Nuclear Lamin A/C Expression Is a Key Determinant of Paclitaxel Sensitivity

ABSTRACT

Paclitaxel is a key member of the Taxane (paclitaxel [originally named taxol], docetaxel/Taxotere) family of successful drugs used in the current treatment of several solid tumors, including ovarian cancer. The molecular target of paclitaxel has been identified as tubulin, and paclitaxel binding alters the dynamics and thus stabilizes microtubule bundles. Traditionally, the anticancer mechanism of paclitaxel has been thought to originate from its interfering with the role of microtubules in mitosis, resulting in mitotic arrest and subsequent apoptosis. However, recent evidence suggests that paclitaxel operates in cancer therapies via an as-yet-undefined mechanism rather than as a mitotic inhibitor. We found that paclitaxel caused a striking break up of nuclei (referred to as multimicronucleation) in malignant ovarian cancer cells but not in normal cells, and susceptibility to undergo nuclear fragmentation and cell death correlated with a reduction in nuclear lamina proteins, lamin A/C. Lamin A/C proteins are commonly lost, reduced, or heterogeneously expressed in ovarian cancer, accounting for the aberration of nuclear shape in malignant cells. Mouse ovarian epithelial cells isolated from lamin A/C-null mice were highly sensitive to paclitaxel and underwent nuclear breakage, compared to control wild-type cells. Forced overexpression of lamin A/C led to resistance to paclitaxel-induced nuclear breakage in cancer cells. Additionally, paclitaxel-induced multimicronucleation occurred independently of cell division that was achieved by either the withdrawal of serum or the addition of mitotic inhibitors. These results provide a new understanding for the mitotis-independent mechanism for paclitaxel killing of cancer cells, where paclitaxel induces nuclear breakage in malignant cancer cells that have a malleable nucleus but not in normal cells that have a stiffer nuclear envelope. As such, we identify that reduced nuclear lamin A/C protein levels correlate with nuclear shape deformation and are a key determinant of paclitaxel sensitivity of cancer cells.

INTRODUCTION

Paclitaxel (originally named taxol) is a key drug widely used to treat a broad range of cancers, including epithelial ovarian cancer, with significant success (1–3). Currently, a cisplatin or carboplatin/paclitaxel regimen following debulking surgery is the standard frontline chemotherapy for ovarian cancer (1–4), and a dose intensive regimen of paclitaxel is also used in salvage treatment following recurrence as a second-line drug (5, 6). At the molecular level, paclitaxel binds to and alters tubulin’s dynamics of disassociation, resulting in stabilization of microtubule bundles (7–9). The activity in causing mitotic arrest by interfering with microtubule function, which subsequently leads to apoptosis, is commonly accepted to be the mechanism of paclitaxel for killing cancer cells (10–12). Recent observations also suggest that paclitaxel promotes cancer cell death by causing multipolar spindle formation and ensuing missegregation of chromosomes (13–15). Although these two mechanisms have strong support from cell culture studies (7, 8) and have been observed in mitotically active primary breast cancer cells (14, 15), they do not account for the observation that the majority of tumor cells in vivo are sensitive to the drug while only a small fraction of the cells are proliferative (16–19). Thus, the clinical efficacy of paclitaxel does not correlate with mitotic arrest or apoptosis, a concept known as the proliferation rate paradox of antimitotic agents (20). This notion also casts doubt on paclitaxel’s classification as a mitotic inhibitor when used in vivo (18–22).

Paclitaxel is surprisingly successful in treatment for several major malignancies (3, 10, 23). For ovarian cancer therapy, the initial response is generally adequate, but with rare exception, resistance subsequently develops. Overcoming drug resistance is considered to be the major problem that needs to be resolved (24–26). Several mechanisms have been identified for paclitaxel resistance; mutations and altered expression of tubulin and its binding proteins (27–29) and expression of members of the ATP-binding cassette family multidrug resistance (MDR) transporters (2, 30) are considered major problems. Additionally, expression of microtubule modulating proteins may contribute to drug resistance (29). Mutations in tubulin have been found only in laboratory cell lines but not in primary tumors (28). Nevertheless, a common mechanism responsible for acquired drug resistance in ovarian cancer has not been identified for the majority of cases, which contributes to the current predicament that no practical strategy is available to overcome paclitaxel resistance, despite active and continuous research on this issue (12, 19, 30). Nearly all patients relapse within 2 years and the recurrent cancer eventually becomes resistant to paclitaxel.

Enlarged and deformed nuclei are characteristic of cancer cells, and their aberrant nuclear morphology correlates with malignancy and is used as a diagnostic and prognostic indicator in Pap smears to predict tumor cell malignancy (31–33). Our previous studies suggest that defects in nuclear structure, such as the loss or reduction of lamin A/C, are the basis for altered nuclear morphology and are significant for producing aneuploidy in ovarian cancer initiation and progression (34–36). Loss of lamin A/C leads to the deformable nuclear envelope of ovarian cancer cells, and the interaction of cellular microtubule filaments with the nuclear lamina determines the aberrant nuclear shape (31, 34, 35). In particular, we observed that the malleable nuclear envelope of neoplastic cells frequently breaks off to produce micronuclei and the nucleus becomes fragmented during nonmitotic phases (34, 35). These findings prompted us to investigate the role of nuclear envelope defects in the sensitivity of the cancer cells to paclitaxel. Here, we report a potentially important discovery regarding the nonmitotic mechanism of cell killing by paclitaxel.

RESULTS

Paclitaxel induces multiple micronucleation in cancer cells.

We tested a panel of ovarian cancer cells (A2780 and OVCAR3, -4, -5, and -10) for the impact of paclitaxel and/or carboplatin on nuclear morphology and cell killing. These five cell lines were used for various experiments, and representative results from one or more lines are shown for individual tests. In multiple experiments conducted to establish a dosage curve of cell killing, all the cell lines showed similar dose responses to carboplatin and paclitaxel, with a K_d (dissociation constant) of 0.25 μM and 0.3 nM, respectively. To model the combination of the two drugs in ovarian cancer therapy, a protocol was designed: the cells were treated with paclitaxel and/or carboplatin for 2 days, and then the drugs were removed to allow recovery of the remaining cells (Fig. 1A). Upon treatment for 2 days with paclitaxel and/or carboplatin, cell numbers were reduced to 40% to 60% of the starting values in day 1, and results are shown for A2780, OVCAR3, and OVCAR5 cells (Fig. 1B to D). Cell growth recovered rapidly after removal of carboplatin, but paclitaxel-treated cells were not able to recover and the cell number continued to decline even 4 days after paclitaxel was removed (Fig. 1B to D). Thus, while cell growth recovers after carboplatin is removed, growth suppression is irreversible and cell death occurs with paclitaxel treatment.

FIG 1.

Paclitaxel induces multimicronucleation and ovarian cancer cell death. (A) A protocol was designed to mimic administration of chemotherapeutic drugs: cells were plated for 1 day and treated with drugs (control, paclitaxel, and/or carboplatin) for the next 2 days, and then the drugs were removed to observe the recovery. In the experiments, a panel of ovarian cancer cell lines was exposed to paclitaxel (1 nM) and/or carboplatin (1 μM). (B to D) A2780 (B), OVCAR3 (C), and OVCAR5 (D) cells were treated according to the protocol. Cell numbers were determined by MTS-1 proliferation assay in triplicate and shown as the mean ± standard deviation. The results are representative of data from 4 similar experiments. (E) A2780 cells were treated with paclitaxel (1 nM) alone, carboplatin (1 μM) alone, or both paclitaxel and carboplatin for 2 days. The cells were analyzed by immunofluorescence microscopy for lamin B1 to visualize the nuclear envelope. (F) Primary normal HOSE cells, immortalized human ovarian HIO-80 cells, and cells from ovarian cancer cell lines (OVCAR3, -4, and -5) were cultured without (control) or with paclitaxel (1 nM) alone or in combination with carboplatin for 2 days and analyzed by immunofluorescence for lamin B1. (G) OVCAR3 cells were treated with paclitaxel (1 nM) for 0, 1, or 5 days (upper panel). Alternatively, after the cells were treated with paclitaxel for 2 days, the drug was removed and the cells continued to be cultured for an additional 1, 2, or 5 days (lower panel). The cells were then analyzed by immunofluorescence microscopy for lamin B1 and alpha-tubulin and counterstained for nuclear DNA with DAPI.

The nuclei of paclitaxel-treated cancer cells exhibited a unique morphology, where several smaller nuclei, rather than a contiguous envelope, were clustered together as shown in examples stained for the nuclear envelope protein lamin B1 (Fig. 1E and F). Carboplatin by itself had no effect on nuclear morphology and did not synergize with paclitaxel to increase nuclear abnormalities, as shown for A2780 cells (Fig. 1E), and is shown only in combination with paclitaxel for other cell lines (Fig. 1F). We coined the term “multimicronucleation” to describe the effect of paclitaxel on the ovarian cancer cells, to distinguish “nuclear fragmentation” as used in describing apoptosis. These aberrant nuclear features persisted for hours to days and differed from chromatin condensation and nuclear fragmentation that occur in cells undergoing apoptosis. Consistently, noncancerous cells, such as human ovarian surface epithelial (HOSE) and human immortalized ovarian (HIO) epithelial cell lines, were much less sensitive to paclitaxel-induced nuclear changes (Fig. 1F). Here, the doubling time for HOSE cells was 48 to 72 h and cell density was sparse in culture, though the simian virus 40 (SV40)-“immortalized” cells were more proliferative, as described previously (34).

Following treatment with paclitaxel for even 1 day, the nuclei of cancer cells were clearly multimicronucleated and became progressively more extensive with time (Fig. 1G, upper panel). The nuclei remained aberrant following the removal of paclitaxel for several days, as shown in an example for OVCAR3 cells up to 5 days after paclitaxel was removed (Fig. 1G, lower panel).

Thus, exposure to paclitaxel causes nuclear breakage in cancer cells but not in noncancer cells. The fragmented, lobulated nuclei that result from paclitaxel treatment likely account for cell growth suppression and death that continue after the removal of the drug. We speculate that differential sensitivity of a nucleus to paclitaxel-induced multimicronucleation is a factor in selectivity for paclitaxel in cancer versus benign cell types. Analyses of the impact of paclitaxel over time indicated that paclitaxel induced nuclear changes starting around 3 h after addition, and the multimicronucleated cells accumulated in a time-dependent manner and reached nearly 100% by 2 to 3 days. Additionally, no substantial difference was noted in the effects on cells for paclitaxel concentrations ranging from 1 to 100 nM. Likely, this is because that paclitaxel is taken up and concentrated several hundred-fold inside cells, due in part to sequestration by microtubules (polymerized tubulins), where concentrations are estimated to be around 10 to 20 μM within cells (37).

Multimicronucleation associates with paclitaxel-induced microtubule stabilization and bundling.

The observed multimicronucleation of neoplastic cells was a unique property of paclitaxel, which stabilizes microtubule filaments but not other microtubule or beta-actin modulating agents (Fig. 2A). As shown in OVCAR5 cells as an example, paclitaxel induced nuclear breaking; however, nocodazole, which destabilizes microtubules, and jasplakinolide, which promotes actin polymerization (38), had no significant impact on nuclear morphology compared to that of paclitaxel (Fig. 2A). As a control, paclitaxel induced microtubule filament bundling in noncancerous HIO80 immortalized human ovarian epithelial cells, though the nuclei appeared distorted but not fragmented (Fig. 2B). All ovarian cancer cell lines tested had similar responses to these actin and microtubule interfering agents. In comparison, paclitaxel consistently induced multimicronucleation (Fig. 2C). Following paclitaxel treatment, microtubules showed strong bundles and became disorganized, showing no obvious microtubule organizing center, as shown by immunostaining (Fig. 2C).

FIG 2.

Microtubule stabilization associates with multimicronucleation. (A) OVCAR5 ovarian cancer cells were treated with paclitaxel (1 nM), nocodazole (100 nM), and jasplakinolide (10 nM) for 24 h and then analyzed by immunofluorescence microscopy. The upper panel shows only the lamin B staining to visualize the outline of the nuclear envelope. The lower panel shows the merged images for lamin B, alpha-tubulin, and DAPI. (B and C) Nontumorigenic (human immortalized ovarian surface epithelial) HIO80 (B) or OVCAR5 (C) cells were treated with paclitaxel (1 nM) for 24 h. The cells were analyzed by immunofluorescence microscopy for lamin B and alpha-tubulin, and the nuclei were counterstained with DAPI. (D) Ovarian cancer cells were treated with paclitaxel (10 nM) or epothilone (5 nM) for 24 h. The cells were analyzed by immunofluorescence microscopy for lamin B and alpha-tubulin, and the nuclei were counterstained with DAPI.

To further verify if the observed multimicronucleation was the consequence of microtubule stabilization caused by paclitaxel, we tested epothilone B, a compound with similar activity to that of paclitaxel in stabilizing microtubules. The results indicated that epothilone B induced microtubule changes and produced a similar degree of nuclear change as paclitaxel (Fig. 2D). Thus, multimicronucleation is a consequence of microtubule stabilization.

As shown at higher magnification of both nonconfocal and confocal microscopy imaging in OVCAR3 cells, the paclitaxel-stabilized microtubule filaments closely associated with fragments of the nucleus broken away from the main nuclear body (Fig. 3A and B). The micronuclei appeared to be physically attached to the rigid microtubule bundles, though the images cannot resolve the linkage mechanism. In some cases, microtubule bundles were observed to surround each of the micronuclei formed (Fig. 3C). Thus, a likely interpretation is that the breaking up of nuclear envelope to form nuclear fragments by paclitaxel is the result of stabilization of microtubule filaments, which then physically pull on and apart the malleable cancer nucleus.

FIG 3.

Paclitaxel-induced microtubule bundles associate with nuclear fragments. Examples of OVCAR3 ovarian cancer cells treated with paclitaxel (1 nM) for 48 h were shown in high magnification. (A) The cells were analyzed by immunofluorescence microscopy for lamin B1 and alpha-tubulin, and the nuclei were counterstained with DAPI. Wide-field images are shown. (B) The cells were analyzed by confocal immunofluorescence microscopy for lamin A/C and alpha-tubulin, and the nuclei were counterstained with DAPI. (C) Another example of OVCAR3 cells treated with 10 nM paclitaxel for 2 days shows that microtubules surround each of the micronuclei generated.

Lamin A/C deletion sensitizes cells to paclitaxel-induced nuclear breakage.

We reasoned that lamin A/C-negative (or low) cells have a malleable nuclear envelope and that the nuclei are more predisposed to breaking when microtubules are rigid, while lamin A/C proteins strengthen the envelope to reduce deformation by force. Thus, a nuclear envelope structure defect caused by the loss/reduction of structural protein lamin A/C may accentuate nuclear deformation and fragmentation exerted by force from rigid microtubule bundles in the presence of paclitaxel. To test this idea, we examined the sensitivity of lamin A/C-null cells to paclitaxel for nuclear deformation. We produced and verified a cre-lox mediated conditional knockout mouse line using three clones of lamin A (lmna) floxed targeted ES clones from EUCOMM. The conditional knockout was determined so that the mutant lmna allele could be deleted by the Sox2-cre transgene in cultured cells and in live transgenic mice to generate lamin A/C-null allele (39) (Fig. 4A). These lmna^df/df mice have a similar phenotype to that of the previously reported knockout (39, 40). The freshly isolated lamin A/C-null mouse ovarian surface epithelial (MOSE) cells, though retarded for growth in culture, underwent significant nuclear shape perturbation with paclitaxel treatment (Fig. 4B and C). We also isolated MOSE cells from the p53R172H mutant mouse, which is the structural mutant homologous to human p53 codon 175 null mutant. Paclitaxel also induced multimicronucleation in the p53 M/M cells but not to the same extent as in the lamin A/C-null cells (Fig. 4B and C). In contrast, control MOSE cells isolated from wildtype littermates were largely resistant to paclitaxel-induced nuclear changes (Fig. 4B and C). It should be noted that the lamin A/C-null cells exhibited nuclear shaped deformation even in the absence of paclitaxel, and paclitaxel exacerbated the degree of deformation. Since the primary lamin A/C-null MOSE cells grew poorly in culture and were limited (35), we generated lamin A/C-null and p53R172H double mutant cells for further study of the lamin A/C-negative cells, since the p53 mutation immortalizes the cells in culture (Fig. 4D). Deletion of lamin A/C showed little effect on lamin B protein level, though expression of cre led to high expression and accumulation of the mutant p53 protein (Fig. 4D). Lack of induction of cyclin inhibitor p21, the downstream effector of p53, indicated the loss of function for the p53 mutant protein (Fig. 4D). The nuclei from isolated lamin A/C-null and p53-mutant MOSE cells became highly sensitive to paclitaxel (Fig. 4E to G). Following treatment with paclitaxel but not carboplatin, most cells contained a larger main nuclear body surrounded by several (to many) smaller micronuclei (Fig. 4E), where 90% to 95% of the cells were quantified to contain multiple nuclei and micronuclei following treatment with paclitaxel or paclitaxel plus carboplatin (Fig. 4F). In some examples (Fig. 4G), paclitaxel caused the breakage of nuclei into numerous smaller spheres to an extent similar to those seen in human ovarian cancer cells exposed to paclitaxel.

FIG 4.

Deletion of lamin A/C sensitizes mouse ovarian surface epithelial cells to paclitaxel-induced multimicronucleation. (A) MOSE cells were obtained from lamin A (lmna) floxed control and Sox2-cre-mediated deleted (+Sox2-cre) mice, and lysates were analyzed for the presence of lamin A/C by Western blotting. Beta-actin was used as a protein loading control. The mutant lmna allele was efficiently deleted by the Sox2-cre transgene in vivo. (B) Primary MOSE cells isolated from wild-type (WT), conditional knockout lamin A/C-null [lmna (−/−)], and p53 mutant [p53 (M/M)] mice were cultured and treated with paclitaxel (1 nM) for 24 h. The nuclear morphology was observed by immunostaining for lamin B1. (C) Cells with multilobular nuclei were quantified in 3 samples each and shown as the mean ± standard deviation. Paclitaxel induced the greatest changes in lamin A/C-null and p53 mutant MOSE cells, as indicated; *, P = 0.007, **, P = 0.0001, and ***, P = 0.0005. (D) MOSE cells were obtained from female conditional lamin A/C-null p53 mutant mice, and isolated cells were treated in culture with Adv-Cre (+ cre) to mediate the gene deletion. Lysates from cells were analyzed by Western blotting for protein expression, as indicated, and normalized to actin levels. “*” indicates a nonspecific band recognized by the p53 antibody. (E) Primary lmna^−/− p53 mutant (M) MOSE cells were treated with paclitaxel (Pacl, 1 nM) and/or carboplatin (Carb, 1 μM) for 24 h. The nuclear morphology was observed by staining for lamin B and DAPI. (F) The effect of treatment with paclitaxel, carboplatin, or combined paclitaxel and carboplatin on multimicronuclei presence is shown as the mean ± standard deviation of 3 samples each, with significance found for paclitaxel treatment as indicated; *, P = 0.0001 and **, P = 0.005. (G) A representative example is shown for primary lmna^−/− p53 mutant MOSE cells treated with paclitaxel (Pacl, 1 nM) for 48 h. The nuclear morphology was observed by staining for lamin B1 and DAPI.

Expression of lamin A protein enables resistance to paclitaxel-induced nuclear fragmentation.

Previously, we have shown that the nuclear lamina protein, lamin A/C, is absent or low in most ovarian cancer cell lines but is strongly expressed in HOSE cells (34). The loss and reduction of lamin A/C proteins in A2780 and OVCAR3, -4, -5, and -10 cells were shown by Western blotting and reported previously (34).

To test if an increase of lamin A/C could prevent paclitaxel-induced nuclear breakage, several lines of ovarian cancer cells (OVCAR3, -5, and -8 and A2780) were transfected with lamin A-red fluorescent protein (RFP). In multiple repeated transfections, the cells expressing lamin A-RFP ranged from 10% to 30% of the cell population. Following treatment of the mixed transfected cell populations with paclitaxel for 24 h, lamin A-RFP-expressing cells had a much lower percentage of fragmented nuclei compared to that of the population of nonexpressing cells (Fig. 5). In a representative example of OVCAR3 cells, around 92% of lamin A-RFP-negative cells showed nuclear breakage, compared to 23% of cells expressing lamin A-RFP (Fig. 5A and B). Similar results were observed in A2780 (Fig. 5C and D) and other ovarian cancer lines. As controls, the same plasmid vector, which was used to express histone H2B-GFP, did not protect the cells from paclitaxel-induced multimicronucleation. Thus, the transfection and overexpression experiments indicate that high expression of lamin A enables resistance of the cancer cells to paclitaxel-induced nuclear multimicronucleation. Thus, we conclude that the loss of lamin A/C likely sensitizes the nucleus to paclitaxel-induced breaking and the level of lamin A/C expression may determine the resistance of cells to paclitaxel.

FIG 5.

Overexpression of lamin A/C confers paclitaxel resistance. (A) OVCAR3 cells were transfected with lamin A-red fluorescence protein fusion construct (RFP-LA), treated with paclitaxel (1 nM) for 24 h, and analyzed for lamin B, lamin A/C, and RFP. (B) The percentage of nuclei that were fragmented or intact in control and RFP-lamin A-expressing cells were estimated in >50 cells, with error bars indicating standard deviations. Paclitaxel induced significant nuclear fragmentation in control (P = 0.0001) and RFP-lamin A-expressing cells (P = 0.002). Similar results were obtained in experiments performed in A2780 cells (C and D). Quantitation indicates expression of RFP-lamin A rendered the cells significantly resistant to paclitaxel-induced nuclear breaking, where * indicates P = 0.0016 for OVCAR3 in panel B and ** and P = 0.002 for A2780 cells in panel D.

Paclitaxel-induced multimicronucleation occurs independently of cell proliferation.

To test if mitosis is essential for paclitaxel-induced nuclear fragmentation, cell cycle inhibitors and a serum-free culture condition were used to prevent the initiation of mitosis. Using a WST-1 proliferation assay, we determined that the growth of ovarian cancer cells (OVCAR3 and -8 and A2780) largely halted over the 3-day period during which the cells were deprived of serum. In the absence of serum, the A2780 cell number decreased by day 4 (Fig. 6A), and the growth of OVCAR3 and OVCAR8 cells was reduced to about 10% of that determined in the presence of serum (Fig. 6B and C). Addition of paclitaxel caused further reduction in cell numbers in all lines (Fig. 6A to C). In the absence of serum and cell proliferation, however, paclitaxel caused extensive nuclear fragmentation to the same degree or more than in the presence of serum. In either condition, 50% to 70% of all nuclei showed severe multimicronucleation in all cell lines treated with paclitaxel (Fig. 6D), as shown by representative examples (Fig. 6E). The results indicate that cell division/mitosis is not essential for paclitaxel to invoke nuclear breakage.

FIG 6.

Paclitaxel induces a similar degree of multimicronucleation in cells grown in medium with or without serum. (A to C) OVCAR3, OVCAR8, and A2780 ovarian cancer cells were cultured in control (10% serum) or serum-free medium for 24 h and then treated with paclitaxel (Ptx, 10 nM) (day 2) in medium without serum for another 48 h (day 4). MTS-1 proliferation assay was used to determine the relative cell growth for the different conditions in triplicate, with standard deviations indicated. (D) At day 4, the cells were imaged by immunofluorescence microscopy for lamin B1 staining to visualize nuclear morphology. The percentages of intact and fragmented nuclei were quantified in 3 randomly taken 20× immunofluorescence images, with error bars indicating standard deviations. The differences in nuclear breakage between cells incubated with or without serum are not significant (P > 0.05) in all lines. (E) Representative images of OVCAR3, OVCAR8, and A2780 ovarian cancer cells are shown for nuclear morphology of lamin B1 staining.

We also tested several mitotic inhibitors with various mechanisms of action, including hydroxyurea (ribonucleotide reductase inhibitor), methotrexate (dihydrofolate reductase inhibitor), mitomycin C (bis-electrophilic intermediate for DNA alkylation), and aphidicolin (DNA α/δ polymerase inhibitor), to determine if paclitaxel-induced fragmentation persists when mitosis is prevented. The effectiveness of these mitotic inhibitors to inhibit cell proliferation was first determined by treating cells at increasing dosages, and the optimal concentration was established as the lowest that significantly diminished growth of the cancer cell lines. For hydroxyurea, methotrexate, mitomycin C, and aphidicolin, the optimal concentrations were 2.5 mM, 50 μM, 2 μM, and 10 μM, respectively, as shown in OVCAR3 cells (Fig. 7A to D). All other lines had similar dose responses to those of these mitotic inhibitors. In several subsequent experiments, ovarian cancer cell lines (OVCAR3 and -8 and A2780) were treated with the optimal concentrations of the mitotic inhibitors and paclitaxel; paclitaxel was added after the cells were first pretreated with mitotic inhibitors for 1 day, and the cells and nuclear morphology were analyzed followed paclitaxel treatment for 2 days (mitotic inhibitors for 3 days). In these experiments, multimicronucleation was significant but variable, ranging from 10% to 44%, in cells treated with paclitaxel alone or with mitotic inhibitors (Fig. 7E), as shown in examples of nuclear lamin B1 staining in the absence (Fig. 7F) or presence (Fig. 7G) of paclitaxel. None of the mitotic inhibitors (hydroxyurea, methotrexate, mitomycin C, and aphidicolin) alone caused significant nuclear changes (Fig. 7F). Generally, addition of mitotic inhibitors reduced the percentages of cells with fragmented nuclei induced by paclitaxel, and especially the differences reached statistically significant for mitomycin C (MMC) in all three cell lines analyzed (Fig. 7E). Nevertheless, a significant degree of paclitaxel-induced multimicronucleation was observed in all cases, in which cell growth and mitosis were essentially eliminated. Therefore, we conclude that paclitaxel is able to induce multimicronucleation through a mitotis-independent mechanism.

FIG 7.

Paclitaxel induces multimicronucleation in the presence of mitotic inhibitors. (A to D) A2780 ovarian cancer cells were examined for growth suppression over a dose range of mitotic inhibitors. The cells were cultured in medium for 4 days in the presence of various concentrations of hydroxyurea (HU) (A), methotrexate (MTX) (B), mitomycin C (MMC) (C), or aphidicolin (APC) (D). The relative cell number was determined by MST-1 proliferation assay with error bars indicating standard deviations. Similar dose curves of the mitotic inhibitors were found for several additional cell lines tested, including OVCAR3, -5, and -8 and A2780. (E) OVCAR3 and -8 and A2780 ovarian cancer cells were treated with hydroxyurea (HU, 2.5 mM), methotrexate (MTX, 50 μM), mitomycin C (MMC, 2 μM), or aphidicolin (APC, 10 μM) for 24 h, and then paclitaxel (ptx, 10 nM) was added for another 48 h in the presence of the same mitotic inhibitors. After incubation of the cells with mitotic inhibitors for 3 days and with or without paclitaxel for the last 2 days, nuclear morphology was then captured by immunofluorescence staining of lamin B1. The percentages of intact and fragmented nuclei were quantified in 3 randomly taken 20× immunofluorescence images, with standard deviations indicated. The differences in nuclear breakage between paclitaxel and combination with mitomycin C were found to be statistically significant (*), with the P value to be 0.011 for A2780, 0.010 for OVCAR3, and 0.02 for OVCAR8. The differences are also statistically significant (*) for hydroxyurea (P = 0.014) in OVCAR3 and for hydroxyurea (P = 0.002) and APC (0.002) in OVCAR8 cells. The differences are not significant (P > 0.05) in the rest of the cases. (F) Representative nuclear images of OVCAR3, OVCAR8, and A2780 ovarian cancer cells treated with mitotic inhibitors are shown for lamin B1 staining. (G) Representative nuclear images of OVCAR3, OVCAR8, and A2780 ovarian cancer cells treated with mitotic inhibitors plus paclitaxel are shown for lamin B1 staining.

DISCUSSION

The results lead us to suggest that the more malleable nuclei of neoplastic, lamin A/C-low or lamin A/C-negative cells are more susceptible to breakage by the force of microtubule filaments, especially when microtubule bundles are stabilized and rigid by persistent paclitaxel binding (Fig. 8). Based on these findings, we propose that, in addition to interfering with microtubule dynamics and consequently mitosis, another mechanism for the anticancer activity of paclitaxel is to cause physical breaking of the more malleable cancer nucleus into fragments/micronuclei. With the force of rigid microtubule bundles stabilized by paclitaxel, fragments of nuclear envelope are pulled off through the connecting LINC complexes (36, 41–43). It appears that the thinner and softer actin microfilaments are not sufficiently robust to cause physical breaking of nuclear envelope. We reason that the loss or reduction of the nuclear lamina protein lamin A/C provides another selectivity (in addition to increased mitosis) for paclitaxel toward neoplastic cells compared to that toward benign cells.

FIG 8.

A nuclear envelope structural defect is common in ovarian carcinomas and is also a key determinant of sensitivity to paclitaxel and microtubule-stabilizing agents. Based on preliminary studies in our lab, we suggest a mechanism for paclitaxel killing of cancer cells, that paclitaxel induces nuclear breakage and the formation of micronuclei in malignant cancer cells that have a more malleable nuclear envelope. Additionally, we propose that the level of nuclear lamin A/C protein (which inversely correlates with the degree of nuclear shape deformation) is a determinant of paclitaxel (and other microtubule-stabilizing agents) sensitivity in cancer cells. (A) In normal, lamin A/C-positive cells, paclitaxel stabilizes microtubules and the rigid microtubule bundles cause only limited deformation of the nuclei. (B) However, cancer cells have a more malleable nucleus (as a result of loss of lamin A/C or other nuclear envelope structural proteins), and paclitaxel (and other microtubule-stabilizing agents) induces the formation of rigid and disorganized microtubule bundles, which causes fragmentation and breakage of the attached nucleus and ultimately kills the malignant cells (but not normal cells).

Although it is generally accepted that paclitaxel kills cancer cells by altering microtubule dynamics, which then induces mitotic arrest and subsequent apoptosis (11, 12, 19), paclitaxel has been observed and reported to induce the formation of multiple micronuclei (44, 45). Here, we describe that a dramatic impact of paclitaxel on cancer cells is the physical breaking of nuclei, so-called the multimicronucleation. Previously, a report described the observation of aberrant, multipolar mitotic division and suggested that paclitaxel kills cancer cells by inducing aberrant chromosome distribution in daughter cells rather than by mitotic arrest (13–15). Multipolar division could account for some fraction of the observed “multimicronucleation.” However, the distinct nuclear morphology suggests that breaking off nuclear buds in nonmitotic cells is another mechanism for the multimicronucleation induced by paclitaxel. Based on experiments described here using serum-free conditions or mitotic inhibitors added prior to paclitaxel treatment, about half of the nuclear changes are due to nonmitotic mechanisms and half are consequences of aberrant mitosis. The effect of paclitaxel in nonmitotic cells is consistent with the idea that paclitaxel functions in both mitosis and interphase of the cell cycle for its clinical activity in cancer therapy (16, 17). In many of the cases examined here and elsewhere (13–15), the cells of fragmented nucleus are assumed to subsequently undergo apoptotic cell death; however, the death mechanism of the nuclear fragmented cells has not yet been thoroughly analyzed. One idea is that cell innate immunity pathways activated by the multiple micronuclei may be involved (46). The formation of multiple micronuclei also leads to DNA damage, which is inefficiently repaired in micronuclei (47). Furthermore, it is possible that a few of those multinucleated cells that survive evade immune response pathways and develop into a highly resistant, heterogenic cancer population, because mitotic checkpoints and DNA repair mechanisms are out of sync and cannot mount an effective response to repair the damaged DNA in micronuclei (47, 48). This is a possible mechanism that resistance commonly develops in the recurred cancer and why chemotherapy ultimately fails.

The specificity of paclitaxel to kill malignant but not noncancerous cells includes the high proliferation rate of cancer cells, as paclitaxel targets microtubule functions in mitosis (9–11, 19). Our current finding adds an additional mechanism regarding specificity: the malleability of cancer nuclei due to structural changes, such as from the loss/reduction of lamin A/C protein, makes the malignant cells more susceptible than nonneoplastic cells to be killed by paclitaxel. It is well documented that nuclear envelope rupture is determined by the structure and strength of the nuclear envelope, and the major contributors in regulating structure and integrity are lamins (47). Most of the strength versus deformability is contributed by the A/C-type lamins, whereas B-type lamins have less effect on nuclear stiffness (49, 50). However, since B-type lamins are membrane tethers because of the C-terminal farnesylation that anchors them in the inner nuclear membrane, they remain with the membrane during mitosis (47). B-type lamins may also contribute to nuclear membrane disruptions during interphase, where breaks in the lamina may associate with micronuclei containing chromatin and cytoplasmic membrane blebbing (51). In our studies, lamin B levels appeared to remain consistent in the cell lines we examined (34), but differential expression of lamins and their contribution to micronuclei formation and paclitaxel sensitivity are issues that need to be considered further (47). Moreover, since both lamin A/C and lamin B deficiencies have noted consequences on DNA repair, their aberrant expression potentially affects multiple tumor-promoting mechanisms (47, 48). Our current study adds that the lamins may also play roles in evading cell death and development of drug resistance in chemotherapy.

Particularly, the current finding provides a feasible explanation for the suspected nonmitotic mechanisms of paclitaxel in anticancer activity (16, 17, 22). The physical breaking up of the nucleus with paclitaxel treatment causes death of the cells and is also consistent with the idea that paclitaxel-mediated killing is p53-independent (12). This is important for an active drug to treat ovarian cancer since the majority of high-grade serous carcinomas have a mutated p53 (52).

In summary, we report here a mechanism for the cytotoxicity of paclitaxel toward ovarian cancer cells, that paclitaxel causes drastic physical breaking up of nuclei in malignant ovarian cancer cells but not in normal cells. We conclude that the malleable cancer nucleus, which results often from reduced lamin A/C protein, is a key determinant for sensitivity of malignant cells to the microtubule-stabilizing agents. Thus, increased lamin A/C expression is a possible mechanism for resistance to paclitaxel, and lamin A/C expression may be a marker for drug resistance. As paclitaxel is a key drug in the current treatment of ovarian and several other major cancer types, the new understanding allows rational design of drug combination for cancer treatment, reveals a potential drug resistant marker and mechanism, and provides a cancer cell killing strategy for further development of new anticancer agents that act by physically breaking the malleable cancer nuclear envelope.

MATERIALS AND METHODS

Reagents.

Paclitaxel, epothilone B, nocodazole, jasplakinolide, carboplatin, and S-phase inhibitors (hydroxyurea, methotrexate, mitomycin C, and aphidicolin) were purchased from Sigma-Aldrich, Inc. Cdk4/6 inhibitor IV (CAS 359886-84-3) was from Calbiochem. Stock solutions were made in dimethyl sulfoxide (DMSO) for carboplatin (100 μM), paclitaxel (100 μM), and the S-phase inhibitors methotrexate (5 mM) and aphidicolin (25 mM). Stock concentrations of hydroxyurea (0.2 M) and mitomycin C (10 mM) were made in sterile distilled H₂O. Tissue culture flasks (trademark Falcon), tissue culture media, trypsin, and 100× antibiotic-antimycotic solution (Cellgro, Mediatech, Inc.) were purchased from VWR (Springfield, NJ). TRIzol reagent and Lipofectamine 2000 transfection reagent were purchased from Invitrogen (Thermo Fisher Scientific) (Carlsbad, CA). For immunofluorescence microscopy, Alexa Fluor 488- and 555-conjugated secondary antibodies were purchased from Life Sciences (Eugene, Oregon). Primary antibodies: anti-lamin A (1:400, H-102, rabbit polyclonal IgG), mouse monoclonal anti-lamin B (1:300, sc373918), and goat polyclonal anti-lamin B (1:400, sc6216) were from Santa Cruz Biotechnology Inc.; rabbit polyclonal anti-lamin B1 (1:1,000, ab16048) and mouse monoclonal anti-alpha-tubulin (1:500, 66031) were from Abcam and Proteintech, respectively. A mouse monoclonal anti-alpha-tubulin was also purchased from Sigma-Aldrich, Inc. (St. Louis, MO).

Cell culture.

Primary human ovarian surface epithelial (HOSE) isolated and provided by Andrew K. Godwin (Fox Chase Cancer Center) were cultured in medium containing 6 g/liter HEPES, 15% fetal bovine serum (FBS), 1× antibiotic-antimycotic, and insulin, as reported previously (34). Three lines of human “immortalized” ovarian (HIO) epithelial cells were obtained by transfection of the HOSE cells with SV40 to prolong their life span in culture (34). Ovarian cancer cells, including A2780, OVCAR3, -4, -5, -8, and -10, were used in the experiments. These cells were cultured in RPMI 1640 medium supplemented with 10% FBS and 1× penicillin-streptomycin, as described previously (34). For serum-free experiments, cells were cultured in RPMI with 1% bovine serum albumin (BSA) and 1× penicillin/streptomycin.

Primary mouse ovarian surface epithelial (MOSE) cells were isolated from 2 to 3 ovaries of mice of 1 to 6 months of age by limited digestion with 0.25% trypsin-EDTA solutions (CellGro) at 37°C for 30 min, as described previously (34, 35). The cells released were harvested for culturing and were found to be more than 90% epithelial in origin as characterized by cytokeratin staining. These primary cells were used for experiments following a brief culture in Dulbecco modified Eagle medium (DMEM) containing 4% FBS, 1× insulin-transferrin-selenium (ITS), 1× nonessential amino acids, and 1× antibiotic-antimycotic solution and expansion for 4 to 7 days. All cells were maintained at 37°C in a humidified atmosphere of 5% CO₂.

Proliferation of cells in culture was analyzed using the cell proliferation reagents WST-1 kit from Roche, which measures the conversion of the tetrazolium salt WST-1 into a soluble formazan dye by mitochondrial dehydrogenase enzymes. The production of the formazan dye correlates with the number of metabolically active cells in culture, and detection is measured spectrophotometrically at 450 nm. Since cells are not directly counted, the results are expressed as the relative cell number over time. Cells were plated at 2.5 × 10³ cells in 96-well dishes in triplicate, the assay followed the manufacturer’s protocol, and experiments were repeated twice.

Generation of lmna (lamin A/C gene) knockout mice and cells.

We generated and confirmed a line of lamin A/C conditional knockout mice from 3 clones of lamin A (lmna) floxed targeted ES clones from EUCOMM. We verified that the mutant lmna allele can be deleted by Sox2-cre transgene and generated lamin A/C-null mice (39). These lmna (df/df) mice have a similar phenotype to that of the previously reported knockout (40). The mutant mice are growth retarded and die around 4 to 6 weeks of age due to muscular dystrophy that causes feeding and heart deficiencies. Heart muscles of the lmna (df/df) mice contain abnormal nuclei, and the deformed nuclear morphology is apparent in isolated cells (38, 39).

The mutant mice were crossed with Tp53R172H mutant (fl/fl) (p53M) mice to generate lmna^fl/fl; p53M^fl/fl mice. The p53 mutant mice were obtained from Jackson Lab (strain number 129S-Trp53^tm2Tyj/J; stock number 008652; also known as p53LSL.R172H 129svj) and maintained in the C57BL/6J background (Jackson Lab; stock number 000664). These mice carry a conditional point mutant allele of the transformation-related protein 53 gene (p53R172H, which is the structural mutant homologous to human p53 codon 175). The conditional allele is functionally equivalent to a null mutation. Cre-mediated recombination leads to deletion of a transcriptional termination sequence (Lox-Stop-Lox) and expression of the oncogenic protein.

Ovarian surface epithelial cells (MOSE) were prepared, and gene deletion was mediated using Adv-Cre to transfect the MOSE in culture. The lamin A/C-deficient MOSE cells had reduced ability for proliferation, but the growth ability was restored by p53 mutation (35, 39). These lmna^df/df; p53^df/df MOSE cells became immortalized in culture (over 30 passages now) and were tumorigenic when implanted in mice (35, 38). All experiments using mice were performed in accordance with federal and state guidelines for the ethical and humane use of laboratory animals, and the experimental animal usage was reviewed and approved by the University of Miami Institutional Animal Care and Use Committee (IACUC).

Immunofluorescence microscopy.

For immunofluorescence microscopy, adhered cells on glass coverslips were washed twice with phosphate-buffered saline (PBS) at room temperature, fixed with 4% paraformaldehyde for 15 min, and permeabilized with 0.1% Triton X-100 for 5 min. The cells were washed 3 times with PBS, blocked with 5% BSA in PBS containing 0.1% Tween 20 for 1 h, and incubated overnight at 4°C with primary antibodies in 5% BSA in PBS. Alexa Fluor 488-conjugated (green fluorescence) or Alexa Fluor 555-conjugated (red fluorescence) secondary antibodies were used. Cells were washed 3 times, counterstained with DAPI (4′,6-diamidino-2-phenylindole), and then mounted and sealed in ProLong Gold antifade reagent (Thermo Fisher Inc.). Immunofluorescence stainings were viewed with widefield microscopy using a Plan-Apochromatix 100× objective lens (oil immersion, numerical aperture [NA] 1.4) on inverted Zeiss Axio Observer Z1 using AxioVision 4.8 software. As indicated, some immunofluorescence stainings were viewed using a Plan-Apochromat 63× oil immersion objective lens (NA 1.4). Images were acquired using a monochrome Zeiss Axio Cam MRm CCD camera. Confocal imaging was performed on a Zeiss LSM.510/uv Axiovert 200M inverted laser scanning confocal microscope operated by Zeiss Zen software using the plan-Apochromat 63×/1.4 NA oil DIC M27 objective.

Expression of RFP-lamin A in ovarian cancer cells.

RFP-lamin A plasmids were purified by maxi-prep according to the manufacturer’s protocol (Qiagen). RPF-lamin A was constructed by inserting lamin A cDNA generated by PCR with a 5′-XhoI restriction site and 3′-BamHI restriction site into the RFP-C1 vector; the RFP-C1 vector was derived from the enhanced green fluorescent protein (EGFP)-C1 plasmid (Clontech, now TaKaRa Bio USA, Inc.), with red fluorescent protein (RFP) inserted into the green fluorescent protein (GFP) site, as described previously (53). Ovarian cancer cells were plated at 2.25 × 10⁵ cells per well on glass coverslips in 12-well dishes, and the following day cells were transfected with 1.5 μg/well RFP-lamin A plasmid using Lipofectamine 2000. One day after transfection, paclitaxel (5 to 10 nM final) was added and cells were incubated for two additional days. Cells were fixed and immunostained according to standard protocol (above) using the primary antibodies rabbit anti-lamin B1 pAb (Abcam ab16048, 1:1,000 dilution) and mouse anti-lamin A/C MAb (Santa Cruz sc-7292, 1:200 dilution) and secondary Alexa Fluor 488-conjugated anti-rabbit and Alexa Fluor 647-conjugated anti-mouse antibodies. The number of cells positive for RFP-lamin A was calculated as a percentage of the total cells present in five fields per sample; experiments were repeated twice with similar results.

Statistical methods.

The Student’s unpaired t test calculator offered online from GraphPad was used to compare and test for significance between two groups, where P < 0.05 is considered statistically significant. Results are plotted as mean ± standard deviation (SD). Figures were plotted using Microsoft Excel and assembled using Microsoft Photoshop CS6.