Mitosis in the fission yeast Schizosaccharomyces pombe: a comparative study with light and electron microscopy

Observations on dividing fungal nuclei with the electron microscope are not yet extensive, and the literature based on light microscopy is filled with contradictions. Nevertheless, the evidence available seems to indicate that in many instances, the geometry of mitosis in fungi differs from the ‘classical’ pattern. Pickett-Heaps (1969) has suggested that variations in the patterns of mitosis which are encountered in different groups of fungi may represent ‘relic’ stages in the evolution of the more complex mitotic mechanisms in cells of plants and animals. Detailed studies of a wide range of fungi are needed for an assessment of this interesting generalization.

The fission yeast, Schizosaccharomyces pombe, a fungal organism now increasingly studied by cell physiologists and geneticists (Mitchison, 1970; Leupold, 1970) is also well suited for observations on mitosis. Large numbers of dividing cells are readily obtained, and the mitosis of the single central nucleus is related in a regular manner to the cell’s cycle of growth and division (Mitchison, 1970). Dividing nuclei can thus be expected in cells exceeding a certain length.

Previous studies of mitosis in S. pombe have been carried out exclusively with the light microscope. Early literature on the subject has been reviewed by Schopfer, Wustenfeld & Turian (1963). These authors, on the basis of observations made on cells stained with Giemsa solution after hydrolysis, concluded that ‘mitosis appears to be accomplished without the help of a spindle apparatus and without the formation of typical metaphase plates’. Using acid fuchsin, one of us (C. R.) has previously demonstrated the presence of an intranuclear fibre in dividing nuclei of this yeast. A photomicrograph of this fibre, together with several illustrations of chromosome configurations at mitosis similar to those shown by Schopfer et al. (1963), and a timelapse sequence of living dividing nuclei with phase-contrast microscopy, have been contributed by C.R. to a recent monograph on S. pombe by Mitchison (1970).

Until now, fixatives used in studies of the fine structure of S. pombe (MacLean, 1964; Schmitter & Barker, 1967; Osumi & Sando, 1969; Oulevey, Deshusses & Turian, 1970; Heslot, Goffeau & Louis, 1970), have been unsuitable for the preservation of nuclear organization. We have now made observations with the electron microscope which confirm and add to the previous knowledge of mitosis in this organism based on light microscopy. We have also made new observations on the structure and behaviour of the mitochondria and certain cytoplasmic inclusions.

Most of our observations have been made on cells of 2 morphologically indistinguishable diploid strains of S. pombe which were kindly supplied by Dr H. Gutz of the University of Texas. Because diploids are considerably larger than haploids, they were easier to examine and photograph with light microscopy. However, we have satisfied ourselves that nuclear structure and behaviour in normal wild-type haploids is the same as in diploids. The yeasts were maintained and propagated for all purposes at room temperature on a single medium consisting of Difco yeast extract 0·5 g, glucose 2·0 g, and agar 1·5 g per 100 ml of water.

We have used ‘spreading drop ‘slide cultures in 21 % gelatin prepared according to Robinow & Marak (1966).

After initial difficulties, the following reliable method was adopted for transferring sufficiently large numbers of growing and dividing cells from monolayers on agar to coverslips. A fairly heavy innoculum was streaked 2 or 3 times over a peripheral segment of a Petri dish and left overnight at room temperature. The next morning the cells were spread over the rest of the dish using a bent glass rod wetted with water. Eight hours later, some of the fresh crop of cells was scraped into a narrow ridge with the edge of a coverglass. The strip of agar covered by the heaped up cells was cut out, lifted from the dish, and placed, yeasts down, on the edge of a 22 × 22 mm coverslip. A small amount of fresh egg white was streaked alongside the slab of agar with a fine wire loop. The agar strip was then swiftly pushed sideways across the coverslip, leaving a thin film of cells dispersed in egg white on the coverslip which was immediately plunged into fixative contained in Columbia staining jars.

Two fixatives were used: Helly’s (mercuric chloride 5 g, potassium dichromate 3 g, water 100 ml; to 10 ml of which 0·6 ml of 37% formaldehyde (formalin) is added just before use); and formalin-acetic acid-alcohol (FAA) (formalin 5 ml, glacial acetic acid 5 ml, 95 % ethanol 50 ml, water 40 ml). Fixation with Helly’s mixture, which is unstable, was not extended beyond 10–20 min but preparations were sometimes left for longer periods in FAA. After both fixations the preparations were rinsed and stored in 70 % ethanol.

The HCl-Giemsa procedure recommended by Robinow in Mitchison (1970) proved satisfactory provided that decolorization with acidified water was carried far enough and the buffer-mounted specimen was adequately flattened with the help of the device described by Miller & Colaiace (1970). Transparent Giemsa preparations gave the same information as can be obtained with the aceto-orcein technique described in Mitchison (1970) but yielded more distinct photographs than the latter. Separation of the chromosomes was most easily achieved in material fixed in FAA, but the small portions of chromosomes that extend into the nucleolus (see below) are more obvious after Helly fixation.

These were selectively stained with acid fuchsin in 1% acetic acid (Robinow & Marak, 1966). After 4 min in a 1:4000o dilution of this stain, spindles and nucleoli stood out sharply against the unstained chromatin regions of nuclei in Helly-fixed preparations. Stained specimens were mounted over a drop of 1 % acetic acid. Cells fixed in FAA had no appreciable affinity for acid fuchsin, even when the dye was used in relatively high concentrations (e.g. i: 1000 for 10 min).

Helly-fixed cells were stained in two ways: (i) In toluidine blue, for i min in a 0·005 % aqueous solution which was acidified just before use with 1 drop of 0·01 N HC1 per 10 ml of stain. Stained specimens were mounted over a drop of acidified water (one drop of 0·01 N HC1 per 10 ml water), (ii) In Sudan black B, 0·5 % in ethylene glycol. Unstained specimens were directly mounted over a drop of the stain.

We used a Bausch and Lomb tungsten ribbon lamp and a Carl Zeiss microscope equipped with a VZ condenser (n.a. 1·4) adjustable for both phase-contrast and ordinary microscopy. A Kodak Wratten filter No. 11 was used in phase-contrast microscopy. For the photography of stained preparations this was replaced by an interference filter maximally transmitting at 546 nm. Photographs were taken with × 100 objectives (fluorite and apochromat for phasecontrast and ordinary microscopy respectively) in conjunction with a × 16 compensating eyepiece, on cut film (Kodak Super-Panchropress for phase-contrast; Ektapan for stained preparations) carried by a bellows at a distance above the eyepiece which provided an initial magnification of × 1800.

We used a modification of the method of Kamovsky (1965). Monolayers of growing cells in Petri dishes were flooded with Kamovsky’s formaldehyde-glutaraldehyde mixture. The cells were scraped off the plates, centrifuged, and resuspended in fresh fixative for 14 h at room temperature. The fixative was removed with 8–10 washes of 0·1 M cacodylate buffer at pH 7·2. The cells were post-fixed for 6 h at room temperature in 1·33 % OsO₄ in collidine buffer. They were then washed twice in collidine buffer and twice in distilled water before being stained in 0·5 % aqueous uranyl acetate for 2·5 h. The cells were next suspended in molten 1·5 % water agar at 47 °C and quickly centrifuged into a pellet before the agar could solidify. After cooling, the pellet of cells embedded in solid agar was cut into o-j-mm cubes. These were dehydrated in a graded ethanol series followed by anhydrous acetone and embedded in Araldite 6005 (Richardson, Jarett & Finke, 1960). Silver or grey sections, cut with a diamond knife, were mounted on carbon films and double-stained for 20 min with 1 % aqueous uranyl acetate followed by 8 min with lead citrate (Reynolds, 1963). Specimens were viewed with a Philips EM 200 electron microscope at 60 kV.

The cytoplasm of log-phase cells is closely packed with ribosomes and has many vacuoles scattered throughout (Figs. 20, 21, 39). With the electron microscope it can be seen that these vacuoles contain an extremely electron-dense material and are bounded by a unit membrane (see especially Fig. 38). The dense material does not fill the vacuole completely and loose coils of unit membrane are often present in the rest of the vacuole space. Phase-contrast microscopy of living cells shows the vacuoles as dark spheres about 0·4–0·6 μm in diameter (Figs. 1-8). After fixation for light microscopy, the vacuole contents stain red with toluidine blue at a low pH, a characteristic which may indicate the presence of polyphosphate (Keck & Stich, 1957). The vacuoles do not stain with Sudan black B (compare Figs. 31, 32) suggesting that no unbound lipid is present.

The cytoplasm of growing cells also contains a few spherical inclusions about 0·2–0·3 μm in diameter which tend to be in clusters at the ends of the cell. These have a low electron density and are not membrane-bound (Figs. 21, 39). Inclusions of the same size and relative quantity per cell are recognizable in the light microscope. With phase-contrast microscopy these appear as bright refractile droplets (Fig. 1). With bright-field microscopy it can be seen that these spheres stain blue with Sudan black B (Fig. 31). They are presumably lipid inclusions.

The mitochondria are short in cells of the early growth phase (Fig. 20) and become longer as the cell grows. In long cells (which are close to nuclear division), the mitochondria may stretch the entire length of the cell (Fig. 21), or they may be highly branched. They remain long during mitosis and appear to divide just before the inward growth of a transverse septum results in the formation of 2 daughter cells. Continued fragmentation of the mitochondria after closure of the septum (Fig. 41) seems to produce the short mitochondria that are characteristic of cells in the early growth phase.

The mitochondria in cells from growing cultures have a few long convoluted cristae surrounded by a fine grey matrix and ribosome-like particles (see especially Figs. 25–27). They also have mesosome-like membrane invaginations (Figs. 22–24). These membranes often seem to be covered with amorphous electron-dense material and they usually encircle an electron-transparent region. There may be several of these membranous inclusions in one mitochondrion (Fig. 21).

The regular association of a mitochondrion with the nucleus is dealt with below in the section which describes nuclear features seen with electron microscopy.

Phasecontrast microscopy has provided information about the sequence and timing of mitotic events and a standard by which to judge the quality of preservation achieved by the fixation method used for electron microscopy.

The nucleus in long cells just before division is spherical or ovoid and has a dark grey, excentrically placed nucleolar region and a transparent region which appropriate staining shows to be the site of chromatin (described below), although we see no indication of chromosomes (Fig. 1).

The first visible sign of the beginning of mitosis is a slight decrease in the density of the nucleolus. Next, a faint grey line appears across the chromatin region (Fig. 2). Electron microscopy and the evidence from stained preparations, to be described below, allows us to identify this grey line as the intranuclear spindle. Subsequently, there is a marked nuclear elongation into a dumbbell shape through intermediate stages as shown in Figs. 3–5 and 7. The nucleolus persists and is stretched out in the interior of the elongated nucleus.

Shortly after the slender dumbbell stage has been reached, there is a period of low visibility which may be due partly to a change in the consistency of the nuclear contents at this time, and partly to nuclear movements in addition to a straight gliding apart of the dumbbell ends. During this indistinct period, separation into daughter nuclei apparently takes place.

By the time 2 separated nuclei finally become clearly visible, their volume seems to be much greater than were the rounded ends of the dumbbell (compare Figs. 5, 7, with 6, 8). The nucleoli in these daughter nuclei are both excentrically placed but are tilted towards opposite sides of the cell (Figs. 6, 8). The latter observation is consistent with our conclusions based on electron microscopy, that nuclear separation involves pivoting movements of the dumbbell ends in opposite directions. The sense of direction of these proposed movements is indicated by arrows in Fig. 7.

Observations by light microscopy on cells stained with Giemsa. Giemsa’s stain shows clearly that the nucleolus and the bulk of the chromatin occupy distinctly separate nuclear regions. In sufficiently flattened, transparent preparations of FAA-fixed cells, the main mass of chromatin in resting nuclei can be resolved in most instances into a cluster of 5–7 short, thick bent rodlets which we regard as chromosomes (Figs. 9, 10). Their number seems to be the same in resting nuclei of the cells of haploid and diploid strains.

The intranuclear spindle which is revealed by phase-contrast microscopy in living dividing nuclei is not visible in hydrolysed Giemsa preparations.

The chromosomes are found contracted and aggregated into 2 groups already at an early stage of mitosis when the nucleus just begins to elongate and the shape of the nucleolus starts to become irregular (Figs. 11, 12). The details of the process of chromosome segregation which must be taking place at this time have remained obscure.

The rest of mitosis seems to involve increasingly wider separation of the 2 daughter chromosome clusters which, during this phase, are usually seen to diverge like spread fingers from 2 points of origin (in the geometrical sense) at opposite poles of the elongated nucleus (Figs. 13–15). At the same time the nucleolus is pulled apart into 2 tapering masses which remain in contact with the chromosomes at their base.

At all stages of the nuclear cycle the region occupied by the nucleolus contains some chromatin in the form of either short rods, as in the recently reconstituted nuclei of Fig. 16, or winding threads as in Fig. 17. We regard these objects as intranucleolar chromosomes or portions of chromosomes embedded in the nucleolus. They are invariably less chromatinic than the main mass of chromosomes aggregated beside the nucleolus, except in the relatively small compact nuclei of non-dividing cells (from cultures several days old) where the nucleoli are regularly seen to contain 2 distinct, presumably contracted chromatinic bodies which are as deeply stainable (if not more so) as the main mass of chromatin (Figs. 18, 19).

The following account describes changes in nuclear morphology which occur during the growth cycle of S. pombe as seen with electron microscopy. We have used 2 criteria for arranging these observations on fixed cells into a life-like sequence. First, characteristics such as cell length and the presence or absence of a septum which, in S. pombe are correlated with the growth cycle (Mitchison, 1970), and secondly, our knowledge of mitosis in living cells.

The early interphase nucleus is spherical, with an excentrically placed nucleolar region containing ribosome-like particles dispersed among amorphous electron-dense material, and a chromatin-containing region which is of uniformly low electron density, showing nothing recognizable as chromosomes (Fig. 20). Located on the outside of the nuclear envelope, adjacent to the chromatin-containing region and roughly opposite the position of the nucleolus is a structure which, for reasons to be outlined in the Discussion, we call the KCE (kinetochore equivalent). The position of this organelle at interphase is always closer to one of the side walls of the cell than it is to the cell’s longitudinal axis, and it is always lying in a narrow ribosome-free zone which is bounded by the outer nuclear membrane on one side and a mitochondrion on the other. Serial sections of the early interphase KCE show that it is a curved disk about 220 nm in diameter with a bulge in the centre of its concave side which is in contact with the nuclear envelope (Figs. 25, 26). This curved disk has 2 regions: an electron-dense portion consisting of at least 3 layers plus the bulge, and a fuzzy grey portion covering the convex surface of the electron-dense region. There is also some amorphous electron-dense material inside the nuclear envelope, beneath the site where the KCE appears to be attached, which has the effect of making this region stand out from the rest of the envelope.

At late interphase, the nucleus is enlarged and usually ovoid (Fig. 21). The KCE at this stage is in the same position but its morphology is different from that of the previously described interphase KCE. It has become longer and more bar-shaped and it lacks the fuzzy grey surface portion (Fig. 27). Like the early interphase KCE, this type of KCE is accompanied by amorphous electron-dense material underlining the inner aspect of the nuclear envelope.

On rare occasions we have observed what appears to be a divided KCE located beside an otherwise normal-looking late interphase nucleus. This structure can best be described as 2 short electron-dense bars which lie very close and almost parallel to each other in a ribosome-free identation of the nuclear envelope, with their long axes at right angles to the membrane surface (Fig. 28).

Nuclei in the earliest stages of mitosis that are comparable to what we have seen in the light microscope, look similar to interphase nuclei but are characterized by the presence of a short peripheral spindle which runs through the chromatin region and joins 2 KCEs located a short distance apart on the outside of the envelope; the short segment of envelope between the KCEs always appears to be convoluted or bulged (Figs. 29, 30, 33). However, over the greater part of their surface, these nuclei still retain the rounded or oval contours that we see in interphase nuclei. The peripheral spindle seems to be a narrow bundle of parallel microtubules which is slightly curved when the KCEs are a short distance apart (Fig. 29), and straight when the KCEs are more widely separated (Figs. 30, 33). At this peripheral spindle stage and at all subsequent stages of spindle formation, the KCEs have a non-layered, disk-like appearance and seem to be closely pressed to the nuclear envelope. As in interphase, KCEs at the ends of intranuclear spindles are located in ribosome-free zones and each is closely associated with a mitochondrion. These KCEs have approximately the same diameter as the curved, disk-shaped KCEs of early interphase but are considerably longer than the 2 component bars of the proposed divided KCE described above. The fact that we have not observed any of the intermediate steps which must occur between the divided KCE stage (Fig. 28) and the short, curved, peripheral spindle stage (Fig. 29), may indicate that this change occurs rapidly.

A later stage in mitosis is characterized by a nucleus which has a narrow, elongated, almost rectangular shape (Fig. 35). The nucleolus persists and is stretched out in the interior of the nucleus. The spindle microtubules occupy the longitudinal axis and seem to run parallel to each other, without interruption, between 2 KCEs located at opposite poles of the nucleus.

Subsequently the rectangular-shaped nucleus becomes further elongated into a dumbbell shape, with 2 rounded ends joined by a long narrow channel containing the spindle microtubules (Figs. 37–39). These microtubules are stretched in a long, straight, parallel bundle between the widely separated KCEs located at opposite dumbbell ends. Before the long dumbbell stage (Fig. 39) is reached, the nucleolus apparently becomes divided between the rounded daughter ends so that no nucleolar material is found in the narrow connecting channel. The whole dumbbell-shaped unit is surrounded by an intact nuclear envelope. This membrane is typically double in most places, but along the envelope of the spindle channel there are several minute regions of high electron density which can be resolved into fold-like arrangements, with 4 unit membranes instead of 2 (arrows Fig. 37 and insets). As reported by Oulevey et al. (1970), and shown in Fig. 39, the cell septum is sometimes beginning to form at this stage, before nuclear separation is complete.

Two of the final steps in the division process are the breaking down of the spindle channel, and the return of each KCE from its central position at a pole of the division axis to its interphase position on the surface of the nucleus facing a side wall of the cell (as in Figs. 20, 21). Our observations seem to indicate that both these steps are achieved at the same time by a pivoting movement of each daughter nucleus. This movement may be inferred from the shape and position of daughter nuclei in cells with an almost completed or a fully completed septum. Whereas the daughter ends of the dumbbell-shaped unit are elongated in the pulling-out direction, with each portion of the divided nucleolus located nearly opposite its respective polarly positioned KCE, a newly separated daughter nucleus is usually seen to be elongated almost perpendicularly to the division axis, with the nucleolus now opposite the KCE in its new interphase position at the side of the nucleus (compare shape of nuclei and the positions of the nucleoli and KCEs in Figs. 39 and 41). The 2 KCEs (and nucleoli) of recently separated nuclei are always seen to lie facing opposite side walls of the cell, suggesting that final separation of daughter nuclei may involve pivoting movements in opposite directions by end portions of the dumbbell shaped nucleus.

After separation has occurred, mitochondria which accompany the polar KCEs in the dumbbell stage are now consistently located in close apposition to the KCEs in their new interphase positions (compare again Figs. 39 and 41). At this stage the KCE once more becomes the curved, clearly layered disk which we have described as a characteristic of early interphase cells.

The relative positions of the spindle and nucleolus at every stage in the division process are closely similar to those we have observed in the electron microscope (compare Figs-33, 34: 35, 36: 39, 40). This technique confirms our electron-microscope observations that the spindle in S. pombe is, in reality not spindle-shaped but is instead a parallel bundle of microtubules, since, the ‘spindle’ stained with acid fuchsin has the shape of a thin wire. This is in contrast to the spindles in Aspergillus (Robinow & Caten, 1969) which have a distinct, tapered, cigar-like shape.

At stages when the nucleus is elongated and the spindle occupies the longitudinal division axis, the wire-like spindle of S. pombe has prominent knobs on the ends. The size of these knobs seems to correspond closely to the size of the KCE plus the ribosome-free zone as seen with electron microscopy at comparable stages of nuclear elongation.

The small bead-like ‘spindle initials’ which are a regular feature of the resting nuclei of Saccharomyces and Aspergillus (Robinow & Marak, 1966; Robinow & Caten, 1969) were not seen in the resting nuclei of S. pombe.

Since the osmophilic structures that have now been seen several times at the poles of mitotic and meiotic intranuclear spindles in ascomycetes and basidiomycetes have been given different names by different authors, as set out in Table 1, we must justify our use of Girbardt’s term ‘kinetochore equivalent’ (KCE) for the spindle-associated organelle in S’, pombe. All the names, except KCE, imply that the spindle-associated structures at the poles of dividing fungal nuclei are analogous to centrioles, since even the term ‘centrosome’ originally stood for a specialized region in the cytoplasm of animal cells containing a pair of centrioles (see Heidenhain, 1907, for a critical review of this once much-discussed subject). To consider these spindle-associated organelles in ascomycetes and basidiomycetes as modified centrioles one must assume that they have the same function as centrioles. As Pickett-Heaps (1969) has persuasively argued, the widely held belief that centrioles form spindles is based largely on circumstantial evidence and is unsound, since many cells without centrioles are able to form spindles. Then again, in certain cells that have large numbers of centriole-like basal bodies available to them, such as the ciliated protozoa, centrioles have never been observed at the poles of either mitotic or meiotic spindles. Moreover, even in those cells where a centriole is associated with the spindle, the microtubules are never directly connected to it but appear to arise either from small bar-shaped ‘centriolar satellites’, strikingly illustrated by Szollosi (1964) and de Harven (1968), or from the amorphous electron-dense material which has been frequently noted in the vicinity of centrioles (e.g. by Buck, 1967). All this evidence seems to indicate that the function of the centriole is not directly related to spindle formation, and it would seem inappropriate to use terms relating to centrioles for the organelles in fungi which are more consistently and more intimately associated with microtubules than are centrioles and do therefore seem to play a more likely role in spindle formation.

Pickett-Heaps (1969) proposes that the organelles in fungi are more accurately described as ‘Microtubule organizing centres’, a general term which he also used for similar electron-dense microtubule-associated regions in plant and animal cells including, among many examples, the kinetochore regions of chromosomes and the small electron-dense ‘nucleating centres’ in ectodermal cells of sea-urchin blastulae (Tilney, 1968). Experimental proof that the ‘nucleating centres’ are indeed sites of microtubule assembly has since been provided by Tilney & Goddard (1970).

Although ‘microtubule organizing centre ‘appears to be an acceptable general term for spindle-forming organelles in fungi, we propose that use of the specific term ‘kinetochore equivalent’ is justifiable in S. pombe for the following reasons: (i) The organelle in 5. pombe appears to organize microtubules. Many observations (reviewed by Luykx, 1970) suggest that this is also true of the kinetochore region of chromosomes, (ii) The structure of the organelle resembles that of a true kinetochore. The layered appearance of the organelle in S. pombe, and even more so of the comparable ‘centriolar plaque’ in Saccharomyces (see fig. 28 in Matile, Moor & Robinow, 1969) is reminiscent of the distinctly layered kinetochores of certain animal chromosomes (Jokelainen, 1967; Brinkley & Stubblefield, 1966, and others) and of the chromosomes in Oedogonium (Pickett-Heaps & Fowke, 1969). (iii) The organelle in S. pombe is attached to the chromosomes. Our evidence for this statement is, admittedly, not conclusive since the chromosomes of S. pombe cannot be seen in the electron microscope. However, comparison of the chromosome configurations in stained preparations with electron micrographs of dividing nuclei (e.g. compare Fig. 13 with Fig. 35), has left us with the impression that the chromosomes move apart in 2 clusters because they are attached to 2 opposing points on the nuclear envelope, and that these points are the spindle-associated organelles. Therefore we suggest that chromosomes are attached to the organelle, just as chromosome material is attached to a true kinetochore.

Support for this hypothesis is provided by our observations that the spindle in the light microscope appears to have the shape of a thin wire instead of a spindle shape, and that in the electron microscope it seems to consist of microtubules which run parallel to each other without interruption between the 2 organelles. Apparently, no discontinuous microtubules are present like the ones shown by Heath & Greenwood (1970) in Saprolegnia (in which, as in S. pombe, chromosomes are not visible with electron microscopy). One would expect to see similar discontinuous microtubules at anaphase in 5. pombe if the chromosomes were not, as we suggest, directly attached to the terminals of the microtubule bundle.

Thus, in the light of these arguments, we have adopted Girbardt’s term ‘kinetochore equivalent’ for the electron-dense organelles found at the ends of the intranuclear spindle in S. pombe.

This term is definitely not applicable to the spindle-associated organelles of all fungi. There is no doubt from the observations of Lu (1967) on Coprinus and of Wells (1970) and Zickler (1970) on Ascobolus that during meiosis in these fungi, separate chromosomes and chromosomal fibres exist. We consider that their so-called ‘centrosomes’ and ‘centriolar plaques’ may be best termed ‘microtubule organizing centres’ and may be analogous to centriolar ‘satellites’ instead of to kinetochores.

The number of linkage groups in S. pombe is 6 according to the extensive studies of Flores da Cunha (1970). This is in good agreement with the number of chromatinic rodlets in interphase nuclei (5–7), which we regard as chromosomes. The fact that approximately the same numbers are seen in the nuclei of diploid and haploid cells recalls the experience of Robinow & Caten (1969) with diploid and haploid Aspergillus nuclei, and suggests that in S. pombe as in Aspergillus, there may be somatic pairing of chromosomes. For a detailed treatment of this subject, which is beyond the scope of the present paper, the reader is referred to the extended discussions by Brown & Stack (1968) and Stack & Brown (1969).

We do not know whether the nucleolar chromosomes are separate from, or parts of, the main mass of chromosomes which we believe to be attached to the KCE. We feel certain however, that they are identical with the paired chromatinic elements which Schopfer et al. (1963) described as ‘possible centrosomes’. They are obviously not the same as the spindle-forming organelles discussed in the present work.

To clarify our discussion of these matters, the stages of mitosis in S. pombe are summarized in Table 2.

We propose that the 2 KCEs which are seen at opposite poles of short peripheral spindles in Stage II nuclei arise from the division (via the parallel-bar stage) of the single KCE associated with the interphase nucleus, and that they have been moved apart to their Stage II positions by a differential membrane growth of a region of nuclear envelope between them. We base this conclusion on the convoluted or bulging appearance of the short stretch of nuclear circumference between the KCEs at Stage II, which seems to indicate that this region has expanded, while the rest of the nuclear envelope has not and has therefore retained its round or oval interphase contours.

Further, we propose that this differential membrane growth could also achieve the change from Stage II to Stage III, since membrane growth exclusively in the region bounded by the peripheral spindle would ultimately move the KCE attachment points into polar positions and shift the excentrically placed nucleolus into a central position without any actual nucleolar movement. This is suggested by our phase-contrast observations (compare the position of the nucleolus in Figs. 1–3).

After Stage III has been achieved by differential membrane growth, Stage IV may be the expression of a general expansion of the nuclear envelope. This could cause a further separation of the KCEs since pressures in the cytoplasm resulting from the close proximity of the cell’s side walls would ensure that the growing nucleus expanded towards the ends of the cell. If it is assumed that the chromosomes and perhaps the nucleoli, too, via the nucleolar chromosomes, are in some way all linked to the 2 KCEs then, as the KCEs are moved farther and farther apart, there is eventually not enough nuclear material left in the central region of the nucleus to prevent the collapse of the nuclear envelope into a tight-fitting sleeve around the central part of the spindle.

The fold-like regions along the nuclear envelope of the spindle channel at Stage IV can be interpreted in 2 ways. They could be localized regions of membrane growth which, had the cell not been fixed, would soon have slid apart into an elongated stretch of normal double membrane. However, we do not know of any published description of membrane growth in this way, and we do not see these fold-like areas at any other stage in the elongation process. Therefore it seems more likely that increased resistance to movement encountered by the daughter ends of the nucleus at Stage IV as they approach the ends of the cell may cause the membrane, which is continuing to expand in the spindle channel region, to buckle back on itself.

We have rejected the possibility that the spindle is generating the force to cause the change from Stage I to Stage IV. Our observations show that the KCE especially during spindle formation, is closely associated with the nuclear envelope and appears to be attached. In Saccharomyces, attachment of the comparable ‘centriolar plaques’ has been even more clearly established by Robinow & Marak (1966), who showed that they were ‘set into enlarged pores of the nuclear envelope’. If the growing spindle were generating the force to separate the membrane-attached KCEs, the shape of the nucleus at each stage would be different from what we have observed. At Stage II we would expect the segment of envelope bounded by the peripheral spindle to be flatter, not more bulging than the rest of the nuclear envelope. At Stage III and IV we would expect the growing spindle to push out the poles of the nucleus into sharp points. On the contrary, we have observed that the ends of nuclei at Stage III and IV are rounded, with the KCEs sitting in slight membrane depressions.

We consider that spindle growth is keeping in step with rather than causing KCE separation, and that the spindle microtubules hold the separated KCEs rigidly apart. Microtubules are known to provide rigidity in many types of cells and, indeed, in mitosis of the dinoflagellate nucleus, there is good evidence that this is their chief function, while the chromosomes are moved apart by growth of the nuclear envelope (Kubai & Ris, 1969). Since the chromosomes seem to be attached to the KCE in S. pombe, the proposed membrane growth which moves KCEs apart may be causing genome separation just as it does in dinoflagellates and in the primitive bacterial mitotic process (Ryter, 1968).

Between Stage IV and Stage V, an interval of poor visibility intervenes in living cells seen with phase-contrast microscopy and we are therefore unclear about the details of daughter nuclei separation. It seems very likely that the microtubule precursors may have become depleted by the time the long spindle of Stage IV has become assembled and that the spindle may therefore start to break down shortly after this stage has been reached. However, the consistent pivoting of daughter dumbbell ends in opposite directions, as evidenced by the shape of daughter nuclei and the positions of their KCEs and nucleoli in our electron micrographs of Stage V, together with our phase-contrast observations at this stage which show the nucleoli at opposite sides of the daughter nuclei, suggest that other forces besides spindle breakdown must be at work between Stages IV and V. Girbardt (1968) has been led by cinematographic evidence to postulate the existence of ‘cytoplasmic forces ’ which act on the KCEs of dividing hyphal nuclei of Polystictus. Perhaps such forces also act on the KCEs of S. pombe after spindle breakdown. It needs only the further assumption of an intrinsic polarity of the organization of the KCEs to account for their being moved in opposite directions by the putative cytoplasmic forces.

Our view of the KCE attached to an expanding nuclear envelope as the main instrument of chromosome separation, leads us to support Pickett-Heaps (1969) in his belief that among the fungi ‘relics of primitive ultrastructure’ may be found preserved and functioning.

Mitochondria have not been seen in close association with the spindle-forming organelles of other fungi. In contrast, every KCE that we have seen in S. pombe, both in interphase and at mitosis, has been located in a narrow ribosome-free zone which is separated from the rest of the cytoplasm by a mitochondrion. Since there is no reason to suspect that the mitotic process in S. pombe requires more energy than it does in other fungi, we believe that the KCE-associated mitochondria may be playing another role in addition to energy production. At present, we cannot suggest what this role might be.

It is difficult to know whether the unique mesosome-like membranes inside the mitochondria of log-phase S. pombe cells are real structures or artifacts of fixation. We feel certain, however, that they correspond to the electron-transparent ‘nucleoid ‘regions described by Osumi & Sando (1969) in S. pombe mitochondria after permanganate fixation, because these ‘nucleoids’ are similar in size and position to our ‘mesosomes’ and it was not clearly demonstrated that DNA was present in them. A full consideration of this subject is beyond the scope of the present work. However, it is interesting to note that these structures are not present in the mitochondria of normal stationary-phase cells (our own observation; and Heslot et al. (1970)). This seems to indicate that these mitochondrial regions have some functional significance.

This work has been supported by a grant to C. Robinow by the Medical Research Council of Canada.

The authors thank Miss lulia Wang for reliable and enthusiastic technical assistance and Mr John Marak for his skill and resourcefulness in keeping the electron microscope working at peak efficiency.

Figs. 1-4 are of the same cell. Figs. 5, 6 are of another cell, × 3600. The interphase nucleus (Fig. 1) has 2 distinct regions. One is a dark grey, excentrically placed nucleolus (n) and the other is a region of low density which is known from stained preparations to be the site of chromatin (chr, see Figs. 9, 10). Figs. 2–4 illustrate the transformation of this interphase nucleus into an elongated, almost rectangular nucleus with the nucleolus stretched out in a central position. In Fig. 2 the spindle (sp-between arrows) can be seen as a fine grey line crossing the nuclear region of low density. Fig. 2 was taken 13 min after Fig. 1. The intervals between Figs. 2 and 3 and between Figs. 3 and 4 are 4 and 6 min, respectively. Fig. 5 shows an early dumbbell stage. The nucleolus is stretched out in the constricted central portion, and 2 regions of low density (chr, see Fig. 13) can be seen at opposite ends of the nucleus. In Fig. 6 nuclear division has been completed and a transverse septum (s) has started to form (between asterisks). Note how the nucleoli of the 2 daughter nuclei are facing opposite side walls of the cell. The time interval between Figs. 5 and 6 is 52 min. Two types of cytoplasmic inclusions are visible in these living cells: refractile lipid droplets (l) and dense spherical vacuoles (v).

Figs. 7, 8. Interpretive drawings of Figs. 5, 6, respectively. The arrows indicate how we think the ends of the dumbbell will pivot when final separation of daughter nuclei takes place (see text).

Figs. 9-19. Distribution of chromosomes in resting and dividing nuclei of S. pombe shown by Giemsa staining after hydrolysis. For Figs. 9–17 the fixative was Helly’s, for Figs. 18, 19 it was FAA. × 3600.

Fig. 25-29. Sequence showing the change in KCE morphology between early interphase and the initial stages of spindle formation at mitosis. A portion of the nucleus can be seen towards the bottom of each micrograph. The electron-dense KCEs at each stage are located on the outside of the nuclear envelope in a ribosome-free zone and are accompanied by a mitochondrion (m) on their cytoplasmic side, × 80700.

Figs. 33, 34. Closely comparable electron and light micrographs showing the relative positions of nucleoli (n) and spindles (sp) in dividing nuclei which are not yet markedly elongated and still have excentrically placed nucleoli, as in interphase.

Figs. 35, 36. Closely comparable light and electron micrographs shotting the relative positions of the nucleolus (n) and spindle (sp) in dividing elongated nuclei which have a more or less rectangular shape.

Figs. 37, 38. Two portions of a dumbbell-shaped nucleus shown whole in Fig. 39.

Figs. 39, 40. Closely comparable light and electron micrographs showing the relative positions of the nucleolus (n) and the spindle (sp) in dividing nuclei which have elongated into a dumbbell shape. The nucleolar material is divided between the daughter dumbbell ends which are joined by a long spindle.