Vanadate inhibits both intercellular adhesion and spreading on fibronectin of BHK21 cells and transformed derivatives

The vanadate ion, a phosphate analogue, inhibits numerous enzymes, including some on the glycolytic pathway (Ramasarma and Crane, 1981; Nechay et al. 1986). A few enzymes, such as flagellar dynein (Gibbons et al. 1978), are sensitive to vanadate at concentrations as low as 10⁻⁷M. In view of the central metabolic role of phosphate, it might be expected that when vanadate is added to the medium surrounding intact animal cells in culture, it would inhibit many cellular activities, including the generation of ATP. Instead, at external concentrations less than 10⁻⁴M it causes a variety of specific regulatory changes in intact cells. For example: it was found to stimulate DNA synthesis in quiescent 3T3 cells (Smith, 1983), to induce various features of the transformed state in NRK-1 rat kidney cells (Klarlund, 1985), interleukin-independent growth of a normally factor-dependent cell line (Tojo et al. 1987), and to induce angiogenesis by bovine capillary endothelial cells (Montesano et al. 1988). Recently, Moria et al. (1989), found that 5×10⁻⁵M vanadate caused reversible arrest of 3T3 cells in G₂, but did not affect S-phase, assembly of the mitotic spindle, chromosome segregation or cytokinesis. It remains unclear whether these selective effects are due to low intracellular concentrations of vanadate itself, to potent inhibition by the expected intracellular reduction product vanadyl (Macara et al. 1980) or even to actions at the cell surface (Nechay et al. 1986).

Perhaps because some tyrosine-specific phosphatases are very sensitive (Lau et al. 1989), vanadate dramatically increases the level of phosphorylation of many proteins at tyrosine residues, in both transformed cells (Brown and Gordon, 1984; Yonemoto et al. 1987) and untransformed cells (Klarlund et al. 1988; Moria et al. 1989; Marchisio et al. 1988). Thus many of the effects of vanadate on intact cells can be interpreted as consequences of such increased phosphorylation, although it may be unwise to ignore other possible sensitive targets.

The aggregation of BHK21 cells in suspension is inhibited by agents that depolymerize microtubules (Waddell et al. 1974). Whilst investigating this, we discovered that both aggregation, and spreading of the same cells on a fibronectin substratum, are inhibited by vanadate. In the latter assay, transformed derivatives of BHK cells are more sensitive, probably because BHK21 cells, once spread, remain so in vanadate, whereas spread transformed cells are susceptible to rounding by very low concentrations.

BHK21 clone 13 cells and derived lines were grown and resuspended as described by Edwards et al. (1987). Polyoma-transformed cells, originally derived from PyY were the line recloned by us as previously described (Edwards et al. 1979). ASV-transformed cells were a line transformed by the Schmidt-Ruppin strain of avian sarcoma virus (Macpherson, 1965).

The balanced saline Hanks’–Hepes (HH) was similar to Hanks’, but buffered with 10 mM Hepes at pH7.4 (Edwards et al. 1987). Hepes–Saline (HS) denotes the same medium minus divalent cations.

Bovine plasma fibronectin was purified from calf serum by affinity chromatography on gelatin-Sepharose (Engvall and Ruoslahti, 1977), and used to coat glass by dilution (to 25 μg ml⁻¹) directly into HH from solution (1 mg ml⁻¹) in 8M urea.

Aggregation was followed using a Coulter counter (model Z_B) with a 200 μm aperture, to follow the reduction in total particle count from the initial value N₀ to the value at time t, N_t, as described by Edwards and Campbell (1971) and discussed in detail by Edwards (1973). A 4.0 ml cell suspension (10⁶ cells per ml) was incubated in 10-ml silicone-treated conical flasks, at 37 °C in a reciprocating shaker (stroke 4 cm; 92 min⁻¹). For counting, 0.1-ml samples were diluted in 20 ml ice-cold 0.9% sodium chloride.

Area measurements were made on cells fixed with formaldehyde and stained with 0.1% Kenacid Blue, using a BEEB Video Digitizer interface and BBC microcomputer, as described by Edwards et al. (1988). Rounding of cells was scored unfixed, as follows: 5×10⁴ cells in 1ml HH were added to each 13 mm, fibronectin-coated coverslip. Cells were incubated at 37 °C, and scored at hourly intervals. A series of adjacent fields was examined across the maximum diameter of the coverslip, to monitor loss of cells and measure the proportion of rounded and spread cells. Cells were scored as rounded when seen solely as bright spheres in phase contrast (×25 objective).

Vanadate: ammonium metavanadate was added to cell suspensions from a 4mM stock in HS, stored at pH 8.0; erythro-9-(2- hydroxy-3-nonyl)-adenine (EHNA, Sigma Chemical Co., Poole, Dorset) was dissolved in HH at 2.67 mw, and the highest working concentration of 2mM achieved by adding to 3.0ml of this, 4.0×10⁶ cells in 1.0ml HH. Ouabain (Sigma) was used from a 10 mM stock in HH.

Short-term aggregation of BHK21 cells is most sensitive to microtubule and metabolic inhibitors when the cells are suspended from pre-confluent cells. As cultures become denser, aggregation that is resistant to these inhibitors becomes apparent. Comparison of Fig. 1A with B shows that the same applies to inhibition by 10⁻⁴M vanadate.

Although the inhibitors could start to enter the cells during the prior-to-aggregation interval of 5–10 min at 0°C, it is worth noting that inhibition by vanadate (as well as colchicine) was always effective from the start of aggregation. The rapidity of certain effects of vanadate has previously attracted comment (Nechay et al. 1986). Fig. 2 shows that for a series of cultures of different density, sensitivity to the two agents varies in parallel over a large range. Not surprisingly, when the concentration dependence of inhibition by vanadate was investigated with a series of cell suspensions, overall sensitivity varied. It was therefore convenient to plot inhibition relative to 10⁻⁴M (saturating) colchicine (Fig. 3). Inhibition is half-maximal with vanadate added at about 10⁻⁵M. Data from other times gave no suggestion that lower concentrations are slower to take effect. Inhibition of aggregation by vanadate is reversible: cells preincubated in vanadate, then washed free of the inhibitor, showed a second cycle of vanadate-sensitive aggregation. Inhibition by colchicine, by contrast, was not reversed in this way, as expected from the irreversibility of its binding to tubulin (Fig. 4).

In attempts to narrow down the possible sites of action, we tested a number of other inhibitors of candidate vanadate-sensitive processes. The dynein inhibitor EHNA (Bouchard et al. 1981) did inhibit aggregation, although less effectively than vanadate. Its effectiveness may have been limited by entry, since inhibition was not usually noticeable at 15 min, and insolubility prevented us using higher concentrations. Other effects besides dynein inhibition have been reported for EHNA, including cytochalasin-like effects (Schliwa et al. 1984). The latter are unlikely to account for inhibition of aggregation, since cytochalasin itself promotes rather than inhibits aggregation in this system (Edwards et al. 1975). Ouabain (an inhibitor of sodium/potassium-ATPase) had no effect. Major alterations to Na⁺ and K⁺ gradients, using both monensin (ionophore for monovalent cations) and replacement of external Na⁺ by K⁺, caused some inhibition. No further inhibition was obtained by combining these treatments, but in the presence of monensin, vanadate produced further inhibition (Table 1). It is therefore most unlikely that inhibition of Na⁺/K⁺-ATPase accounts for inhibition of aggregation by vanadate. Although inconclusive, these results are consistent with the possibility, see below, that the target is cytoplasmic dynein.

Vanadate is believed to enter erythrocytes through a route shared with phosphate (Cantley et al. 1978). We therefore included in some experiments a 100-fold excess of phosphate over vanadate. In no case did this reduce the inhibitory effects of vanadate (Table 1). If, as seems likely, vanadate must enter cells to produce its effects, the route of entry remains unclear.

The mean projected areas attained by both BHK21 cells and transformed derivatives spreading on fibronectin are decreased by vanadate. It was noticeable that if this assay was scored at 180 min rather than 30, the spreading of transformed cells appeared to be sensitive to lower concentrations (Fig. 5). However, time courses have shown that this is not in fact a greater sensitivity of initial spreading. At 10⁻⁶M and below, both cell types spread essentially normally, but the mean area of transformed cells then slowly decreased (Fig. 6). Untransformed cells do not behave in this way. This was clearly seen in experiments testing reversibility of vanadate inhibition. Although both cell types recovered almost completely from vanadate inhibition during 3h following its removal, there was a very striking difference in the ability of vanadate to cause rounding of pre-spread cells (Figs 7 and 8). Untransformed cells resisted this even at high concentrations such as 10⁻⁴M, whereas polyoma-transformed cells were rounded by vanadate at 10⁻⁶M. Similar rounding was observed with cells transformed by ASV (not shown). The rounding data of Fig. 8 were obtained by scoring live cells by phase-contrast microscopy for the percentage of brightly refractile rounded cells, since at these long times, many of the cells were too loosely attached to remain in place throughout fixing and staining for area measurements. To measure the concentration dependence of this reversal of spreading by vanadate, cultures fixed and stained at 90 min were used, at which time measurable reductions in area were found, but relatively few cells were completely rounded and therefore liable to be lost. Fig. 9A shows two of the area distributions from which the concentration dependence in Fig. 9B was derived. 10⁻⁶M vanadate was clearly effective, results at 10⁻⁴ M were very variable between experiments.

How does vanadate reversibly inhibit cell-cell and cell-substratum adhesion, at concentrations as low as 10⁻⁶M (sometimes 10⁻⁷M) in short-term assays? Inhibition is most unlikely to be due to non-specific inhibitory effects on metabolism, such as depletion of ATP. Consistent with the specific G₂ blockade reported by Moria et al. (1989), our own unpublished work has shown a complete lack of inhibition by vanadate in various assays of metabolic function measured over 3h. Thus even 10⁻⁴M vanadate did not inhibit incorporation of [³⁵S]methionine into protein, [³H]thymidine into DNA, or secretion into the medium of pre-labelled proteins.

The parallel between colchicine and vanadate, and inhibition by EHNA, strongly suggest that vanadate inhibits aggregation by inhibiting a microtubule-associated ATPase such as cytoplasmic dynein. Inhibition of such an ATPase could also account for the retraction of intermediate filaments in BHK21 cells exposed to 5×10⁻⁷M vanadate (Wang and Choppin, 1981). How microtubule function could make cells adhesive remains a problem. In the aggregation assay, microtubules are required continuously, since incubation of cells at 37 °C prior to aggregation fails to reduce sensitivity to colchicine or vanadate (data not shown). Their role is therefore unlikely to be restoration of trypsin-cleaved adhesion molecules. It could be the maintenance of functional polarity in the cells (Bershadsky and Vasiliev, 1989) dependent on both microtubules and associated ATPases (reviewed by Kelly, 1990). A zone opposite the microtubule-organizing centre and Golgi, topologically equivalent to the leading lamella of a locomoting fibroblast (perhaps a site of new membrane insertion) could behave as a sticky patch. A similar phenomenon could underly the sensitivity to nocodazole of aggregation of mouse L-M cells transfected with a hybrid L-CAM/N-CAM cDNA (Jaffe et al. 1990), and the polarization (Golgi towards target) of killer lymphocytes adhering to their target cells (Geiger et al. 1982).

Given the circumstantial evidence relating tyrosine phosphorylation to diminished adhesiveness and rounding of transformed and mitotic cells (Burridge et al. 1988), the target of vanadate in this case could be a protein tyrosyl phosphatase. Marchisio et al. (1988) found that 10⁻⁶ to 10⁻⁴ M vanadate added to BHK21 cells grown overnight on fibronectin caused a dramatic increase in phosphorylation of many proteins at tyrosine residues. A minority of the cells lost stress fibres, and acquired punctate adhesive structures referred to as podosomes, both features of transformed cells. These changes were interpreted as phenotypic transformation resulting from the increased tyrosine phosphorylation. A similar mechanism could also explain the selective rounding by vanadate of transformed cells, given that both ASV- and polyoma-transformed cells (Yonemoto et al. 1987) may be expected to show a larger increase in tyrosine phosphorylation in response to vanadate than untransformed cells. Thus vanadate-stabilized tyrosine phosphorylation may cause weakening and disassembly of associations between integrins and fibronectin and/or intracellular structures. Grounds for caution are: first, that the rounding induced by vanadate is much slower than by an agent such as EDTA acting directly on integrins (Edwards et al. 1987); and second, that cells can be rounded by oncogenes that do not encode tyrosine kinases (Hunter, 1989). Both suggest that the actual mechanism of rounding may lie several steps downstream from increases in tyrosine phosphorylation, whether induced by oncogenes or by vanadate.

Clearly, the sensitive enzymes responsible for inhibition of cell-cell adhesion and spreading by vanadate cannot yet be identified. However, aspects of the models suggested here are testable, and as the intracellular events influencing these adhesive phenomena become better understood, inhibition by vanadate is likely to prove informative.