Introduction

It is known that HGF/SF-induced spreading requires changes in the actin cytoskeleton (Ridley et ah, 1995, Dowrick et ah, 1991). In addition, intercellular junctions have to be disrupted to allow cell scattering. However, how HGF/SF disrupts intercellular junctions to enable cells to detach from each other has not been investigated yet. In epithelial cells, the lateral plasma membranes of adjacent cells interact via adherens, tight junctions and desmosomes (Ben-Ze'ev, 1997). Previous reports studying disruption of intercellular junctions have concentrated on v-src expression in MDCK cells. Forced expression of low levels o f v-src in polarized M DCK epithelial cells results in the disruption of adherens junctions, but tight junctions and desmosomes are retained (Warren and Nelson, 1987). In MDCK cells as well as chicken embryo and rat fibroblasts, it has been reported that disruption o f adherens junctions correlates with increased tyrosine phosphorylation of adherens junction proteins, principally p-catenin (Matsuyoshi et al., 1992, Behrens et al., 1993, H am aguchi et al., 1993). In contrast, no change in the expression or phophorylation state of E-cadherin, a major cell-cell adhesion molecule in epithelial cells was observed in MDCK cells following HGF/SF addition (Weidner et al., 1990).

Therefore, to study how HGF/SF mediates the disruption of intercellular junctions, the HGF/SF effect on localization of intercellular junctions was mainly studied in this first Chapter. In addition, further experim ents tried to evaluate the role of tyrosine phosphorylation of adherens and tight junction proteins in cell scattering.

3.2 Results

3.2.1 Characterization of the effects of recombinant HGF/SF on MDCK c e lls

3.2.1.1 Titration of the concentration of the recombinant HGF/SF

In the preceding study recombinant human HGF/SF was obtained from the supernatant of sf9 cells infected with a baculovirus carrying the human HGF/SF cDNA (a gift from G. Gaudino, University of Turin). The supply of HGF/SF was therefore limited. However,

when this study was started the first recombinant HGF/SF was developed by R&D Systems. The DNA sequence encoding 728 amino acid residue variant of the human pro- HGF/SF (Nakamura et ah, 1989) had been expressed in sf21 insect cells using a baculovirus expression system by RD Systems. They measured the biological activity of the recombinant human HGF/SF by its ability to stimulate 3H-thymidine incorporation in the HGF responsive monkey epithelial cell line, 4MBr-5 (Rubin et al., 1991).

The insect cell derived recombinant HGF preparation is a mixture of predominantly single HGF, as well as some heterodimeric HGF. Previous studies have shown that single chain HGF and heterodimeric HGF are equally active in in vitro bioassays due to either the production of the protease by the cell culture or the presence o f a apparently ubiquitous protease in serum. Indeed, there has been a serine protease in serum identified which processes single chain HGF/SF into its active heterodimeric form (Miyazawa et al., 1993).

Therefore, it was necessary to test if this recom binant HGF/SF could induce cell scattering on the M DCK cell strain used in the preceding study (Ridley et al., 1995). First, the optimal concentration for the HGF/SF response (spreading and scattering of M DCK cells) was determined. MDCK cells were seeded at 1 x lO^cells per well (diameter, 1.8cm) on glass coverslips. Two days after seeding, cells were transferred from 10% FCS to 0.5% FCS for 24 h before addition of different concentrations o f the growth factor into the medium on day 3 after seeding. On day 3, before growth factor addition, M DCK cells formed colonies o f variabale size o f around 10-50 cells. Cells within the colony were in close contact with each other and displayed discrete edges with very few occasional membrane ruffles. As it was noticed that cells responded better to growth factor stimulation in low percent serum, HG F/SF stim ulation was always performed in 0.5% FCS in subconfluent M DCK cells. A fter overnight incubation with HGF/SF (16 h), the number of colonies, in which most o f the cells had detached from each other (scattered), were counted using phase contrast light microscopy.

The optimum concentration where most colonies (50%-100%) were scattered after 16 h was found to be 10 ng/ml of HGF/SF (Table 3.1). Consequently, this concentration was used in all subsequent experiments. At concentrations below 5 ng/ml cells did not respond and above 10 ng/ml the response got weaker. This might be due to the fact that at concentrations above 10 ng/ml, the receptors were saturated and at higher concentrations were down-regulated. Down-regulation o f the HGF receptor has been previously reported in acute lung injury 12 h after secretion o f HGF/SF (Yanagita et al., 1993). Alternatively, as with higher concentrations the dilution factor o f the HGF/SF stock was smaller, toxic substances contained in the sample could have had an effect.

In several experiments it was observed that not all cells responded equally to HGF/SF. There probably exist more potent forms of the growth factor in vivo as its recombinant version might not be folded and glycosylated in the same way as in vivo secreted HGF/SF. Furtherm ore, the E D50 to induce 3H -thym idine incorporation 4M Br-5

epithelial cells was found to be 20-40 ng/ml by RD Systems. In contrast, in this work, it was found that concentrations of 10 ng/ml were optimal for M DCK cell scattering overnight. Table 3.1 Titration of HGF/SF cone (ng/ml) 0 1 2 5 1 0 15 2 0 30 scattered cells - - - 4-4- ₄-₄-₄- ₄-₄- 4- - 0-10% of cells scattered +: 10-30% ++: 30-50% +4-4-: 50-100%

3.2.1.2 Timecourse of the HGF/SF response

In order to monitor the timecourse of cell scattering in response to the insect-cell derived recombinant HGF/SF, time lapse video microscopy was used. As previously described, it was observed that the motile response to HGF/SF could be divided in two stages (Stoker and Perryman, 1985, Ridley et ah, 1995): cell colonies spread centrifugally during the first 4 h to 6 h after addition of HGF/SF followed by the breakdown of cell­ cell contacts and scattering of the colony at 8 h to 24 h (Table 3.2). If cells were further stimulated through addition of new HGF/SF (HGF/SF seemed to be degraded after 24 h) cells continued to scatter until all colonies were scattered and were even more elongated and fibroblastoid in their cell shape (progressive scattering).

Table 3.2 Timecourse of HGF/SF-induced scattering

time (h) 0 2 4 6 8 12 16 24

Phase contrast pictures in Fig. 3.1 A illustrate the two stages o f the HGF response: unstimulated cells grew as a compact colony when grown subconfluent (a) and then spread following addition of HGF/SF for 4 h (b) and had detached from each other after 16 h (c).

3.2.1.3 Changes in the actin cytoskeleton induced by HGF/SF

HGF/SF has been reported to induce cell migration and changes in the actin cytoskeleton of MDCK cells which have been previously studied in detail by (Ridley et al., 1995) and are shown in Fig. 3. IB. Unstimulated MDCK cells had strong actin cables at the edge of colonies and contained stress fibers (Fig. 3. IB a). After 4 h of HGF/SF-stimulation (Fig. 3 .IB b), these actin structures were disassembled and replaced by lamellipodia and membrane ruffles (Fig. 3 .IB, b, c and d). At later time points (8 h-16 h) cells also started to detach from each other (Fig. 3 .IB, c and d).

3.2.2 HGF/SF-induced changes in E-cadherin and p-catenin localization

E-cadherin is a transmem brane protein which defines the adherens junction through homophilic interaction of its extracellular domains while the intracellular domains directly interact with p-catenin, (Ben-Ze'ev, 1997). In addition to changes in actin organization, the migratory response of MDCK cells to HGF/SF also requires the disruption of cell-cell junctions. Cell-cell contact in MDCK cells is maintained via tight junctions, desmosomes and adherens junctions. Therefore stimulated and unstimulated cells were analysed for the integrity of their adherens junctions using immunofluorescence techniques where the cells were fixed and stained for E-cadherin or p-catenin. Fig. 3.2 shows that p-catenin and E- cadherin localized to intercellular junctions in unstimulated MDCK cells (Fig. 3.2, a and b). W hen serial sections were taken, it was realized that E-cadherin and p-catenin localized to sites in the basolateral membrane which were more apical located in comparison to the stress fibres in actin filament stainings (Fig. 3.1.B).

Fig. 3.1