Differential Mitotic Responses to Microtubule-stabilizing and -destabilizing Drugs¹

A variety of antimitotic agents interact in the tubulin/microtubule system, a validated target for cancer chemotherapy. Taxol, the epothilones, and discodermolide have a binding site in β-tubulin in the microtubule polymer, and stabilize microtubules promoting the formation of microtubule bundles (1, 2, 3). In contrast, vinblastine, colchicine, and nocodazole bind primarily to αβ−tubulin dimers and prevent assembly of microtubules. However, low concentrations of both Taxol and vinblastine suppress growing and shortening at the ends of microtubules and appear to block mitosis by dynamically stabilizing spindle microtubules (4, 5).

Cell death occurs in cancer cells treated with Taxol (6), and the question arises as to the relationship between mitotic block and cell death. Although Taxol-induced apoptosis may occur after a prior mitotic arrest (7), it could also be induced independently of G₂-M arrest (8). Previous studies with lung carcinoma A549 cells found that low concentrations of Taxol inhibited cell proliferation without blocking cells at mitosis (9) and induced a large population of hypodiploid cells seen close to the G₁ peak (10). It was speculated that the aneuploid cells might originate from aberrant mitosis (10), although the mechanism remained to be determined.

Because suppression of spindle dynamics by low concentrations of drugs is a common mechanism responsible for mitotic block, induced by both microtubule-stabilizing and -destabilizing agents (4, 5), we questioned whether other microtubule-interacting drugs would, like Taxol, induce aneuploid cells without blocking cells at mitosis. The ability of six antimitotic drugs to initiate aberrant mitosis and to induce an aneuploid population of A549 cells has been analyzed. Our results indicate that the stabilizing drugs, but not the destabilizing agents, induce aberrant mitosis through multipolar spindle formation that results in aneuploid populations of cells.

Human non-small cell lung carcinoma cells, A549, were cultured as described previously (10). A549.EpoB40, an EpoB³-resistant cell line, was isolated from A549 cells in our laboratory, and is maintained in 40 nm EpoB (11). Monolayer cultures were grown in 7% CO₂ and passaged at intervals of 4 days. Fluorescent dyes rhodamine-phalloidin and Alexa 488-conjugated antimouse IgG were from Molecular Probes (Eugene, OR). All other chemicals were purchased from Sigma Chemical Co.

Distribution of DNA content in A549 cells treated with microtubule interacting drugs was determined by flow cytometry (10). Briefly, cells treated with drugs were fixed in 70% ethanol and resuspended in PBS containing PI and DNase-free RNase A. Forward and orthogonal scatter lights, as well as red fluorescence were analyzed by fluorescence-activated cell sorting. A dual parameter dot plot of the integrated area verses the width of the fluorescence pulse was displayed to exclude cell aggregates from cell counting and analysis. The histogram of DNA distribution was modeled as a sum of G₁ (Fig. 1,B 2N), G₂-M (Fig. 1,B, 4N), S phase, and an aneuploid population close to the G₁ peak, by using ModFitLT software (Fig. 1 B).

After growth for 48 h on coverslips, subconfluent A549 cells were incubated with drugs at the indicated concentrations for 18 h. Cells were made permeable and fixed (10). After blocking, cells were incubated with mouse monoclonal α-tubulin antibody for 1 h at 37°C. The secondary antibody, Alex 488-conjugated goat antimouse IgG, was added to the cells for 1 h, in the presence of rhodamine-phalloidins for actin staining. Chromosomes were stained with 1 μg/ml DAPI in PBS. After washing with PBS, the slides were mounted and sealed. Fluorescent images were acquired with an Olympus high-resolution microscope, connected to a CCD camera. Tricolor images were merged using Photoshop 5.5. Approximately 100 mitotic cells were counted to calculate the number of spindles per cell.

It was previously noted that treatment of human lung carcinoma A549 cells with low concentrations of Taxol (nm) resulted in a hypodiploid population of cells in the absence of a G₂-M block (10). This phenomenon was examined with two other microtubule-stabilizing agents, EpoB and discodermolide. A large population of hypodiploid cells, close to the diploid 2N peak, was observed on cell cycle analysis after incubation of A549 cells with increasing concentrations of EpoB for 18 h. In addition, a small shoulder was seen to the right of the 2N DNA (Fig. 1,A). The hypodiploid population and the shoulder, taken together, can be best fitted by a symmetric distribution near the G₁ peak, in addition to the standard distribution of the cell cycle (Fig. 1,B). A dose-response curve is shown in Fig. 2 that compares induction of the aneuploid population with the G₁ and G₂-M phases of the normal cell cycle for six microtubule interacting agents. If the hypodiploid portion alone were plotted instead of the whole aneuploid population, the dose-response curve would be the same, although with lower percentages of cells.

All three microtubule-stabilizing agents, Taxol, EpoB, and discodermolide induced aneuploid populations in A549 cells. The aneuploid population was most apparent before an increase in the G₂-M population. This population, including both the hypodiploid cells and the cells represented by the right shoulder of the 2N peak, disappeared at higher concentrations of the stabilizing drugs when the cells were blocked at mitosis (Fig. 2). Treatment with 0.2 mm H₂O₂ or 2 μm geldanamycin, both of which caused mitotic arrest, together with 10 nm EpoB for 18 h, decreased the EpoB-induced aneuploid population (data not shown). Induction of aneuploid cells evidently requires mitotic cell cycle progression, suggesting that aneuploid populations, amid normal G₁ cells, originate from aberrant mitosis (12). In contrast to the stabilizing drugs, three microtubule-destabilizing agents including vinblastine, colchicine, and nocodazole did not induce a large population of aneuploid cells in the absence of mitotic block (Fig. 2).

Mitoses were examined by fluorescent microscopy, and four kinds of spindle structures were observed in different cells before division (Fig. 3,A). Some cells (Fig. 3, A.1) contained a ring of condensed chromosomes with a monopolar spindle pole in the center. Bipolar spindles (Fig. 3, A.2) were present not only in normal cells, but also in drug-treated cells. Lagging chromosomes close to spindle poles were occasionally found in drug-treated metaphase cells. In the absence of proper spindle attachment, lagging chromosomes may activate a spindle checkpoint and block anaphase transition (13). Tripolar (Fig. 3, A.3 ) and quadripolar (Fig. 3, A.4) spindles, with abnormal chromosome alignments, also were seen in cells treated with the stabilizing drugs. Multipolar spindles have been observed previously in cells treated with taxanes (14, 15). For a detailed view of bipolar, tripolar, and tetrapolar spindles and their associated chromosomes in EpoB-treated cells, three-dimensional movies have been generated from confocal fluorescent microscopy and are available from the authors or on the Internet.⁴ (QuickTime software is required to view the files.)

The spindle structures were compared at those drug concentrations at which the G₁ population was initially decreased (Fig. 2). Multipolar spindles (>2) were noted only in cells treated with the stabilizing drugs, but not in cells treated with the destabilizing agents (Fig. 3,B). For cells treated with the destabilizing drugs, such as HeLa cells incubated with Vinca alkaloids (16), there was an increase in the incidence of mitotic cells with monopolar spindles. It is important to point out, however, that cells containing the same number of spindles may have subtle differences resulting from different drug treatments. For example, monopolar spindles in the cells treated with the destabilizing drugs displayed distinct astral microtubules irradiating from a nucleation center (data not shown). In contrast, cells treated with the stabilizing agents have a condensed, ball-shaped microtubule center (Fig. 3, A.1).

As long as chromosomes can congress and align, although they are distorted (Fig. 3, A.3  and A.4), the presence of multipolar spindles does not alert the checkpoint controlling the exit of mitosis (17). Mitotic cells with multipolar spindles resulted in aberrant cell divisions. A few cells gave rise to unequal division (Fig. 3, C.1), which would produce not only hypodiploid cells but also G₁ cells with more than 2N DNA. Tripolar cell divisions were significantly increased in cells treated with microtubule-stabilizing drugs compared with control cells (Fig. 3, C.2 and C.3 ). At the drug concentrations used in Fig. 3,B, ∼15% of the telophase cells, after incubation with Taxol or EpoB, produced three daughter cells connected by multiple midbodies, whereas the remaining 85% of the telophase cells produced two daughter cells. With discodermolide, 25% of the telophase cells produced tripolar cell division. In contrast, no tripolar cell division was observed in cells treated with the destabilizing drugs at the concentrations indicated in Fig. 3 B, nor in cells treated with 9 and 11 nm of vinblastine (data not shown). Both tripolar cell division and multipolar spindles occurred only in cells treated with the stabilizing drugs. A higher percentage of tripolar cell division might have been expected if all mitotic cells with tripolar spindles had progressed to tripolar cell division. However, the number of tripolar cells may be underestimated if one cell was separated early from the two other connected cells. Our results suggest that a significant percentage of mitotic cells contained multipolar spindles that led to aberrant cell division and produced aneuploid G₁ cells. Multipolar spindles have been observed in many tumor cells and are responsible for the induction of aneuploidy (18).

A significant aneuploid population was induced by treatment with 8 nm Taxol for 18 h (Fig. 2). In response to an additional treatment with 36 nm Taxol that blocked cells in mitosis, the accumulation of cells at the G₂-M phase was significantly slower compared with cells without a preincubation (data not shown). This suggests that part of the cells were not cycling after preincubation with Taxol. Taxol may have activated the G₁ checkpoint genes p53 and p21 (9), and induced a G₁ block in cells after division (19). Consistent with previous observations in Taxol-treated cells (20, 21), drug sensitivity for microtubule-stabilizing agents corresponds to the induction of aneuploid G₁ cells. The drug concentrations required for 50% inhibition of cell growth after a 72-h incubation (IC₅₀) with Taxol and discodermolide were 3.4 and 9.5 nm, respectively. These concentrations were closer to the concentrations of drug required for the initiation of aneuploidy than for the induction of mitotic arrest (Fig. 2). Interestingly, EpoB had an IC₅₀ value of 0.57 nm, which was even lower than the drug concentration (3.3 nm) needed for one-half maximal induction of aneuploidy (Fig. 2). This suggests that induction of an aneuploid population may not fully account for the EpoB cytotoxicity, and that other mechanisms may be involved. Alternatively, more accurate methods such as fluorescence in situ hybridization may be required to measure aneuploidy induction by EpoB at concentrations close to its IC₅₀ value.

To further examine the significance of aneuploidy induction on EpoB sensitivity, an EpoB-resistant cell line, A549.EpoB40, was examined. This cell line has a β-tubulin mutation at amino acid 292 and is 95-fold resistant to EpoB compared with the parental cells (11). Although a mitotic block did occur in response to increasing concentrations of EpoB, there was a large decrease in the aneuploid population compared with that in the drug-sensitive cells (compare Fig. 4,A with EpoB in Fig. 2). In contrast, the A549.EpoB40 resistant cell line that does not demonstrate cross-resistance to discodermolide (11), exhibited a large increase in the aneuploid population in response to discodermolide (Fig. 4 B). These results indicate that EpoB drug resistance is partly determined by the inhibition of aneuploid induction.

Our observations indicate that aneuploidy is a result of aberrant mitosis and contributes to cell death in drug-sensitive cells. Aberrant mitosis may confer a selective killing of malignant cells by microtubule-stabilizing drugs, because induction of aneuploidy through aberrant mitosis creates chromosomal breakage, interchromosomal concatenation, and a lethal genetic imbalance (22). Tumor cells normally display a marked genetic instability caused by amplification of centrosomes and formation of multipolar spindles (18). As evidenced by a high level of apoptosis, DNA replication in these cells operates near the error threshold for cell viability (23). Therefore, a tumor cell cannot sustain further disruption of its genome from increased aberrant mitosis attributable to microtubule-stabilizing drugs. However, if microtubule-stabilizing agents initiate multipolar spindles and aberrant mitosis in normal somatic cells, they could destabilize the normal cell genome as well, thereby being potentially mutagenic. It would be of interest to investigate whether different mitotic responses occur in normal and cancer cells in response to low concentrations of microtubule-stabilizing drugs.

In contrast to mitotic arrest, aberrant mitosis was induced only by microtubule-stabilizing drugs and not by destabilizing agents. This suggests that suppressing spindle dynamics cannot fully account for aberrant mitosis. In addition, a complete mitotic block, but not aneuploidy, was induced by higher concentrations of EpoB in A549.EpoB40-resistant cells (Fig. 4). This further supports the concept that the mechanisms responsible for mitotic block and aberrant mitosis may not be the same. Interference with the function of centrosomes may contribute to aberrant mitosis (15), because Taxol has been reported to preferentially bind to centrosomes (24). Preliminary studies using antibodies against γ-tubulin and pericentrin found two major centrosomes and additional minor ones in some aberrant mitotic cells after treatment with 4.8 nm EpoB (data not shown). Although centrosomes may not be required for the function of spindle poles, the frequency of midbody formation and successful division is higher when centrosomes are present (25). Centrosome amplification has been associated with induction of multipolar spindles and aneuploidy in cancer cells (18). Further research is needed to elucidate the molecular mechanisms of multipolar spindle induction by the microtubule-stabilizing drugs.

In summary, we have demonstrated that the response to different microtubule-interacting agents can be distinct. The occurrence of aberrant mitosis and aneuploidy is dependent on both the mechanism of drug action and the concentration of drug. The sensitivity of cells to microtubule-stabilizing drugs is mainly the result of aberrant mitosis and not mitotic block.

We thank Dr. C-P. H. Yang from our laboratory for isolating the A549.EpoB40 resistant cell line, Michael Cammer for help with confocal microscopy and three-dimensional movies, and Drs. Samuel Danishefsky and Amos B. Smith III for their help.