Serial sections of mitotic spindles of the marine cryptophycean alga, Cryptomonas, were analysed to determine what types of microtubules they contained and which of these microtubules came close enough to each other (50 nm or less) for the commonly described crossbridging to be possible. Interpolar microtubules were rare (≤ 1 %) but from prometaphase through anaphase there was a substantial interpolar framework of free and polar microtubules which came close enough to one another to cross-bridge and generate anaphase spindle elongation by intermicrotubule sliding. However, such elongation would also require some concomitant polar microtubule polymerization. In contrast, only about 12% of the chromosomal microtubules came within bridging distance of interpolar framework microtubules. Thus, assuming that microtubules were accurately fixed in their in vivo positions, crossbridging between chromosomal and interpolar framework microtubules is unlikely to function in chromosome-to-pole movement. In all stages the great majority of chromosomal microtubules were found to extend all the way from the chromatin to the pole. Although the identification of intermicrotubular bridges in section is often ambiguous, the best and most frequent examples of bridges in the present material were between polar microtubules and between chromosomal microtubules but not between the 2 different categories. The spindles also contained 5-nm- diameter microfilaments associated mainly with chromosomal microtubules and occasionally with polar microtubules. A specific model for the possible involvement of these filaments in mitosis is presented.
There are at least 2 major reasons for studying spindle ultrastructure in detail. One is that a large amount of data on the types and positioning of microtubules of spindles of a number of organisms is needed as a basis for developing experiments and hypotheses. A second is that certain models for chromosome movement, those involving intermicrotubule bridging, postulate a certain spatial arrangement of microtubules and it should be possible to determine whether this arrangement occurs. These models are of 2 types. The first is the sliding filament type proposed by McIntosh, Hepler & Van Wie (1969) and in modified form by Nicklas (1971). The model of McIntosh et al. suggests that spindle elongation is due to interpolar microtubules (see Fig. 1 for terminology) of opposite polarity sliding against each other, and that movement of the chromosomes to the poles is due to the sliding of kinetochore microtubules against the interpolar microtubules. In the Nicklas model interdigitating microtubules of opposite polarities are somehow fastened together and chromosomes move by kinetochore microtubules sliding against the interdigitating microtubules. In both models bridges between microtubules are seen as force producers. The second type of model is the zipper hypothesis of Bajer (1973). This model postulates that the movement of the chromosomes to the pole is due to the zipper-like formation of lateral links between kinetochore microtubules and an interpolar framework of microtubules causing a fanning out of the kinetochore microtubules and a net transport of the chromosome toward the pole. Both of these models are supported by observations that there are intermicrotubular bridges in the spindles of several organisms (e.g. Hepler & Jackson, 1968; Wilson, 1969; Hepler, McIntosh & Cleland, 1970; McIntosh, 1974) but the quantity and distribution throughout the spindle of these bridges remains unknown. A specific postulate of either model is that kinetochore microtubules must bridge to a framework of interpolar or interdigitating microtubules.
It should be possible to use electron microscopy of serial sections to test these hypotheses directly. In other words we should be able to see if the postulated microtubule configurations and intermicrotubular bridges really occur. There are, however, certain difficulties with this approach. One is that a positive result would be ambiguous. The presence of the postulated bridges would be consistent with but not prove that they were functional in movement. A second problem is that observers often differ in the identification of bridges, advocates often finding them where sceptics do not. This problem can be partially surmounted by using proximity of microtubules, and thus potential for cross-bridges, as a substitute for actual cross-bridges. Intermicrotubule bridges in spindles have been reported as being between io and 40 nm in length (McIntosh et al. 1969). If we choose a value larger than this, 50 nm for instance, we can say with certainty that microtubules which do not come closer than this distance do not possess bridges of the type reported in spindles and on which the cross-bridging hypotheses have been based. Again a positive result would be ambiguous because microtubules coming closer than 50 nm would not necessarily have to bridge and even if bridging occurred it would not necessarily be functional in movement. A third problem with this approach is that many of the spindles commonly used for mitotic study have general or localized regions of very high microtubule density (see abb. 2 a, Fuge, 1973; fig. 5, LaFountain, 1976, for example). This makes it very difficult to track microtubules and it also means that because of sheer microtubule density there is going to be an intermingling of kinetochore and interpolar framework microtubules. In order to test the validity of cross-bridging hypotheses, one would need a spindle with a microtubule density low enough to enable individual microtubules to be plotted with confidence and for random intermingling to be minimal. Other criteria for the selection of a suitable spindle would be that the spindle should be small (preferably), easily fixed, and exhibit as many characteristics in common with other organisms as possible. An organism on which we were working, Cryptomonas, appeared to satisfy most of these criteria.
We ask three questions in this study. First, what types of microtubules are present? Second, in what relative numbers are these types present? Third, do they come close enough for cross-bridging to occur? Other observations made in this study help elucidate the structure of spindles and help determine the validity of other hypotheses concerning the movement of chromosomes.
MATERIALS AND METHODS
The organism used was Cryptomonas sp., originally collected from the Gulf Stream and identified as Rhodomonas lens. Details of taxonomy, culture and fixation are found in Oakley & Bisalputra (1977). Briefly, initial fixation was in 1 % glutaraldehyde, 0·4 M sucrose, and 0·1 M phosphate buffer, pH 7·4. The glutaraldehyde was obtained from Ladd Industries and was checked spectrophotometrically for purity before use. Postfixation was in 1 % OsO4 in the same buffer, dehydration in ethanol, and embedding in Spurr’s resin.
For serial sectioning, cells were flat-embedded in the following manner. A single drop of resin, containing suspended cells, was placed on a slide coated with Glasskote (Manostat Corp.). Similarly coated No. 1 coverslips were placed on the slide on either side of the drop of resin and a third was suspended between the other two, thus forming a layer of resin one coverslip thick. After curing, the coverslips were removed and the thin layer of resin peeled off. The cells had settled to the bottom surface of the sheet of resin and could be seen clearly under a compound microscope. Dividing cells were selected and photographed using Nomarski interference contrast optics. They were then excised and remounted in an orientation which gave precise longitudinal sections of spindles. Final trimming of the block was performed on an ultramicrotome so that the height of the block face (about 15 μm or less) was only slightly greater than the diameter of the cell. Ribbons of approximately 50-nm-thick sections were picked up on Formvar-coated single-slot grids and stained with aqueous uranyl acetate and lead citrate. Section thickness was determined initially on the basis of interference colour and more accurately by the number of sections needed to pass through a number of spherical vesicles within each cell. Photographs of spindles were taken on a Philips EM 201 electron microscope using 3 5-mm roll film and printed to give final magnifications of 5 0 000-7 5 0 00 times. Microtubules were plotted using cellulose acetate overlays.
Because the validity of the results is dependent upon the ability of the method of microtubule plotting to determine continuity of microtubules from one section to another, and to determine intermicrotubule distances, it is necessary to explain briefly possible sources of error and to indicate their significance. In spindles with high microtubule densities tracking of microtubules from transverse sections is, in general, more accurate than from longitudinal sections. With the present material, however, it was considerably easier to obtain accurately oriented longitudinal sections than transverse sections and because of the relatively low microtubule density and the presence of numerous nearby cytoplasmic markers it proved possible to track microtubules with as much confidence from longitudinal as transverse sections. It is not possible to interpolate across missed or obscured longitudinal sections, however, so only intact series with no missing or obscured sections were used. An advantage of longitudinal sections is that data from intact series from portions of spindles are useful, since all of the microtubules except for those present on the end sections of the series can be tracked along their entire length. The data presented here are of this type. In order to avoid bias incomplete microtubules were not included in the data. In any type of microtubule tracking certain ambiguities arise and those relevant to this work are shown in Fig. 2. The problem shown in Fig. 2A is minimized by the low microtubule density. This problem would not, in fact, seriously influence the major conclusions of the work because it would cause an overestimation, not underestimation, of the numbers of continuous or inter-polar complex microtubules (as defined in Fig. 1). The potential problem of tracking directly superimposed microtubules (Fig. 2 B) was overcome by only using sections less than 50 nm thick. With a microtubule diameter of 20 nm and an observed surrounding zone of exclusion of 15 nm it would be impossible to have two superimposed microtubules in one 50-nm-thick section. The problem of a finite section thickness does affect the measurement of intermicrotubule distances, as shown in Fig. 2C, but such errors would produce an overestimate of the number of close microtubules. Likewise the ambiguous case shown in Fig. 2D was always decided in favour of closeness, thus again overestimating the incidence of close microtubules.
It should be emphasized that, in spite of the above comments, in the vast majority of cases both continuity (or otherwise) and intermicrotubule distances were unambiguous. If there was ever any ambiguity the decision was always in favour of continuity and closeness. Also the process of data collection incorporated as many blinds as possible so that subjective judgement would not affect the results. Thus microtubules on all sections were identified before tracking was begun and analysis of data was not begun until all of the data for a particular spindle were collected.
Outline of mitosis
A general account of mitosis and cytokinesis in Cryptcmcnas is being published elsewhere (Oakley & Bisalputra, 1977); thus, only a summary will be given here. During prometaphase the chromatin is scattered along the spindle (Fig. 3) but by metaphase it forms a dense 0·7-μm-thick equatorial plate (Fig. 4). This chromatin plate occupies the entire width of the spindle and contains numerous chromatin-free tunnels of less than 0·1 μm diameter through which some spindle microtubules run. This plate splits into two 0·4μm-thick plates which separate during anaphase (Fig. 5), then round up during telophase (Fig. 6). The metaphase spindle is barrel-shaped, with broad, diffuse poles which are only slightly smaller in diameter than the equator. Despite the absence of any obvious polar structures, the poles are readily defined as a zone beyond which microtubules do not extend. The equatorial diameter of the spindle is about 3·5 μm and the pole-to-pole distance is about 2·5 μm. Chromosomal microtubules (Figs. 12–15, p. 66) insert singly in the chromatin mass, which shows no differentiation at the insertion points except for a small extension toward the poles as if a poleward force were applied to it. During anaphase the spindle elongates to a maximum pole-to-pole distance of 4·5 μm, the chromatin-to-pole distance reduces to o and the spindle diameter decreases to about 1·3μm. The whole process from prometaphasc through cytokinesis lasts approximately 10 min.
Interrelationships among spindle microtubules
The data presented here are from intact series of longitudinal sections of substantial portions of 5 spindles: 1 prometaphase, 2 metaphase, 1 anaphase and 1 anaphase-telophase. Since the sample size is small, one must be cautious in generalizing about changes in microtubule distribution from stage to stage, since we have no idea of how much variation is present between, for example, anaphase stages in different cells or between early and late anaphase stages in the same cell. Moreover, it is not possible to compare absolute microtubule numbers between various stages because the data are not from entire spindles. Certain features of microtubule distribution occurred in all spindles examined, however, and these features can be set forth with confidence.
The Cryptomonas spindle contains all of the types of microtubules previously reported (e.g. Fuge, 1974; Heath, 1974; McIntosh, Cande & Snyder, 1975). The terminology used here is outlined in Fig. 1. Free microtubules are numerous but, being typically short, they constitute a small portion of the total polymerized microtubules. Throughout mitosis polar microtubules range from 0·1 μm to lengths close to the interpolar distance. Long and short, free and polar microtubules occur singly (i.e. more than 50 nm away from any other microtubule along their entire length) but they often also come close to each other in combinations shown in Fig. 1. The frequencies of these combinations at different mitotic stages are shown in Table 1, together with the frequencies of chromosomal and interpolar microtubules. From this table it is clear that; (a) interpolar microtubules are rare; (b) effective interpolar complexes of types B and c (Fig. 1) are more common and, when summed, contain almost as many polar microtubules as there are chromosomal microtubules. The frequency of interpolar complexes also appears to decrease during anaphase and into telophase. Note also that, as seen in Table 2, the amount of overlap between interdigitating microtubules (i.e. those shown in Fig. 1? and column G, Table 1) during early stages of division (prometaphase through metaphase) is not as great as the observed amount of anaphase spindle elongation (ca. 2 · 0 μm). In fact, except for a single pair in the prometaphase spindle, there are no examples of interdigitating microtubules with overlapping regions great enough to account for spindle elongation and the average amount of overlap is about 0 · 5 μm, only about a quarter of the observed elongation.
As with the above types of microtubules, so the chromosomal microtubules may, or may not, come close to other spindle microtubules. The frequencies with which chromosomal microtubules come close to other microtubules are shown in Table 3. In this Table it is clear that not only do approximately 50% of the chromosomal microtubules not come close to other microtubules at all, but also that few of the close pairings that occur are between chromosomal microtubules and any form of interpolar complex microtubule.
Spindle microtubule length changes
The distributions of lengths of chromosomal microtubules are shown in Fig. 7. There are obvious changes in both the average length and variability in length of the chromosomal microtubules during the course of mitosis. These changes reflect one major fact, however: that at all stages nearly all chromosomal microtubules extend to the pole. In prometaphase the chromatin is scattered along the spindle, and clumps of chromatin near a pole have short chromosomal microtubules which extend to the near pole and long microtubules which extend to the far pole. Thus there is greater variability in length at prometaphase than at the other stages. Likewise the movement of the chromatin to the pole is reflected by the shortening of the chromosomal microtubules in anaphase. The variability between the 2 metaphase spindles could be due to cell-cell variation or to time-dependent variability (i.e. early vs. late metaphase).
As shown in Table 2, there is no substantial change in the average length of the polar microtubules during mitosis. However, these average figures conceal the fact that some polar microtubules lengthen substantially during anaphase since by anaphase/telophase there are some polar microtubules whose lengths greatly exceed the maximum observed metaphase interpolar distance. This means that, since the average length does not change greatly, there is either shortening of some of the short polar microtubules or, what we believe to be more probable, the elongating microtubules make up only a small fraction of the total polar population; thus they do not significantly alter the average. This possibility cannot be analysed in detail unless very large numbers of spindles are analysed to eliminate cell-cell variability. The decrease in number of polar microtubules at anaphase/telophase could be accounted for in part by loss of microtubules but is probably mainly due to loss of the polar association of one end of each microtubule, so that those tubules would be identified as free, not polar.
Cross-bridges and microfilaments
In many situations it is difficult to demonstrate unambiguously inter-microtubule cross-bridges. This is true in the Cryptomonas spindle but structures which may be intermicrotubule cross-bridges do occur (Fig. 8). However, they have been observed only between adjacent chromosomal microtubules and between polar microtubules emanating from opposite poles; they have not been found between microtubules of different categories (e.g. chromosomal-interpolar or chromosomal-polar). Many microtubules of diverse categories are decorated by material comparable to the ‘cross-bridges’ described above, but this material ‘links’ the microtubule to the matrix of the spindle, rather than to any identifiable structure. However, in the region of the tunnels through the chromatin, there are suggestions of cross-bridging between the chromatin and polar microtubules (Fig. 9).
In addition to the rather ill-defined ‘cross-bridges’, the spindles also contain microfilaments which have a diameter of about 5 nm (Figs. 10, 11). These filaments were detected only between pairs of polar microtubules from opposite poles or pairs of chromosomal microtubules (Figs. 10, 11). Attempts to identify these microfilaments with heavy meromyosin labelling were unsuccessful because the glyceri- nation procedure caused excessive cell damage.
In order to explain chromosome-to-pole movements, cross-bridge models for mitosis (e.g. McIntosh et al. 1969; Nicklas, 1971) require cross-bridges between closely spaced chromosomal microtubules and interpolar complex microtubules of one type or another. Assuming that the in vivo arrangement of microtubules is accurately preserved by the fixation procedure (see below), the present results tend to contradict the idea that intermicrotubule bridging functions in chromosome-to-pole movement, since only approximately 12% of all chromosomal microtubules, including those in an anaphase spindle, come close to interpolar complex microtubules (Table 3). While it is quite possible that bridges to 12% of the chromosomal microtubules might be able to provide enough force to move the chromatin, it appears intuitively unlikely that nearly 90% of the chromosomal microtubules are inactive. Moreover, the fact that chromosomal microtubules come close to interpolar complex microtubules does not mean that they are bridged. If the 2 types of microtubules were distributed randomly one would expect a certain frequency of association and this frequency is calculable in the following manner. The cross-sectional area of the portion of the spindle analysed in a plane half way between the chromatin and the pole is easily calculable (from spindle width and section thickness), as is the area around chromosomal microtubules within which the centres of other microtubules would have to fall to have an edge-to-edge spacing of 50 nm or less. For the anaphase spindle analysed here these values are 4 · 59 and 1 · 37μm2 respectively, thus 1 · 37/4-59 = 0 · 30 = 30% of the cross-sectional area of the spindle is within bridging distance of a chromosomal microtubule. Since there are 68 interpolar complex microtubules in this spindle one would expect 0 · 30 (68) = 20 · 3 instances of interpolar microtubules close to chromosomal microtubules. Since there are 103 chromosomal microtubules in this area one would expect 20 · 3/103 = 19·7% of the chromosomal microtubules to be close to an interpolar complex microtubule on a random basis. The observed value is only 9%. Thus even allowing for the possibility that two or more interpolar complex microtubules might come close to one chromosomal microtubule and for impossibly large errors of measurement, it is apparent that there is no preferential association of chromosomal and interpolar complex microtubules and there may even be a tendency not to associate.
These data also suggest that a large amount of ‘zipping’ does not occur between chromosomal and interpolar complex microtubules, as postulated by Bajer & Molè- Bajer (1975 and earlier). Zipping could, however, occur between chromosomal microtubules and spindle matrix or some other spindle element. Our results do not contradict, nor do they significantly support other models for chromosome movement (Inoué & Sato, 1967; Dietz, 1972; Heath, 1975). It is interesting to note that, unlike the situation in Pales (Fuge, 1973), the chromosomal microtubules in Cryptomonas extend to the poles. This fulfils one of the requirements of the assembly-disassembly hypotheses (Inoué & Sato, 1967; Dietz, 1972). The 5-nm filaments are predicted by the work of Forer (1974) but are not sufficiently abundant, nor well preserved, to add much to his previously expressed ideas. While the present results tend to contradict previous sliding filament models for mitosis, Fig. 16 illustrates an alternative sliding filament model which is consistent with these results and which makes a specific prediction for the way in which filaments, actin or otherwise, could be involved in movement of chromosomes to spindle poles.
The rarity of interpolar microtubules in the Cryptomonas spindle tends to make the model of Inoué & Sato (1967) unattractive as an explanation for spindle elongation. Likewise the cross-bridge model of McIntosh et al. (1969), in its original form, requires sufficient metaphase overlap of cross-bridged polar microtubules to account for anaphase spindle elongation. This degree of overlap, and the necessary reduction in overlap during anaphase, is not present in the Cryptomonas spindle; thus this model is also inappropriate in its simplest form. However, a modification of this model to allow polar microtubules to polymerize as they slide is consistent with the present results and is an attractive explanation for anaphase spindle elongation in Cryptomonas. It should be noted that the simultaneous polymerization of polar microtubules and depolymerization of chromosomal microtubules observed in Cryptomonas and required by the above model requires a very fine localized control of microtubule polymerization, a feature which must obviously be considered in any postulated mechanism for control of mitosis.
The validity of the present results, and the above conclusions drawn from them, are dependent on the assumption that there has been no loss, shortening or movement of microtubules during preparation for electron microscopy. This assumption is common to all ultrastructural analyses of spindles but it is impossible to justify completely because there is no other technique which can give the same information at the necessary level of resolution. However, a number of points should be made. (a) The present results are comparable with other ultrastructural work on spindles because they have shown all known categories of spindle microtubules. At least no entire category of microtubule has been lost, (b) We do have close spacing and evident bridging between at least some members of relevant categories of microtubules, so that any artifactual loss of cross-bridges is partial at worst. It seems more likely that all, not some, of the bridges would be lost if any were lost due to preparation damage, (c) In order to explain the difference between the present results and those predicted by cross-bridge hypotheses, it would be necessary to postulate either substantial loss of interpolar complex microtubules or loss of crossbridges and subsequent separation of microtubules. The latter point is discussed above (point (b)). Because almost all interpolar complex microtubules must pass through the tunnels in the chromatin mass, substantial loss of these microtubules would probably leave some empty tunnels. Out of 231 tunnels observed in various spindles, none was empty. Furthermore, the tunnels typically have very little space from which microtubules could have been lost, so that any hypothetical loss must have been accompanied by concomitant closing of the tunnels through chromatin motility, an unlikely event. Thus, although this is not proven, we conclude that our results are probably a fair representation of the in vivo condition.
While observations on one organism are of intrinsic interest, it is desirable to be able to draw general conclusions applicable to a broad range of other organisms. We believe that the above conclusions made for the Cryptomonas spindle may fairly be extrapolated to higher organisms for the following reasons: (a) The Cryptomonas spindle contains all of the categories of microtubules found in mammalian spindles (e.g. McIntosh et al. 1975). (b) The stages of mitosis in Cryptomonas are very similar to those found in higher organisms, (c) Both spindle elongation and movement of chromatin to the poles during anaphase are characteristics of both Cryptomonas and higher organisms. (d) The rate of anaphase chromosome movement (Oakley & Bisalputra, 1977) is comparable to that reported for higher organisms (Mazia, 1961). The only obvious, possibly significant, difference between the spindles of Cryptomonas and higher organisms is the packing of the chromatin into plates rather than separate chromosomes. However, on balance it seems probable that there is functional homology between the spindles of Cryptomonas and higher organisms.
Given the above qualifications, the present method of analysis has produced a number of observations which help differentiate between extant hypotheses for mitosis. Whilst time-consuming, it is hoped that the type of questions which can demonstrably be answered by this type of analysis will encourage others to investigate similarly a range of different organisms, so that the generality of the present conclusions can be established or refuted.
We gratefully acknowledge the excellent help of Elizabeth Callan with the analysis of the data and Dorothy Gunning with the typing of the manuscript. The work was supported by a grant from the National Research Council of Canada to I.B.H.