A series of monoclonal antibodies (mAbs) (all of murine origin) was used in this study. Anti-human αv (L230; Weinacker et al, 1994) was prepared in our laboratory from hybridoma cells obtained from the American Type Culture Collection (Rockville, MD, USA). The anti-αvβ6 (E7P6 and R6G9; Weinacker et al, 1994), anti-αvβ6 (10D5; Huang et al, 1998), anti-α5β1 (P1D6) and anti-β1 (P4C10) antibodies were purchased from Chemicon International (Harrow, UK). Anti-αvβ5 (P1F6) was obtained from Life Technologies, Paisley, UK. W632 (anti-MHC class I) was a kind gift from W Bodmer (IMM, Oxford). FITC- and horseradish peroxidase-conjugated rabbit anti-mouse antibodies were purchased from Dako (High Wycombe, UK). Plasma fibronectin, TGF-β1 latency associated peptide (LAP) and BSA were purchased from Sigma Chemical Co (Poole, Dorset, UK).

Using cDNA transfection techniques we previously created a panel of cell lines expressing various levels of αvβ6 (Thomas et al, 2001b). H357 is an αv-negative oral squamous carcinoma cell line (Prime et al, 1990; Sugiyama et al, 1993) from which the V3 cell line was generated by transfection of αv cDNA (Jones et al, 1996); V3 cells predominantly express the αvβ5 heterodimer. This cell line was retrovirally infected with β6 cDNA creating the VB6 cell line, which has high αvβ6 expression. A null transfectant control cell line for the VB6 cells (C1) was also generated at this time (Thomas et al, 2001b). In light of the report of Park et al (2000) showing the co-activity of activated Ras and TGF-β, it is perhaps worth noting that H357 cells, and the derivative cell lines described above, all retain sensitivity to the growth inhibitory effects of TGF-β (Paterson et al, 1995; data not shown) despite the fact that H357 cells contain c-Ha-ras gene point mutations at the activating codons 13 and 61 (Yeudall et al, 1993).

Cells were grown in standard keratinocyte growth medium (KGM) as described (Sugiyama et al, 1993; Jones et al, 1996). KGM comprised α-MEM containing 10% foetal calf serum (Globepharm, Surrey) supplemented with 100 IU l⁻¹ penicillin, 100 μg l⁻¹ streptomycin and 2.5 μg l⁻¹ amphotericin B (Gibco BRL), 1.8×10⁻⁴ M adenine, 5 μg ml⁻¹ insulin, 1×10⁻¹⁰ M cholera toxin, 0.5 μg ml⁻¹ hydrocortisone and 10 ng ml⁻¹ epidermal growth factor (Sigma). All cells were tested routinely for mycoplasma.

Subconfluent cells were washed twice with PBS and harvested by trypsin/EDTA (0.25% w v⁻¹, 5 mM). Cells were washed once in PBS containing 10% FCS. Cells were incubated with primary antibody for 40 min at 4°C and washed twice with PBS. FITC-conjugated secondary antibody was applied to the cells for 30 min at 4°C. Briefly, cells were washed twice with PBS and resuspended in 0.5 ml PBS with 10% FCS. Labelled cells were scanned on a FACSCalibur cytometer (Becton-Dickinson) and analysed using Cellquest software, acquiring 1×10⁴ events.

A 50 μl solution of LAP at a concentration of 0.25 μg ml⁻¹ was added to 96-well plates (Falcon 3912; Becton Dickinson) and incubated at 37°C for 1 h. After incubation wells were washed with PBS then blocked with 0.1% BSA at 37°C for 30 min. Control wells were incubated with 0.1% BSA. Cells were chromium [⁵¹Cr] labelled (Brunner et al, 1976) and resuspended in PBS containing 2.5 mM EDTA for 10 min. Cells were washed and resuspended in cold α-MEM (1.5×10⁴ cells per well). For blocking experiments, cells were incubated with specific antibodies (as described in Results) for 10 min on ice in each well. Plates were incubated at 37°C for 30 min. Non-adherent cells were removed by flooding plates with PBS supplemented with 1 mM CaCl₂ and 0.5 mM MgCl₂. After two washes, the plates were cut into individual wells and the radioactivity associated with each well was determined in a gamma counter (1261 Multigamma; LKB Wallac, Bromma, Sweden). The per cent adhesion was expressed as the adherent cell radioactivity as a proportion of the total cell input. The non-specific adhesion (attachment to wells coated with BSA) was subtracted. Experiments were repeated on six occasions in quadruplicate, with similar results.

Twenty-four-well plates were coated with fibronectin (10 μg ml⁻¹) or LAP(0.25 μg ml⁻¹). A 200 μl solution was added to the wells and incubated at 37°C for 1 h. After incubation wells were washed with PBS and blocked with 0.1% BSA for 30 min. 10⁵ cells in α-MEM were seeded into each well. For blocking experiments, cells were incubated with anti-αvβ6 antibody (10D5) or an irrelevant control antibody (W632; anti-MHC CLASS 1) for 30 min at 4°C and plated in medium containing an excess of antibody. Supernatant was sampled after 24 h at which time a cell count was carried out. Conditioned medium was cleared of cells and debris by centrifugation at 4000 r.p.m. for 10 min.

MMP-9 activity was analysed using SDS–PAGE-substrate gels. Gelatin (bloom 300, Sigma) was added to a 12% acrylamide separating gel at a final concentration of 1 mg ml⁻¹. To each gel, supernatant samples containing equal volume/cell number were mixed with non-reducing sample buffer (62.5 mM Tris-HCl, pH 6.8, 10% glycerol, 2% SDS, 0.1% bromophenol blue) and added to the gel without boiling. MMP-9 standard was run on each gel. Following electrophoresis, gels were washed twice in 2.5% Triton X-100 for 30 min at 37°C to remove the SDS. Gels were incubated at 37°C overnight in developing buffer containing 50 mM Tris-HCl, 0.2 M NaCl, 5 mM CaCl₂ and 0.02% Triton X-100. Gels were stained with 0.5% Coomassie blue G250 in 30% methanol, 10% glacial acetic acid for 30 min and destained in the same solution without Coomassie blue. Gelatin-degrading enzymes were identified as clear bands against the blue background of the stained gel. Images of stained gels were captured under illumination using the UVP Imagestore 5000 (Ultra-Violet Products UK) and exported for use on a PC using the Scion image program (Scion Corp; based on the Macintosh NIH Image program developed at the National Institutes of Health, USA). The intensity of the bands was measured by densitometric analysis and comparisons made within each gel to determine relative changes in MMP activity. Data for each zymogram were expressed as relative changes in MMP activity and these relative changes compared with repeat experiments. Experiments were repeated a minimum of three times in duplicate or triplicate.

Haptotactic cell migration assays were performed using matrix coated polycarbonate filters (8 μm pore size, Transwell®, Beckton Dickinson). The membrane undersurface was coated with fibronectin (10 μg ml⁻¹) or LAP (0.25 μg ml⁻¹) in PBS for 1 h at 37°C and blocked with migration buffer (0.1% BSA in α-MEM) for 30 min at 37°C. For blocking experiments, cells were incubated with antibody for 30 min at 4°C prior to seeding. The lower chamber was filled with 500 μl of migration buffer, following which cells were plated in the upper chamber of quadruplicate wells, at a density of 5×10⁴ in 100 μl of migration buffer and incubated at 37°C for 12 h. Following incubation, the cells in the lower chamber (including those attached to the undersurface of the membrane) were trypsinised and counted on a Casy 1 counter (Sharfe System GmbH, Germany).

Cell invasion assays were performed using Matrigel coated polycarbonate filters (8 μm pore size, Transwell®, Beckton Dickinson). Matrigel (70 μl; 1 : 2 dilution in α-MEM) was added to the upper membrane and allowed to gel for 1 h at 37°C. For blocking experiments, cells were incubated with anti-integrin antibody for 30 min at 4°C prior to seeding. To act as a chemoattractant, 500 μl of KGM was placed in the lower chamber. Cells were plated in the upper chamber of quadruplicate wells at a density of 5×10⁴ in 200 μl of α-MEM and incubated at 37°C for 72 h. The cells in the lower chamber (including those attached to the undersurface of the membrane) were then trypsinised and counted on a Casy 1 counter (Sharfe System GmbH, Germany).

Data are expressed as the mean±s.d. of a given number of observations. Where appropriate, one way analysis of variance (ANOVA) was used to compare multiple groups. Comparisons between groups were by Fishers PLSD (set at 5% significance). A P value of <0.05 was considered to be significant.

Figure 1Figure 1

Flow cytometric analysis of integrin expression by cell lines. The geometric mean fluorescence (arbitrary units, log scale) as measured by flow cytometry of cells labelled with anti-integrin antibodies is shown. Negative control had secondary antibody only and has been subtracted from the results. Figure shows a representative experiment (ND=not detected). Flow cytometry confirmed high αvβ6 expression by VB6 cells and that C1 null transfectant cells express low levels of endogenous αvβ6. H357 cells are αv- negative. The cell lines express similar levels of β1 integrins.

Preliminary studies indicated that optimal adhesion to LAP was achieved at a concentration of 0.25 μg ml⁻¹ (Figure 2AFigure 2

Cell adhesion to LAP is αvβ6-dependent. Chromium [⁵¹Cr]- labelled cells (1.5×10⁴) were added to LAP-coated 96-well plates containing an irrelevant control antibody (W632 anti-MHC class 1) or test antibodies against αv (L230), αvβ5 (P1F6), αvβ6 (10D5), β1 (P4C10) and α5β1 (P1D6). Background binding to BSA has been subtracted from the results. Figures show representative experiments performed in quadruplicate. Error bars represent standard deviation. (A) VB6 cells show increased adherence LAP. β6 transfected cells (VB6), control cells (C1) and αv-negative cells (H357) were plated onto varying concentrations of LAP. VB6 cells showed significantly increased adhesion compared to C1 cells (which express low levels of endogenous αvβ6). H357 cells did not adhere. (B) Adhesion to LAP is αvβ6-dependent. VB6 and C1 adhesion to LAP was inhibited by anti-αvβ6 antibody or anti-αv antibody. Antibodies against αvβ5, α5β1 or β1 produced no effect. These data suggest that adhesion of VB6 and C1 cells is modulated solely through αvβ6.

VB6 cells show increased αvβ6-dependent migration towards fibronectin relative to the other cell lines (Thomas et al, 2001b). To determine whether VB6 cells migrated towards LAP, haptotactic migration assays were performed using LAP-coated Transwell filters. VB6 cells migrated less well towards LAP than fibronectin (a relative reduction of 53% over four experiments; P=0.039; data not shown). However, migration towards LAP was increased significantly in VB6 cells compared with C1 cells (P=<0.001; (Figure 3AFigure 3

Cell migration towards LAP is αvβ6-dependent. Cells were allowed to migrate towards LAP in haptotactic migration assays. To assess integrin specificity of migration, integrin-blocking antibodies against αv (L230), αvβ5 (P1F6), αvβ6 (10D5), or a control antibody (W632) were added to VB6, C1 and H357 cells prior to plating into wells. Following 8 h incubation, the cells in the lower chamber (including those attached to the undersurface of the membrane) were trypsinised and counted on a Casy 1 counter (Sharfe System GmbH, Germany). Results for the cell lines are expressed relative to VB6 migration following incubation with an irrelevant control antibody (=100). Figures show representative experiments performed in quadruplicate. Error bars represent standard deviation. (A) VB6 cells show increased migration towards LAP. Comparison of migration of VB6, C1 and H357 cells towards LAP. Migration is higher in the αvβ6 expressing VB6 cells. αv-negative H357 cells did not migrate significantly. (B) Migration towards LAP is αvβ6-dependent. Migration of VB6 and C1 cells was inhibited completely by antibody inhibition of αvβ6 or αv. Antibodies inhibiting αvβ5 produced no effect. These data suggest that migration of VB6 and C1 cells is modulated solely through αvβ6.

Expression of MMP-9 on BSA-, fibronectin-, and LAP-coated tissue culture plastic was examined by zymography. Zymography on supernatant samples from cells grown in serum-free medium showed that the cells produced predominantly MMP-9 and that this enzyme was mainly present in proenzyme form. Similar to the effect observed on fibronectin (Thomas et al, 2001c), VB6 cells showed significant upregulation of MMP-9 when plated on LAP (P=<0.001; Figure 4A,CFigure 4

Binding of αvβ6 to LAP induces upregulation of MMP-9. Zymography for MMP-9. Cells were grown for 24 h in additive-free medium before supernatant sampling and cell counting. Samples containing equal volume per cell number were run on each gel with MMP-9 control. The intensity of the bands was measured by densitometric analysis and comparisons made within each gel to determine relative changes in MMP activity. (A) Zymogram showing upregulation of MMP-9 expression by VB6 cells when plated on fibronectin (FN) and LAP relative to BSA-coated plastic. Control is pro-MMP-9. (B) Zymogram showing the inhibition of MMP-9 expression by VB6 cells plated onto LAP following blockade with anti-αvβ6 antibody 10D5 relative to an irrelevant control antibody (Ctl; W632 (anti-MHC class 1)). (C) Densitometric analysis of zymograms from multiple separate experiments (n=4) showing MMP-9 expression by VB6 cells on BSA, LAP following incubation with an irrelevant control antibody (Ctl; W632 (anti-MHC class 1)) and LAP following blockade with anti-αvβ6 antibody (10D5). Results are expressed relative to MMP-9 expression by VB6 cells on BSA-coated plastic (=100). Error bars represent standard deviation. These data suggest that αvβ6-LAP interaction modulates MMP-9 expression in VB6 cells.

VB6 cells bind to fibronectin through both αvβ6 and α5β1 integrins such that to completely inhibit adhesion to this substrate, a combination of blocking antibodies against both integrins is required (Thomas et al, 2001b). To determine whether LAP competitively inhibits αvβ6-dependent binding to fibronectin, VB6 cells were incubated with LAP (0.25 μg ml⁻¹) or antibodies against αvβ6 (10D5) or α5β1 (P1D6) prior to plating. Treating the cells solely with LAP, or antibodies against αvβ6 or α5β1 did not inhibit adhesion to fibronectin significantly (Figure 5AFigure 5

Soluble LAP inhibits αvβ6 interaction with fibronectin. (A) LAP inhibits VB6 adhesion to fibronectin. Chromium [⁵¹Cr]-labelled VB6 cells were added to fibronectin-coated 96-well plates containing an irrelevant control antibody (W632) or test antibodies against αvβ6 (10D5), α5β1 (P1D6) and LAP (0.25 μg ml⁻¹) or combinations thereof. Figure shows a representative experiment performed in quadruplicate. Results are expressed relative to binding in the presence of a control antibody (=100). Background binding to BSA has been subtracted from the results. Error bars represent standard deviation. VB6 cells adhere to fibronectin through αvβ6 and α5β1 and a combination of antibodies against both integrins is required to block adhesion completely (Thomas et al, 2001b). Treating the cells solely with LAP, or antibodies against αvβ6 or α5β1 did not inhibit adhesion significantly (neither did a combination of LAP and anti-αvβ6). However, combinations of anti-α5β1 and LAP or anti-αvβ6 and anti-α5β1 inhibited adhesion completely. These data confirm that soluble LAP produces a similar effect to anti-αvβ6 antibody and suggests that LAP may alter keratinocyte binding to fibronectin through its interaction with the αvβ6 integrin. (B) LAP inhibits VB6 migration to fibronectin. To determine the effect of LAP on cell migration, the cells were treated with varying concentrations of LAP peptide prior to plating and then allowed to migrate for 8 h before counting. A representative experiment performed in quadruplicate is shown. Results are expressed relative to migration in the presence of a control antibody (=100). Error bars represent standard deviation. Haptotactic migration of VB6 cells towards fibronectin is modulated through αvβ6 and α5β1 integrins and maximal inhibition of migration is produced by blocking both integrins (Thomas et al, 2001b). Migration of VB6 cells towards fibronectin was inhibited by a concentration of LAP as low as 0.1 μg ml⁻¹ (51% inhibition). At higher concentrations a further degree of inhibition was seen (0.25 μg ml⁻¹=63% inhibition). This level of inhibition was similar to that produced by anti-αvβ6 antibody (54% inhibition) and did not differ significantly from a combination of LAP and anti-αvβ6 antibody (52% inhibition). Complete inhibition of migration was produced using a combination of soluble LAP with anti-α5β1 antibody (or a combination of anti-αvβ6 with anti-α5β1 antibody). Soluble LAP had no effect of the migration of H357 αv-negative cells. These data confirm that soluble LAP produces a similar effect to anti-αvβ6 antibody and suggest that LAP may inhibit αvβ6-dependent keratinocyte migration towards fibronectin. (C) LAP inhibits αvβ6-dependent invasion through Matrigel. Cell invasion assays were performed over 72 h using Matrigel coated polycarbonate filters. To assess the effect of soluble LAP invasion, cells were treated with LAP (0.5 μg ml⁻¹) for 30 min at 4°C prior to plating. Following incubation the cells in the lower chamber (including those attached to the undersurface of the membrane) were trypsinised and counted on a Casy 1 counter (Sharfe System GmbH, Germany). Figure shows a representative experiment performed in quadruplicate. Error bars represent standard deviation. Pre-incubation of VB6 cells with soluble LAP inhibited invasion (62% inhibition), producing a similar level of inhibition to anti-αvβ6 antibody (58% inhibition). Soluble LAP had no effect on invasion of H357 αv-negative cells or C1 control cells which express low levels of endogenous αvβ6. These data suggest that soluble LAP may inhibit αvβ6-dependent cell invasion.

VB6 cells show increased αvβ6-dependent invasion through Matrigel relative to C1 control cells (Thomas et al, 2001b,c). To determine whether LAP inhibits αvβ6-dependent invasion, VB6 cells were incubated with LAP (0.5 μg ml⁻¹) or antibodies against the αv subunit, αvβ6 or α5β1. Figure 5C confirms that prior incubation of VB6 cells with LAP inhibits invasion by VB6 cells (62% inhibition; P=<0.001), producing a similar level of inhibition to that obtained with anti-αvβ6 antibody (58% inhibition). Invasion of H357 cells (which invade at a similar level to VB6 cells through non-αvβ6 mechanism) or C1 cells was not affected by LAP. In addition, to confirm that cell invasion was not affected by endogenously secreted TGF-β1 or by TGF-β1 present within serum, we repeated the assays incorporating a TGF-β1 blocking antibody or TGF-β1 soluble receptor. Neither of these TGF-β inhibitors produced an effect on invasion by VB6 cells (data not shown). These data confirm that soluble LAP may specifically inhibit αvβ6-dependent cell invasion.