Role of Cell Shape in Determination of the Division Plane in Schizosaccharomyces pombe: Random Orientation of Septa in Spherical Cells

Abstract

The establishment of growth polarity in Schizosaccharomyces pombe cells is a combined function of the cytoplasmic cytoskeleton and the shape of the cell wall inherited from the mother cell. The septum that divides the cylindrical cell into two siblings is formed midway between the growing poles and perpendicularly to the axis that connects them. Since the daughter cells also extend at their ends and form their septa at right angles to the longitudinal axis, their septal (division) planes lie parallel to those of the mother cell. To gain a better understanding of how this regularity is ensured, we investigated septation in spherical cells that do not inherit morphologically predetermined cell ends to establish poles for growth. We studied four mutants (defining four novel genes), over 95% of whose cells displayed a completely spherical morphology and a deficiency in mating and showed a random distribution of cytoplasmic microtubules, Tea1p, and F-actin, indicating that the cytoplasmic cytoskeleton was poorly polarized or apolar. Septum positioning was examined by visualizing septa and division scars by calcofluor staining and by the analysis of electron microscopic images. Freeze-substitution, freeze-etching, and scanning electron microscopy were used. We found that the elongated bipolar shape is not essential for the determination of a division plane that can separate the postmitotic nuclei. However, it seems to be necessary for the maintenance of the parallel orientation of septa over the generations. In the spherical cells, the division scars and septa usually lie at angles to each other on the cell surface. We hypothesize that the shape of the cell indirectly affects the positioning of the septum by directing the extension of the spindle.

Cell polarization is a fundamental requirement of cell growth, division, and differentiation. The polar growth ensures directed cell extension, the correct organization of cytoskeletal structures, and the development of proper cell shape (for recent reviews, see references 8, 16, 23, 25, and 35) and orients the mitotic spindle (reviewed in references 22, 36, and 48). In most animal and plant species, the mitotic spindle then determines the plane of division (39, 52). In the budding yeast Saccharomyces cerevisiae, the division plane is independent of spindle orientation because it is predetermined by the position of the bud (18). The division takes place at the narrow neck between the mother and bud cells. The site of bud formation is determined by the location of a previous one, and the axis of the spindle is oriented by directing cytoplasmic microtubules running from the spindle pole body into the bud (17, 18).

Schizosaccharomyces cells provide a tractable system for investigating cell polarity, because these cells grow in a highly polarized manner. In the unicellular yeast phase, their cells grow by polar extension at their tips and divide by medial septation followed by the splitting of the septum, a process frequently called fission (20). The newly born cell of Schizosaccharomyces pombe defines its poles (establishing its own polarity) and starts growing in one of three possible modes: old-end extension, new-end extension, or bipolar extension (31, 32, 43). In liquid Edinburgh minimal medium (EMM), most cells begin to grow at their old ends (the ends that existed in the previous cell cycle as the ends of the mother cells) and launch growth at the opposite new ends only with considerable delay (30, 31). The cells of Schizosaccharomyces japonicus also grow bipolarly, but they change to a unipolar extension pattern when converting to hyphae. In the hypha, the tip cell extends at its old end, whereas the rest of the cells grow at their new ends (45).

Polarity establishment in fission yeast cells is a complex process with numerous players. Upon cell division, the newborn cell reestablishes the interphase microtubular array by developing microtubules spanning the gaps between the cell ends along the longitudinal axis of the cell (for a review, see reference 12). The building of this structure is believed to be a major element of polarity establishment (29) and seems to involve a sort of “inherited structural memory” provided by the mother cell (42). Most cytoplasmic microtubules are nucleated by the cytoplasmic microtubule organizing center located near the splitting septum of the mother cell (the emerging new end of the daughter cell) and extend towards the opposite, so-called old end of the young cell (reviewed in reference 12). It has been shown that their extension is largely directed by the shape of the young cell determined by the rigid cell wall inherited from the mother cell (42). This external skeleton constrains the cytoplasmic microtubules to grow from the microtubule organizing center towards the opposite old cell end, where they converge. The microtubules deliver Tea1p, a protein associated with their growing tips and necessary for growth initiation to the old end (29). Most of the actin accumulates at the growing poles (26). Although these events have actually polarized the cytoplasm and marked a site for growth, the cell still has a function to perform before launching growth. It probes the cell wall of its end for suitability for growth. If the wall is covered with septal material left behind by incomplete cell separation, the cell shifts the site of growth to a subapical or lateral position (43). The subapical growth then elicits a bending of cell shape, the microtubular array, and the cell axis (42). Upon transition from G₂ to M, the interphase cytoskeleton breaks down and a mitotic spindle is formed which extends along the long axis of the cell (12). The septum formed midway between the cell poles is oriented perpendicularly to this axis, thus ensuring that the daughter nuclei are separated by the division process. A model published recently proposes that the site of septum formation is determined by the position of the premitotic nucleus (2). However, the septation pattern of the mutant sep2-SA2 indicates that a polarly generated signal might also be involved (10).

Since the inherited shape predetermines the direction of cell extension, the axis that connects the growing poles lays parallel to the axis of the mother cell. We hypothesize that this parallelism of axes causes the division (septation) planes, which are always perpendicular to the longitudinal axes, to also be parallel to one another over generations of dividing cells. A test of this possibility can be the comparison of septation planes in consecutive generations of spherical cells that do not inherit morphologically predetermined poles. Conversion of the standard cylindrical shape to round can be elicited by removal of the cell wall (protoplasting) (44), treatment with drugs (e.g., reference 33), or mutations. Both protoplasting and drug treatments are very drastic interventions which seriously affect or abolish propagation. The round mutants that can propagate seem more suitable for the investigation of division plane orientation over generations. However, most round mutants described so far show the spherical morphology as a lethal phenotype under conditions restrictive for growth or are heterogeneous in morphology (e.g., references 6, 7, 37, 38, 46, and 51). To have propagating cultures with highly homogeneous spherical morphology, we isolated novel nonconditional mutants whose cells grow by isotropic extension and form spherical daughters when they divide. In this study, we report the isolation and genetic and cytological analysis of four mutants defining four novel genes. In contrast to the cylindrical wild-type cells, these spherical cells form poorly polarized interphase cytoskeletons and show highly randomized division plane orientation. Their septa are usually laid down at angles to the septation planes of their mother cells. This septation pattern indicates that the shape of the S. pombe cell plays an important role not only in the polarization of the interphase cytoskeleton (42), but also in the orientation of spindle extension and the division plane.

MATERIALS AND METHODS

Strains and media.

S. pombe strains used in this study are listed in Table 1. Yeast extract liquid, yeast extract agar, malt extract agar (MEA), and synthetic minimal agar were prepared as previously described (41).

TABLE 1.

List of strains

Strain	Genotype	Origin or reference
0-1	L972 h− (wild type)	U. Leupold
0-77	SA24 h+S	L. Heim
0-177	ade3-58 h90	U. Leupold
0-39	leu1-32 h90	U. Leupold
2-416	cwg1-1 leu1-32 ura4d-18 h−	37
2-999	cwg1-1 leu1-32 h90	This study
2-418	cwg2-1 ura4d-18 h−	37
2-998	cwg2-1 leu1-32 h90	This study
2-726	orb1-13 ade6-M216 leu1-32 h−	51
2-727	orb2-34 ade6-M216 leu1-32 h−	51
2-728	orb3-167 ade6-M216 leu1-32 h−	51
2-729	orb5-19 ade6-M216 leu1-32 h−	51
2-730	orb6-25 ade6-M216 leu1-32 h−	51
2-731	orb7-34 (cwg2) ade6-M216 leu1-32 h−	51
2-732	orb8-35 ade6-M216 leu1-32 h−	51
2-733	orb9-46 ade6-M216 leu1-32 h−	51
2-734	orb10-9 ade6-M216 leu1-32 h−	51
2-735	orb11-59 ade6-M216 leu1-32 h−	51
2-736	orb12-5 ade6-M216 leu1-32 h−	51
3-384	ral1 leu1-32 ade6-216 h90	6
3-385	ral2 leu1-32 ade6-216 h90	6
3-386	ral3 leu1-32 ade6-216 h90	6
3-387	ral4 leu1-32 ade6-216 h90	6
2-997	sph1-1 h−	38
2-996	sph1-1 leu1-32 h90	This study
2-752	sph2-3 ade3-58 h90	This study
2-754	sph3-7 ade3-58 h90	This study
2-756	sph4-12 ade3-58 h90	This study
2-758	sph5-15 ade3-58 h90	This study
3-94	ste5(ras1)-JM70 leu1-32 h90	7

Isolation of spherical mutants.

Cells of overnight cultures of ade3-58 h⁹⁰ cultivated in yeast extract liquid were spread on MEA plates, were irradiated with UV (90% kill), and were incubated for 7 days at 25°C. The colonies formed were treated with iodine vapor to distinguish between sporulating and sterile colonies (24). Colonies containing spores stain dark with iodine. The iodine-negative colonies were isolated, inoculated on yeast extract agar, and incubated at 25°C for 48 h. The mutants showing spherical morphology were identified by microscopic examination.

Methods of genetic analysis.

Diploids were constructed by protoplast fusion (44). All other genetic methods, including random spore analysis and the construction of double mutants, were essentially the same as those described by Gutz et al. (11).

Cytological methods.

Septa and division scars were visualized by calcofluor (19). F-actin was stained with rhodamine-conjugated phalloidin (26). Tubulin staining was performed as described in reference 13. The primary antibody was the TAT1 anti-α-tubulin mouse monoclonal antibody (a gift from Keith Gull). A fluorescein isothiocyanate-conjugated goat anti-mouse immunoglobulin G antibody (Sigma F-0257) was used as secondary antibody. For Tea1p immunolocalization, affinity-purified anti-Tea1p antibody provided by J. Mata and P. Nurse was used (28). Fluorescein-linked anti-rabbit antibody (Amersham) at a 1:150 dilution was used as secondary antibody. Affinity-purified Sad1 antibodies (AP9.2; a gift of I. Hagan) were used to visualize the spindle pole bodies (SPBs) as described in reference 15. The preparations were examined with an Olympus BH2 fluorescence microscope. Freeze-substitution electron microscopy was carried out as described previously (53). For scanning electron microscopy, the cells were fixed with glutaraldehyde-osmium tetroxide, were dehydrated, and were critical-point dried (47). Freeze-etching was performed as described previously (49).

Tests of stress response.

The response of cells to hyperosmotic conditions, the presence of high concentrations of salts, and hyperoxidative and heat stress conditions was tested as described previously (9).

RESULTS

Isolation and characterization of spherical mutants.

Based on the assumption that cells which grow by isotropic extension (forming completely spherical cells) would be unable to form conjugation tubes, we isolated sterile mutants from the homothallic ade3-58 h⁹⁰ strain. One hundred sixty iodine-negative, nonsporulating colonies were isolated. Microscopic examination of the isolates revealed 22 mutants which displayed a high frequency of round cells. Four of these mutants showed highly uniform morphology (over 95% of the cells were perfectly spherical in overnight cultures grown at 25, 30, or 35°C [Fig. 1]). None of them formed conjugation tubes when mixed with the heterothallic h⁻ (L972) or h⁺ (SA24) tester strains on the sporulation medium MEA. These isolates were backcrossed to the morphologically wild-type leu1-32 h⁻ by protoplast fusion. The spherical cell shape and the sterility segregated together; thus the two defects were caused by single chromosomal mutations in each case. The spherical cells had a somewhat reduced growth rate, could easily be broken by pressing a coverslip on a drop of suspension placed on a microscope slide, and showed an increased sensitivity to the presence of CaCl₂ in the medium (the inhibitory concentrations of CaCl₂ were 600 mM and 150 to 200 mM for the wild-type L972 and the mutants, respectively). No difference was detected between the mutants and the wild type in their response to the physiological stresses imposed by KCl, NaCl, MgCl, mannit, sorbit, hyperoxidative (H₂O₂) conditions, or heat shock.

FIG. 1.

Phase-contrast morphology of cells. (a) Wild-type L972. (b) sph2-3. (c) sph5-15. Bar represents 2 μm.

Since the mutations recombined with the leu1-32 marker, we could isolate recombinants which were then used for hybridizing the isolates with each other and for constructing homozygous diploids. All homozygous diploids formed spherical cells which sporulated on the sporulation medium MEA, indicating that the meiosis-sporulation pathway was not affected by the mutations. The heterozygous diploids had cylindrical morphology, demonstrating that the mutations were recessive. In a random spore analysis, all of them segregated haploids with wild-type morphology; thus their mutations were not allelic. The isolates were also tested for complementation with mutations in the genes cwg1, cwg2, orb1 to orb12, ral1 to ral4, sph1, and ste5(ras1) (see Table 1), which are known to make the cells more or less round or oval (6, 7, 37, 38, 51). To overcome sterility, protoplast fusion was used for hybridization. All diploids showed wild-type morphology and segregated wild-type haploids, indicating that the mutations of the four isolates were not allelic with the mutations of the testers. From the results of the genetic analysis, we inferred that the four isolates carried nonallelic mutations which defined four novel morphogenetic genes. We designated them sph2-3, sph3-7, sph4-12, and sph5-15.

Isotropic extension in the mutant cells.

S. pombe cells grow by polar extension during the interphase of the cell cycle and by rounding the cell ends during septum cleavage (20). The sph⁻ mutants grow and propagate as spheres. We found less than 5% of cells with somewhat elongated shape, and cylindrical forms were seen very rarely (Fig. 1 and 2). The spherical cells displayed a continuous range of size, in which the separating sister cells represented the category of the smallest cells and the septated cells formed the class of the largest cells. Thus, the sph⁻ cells grew predominantly by isotropic extension. They had somewhat increased volume compared to the wild type and were easily lysed when pressed on the slides during microscopic observation.

FIG. 2.

Cell separation in spherical cells. Calcofluor-stained sph2-3 (a) and sph5-15 (b) cells. The images of two dividing cells are interpreted in the drawings below the microphotographs. ps, splitting primary septum; ss, secondary septum (the emerging new end). Bar represents 1 μm.

Septum cleavage takes place in the wild type by gradual centripetal degradation of the primary septum, the central layer of the mature septum (21). The primary septum can be visualized by calcofluor, a fluorescent brightener (19). Its dissolution is then followed by a fast rounding of the emerging cell end (secondary septum). In the sph⁻ mutants, the two processes took place in a different order: large parts of the separating sister cell ends remained covered with bright calcofluor-positive septal material after physical separation (Fig. 2). Obviously, the cell end rounded so drastically that it split the primary septum before it could become degraded.

Apolar interphase cytoskeleton.

In the growth phase, the wild-type cell has a polar cytoskeleton which can be made visible by staining its components such as tubulin, actin, Tea1p, etc. The most specific polar markers are actin (26) and Tea1p, a protein associated with the growing tips of the microtubules (28). To investigate the polarity of the sph⁻ cells, we stained actin and Tea1p and examined their distribution (Fig. 3). Three patterns were visible: in the few oval cells, both proteins were concentrated at the cell tips (Fig. 3d and h); in a small percentage of the spherical cells, some accumulation of the proteins was seen in distinct intracellular localities (Fig. 3g); but in the vast majority of the spherical cells, their distribution was random (Fig. 3b, c, e, and f). The nonspecific localization of Tea1p indicated that the cytoplasmic microtubules of most spherical cells were not oriented to particular sites and thus did not form polar arrays. This was verified by staining tubulin. As shown in Fig. 3j to t, the microtubules of the interphase cells (single SPB) in the spherical mutants appeared to form a disorganized crisscross pattern rather than the wild-type end-to-end pattern.

FIG. 3.

Distribution of cytoskeletal components in cells. (a) Polar accumulation of Tea1p in wild-type L972 cells. Apolar distribution of Tea1p in spherical cells of sph2-3 (b) and sph3-7 (c). (d) Distribution of Tea1p in an oval cell of sph2-3. Apolar distribution of F-actin in spherical sph2-3 (e) and sph3-7 (f) cells. (g) An sph2-3 cell with uneven distribution of F-actin. (h) Polar accumulation of F-actin in oval sph2-3 cells. (i) F-actin arranged along the septum in a dividing sph2-3 cell. Microtubules in sph2-3 (j), sph3-7 (k), sph4-12 (l), and sph5-15 (m) cells are shown. (n to r) SPBs in the cells are shown in panels j to m. (s) Microtubules and (t) SPBs in wild-type L972 cells. Note that the lower cell shown in panels s and t has two SPBs connected with a short spindle laid down parallel to the longitudinal cell axis. Bar represents 1 μm.

Diagonal septation and the orientation of the division plane.

The orientation of septation plane in the wild-type fission yeast cells is always such that the septum bisects the axis spanning the gap between the two separating nuclei, thus ensuring the equal partitioning of daughter chromosomes. Since the nuclei separate along the longitudinal axis of the cell (14), the septation plane is always perpendicular to this axis. The spherical cells have no morphological axis to determine the direction of nuclear separation. As shown in Fig. 1, this does not prevent division, and a septum is usually laid down in a diagonal plane. We asked whether the selection of division planes follows a rule or takes place in a random fashion. To investigate this question, we made use of the phenomenon that each division leaves behind a scar on the cell surface. The division scars can be examined by electron microscopy or by fluorescence microscopy upon staining the cell wall with calcofluor (20). They appear as weakly fluorescent bands when viewed with the fluorescence microscope (Fig. 4) and as surface ornamentation when viewed with the electron microscope (Fig. 5 and 6). The wall of a fission yeast cell always has at least one division scar (19). This scar marks the site where the mother cell divided and shows how the division plane was oriented. In the wild type, the plane of the scar is perpendicular to the longitudinal axis of the cell, indicating that the division plane in the mother cell was also perpendicular to the longitudinal axis. If more scars are visible on a wild-type cell, they lie parallel to each other and perpendicular to the cell axis (Fig. 4a, 5a, and 6a), which demonstrates that the orientation of the division planes is kept fixed over generations. As shown in Fig. 4b to d, Fig. 5b to e, and Fig. 6c to e, the scars of the spherical sph⁻ cells usually intersect each other or lie at various angles to each other and the new septum.

FIG. 4.

Division scars as seen on calcofluor-stained cells. (a) Wild-type L972 cells. Cells of the mutants sph2-3 (b), sph3-7 (c), and sph4-12 (d). Arrowheads show division scars (borderlines between wall regions arisen from secondary septa and wall regions produced by cell growth) of selected cells. Bar represents 1.5 μm.

FIG. 5.

Visualization of division scars by scanning and freeze-etching electron microscopy showing the position of division scars. Scanning electron microscopic images of the wild-type L972 (a) and the mutant sph2-3 (b) cells. Freeze-etching microscopic images of sph2-3 (c) and sph4-12 (d and e) cells. Bars represent 1 μm (a and b), 350 nm (d), and 600 nm (c and e), respectively.

FIG. 6.

Division scars on the cell wall of freeze-substituted cells. Examples of scars are marked with s. (a) A wild-type L972 cell showing two division scars. Each scar is visible as a pair of wall protuberances in opposite positions of the lateral cell wall. (b) An sph2-3 cell with one division scar (one pair of surface protuberances). (c) An sph2-3 cell with two division scars (three surface protuberances provided by two nonparallel scars). Note that this cell shows multiple septum initiation (arrowheads). (d) An sph5-15 cell with two scars (surface ornamentation similar to that of the cell shown in panel c). (e) Separating sph5-15 cells. The cell on the left side has one division scar, whereas its sister has two scars that intersect each other. Magnifications: 10,000× (panels a, b, and d), 12,000× (panel c), and 8,000× (panel e).

Multiple-septum initiation in sph2-3.

A recently published model proposes that the position of the premitotic nucleus determines the site of division, because the contractile ring is always formed on the cortex that overlays the nucleus (27). If this is the case, a spherical cell must cope with the problem of selecting one of the many possible cortical rings that all overlay the nucleus and are at various angles to each other. In principle, if the nucleus is in the center of the cell, any diagonal plane of the cell might become the division plane. Calcofluor staining, however, revealed only single septa in the dividing cells (Fig. 4). The sph2-3 mutant, however, showed a striking phenotype: its cells usually contained several wedge-shaped wall protuberances beside the complete septum (Fig. 6c). Their immature structures suggested that they might have been the products of false or aborted septum initiations.

DISCUSSION

We have previously proposed that the shape which the newly born S. pombe cell inherits from the mother cell predetermines the polarity of its growth by directing the extension of the microtubules when the new cytoplasmic cytoskeleton is being formed (42). The observations presented here indicate that the cell shape is also involved in the determination and orientation of the division plane, most probably by orienting the elongation of the spindle.

To study the role of cell shape in the determination of division planes, mutants which form completely spherical cells were isolated. Although S. pombe shows a regular cylindrical cell morphology and divides by medial transaxial septation, it can cope with a wide range of morphological deviations (20, 42, 43, 46). One type of the morphological aberrations is the rounding of the cell shape. The cylindrical cell can be rounded, for example, by spheroplasting (44) or by cultivating in the presence of aculeacin A (33). Neither treatment is lethal if applied briefly: both the spheroplasts and the aculeacin-treated cells can grow and divide and gradually restore the standard cylindrical morphology. Mutants carrying mutations in genes with diverse functions can form very short, almost round cells (e.g., wee1⁻ and cdc2w) or cultures with high proportions of oval and/or spherical cells (e.g., cwg, sph1, orb, ras, ral, sts, etc.) (5–7, 34, 37, 38, 50, 51). In many of them, the morphological conversion is temperature sensitive and associated with lethality. The sph⁻ mutants described here have almost exclusively spherical cells which divide into spherical progeny cells, and this phenotype is not dependent on the incubation temperature. Although the septum halves their cells into two hemispheres, the emerging daughter cells form into spheres by the end of cytokinesis. This is most probably due to the cytoplasmic turgor that bulges out the secondary septa simultaneously with the progression in cell separation. The physical force is so drastic that it rips the primary septum in two layers visible as calcofluor-positive material on the separating ends of the daughter cells. In the wild type, the new cell end also rounds, but only after the dissolution of the primary septum (19).

Shortly after the completion of cell separation, the daughter cells begin to grow. In the wild type, the growth is confined to the poles, which ensures that the cell shape remains cylindrical (30–32). In contrast, the sph⁻ cells extend over the whole surface, which suggests that they may have no poles for growth. The apolar distribution of Tea1p, tubulin, and F-actin corroborates this conclusion. Tea1p is associated with the ends of the cytoplasmic microtubules and is supposed to direct the growth machinery by tagging a region of the cortex as a growth site (28). F-actin is located at the growth sites (26). Their highly randomized distribution in the sph⁻ cells indicates that both the microtubular and the actin cytoskeletons must be poorly polarized or apolar and might result in diffuse growth over the whole cell surface. This might be the case in the cells which stain homogeneously when treated with calcofluor. The alternative possibility is that the cell manages to concentrate most of its growth to particular sites, but cannot elongate, because the pressure of the cytoplasmic turgor keeps the shape spherical by causing an isotropic swelling. The increased size and fragility of the mutant cells implies that their cell wall is less rigid and thus more flexible than that of the wild-type cells. This might have happened in the cells that showed distinct bright regions (probably growing sites) on their walls when stained with calcofluor (an example is shown in Fig. 4d).

The sph⁻ mutants were isolated as sterile, iodine-negative colonies. Microscopic examination revealed that they were defective in mating. Sexual mating requires intercellular communication by pheromones (for a review, see reference 3) and the production of a mating projection by unidirectional extension of the cell towards its partner (1). The round sph⁻ cells cannot form conjugation tubes. The rest of the sexual program, meiosis and sporulation, which do not require the direct involvement of the cell wall (40) but depend on pheromone signals (4), are not affected, indicating that the mating defect is also due to the inability of the cells to extend polarly rather than to a defect in pheromone communication.

The poles of the cylindrical wild-type cells define a longitudinal axis. The septum is placed midway between the poles so that its plane is perpendicular to the cell axis (20) and to the spindle extending parallel to the cell axis (14). This regularity ensures that the septum severs the cell between the separating mitotic nuclei and divides it into two uninucleate siblings. The spherical sph⁻ cell, however, has no morphological poles to define an axis. In spite of this, it can place a septum which halves the cytoplasm and separates the postmitotic nuclei. Thus, the cylindrical shape and its poles are not essential for correct determination of the division plane.

This conclusion is consistent with the model, suggesting that it is not the cell poles but the position of the premitotic nucleus that determines the site of septum initiation: the septum begins to form from the cortex that overlays the nucleus (27). However, in a spherical cell with the premitotic nucleus in the center, the whole cortex overlays the nucleus and thus any diagonal plane is equally probable for septum development. In spite of this, the sph⁻ cells form single septa. The mutant sph2-3 is peculiar in that its cells perform multiple events of septum initiation, which might reflect a sort of hesitation of the initiation machinery in the selection of the site for septum synthesis. However, most attempts turn out to be false or abortive and only one septum is completed in each cell. The ability of the spherical cells to define division planes that separate the daughter nuclei indicates that other factors besides the position of the premitotic nucleus must also participate in the placement of the septum. Similar conclusions were recently drawn from the septation pattern of the sep2-SA2 mutant. Its cells divide by transaxial septation like the wild type, but are prone to form twin septa, particularly under conditions that make them longer. It was hypothesized that sep2-SA2 mutation impairs the generation of the gradient of a hypothetical inhibitory polar signal. Septation can be initiated at the site(s) where the level of the inhibitor falls below a critical value (10). Since the morphologically apolar spherical cells can form septa, we propose that the signal is not generated by the cell poles but by an intracellular polar structure. By analogy with animal cells, this structure could be the spindle formed in the mitotic nucleus. Numerous observations suggest that in animal cells the mitotic spindle determines the position of the cleavage furrow between the spindle poles so that the division plane is perpendicular to the long axis of the spindle (see references 39 and 52 for reviews). The cleavage furrow bisects the mitotic apparatus, thus ensuring the equal partitioning of daughter chromosomes. A similar mechanism might also operate in S. pombe. The axis of the extending spindle and its poles, and perhaps the SPBs, might define the perpendicular plane along which the septum will be formed. In spherical cells the spindle can extend in any direction. Therefore, division planes in spherical cells can be laid down at angles to the division planes of their progenitors. The wild-type cells are elongated, and thus provide more space for spindle elongation along the longitudinal axis than in any other direction. Most probably, it is this spatial constraint that later causes the spindle to expand parallel to the longitudinal axis of the cell. Due to this parallelism, the septum will be laid down parallel to the septation plane of the mother cell. The proposed role of the cell shape in spindle orientation and septum placement is consistent with the earlier observation that the branched hyphal cells of the sep1-1 mutant were prone to form septa not perpendicular to the cell wall from which they grew (43). In bent cells, the mitotic nuclei separate along the bent cell axis, and thus the septum laid down perpendicularly to the axis is not necessarily perpendicular to the cell wall.

The results presented here and in earlier papers (10, 14, 33, 41, 42) indicate that the growth polarity (selection of sites for growth) and division polarity (orientation of the spindle and positioning of the septum) of fission yeast cells are largely predetermined by their cylindrical shape, which directs the extension of both the cytoplasmic microtubules and the spindle.