Growth Polarity And Cytokinesis In Fission Yeast: The Role Of The Cytoskeleton

The use of simple model systems to study complex problems is a familiar strategy in biological research. Recent developments in molecular genetics have reinforced the value of organisms such as yeasts in the study of fundamental cellular processes. This is particularly true in the case of the cytoskeleton. Although the small size of yeast cells and the presence of a cell wall present obstacles to traditional approaches to the organization of cytoplasmic filament systems such as immunofluorescence microscopy, these have now been largely overcome (Kilmartin & Adams, 1984; Adams & Pringle, 1984). Coupled with the analysis of cloned actin and tubulin genes (Thomas et al. 1984; Yanagida et al. 1985) this opens up a combination of approaches to cytoskeleton structure and function that is possible in few other organisms.

An additional attraction of yeasts as experimental systems is the availability of a large number and variety of mutants that affect the yeast cell division cycle (Pringle & Hartwell, 1981; Nurse, 1981). Thus, the relationship of the cytoskeleton to other cell cycle events is also open to direct study. Two yeasts in particular have been the focus of most attention, the budding yeast Saccharomyces cerevisiae and the fission * yeast Schizosaccharomyces pombe. Of the two, S. pombe may prove to be the better model, if only because its cell division cycle more closely resembles that of higher eukaryotes (Nurse, 1985).

Fission yeast cells grow only at their ends (Johnson, 1965; Streiblova & Wolf, 1972; Mitchison & Nurse, 1985). In newly divided cells growth occurs solely at the old end, that is, the end that existed prior to septation. At a point in the G₂phase of the cell cycle termed NETO (new-end take off) bipolar growth is initiated. This is maintained until the end of G₂, at which point growth ceases and the sequence of mitosis, cytokinesis and cell separation begins. In this paper we describe the relationship of the cytoskeleton to these changes in growth polarity. We also address the role of the cytoskeleton in establishing the division plane in these simple eukaryotes, both in wild-type cells and in various cell division cycle (cdc) mutants.

Wild-type S.pombe strain 972h^- and the mutant strains cdc 10, I29h^-, cdc11, 136?^- andc√cl3, 117h^- (Nurse et al. 1976) were kindly supplied by Dr Paul Nurse; strain ben?, D3 (Roy & Fantes, 1982) was kindly supplied by Dr Peter Fantes. Cultures were grown as previously described (Marks & Hyams, 1985). Temperature-sensitive cdc strains were grown at 25°C to mid-log phase prior to arrest at 36°C for 5–7 h. The cold-sensitive mutant 6e?4 was grown at 36°C to mid-log phase and blocked at 22·5°C for 24h. Intracellular F-actin distribution was determined by phalloidin staining (Wulf et al. 1979; Wieland & Govindan, 1974) according to Marks & Hyams (1985.). Nuclear morphology, and thus the position of the cell in its mitotic cycle, was determined using DAPI (4’-6-diamidino-2-phenylindole; Williamson & Fennell, 1975). Calcofluor white was used to reveal both the cell wall and the septum (Darken, 1961). Tubulin staining of wild-type cells using the monoclonal antibody to yeast tubulin, YOL 1β4 (Kilmartin et al. 1982), was performed essentially after the method of Kilmartin & Adams (1984).

A field of S. pombe cells stained with rhodamine-conjugated phalloidin as a probe for F-actin is shown in Fig. IB. Fluorescence is seen either at one end of the cell, at both ends or at the cell equator. End staining is mainly in the form of dots whereas equatorial actin is resolved as a filamentous ring. The different staining patterns may be ordered with respect to the cell cycle by reference to Fig. 1A, which shows the same field of cells stained with the cell wall stain Calcofluor and the DNA probe DAPI. Newly divided S. pombe cells grow only at the old end (Mitchison & Nurse, 1985). The two cell ends can be distinguished by their affinity for Calcofluor, the old (growing) end staining brightly whilst the new (non-growing) end appears as a dark, unstained hemisphere (Fig. 1A, cell 1). When new-end growth is initiated (NETO) this dark region is internalized and appears as a birth scar on the cell surface (Fig. 1A, cell 2). Comparison of the phalloidin staining patterns reveals that the distribution of actin coincides precisely with the polarity of cell growth; namely, actin resides at the single growing end before NETO and at both growing ends after NETO (compare cells 1 and 2 in Fig. IB). The transition from one-end to two-end staining also sees the transient appearance of fine filaments of F-actin possibly extending the length of the cell (cell3, Fig. IB).

Double-end actin staining is maintained until the onset of mitosis (cell 4, Fig. 1A), which coincides with the completion of end growth. Actin now disappears from the poles but reappears as a ring at the cell equator where it overlies the dividing nucleus (cell 4, Fig. 1A,B). The position of the actin ring anticipates the deposition of the septum, which stains intensely with Calcofluor (cell 5, Fig. 1A). As the septum grows centripetally the appearance of the equatorial actin changes from a filamentous ring to clusters of dots (cell 6, Fig. IB). At the completion of septation the remnants of the equatorial actin lie at the new ends of the two daughter cells (cell 7, Fig. IB). Since growth will be initiated at the opposite (old) end, actin must rapidly relocate to the other end of the cell before the next cell division cycle can begin. This transition is again accompanied by the transient appearance of fine actin fibres (not shown). The complete sequence of actin distribution through the S. pombe cell cycle is summarized in Fig. 2.

Further evidence for the relationship between actin and cell growth and division in fission yeast has been obtained from various temperature-sensitive cdc mutants. These become arrested at a specific point in the cell division cycle when grown at the restrictive temperature although their metabolic processes are maintained and they continue to elongate (Bonatti et al. 1972; Nurse et al. 1976). Fig. 3 shows Calcofluor and phalloidin images of cdclO, which arrests in G₁, i.e. prior to NETO. Intense actin staining is seen at the growing old end although a few dots are also seen at the new end (Fig. 3B). Correspondingly, although Calcofluor staining indicates that growth is predominantly at the old end, a small amount of cell wall deposition at the new end is also detectable (Fig. 3A). None of the cells display equatorial actin nor do they form septa. Similar images of c<?l3 are shown in Fig. 4. This mutant arrests in mitosis although under certain conditions a proportion of the cells leak through the temperature block and form multiple, aberrant septa (Fig. 4A). Growing cell ends (as judged by Calcofluor staining) again reveal intense actin fluorescence in the form of both dots and fibres (cells 1, 2 and 3 in Fig. 4B). Actin dots also occur in the region of the multiple septa in cell 4 whilst cells 5 and 6 reveal the equatorial actin ring, which anticipates the septum. The contiguous nature of the ring is particularly clearly seen in these cells. In cell 1, the equatorial actin is largely dispersed following the formation of a septum and intense staining is again seen at the cell poles.

Fig. 5 shows the relationship of the equatorial actin ring to nuclear position in an early septation mutant cdc 11. These cells are unable to undergo cytokinesis at the restrictive temperature although nuclear division and cell elongation continue. At the first mitosis following the temperature block the daughter nuclei return to a point either side of the middle of the cell (cell 1, Fig. 5B). Reinitiation of end growth is indicated by the bipolarity of actin staining (cell 1, Fig. 5A). Although these cells do not form a septum, a pair of actin rings forms at the positions occupied by the two nuclei prior to mitosis (cell 2, Fig. 5A). The nuclei do not appear to be physically connected to their respective actin rings since they are able to move away from their original position at the second mitotic division (cell 2, Fig. 5A,B). At the next mitosis actin rings appear at the position occupied by each of the four daughter nuclei (not shown).

Although not excluding other possibilities, we have begun to investigate whether the changing patterns of actin described above are dependent on the presence of cytoplasmic microtubules. S. pombe cells stained with anti-tubulin antibody are shown in Fig. 6. During interphase, groups of microtubules extend between the two ends of the cell. As the cell enters mitosis, these cytoplasmic microtubules disappear, to be replaced by an intranuclear spindle. The precise details of the cell cycle rearrangements of tubulin in 5. pombe will be presented elsewhere (I. M. Hagan, P. Nurse & J. S. Hyams, unpublished data). Since anti-microtubule drugs have only a limited effect on fission yeast (Burns, 1973; Walker, 1982; our unpublished results) we have attempted to destabilize cytoplasmic microtubules by genetic means. The cold-sensitive, benomyl-resistant mutant ben4 is unable to undergo cell division at 2O°C (Roy & Fantes, 1982). When cold-grown cells are stained with phalloidin, a dramatic rearrangement of actin is observed. Instead of multiple small dots at the ends of the cells, a single large dot is frequently present. Most obviously, however, much of the cellular F-actin is seen to be associated with the nucleus. Predominantly, this is in the form of a continuous ring, although sometimes this is twisted to form a figure eight and often bears a tail like a hangman’s noose (Fig. 7). These various perinuclear configurations are present in up to 90% of the cells.

The precise manner of cell growth in S’, pombe coupled with the availability of mutants affecting both the cytoskeleton and the cell division cycle make this a most attractive organism in which to investigate the structural rearrangements and interactions of cytoskeletal proteins through the cell cycle and the mechanisms whereby these are integrated with other cellular events. In this paper we have shown that the two major growth transitions in the S. pombe cell cycle, that is, from monopolar to bipolar cell growth early in G₂ (NETO), and the cessation of end growth and the initiation of cell division are accompanied by corresponding rearrangements of F-actin. These findings, which were originally established in wild-type cells (Marks & Hyams, 1985), have been confirmed here by the use of cell division cycle mutants arrested at different points of the cell cycle by growth at the restrictive temperature (Nurse et al. 1976). In cdc10 both actin and cell growth were predominantly monopolar (consistent with the known execution point of this mutant before NETO). It was noticeable, however, that a small amount of actin staining, and a corresponding degree of Calcofluor staining, was always detectable at the opposite pole. A small amount of new-end growth prior to NETO has been detected in wild-type cells although its extent has been difficult to assess, partly because it represents only a minor contribution to total cell growth and also because the methods of analysis used to date are relatively crude (Mitchison & Nurse, 1985). The problem may be resolved when more sophisticated methods for detecting zones of cell expansion in yeasts are applied to S. pombe (Staebell & Soli, 1985), or through the use of mutants such as cdc10, which can be held prior to NETO for extended periods. Preliminary observations of cdc11 have shown that the normal relationship between nuclear division and the formation and disappearance of the equatorial actin ring is maintained even in the absence of septation. Mitchison & Nurse (1985) noted that cdc11 showed pulses of cell growth interspersed with periods of quiescence. The fact that actin disappears from the cell ends with each cycle of ring formation provides an explanation for their observations. The relationship between the actin ring and septation is also clearly shown in cdc13, where the ring cycle and nuclear cycle become uncoupled and multiple septa are laid down in the absence of nuclear division.

At present we cannot distinguish whether the coincidence of actin and polarized cell growth is a cause or an effect relationship. In the case of septation, however, the situation is unambiguous, actin appearing at the cell equator prior to the deposition of Calcofluor-staining material. Both end growth and septation involve the deposition of new cell wall macromolecules, albeit of different chemical composition (Bush et al. 1974; Horisberger & Rouvet-Vauthey, 1985). In fungi this requires the mobilization of vesicles containing wall precursors to the growing region (McClure et al. 1968; Grove, 1978). Vesicles associated with the poles and septa of S. pombe cells have been reported (Oulevey et al. 1970; Johnson etal. 1973), and this suggests a possible explanation for the dot-like nature of actin staining. Although at present there is no evidence for the association of actin with cell wall vesicles in yeasts, vesicles coated with fine filaments of the approximate dimensions of F-actin have been observed in a filamentous fungus (Hoch & Howard, 1980). The role of actin in wall deposition may not be finally clarified until cell wall vesicles are available for biochemical study. The localization of myosin may also provide important clues (Watts et al. 1985) as will drugs that selectively interfere with cell wall morphology (Miyata et al. 1985, 1986), and experiments to these ends are in progress. Although the nature of the dot staining remains to be resolved, it is obviously a common feature of fungal cells, having been seen in three ascomycetous yeasts (Sa. cerevisiae, Sa. uvarum and S. pombe) (Kilmartin & Adams, 1984; Adams & Pringle, 1984; Marks & Hyams, 1985) and a filamentous basidiomycete (Uromyces phaseoli) (Hoch & Staples, 1983). This represents a broader sample than may initially be apparent, in view of the evolutionary divergence between budding and fission yeasts (Huysmans et al. 1983) and the fact that their mode of division is quite different. Cytokinesis in budding yeast is assymmetric and involves a chitin ring, whereas fission yeasts divide symmetrically and lack chitin (Bush et al. 1974).

The nature of actin staining as well as its position undergoes a marked change at the initiation of mitosis. Dot staining at the cell ends is replaced by a filamentous ring, which presumably occupies the thin layer of cytoplasm between the nucleus and the cell membrane (Streiblova & Girbardt, 1980). Our finding that a ring of actin anticipates the formation of the septum in S’, pombe is consistent with earlier studies of cytokinesis in fungi, which have provided ultrastructural evidence for the presence of such a structure (Girbardt, 1979). The intimate relationship between the nucleus and the actin ring is most clearly demonstrated by cdc11, which can go through multiple nuclear divisions at the restrictive temperature without undergoing cytokinesis (Nurse et al. 1976). An actin ring forms in association with each daughter nucleus at the start of mitosis; however, as the nuclei divide this spatial relationship is lost. The fact that cdc11 forms apparently normal actin rings and yet fails to lay down any detectable septal material may well be of value in establishing the coupling of these two events.

The relationship between the nucleus and the actin ring may also be addressed by our findings with ben4. This mutant has the classic phenotype of a tubulin gene mutation, namely, it is benomyl-resistant and cold-sensitive (Roy & Fantes, 1982). Genetic studies have shown, however, that the ben 4 gene is distinct from the known tubulin genes of 5. pombe (P. Fantes, personal communication; see Yanagida et al. 1985). The exciting possibility therefore exists that ben 4 codes for a protein that interacts with microtubules, i.e. a microtubule-associated protein (MAP). This is supported by immunofluorescence staining of ben 4 cells with anti-tubulin antibody, which reveals an apparently normal array of cytoplasmic microtubules (our unpublished results). The most conspicuous cytological feature of cold-treated ben 4 cells is the presence of a perinuclear F-actin ring, reminiscent of the rings of intermediate filaments that form around the nuclei of cultured cells treated with colchicine (Goldman, 1971). A possible explanation for the origin of these structures is that, like the β-tubulin mutant nda 3 (Hiraoka et al. 1984), cold-treated ben4 cells arrest at mitotic prophase. Since this is the time at which the equatorial actin ring appears, the perinuclear rings in ben4 may be the equatorial ring displaced from its normal location. Although Roy & Fantes reported that ben4 does not exhibit classical cell cycle arrest, we have used a different temperature from that used in their study and this can have a profound effect on the phenotype (Hiraokaet al. 1984). Clearly, examination of the distribution of actin in known tubulin mutants of 5. pombe will be of value in further establishing the nature of the ben4 mutation as well as clarifying the relationship between actin and tubulin in this organism.

Investigations of the role of microtubules in S. pombe will also be considerably aided by the introduction of the immunofluorescence techniques described here. Spindle microtubules have previously been demonstrated in this way (Hiraoka et al. 1984) but this is the first report of the visualization of cytoplasmic microtubules. These are much more abundant than has previously been appreciated from electron microscopy (Hereward, 1974; Streiblova & Girbardt, 1980; King & Hyams, l982 a,b; Tanaka & Kanbe, 1986) and extend from one end of the cell to the other. Cytoplasmic microtubules are clearly involved in the establishment and maintenance of cell morphology since treatment of ñssion yeast with anti-microtubule drugs results in a variety of morphological changes (Walker, 1982), as does the disruption of microtubules by means of tubulin gene mutations (Toda et al. 1983; Hiraoka et al. 1984). Whether microtubules are actively involved in the transport of cell wall precursors or merely establish general cell polarity is at present unknown. However, the fact that bud expansion in 5. cerevisiae proceeds in the absence of microtubules (Pringle et al. 1984) encourages us to favour the latter alternative. Irrespective of this, our results have clearly shown that the distribution of F-actin in fission yeast is controlled in a cell cycle site-specific manner. Whether understanding of this regulation will emerge through the wider study of cell cycle controls (Hayles & Nurse, 1986) or, more specifically, through the use of cloned actin genes and the identification and characterization of the actin binding proteins remains to be seen.

We thank Professor Th. Wieland and Dr John Kilmartin for the generous gifts of phalloidin and tubulin antibody, respectively. This work was supported by Action Research for the Crippled Child and the Science and Engineering Research Council.