Cytoskeletons from a mutant of Dictyostelium discoideum with flattened cells*

Amoebae of Dictyostelium discoideum are highly motile cells that, like granulocytes, rapidly change their shape and directionality in response to external factors. These cells lack structures like stress fibres that stabilize the organization of the cytoskeleton. For the preparation of their delicate cytoskeletons, aggregating cells of D. discoideum are not optimally suited, since they do not spread as extensively and do not adhere to a substratum as tightly as fibroblasts.

In this paper we describe a mutant of D. discoideum, HG403, whose cells spread strongly on a surface. The flattened cells of the mutant are able to aggregate and to form fruiting bodies, although the cells are less strongly polarized during aggregation and, consequently, streams are less well developed than in the wild type. The mutant cells are capable of responding chemotactically, and within the streams they adhere tightly to each other. HG403 cells give rise to cytoskeleton preparations in which the microtubular apparatus and the peripheral microfilament layer are well preserved. The mutant is thus suited to study relationships of structure and function in chemotactically responding amoeboid cells.

Mutant HG403 was selected as a strain with slightly atypical aggregation patterns after ultraviolet irradiation of HG313. This mutant was obtained by nitrosoguanidine treatment of the AX2 strain, and was selected on basis of the resistance of its cell aggregation to 3′,5′-cyclic-adenosine phosphorothioate (cAMPS). This phosphodiesterase-resistant cyclic AMP analogue inhibits cell aggregation in wild-type NC4 and in its axenically growing derivative, AX2. In the absence of cAMPS, HG313 aggregates normally. Mutant 403 carries the cAMPS-resistance of its parent, but in the context of the present paper this mutation is of no interest.

For studying development (Fig. 1), the strains were cultivated with Escherichia coli B/2 on nutrient agar plates containing 0·1 % glucose, 0·1 % bacteriological peptone (Oxoid) and 2% agar in 17 mM Soerensen phosphate buffer, pH 6·0. For electron microscopy, AX2 or HG403 cells were grown in suspension at 23 °C either with 1×10¹⁰ cells per ml of E. coli B/r in the phosphate buffer, pH 6’0, or axenically as described previously (Malchow et al. 1972). The cells were harvested during the exponential growth phase, washed and shaken for various times in the phosphate buffer as indicated in the figure legends.

For chemotaxis experiments, HG403 or wild-type V12/M2 cells were starved in phosphate buffer, transferred onto a glass coverslip and stimulated with micropipettes filled with 1 × 10⁻³ M-cyclic AMP essentially as described (Gerisch et al. 1975a; Gerisch & Keller, 1981).

For scanning electron microscopy, cells were fixed on glass coverslips in a mixture of 1-2-5 % glutaraldehyde and ·01–0·02% OsO₄ in the pH 6·0 phosphate buffer for 15 min on ice in the dark and postfixed in 2 % glutaraldehyde on ice. After dehydration through a graded series of ethanol the specimens were dried by the critical point method with CO₂, sputter coated with gold and viewed in a Jeol scanning electron microscope JSM 35 C at 25 kV.

Cells attached to glow-discharged, carbon-coated Formvar films on gold grids were extracted and fixed in mixtures of Triton X-100 and glutaraldehyde (Hôglund et al. 1980). For Figs 5 to 9A-C the mixture consisted of 1 % Triton X-100 and 0·05 % glutaraldehyde in PEM buffer, pH 6·5, for Fig. 9D the concentrations of both Triton and glutaraldehyde were 0·5%. The PEM buffer was similar to a buffer used by Small & Celis (1978) and Galvin et al. (1984) for cytoskeleton preparations; it consisted of 20mM-KCl, 2–4mM-EGTA, 2mM-MgSO₄ and 10 mM-PIPES, pH6·5. The grids carrying the cells were incubated for 10–20 mm in the extraction-fixation solution, slowly moving them with tweezers for the first minute. The cytoskeletons were postfixed in 0·5–2% glutaraldehyde in PEM buffer and washed in the buffer.

Cytoskeletons postfixed for 15 min with 0·5 % glutaraldehyde were treated with 2 mg per ml of borotetrahydride according to Weber et al. (1978). Subsequently they were incubated in 1 % non-immune goat serum plus 0·1 % bovine serum albumin in PEM buffer, pH 8·2, and labelled with rat monoclonal anti-yeast tubulin antibody YL 1/2 (Kilmartin et al. 1982), coupled directly as described by de Mey (1983) to 5 nm colloidal gold prepared according to the phosphor-gold method described by Slot & Geuze (1981). After washing three times for 20min in PEM buffer, pH 8·2, containing 0·1 % bovine serum albumin, and rinsing in bidistilled water, the specimens were negatively stained in the cold with 2% ammonium phosphotungstate, pH 7·6. The nonimmunolabelled preparations shown in Fig. 9 were negatively stained with 1–2% sodium silico-tungstate at room temperature, essentially as described for tissue culture cells by Small & Langanger (1981), using Bacitracin as wetting agent before staining.

Grids carrying the cytoskeletons were rinsed in bidistilled water and briefly in 10% methanol. No. 5 tweezers (Dumont) with a clamped grid were plunged 10 cm deep into liquid propane by free fall from a height of half a metre. The cytoskeletons were freeze-dried in a Balzers BAF 301 apparatus and rotary shadowed at an angle of 25° with platinum/carbon, and an angle of 90° with carbon using electron beam guns. Micrographs were taken in a Jeol JEM-100 CX electron microscope at 80 kV.

Cells of mutant HG403 formed streams during aggregation, but these streams were usually less well developed than those shown in Fig. 1A, and even the ones shown were less regularly branched than those of AX2 or other wild-type strains (Fig. 1C). Under the conditions used, fruiting bodies were smaller than in AX2 (Fig. 1B,D), but they were well proportioned into stalk and spore head.

The altered aggregation pattern was associated with an atypical shape of the cells. The aggregating cells of HG403 tended to flatten by spreading extensively on a substratum. They were less elongated than the almost cylindrical cells in wild-type aggregates (Fig. 2). Moreover, the mutant cells often extended pseudopodia in many directions, so that their polarity remained unclear. They adhered tightly to each other during aggregation, and often did so over extended areas of their surface, rather than in a clearly polar fashion comparable to the end-to-end adhesion of aggregating wild-type cells. Often surface extensions of two adjacent cells were abundantly indented at the areas of contact (Fig. 3).

When stimulated through a micropipette filled with 1×10⁻³ M-cyclic AMP, mutant cells accumulated at the tip of the pipette, although they needed a longer time than wild-type cells. The responding cells assembled into streams in HG403 as in the wild type, but the pattern of streams was less well developed in the mutant.

Fig. 4 shows the chemotactic response of a single HG403 cell (A—I) in comparison to a typical response of a wild-type cell (K-O). When exposed to the steep gradient of cyclic AMP built up around the tip of the micropipette, wild-type cells often respond by changing their polarity (Gerisch et al. 19756; Swanson & Taylor, 1982), extending a new front to the source of cyclic AMP and retracting their previous front. HG403 cells responded similarly, but their ability to stabilize directionality, polarity and shape of the cells seemed to be reduced. The mutant cells tended to protrude many pseudopodia simultaneously, or became extremely stretched in the direction of the gradient (Fig. 4F). All this behaviour is different only in degree from that of wild-type cells. Extension of multiple pseudopodia is also observed in wild-type cells; there it occurs transiently at the beginning of reorientation and is followed by a stationary state in which a single pseudopod maintains dominance.

For studies on the organization of the cytoskeleton, the flattened cells of HG403 provide nearly ideal material since they yield quasi two-dimensional cytoskeleton preparations, yet aggregate and develop in a fashion fairly close to normal.

HG403 cells were allowed to attach to carbon-coated grids and were subsequently treated with Triton X-100 and glutaraldehyde, In the cytoskeletons obtained, the microtubules were well preserved and could be labelled with a gold-conjugated antitubulin antibody (Fig. 5). We have counted between 32 and 48 microtubules originating from single microtubule-organizing centres (MTOC) that were located close to a nucleus. The microtubules were embedded in a loose network of filaments. Numerous remnants of mitochondria were attached to the microtubules or fixed in the network of filaments between them (Fig. 6).

At the periphery of the cells a dense network of microfilaments was preserved in the cytoskeleton preparations. This network had an exceptionally tight texture at the end of a cell, which was most probably its moving front (Fig. 7). The area of this network was free of mitochondria and thus corresponded to the hyaline ectoplasm seen in the light microscope where, as judged by labelling with fluorescent phal-loidin, F-actin is accumulated (Rubino et al. 1984). A similar network was seen in portions of the cells that were in contact with the substratum (Fig. 8). This network resembled the system of actin filaments associated with the substratum-attached portion of the cell membrane, as demonstrated by Clarke et al. (1975) by shearing off the upper portion of the cells.

During the first 20 min of spreading of suspended cells on a surface, the cells protruded spike-like extensions that were attached to the glass along their entire length. The skeleton of these spikes was visualized by treatment with glutar-aldehyde-Triton X-100 and negative staining. It consisted of bundles of filaments with the typical appearance of F-actin (Fig. 9A—C). These microfilament bundles were laterally connected by a dense network of microfilaments. When extraction of the cells was stopped before its completion by decreasing the Triton X-100 and increasing the glutaraldehyde concentration, the microfilament bundles could be seen to extend within the spikes up to their tips (Fig. 9D).

D. discoideum cells are well suited for studying the control of motility in amoeboid cells. In the aggregation-competent stage these cells respond sensitively to cyclic AMP. Their chemotactic response implies functional reorganization of the contractile system within less than 5-7 s, since the cells change their polarity within that time when an external gradient of cyclic AMP is imposed (Gerisch et al. 19756; Swanson & Taylor, 1982). Investigation of the actual changes in the contractile system that are coupled to the activation of cell-surface receptors by the attractant requires a combination of biochemical and other methods. These include structural studies on the cytoskeleton and immunolabelling of proteins that, like actin-binding proteins, regulate the assembly of its components. The structural aspect is of particular importance since, in responding chemotactically, a cell has to regulate its activities in a spatial manner, extending pseudopodia in one direction while contracting at its opposite end. For the investigation of cell structures and their changes during chemotactic orientation, aggregating D. discoideum cells are not optimally designed because of their cylindrical shape and their poor spreading on a substratum.

To study the motility and behaviour of aggregating cells (Gerisch, 1964; Heunert, 1973) and changes in the localization of myosin after stimulation with cyclic AMP (Yumura & Fukui, 1985), cells have been brought into a more two-dimensional condition by pressing them between the surfaces of a glass coverslip and an agar layer. Cells of HG403, the mutant described in this paper, do not need this sandwich technique for producing well-spread, flat, albeit aggregating cells. The cytoskeleton preparations obtained demonstrate the usefulness of the mutant for structural studies. They show the organization of the microtubular apparatus and its relationship to the microfilament system. Furthermore, they show the association of microfilaments into a dense cortical network and into bundles forming the core of spike-like extensions of the cells, comparable to the pseudodigits observed by Eckert et al. (1977) in ultrathin sections.

The applicability of immunolabelling has been shown for tubulin and should be possible also for other proteins that constitute the cytoskeleton or are associated with these proteins, such as those regulating the assembly state of actin. Immunolabelling may also help to identify intermediate filaments in the cytoskeleton preparations. Primarily on the basis of biochemical evidence and the reactivity of anti-vimentin antibodies with two polypeptides of D. discoideum, the presence of intermediate filaments in this organism has been postulated (Koury & Eckert, 1985). It is possible that some or most of the filaments seen in the area between the microtubules are in fact intermediate filaments, but this remains to be proven.