Binding of parbendazole to tubulin and its influence on microtubules in tissue-culture cells as revealed by immunofluorescence microscopy

An important range of compounds based on the benzimidazole nucleus have become useful fungicides and anthelminthics. Benomyl, thiabendazole and methyl benzimidazol-2-yl carbamate are important fungicides (Davidse, 1975), whilst mebendazole, cambendazole and albendazole are used as anthelminthics (Borgers, De Nollin, Verheyen, de Brabander & Thienpont, 1975). A similar compound, nocodazole has been shown to have anti-tumour activity (de Brabander, Van De Veire, Aerts, Guens, Borgers, Desplenter & De Creé, 1975). There is increasing evidence to show that this range of compounds have a similar mechanism of action, which is primarily mediated by an inhibition of microtubule-based functions. Davidse (1973) was able to show that benomyl blocked mitosis in the fungus Aspergillus nidulans. Mutants that were resistant to benomyl have been found to have lower binding affinities than either the wild type or a super-sensitive mutant (Davidse & Flach, 1977). Other workers have shown that benomyl-resistant mutants of A. nidulans possess altered tubulins as determined by two-dimensional gel electrophoresis (Morris, 1980).

Direct evidence of the interaction of one member of this class of compounds with tubulin has come from a study of nocodazole. This drug has been shown to bind to tubulin and to inhibit the binding of colchicine (Hoebeke, Van Nijen & de Brabander, 1976). A recent report from this laboratory (Ireland, Gull, Gutteridge & Pogson, 1979) has shown that the assembly of mammalian brain microtubules in vitro is inhibited by a wide range of these benzimidazole carbamates. Nocodazole was found to be a potent inhibitor of microtubule assembly in vitro although the anthelminthic compound parbendazole was even more effective. Colchicine is a much used inhibitor in studies of microtubule-mediated processes. There is increasing evidence that the benzimidazole carbamates may be equally useful in these studies and may even have advantages over colchicine where the drug action is to be reversed or in systems that are peculiarly insensitive to colchicine (for example, many of the lower eukaryotes; Mir & Wright, 1978; Welker & Williams, 1980; Quinlan, Roobol, Pogson & Gull, 1980). We have, therefore, studied further the influence of parbendazole on both tubulin and microtubules to assess the nature and extent of its inhibitory activity.

Parbendazole was radiolabelled in solution using tritium gas, at the position indicated by an asterisk in Fig. 1, by the Radiochemical Centre, Amersham, England. It was purified to 99·05% using thin-layer chromatography to give a specific activity of 167·4 μCi/mg.

Pure tubulin was obtained from sheep brain by 2 cycles of assembly and disassembly in vitro (Dentler, Granett & Rosenbaum, 1975) followed by phosphocellulose chromatography to remove any associated proteins. Immediately prior to use the protein was centrifuged at 130000 g for 30 min to remove any aggregates. It was used at a protein concentration of 0·2 mg/ml in 0·025 M Pipes buffer (piperazine N-N′-bis(2-ethane sulphonic acid), 0·5 mM EGTA (ethylene glycol-bis(β-aminoethyl etherj-N′.N’-tetraacetic acid), 0·25 mM Mg₂SO₄, 0·1 mM GTP. Drug binding was determined by equilibrium dialysis using concentrations of parbendazole between 0·1 μM and 4 μM, and 2% (v/v) DMF (dimethyl formamide) as a carrier. Equilibrium was achieved by constant stirring for 2 h at 26 °C, bovine serum albumin being used as a standard. 200 μl aliquots were counted in PCS (Hopkin and Williams) in a Beckman 25-200B liquid scintillation counter.

Vero cells, an established cell line derived from African Green Monkey kidney (a gift from Dr F. Watt, University of Oxford, U.K.) were seeded in DMEM (Dulbecco’s modification of Eagles’ medium; Flow Laboratories, Irvine, U.K.) supplemented with 10% (v/v) foetal calf serum (Gibco Biocult, Paisley, U.K.) onto glass coverslips in multiwell dishes (Sterilin). They were allowed to settle, and spread for 2·5 h in a humid atmosphere at 37 °C. After this time the medium was changed to DMEM containing 2, to or 20 μM parbendazole and 1% (v/v) DMSO (dimethyl sulphoxide). Controls contained 1% (v/v) DMSO or had no additions.

Indirect immunofluorescence was performed according to Osborn & Weber (1977) and using an antibody raised against purified tubulin. The second antibody was a fluorescein conjugate of an antibody, raised in a goat, against rabbit IgGs (Miles Laboratories Ltd, Slough, Bucks, U.K.).

The antitubulin antibody was raised in rabbits by 3 fortnightly injections of 2 mg of antigen in Freund’s adjuvant followed by monthly boosters of 1 mg without adjuvant. The antigen was purified by 2 cycles of assembly and disassembly (Dentier et al. 1975) followed by 2 cycles of assembly in the presence of 10% (v/v) DMSO (Himes, Burton & Gaito, 1977) to reduce the microtubule-associated protein content. The tubulin was then assembled at 37 °C and crosslinked with 1% glutaraldehyde (Fuller, Brinkley & Boughter, 1975) and stored in liquid nitrogen.

Pure tubulin was prepared by 2 cycles of assembly and disassembly followed by phosphocellulose and DEAE-Sephadex chromatography. This tubulin was coupled to Sepharose 4B (Pharmacia) by a standard method (March, Parikh & Cuatrecasas, 1974; Fuller et al. 1975). Monospecific antibody was isolated by passage of whole serum through the immunosorbent column. The bound fraction was eluted with glycine/HCl buffer, pH 2·7 (Fuller et al. 1975) and concentrated by precipitation with ammonium sulphate.

Coverslips were viewed with a Zeiss microscope equipped with epifluorescent optics. Photographs were taken with an oil-immersion objective (63 ×) on Ilford HP5 film.

Increasing concentrations of parbendazole (Fig. 1) progressively inhibited the assembly of microtubules in vitro. Inhibition of 50% of the assembly of brain microtubules was achieved with 3 μM, whilst virtually all assembly was inhibited by 10 μM parbendazole. The kinetics of inhibition were identical to those achieved with colchicine (Ireland et al. 1979).

The possible binding of parbendazole to tubulin was studied using highly purified mammalian brain tubulin produced by an assembly-disassembly protocol followed by phosphocellulose column chromatography. The maximum concentration of parbendazole used in these binding studies was restricted to 4 μM. Essentially, this was because at concentrations greater than this parbendazole tended to precipitate onto the dialysis membranes or other supports used in binding assays. Increasing concentrations of radiolabelled parbendazole were incubated with known amounts of the purified tubulin in a standard equilibrium dialysis assay. Bovine serum albumin was used as a control. Results from this study indicate a specific binding of parbendazole to tubulin and are shown in Fig. 2 as a Lineweaver-Burk plot. Computer analysis of these results gave a maximum binding value of 0·72 mole parbendazole per mole tubulin. This result was in close agreement with the value calculated from Eadie-Hofstee and direct plots.

Untreated Vero cells possess a normal distribution of microtubules when stained with antitubulin antibody (Weber, Pollack & Bibring, 1975). In particular, most microtubules appeared to arise from a discrete microtubule-organizing centre (MTOC) close to the nuclear envelope (Figs. 3 and 4). Microtubules which radiated from this structure transversed the cytoplasm and usually terminated in the cell periphery close to the plasma membrane (Fig. 3).

Within 30 min of treatment with parbendazole microtubules have withdrawn from the peripheral area of the cell (Fig. 5). At all drug concentrations studied this early (30-min) disassembly of the microtubule in the cell periphery was associated with blebbing of the plasma membrane and an increased level of general diffuse fluorescence in the cell (Figs. 5, 6). Incubation with parbendazole for 3 to 4 h caused complete rounding up of the cells and a further increase in cytoplasmic fluorescence. This meant that no details were discernible within the cell, not even the nucleus. For this reason we have not included photographs of these cells. However, after 20 h of treatment (Figs. 7−10) the cells had flattened once more, assumed a typical fibroblastic shape and did not show plasma membrane blebbing. Cells treated with between 2 μM and 20 μM parbendazole for this length of time or longer (up to 45 h) showed an almost complete absence of microtubules (Figs. 11, 12). The 1 or 2 fluorescent structures which did occur were always associated with the centrioles of the cell, which were clearly seen as 2 small structures close to the nucleus (Figs. 7-10). Usually these structures were associated with the end of one of the centrioles. Without examination by electron microscopy it is impossible to tell if these are single microtubules, primary cilia or remnants of the latter.

Parbendazole is an extremely potent inhibitor of microtubule assembly in vitro and our results indicate that this inhibition is mediated by binding of this drug to tubulin. Binding studies with pure tubulin indicate that the number of binding sites per tubulin molecule (110000 molecular weight dimer) is 0·7, suggesting that parbendazole binds to the tubulin dimer with a mol : mol stoichiometry. Colchine binding to tubulin is also generally accepted to be 1 mole per mole of tubulin dimer (Wilson et al. 1974), although actual experimental values obtained range between 0·6 and 1·07 (Wilson et al. 1974; Wilson & Meza, 1973 ; Bryan, 1972). The specific binding of parbendazole to tubulin, taken with similar earlier reports of nocodazole binding, suggest that the benzimidazole carbamates as a group possess tubulin binding properties.

Parbendazole effectively depolymerizes microtubules in the Vero cells used in this project. Using concentrations between 2 μM and 20 μM we were able to detect depolymerization of microtubules within 30 min. In most cell lines studied the cytoplasmic microtubules radiate from discrete MTOCs - the major cytoplasmic MTOC being the pericentriolar material (Gould & Borisy, 1977). During parbendazole treatment the microtubules disappeared initially from peripheral regions of the cell. Complementary experiments using cells that are reassembling microtubules after drug treatment or cold treatment have shown that the direction of growth of these microtubules is from the MTOC to the cell periphery (Osborn & Weber, 1976). These findings are understandable in the light of recent evidence that cytoplasmic microtubules do possess an intrinsic polarity. Bergen, Kuriyamo & Borisy (1980) have shown that microtubules assembled onto isolated centrosomes elongate at a rate corresponding to the addition of subunits to only one end. These authors also conclude that the subunits are added to the end of the microtubule distal to the MTOC. Similar reasoning that microtubule depolymerization occurs only at this distal end is supported by the assessment of various drug treatments on intact cells (Weber & Osborn, 1979).

Treatment of Vero cells with parbendazole for 3 h or more leads to an almost complete depolymerization of cytoplasmic microtubules. Immunofluorescence of these cells showed that the 2 centrioles are still usually situated close to the nucleus. One or two short microtubules radiate from one of the centrioles. Most of the cyto-plasmic microtubules of mammalian cells are nucleated onto the pericentriolar material, although the centrioles themselves have been shown to be capable of nucleating small numbers of microtubules (Gould & Borisy, 1977). The one or two microtubules which radiate from the centriolar region in parbendazole-treated cells usually arise from only one of the two centrioles. In some cell types in culture primary cilia are a common feature (Osborn & Weber, 1976; Albrecht-Buehler & Bushnell, 1980), and in these cell lines it is usually only one of the two centrioles which acts as a basal body during ciliogenesis. The structures which we see after parbendazole treatment of cells could be individual microtubules, primary cilia or remnants of the latter. Whichever is the case they appear to be less sensitive to drug-induced depolymerization than the majority of cytoplasmic microtubules. The fact that the 2 centrioles are usually found close to the nuclear envelope in the parbendazole-treated cells indicates that they are able to maintain their position in the absence of the majority of cytoplasmic microtubules.

During parbendazole treatment the cells round up and respread, even though no microtubules are present. This implies that microtubules are not necessary for cell spreading and that this is the function of another cytoskeletal system. Microfilaments are the most likely candidates as their distribution has been shown to remain unchanged during colchicine treatment of spread cells and colchicine-treated cells in suspension are capable of spreading onto a surface (Goldman, 1971).

The specific binding of parbendazole to tubulin and its effective action on microtubules in vitro and in vivo suggest that it and other benzimidazole carbamates will be useful drugs in studies of microtubule-mediated functions. In many situations, use of parbendazole or another benzimidazole carbamate as well as colchicine should provide stronger evidence for microtubule involvement. However, in some instances, the benzimidazole carbamates have distinct advantages over colchicine. Lower eukaryotes are peculiarly insensitive to colchicine whilst their microtubule systems appear particularly sensitive to parbendazole and certain other benzimidazole carbamates (Mir & Wright, 1978; Welker & Williams, 1980; Davidse, 1973). Also, colchicine binding to tubulin is not easily reversed, whilst that of the benzimidazole carbamates is rather easily reversible. These compounds therefore, appear likely to be useful in synchronizing mammalian cell cultures (Zieve, Turnbull, Mullins & McIntosh, 1980).

We thank Dr J. Thorpe for purification of the radiolabelled parbendazole. J.C.H. and R.A.Q. are supported by SRC studentships. This work was funded by grants from the Wellcome Trust and the Royal Society.