken together these results suggest that 64 and 1 integrins present in retraction fibers are not associated with the cellular cytoskeleton.

Stabilization of Membrane Extensions by 4-EGFP Is Not Required for Cell Migration

In several reports it has been shown that 64 associates with actin and is localized at the leading edge of invading carcinoma cells, where it is assumed to contribute to migration by stabilizing filopodia and lamellipodia (Rabinovitz et al., 1997; O'Connor et al., 1998; Goldfinger et al., 1999; Decline and Rousselle, 2001). We, therefore, studied the localization of 64 in PA-JEB/4-EGFP keratinocytes migrating into a wound bed. In the wound area the migrating cells were of two types. One type contained large lamellipodia with 4-EGFP evenly distributed over the cell (Figure 6), whereas in the other type 4-EGFP appeared at the leading edge. These cell types most likely represent different phases in the migration of the cell, the first type representing a cell that is actually migrating, whereas the migration of the cells with 4 containing adhesion sites at their base has slowed down or perhaps even completely stopped. Although in the more stationary cells both 31 and 4-EGFP are present at the leading edge, they are clearly not colocalized, 31 being closer to the leading edge than 4-EGFP. Similar results were obtained with PA-JEB cells expressing EGFP-4. To investigate whether the hemidesmosomal structures formed during migration are type I or type II hemidesmosomes, PA-JEB/4-EGFP and PA-JEB/EGFP-4 keratinocytes crossing the wound bed were incubated with antibodies against plectin or BP180. As shown in Figure 6, although plectin was always found together with the 4-EGFP and EGFP-4 clusters at the leading edge of the keratinocytes, BP180 was only occasionally colocalized with them. Thus, PA-JEB/4-EGFP and PA-JEB/EGFP-4 keratinocytes form both type I and type II hemidesmosomes during migration.

View larger version (48K): [in this window] [in a new window]   Figure 6.   During migration in a wound bed PA-JEB/4-EGFP and PA-JEB/EGFP-4 form type I or type II hemidesmosomes. Localization by immunofluorescence of the hemidesmosomal components 4, plectin, and BP180 in PA-JEB/4-EGFP or PA-JEB/EGFP-4 keratinocytes crossing the wounded area after 48 h. Note the different cell types present. Fan-shaped cells with 4-EGFP and EGFP-4 equally distributed over the cells (resulting in PA-JEB/EGFP-4 with hardly visible green fluorescence) most likely represent truly migrating cells. The other cells likely just stopped migrating and both 4-EGFP and EGFP-4 appear at the leading edge of the cell, where 31 is also present. Plectin is always colocalized with 4-EGFP and EGFP-4 at the leading edge, whereas BP180 is occasionally present. Thus both type I and type II hemidesmosomes are formed by keratinocytes migrating into a wound bed. Bar, 20 µm, except for (A), where it indicates 10 µm.

The Integrin 64 Slows Down Migration of Keratinocytes in Response to Wounding

The finding that 64 induces the formation of hemidesmosome-like structures during migration of PA-JEB/4-EGFP keratinocytes prompted us to further investigate the role of this integrin in cell migration using in vitro wound healing assays. PA-JEB/4-EGFP and PA-JEB/EGFP-4 keratinocytes were grown to confluence after which a scratch wound was introduced in the monolayer. Phase contrast micrographs taken at 48 h demonstrated that the migration of PA-JEB/4-EGFP keratinocytes is slower than that of PA-JEB/EGFP-4 keratinocytes, implicating an inhibitory effect of 4 on cell migration when it can bind to its ligand LN-5 (Figure 7). Indeed, the mAb GoH3 known to block the adhesion of 64 to LN-5 enhances migration of PA-JEB/4-EGFP cells. Furthermore, when the interaction of 4-EGFP with LN-5 is inhibited by the mAb GoH3, the 4-EGFP hemidesmosomes adopted the appearance of EGFP-4 hemidesmosomes (see characteristic patterns of 4-EGFP- and EGFP-4-hemidesmosomes in Figure 2A), the clusters of hemidesmosomes were more clearly stained and defined, and there was no 4-EGFP in retraction fibers.

View larger version (75K): [in this window] [in a new window]   Figure 7.   Cell motility is inhibited by binding of 64 to plectin. PA-JEB/4-EGFP, PA-JEB/EGFP-4 and PA-JEB/4R1281W-EGFP keratinocytes were grown to confluence on glass coverslips and treated for 2 h before wounding with 10 µg/ml mitomycin C to prevent proliferation of cells. The monolayers were then wounded and cell migration was determined after 48 h in the presence or absence of 6 blocking antibodies. Expression of the LN-5 binding 4-EGFP chimera in PA-JEB cells slows down cell migration. Blocking of the 64-LN-5 interaction by using the 6 blocking antibody GoH3 at a concentration of 25 µg/ml, enhanced migration of PA-JEB/4-EGFP keratinocytes and made it comparable to that of PA-JEB/EGFP-4 cells. There was no effect of 64 on migration when a mutant of 4 (4R1281W), incapable of interacting with the cytoskeleton was used. Blocking of the 64-LN5 interaction did not promote migration. In the bar graph, results are expressed as mean ± SD (n = 3).

Because 64 can interact with plectin and thereby become connected to the cytoskeleton, it is possible that the inhibition of migration by 64 is not due to the increased adhesion to LN-5 but rather to the stabilization of this adhesion as a result of the interaction of 64 with the cytoskeleton. To investigate this, we generated an EGFP version of a mutant 4 subunit, 4^R1281W, that can bind to LN-5 but not to plectin (Geerts et al., 1999; Koster et al., 2001). We found that when this mutant is stably expressed in 4-negative PA-JEB keratinocytes, migration was hardly affected. In fact, the PA-JEB/4^R1281W-EGFP keratinocytes move as fast as PA-JEB/EGFP-4 keratinocytes (Figure 6). We conclude that the inhibition of migration by 64 is due to the stabilization of the bond of 64 and LN-5 through the interaction of 4 with plectin.

FRAP Analysis of EGFP-tagged Fusion Proteins in Live Keratinocytes

FRAP was used to determine the dynamics of 4-EGFP and EGFP-4 in hemidesmosomes of live keratinocytes. PA-JEB/4-EGFP and PA-JEB/EGFP-4 cells were examined by time-lapse two-photon excitation microscopy (Figure 8). After photobleaching, fluorescence of 4-EGFP had recovered for 32% (±2%) within the first 4 min and for 56% (±6%) after 10 min. The recovery of EGFP-4 occurred much more slowly, i.e., 12% (±1%) after 10 min. There was no fast recovery in the first 4 min, and the rate of the EGFP-4 recovery remained constant throughout time. Thus, in hemidesmosomes of live cells, EGFP-4 is less dynamic than 4-EGFP. Because the 4-EGFP and EGFP-4 chimeras only differ in their capacity to bind the ligand LN-5, this suggests that interaction with ligand increases the dynamics of 64. To investigate whether indeed the dynamics of 4 are dependent on its interaction with LN-5, binding of 4-EGFP to LN-5 was blocked by adding the 6 blocking antibody GoH3 to PA-JEB/4-EGFP keratinocytes, a treatment that results in a distribution of hemidesmosomes comparable with that in PA-JEB/EGFP-4 (see above). When these cells were subjected to FRAP analysis, recovery of fluorescence occurred as slowly as that in the case of EGFP-4. Thus, the bond between 64 and LN-5 is not stable and can be broken, which makes the integrin a dynamic protein. In PA-JEB/4^R1281W-EGFP cells, clusters of 4^R1281W-EGFP can be observed. However, these clusters cannot be considered to be hemidesmosomes, because plectin, an essential component of hemidesmosomes, cannot bind to this 4 mutant. FRAP analysis of 4^R1281W-EGFP was performed to investigate the contribution of the interaction of 4 with the cytoskeleton to the dynamics of 4. The recovery of fluorescence was faster with 4^R1281W-EGFP than with 4-EGFP. Within the first 4 min, fluorescence recovers for 58% (±4%), and eventually the recovery reaches 77% (±5%) after 10 min. Diffusion coefficients for the different proteins were not calculated because the fluorescence of EGFP-4 and 4-EGFP+GoH3 hardly recovered and because the fluorescence recovery curves for 4-EGFP and 4^R1281W-EGFP do not fit with a single exponential (see MATERIALS AND METHODS). However, the observed differences between the different recovery curves are consistent and striking. Thus, the bond between 64 and LN-5 is less easily broken (i.e., less dynamic) when 64 is associated with cytoskeleton.

View larger version (96K): [in this window] [in a new window]   Figure 8.   Fluorescence recovery after photobleaching (FRAP) of EGFP chimeras is dependent on LN-5 binding. PA-JEB/EGFP-4, PA-JEB/4-EGFP, PA-JEB/4R1281W-EGFP keratinocytes and PA-JEB/4-EGFP cells that had been treated with the 6 blocking antibody GoH3 were photobleached for 1 s using 100% laser power at t = 0. Fluorescence recovery was subsequently monitored by acquiring images every 15 s for a period of 10 min. Top panels: fluorescence images of cells before bleaching, immediately after bleaching, and 10 min after bleaching. The bleached area is depicted by a white oval. Bottom panels: representative recovery curves of fluorescence in the bleached area. The fluorescence was related to the initial fluorescence (set at 100%) and corrected for loss of fluorescence due to bleaching and scanning procedure. The presented figures represent a typical experiment of which four that were performed with similar results. After photobleaching, fluorescence of 4-EGFP had recovered for 32% (±2%) within the first 4 min and for 56% (±6%) after 10 min. The recovery of EGFP-4 occurred much more slowly, i.e., 12% (±1%) after 10 min, and was similar to the recovery of 4-EGFP incubated with mAb GoH3, 13% (±0.5%). The 4R1281W-EGFP shows the fastest recovery of fluorescence compared with 4-EGFP. Within the first 4 min the fluorescence recovers for 58% (±4%). Eventually, leading to a recovery of 4R1281W-EGFP fluorescence of 77% (±5%) after 10 min. Thus, the dynamics of the interaction of 4 with LN-5 is suppressed by the association of 4 with the cytoskeleton.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The integrin 64 is an essential component of hemidesmosomes and is necessary for tightly anchoring keratinocytes to the extracellular matrix (Jones et al., 1998; Borradori and Sonnenberg, 1999). When epithelial cells are induced to migrate in response to wounding, they lose their hemidesmosomes, probably to reduce their strong adhesion to the substratum (Riddelle et al., 1992; Gipson et al., 1993). In carcinoma cells, 64 contributes to migration and invasion by activation of PI-3 kinase signaling (Chao et al., 1996; Shaw et al., 1997). Thus 64 plays a dual and apparently paradoxical role because it is essential both to keep cells stationary and to promote migration.

In this study, we analyzed the dynamics of 64 in stationary and migrating keratinocytes by expressing a 4 subunit tagged with EGFP, in 4-deficient keratinocytes. Furthermore, the involvement of ligand binding in the dynamics of 64 was investigated by using an EGFP-4 chimera that is unable to bind to LN-5. Time-lapse videomicroscopy demonstrated that in moving and dividing keratinocytes, 4-EGFP and EGFP-4 hemidesmosome-like structures are assembled and redistributed within minutes. Unfortunately, we could not determine whether these hemidesmosomal structures are type I or type II hemidesmosomes, because the other hemidesmosomal components plectin, BP180, and BP230 could not be made visible during the time-lapse recordings.

During migration, the leading edge of lamellipodia and filopodia is the site where new adhesions are formed. Both in randomly moving cells and in cells that migrate into cleared areas after the monolayer has been wounded, clustered 4-EGFP appears at the leading edge of the keratinocytes, as has also been shown by others (Goldfinger et al., 1999). The LN-5 binding integrin 31 was also concentrated at these sites and was even closer to the leading edge of the cell than 4-EGFP. Several reports have suggested a role of 64 in stabilizing newly formed filopodia and lamellipodia in order to facilitate migration (Rabinovitz and Mercurio, 1997; O'Connor et al., 1998). However, we show that EGFP-4, which cannot interact with LN-5, also becomes concentrated at these sites. This suggests that 4 is involved in a different process, i.e., the formation of new hemidesmosomes. This assumption is supported by the colocalization of plectin and, in some cells of BP180 and BP230, with EGFP-4 at these sites. In contrast, in highly motile cells displaying a characteristic fan-shaped morphology, neither 4-EGFP nor EGFP-4 are clustered but are diffusely distributed throughout the cell. The mechanism responsible for the localization of EGFP-4 at the leading edge is not known, but we assume that it is targeted by a direct or indirect association with 31. A role of the latter integrin in hemidesmosome formation has previously been suggested (Nievers et al., 1998, 2000; Sterk et al., 2000) and is supported by the strong reduction in the number of hemidesmosomes in mice that do not express 1 in the skin (Brakebusch et al., 2000; Raghavan et al., 2000).

Cell migration not only depends on the formation of cell-matrix adhesion by the moving cell, but also on breaking existing adhesions at the rear end of the cell. (Regen and Horwitz, 1992; Palecek et al., 1998). If these adhesions are strong, they will be less easily broken, which will result in the appearance of retraction fibers, causing migration to be slowed down. In PA-JEB/4-EGFP keratinocytes many retraction fibers can be observed that originate from hemidesmosomes. These retraction fibers contain 4-EGFP, but no plectin, which suggests that in them the linkage of 4 with the cytoskeleton is fractured. Indeed, the intermediate filament proteins keratin and vimentin, to which plectin can bind, are not present in these retraction fibers. Incidentally, some of the 4 is ripped from the membrane and remains attached to the LN-5 matrix, leaving "footprints" of a migrating cell.

The detachment of plectin from 4 during the transition of hemidesmosomes into retraction fibers may be due to either mechanical stress or signaling events. Several studies have shown that phosphorylation of the 4 subunit by protein kinase C is associated with a redistribution of 64 from the hemidesmosome to the cytosol (Alt et al., 2001) and/or to F-actin-rich cell protrusions (Rabinovitz et al., 1999). Furthermore, the link between 4 and plectin might be cleaved by proteases such as the Ca²⁺-dependent protease calpain. This stimulates rear end release of CHO cells (Palecek et al., 1998) by cleavage of cytoskeletal linkages. In fact, calpain cleavage sites are present in the cytoplasmic domain of the 4 subunit (Giancotti et al., 1992). The reverse reaction, the reassociation of plectin with 64 occurs when retraction fibers convert into hemidesmosomes, which occurs prominently after mitosis (see Figure 4A).

Our study clearly shows that the introduction of 4-EGFP into a cell that lacks 4 slows down the migration of that cell. Previous work has shown that the proteolytic processing of the 3 chain of LN-5 by plasmin produces an LN-5 molecule that induces the assembly of hemidesmosomes and impedes cell migration (Goldfinger et al., 1998). On the contrary, proteolytic processing of the 2 chain of LN-5 has been associated with the induction of cell migration (Giannelli et al., 1997