Actin in spindles of Haemanthus Katherinae endosperm :IX. general results using various glycerination methods

Actin has been identified in mitotic and meiotic spindles, using electron microscopy, and actin has been identified in mitotic spindles using fluorescence microscopy, both in conjunction with antibodies against actin and in conjunction with fluorescently labelled heavy meromyosin or heavy meromyosin subfragment 1 (review in Forer, 1978 a). While electron-microscopic detection of actin-heavy meromyosin ‘arrowhead complexes’ gives unambiguous identification of actin (review in Forer, 1978 b), detection of actin using fluorescence techniques has not yet allowed unambiguous identification of spindle actin : ambiguities are present with the fluorescence techniques, as used to date. For example, it is possible that the antibodies against actin or the fluorescently labelled subfragment 1 reacted with cellular components other than actin in the experiments so far reported (review in Forer, 1978a; see also Ishiura, Shibata-Sekiya, Kato & Tonomura, 1977).

Even accepting the electron-microscopic evidence that actin is present in spindles, this does not necessarily mean that actin functions in producing force for chromosome movement. Indeed, several authors have suggested that actin seen in the spindle originally was present outside the spindle and moved into the spindle only during glycerination, while others have suggested that actin seen in the spindle is ‘trapped’ there during formation of the spindle, and hence has no role in chromosome movement (review in Forer, 1978a). How does one decide whether actin is a ‘passive’ spindle component or, alternatively, whether actin functions in producing force for chromosome movement?

Determining the locations of actin in the spindle might help decide whether actin has an active role in chromosome movement. The forces causing chromosomes to move polewards during anaphase arise from some component(s) associated with the light-microscopically identified spindle fibres which extend between chromosomal kinetochores and poles, i.e. the chromosomal spindle fibres (reviews in Schrader, 1953 ; Mazia, 1961; Forer, 1969, 1974; Nicklas, 1971, 1975). Thus, if actin is in the spindle because it is ‘trapped’ there or because it ‘relocated’ during treatment with glycerol and heavy meromyosin (HMM), then one would expect that actin filaments may or may not be associated with chromosomal spindle fibres. If on the other hand, actin functions in producing force for chromosomal movement, then actin must be associated with each and every chromosomal spindle fibre (and/or chromosome). To see if actin is associated with each and every chromosomal spindle fibre, we have looked electron-microscopically at actin filaments in Haemanthus endosperm cells after treatment with glycerol and heavy meromyosin. We have chosen electron-microscopic study of actin, rather than immunofluorescent study of actin or study of actin using fluorescent-HMM, because of ambiguities associated with the fluorescence techniques as used to date (see Forer, 1978 a), and because resolution using electron microscopy is greater than that using light microscopy. We have chosen Haemanthus endosperm cells because these cells contain little extra-spindle actin; this minimizes the possibility that actin found in the spindle had moved into the spindle during glycerination.

We have also looked light microscopically at cells undergoing glycerination, and in these experiments we used both Haemanthus endosperm cells and crane-fly spermatocytes. We used Haemanthus cells for comparison with the electron-microscopic observations; we used crane-fly spermatocytes because individual chromosomal spindle fibres are visible in these cells. These light-microscopic observations are limited in that one cannot look directly at actin in living cells using light microscopy. Nonetheless, there is some value in such observations, and we report herein phasecontrast microscopic observations and polarization-microscopic observations on glycerinated cells, and on cells undergoing glycerination using various glycerination methods. Electron-microscopic observations on similarly treated cells are also reported. In the following article (Forer, Jackson & Engberg, 1979) we describe the distribution of actin filaments in chromosomal spindle fibres as determined from analysis of serial sections.

Crane flies (Nephrotoma suturalis Loew) were reared in the laboratory. Living meiosis-I spermatocytes were held in place in a fibrin clot using procedures published elsewhere (Forer, 1972). Cells were studied using polarization microscopy, both before and after glycerination.

Haemanthus katherinae (Baker) plants were grown in greenhouses, and endosperm samples were collected from immature fruits about 4–5 weeks after pollination. Preparations in Hanover used fruits taken directly from plants. Preparations in Toronto used either fruits taken directly from plants or fruits which had been removed from plants in Hanover and then shipped to Toronto. Such removal of fruits from plants does not destroy the capacity of endosperm cells to divide (Jackson, in preparation).

To study Haemanthus endosperm cells during glycerination, living cells were held in place using a fibrin clot (see Forer, 1972). HoNine fibrinogen (Miles laboratory, fraction I, plasminogen free, 90% clottable) was dissolved in 25 ml of 3.5% glucose to which had been added 1 ml of standard salts solution (0.05 M KC1, 0.005 M MgCl₂, and 0.005M Sørensen’s phosphate buffer, pH 6.8), at room temperature, at a final concentration of 20 mg of fibrinogen per ml. The nondissolved material was removed by filtration through Whatman filter paper. Small aliquots (< i ml) were stored frozen at —20 °C. Individual aliquots were thawed and stored at room temperature before use; individual aliquots were thawed only once, and were then discarded. Bovine thrombin (Sigma, 500 units, containing 0-15 M NaCl and 0 05 M sodium citrate if dissolved in i ml of H₂O) was dissolved in 5 ml of 3-5% glucose to give a final concentration of 100 units of thrombin per ml. Small aliquots were stored at-20 °C. Individual aliquots were thawed and stored on ice before use, and individual aliquots were used repeatedly. To embed the living cells in a clot, living cells were mixed with the fibrinogen, and then thrombin was added to polymerize the fibrin, as follows.

Holes (22 mm diameter) were drilled in 1 mm × 50 mm × 75 mm glass slides, and glass coverslips were placed under the holes and affixed to the glass, forming i-mm-deep wells. A 50-μ 1 drop of fibrinogen was placed on the coverslip on the inside of the well (the slide side of the coverslip), by means of a ‘Pipetman ‘pipette; then endosperm from one seed was thoroughly mixed into the fibrinogen, 15 μ 1 of thrombin was rapidly mixed into the fibrinogen-cell mixture (using a Pipetman pipette), and a coverslip was placed on top of the slide, to cover the hole and prevent evaporation. (This coverslip was placed loosely on top of the hole, and not affixed to the slide.) A clot formed within a few minutes, and under these conditions endosperm cells divided normally for hours. We studied such preparations during glycerination, using phase-contrast microscopy (using a Nikon 40 × phase-contrast lens, N.A. 0 6). To do this, the slides were placed cell side down on the stage of an inverted microscope, and cells were photographed on 16-mm film at intervals of 4 – 25 s. Whilst continuing to photograph the cells, the upper coverslip was removed, and the glycerol-containing solution was added to the well, to fill up the well. Thus, the cell was photographed before the glycerination, during the glycerination procedure, and for varying times afterwards (up to 45 min), until no further changes seemed to take place. (Sometimes the glycerol solution in the well was replaced by fresh glycerol solution.)

Haemanthus endosperm cells were also studied after glycerination, using polarization microscopy. Haemanthus endosperm was mixed thoroughly with the glycerol-containing solution in question, a slide was made as rapidly thereafter as possible, and cells were studied in a polarizing microscope (see Forer & Zimmerman, 1974, for optical conditions), and photographed using a 35-mm camera.

Various glycerination solutions were used. 50% glycerol-standard salts solution is a 1 + 1 mixture of glycerol (Fisher) plus standard salts solution (0.05 M KC1, 0.005 ^M MgCl₂, and 0.005 M Sorensen’s phosphate buffer, pH 6-8). 25% glycerol-standard salts solution is a 1+3 mixture of glycerol plus standard salts solution. 6% glycerol-standard salts solution is a 1 + 15 mixture (glycerol plus standard salts solution). 6% glucose-standard salts solution is 6 g glucose in 100ml standard salts solution. When the glycerol-standard salts solution inquestion contained Triton-X-100 (obtained from British Drug Houses Ltd.), the solutions were mixed as above except that the volume of the standard salts solution was reduced and a corresponding volume of stock i% Triton in standard salts solution was added to achieve the required final concentration of Triton.

Procedures for negatively staining actin or actin-HMM complexes were as given in Forer (1978 b) : actin and HMM were mixed in the solutions in question, placed on a grid, rinsed with bacitracin, and negatively stained using 1% uranyl acetate in water.

HMM and HMM-subfragment-1 (Si) were prepared as in Appendix II and Appendix III B of Forer (1978 b), actin was prepared as in Appendix IV of Forer (1978 b), and concentrated stock solutions were stored at —20 °C in 25% glycerol-standard salts until use.

Cells were studied electron microscopically after embedment in Araldite and subsequent sectioning. These cells were prepared using one of the following procedures. (1) Some samples were glycerinated with 50% glycerol-standard salts, centrifuged and resuspended in HMM in standard salts solution, fixed by addition of an equal volume of 4% glutaraldehyde (TAAB, from Marivac Services Ltd., Halifax, N.S.) in standard salts solution, and immediately centrifuged and resuspended in 2% glutaraldehyde in standard salts solution, as described previously (Forer & Jackson, 1976). (2) Other samples were treated with a new procedure: endosperm was suspended in a glycerol solution containing HMM or Si, with or without Triton, and 30 min later these samples were fixed by addition of an equal volume of 4% glutaraldehyde in the same glycerol solution; then the cells were immediately centrifuged out of solution and the pellet of cells was resuspended in 2% glutaraldehyde in the same glycerol solution. (Otherwise, without the centrifugation after glutaraldehyde was added to the HMM solution, the glutaraldehyde caused the HMM or Si to form a precipitate, and this precipitate interfered with subsequent processing of the cells and identification of the cells in the final flat embedment.) The rationale for the changedprocedure is that, unlike the original procedure, in the newer procedures the cells are suspended once in a solution, and are not centrifuged or otherwise handled harshly until after fixation with glutaraldehyde. The solutions used for the newer procedure contained i mg/ml of HMM (or Si) and the HMM was in 25% glycerol-standard salts solution or 6% glycerol-standard salts solution, and sometimes the solutions contained Triton, in final concentrations as specified in the Results section. We studied cells glycerinated with 25% glycerol which contained 0.01, 0.05, 0.1, 0.25, 0.5 or 10% Triton.

After samples were fixed in glutaraldehyde they were stored in glutaraldehyde for varying lengths of time, from hours to weeks, and sometimes samples in glutaraldehyde were shipped from Hanover to Toronto. Then cells were post-fixed for 45 min-i h with 1% OsO₄ in standard salts solution, rinsed with H,O, flat-embedded in 1.2% agar at ∼ 40 °C, and after the agar hardened the thin agar sheet was cut into small pieces (∼ 3 mm × 6 mm). The small pieces of agar either were placed in 1% uranyl acetate in H₂O, overnight, and then dehydrated through a graded series of acetone solutions (beginning with 50% acetone) and into Araldite, or they were placed directly into 50% acetone and then dehydrated through to 100% acetone and placed in Araldite. The Araldite was hardened in a 60 °C oven for 48–72 h.

Individual cells of the proper stage were identified in the flat sheets of Araldite; they were then marked, drilled out of the Araldite, and mounted (with Epoxy glue) onto the end of a block (see Zimmerman, Zimmerman & Forer, 1977, for details of the procedure). The cells were sectioned with a diamond knife, using a Porter-Blum Ultramicrotome; serial sections were placed onto films in the centres of single-hole (slot) grids, stained with 1% uranyl acetate (Watson, 1958) followed by lead citrate (Fiske, 1966), and were studied using either a Philips EM 200 or a Philips EM 201 electron microscope, operated at 60 kV.

We studied the effects of 25% glycerol and of Triton on the interaction between HMM and actin. When actin and Si (or HMM) were mixed together in 25% glycerol-standard salts solution, ‘arrowheads’ seen in negatively stained preparations were identical to those seen when actin and HMM were mixed in standard salts solution (Fig.1 A). Addition of Triton to the 25% glycerol did not seem to interfere with this reaction, in agreement with previous work (e.g. Mooseker & Tilney, 1975; Tawada, Yoshida & Morita, 1976), up to the highest concentration of Triton-X-100 tested, which was 0.75% (Fig. IB). Thus, the use of 25% glycerol-standard salts solution with added Triton and HMM would seem to be suitable for identification of actin in situ.

Fibrin-embedded endosperm cells were studied during glycerination, using phasecontrast microscopy. In preliminary experiments, endosperm cells were placed in a glycerol-containing solution, and then cells were observed 10–15 min later, using phase-contrast microscopy. These experiments together with electron-microscopic observations described below indicated that cells were not made permeable by 25% glycerol when the concentration of Triton was less than 0.02%. Hence, phasecontrast microscopic observations of living cells during glycerination were made on cells glycerinated using 25% glycerol with ⩾0.02% Triton; we also studied cells glycerinated using 50% glycerol, which, from previous experience (Forer & Jackson, 1976), we knew made Haemanthus endosperm permeable, and we studied cells glycerinated using 6% glycerol.

We conclude from studying cells during glycerination that glycerination initially causes cells to shrink rapidly ; after some time the cells expand again, to near original size (though generally to less than the original size). We also conclude that the relative positions of the chromosomes are maintained during glycerination. These conclusions are discussed in turn.

Cell shrinkage, as studied in detail in 23 cells, usually occurred rapidly and minimum dimensions were generally reached within 30 s-2 min after addition of glycerol. [Of the 23 cells in question, 2 were glycerinated with 50% glycerol; 3 with 25% glycerol, 0.05% Triton; 2 with 25% glycerol, 0.3% Triton; 3 with 25% glycerol, °’5% Triton ; 3 with 25% glycerol, 0.75% Triton ; 5 with 6% glycerol, 0.25% Triton ; 1 with 6% glycerol, 0.3% Triton; 2 with 6% glycerol, 0-5% Triton; and 2 with 6% glycerol, 0.9% Triton.] The extent of the shrinkage was estimated in 2 ways: (a) ciné films were projected onto paper, tracings of cell shapes were made, and then the tracings were cut out and weighed; or (b) we measured the linear dimensions of the cells along the spindle pole-to-pole axis. In different treatments cells shrank to 4093% °f their original area, or 44–93% of their original length. The cells expanded after variable lengths of time, ranging from 1 to 30 min, reaching sizes of 60–100% of the original areas, or 70-100% of the original lengths. When cells expanded, there was a sudden, rapid swelling, and sometimes this was preceded by a gradual increase in dimensions (Fig. 2). There was a great deal of variation in results, even with the same treatment, so from the quantitative data on shrinkage and recovery we cannot confirm or deny our general subjective impression that addition of Triton reduces distortions in cells due to glycerination. We are able to draw 2 conclusions: (1) that cells treated with 6% glycerol (with Triton) shrink less and expand to nearer the original size than cells treated with 25% glycerol (with Triton) or with 50% glycerol, and (2) that treatment with 50% glycerol seems to be harsher than treatment with 25% glycerol (with Triton) or 6% glycerol (with Triton), because cells treated with 50% glycerol generally lost all internal contrast when they shrank (Fig. 3): this rarely occurred during treatment with 25% glycerol or with 6% glycerol. Cells treated with 6% glucose containing 0.3% Triton shrank slightly, and then, after 2–3 min, expanded and disintegrated, so that no cell structures were identifiable at all. Hence, 6% glucose is not suitable for stabilizing cells.

Chromosome positions were studied carefully in 6 favourable cells (in metaphase and anaphase): in all cells the relative chromosome positions seen after shrinkage and expansion were similar to those seen before glycerination (e.g. Figs. 3-6). The cells were not flat enough, however, to analyse the positions of more than a fraction of the chromosomes in each cell. Other cells were studied in less detail than the 6; in these other cells the only large changes which occurred during glycerination were due to decreases in the interkinetochore distances (in anaphase cells). As the chromosome groups moved closer together the chromosome arms pushed against each other and, as a result, after glycerination the chromosome arms extended outwards from the spindle, whereas before glycerination they were parallel to the spindle axis. In all the cells studied, after shrinkage and the sudden expansion the chromosomes were clearer and appeared in greater contrast than before glycerination (e.g. Figs. 3-6).

We conclude from the phase-contrast observations of cells during glycerination that whilst cell shrinkage does occur, general spindle structure would seem to remain intact, because chromosome positions are maintained. This conclusion was confirmed by study of glycerinated cells using polarization microscopy: cells were glycerinated either with 50% glycerol, or with 25% glycerol plus 0.3% Triton. In both cases, cells were observed as early as 1.5 min after glycerination (Fig. 7). All spindles were biréfringent, from the earliest time after glycerination (1.5 min) to 24 b after glycerination (Fig. 8 B). Spindles were also biréfringent when HMM was included in the glycerination medium (Fig. 8A). Thus spindle birefringence remains through glycerination, confirming that at least some spindle organization is maintained during glycerination.

In order to see if individual chromosomal spindle fibres remained biréfringent during the glycerination procedure we studied crane-fly spermatocytes; in these cells individual chromosomal spindle fibres can be clearly seen. We studied spermatocytes during glycerination with 50% glycerol: individual chromosomal spindle fibres remain biréfringent during glycerination (Fig. 9). Thus, in summary, some shrinkage occurs during glycerination, and it is possible that some ‘relocation’ might occur, but chromosomal spindle fibres remain intact, as evidenced by polarization-microscopic observations, and general spindle organization seems to remain intact, as evidenced by chromosomal positions after glycerination being similar to those before glycerination.

As adjudged using phase-contrast microscopy, the best procedure for glycerinating cells seemed to be to use 6% glycerol (containing 0.25% Triton). Such cells were unsuitable as studied electron microscopically: in 3 cells, spindle microtubules were either completely absent, or were largely absent, and the few remaining microtubules were hard to see clearly because of adhering electron-dense material. When microtubules were completely absent, electron-dense material was seen which was more-or-less linearly arranged (Fig. 10). Furthermore, actin-containing filaments were not seen, and the chromosomes and kinetochores appeared distorted compared to those in not-treated, glutaraldehyde-fixed cells. In not-treated cells the kinetochores appear as ‘grey’ balls inserted in ‘black’ chromosomal sockets (e.g. Bajer, 1968a, b), whereas in cells glycerinated with 6% glycerol (containing 0.25% Triton) the kinetochores and chromosomes were equally black (Fig. 10).

The next-best procedure for glycerinating cells (as adjudged using phase-contrast microscopy) seemed to be to use 25% glycerol: these cells were suitable for electron-microscopic study of actin and of microtubules when sufficient Triton was included in the glycerination medium to make the cells permeable to HMM. Without Triton the cells seem to remain impermeable to HMM, for no HMM-actin complexes (decorated filaments) were seen; indeed, in one cell the cytoplasmic membrane seemed to remain completely intact and macrotubules, 35 nm in diameter, were seen in the spindle together with microtubules (Fig. 11). With 0.01% Triton, too, the cells seemed to remain impermeable to HMM. When Triton was added to the 25% glycerol in concentrations between 0 05 and 1.0% then we saw both microtubules and actin-HMM (decorated) filaments in spindles. When the Triton concentration was in the range of 0.05–1.0% there was little to choose from in electron-microscopical appearance of the sectioned cells, and these cells looked identical to those glycerinated with 50% glycerol. Hence the descriptions which follow pertain to all treatments with 25% glycerol and to treatment with 50% glycerol. The only differences noted between different treatments is that at higher concentrations of Triton (i.e. > 0.5%) the microtubules often appeared extremely wavy and of low contrast, and it was difficult to see individual actin filaments clearly. These conclusions must remain somewhat tentative, however, because there was some variability in our procedure which we have been unable to control or to identify: nearly half the cells sectioned were not usable for detailed study of serial sections because electron-dense material obscured many of the microtubules and the actin filaments. Our descriptions therefore are based on detailed study of those cells in which microtubules were seen clearly.

Kinetochore structure after glycerination is similar to that seen in non-treated cells: kinetochores consist of ‘balls’, ∼ 0.7 μ m in diameter, found in recesses in the more electron-dense chromosomes (‘cups’ or ‘sockets’) (see Bajer, 1968a, b;,Bajer & Molé-Bajer, 1971). The kinetochore cup appears as a mixture of less-dense and more-dense components (Fig. 12 A), as in non-treated cells (Bajer, 1968 a, b;Bajer & Molé-Bajer, 1971). Microtubules (and actin-containing filaments) terminate in the kinetochore (e.g. Fig. 12 A), as we describe in detail in the subsequent paper (Forer, Jackson & Engberg, 1979).

Spindle polyribosomes are seen in some cells (review in Fuge, 1977), including Haemanthus endosperm (Bajer & Molé-Bajer, 1969). In electron micrographs of Haemanthus endosperm spindles one sometimes sees helical polyribosomes (e.g. see fig. 14 of Bajer & Molé-Bajer, 1969). The glycerination procedure we used did not disrupt the helical arrangement of polyribosomes seen in the spindle region: helical polyribosomes were often seen in glycerinated, HMM-treated Haemanthus endosperm spindles, oriented in various directions with respect to the spindle microtubules (Fig. 13).

Actin-containing filaments are seen in varied locations in spindles. Some single actin-containing filaments are associated with (i.e. within 50 nm of) kinetochore microtubules (i.e. those which attach to kinetochores) and others are associated with non-kinetochore microtubules (i.e. those which do not attach to kinetochores; see Jensen & Bajer, 1973); in both cases the filaments are seen both in amongst the microtubules in the bundles and on the outside of the bundles. Groups of actin-containing filaments (usually 3–4 filaments) are also associated with kinetochore microtubules and non-kinetochore microtubules. Most of the actin-containing filaments in Haemanthus endosperm spindles are single, rather than in groups, and the actin-containing filaments associated with microtubules are generally oriented parallel to the microtubules. Some actin-containing filaments (both single filaments and groups of filaments), however, are seen in regions of the spindle in which no microtubules are seen. These are oriented either parallel to the spindle pole-to-pole axis, or at varying angles to this axis, and some were seen that were perpendicular to the pole-to-pole axis, even in the half-spindle region (between chromosomes and poles).

Some actin-containing filaments were seen outside the spindle, but there were relatively very few of these compared to those seen in the spindle region: our impression is that the number of actin-containing filaments outside the spindle would be less than 1% of those in the spindle. Those at the cell periphery were generally arranged parallel to what one would guess would be the cell surface and these were generally not single filaments but rather groups of 3–4 filaments. These might be the remnants of actin-containing filaments which originally formed a layer around the cell periphery (i.e. a ‘cortex’), but we certainly cannot conclude this from our data.

If these are the remnants of a cortex, we would guess that the rest of the cortical actin-containing filaments (if they existed) went outwards from the cell, as the cell re-expanded after glycerination, because similar groups of filaments commonly were seen extending outwards from the peripheries of the cells in question, at large angles and even perpendicular to what one would guess would be the cell surface.

We consistently see associations between actin-containing filaments and spindle microtubules, similar to those described by Schloss, Milsted & Goldman (1977) in PtKi cell spindles. In fortuitous sections one sees actin-containing filaments very close to microtubules, parallel to the microtubules (Figs. 12B, 14, 15 c, D); indeed, the actin-containing filaments are difficult to detect in amongst microtubules, and we have required much practice in order to do so. For sections cut at different angles relative to the microtubules and actin-containing filaments, the filaments can be obscured by the microtubules: in many cases filaments are seen only when the microtubules leave the plane of the section, and, because of their close association with the microtubules, the actin-containing filaments often appear to extend from the end of the microtubules as the microtubules leave the plane of the section (Figs. 12B, 14c, D, 15 B, c). That the microtubules are indeed obscuring the actin-containing filaments in such cases is substantiated by studying serial sections: when actin-containing filaments appear to extend from the ends of microtubules as the microtubules leave the plane of section, in the adjacent section the microtubules enter the section at the same place they left the previous section, and actin-containing filaments appear to extend from opposite ends of the same microtubules, as the microtubules leave the plane of this next section. The same thing occurs in subsequent sections (e.g. Fig. 16).

There are other indications of association between actin-containing filaments and microtubules. For example, actin-containing filaments remain parallel to spindle microtubules even when the latter are gently curved (Figs. 12B, 15B, 16); in some cases, actin-containing filaments remain closely parallel to microtubules even through sharp curves (Fig. 15 A). In the particular case illustrated in Fig. 15 A, the actin-containing filament remains about 10 nm from the microtubule as the microtubule bends through a curve which is a somewhat distorted sine wave of amplitude about 70 nm and wavelength about 400 nm : this is nearly the same shape as the sharply curved microtubules described by Jensen & Bajer (1969) in non-treated cells (they described sinusoidal microtubules with wavelengths around 300 nm and amplitudes 25-50 nm). In summary, then, there seems to be close association between actin-containing filaments and microtubules in Haemanthus endosperm. Thus, because of the large numbers of spindle microtubules, one cannot see all the actin-containing filaments throughout the lengths of these filaments, and, as illustrated (Fig. 16), actin-containing filaments are often obscured by microtubules. Hence,-we are unable to identify all the actin-containing filaments present in the spindle, especially those found in microtubule bundles.

We have described general features of actin-containing filaments in Haemanthus endosperm cells, as studied electron-microscopically after glycerination using various glycerination procedures. To see if the glycerination procedures might have produced artefactual localizations, we made light-microscopic observations on cells undergoing glycerination: with all the glycerination methods the cells shrink upon initial exposure to glycerol and then, after a variable time, the cells expand again, to near their original size. We have been unsuccessful in finding a glycerination method in which shrinkage does not occur; but that such changes do occur, might emphasize the cautions of others (Porter, 1973; LaFountain, 1974, 1975; Nicklas, 1975; reviewed in Forer, 1978 a) that actin-containing filaments found in the spindle might have moved into the spindle during treatment with glycerol and HMM. Only by some method such as pre-fixing cells with glutaraldehyde prior to adding HMM (Ohtsuki, Manzi, Palade & Jamieson, 1978) might one be able to overcome the problem of relocation during the procedure, but even with such treatment there is solubilization (and hence movement) of a certain amount of cytoplasmic protein (Ohtsuki et al. 1978).

Even with the changes in cell sizes during glycerination that we have described, however, this does not necessarily mean that actin-containing filaments do relocate during the procedure; to consider the opposite case, if we found a procedure in which cells remained of constant size, with no visible changes during glycerination, this would not necessarily mean that actin-containing filaments do not relocate during the procedure. The difficulty is that with present techniques one cannot look directly at actin-containing filaments in cells in vivo [except, of course, for arrays such as the ‘stress fibres’ described in other cells (review in Goldman, Yema & Schloss, 1977)]. Hence one cannot determine from any such light-microscopical observations whether or not actin-containing filaments move out of the spindle or into the spindle during glycerination. But one can study the structure of the individual organelle in question during the glycerination.

Spindle structure seems to remain relatively intact during glycerination, despite the changes in cell size. Evidence for this in Haemanthus endosperm cells is that spindles are biréfringent after extended glycerination (Figs. 7, 8 B); that biréfringent spindle fibres are seen 1.5 min after the start of glycerination, at a time at which, in phase-contrast microscopic observations (Fig. 3), the cells are of minimum size and of little internal contrast; that biréfringent spindle fibres are present when HMM is included in the glycerination medium (Fig. 8A); and that after the cells shrink and expand the relative chromosome positions remain the same as before glycerination (Figs. 3-6). Further, individual chromosomal spindle fibres remain birefringent during glycerination of crane-fly spermatocytes (Fig. 9). Finally as studied electron-microscopically, general features of spindle organization in glutaraldehyde-fixed nonglycerinated Haemanthus endosperm cells are seen in glycerinated Haemanthus endosperm cells. Some such features are that spindles contain both kinetochore microtubules and non-kinetochore microtubules; that several microns from the kinetochores the kinetochore microtubule bundles ‘intermingle’ with sheet-like arrays of non-kinetochore microtubules (Forer et al. 1979), as described in non-treated cells by Bajer (1968 a, b) and Jensen & Bajer (1973); that the bundles of non-kinetochore microtubules often branch (Forer et al. 1979), also as described in untreated cells by Bajer (1968a, b) and Jensen & Bajer (1973); and, finally, that helical polyribosomes seen in spindles in untreated Haemanthus endosperm are also seen in spindles in glycerinated Haemanthus endosperm (Fig. 13), and hence the glycerination did not disrupt the helical nature of the polyribosomes. In sum, these data suggest strongly that spindle organization in general is maintained during glycerination, and that spindle fibres in particular remain structurally intact during the glycerination procedure.

Data on the arrangements of actin-containing filaments relative to the bundles of kinetochore microtubules are relevant to the question of whether actin-containing filaments found in spindles have relocated during glycerination; these data and discussion of that question are presented in the subsequent paper (Forer et al. 1979).

Our data suggest that actin-containing filaments interact with microtubules. The evidence that suggests this is descriptive, in that actin-containing filaments are often within 10–20 nm of microtubules, are parallel to spindle microtubules, and are so closely apposed to microtubules that the microtubules obscure the actin-containing filaments: the actin-containing filaments are often visible only after the microtubules leave the plane of the section. In considering possible interactions between actin-containing filaments and microtubules, it is relevant to consider the possible counterargument that actin moves into the spindle during the glycerination procedure and is aligned by the microtubules, much as any long molecule would be aligned by flowing into a system of aligned rods. Evidence against this counter-argument is: (i) very little extra-spindle actin is seen in Haemanthus endosperm (Forer & Jackson, 1976, and this report), so there is no known source of extra-spindle actin. (2) Even were there extra-spindle actin, such a flow of actin-containing filaments might produce alignment of filaments, more-or-less in the direction of the spindle axis, but it would not produce microfilaments that are closely parallel to microtubules (Figs. 12, 14–16), and that even appear as extensions of microtubules when the microtubules leave the plane of the section (Figs. 14-16). This argument has even more force when one considers microfilaments parallel to microtubules that bend gently (Figs. 12, 15B, 16) and especially microfilaments parallel to microtubules that bend sharply (Fig. 15A): such arrangements certainly imply that there are strong interactions between microtubules and microfilaments. (3) Similar arrangements were seen by Schloss et al. (1977) in spindles in glycerinated HMM-treated PtKi cells, and, perhaps even more importantly, in spindles in non-glycerinated glutaraldehyde-fixed cells as well; Schloss et al. (1977) also suggested that microfilaments interacted with microtubules. In sum, then, our data strongly suggest that actin-containing filaments interact with spindle microtubules.

Another argument for actin-microtubule interactions was presented in Forer (1974). This argument essentially is that, in some cases, HMM seems to affect microtubules; since it seemed unlikely that HMM would affect microtubules directly, it was argued that actin probably interacted with microtubules and HMM probably affected microtubules via interactions with actin. Recent data have probably made this argument less compelling: Ishiura et al. (1977) showed that HMM subfragment 1 (S1) interacts with tubulin (S1 partially inhibits tubulin polymerization), and binds somewhat to polymerized microtubules. Such data suggest that HMM could interact directly with microtubules, but Ishiura et al. (1977) did not present data to rule out the possibility that trace actin might be present in their tubulin preparation, so the case is not yet unequivocal.

Other data might also suggest that actin interacts with microtubules in the spindle. Oakley & Heath (1978) described 5-nm filaments in not-treated Cryptomonas spindles which seemed to interact with spindle microtubules, and they suggested that such filaments might interact with spindle microtubules to produce force. In experiments of Keller & Rebhun (1978), mitotic apparatus (MA) were isolated from sea-urchin zygotes and then placed in the cold : microtubules disappeared from the MA when the MA were placed in the cold, and both tubulin and actin were found in the supernatant after the remains of the MA were centrifuged into a pellet (Keller & Rebhun, 1978). This experiment suggests, perhaps, that actin interacts with microtubules in the MA but leaves the MA when it is no longer held in place by microtubules.

Some experiments on brain microtubules in vitro also suggest that actin interacts with microtubules. Microtubules purified from brain as such (i.e. without depolymerization) contain co-purifying actin even after purification by centrifugation in a step gradient (Mann, Griesel, Fasold & Haase, 1978); when these microtubules are depolymerized and centrifuged at 100000 g for 1 h, which would pellet actin filaments, the actin is absent from the supernatant, suggesting, perhaps, that actin filaments interact with microtubules but are released from the microtubules when the microtubules are depolymerized. Griffith & Pollard (1978) present direct evidence from viscometry and electron microscopy that actin filaments indeed interact with microtubules in vitro, and that, in their experiments, the interaction seems to be mediated by microtubule-associated proteins. In sum, then, various lines of evidence suggest that in spindles and elsewhere there is interaction between microtubules and actin filaments.

We have argued that because actin-containing filaments are parallel to and close to microtubules, the microtubules obscure the microfilaments, and we have illustrated one such case (Fig. 16). As suggested by Schloss et al. (1977), this might be one reason why microfilaments are rarely seen in untreated spindles (review in Forer, 1978 a); that is to say, microfilaments are indeed present, but are obscured by the large numbers of microtubules also present.

We have used a one-step procedure for reacting HMM with actin-containing filaments in situ: the HMM was included in 25% glycerol, and Triton-X-100 was added to make the cell permeable. This avoids changing solutions (e.g. from 50% glycerol to standard salts solution, as in the more usual procedure), and avoids possible deleterious effects of centrifugation or of removing the glycerol. Such a procedure can also be quite rapid (15 min between glycerol-HMM-Triton treatment and fixation with glutaraldehyde) and hence may be of general use.

We have studied clot-embedded living cells as they were being glycerinated. Embedding cells in a clot, to keep them fixed in place whilst the surrounding medium is exchanged, seems to us to be easier than placing cells between agar and agargelatin (Molé-Bajer & Bajer, 1963); this clot procedure, used originally (Forer, 1972) to keep crane-fly spermatocytes fixed in place, might be of general usefulness, especially if one could determine how to obtain flattened cells routinely in a clot. In any event, it might be relevant to point out that while we studied the effects of glycerol on cells held in place on a coverslip by means of a fibrin clot, cells glycerinated in suspension might respond differently from clot-embedded cells. That is to say, it is conceivable that the clot and coverglass act as restraints on the cell during glycerination, and, for example, that some of the variability we observed might be due to different cells being constrained to different degrees by the clot and coverglass.

We acknowledge the skilful technical assistance of Alexandra Engberg, who did all the serial sectioning, and some of the photographing of the sections (using the electron microscope), and of Barbara G. Doyle, who did film analysis and preparation of endosperm cells. This work was supported by grants from the National Research Council of Canada (to A. F.), and from the U.S. National Science Foundation (to W. T. J.).