Neomycin reversibly disrupts mitotic progression in stamen hair cells of Tradescantia

Mitosis is the process that facilitates the equal partitioning of chromosomes. Studies on the mechanochemical basis of anaphase chromosome separation have recently begun to focus on the kinetochore (or the region very close to the kinetochore) as an active component in movement (Mitchison et al. 1986; Gorbsky et al. 1987; Nicklas, 1989), but irrespective of the identity of the motor, the physiological regulation of anaphase chromosome separation remains largely obscure (Wolniak, 1988). During the past few years, changes in cytosolic calcium activity have been postulated to be a major factor in the regulation of mitotic spindle function (Hepler and Wolniak, 1983; Wolniak, 1988; Hepler, 1989). It seems possible that the regulation of calcium activity itself, through some other signalling molecule, may have a direct bearing on events that take place within the spindle.

In a wide variety of animal and plant cells, cytosolic calcium activity has been shown to be regulated by the biochemical cycling of inositol lipids, in the so-called polyphosphoinositide cycle (Abdel-Latif, 1986; Sekar and Hokin, 1986; Berridge, 1987; Einsphar and Thompson, 1990). In cells in which the cycle has best been described an agonist binds to a cell surface receptor and stimulates phospholipase C. The lipase hydrolyzes phosphatidylinositol 4,5-bisphosphate, thereby producing the hydrophilic inositol 1,4,5-trisphosphate and a 1,2-diacylglycerol. The inositol 1,4,5-trisphosphate enters the cytosol, where it binds to an internal membrane (receptor) and induces the release of sequestered calcium (Abdel-Latif, 1986; Sekar and Hokin, 1986; Berridge, 1987). 1,2-Diacylglyercol associates with, and activates, protein kinase C (Bell, 1986; Nishizuka, 1986). The increases in cytosolic calcium and protein kinase C activities, in turn, induce cellular responses through calcium-dependent phosphorylation of specific proteins (Abdel-Latif, 1986; Sekar and Hokin, 1986; Nishizuka, 1986; Berridge, 1987).

Evidence for the possible involvement of polyphosphoinositide cycling in mitosis came originally from studies where lithium was used as an inhibitor of the polyphosphoinositide cycle. By blocking the hydrolysis of inositol 1-monophosphate, this cation has the potential to deplete the precursor pool for further elicitor production (Abdel-Latif, 1986; Sekar and Hokin, 1986; Berridge et al. 1989). Sillers and Forer (1985) first showed the inhibitory effect of this cation on mitosis in sea urchin zygotes. Reversal of the mitotic block was possible with the addition of myoinositol, but not with its stereo isomers, a result suggesting the involvement of polyphosphoinositide cycling in mitotic regulation (Sillers and Forer, 1985; Forer and Sillers, 1987). Recently, we reported timedependent, reversible changes in the rate of mitotic progression as a function of lithium treatment in Tradescantia virginiana (Wolniak, 1987). The addition of lithium to stamen hair cells in late prophase resulted in a block to progression into metaphase; nuclear envelope breakdown almost never occurred. When lithium was added at an earlier stage, during mid-prophase, the cells exhibited nuclear envelope breakdown and became arrested in metaphase. This block to progression could be reversed by post-treatment with myo-inositol (Wolniak, 1987), a result suggesting the involvement of polyphosphoinositide cycling in mitotic regulation. Post-treatment with calcium also was sufficient to reverse the metaphase block. The interval between calcium addition and anaphase onset was significantly shorter than that with myo-inositol, a finding that is consistent with the notion of a regulatory cascade involving the biochemical cycling of inositol lipids via calcium that ultimately leads to anaphase onset. Lithium addition after nuclear envelope breakdown resulted in an increase in the rate of progression through metaphase and anaphase. Since lithium indirectly inhibits the degradation of inositol 1-monophosphate and thereby affects 1,2-diacylglyerol levels, reduced hydrolysis of inositol 1-monophosphate may result in increases in 1,2-diacylglycerol pools (Sekar and Hokin, 1986; Berridge et al. 1989) that may account for precocious entry into anaphase. The multiple, seemingly opposite effects of lithium at different stages of mitosis suggest the existence of a series of similar or identical regulatory events that trigger different cellular responses at different times. Each cascade may involve many of the same elicitors, whose activities oscillate like a pendulum during the course of mitosis (Wolniak, 1987, 1988).

In this study, we have treated stamen hair cells with the aminoglycosidic antibiotic, neomycin, to test more directly whether the polyphosphoinositide signalling pathway is involved in mitotic regulation. Neomycin inhibits phosphatidylinositol 4,5-bisphosphate hydrolysis in a wide variety of eukaryotic cells (Schacht, 1976; Downes and Michell, 1981; Carney et al. 1985; Whitaker and Aitchison, 1985; Tysnes et al. 1987). We have found that the addition of neomycin to cells in prophase prevents the vast majority of them from either progressing through nuclear envelope breakdown or regressing to interphase without division. In cells that do proceed through nuclear envelope breakdown, we have found that neomycin forstalls or prevents anaphase onset as a function of the time of initial exposure to the drug and that the arrested cell remains in metaphase for many hours. Perfusion of cells arrested in metaphase with 1,2-dioctanoylglycerol induces anaphase more than half of the time while 1,3-dioctanoylglycerol is totally insufficient for reversal. Calcium treatment is also sufficient to reverse metaphase arrest in approximately half of the cells we observed, with reversal kinetics that are somewhat slower than those observed with 1,2-dioctanoylglycerol. We attribute this low efficiency of reversal to a loss of capacity for progression after arrest, to differences in the identities of polyphosphoinositide elicitors in animals and plants, and to nonspecific inhibitory effects of neomycin on the cells.

Neomycin sulfate, clindamycin, 1,2-dioctanoylglycerol, 1,3-dioctanoylglycerol and other chemicals were obtained from Sigma Chemical Co. (St Louis, MO). 1,2-Dioctanoylglycerol was also obtained from Molecular Probes, Inc. (Eugene, OR). Purified cutinase was kindly provided by Dr P. E. Kolattukudy (The Ohio State University, Columbus, OH).

Spiderwort plants (Tradescantia virginiana) were maintained m the greenhouse facilities at the University of Maryland at College Park, with an 18-h photoperiod as described previously (Wolniak and Bart, 1985a).

Stamen filaments were dissected from immature flower buds and immersed briefly in a 0.05% solution of Triton X-100 in 15 mM Hepes/15mM KC1 (pH 7.0) to wet the cell surface. To permeabilize the waxy cuticle that surrounds the cell wall, the stamen hairs were pretreated with cutinase (0.lmgml⁻¹) in 15 mM Hepes/15mM KC1 buffer (pH 8.0) for 45–60 min (Hepler, 1985; Chen and Wolniak, 1987,a). The enzyme treatments had no apparent effect on the rate of mitotic progression, but they greatly facilitated the exchange of solutions through the cell wall and thereby reduced variability in the responses observed (Hepler, 1985; Chen and Wolniak, 1987 a).

After the enzyme treatment, the hairs were dissected from stamen filaments in approximately 100 μl of the Hepes/KCl buffer (pH 7.0) on a glass microscope slide and covered with a coverglass. In these experiments, it was essential to know the exact stage of mitosis at the time of drug addition, so a cell at the appropriate stage was located with the microscope, and the preparation was then perfused with the 2 mM neomycin solution in the same buffer. In a typical experiment, a late prophase cell was followed through mitosis and the duration of metaphase and anaphase were determined. After 15–30 min in neomycin, 50 pl of the Hepes/KCl buffer without the drug was perfused beneath the coverglass past the cells.

Dioctanoylglycerol solutions were prepared as concentrated stocks (60 mg ml^-1) in dimethyl sulfoxide (DMSO) and stored frozen (−20°C) in small aliquots. Purified 1,2- and 1,3-diacyl-glycerol spontaneously isomerize in aqueous solution [Ganong et al. 1986; Nomura et al. 1986), making it essential to prepare the aqueous diacylglycerol solutions daily. Both 1,2- and 1,3-dioctanoylglycerol solutions were freshly prepared each day by diluting the concentrated DMSO stocks into 15 mM Hepes/KCl buffer. Neomycin was prepared freshly in 15 mM Hepes/KCl buffer at its final concentration of 2 mM.

All experiments were conducted at 18–22°C. An LR interference filter was in the microscope light path at all times. Mitotic progression was observed in dim, broad-band green light (500–550 nm) from a 60 W tungsten lamp operated at 7–9 VDC with differential interference contrast microscopy (Carl Zeiss, New York) (Wolniak and Bart, 1985a). The objective lens used was a Zeiss 40x strain-free achromat (n.a.=0.85). Rates of chromosome separation were determined by measurements from the videotapes from two pairs of chromosomes that remained in the plane of focus for at least 4–6 min (Larsen and Wolniak, 1990). Photomicrographs were obtained from the time-lapse videotapes, as described previously (Larsen and Wolniak, 1990), without enhancement or frame averaging. The video signal was run into a Seikoshia VP-3500 thermal video processor (Seikosha America Co., Mahwah, NJ), which produced 16 cm × 25 cm still frame images. We then photographed these images using a 6 cm × 7 cm camera (Asahi Pentax Co., Tokyo) with Kodak T-Max 100 film (Eastman Kodak, Co., Rochester, NY). The negatives were printed with a magnification factor of × l.25.

The remarkably predictable rate of progression through mitosis exhibited by stamen hair cells of T. virginiana permits the study of regulatory systems that control the process (Hepler, 1985; Wolniak and Bart, 1985a; Larsen et al. 1989; Larsen and Wolniak, 1990). In this series of experiments, our observed mean metaphase transit time (i.e. the interval between nuclear envelope breakdown and anaphase onset) for untreated cells was 31min±2min (S.D.) and the mean duration of anaphase (i.e. the interval from anaphase onset to the commencement of cell plate vesicle aggregation) was 18min±2min (S.D.) (Fig. 1, line A; Table 1). The mean rate of anaphase chromosome separation, as determined from videotape analysis (Larsen and Wolniak, 1990), was 1.3 jimmin^-1±0.2 jzmmin^-1 (S.D.) (Table 1).

Our treatments of stamen hair cells with neomycin resulted in a series of statistically significant shifts in the rates of mitotic progression and chromosomal separation from these mean values. Not surprisingly (Wolniak, 1987; Larsen et al. 1989; Larsen and Wolniak, 1990), the primary determinant for the observed shifts from control levels was the initial time of treatment with the drug; the results of our treatments of cells with neomycin are presented in terms of the times of initial treatment relative to the timing of nuclear envelope breakdown, an event that serves as an unmistakable marker for entry into prometaphase.

The initiation of either a 15- or 30-minute treatment of stamen hair cells with 2mM neomycin resulted in the failure of over 75% of the cells to undergo nuclear envelope breakdown. In a sample size of over 400 cells treated in this fashion, over 300 of them became arrested in prophase, with heavily condensed chromatin (Fig. 1, line B). Less than 15 of these cells reverted to interphase without dividing within a 2–3 h time frame. We did not attempt to reverse prophase arrest in the vast majority of these cells, because we are unable to predict reliably when (or whether) nuclear envelope breakdown would occur. Nuclear envelope breakdown in untreated stamen hair cells is a rapid event that is easily documented with time lapse video microscopy, but is, nevertheless, an event that we can only predict temporally within a few minutes of its actual occurrence. Cells in late prophase exhibit highly condensed chromatin in a spherical array that can persist unchanged for over an hour after neomycin treatment. Remarkably, all of the morphological indicators in these arrested cells suggest that nuclear envelope breakdown could occur at any time. With the understanding that we lack an a priori sense of the precise timing for nuclear envelope breakdown, we attempted reversals from prophase arrest in a small sample of cells (n=17). Cells pretreated for 15 min with neomycin, were allowed to remain in what appeared to be a late prophase configuration for 30 min before we perfused a Hepes/KCl buffer solution containing 0.5μgml⁻¹ 1,2-dioctanoylglycerol beneath the coverglass. None of the cells in the sample either progressed through nuclear envelope breakdown or reverted to interphase for at least 60 min after the perfusion.

While 75% of the cells became arrested in prophase, 25 % of the cells in our sample progressed through nuclear envelope breakdown into prometaphase and metaphase (Fig. 2). This group comprised a total of 110 cells; 83 of the cells shown in Fig. 2 were treated with neomycin prior to nuclear envelope breakdown and 37 cells were treated after nuclear envelope breakdown. The metaphase transit time (Fig. 2), the duration of anaphase, and the rate of chromosomal separation were all dependent upon the interval separating the time of initial exposure to neomycin during prophase and the time of nuclear envelope breakdown. Cells treated 0–5 min prior to nuclear envelope breakdown exhibited metaphase transit times not significantly different than those of control cells. These cells progressed through anaphase normally; there was no statistically significant difference between them and control cells, in terms of either the interval from anaphase onset to cell plate vesicle aggregation or the velocity of chromosome separation. In striking contrast, the initiation of treatment with 2 mu neomycin somewhat earlier in prophase, 5–25 min prior to nuclear envelope breakdown, resulted in a significant extension of the metaphase transit time, to approximately 48 min from a normal of 31 min. Even more pronounced, however, was the reduction in the rate of chromosomal separation, to ∼0.6μmmin⁻¹ from a normal rate of ∼1.3gmmin⁻¹ (Table 2). The reduction in the velocity of chromosomal separation was accompanied by an increase in the length of the interval from anaphase onset to cell plate vesicle aggregation to approximately 27 min. Cell plate vesicle aggregation is an event that normally occurs 18–20 min after the start of anaphase. When the plate did form, it was noticeably thinner than normal, but ultimately, after a period of several hours, it was not discernably different from comparably new walls separating untreated cells.

The initiation of a 15- or 30-min treatment with neomycin even earlier in prophase, 30–90 min prior to nuclear envelope breakdown (when it occurred, in 25 % of the cells observed) resulted in metaphse arrest (Fig. 1, line C). The cells entered metaphase with a well-formed spindle, but were unable to progress into anaphase or to revert to interphase without dividing for several hours. In a continuous 11-h observation of a cell arrested in metaphase, the mitotic spindle was clearly present as a biconical clear zone for 3–4 h and then it apparently disintegrated. The chromosomes remained condensed and the nuclear envelope failed to reform for an additional 2 h. During metaphase arrest in this and most other cells, spindle rotation away from the longitudinal axis of the cell was commonly observed and, in the extreme cases, the spindle axis ended up perpendicular to the longitudinal axis of the cell.

As a positive control for our neomycin treatments, we incubated cells for 30 min with 2 mM clindamycin, another aminoglycoside antibiotic, at various and comparable times during prophase and metaphase. Irrespective of the time of treatment with clindamycin, we detected no discernable difference in the metaphase transit time, the duration of anaphase, or the rate of anaphase chromosome separation from those values observed in untreated cells (Fig. 1, line D; Table 1). The mean values for these mitotic parameters in clindamycin-treated cells are significantly different from those obtained with neomyin treatments.

For our attempts at reversal of metaphase arrest, we limited ourselves to the set of neomycin treatment during prophase, 30–90 min prior to nuclear envelope breakdown. Once nuclear envelope breakdown occurred, these cells always became arrested in metaphase, remaining there with an intact spindle for over 2 h. We judged these cells to be arrested in metaphase when they failed to enter anaphase 45 min after nuclear envelope breakdown, before attempting to induce anaphase. Some of our other treatments with neomycin (below) resulted in an extension, but not in the true arrest, of progression through metaphase.

Metaphase arrest induced by neomycin could be reversed by post-treatment with 60 μgml⁻¹ or 0.5 μg ml⁻¹ 1,2-dioctanoylglycerol, added 45 min after nuclear envelope breakdown. Perfusion with 60 μgml⁻¹ of the diglyceride resulted in anaphase onset in 65 % of the cells observed (11/16 cells in the sample, Table 2). The interval from anaphase onset to cell plate vesicle aggregation was increased to approximately 30 min (Fig. 1, line E; Table 2), from its normal 18–20 min length in untreated or in clindamycin-treated cells. The increased length of anaphase was accompanied by a significant reduction in the velocity of chromosome separation, to ∼0.5 mmin⁻¹ (Table 2). Normally, we would expect to see chromosomes separating at 1.3μmmin⁻¹. Just over 50% of the cells treated with 0.5 μgml⁻¹ 1,2-dioctanoylglyerol (8/15 cells in the sample) entered anaphase after metaphase arrest, with anaphase onset occurring approximately 18 min after addition of diglyceride (Fig. 1, line F; Table 2). The interval from anaphase onset to cell plate vesicle aggregation in these cells was ∼36 min with an average chromosome separation velocity of 0.6 μm min⁻¹ (Table 2). Surprisingly, the initial appearance of cell plate vesicles in the spindle midzone occurred with either of the 1,2-dioctanoylglycerol treatments only 12–15 min after anaphase onset, while the cells were still in mid-anaphase, but a clear linear array of plate vesicles comparable to that in untreated cells at ∼20 min after anaphase onset did not appear until much later at 30–36 min after anaphase onset, which is 10–16 min later than normal. Following their initial appearance in the mid-anaphase spindle midzone, the vesicles were interspersed among the separating chromosomes. As the chromosomes became further separated, the vesicles resided in a broad band in the spindle midzone (Fig. 3E). Then, 15–20 min after their initial appearance in the spindle, the vesicles aggregated into a linear (actually planar) array. This array is unusual in the sense that it was extremely thin, and the coalescence of these vesicles resulted in the formation of an anomalously thin cell plate (Fig. 3F). Our attempts to reverse neomycin-induced metaphase arrest with the 1,3-isomer of dioctanoylglycerol were completely unsuccessful (Fig. 1, line G). The cells that were not rescued from metaphase arrest usually exhibited spindle collapse at various intervals after the addition of the diglyceride, and then, in contrast to prophase arrest, a gradual reversion to interphase.

Reversal of metaphase arrest could also be induced by the perfusion of calcium chloride beneath the coverglass past the cells. The addition of calcium chloride was sufficient to reverse the block to progression only half of the time, however. Anaphase onset occurred in these cells approximately 16 min after the addition of the cation. Here, the interval between anaphase onset and cell plate vesicle aggregation averaged 33 min (Fig. 1, line H), and the mean chromosomal separation velocity was ∼0.8 μm min⁻¹ (Table 2). Again, the initial appearance of cell plate vesicles occurred while the cells were in midanaphase, while the chromosomes were not fully separated, and the plate they ultimately formed was anomalously thin. Treatment with magnesium chloride was insufficient to reverse metaphase arrest induced by neomycin (Fig. 1, line I).

The addition of 2 mM neomycin to cells in prometaphase or early metaphase, 0–15 min after nuclear envelope breakdown, resulted in an extension of metaphase, but not its arrest, to 37min±5min (S.D.). Once these cells entered anaphase, they progressed normally; the interval from anaphase onset to cell plate vesicle aggregation was ∼20 min and the rate of chromosomal separation was ∼1.1 μm min⁻¹ (Table 1). The cell plates in these cells were not discemably different from those in untreated (or clindamycin-treated) cells.

In contrast, the addition of neomycin to cells already well into metaphase, 16–29 min after nuclear envelope breakdown, resulted in a significant extension of the metaphase transit time, to ∼58 min. We did not attempt to induce anaphase onset in these cells by the addition of 1,2-dioctanoylglycerol or calcium, because we were unable to predict reliably when the cells would spontaneously enter anaphase. The duration of these drug treatments was 15 min. In addition to its effects on the metaphase transit time, the drug prolonged the interval from anaphase onset to cell plate vesicle aggregation, to an average of 28 min (Table 1). Chromosomal separation rates were again reduced from the normal mean of ∼ 1.3μmmin⁻¹ to 0.8 μm min⁻¹ (Table 1).

Remarkably, the initial addition of neomycin to cells in late metaphase, 30 or more min after nuclear envelope breakdown was followed by mitotic progression that was absolutely normal, in terms of the interval from nuclear envelope breakdown to anaphase onset, the interval from anaphase onset to cell plate vesicle aggregation, the rate of chromosome separation (Fig. 2), and the appearance and thickness of the cell plate.

Neomycin is an aminoglycoside antibiotic that is commonly used to block translation by 70S ribosomes (Hershey, 1977). Neomycin also binds to phosphatidylinositol 4,5-bisphosphate in eukaryotic cells (Au et al. 1986), and has been reported to inhibit agonist-stimulated phosphoinositide turnover (Downes and Michell, 1981; Carney et al. 1985; Whitaker and Aitchison, 1985; Tysnes et al. 1987). It is through this mechanism that the drug has been used to study polyphosphoinositide-mediated events in several cell types (Carney et al. 1985; Streb et al. 1985; Vergara et al. 1985; Prentki et al. 1986; Swann and Whitaker, 1986; Tysnes et al. 1987; Cemescu et al. 1988). Our previous studies on mitotic progression involving lithium treatments (Wolniak, 1987; Chen and Wolniak, 1987,b) led us to suspect that one or more episodes of polyphosphoinositide cycling may occur during metaphase. Our pharmacological studies on mitotic progression in stamen hair cells have led us to generate a hypothetical model for mitotic regulation (Wolniak, 1988), involving multiple oscillatory episodes of calcium-dependent and -independent protein phosphorylation followed by waves of dephosphorylation, with the same cast of regulatory characters activating different processes in the cell at different times. Our working hypothesis in this study is that polyphosphoinositide cycling is part of this oscillating regulatory cascade that culminates in anaphase. We treated stamen hair cells with neomycin to see if stagespecific arrest could be reversed specifically by the polyphosphoinositide elicitor 1,2-dioctanoylglycerol or by calcium, thereby linking the regulatory pathway to the process.

We show that neomycin alters the rate of mitotic progression in stamen hair cells of T. virginiana and that the primary determinant for the response by the cell is the time of initial exposure to the drug. Neomycin induces arrest in prophase, where cells can neither proceed to metaphase nor revert directly to interphase without going through division. We did not attempt reversals of prophase arrest in the vast majority of cells observed because nuclear envelope breakdown serves as our temporal benchmark for mitotic progression. In the absence of a clear sense of whether or when nuclear envelope breakdown is to occur, such experiments are largely meaningless. Nevertheless, with this understanding in mind, we found that all of the cells in a small sample arrested in a late prophase configuration by neomycin treatment could not be rescued by post-treatment with 1,2-dioctanoylglycerol. These cells could neither progress to prometaphase nor revert to interphase without mitosis for at least 60 min. By our morphological criteria of progression and under these conditions of attempted reversal, we judge the cells to be irreversibly locked in late prophase. Lithium treatments resulted in irreversible prophase arrest (Wolniak, 1987) that is indistinguishable from the effect observed with neomycin.

Treatments with neomycin introduced at other times during mitosis extend or arrest progression through metaphase; in the latter case, the cells require 2—3h to revert to interphase without entering anaphase. Reversals from metaphase arrest were attempted because rates of progression after reversal can be compared directly with those of untreated, or control-solution-treated cells. We show that metaphase arrest is partially reversible by treatment with 1,2-dioctanoylglycerol or by calcium but not by 1,3-dioctanoylglycerol or magnesium. Because of this reversibility with 1,2-dioctanoylglycerol or calcium, we suspect that the mechanism of inhibition to mitotic progression by neomycin lies in a disruption of polyphosphoinositide cycling rather than in a blockade of translation. In control experiments for the neomycin treatments, we used clindamycin, a related antibiotic, and found it to be totally ineffective in altering the rate of mitotic progression in stamen hair cells.

We have found that neomycin-induced arrest can be partially reversed by treatment with either 1,2-dioctanoylglycerol, a permeant dicylglycerol widely used to study protein kinase C-mediated responses (Davis et al. 1985; May et al. 1986; Nomura et al. 1986), or by calcium chloride. These results support our previous postulation on the regulatory role of polyphosphoinositide cycling in mitosis (Wolniak, 1987, 1988; Larsen et al. 1989; Larsen and Wolniak, 1990). Since protein kinase C can be activated by the 1,2-, but not by the 1,3-diacylglycerol isomer (Bell, 1986; Nishizuki, 1986; Ganong et al. 1986), our results suggest that protein phosphorylation mediated by this enzyme (or enzyme class) might be an important step in the regulation of spindle function at specific time points before the transition from metaphase to anaphase (Larsen and Wolniak, 1990). Biochemical analyses have shown that lipid- and calcium-dependent kinases exist in plants (Harmon, 1990), and consensus hybridization of plant cDNA libraries indicates that protein kinase C homologs may be present, though different from animal protein kinase C proteins in some of their regulatory domains (Lawton et al. 1989). 1,2-dioctanoylglycerol and calcium are each effective only about half of the time in reversing metaphase arrest, results that may be based on these differences in regulation, and partially on the exact timing of our reversal treatments relative to the stage of mitosis. Previously, we (Chen and Wolniak, 1987 b) found that only during an extremely short interval near the onset of vesicle aggregation could we prevent anomalous cell plate vesicle dispersion after earlier lithium treatment during anaphase. The interval ended much earlier in the period of vesicle aggregation than one might otherwise suspect. It is possible that the 50% reversal rates we report here reflect the ending of brief signaling intervals during which cells retain the capacity to enter anaphase. In this scenario, attempts at reversal thereafter would be unsuccessful. Curiously, there is no morphological change in the cell arrested in metaphase that accompanies this possible loss in capacity to progress until 1–2 h later, when the spindle collapses. In these cases, the cell retains the capacity to revert to interphase without mitosis, a capacity that is not apparent in cells arrested in prophase, just prior to nuclear envelope breakdown. We chose to delay rescue attempts until 45 min after nuclear envelope breakdown because we wanted to be certain that the cells were actually arrested in metaphase.

The interval between reversal solution addition and anaphase onset was similar for calcium and 1,2-dioctanoylglycerol, suggesting that the two treatments may operate through a common pathway, either to supplant or reverse the block to progression through metaphase. If neomycin blocks phospholipase C activity that is necessary for anaphase by affecting its binding to phosphatidylinositol substrates, and this phospholipase C activity produces both a phosphorylated inositide and a 1,2-diacylglycerol as elicitors, our reversal experiments, involving or affecting only one part of the bifurcating polyphosphoinositide cascade, may already be operating at maximal efficiency. It is possible that both types of elicitors are necessary in a cascade that precedes anaphase, and the addition of either the diglyceride or calcium is insufficient to overcome one or more necessary steps for sister chromatid separation that is normally activated by the other elicitor pool.

Treatment with lithium can induce metaphase arrest in stamen hair cells that is reversible by treatment with myoinositol or calcium (Wolniak, 1987). Although the effect of lithium may involve the inhibition of polyphosphoinositide cycling, the cells arrested in metaphase exhibited a pattern of reversal that is different from those arrested with neomycin. Cells arrested by lithium treatment reverted back to interphase within ∼70 min, and a calcium perfusion past the cells arrested in metaphase was followed by entry into anaphase after 2–5 min (Wolniak, 1987). For neomycin-arrested cells, the interval between calcium addition and anaphase onset was approximately 17 min. The difference between the two reversal intervals may indicate different sites of action for each inhibitor and for calcium. Lithium appears to inhibit the recycling of inositol 1-monophosphate (Abdel-Latif, 1986; Sekar and Hokin, 1986; Berridge et al. 1989), while neomycin appears to inhibit the production of inositol 1,4,5-trisphosphate and 1,2-diacylglycerol (Downes and Michell, 1981; Carney et al. 1985; Whitaker and Aitchison, 1985; Tysnes et al. 1987). Thus, phospholipase C is unaffected by lithium, and the synthesis of inositol 1,4,5-trisphosphate and 1,2-diacylglycerol from the hydrolysis of phosphatidylinositol 4,5-bisphosphate may continue unabated if the inositol pool does not form a limiting factor for the production of these elicitors (Abdel-Latif, 1986). The subthreshold amount of elicitors required for anaphase may be sufficient to induce the reversion to interphase or may potentiate the reversal of metaphase arrest by calcium. In these and our earlier experiments (Wolniak, 1987), we cannot discount the possibility that other, nonspecific effects of neomycin (Prentki et al. 1986) and/or lithium (Abdel-Latif, 1986) may underlie differences in the occurrence of metaphase arrest, and in the rate or extent of its reversal.

In contrast to neomycin treatments, metaphase arrest in stamen hair cells induced by quin2 (Wolniak and Bart, 1985a) or nifedipine (Wolniak and Bart, 1985b) lasted for 70–100 min before spontaneous reversion to interphase. Metaphase arrest that follows neomycin treatment appears to be distinct from that induced by quin2 or nifedipine; after neomycin, the cell can maintain a metaphase plate that appears normal for an extended period (usually longer than 2 h) before apparent spindle collapse, nuclear envelope re-formation or chromatin decondensation. It appears that reversion to interphase requires a step that is blocked by neomycin but not by these other treatments. Since quin2 and nifedipine apparently induce metaphase arrest by inhibiting a necessary calcium influx for anaphase (Hepler, 1985; Wolniak and Bart, 1985a,b), and since the elimination of a calcium influx in arrested cells does not prevent reversion (Wolniak and Bart, 1985a,b), it is possible that calcium plays only an indirect role in nuclear envelope reformation and/or chromosome decondensation during reversion to interphase.

There are several recent reports that 1,2-dioctanoylglycerol treatment may induce calcium mobilization, directly increasing cytosolic calcium activity (Ebanks et al. 1989; Larsen and Wolniak, 1990), and thereby providing the common pathway for the reversal of metaphase arrest mentioned above. If the 1,2-dioctanoylglycerol added to our arrested cells is acting solely as an elicitor of an increase in cytosolic calcium activity, then we would expect a higher reversal rate for the addition of the cation alone, because calcium appears to enter the cytosol rapidly from the external space during reversal from metaphase arrest after chelator treatment in the absence (Hepler, 1985) or presence of ionophores (Wolniak and Bart, 1985a), or after photo-oxidation of calcium channel blockers (Wolniak and Bart, 1985b). In addition, a rise in cytosolic calcium induced by 1,2-dioctanoylglycerol might be expected to promote rapid chromosome-separation velocities, because rates of separation in stamen hair cells have been shown in microinjection studies to be directly related to cytosolic calcium activity (Zhang et al. 1990). Neither effect was observed. Although we cannot rule out calcium mobilization as a consequence of treatment with 1,2-dioctanoylglycerol, we believe that the diglyceride may act through a variety of mechanisms, possibly through the activation of a protein kinase (Larsen and Wolniak, 1990). Very recently, Levin and coworkers (1990) demonstrated that a mutation manifested conditionally as a disruption of protein kinase C in budding yeast blocked both cell division and cell growth, somewhere in mid/late G_a of the cell cycle. The cessation of division and growth was far more pronounced and earlier than that observed with the disruption of the cell cycledependent protein kinase, p34^cdc2 (Draetta and Beach, 1989; Norbury and Nurse, 1989), indications that protein kinase C plays an essential regulatory role in both processes and may operate both upstream from, and, in parallel with, p34^cdc2 kinase.

Our current studies with neomycin suggest that polyphosphoinositide cycling probably occurs at specific times during metaphase, preceding anaphase onset, and possibly during prophase, preceding nuclear envelope breakdown. The presence of neomycin prior to or during these periods results in extension of the metaphase transit time, or metaphase arrest, or prophase arrest, respectively. We believe it is significant that once the neomycin-sensitive step during early/middle stages of both prophase and metaphase is complete, the addition of neomycin in late prophase or in late metaphase has no effect on subsequent mitotic progression. Chromosome-separation rates are also dramatically slowed down by the presence of neomycin, but, curiously, only when the cells have been rescued from metaphase arrest or when the drug is added more than 4–5 min prior to anaphase onset. It is possible that the cell plate anomalies we observe are related to the slowed anaphase chromosome movement; the presence of chromosomes in the spindle midzone during the earliest stages of phragmoplast formation and cell plate vesicle movement may alter the process and prevent normal vesicle aggregation from occurring. The broad band of vesicles (Fig. 3E) may be a reflection of improper associations between the vesicles and the phragmoplast microtubules or, alternatively, a malformed phragmoplast. The thin plate may be the result of a reduced number of vesicles being incorporated into the forming partition between the daughter cells.

Correlated with our postulation on the possible involvement of protein phosphorylation prior to the metaphase/ anaphase transition are several immunological studies that suggest that the state of phosphorylation of specific metaphase spindle components changes drastically during mitosis, rising during metaphase and then declining by early anaphase. The monoclonal antibody, MPM-2, originally directed against phosphorylated proteins in mitotic HeLa cells (Davis et al. 1983), stains spindle poles and kinetochores of metaphase cells (Vandre et al. 1984). The signal in metaphase cells is significantly brighter and more specifically localized than that observed during anaphase (Vandre et al. 1984). At least some of the phosphorylation detected by MPM-2 can be attributed to the activity of MPF (i.e. maturation or mitosis promoting factor, a cytoplasmic extract from dividing cells that contains an active kinase homologous with the cell cycle protein kinase known as p34^cdc2; Kuang et al. 1989). In addition, the changes in the phosphorylation status in certain centrosomal components may be related to the microtubule organizing activity in the structure at different stages of mitosis (Vandre and Borisy, 1989; Centzone and Borisy, 1990). The decrease in apparent phosphorylation may be caused by a rise in phosphatase activity during anaphase (Vandre and Borisy, 1989). Increases in phosphatase activity starting in mid-late metaphase (Larsen and Wolniak, 1990) and continuing through anaphase onset are likely (Cyert and Thorner, 1989) and perhaps necessary for sister chromatid separation (Booher and Beach, 1989; Okhura et al. 1989). Clearly, the activity of the cell cycle kinase p34^cdc2 drops around the time of the metaphase/anaphase transition, as a consequence of the dissociation and degradation of cyclin from the catalytic kinase subunit (for reviews see Norbury and Nurse, 1989; Draetta and Beach, 1989).

The occurrence of a p34^cdc2 kinase homolog (John et al. 1989; Feiler and Jacobs, 1990), and its association with other proteinaceous factors (i.e. a putative cyclin) for high histone Hl kinase activity has now been demonstrated in plant cells (Feiler and Jacobs, 1990). We do not know if neomycin affects the activity of p34^cdc2 kinase. In organisms and extracts where its activity has been well characterized, this kinase generally appears not to be calcium-dependent (Moreno et al. 1989). In living cells, it is conceivable that some of its activities on mitotic substrates may be supplanted by the kinds of manipulations we have performed on calcium or calcium-dependent regulatory elicitors, and their consequent activation of other pathways may be sufficient to promote or forestall mitotic progression (Larsen and Wolniak, 1990). In addition to p34^cdc2, a calcium/calmodulin-dependent protein kinase activity exhibits maximal activity at the metaphase/anaphase transition in sea urchin spindle isolates (Dinsmore and Sloboda, 1988). The major substrate for this kinase appears to be a 62×10³M_r protein, which, upon phosphorylation, sensitizes the spindle microtubules to calcium-induced depolymerization. Obversely, the necessity of kinase activity for the metaphase/anaphase transition can be demonstrated by experiments in which the inhibition of the kinase activity forestalls anaphase onset. In PtK! cells, the microinjection of an inhibitor protein or an antibody to cyclic AMP-dependent protein kinase during prophase prolongs the duration of metaphase (Browne et al. 1987). Collectively, these results suggest that more than one type of kinase may be present and responsible for the phosphorylation of specific sets of substrates, and that each set of substrates may have a different role during the transition period between metaphase and anaphase. The specific phosphorylation of mitotic substrates during prometaphase and metaphase by several kinases working in parallel, followed by waves of phosphatase-mediated dephosphorylation late in metaphase may underlie the regulation of progression into anaphase.

We are grateful for the support that this work has received from the NSF through research grant DCB-8700422, and for support from the Maryland Agricultural Experiment Station, through Hatch Project MD J-136 (Article number A-4868, Contribution number 7899) and MAES Competitive grant J-003. We are also grateful to Dr P. E. Kolattukudy for providing us with samples of purified cutinase.