INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Cell migration is crucial to embryonic development, tissue remodeling, and cancer invasion. To migrate properly, cells must integrate multiple incoming signals. Once committed to migration, they coordinately regulate, both spatially and temporally, surface receptors and cytoskeleton to generate traction and movement (Huttenlocher et al., 1996; Lauffenburger and Horwitz, 1996). Migration usually occurs over extracellular matrix (ECM) and is accompanied by characteristic morphological changes. Cell protrusions (e.g., filopodia or lamellipodia) are sites where adherence contacts for traction generally are formed. To accomplish forward movement, there must be a balance between the establishment of plasma membrane-ECM adherence contacts at the cell leading edge and their coordinated asymmetric release at the cell trailing edge (Huttenlocher et al., 1995). Although significant advances have been made in identifying molecules involved in cell migration, molecular mechanisms are poorly defined.

Integrins are a class of cell surface receptors that recognize ECM proteins and cellular ligands. They play an important role in migration because they can generate traction by forming mechanical transmembrane links between ECM and cytoskeleton (Lauffenburger and Horwitz, 1996). In addition, ligation of integrin receptors by ECM may transduce signals (Hynes, 1992) interfacing with a cascade of second messengers (Schwartz et al., 1995). These two aspects of integrin function are likely linked; i.e., signaling events may regulate or influence interactions with cytoskeleton or ECM, and vice versa. Integrins consist of two transmembrane subunits,  and . To date there have been more than 14  and 8  subunits identified in humans (Hynes, 1992). To define postreceptor occupancy events that mediate mechanical strength and signaling, considerable efforts have focused on the short cytoplasmic tails of both  and  chains. The  chain cytoplasmic tails interact directly with cytoskeletal components, e.g., talin (Horwitz et al., 1986; Knezevic et al., 1996). They may also be involved in regulating signaling (Guan et al., 1991; Lewis and Schwartz, 1995) and cell migration (Filardo et al., 1995).

The  chain cytoplasmic tails have been implicated in setting the activated versus resting state of integrins, presumably via as yet undefined signaling events that affect integrin conformation (O'Toole et al., 1994). The 2 cytoplasmic tail may interact with actin directly (Kieffer et al., 1995), and the 4 cytoplasmic tail was implicated in supporting random migration, spreading, and adhesion (Chan et al., 1992; Kassner et al., 1995).

The  subunits of the laminin-binding integrins 3, 6, and 7 (Hynes, 1992) are found as two isoforms, A and B, that differ by the cytoplasmic domains (Tamura et al., 1991; Mercurio, 1995). Expression of the 6A or B tail (Figure 1), probably based on alternative exon splicing, is developmentally regulated in a position-specific manner (Collo et al., 1995; Thorsteinsdottir et al., 1995). The 6A1 isoform induced a more migratory phenotype than 6B1 in transformed macrophages (Shaw and Mercurio, 1994). The 7 integrin supported migration in human 293 kidney cells (Echtermeyer et al., 1996) or MCF7, human mammary epithelial cells (Yao et al., 1996), but no major differences between the two isoforms were observed.

View larger version (8K): [in this window] [in a new window]   Figure 1.   Alternate cytoplasmic tails of the 6 integrin subunit. Shown is the single letter amino acid sequence of the 6A and 6B cytoplasmic tails beginning at amino acid 988. The extracellular domains are identical from amino acid 1 through 994 after which the sequences are completely divergent. An arrowhead marks the point where the sequences begin to differ.

This laboratory showed that 61 is the receptor for laminin (Ln)-1 in mouse embryonic stem (ES) cells and that undifferentiated ES cell lines, including ES1, expressed exclusively the 6B isoform (Cooper et al., 1991). ES cells injected into blastocysts can participate in formation of any tissue (Robertson, 1987). Furthermore, they can be induced to differentiate in culture and differentiation occurs concomitantly with expression of the 6A isoform (Cooper et al., 1991). In the developing mouse embryo, 6B was the only isoform expressed until day 8.5 of development, at which time 6A became detectable, but only in the developing heart as a gradient increasing from the outer to the inner myocardial layers (Collo et al., 1995).

In this article, we describe that expression of 6A in ES1 cells induced filopodia extension and migration on Ln-1. Intriguingly, these effects were independent of 6A adhesive functions, because they occurred also on substrates that did not require 6A for ES cell adhesion, i.e., Ln-5 and Fn. However, although 6A did not have to engage with an ECM ligand to induce migration, it participated in molecular interactions inhibitable by anti-6 antibodies and mimicked by clustering. Furthermore, 6A1-induced motility, but not cell adhesion, was inhibited by antibodies to CD81, a member of the tetraspanin family of cell surface molecules recently found to form complexes with certain integrins including 61 (Berditchevski et al., 1996).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cells and Transfections

ES1 cells are a subline isolated by growing D3 ES cells without a fibroblast feeder layer in the presence of leukemic inhibitory factor (LIF) on tissue culture plates coated with denatured gelatin (Cooper et al., 1991). ES1 cells were routinely passaged by using 1× trypsin/EDTA (Sigma, St. Louis, MO) and grown in DMEM (BioWhittaker, Walkersville, MD) supplemented with 10% fetal calf serum (Gemini Bio-Products, Calabasas, CA), 0.0007% 2-mercaptoethanol, 2 mM glutamine (Irvine Scientific, Santa Ana, CA), and LIF supplied as a 1:10,000 dilution of supernatant from CHO cells transfected with LIF (Abe et al., 1991; 10% CM). Cells used for adhesion assays were harvested by using 2.5 mM EDTA in HBSS (Life Technologies, Gaithersburg, MD) followed by extensive washing in DMEM. For expression, full-length human 6A or 6B cDNA was subcloned into the expression vector pBJneo. 6 expression was driven by the SR promoter (Takebe et al., 1988), which is composed of the simian virus 40 early promoter and the R-U5 region from the long terminal repeat of human T-cell leukemia virus type I. For transfection, 1 × 10⁷ ES1 cells, from early passages (passages 15-19), were trypsinized, washed twice with EP buffer [20 mM N-(2-hydroxyethyl)piperazine-N-(2-ethanesulfonic acid), pH 7.0, 137 mM NaCl, 5 mM KCl, 0.7 mM Na₂HPO₄, 6 mM D-glucose, 0.1 mM 2-mercaptoethanol, 2.5 mM NaOH] and electroporated with 20 µg of DNA at 250 µF and 250 V (Bio-Rad, Hercules, CA). Cells were incubated on ice for 10 min and then cultured at 37°C in 5% CO₂ in 10% CM. After 24 h, selection medium containing 300 µg/ml G418 (Life Technologies) was added to the cultures and changed every other day. Approximately 10-14 d after transfection, G418-resistant colonies were trypsinized and expanded.

Flow Cytometry

For flow cytometry (FACS) analysis, cells were harvested and washed with 5% fetal calf serum in HBSS. Primary monoclonal antibody (mAb)BQ16, anti-human 6, (Liebert et al., 1993; ascites fluid diluted 1:500), 2B7 (Shaw et al., 1993; 10 µg/ml), or EA1 (Ruiz et al., 1993) (10 µg/ml)was bound to cells for 30 min on ice. Cells were washed, resuspended in goat anti-mouse-fluorescein isothiocyanate (FITC; Sigma) or goat anti-rat-FITC (1:40 dilution; Calbiochem, San Diego, CA), and incubated 30 min on ice. Cells were again washed, resuspended in 2% fetal calf serum in HBSS, and either analyzed on a FACScan or BQ16-stained cells were sorted on a FACSort machine (Becton Dickinson, San Jose, CA).

Morphological Analysis

Glass coverslips were coated with 20 µg/ml Ln-1 or Fn in HBSS for 30 min at 37°C. Cells (4 × 10⁴ cells) were grown in migration medium (DMEM, 2 mM glutamine, 1 mM sodium pyruvate, LIF) on coated coverslips overnight at 37°C. Where mAb treatments were used, cells were incubated at room temperature with 10 µg/ml GoH3 in migration medium for 30 min before plating. For cross-linking experiments, cells were incubated at room temperature with 10 µg/ml GoH3 for 15 min, and then 10 µg/ml affinity-purified goat anti-rat immunoglobulin (Calbiochem) was added for an additional 15 min before plating. After 18 h, cells were washed with phosphate-buffered saline (PBS), fixed in 2.5% paraformaldehyde in PBS, and mounted in Immunofluore (ICN, Costa Mesa, CA). Cells were visualized in phase contrast by using a Zeiss Axiovert microscope and photographed on TMAX 400 ASA film.

Immunoprecipitation/Western Blot Analysis

Detergent lysates were prepared from transfected ES cells. Briefly, cells were trypsinized, blocked with trypsin inhibitor, washed with PBS, and lysed with 2% Renex in PBS containing 0.174 µg/ml phenylmethylsulfonyl fluoride, 0.7 µg/ml pepstatin A, 0.5 µg/ml leupeptin, and 2 µg/ml aprotinin. Lysates were incubated on ice for 1 h then centrifuged at 40,000 rpm for 1 h. Lysates were precleared with Sepharose beads and then immunoprecipitated overnight at 4°C with the anti-1 mAb 9EG7 (Lenter et al., 1993). For immunoprecipitation, 9EG7 was covalently coupled to CNBr-activated Sepharose (Pierce, Rockford, IL). Immunoprecipitated proteins were washed with 50 mM Tris(hydroxymethyl)aminomethane hydrochloride, pH 7.4, 0.5 mM NaCl, and 0.1% Tween 20, eluted, and separated on 4-20% SDS-PAGE gradient gels (Novex, San Diego, CA) under reducing and nonreducing conditions, and then transferred to polyvinylidene difluoride (Millipore, Bedford, MA). Coprecipitation of the transfected human 6A or 6B with mouse 1 was detected with polyclonal antibodies raised against the cytoplasmic tail of 6A (6845) or 6B (0530) that had been affinity purified on an 6A peptide column (Collo et al., 1995) or similarly on an 6B peptide column. Affinity-purified antibodies were detected with a secondary antibody coupled to horseradish peroxidase (Amersham, Arlington Hts., IL) followed by ECL detection.

Migration Assays

Transwell filters (8.0 µm, pore size; Costar, Cambridge, MA) were coated for 4 h at 37°C with various concentrations of Ln-1 (Life Technologies), 40 µg/ml human Fn (Life Technologies), 1 µg/ml Ln-5 (kind gift from Dr. M. Fitchmun, Desmos, San Diego, CA) diluted in 1 mg/ml ovalbumin/HBSS. Cells (1 × 10⁴ cells/filter) were plated on the uncoated side of the filter in migration medium (DMEM, 2 mM glutamine, 1 mM sodium pyruvate, LIF). For anti-integrin, anti-Ln-1, anti-Ln-5, or anti-CD81 antibody blocking experiments, cells were incubated at room temperature with 10 µg/ml GoH3, 1:100 dilution of supplied concentration of affinity-purified anti-Ln-1 (Sigma), 25 µg/ml CM6 (Plopper et al., 1996), 75 µg/ml 2F7 (Boismenu et al., 1996; hamster anti-mouse CD81), or 75 µg/ml hamster anti-Trinitrophenol (TNP) (PharMingen, San Diego, CA) in migration medium for 30 min before plating on filters. For monensin treatment, ES6A, ES6B, or ESneo cells were incubated in migration medium or migration medium including 0.01, 0.1, or 1 µM monensin (Sigma). For cross-linking experiments, cells were incubated at room temperature with 10 µg/ml GoH3 for 15 min, and then 10 µg/ml affinity-purified goat anti-rat (Calbiochem) was added for an additional 15 min before plating on filters. Cells were maintained at 37°C in a humidified incubator containing 5% CO₂ for 18 h, and then cells were fixed and stained by using the Diff-Quik stain kit (Baxter, McGaw Park, IL). The uncoated side of each filter was wiped with a cotton-tipped applicator to remove cells that had not migrated though the filter. Filters were viewed under bright-field optics. To quantify migration, stained cells were counted in four fields (by using a 40× objective) from each of two filters for each condition. Results of representative experiments are expressed as the number of cells counted in each field (mean ± SD).

Adhesion Assays

Untreated 96-well plates (Sarstedt, Newton, NC) were coated for 4 h at room temperature with mouse Ln-1 (Life Technologies; 20 µg/ml), Ln-5 (1 µg/ml), or with human Fn (Life Technologies; 20 µg/ml). All proteins were diluted in 100 mM carbonate buffer, pH 9.3. Plates were then washed twice with PBS containing 0.2% Tween 20 and blocked 1 h with Blotto (5% nonfat dried milk in PBS and 0.2% Tween 20). Cells were collected by treatment with 2.5 mM EDTA in HBSS, washed twice with DMEM, 1% bovine serum albumin, and then plated (1.2 × 10⁵ cells/well) in DMEM, 1% bovine serum albumin, and 25 mM N-(2-hydroxyethyl)piperazine-N-(2-ethanesulfonic acid), pH 7.2. For anti-integrin antibody blocking experiments, cells were incubated at room temperature with 10 µg/ml GoH3 for 30 min before addition to plates; blocking antibodies were present during plating. Plates were kept at 37°C in a humidified incubator containing 10% CO₂ for 30 min. To remove unbound cells, wells were then filled with PBS, and the plates were inverted in a tank of PBS and allowed to gently shake for 15 min. Excess PBS was absorbed from the wells by inverting plates on paper towels. Bound cells were fixed in 3% paraformaldehyde and PBS and then stained with 0.5% crystal violet in 20% methanol/80% H₂O. Wells were washed with water to remove excess dye, then cells were solubilized in 1% SDS, and the amount of dye was quantified by using a Molecular Devices plate reader set to absorb at 595 nm. Bars represent the average and SD of four replicates.

The adhesion assay that measures strength of adhesion is an adaptation of the assays described by Calof and Lander (1991). Briefly, a sheet of polystyrene was cut to fit a 96-hole silicon gasket (Bio-Rad). The wells of the assay plate were coated with a titration of Ln-1 or heat-denatured bovine serum albumin (BSA) diluted in HBSS, incubated at 4°C for 4 h, then washed, and blocked with 2% heat-denatured BSA at 4°C overnight. Transfected ES1 cells were pulse-labeled with 10 µC of [³⁵S]methionine for 1 h in 1 ml of methionine-free medium and then incubated overnight with the addition of 3 ml of 10% CM. Labeled ES cells were introduced into the assay wells at a concentration of 50,000 cells/well. The plates were immediately centrifuged at 80 × g to synchonize cell contact with the substratum. The cells were allowed to bind for 30 min at 37°C, and then the plates were flooded with warm PBS, sealed, inverted, and centrifuged for 8 min at 80, 400, or 800 × g. The entire plate, still inverted, was submerged in ice-cold PBS and then in fixative (3.7% formaldehyde, 5% sucrose, 0.1% Triton X-100, PBS). After air-drying, the bound radioactivity, representing cell adhesion, was quantified on a Molecular Dynamics PhosphorImager. Each point represents the average and SD of four replicates.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Murine ES1 cells expressing human integrin subunit 6A (ES6A) were obtained by transfection, as described in MATERIALS AND METHODS. Surface expression of human 6A was verified by flow cytometry with two mAbs to human 6 (BQ16 and 2B7) that reacted equally well with ES6A cells (Figure 2A). Furthermore, ES6A cells expressed heterodimers of human 6A associated with endogenous mouse 1, because immunoprecipitates of mouse 1 integrins contained 6A by Western blotting (Figure 2D). Shown as control are flow cytometry analysis of ES1 cells transfected with vectors expressing either human 6B protein (ES6B; Figure 2B) or neomycin-resistance protein (ESneo; Figure 2C) and specific reactivity of the anti-6A antiserum (Figure 2E).

View larger version (34K): [in this window] [in a new window]   Figure 2.   ES6A transfectant cells express human 6A on the cell surface, complexed with endogenous mouse 1. Expression of full-length human 6A or 6B cDNA was analyzed by FACS of ES6A (A), ES6B (B), and ESneo cells (C) with FITC-secondary antibody alone (solid peaks), mAb BQ16 and 2B7 (anti-human 6; dashed/dotted lines), and EA1 (anti-mouse and human 6; solid lines). Cells were analyzed on a FACScan machine (Becton Dickinson). (D) Cell lysates were subjected to immunoprecipitation using Sepharose beads coupled to the anti-1 mAb 9EG7. Beads were eluted in nonreducing (odd-numbered lanes) or reducing (even-numbered lanes) sample buffer, separated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes. Whole lysate (lanes 1, 2, 5, and 6) was electrophoresed along with the anti-1 immunoprecipitations. Lanes 1-4, probed with anti-6A antibody; lanes 5-8, probed with anti-6B antibody. (E) As control to show specificity of the anti-peptide antibodies, whole lysates from ES6A (lanes 1 and 3) and ES6B (lanes 2 and 4) were separated under nonreducing conditions. Lanes 1 and 2, probed with anti-6A antibody; lanes 3 and 4, probed with anti-6B antibody. In both cases (D and E), the arrow indicates a band corresponding to nonreduced 6 polypeptide detected at approximately 140 kDa.

Inspection by light microscopy revealed that ES6A transfectants displayed morphological features unusual for undifferentiated ES cell cultures. ES6A cells formed loose colonies, with many individual cells appearing well separated (Figure 3, top). Overall, morphology was reminiscent of motile cells, because many cytoplasmic protrusions or spikes were evident, recognizable as filopodia and lamellipodia (Figure 3, arrowheads). These morphological features were observed reproducibly in seven independent transfection experiments. They were not due to clonal variations, because ES6A transfectants were grown as bulk cultures that were FACS-selected for surface expression of human 6A (see MATERIALS AND METHODS). Moreover, control ES6B (Figure 3, middle) and ESneo (Figure 3, bottom) transfectants displayed the same morphology as wild-type ES1 cells (which express endogenously only 6B not 6A), i.e., cell cultures contained mostly compact multicellular islands with smooth borders and rare isolated cells (Figure 3, middle and bottom). These results indicated that expression of transfected 6A was likely responsible for the morphological changes of ES6A cell cultures.

View larger version (65K): [in this window] [in a new window]   Figure 3.   Periphery of ES6A cells, but not ES6B or ESneo cells, exhibits lamellipodia and filopodia. Cells were grown on Ln-1-coated glass coverslips overnight, washed with PBS, fixed in 2.5% paraformaldehyde in PBS, and mounted in Immunofluore (ICN). Cells were visualized in phase contrast by using a Zeiss Axiovert microscope and photographed by using TMAX 400 ASA film. Top, ES6A (arrowheads indicate lamellipodia and filopodia); middle ES6B; bottom, ESneo. Bar, 20 µm.

Because the morphology of ES6A cells was reminiscent of motile cells, we performed haptotactic migration assays through porous membranes coated with purified ECM molecules in transwell chambers. Under these conditions, ES6A cells exhibited dose-dependent high levels of migration on purified Ln-1 and Fn (Figure 4, A and B, respectively). Migration was dependent upon the presence of 6A, because control transfectants ES6B and ESneo did not migrate significantly (Figure 4). Furthermore, migration was observed on Ln-5, and migration on each of the purified matrices was blocked by GoH3, a function-blocking mAb to both mouse and human 6 (Figure 5).

View larger version (20K): [in this window] [in a new window]   Figure 4.   ES6A cells migrate on Ln-1 and Fn. Transwell filters were coated for 4 h at 37°C with various concentrations of Ln-1 or Fn. ES6A, ES6B, or ESneo cells (1 × 104 cells/filter) were plated on the uncoated side. Cells were maintained at 37°C for 18 h, and then filters were fixed and stained. To quantify migration, stained cells were counted in four fields (by using a 40× objective) from each of two filters for each condition. Results of representative experiments are expressed as the number of cells counted in each field (mean ± SD).

View larger version (23K): [in this window] [in a new window]   Figure 5.   ES6A cell migration on Ln-1, Fn, and Ln-5 is blocked by anti-6 mAb GoH3. Transwell filters were coated for 4 h at 37°C with 20 µg/ml Ln-1, 20 µg/ml Fn, or 10 µg/ml Ln-5. ES6A, ES6B, or ESneo cells (1 × 104 cells/filter) were plated on the uncoated side. For blocking experiments, cells were incubated at room temperature with 10 µg/ml GoH3 antibody in migration medium for 30 min before plating on coated filters. Cells were maintained at 37°C for 18 h, and then filters were fixed and stained. To quantify migration, stained cells were counted in four fields (by using a 40× objective) from each of two filters for each condition. Results of representative experiments are expressed as the number of cells counted in each field (mean ± SD).

Inhibition of migration on Ln-1 by GoH3 was expected, because 61 is a receptor for Ln-1 in most cell types studied (Hynes, 1992). However, in these cells, 61 is not required as a receptor for Ln-5, and there is no published evidence indicating that 61 may interact with Fn (Hynes, 1992), although migration was blocked by the anti-6 mAb GoH3. In standard adhesion assays, ES6A, ES6B, and ESneo cells adhered readily to Ln-1, Ln-5, or Fn (Figure 6). However, only adhesion to Ln-1 was inhibited by the anti-6 mAb GoH3 (Figure 6). Thus, in ES1 cells, 61 integrin is required as an adhesive receptor for Ln-1 but not for Ln-5 or Fn.

View larger version (40K): [in this window] [in a new window]   Figure 6.   ES6A, ES6B, and ESneo cells adhere to Ln-1 in an 6-dependent manner and to Fn and Ln-5 in an 6-independent manner. Ln-1 (20 µg/ml), Fn (20 µg/ml), and Ln-5 (1 µg/ml) were coated in 0.2 M carbonate buffer onto 96-well plates at 37°C for 4 h. Plates were washed and blocked for 1 h. For blocking with GoH3, cells were pre incubated with 10 µg/ml GoH3 and then plated in the presence of antibody. Cells adhered for 30 min at 37°C, washed, fixed, and stained. Stained cells were solubilized with 1% SDS and absorbance was quantified at 595 nm.

In the case of Ln-1, it was then possible to address the issue as to whether transfected 6A induced migration by changes in short-term cell-substratum adhesion strength, as recently proposed (Pelletier et al., 1996; Palecek et al., 1997). To characterize adhesion strength of ES6A cells on Ln-1, cells were tested in a centrifugal detachment assay. Figure 7 shows that the number of cells adhering to Ln-1 decreased with increasing centrifugal force but that the decrease was the same for ES6A, ES6B, and ESneo cells. We also tested ES6A cells for resistance to detachment under flow conditions (Savage et al., 1996). Our unpublished observations showed no differences from control ES6B and ESneo. These data do not support the idea that transfected human 6A alters adhesion strength to substratum as a mechanism for inducing migration.

View larger version (21K): [in this window] [in a new window]   Figure 7.   ES6A cells do not differ in strength of adhesion to Ln-1. The wells of the assay plate were coated with a titration of Ln-1 or heat-denatured BSA diluted in HBSS, incubated at 4°C for 4 h, then washed, and blocked with 2% heat-denatured BSA at 4°C overnight. [35S]Methionine-labeled ES6A, ES6B, or ESneo cells were plated (10,000 cells/well). The plates were immediately centrifuged at 80 × g to synchronize cell contact with the substratum. The cells were allowed to bind for 30 min at 37°C, and then the plates were flooded with warm PBS, sealed, inverted, and centrifuged for 8 min at 80 × g (top), 400 × g (middle), or 800 × g (bottom). The entire plate, still inverted, was submerged in ice-cold PBS and then in fixative (3.7% formaldehyde, 5% sucrose, 0.1% Triton X-100, PBS). After air drying, the bound radioactivity, representing cell adhesion, was quantified on a Molecular Dynamics PhosphorImager.

On Ln-5 and Fn, migration was also unlikely to be induced by changes in adhesion strength, because 6A1 apparently is not required as an adhesive receptor for these matrices. However, adhesion assays were performed over a shorter time than migration assays. To investigate whether 6A1 may contribute to adhesion to those substrates over a longer period of time, we took advantage of the unique morphology of ES6A cells (Figure 8, A-F). If cells were pretreated with GoH3 and then plated on Fn, the extended cytoplasmic processes (Figure 8A, arrowheads) were retracted completely. These experiments were performed for time periods matching the migration assay. After 18 h, the GoH3-induced retraction of cytoplasmic processes persisted, even though cells were not detached (Figure 8D). This correlated well with the block in migration (Figure 5) and further suggested that transfected 6A induces ES6A migration and morphological changes on Ln-5 and Fn but is not required for adhesion on these substrates.

View larger version (89K): [in this window] [in a new window]   Figure 8.   ES6A filopodia and lamellipodia are blocked by anti-6 antibody GoH3 and are restored by cross-linking GoH3 with secondary antibody. Glass coverslips were coated for 30 min at 37°C with 40 µg/ml Fn. ES6A, ES6B, or ESneo cells (1 × 104 cells/coverslip) were plated. For antibody treatments, cells were incubated at room temperature with 10 µg/ml GoH3 or 10 µg/ml goat anti-rat affinity-pure antibody in migration medium for 15 min. For cross-linking, 10 µg/ml goat anti-rat affinity-pure antibody was added for an additional 15 min to cells that had bound GoH3. No antibodies were removed before plating on the coverslips. Cells were maintained at 37°C for 18 h, then washed with PBS, fixed in 2.5% paraformaldehyde in PBS, and mounted in Immunofluore. Cells were visualized in phase contrast by using a Zeiss Axiovert microscope and photographed using TMAX 400 ASA film. (A, D, and G) ES6A cells: control, GoH3, and cross-linked GoH3, respectively. (B, E, and H) ES6B cells: control, GoH3, and cross-linked GoH3, respectively. (C, F, and I) ESneo cells: control, GoH3, and cross-linked GoH3, respectively. Bar, 20 µm.

It was still possible that ES1 cells deposited Ln-1 or a similar 61 adhesive ligand over the course of the migration assay and that such an adhesive ligand was responsible for the 6A1-induced migration on Ln-5 or Fn. Migration assays were performed on filters saturated with purified matrices and excess ovalbumin to prevent this. Nonetheless, we tested migration in the presence of monensin, a broad inhibitor of secretion. Monensin was toxic to cells in a dose-dependent manner. Therefore, decreases in migration activity may be expected due to cytotoxicity. Because migration on Ln-1 should not require matrix secretion, we took migration levels on Ln-1 as a benchmark for maximum possible migration at any given concentration of monensin. Levels of migration on Ln-5 or Fn were not lower than Ln-1 at any monensin concentration (Figure 9), and our unpublished results showed that at all points, migration was fully inhibitable by GoH3. These results suggested that induction of migration by 6A1 on Ln-5 or Fn did not occur via a deposited adhesive ligand.

View larger version (45K): [in this window] [in a new window]   Figure 9.   Monensin does not abolish 6A-induced migration on Fn. Transwell filters were coated for 4 h at 37°C with 40 µg/ml Ln-1, 40 µg/ml Fn, or 10 µg/ml Ln-5. ES6A, ES6B, or ESneo cells (1 × 104 cells/filter) were plated on the uncoated side of filters in migration medium, or migration medium including 0.01, 0.1, or 1 µM monensin. In all cases, cells were maintained at 37°C for 18 h, and then filters were fixed, stained, and analyzed as previously described.

To validate this point further, we used function-blocking antibodies to either Ln-1 or Ln-5 in migration assays. An anti-Ln-1 antiserum blocked migration on plated Ln-1 but did not affect ES6A migration on Fn or Ln-5 (Figure 10A). CM6, a rat-specific anti-Ln-5 mAb, completely blocked migration on Ln-5 but had no effect on Ln-1 or Fn (Figure 10B). These results demonstrated further that 6A1-induced migration was not due to adhesive ECM deposited by ES6A cells.

View larger version (27K): [in this window] [in a new window]   Figure 10.   ES6A migration is blocked by anti-Ln-1 on anti-Ln-5 antibodies only on the corresponding ECM substrates. Transwell filters were coated for 4 h at 37°C with 40 µg/ml Ln-1, 40 µg/ml Fn, or 10 µg/ml Ln-5. ES6A, ES6B, or ESneo cells (1 × 104 cells/filter) were plated on the uncoated side. For blocking experiments, cells were incubated at room temperature with a 1:100 dilution of anti-Ln-1 antibody in migration medium for 30 min before plating on coated filters (A). For anti-Ln-5 blocking experiments, ES6A cells were incubated at room temperature with 25 µg/ml CM6 or, as control, 25 µg/ml MOPC in migration medium for 30 min before plating on coated filters (B). In all cases, cells were maintained at 37°C for 18 h, and then filters were fixed, stained, and analyzed as previously described.

On the basis of these data, 6A1 induced changes in morphology and migration in ES6A cells without having to engage an ECM ligand. On the other hand, GoH3, a function-blocking anti-6 antibody, did inhibit both morphological changes and migration, raising the possibility that, to cause these effects, 6A1 was involved in molecular interactions that were interrupted by mAb GoH3. To investigate this possibility, we took advantage of the fact that antibody-mediated clustering of integrins can stimulate integrin-initiated functions, e.g., downstream signaling (Pelletier et al., 1992; Schwartz et al., 1995). After GoH3 was allowed to bind ES6A cells, goat anti-rat Ig antibody was added to cross-link the bound GoH3, hence, clustering the GoH3-bound 61. For Ln-1, which required 61 for ES1 cell adhesion (see Figure 6), cross-linked GoH3 inhibited migration to the same extent as GoH3 alone (Figure 11). Remarkably, though, on Ln-5 and Fn, cross-linking reversed the inhibition of migration by GoH3 (Figure 11). Similarly, the morphology of ES6A cells plated on Fn was modulated by cross-linking GoH3: whereas GoH3 alone inhibited the extension of cytoplasmic processes characteristic of ES6A morphology (Figure 8D), cross-linking GoH3 reversed this inhibition and reestablished the motile-like morphology of ES6A cells (Figure 8G). Thus, on substrates that do not require 61 for adhesion (i.e., Ln-5 and Fn), the changes in motility state of ES cells induced by 6A1 may be triggered by the clustering of this integrin.

View larger version (31K): [in this window] [in a new window]   Figure 11.   Anti-6 antibody GoH3 blocks migration on Fn and Ln-5, but cross-linking GoH3 with secondary antibody restores it. Transwell filters were coated for 4 h at 37°C with 40 µg/ml Ln-1, 40 µg/ml Fn, or 10 µg/ml Ln-5. ES6A, ES6B, or ESneo cells (1 × 104 cells/filter) were plated on the uncoated side. For controls, cells were incubated at room temperature with 10 µg/ml GoH3 or 10 µg/ml purified goat anti-rat antibody in migration medium for 15 min. For cross-linking, 10 µg/ml purified goat anti-rat antibody was added for an additional 15 min to cells that had bound GoH3. No antibodies were removed before plating on the filters. In all cases, cells were maintained at 37°C for 18 h, and then filters were fixed, stained, and analyzed as previously described. (A) ES6A. (B) ES6B. (C) ESneo.

To explore further these molecular mechanisms, we tested the effects of antibodies to CD81 on ES cell adhesion and migration. CD81, a tetraspanin protein, was recently shown to physically associate with several integrins, including 6A1 (Berditchevski et al., 1996; Hemler et al., 1996; Rubinstein et al., 1996). By flow cytometry with the anti-CD81 antibody 2F7, wild-type ES cells and the transfectants ES6A, ES6B, and ESneo were positive for surface CD81 (our unpublished results). In adhesion assays, anti-CD81 had no effect on the adhesion of these cells to Ln-1 or Fn (our unpublished observations). In contrast, in migration assays, anti-CD81 strongly inhibited ES6A migration on Ln-1 and Fn (Figure 12), but an isotype-matched immunoglobulin control had no effect. These results indicate that the cell motility induced by 6A1 may occur via a mechanism involving the integrin-associated protein CD81.

View larger version (28K): [in this window] [in a new window]   Figure 12.   Anti-CD81 antibody blocks migration on Ln-1 and Fn. Transwell filters were coated for 4 h at 37°C with 20 µg/ml Ln-1 or 20 µg/ml Fn. ES6A, ES6B, or ESneo cells (1 × 104 cells/filter) were plated on the uncoated side. Cells were incubated with 75 µg/ml hamster anti-CD81 (2F7) or, for controls, cells were incubated with 2% glycerol in PBS (since 2F7 is supplied in 20% glycerol) or with 75 µg/ml hamster anti-TNP antibody in migration medium for 30 min. No antibodies were removed before plating on the filters. In all cases, cells were maintained at 37°C for 18 h, and then filters were fixed, stained, and analyzed as previously described.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Transfecting the 6A integrin isoform into ES cells induced migration and a morphological change on Ln-1, Ln-5, and Fn. The 6A integrin affected the motility state of ES1 cells, without having to engage with the ECM substrate onto which the cells migrated. This effect was blocked by antibodies against CD81, establishing an intriguing link between tetraspanin proteins and integrin-mediated migration.

Expression of 6A changed the morphology of ES1 cells by promoting filopodia and lamellipodia extension and disrupting the compact appearance of cell colonies. ES6A morphology correlated well with increased migration in transwell assays. Both the increased motility and the morphological changes suggest a down-regulation of cell-cell adhesive contacts in ES6A cells. It will be interesting to determine whether 6A1 has any effect on the affinity of, for example, cadherin interactions. Another line of evidence suggested a causal relationship between 6A and morphological changes: on Fn, addition of GoH3 (a function-blocking anti-6 antibody) reversed the morphology of ES6A cells to that of wild-type ES1 cells.

A purely mechanical explanation of migration proposed that cell migration depends upon two crucial parameters, intracellular motile force and receptor/ligand adhesive strength (Lauffenburger and Horwitz, 1996). An optimal ratio between these two parameters would be required for cells to locomote. Recently, we reported that activation of the integrin v3 also alters the migratory behavior of cells possibly by altering the strength of the integrin/ligand interaction (Pelletier et al., 1996). However, we tested this possibility with the ES cell transfectants and found, by two independent assays, no differences in strength of adhesion to Ln-1 induced by 6A expression. Therefore, our data could not be explained satisfactorily on the basis of current theories on the mechanical aspects of migration and suggested additional mechanisms for regulating cell migration.

Our results implicate the cytoplasmic domain of 6 as a regulatory site for migration, because ES cells endogenously express the 6B isoform but do not display the migratory phenotype of ES6A cells. Furthermore, transfection of human 6B, as an additional control, did not induce the migratory phenotype. Structural differences between 6A and 6B are limited to the cytoplasmic domains. It remains to be determined how these structural differences relate to the induction of migration by 6A. There has been a report of differences in tyrosine phosporylation of other molecules induced by ligating 6A versus 6B in transfected macrophages (Shaw et al., 1995b). However, the relationship between these phosphorylation events and migration has not been defined. Additionally, it has not been established whether the two serines that exist in the 6A tail, which can be phosphorylated, play a role in cell migration (Hogervorst et al., 1993; Shaw and Mercurio, 1993).

The following four independent lines of evidence eliminated the possibility that ES6A cells deposited an adhesive ligand responsible for the migration on Fn and Ln-5: 1) anti-Ln-1 antiserum blocked migration on Ln-1 yet had no effect on migration on Fn or Ln-5; 2) an anti-Ln-5 mAb completely blocked migration on Ln-5 yet had no effect on Ln-1 or Fn; 3) vitronectin, collagen IV, or BSA did not support migration (our unpublished observations); and 4) treatment with monensin, an inhibitor of secretion, did not block migration on Fn or Ln-5 any more than the toxic effect observed for ES6A cells migrating on Ln-1. Thus far, on all substrates to which ES cells adhere, ES6A cells showed enhanced migration irrespective of whether 6 is involved directly with adhesion on that substrate.

Our results indicate that adhesion of ES cells to Fn or Ln-5 requires integrins other than 61. By flow cytometry we showed at least two fibronectin receptors on ES cells, 41, and 51 (our unpublished results). For Ln-5, it is likely that 31 is the relevant receptor (Wayner et al., 1993). Unfortunately, the lack of function-perturbing antibodies to these mouse integrins prevented us from positively identifying these integrins as the responsible receptors. When these reagents become available, it should be possible to determine whether in fact those integrins mediate the mechanical aspects of ES6A cell migration on Fn and Ln-5, by generating traction via adhesive interactions with those substrates. Our study predicts that 61 should have an effect on the traction-generating properties of the Fn and Ln-5 receptors.

We found that GoH3 completely blocked migration on Ln-1, Ln-5, and Fn. However, cross-linking of GoH3 with a secondary antibody restored migration on Ln-5 and Fn, i.e., those substrates on which 61 was not required as a receptor for adhesion. This suggests a model whereby expression of 6A is not by itself sufficient to induce migration. Rather, 6A may be required in molecular interactions that can be inhibited by GoH3. It is intriguing, and perhaps unexpected, that clustering of GoH3 reversed this inhibition. Several explanations are possible, because bound GoH3 antibody may interfere with many conceivable mechanisms (e.g., receptor recycling rate) that could be reinstated by clustering. Clustering may mimic the binding of 61 to some counterreceptor that signals cells to become more motile, irrespective of the adhesive ligand on which migration occurs. It might then be expected that distinct structural domains on the 6A subunit may regulate adhesive versus migratory functions. Similarly, Chan and collaborators (Hangan et al., 1997) recently reported a mAb to 61 that inhibits migration but not adhesion. It is possible that the determinant recognized by this antibody participates in the induction of motility we describe herein.

Our data support a correlation between migration and the marked increase in long filopodia in ES6A cells, because these two parameters where either induced or inhibited concurrently. GoH3 alone both inhibited migration and reversed the morphological phenotype. Cross-linking of 6A1 with GoH3 restored both migration and the filopodia. In the haptotactic migration assay, cells presumably extend processes through the pores to detect the matrix on the underside of the filter. Thus, the extension of long filopodia in ES6A cells may be directly responsible for increased migration. Defining the molecular basis for these effects is an area for future investigations.

One important consequence of our data is that inhibition of migration by anti-6 antibodies no longer signifies unequivocally that cells are migrating on adhesive ligands for 61. To our knowledge, this is the first time that a function-perturbing antibody, such as GoH3, was shown to block migration without blocking adhesion. In light of these results, reevaluation of the role of some function-blocking antibodies may be required.

To investigate molecular mechanisms whereby 6A1 influenced migration, we explored the role of integrin-associated proteins, particularly transmembrane ones. Several members of the tetraspanin superfamily, e.g., CD9, CD63, CD81, and CD82, were convincingly documented to form molecular complexes with integrins at the cell surface. Tetraspanin proteins are expressed in many cell types and contain two short cytoplasmic domains (the amino and carboxyl terminus), two unequal extracellular loops, and four membrane spanning segments. Though tetraspanin proteins have been implicated in many cellular activities, including adhesion, migration, and proliferation, their exact function is unknown (Hemler et al., 1996; Maecker et al., 1997). Among the possibilities, they were proposed to act as networking pieces for cell surface proteins, such as signal transducers associated to G proteins, or as ion channels. Any of these functions are conceivable in their part as integrin-associated proteins.

Recently, an antibody to the tetraspanin protein CD81, 2F7, inhibited T cell maturation in fetal thymus organ cultures (Boismenu et al., 1996). We show herein that this same antibody inhibited motility induced by 6A1. These results are consistent with the recent proposals that tetraspanin proteins may influence integrin-mediated migration (Shaw et al., 1995a), primarily based on data showing the enhanced motility of B cell lines transfected with CD9, the inhibition of haptotactic migration with anti-CD9 antibody, and the identification of a tetraspanin as a metastasis suppressor gene (Dong et al., 1995). However, to our knowledge, this is the first report that anti-CD81 antibodies inhibit cell motility.

It is worth pointing out that, in our system, anti-CD81 antibodies blocked the motility induced by an integrin not involved in adhesion to the migratory substrate. Furthermore, the anti-CD81 antibody had no effect on adhesion of ES cells. Therefore, our results are consistent with the prevalent view that tetraspanin proteins may not be involved in modulating integrin-mediated adhesion (Hemler et al., 1996). We propose a model whereby, in ES6A cells, the physical interaction between 6A1 and CD81 triggers signaling events that result in up-regulation of motility and modulation of function of other integrins engaged with migratory substrate. It is also possible that CD81 plays an alternative or additional role at the level of the migratory integrins themselves. Further studies with ES6A cells, aimed at establishing the molecular interactions between CD81 and integrins, as well as their functional consequences, should help clarify the validity of this model.