Function of spindle microtubules in directing cortical movement and actin filament organization in dividing cultured cells

Cell division represents the final phase of the cell cycle, when chromosomes are segregated during mitosis and the cell is partitioned into two daughter cells during cytokinesis. This highly integrated process involves both the regulated assembly-disassembly of microtubules for chromosome movements (see Salmon, 1989a; Wadsworth, 1993; McIntosh, 1994, for reviews), and the dynamic reorganization of cortical actin filaments for cell cleavage (reviewed by Mabuchi, 1986; Salmon, 1989b; Conrad and Schroeder, 1990; Satterwhite and Pollard, 1992; Fukui, 1993; Fishkind and Wang, 1995).

While mitosis and cytokinesis are often treated as distinct events, classic experiments of Rappaport (1986, 1991), Hiramoto (1956, 1971), and others (Wilson, 1925; Dan, 1948; Swann and Mitchison, 1958) have clearly indicated that the mitotic spindle plays a key role in controlling cytokinesis. For example, cytokinesis is inhibited by either removal (Hiramoto, 1956, 1971) or disruption of the mitotic spindle during metaphase and early anaphase (Chambers, 1938; Beams and Evans, 1940; Swann and Mitchison, 1953; Hamaguchi, 1975), or by the presence of physical barriers placed between the spindle and equatorial cortex (Rappaport, 1986, 1991; Cao and Wang, 1996). Moreover, when the mitotic spindle is moved by micromanipulation, the cleavage furrow regresses and then reinitiates at a new position, directly coincident with the midzone of the repositioned spindle (Rappaport, 1985).

Beyond the importance of the spindle in signaling the cortex and determining the position of the cleavage furrow, relatively little is known about the mechanisms that control this process. Results from micromanipulation experiments first suggested that diffusible signals emanating from spindle asters might function as a stimulus for the induction of cleavage (Rappaport, 1986, 1991). While this hypothesis can explain a number of observations with echinoderm embryos, neither the nature of the signal nor the coupling of the stimulation to the cortical actin machinery has yet to be elucidated. In addition, recent observations suggest that the model may not be fully applicable to all animal cells (Cao and Wang, 1996). A second idea for spindle induction of furrowing proposes that the microtubule and actin system in dividing cells may interact with each other, thus providing a direct mechanism for the bipolar spindle apparatus to establish or guide cortical dynamics (Dan, 1948; White and Borisy, 1983; Mabuchi, 1986; Rappaport, 1986, 1991; Bray and White, 1988). While both ideas are plausible and not mutually exclusive, there is presently no direct evidence demonstrating a role for microtubules or other spindle structures in regulating cortical movement or actin filament reorganization in dividing cells.

To determine if microtubules can modulate cortical structure and cell cleavage, we have designed a series of experiments to examine cortical dynamics and actin filament organization in dividing cultured cells following drug-induced disruptions of spindle microtubules. This study was aided by the recent development of several light microscopy techniques that facilitate the acquisition of structural information at the molecular level. Specifically, by utilizing single particle tracking techniques, we have been able to directly follow the movement of cell surface receptors driven by the underlying cortex (Wang et al., 1994). In addition, using fluorescence-detected linear dichroism and digital optical sectioning microscopy, we can determine the predominant orientation and three-dimensional organization of actin filaments in different regions of dividing cells (Fishkind and Wang, 1993).

Our results show that both cortical movements and the corresponding alignment of actin filaments along the spindle axis are severely inhibited by the depolymerization of microtubules during early anaphase. In contrast, stabilization of anaphase microtubules by taxol induces profound changes in the pattern of cell surface movements, coincident with a dramatic reorganization of cortical actin filaments on the dorsal cortex. Together, these results provide direct evidence that spindle microtubules can regulate the activity and organization of actin containing structures in dividing cells, and hence serve an important function in the assembly and contraction of the cleavage furrow.

Normal rat kidney (NRK) cells from the well-spread subclone NRK-2 (Fishkind and Wang, 1993) were cultured and maintained in plastic Petri dishes using Kaighn’s modified F12 medium (F12K, JRH Biosciences, Lenexa, KS) supplemented with 10% fetal bovine serum (JRH Biosciences, Lenexa, KS), 1 mM L-glutamine, 50 µg/ml streptomycin, 50 units/ml penicillin, at 37°C and 5% CO₂. Prior to experiments, cells were plated onto alcohol-dipped, flamed glass coverslips set into 35 mm dishes or secured to microinjection chambers (McKenna and Wang, 1989), and cultured for 48 to 72 hours. Many cells in this subclone maintain their association with neighboring cells during cytokinesis and do not show appreciable degrees of lateral constriction. This facilitates optical analyses of surface receptor movement and cortical actin organization (Fishkind and Wang, 1993; Wang et al., 1994).

Pilot experiments were performed to determine the dose required to disrupt mitotic processes and to cause microtubule disassembly. Cells in metaphase were carefully monitored until anaphase onset, at which time the culture medium in the chamber was rapidly exchanged with medium containing 1-10 µM nocodazole (Sigma Chemical Co., St Louis, MO) or 1-10 µM taxol (National Cancer Institute, Bethesda, MD). Media with 1% dimethyl sulfoxide (DMSO; Sigma) or 1-10 µM baccatin III (an analog of taxol that has no direct effects on microtubules; Manfredi and Horwitz, 1984) were used as controls. Parallel observations were performed with phase optics to assess the degree of inhibition of mitosis and with immunofluorescence staining to determine the extent of microtubule disruption (see below).

Fluorescent beads for single-particle tracking experiments were prepared according to the method of Wang et al. (1994). Briefly, unmodified fluorescent carboxylated beads (L-5221, Molecular Probes, Eugene, OR) were sonicated, washed, and resuspended in phosphate-buffered saline (PBS) containing 10 mg/ml bovine serum albumin (BSA; Sigma, St Louis, MO). Cells were labeled by carefully drawing off the culture medium, briefly rinsing with two exchanges of warm Hanks’ balanced salts, and applying a dilute, sonicated suspension of beads onto the dish. Following a 2-3 minute incubation, the dish was rinsed with several changes of culture medium. Cells were maintained on a warm microscope stage incubator during the labeling and subsequent observation period (McKenna and Wang, 1989).

Metaphase cells labeled with beads were monitored with phase optics until anaphase onset, and then imaged with fluorescence optics every 15 seconds. Cells that divided without extensive lateral constriction during the period of observation were chosen for analysis, in order to simplify the interpretation of bead movement. In most experiments cells were co-illuminated for simultaneous fluorescence and phase-contrast observations. To study the effects of microtubule drugs on cell surface movements, cells were labeled with beads as above and incubated until shortly after anaphase onset, when directional bead movement had occurred for 1-1.5 minutes. The medium was then rapidly replaced with complete F12K containing either 3.3 µM nocodazole, 5 µM taxol, 1% DMSO, or 1% DMSO with 5 µM baccatin III, and further imaged for the duration of the experiment. The drug concentrations of nocodazole and taxol represent the lowest doses required to provide optimal effects on the spindle microtubules of NRK-2 cells as determined from the analysis described above. Analyses of particle movement, including the determination of directionality and speed, were carried out as previously described by Wang et al. (1994), using a combination of movie loops and frame-to-frame particle tracking of single beads.

Cells were fixed using the glutaraldehyde-based method of Small et al. (1981) as previously described (Fishkind and Wang, 1993). Autofluorescence and free-aldehydes were quenched with 0.1% NaBH₄ for 5 minutes, and additional non-specific binding was blocked for 1 hour in PBS/1%BSA. Cells were then stained with a 1:200 dilution of anti-β-tubulin monoclonal antibodies (N357, Amersham Life Sciences, Arlington Heights, IL) for 12-16 hours at 4°C, followed by 3 brief (10 minute) washes in PBS/1%BSA, and incubation with a 1:200 dilution of fluorescein labeled sheep anti-mouse secondary antibody (Sigma) for 2 hours at 37°C. After several PBS washes, cells were stained with 200 nM rhodamine phalloidin (Molecular Probes, Eugene, OR) diluted in PBS for 1 hour, and then rinsed in PBS for an additional 30 minutes before examination. For digital optical sectioning analysis, coverslips were mounted in the anti-photobleach medium of Clark and Meyer (1992) prior to examination.

Measurements on actin filament orientation using FDLD were performed as described by Fishkind and Wang (1993), except that data were acquired with a cooled CCD camera with an EEV chip of 576×384 pixels (Princeton Instruments, Inc., Trenton, NJ). The image was focused on or near the ventral cortex. FDLD values were measured as before by averaging pixel values within a 2.5 µm diameter spot (5 µm² area), positioned either at the center of the cell (i.e. at the equator) or in the subequatorial zone. Cells were analyzed at various stages of mitosis from metaphase to telophase. Measurements from at least 10 different cells per stage were averaged to obtain the mean and standard deviation of the FDLD value for a given region.

The three-dimensional organization of spindle microtubules and actin filaments in dividing cells were examined using digital optical sectioning microscopy as previously described (Fishkind and Wang, 1993). Briefly, a through focus series of optical sections (0.28 µm/section) were acquired from fluorescently stained samples, and out-of-focus fluorescence was removed using computational methods based on the nearest neighbor algorithm (Castleman, 1979; Agard, 1984; Shaw and Rawlings, 1991) and point spread functions derived with our optics. Using custom-based software, stereo pairs were produced from partial or complete stacks of deconvolved sections by projecting the images at angles +10° and-10° from the optical axis. Analysis of data sets were performed with a customized, interactive, motion display program that allows rotation and stereo viewing of reconstructed three-dimensional images.

All experiments were performed on the well-spread subcloned NRK-2 cells to facilitate observations of mitotic and cortical structures. During mitosis, these cells remain attached to the substratum and neighboring cells, such that the ventral surface shows little or no cleavage while the dorsal surface cleaves actively (Fishkind and Wang, 1993). To test the response of NRK-2 spindle microtubules to drug treatments, we applied nocodazole (an agent known to depolymerize microtubules; DeBrabander et al., 1976) and taxol (a drug known to stabilize microtubules and affect the assembly kinetics of tubulin; Schiff and Horwitz, 1980; Wilson and Jordan, 1994) after anaphase onset and evaluated their effects on microtubule structure and spindle function.

In control NRK-2 cells, the early anaphase spindle is composed of two robust asters interconnected by a dense interzonal array of microtubules (Fig. 1a). Upon application of micromolar levels of nocodazole in early anaphase, spindle microtubules rapidly disassemble within 5 minutes (DeBrabander et al., 1986; see Fig. 4c below), coincident with immediate cessation of chromosome separation (Mullins and Snyder, 1981; Fig. 2d-f below). In cells treated with taxol, three-dimensional images of early anaphase microtubules showed attenuated spindles consisting of two compact half spindles of short microtubule bundles emanating from the poles (Fig. 1b). Cells treated with taxol later in anaphase and telophase contained elongated microtubules similar to controls, however the interzonal region showed a number of bundled microtubules (Fig. 1c). In all taxol-treated cells, microtubules appear to converge along the equatorial region, with a pronounced gap of staining between opposing fibers (Figs 1b,c, 5a; Amin-Hanjani and Wadsworth, 1991; Snyder and Mullins, 1993). Note, that the perturbation of microtubules in wellspread NRK-2 cells early in anaphase led to either complete inhibition of cytokinesis as shown by the eventual production of binucleated cells for nocodazole (Mullins and Snyder, 1981; DeBrabander et al., 1986), or delays in the progression of cell division as previously observed for taxol (Amin-Hanjani and Wadsworth, 1991; Snyder and Mullins, 1993).

Given the ability to rapidly manipulate spindle microtubules with nocodazole and taxol, we asked whether these drugs could alter cortical dynamics in dividing cells. Movements of surface receptors on living NRK-2 cells were examined with a single particle tracking assay (Fig. 2), which allows direct observations and measurements of the movement of cell surface receptors associated with the underlying cortical actin cytoskeleton (Wang et al., 1994). As shown previously, surface-bound beads on control cells showed no organized movement during metaphase, but began directional movement (0.68±0.34 µm/minute) toward the equator ∼1 minute after anaphase onset (Fig. 2a-c). These movements were most active over the central region of the spindle where well-aligned microtubules fill the midzone (Fig. 1a). Later in anaphase and telophase, a slight converging pattern of movement caused the concentration of beads toward the center of the cell (Fig. 2c).

In contrast, cells treated with nocodazole 1-2 minutes following anaphase onset showed a rapid inhibition of directional bead movement, simultaneous with the cessation of chromosome separation (Fig. 2d-f) and depolymerization of microtubules. As a result, few beads accumulated at the equator (Fig. 2f). While taxol treatment caused only a slight decrease in the rate of directional bead movement (0.48±0.19 µm/minute, n=51 beads from 8 different cells), there were pronounced changes observed in the pattern of particle movements along the cortex (Fig. 2g-i). Instead of moving more or less directly toward the equator, beads tended to deplete first from the region overlying the spindle, and then began active convergence toward the center of the cell (Fig. 2h,i). The initial zones of depletion correlated well with the characteristic pattern of spindle microtubules (Fig. 1b), while the subsequent converging movement in the midzone was directed toward the focused ends of microtubules at the equator (Fig. 1c).

To determine if microtubule-dependent effects on particle movements reflect underlying changes in the structure of the contractile apparatus, we examined the three-dimensional organization of actin filaments in glutaraldehyde-fixed cells using digital optical sectioning microscopy. Stereo pairs of spindles were also prepared to allow direct comparison of actin and microtubule structures (Fig. 3). Previous studies of control cells indicated an essentially isotropic organization of actin filaments on the dorsal equatorial cortex, and equatorial alignment of actin filaments on the ventral equatorial cortex (Fig. 3b; Fishkind and Wang, 1993). In addition, actin filaments along the dorsal cortex in regions flanking the equator showed preferential orientation along the long axis of the cell (Fig. 3c), similar to alignment of interzonal microtubules in the spindle apparatus (Fig. 3a).

In nocodazole treated cells, the general pattern of actin organization in late anaphase/early telophase cells appeared strikingly different from that of control cells with a similar extent of chromosome separation (Figs 3 and 4). The most pronounced change in actin filament organization was an apparent reduction in the number of actin filaments oriented parallel to the long axis along the dorsal cortex and spindle region (Fig. 4a,b), an observation later confirmed quantitatively with FDLD imaging (see Fig. 6, below). On the ventral cortex along the equator, actin filaments retained an organization similar to that in control cells, however the band of filaments appeared broader and less compact (measured below, FDLD analysis). The rapid loss of anaphase spindle microtubules following nocodazole treatment was confirmed by anti-tubulin staining that showed diffuse fluorescence throughout the cell (Fig. 4c).

In taxol-treated cells, the most dramatic change was an increase in filament alignment along the equator of the dorsal cortex (Fig. 5). By late anaphase/early telophase, a striking band of actin filaments formed along the equator in the central region of the cell (Fig. 5b,c), where surface-bound particles were previously observed to converge and concentrate (Fig. 2i). These highly aligned equatorial filaments were localized in the gap between the ends of microtubule bundles emanating from the two half spindles (Fig. 5a). Many actin filaments appeared to extend from the outer lateral margins of the equator and subequatorial region toward the central region of the ingressing cleavage furrow (Fig. 5b). The pattern of actin organization parallels that of bead movement in taxol-treated cells (Fig. 2g-i), consistent with the idea that surface particle translocations are linked to movement of the underlying cytoskeleton (Wang et al., 1994), and that both are sensitive to modifications of spindle microtubules.

To measure quantitatively the effect of microtubule drugs on the organization of actin, we used FDLD microscopy to determine the preferred orientation of actin filaments (Fishkind and Wang, 1993). As previously shown, metaphase cells generally displayed little preferred orientation of actin filaments except for a slight longitudinal orientation in the equatorial region (Fig. 6a). Upon anaphase onset, actin organization underwent a progressive increase in equatorial alignment along the equator, and longitudinal alignment in subequatorial regions (Fig. 6d,g,j).

Following nocodazole treatment, anaphase cells maintained a preferred alignment of actin filaments along the equator similar to that of controls (Fig. 6b,e,h), however, little or no orientation was detected along the longitudinal direction in the subequatorial region (Fig. 6e, arrows and 6k, bar graph). This result suggests that disruption of microtubules may have prevented or reversed the longitudinal alignment of filaments while allowing equatorial organization to take place. Moreover, as observed with optical sectioning microscopy (Fig. 4), the equatorial band of filaments in late anaphase and telophase cells was slightly wider than that of controls (Fig. 6h, 7.0±2.0 µm versus 4.6±1.0 µm; n=23 and 30 cells, respectively). In taxol-treated cells, the overall pattern of filament organization as indicated by FDLD appeared similar to that in control cells (Fig. 6c,f,i). However, the positive FDLD value along the equator was slightly higher in taxol-treated cells than in controls during the early stage of anaphase (Fig. 6f). The apparent shift in the development of equatorial alignment may be due to the ability of cells to continue the organization of actin filaments following the addition of taxol, or possibly to slight differences in the mitotic staging relative to the separation of chromosomes (Fig. 6l).

The mitotic spindle has long been recognized to play a critical role in establishing the position of the cleavage furrow in dividing cells (Wilson, 1925; Swann and Mitchison, 1958; Rappaport, 1986, 1991). While many hypotheses have been proposed to explain how the spindle might function in delivering diffusible and/or mechanical signals to the cortex (Dan, 1948; Wolpert, 1960; White and Borisy, 1983; Rappaport, 1986, 1991; Devore et al., 1989; Margolis and Andreassen, 1993), little direct experimental evidence has been obtained to elucidate the role of spindle microtubules in regulating actin filament organization. In this study, we have utilized a variety of optical approaches to analyze the potential role of microtubules in controlling cortical dynamics and actin organization leading to the assembly of the contractile apparatus for cyto-kinesis. By combining singleparticle tracking, FDLD measurements, and three-dimensional imaging, together with the use of well-spread NRK-2 cells to facilitate the detection of structures, we have provided new evidence that spindle microtubules promote bipolar cortical flow, enhance actin filament orientation, and help maintain the structural organization of actin during cell division.

The initiation of cytokinesis in tissue culture cells is first detected upon the movement of cortical actin filaments (Cao and Wang, 1990) and associated membrane receptors (Wang et al., 1994) toward the equator following anaphase onset. This process occurs concomitant with well-documented changes in spindle structures, including the simultaneous shortening of kinetochore fibers and elongation of astral and interzonal microtubules (Salmon, 1989a; Wadsworth, 1993; McIntosh, 1994). During the same period of time, there is a dramatic realignment of actin filaments along the spindle axis in subequatorial regions (Figs 3 and 6), that directly correlates with cortical movements to the equator (Fig. 2a-c). The inhibition of both cortical and chromosome movements within 1-2 minutes of the application of nocodazole (Fig. 2d-f) indicates that cortical flow is highly sensitive to the disassembly of anaphase microtubules. At the same time, stereo images and FDLD measurements reveal a failure of actin filaments to align along the spindle axis (Figs 4, 6e,k), demonstrating that microtubules play a critical role in orienting actin filaments during the early phase of cytokinesis.

Further indications of the regulation of cortical movement by microtubules were obtained with taxol, an agent known to cause the stabilization of microtubules in vitro (Schiff et al., 1979; Wilson et al., 1985). When applied to actively dividing cells, taxol appeared to affect the dynamic properties of spindle microtubules, leading to changes in their length, degree of bundling, and three-dimensional organization (Figs 1b,c, 5a; Amin-Hanjani and Wadsworth, 1991; Snyder and Mullins, 1993; Wilson and Jordan, 1994). Interestingly, these effects were also accompanied by dramatic changes in the pattern of cortical movements (Fig. 2g-i), as well as the reorganization of actin filaments on the dorsal cortex of the cleavage furrow (Fig. 5b,c). The rapid clearance of beads from the region overlying the early spindle (Figs 1b,c, 2g,h), together with their strong convergence to the central region of the cell, suggests that taxol stabilized spindle microtubules possess an enhanced capacity to alter the path of cortical movements and the organization of actin filaments. Furthermore, the site of bead convergence contains a high concentration of converging cortical actin filaments, consistent with the idea that bead transport is driven by the movements of the underlying cortex. Most importantly, the region of convergence is characterized by the apparent termination of numerous interzonal microtubule bundles (Figs 2h,i, 5a), suggesting that molecular signals involved in generating cortical movements are localized along microtubules and likely concentrate at or near the ends of these bundles.

The most prominent actin-containing structures in dividing cells is the band of aligned filaments lying along the equator, commonly referred to as the contractile ring. While these filaments have been proposed as the main contractile element for cytokinesis (Schroeder, 1970, 1973; Sanger and Sanger, 1980; Sanger et al., 1994), recent studies indicate that they are more abundant in adherent cells than in round cells, and are concentrated primarily on the bottom, non-cleaving cortex (Fishkind and Wang, 1993). On the dorsal cortex of adherent cells, where active cleavage takes place, actin filaments show a largely isotropic organization, suggesting filament alignments are more likely a response to resistive forces such as cell-cell and cell-substratum interactions rather than a prerequisite for cell cleavage.

In this study we have observed that alteration or disruption of spindle microtubules can also cause pronounced effects on the equatorial alignment of actin filaments. In taxol treated cells there was a striking increase in tightly focused equatorial filaments along the dorsal cortex of the furrow (Fig. 5), while nocodazole-treated cells showed a broader, less well-organized arrangement of equatorial filaments (Figs 4 and 6). As discussed above, one explanation for strong filament alignments in taxol-treated cells may relate to increased concentration of signaling or motor molecules at the end of taxol-stabilized microtubules that drive cortical movements to the equator. Increases in convergent forces at the equator could then promote increases in actin filament alignment as first proposed by White and Borisy (1983). Alternatively, enhanced filament alignment could reflect increases in cortical resistive forces due to stabilized spindle structures (Mickey and Howard, 1995; Vallee, 1995). Similar explanations may account for the broadening of the equatorial actin filament band in nocodazole-treated cells. Hence, disassembly of microtubules could lead to disruptions in the distribution of cortical signals controlling bipolar cortical flow or a decrease in resistive forces (Hamilton and Snyder, 1983) that elicit structural orientation to actin filaments in the cortex.

Based on the dramatic effects of microtubules on cortical movements and actin filament organization during cell division, we propose that cytokinesis in tissue culture cells consists of at least two temporally overlapping phases, with the initial stage being regulated by spindle microtubules that control the longitudinal movement of cortical components into the cleavage furrow, and a second stage involving activation of an actin-myosin based equatorial contraction that serves to cleave the cell into two daughter cells. A third and final stage of cell cleavage in dividing tissue culture cells likely relies on proper formation of the mid-body involving additional activities such as interzonal microtubule motors (Yen et al., 1991; Margolis and Andreassen, 1993; Williams et al., 1995), γ-tubulin mediated microtubule assembly (Julian et al., 1993; Shu et al., 1995), and eventual disassembly of mid-body components and fusion of the plasma membrane (Mullins and Biesele, 1977; Rattner, 1992).

Although little is known about the nature of microtubulecortical interactions, our observations provide some important clues. The most obvious role for microtubules is to stimulate, guide, and drive active movement of surface receptors (Koppel et al., 1982; McCaig and Robinson, 1982) and cortical components (Cao and Wang, 1990; Yonemura et al., 1993; Wang et al., 1994) toward the equator during early anaphase, in order to properly position the assembly of the contractile apparatus. It is possible that the spindle interacts mechanically with the cortex through various accessory proteins such as microtubuleassociated proteins, motors, and other proteins that could serve to link microtubules, intermediate filaments, and actin filaments to the membrane (Pollard et al., 1984; Goslin et al., 1989; Sato et al., 1991; Wang et al., 1993; Katsumoto et al., 1993; Klymkowski, 1995). This idea is supported by recent findings implicating cortical interactions of the microtubules in the positioning and orientation of the spindle in yeast (Palmer et al., 1992; Li et al., 1993; McMillan and Tatchell, 1994) and dividing nematode embryos (Hyman and White, 1987; Hyman, 1989; Hird and White, 1993). Alternatively, microtubules may serve as guiding tracks for the delivery of diffusible signals, signaling organelles, or other organizational cues to the equatorial cortex (Wright et al., 1993; Waterman-Storer et al., 1993; Martineau et al., 1995). Given the large family of microtubule motors already implicated in spindle function (Skoufias and Scholey, 1993; McIntosh, 1994) and numerous signaling pathways that affect cytokinesis (Fishkind and Wang, 1995), the identification of molecules involved in transducing signals should be forthcoming.

While microtubules seem to play an important role in cortical dynamics in dividing tissue culture cells, their specific function in cytokinesis of cells in embryos, yeast, fungi, and slime molds remains less clear. Past ultrastructural work by Asnes and Schroeder (1979) on sea urchin eggs, showing an apparent low density of microtubules near the equatorial cortex during anaphase, first cast some doubt on the stimulatory role of microtubules in cortical events. However, more recent confocal studies have clearly shown high densities of microtubules extending out to the cortex during mitosis (White et al., 1987; Henson et al., 1989), reviving the possibility they may play a specific role in cortical dynamics during cell division in the sea urchin embryo (Dan, 1948, 1954; McCaig and Robinson, 1982). Similarly, the extensive array of microtubules lying parallel to advancing furrow canals in the cellularizing embryo may help function in establishing directional cues for actin-myosin based furrowing (Warn and Warn, 1986; Katoh and Ishikawa, 1989; Foe et al., 1993; Schejter and Wieschaus, 1993; Sullivan and Theurkauf, 1995). Future studies with these and other experimental systems should continue to provide us with new and important insights into the complex process of cytokinesis.

We thank Dr Sally Wheatley, the reviewers, and managing editor for helpful comments on the manuscript, as well as Dr Susan Horwitz for advice on the use of taxol and baccatin III. Additionally, we thank the Drug Synthesis and Chemistry Branch, Developmental Therapeutics Program, Division of Cancer Treatment, National Cancer Institute for supplying taxol and baccatin III, and Drs Richard Vallee and Howard Shpetner for test materials used in pilot studies. This work was supported by grants from the American Cancer Society PF-3758 (D.J.F.), National Institutes of Health GM32476 (Y.L.W.) and Human Frontier Science Program (Y.L.W.).