Taxol Affects Nuclear Lamina and Pore Complex Organization and Inhibits Import of Karyophilic Proteins into the Cell Nucleus¹ Free

Taxol (Paclitaxel) is a plant alkaloid that is commonly used in the treatment of human carcinomas (1). It acts by stabilizing the cellular microtubules, rendering them rigid and less dynamic (2, 3). The pharmacological effects of the drug vary, depending on dose and treatment scheme. When administered at low concentrations (10–100 nm), Taxol induces mitotic arrest (4, 5), inhibits protein prenylation (6), and leads, eventually, to apoptosis (5, 6, 7),_. At micromolar doses, it also promotes synthesis and release of TNF-α³ (8, 9), expression of interleukin 1 (10) and interleukin 8 (11), and activation of early response genes (12).

Recent observations show that Taxol activates the Raf-1 kinase (13) and promotes phosphorylation of bcl-2, a “guardian of microtubule integrity” (14). Increased bcl-2 phosphorylation may weaken bcl-2/bax complexes, unleashing the latter and triggering apoptosis. In addition to that, Taxol and other antimicrotubule drugs are known to affect the shape of the cell nucleus. After exiting mitosis, cells treated with taxanes or Vinca alkaloids often develop lobulated nuclei and multiple micronuclei (15). Interestingly, similar defects are detected when mutants of the nuclear matrix protein NuMA (16) or the Ran/TC4 exchange factor RCC1 (17) are expressed in higher eukaryotic cells.

A major determinant of nuclear form and architecture is the nuclear envelope (18). This is a highly organized assembly comprising: the outer and inner nuclear membrane; the nuclear lamina meshwork, an intermediate filament system that is associated with the inner surface of the nucleus; and the NPCs. The outer nuclear membrane represents an extension of the endoplasmic reticulum, whereas the inner nuclear membrane constitutes a specialized environment that accommodates a unique set of proteins (LBR, LAP1s, and LAP2s). The nuclear lamina, is composed of A- and B-type lamins. These proteins form a polymeric lining that supports the inner nuclear membrane and imparts elasticity to the nuclear envelope. The NPCs are 125-MDa complexes containing 50–100 distinct polypeptides (nucleoporins). They provide the sole means for regulated transport between the cytoplasm and the nucleoplasm and are conserved in all eukaryotic cells, from human to yeast. The nuclear envelope disassembles during mitosis and reassembles at the end of cell division. A number of factors, including the cdc2 kinase and the microtubules, appear to play a role in this process, but the exact mechanism of reversible disassembly has not yet been elucidated (for review see Ref. 19).

To understand how antimicrotubule drugs affect the cell nucleus, we have examined the in situ organization of the NPCs and the nuclear lamina in human carcinoma cells treated with low concentrations of Taxol. Using functional assays, we have also assessed whether Taxol-treated cells could carry out nucleocytoplasmic transport in a normal fashion before they become apoptotic. Observations presented below show that Taxol induces striking nuclear envelope aberrations and compromises macromolecular traffic through the NPC.

Ishikawa cells were maintained in MEM, whereas HeLa cells were grown in DMEM. The various drugs, diluted in culture medium, were applied for 20 h at 37°C. After treatment, the cells were rinsed with fresh medium and incubated for 3–72 h under standard conditions. Viability was determined by trypan blue staining and hemocytometer counting.

Fixation, permeabilization with Triton X-100, and antibody staining were performed exactly as described previously (20, 21). The antibodies used included: an anti-α-tubulin monoclonal (clone DM1A), obtained from Sigma Chemical Co. (St. Louis, MO); the antivimentin polyclonal aV2 (22); the antikeratin monoclonal Lu-5, purchased from Boehringer (Mannheim, GmbH, Germany); the antinucleoporin monoclonal 414 (23); the antinucleoporin p68 autoimmune serum no. 27 (24); an anti-LAP2B polyclonal (21, 25); the anti-lamin B polyclonal no. 16 (20, 21, 22); the anti-lamin A polyclonal no. 163 (21, 26); the anti-all-lamin polyclonal aLI (27); the anti-lamin A/C monoclonal XB-10 (21); a polyspecific serum directed to multiple endoplasmic reticulum membrane proteins (28); and an anti-NFκB65 polyclonal purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Specimens were examined in a Zeiss epifluorescence and a Leica confocal microscope.

Monitoring of living cells after release from Taxol was accomplished using a computer-operated Leica microscope equipped with appropriate low-illumination optics and a stage chamber for culturing cells.

TUNEL assays were performed using an in situ cell death detection kit obtained from Boehringer Mannheim. Annexin V/propidium iodide assays were performed using a kit purchased from Genzyme/Techne (Cambridge, MA).

Passage of cells through the S phase was assessed by BrdUrd incorporation, using a kit from Boehringer Mannheim.

Western blotting and immunoprecipitation from total cell lysates were performed as specified previously (21).

Treatment of human endometrial carcinoma (Ishikawa) cells with 10 nm Taxol resulted in a sustained mitotic arrest. After a 20-h incubation, >70% of the cells possessed condensed chromosomes, disassembled NPCs, and depolymerized nuclear laminae (Table 1 and Fig. 1, b–d and b′–d′). A-type lamins were randomly scattered in the cytoplasm, whereas B-type lamins were relatively concentrated at the spindle poles (compare Fig. 1, c and d; for relevant information see Ref. 20). The geometry of the mitotic apparatus was minimally perturbed (mostly bipolar spindles), and the majority of the chromosomes were aligned at the cell equator (Fig. 1, a and a′). Consistent with a blockade at the metaphase-anaphase transition, prophase and telophase figures could not be detected.

Seven to 24 h after removal of Taxol from the medium, numerous telophase figures appeared in the fields (Fig. 1, e–h), whereas the percentage of metaphase cells was gradually reduced (Table 1). Abortive cytokineses were not evident, but sometimes a nuclear envelope formed around ectopic chromatin particles that had not segregated correctly (Fig. 1, e, e′, g, and g′, small arrows and arrowheads). Trypan blue staining showed that the number of live cells nearly doubles 24 h after release (Table 1), whereas video monitoring confirmed that these cells physically divide passing through anaphase and telophase (Fig. 2, a–d).

In line with the current literature, the cells possessed nuclei with variable morphologies. More specifically, 24 h after removing the drug, roughly half of the population contained oval-shaped, normal-looking nuclei, whereas the other half contained lobulated or multiple micronuclei (Table 1). To assess whether these cells progressed normally through the cell cycle or were arrested at the G₁ restriction point, we incubated the cultures with the base analogue BrdUrd. In the absence of DNA damage (as shown below in Fig. 5,a), incorporation of this compound into the nucleus indicates entry into the S phase and successful crossing of the G₁-S straits. Irrespective of nuclear phenotype, the majority of the cells incorporated BrdUrd within 24 h after removal of Taxol (Fig. 2, e–g). DNA replication by BrdUrd-labeled cells was confirmed by fluorescence-activated cell sorting analysis (data not shown). From these data, we conclude that, upon release from mitotic blockade, drug-treated cells reenter the cell cycle, effectively complete mitosis, and cross the G₁ checkpoint.

To examine whether the nuclear envelope of Taxol-treated cells was normally organized, we used specific antibodies recognizing NPC proteins (nucleoporins) or nuclear lamins. Immunostaining with antinucleoporin antibodies 24 h after release indicated that ∼50% of the cells contained abnormally arranged nuclear pores. Whereas the NPCs of nontreated cells were closely spaced, yielding a “rim” fluorescence pattern (Fig. 3,a), those of the treated cells were organized in large clusters with pore-free areas in between (Fig. 3, b and c). A more systematic survey of the specimens by confocal microscopy confirmed the existence of pore clusters at the nuclear periphery (Fig. 3, e–g) and further revealed that NPC material was sometimes deposited in the cytoplasm (Fig. 3,d). As indicated by morphometric data (Table 1), NPC clustering was more frequently seen among cells with lobulated nuclei and multiple mininuclei. Interestingly, defects in NPC distribution could be detected as early as 1 h after release from Taxol.

Staining with anti-lamin B antibodies revealed numerous “gaps” in the nuclear lamina meshwork of both normal-looking and irregularly shaped nuclei. The lamina deficits imparted a “moth-eaten” appearance to the nuclear lamina and could be better assessed by inspecting the specimens at different focal levels (Fig. 4, b–d and b′–d′). In general, the lamina deficits detected in micronuclei were more extensive than in other types of nuclei, yielding “crescent-like” patterns (Fig. 4,e). This kind of lesions were never detected in nontreated cells that possessed a continuous nuclear lamina meshwork investing precisely the “folds” of the nuclear surface (Fig. 3, a and a′). Although less obvious at a first glance, very small lamina gaps could also be discerned in cells that were in the process of exiting mitosis (Fig. 1, h and h′, arrows), suggesting that these aberrations develop early after release from mitotic blockade. Immunostaining with other antinuclear envelope antibodies yielded essentially the same results, showing lesions in the nuclear lamina, which contained both A- and B-type lamins and aberrant distribution of the integral protein LAP2 (data not shown).

To find out whether the alterations of the nuclear envelope were peculiar to the specific antimicrotubule agent we have been using, we repeated this analysis comparing the effects of Taxol with those of taxotere (a different taxane) and vinorelbine (a Vinca alkaloid derivative). In pilot experiments, we have determined that the cell cycle-arresting effect of 10 nm Taxol could be reproduced by 1–10 nm taxotere or vinorelbine, making such a comparison meaningful. The three antimicrotubule drugs induced nuclear envelope aberrations to almost the same extent (Table 1). Lamina deficits were detected in all types of cells, irrespective of nuclear morphology, whereas NPC clusters were more commonly seen in cells with multiple micronuclei or a single lobulated nucleus.

As has been described above, 24 h after removal of Taxol from the medium, the cells divide and pass through the G₁ checkpoint. However, as indicated in Table 1, the total number of cells remains constant thereafter, whereas the proportion of dead cells reaches the level of 60% 72 h after release. Prompted by these findings, we sought to determine at which point after release programmed cell death commences.

Twenty-four h after removal of Taxol, all cells were TUNEL-negative (Fig. 5a). Their nuclei, deformed or normal-looking, were free of apoptotic lobules, as could be judged by DAPI staining (Fig. 5,a′). Use of an annexin V assay and staining with propidium iodide, both of which detect early apoptotic phenomena (29, 30), fully confirmed these observations (data not shown). The same results were obtained when we examined cells that had been released for 48 h, the only difference being that a small proportion of them (∼10%) were TUNEL-positive (Fig. 5, b and b′). Eventually, most of the cells became TUNEL-positive 72 h postrelease (Fig. 5,c), as the chromatin condensed into lobules and nuclear architecture was abolished (Fig. 5 c′).

To assess when apoptotic degradation of nuclear envelope proteins is initiated, we continued with immunoprecipitation and Western blotting experiments. Cells treated with Taxol and released for 24 h, as well as controls, were extracted with 2% Triton-300 mm NaCl, and the solubilized proteins were immunoprecipitated with anti-lamin B antibodies. As shown in Fig. 6,a, comparable amounts of B-type lamins could be detected in precipitates from treated and untreated cells (compare Lanes N and 24). Western blotting analysis of total cell extracts (Fig. 6,c) and isolated nuclei (Fig. 6,b) derived from cells that had been released for 24 h did not reveal any change in the lamin B pattern, with respect to controls (compare Lanes N and 24). The same was observed with cells released for 48 h (Fig. 6,d, compare Lanes N and 48). However, extensive lamin degradation was noted 72 h after removal of Taxol (Fig. 6 d, compare Lanes N and 72), indicating activation of ICE proteases. From these and previous morphological results, we can safely conclude that the nuclear lamina and NPC abnormalities detected by immunofluorescence microscopy develop well before the advent of apoptosis.

Having established that antimicrotubule agents dramatically alter nuclear envelope organization, we then examined whether Taxol treatment affects macromolecular traffic through the NPC. To assess nuclear import in a controlled fashion, we used a well-established, in vitro system (31). Intact cells, either nontreated or treated with Taxol and released for 24 h into normal medium, were permeabilized by low concentrations of digitonin, an agent known to open the plasma membrane without affecting the integrity of the nuclear envelope. After washing out all cytosolic constituents, the resulting “ghosts” were supplemented with a Xenopus laevis egg extract (which contains soluble import factors), energy (ATP, GTP, and ATP-regeneration system), and FITC-labeled BSA coupled to the SV40 large T antigen NLS. As shown in Fig. 7,b, ghosts derived from nontreated cells efficiently imported the karyophilic ligand into the nucleus. However, ghosts prepared from Taxol-treated cells did not accumulate BSA-NLS into the nucleus under the same assay conditions (Fig. 7,c). Because, in both cases, the necessary soluble factors were provided exogenously, this result strongly suggests that it is NPC itself and not the cytosolic transport machinery that is primarily affected by Taxol treatment. Nuclear import under the conditions of the assay was specific and did not occur when Xenopus cytosol was omitted from the reaction (Fig. 7 a). The inability of Taxol-treated cell models to accumulate BSA-NLS into the nucleus was not due to nuclear envelope lysis during preparation of the ghosts. This could be confirmed by permeabilizing Taxol-treated cells with digitonin and probing with antibodies that recognize nucleoplasmically disposed antigens (e.g., lamins). In this setting, nuclear lamina staining was not observed, indicating that the nuclear membranes were intact (data not shown; for relevant data see Ref. 20).

On the basis of these in vitro findings, we further asked whether Taxol affects nucleocytoplasmic transport under in vivo conditions. To monitor import of macromolecules in living cells, we studied the partitioning of the transcription factor NFκB, which is rapidly translocated from the cytoplasm to the nucleus when cells are exposed to TNF-α. Taxol-treated cells and nontreated controls were first incubated in serum-free medium and then stimulated with 10 ng/ml human TNF-α. After 30 min, they were fixed, and the location of NFκB was determined by indirect immunofluorescence microscopy. Neither normal (Fig. 8, a and a′) nor Taxol-treated (Fig. 8, c and c′) cells imported NFκB in the absence of TNF-α stimulation. However, after exposure to TNF-α, all of the nontreated cells imported NFκB into the nucleus ^{(Fig. 8, b and b′)}. This did not happen with Taxol-treated cells. In that case, whereas the majority (∼90%) of the cells with a normal-looking nucleus successfully translocated NFκB across the nuclear envelope, virtually all of the cells that contained multiple micronuclei and lobulated nuclei failed to do so (Fig. 8, d and d′). This result seems to agree with the fact that NPC defects are much more frequently seen among cells with aberrantly shaped or sized nuclei than among cells with a single, normal-looking nucleus (Table 1).

We have shown here that Taxol and other antimicrotubule agents affect nuclear envelope organization and compromise nucleocytoplasmic transport. Clustering of the NPCs associated with nuclear lamina aberrations has also been seen in neuronal cells of Drosophila melanogaster upon disruption of the Dm₀ lamin gene (32), whereas pore clustering and mRNA export defects have been detected in yeast strains expressing mutant nucleoporins (Ref. 33 and references therein). Whether these defects and the aberrations caused by antimicrotubule agents share a common pathophysiological basis is not clear at this time.

Apoptotic cleavage of the nuclear lamins upon treatment of human breast carcinoma cells with Taxol has been reported by McCloskey et al. (34). However, our observations in human endometrial and cervical carcinoma cells indicate a different type of defect that develops well before degradation of nuclear envelope constituents. The possibility that the Taxol-induced defects arise from apoptosis can be ruled out because TUNEL and annexin V assays do not reveal early alterations in DNA structure or plasma membrane organization at time points when the NPCs and the nuclear lamina are clearly disorganized. Furthermore, physical degradation of nuclear lamins, which is diagnostic for ICE protease activation, is not detected until 72 h after release. Although we cannot exclude by the currently available techniques that a small amount of the lamins (i.e., 1–2%) are proteolyzed early after treatment, it is important to note that most the treated cells divide successfully, pass the G₁ checkpoint, and transit through the S phase after release into normal medium. This would be unlikely for cells that have initiated the process of programmed death.

Considering the current literature, one could speculate that nuclear envelope lesions develop as a result of decreased farnesylation of the nuclear lamins. Although this would be consistent with the inhibitory effects of Taxol on protein prenylation (6), it does not fit the fact that lamins assemble normally when isoprenylation is blocked by the specific farnesyl transferase inhibitor BZA-5B (35).

Nuclear lamina and NPC alterations could also be a consequence of increased phosphorylation mediated by mitogen-activated protein kinase and other protein kinases that are activated by antimicrotubule drugs (36). Along the same lines, the structural abnormalities induced by Taxol are morphologically similar to microtubule-dependent lamina lesions and nuclear shape irregularities that develop in fibroblastic cells overexpressing a dominant-negative form of the ATR kinase (37) after treatment with hydroxyurea. ATR, a protein related to the ataxia-telengiectasia gene product (ATM), is a member of the phosphatidylinositol kinase family and an important component of several cell cycle checkpoints (38). Whether nuclear envelope proteins constitute an end target of these enzymes is an exciting possibility that needs to be addressed in future studies.

Apoptosis and metabolic effects aside, we should finally consider the consequences of altered microtubule dynamics on nuclear envelope reassembly because taxanes and Vinca alkaloids, when given at nanomolar doses, lower dynamic instability without significantly affecting the mass of tubulin polymer inside eukaryotic cells (15). When administered at a dose of 10 nm, Taxol is taken up and concentrated >800-fold by binding to cellular microtubules (4, 5). At this concentration (∼8 mm), the drug decreases exchange of tubulin subunits at polymer ends (dynamicity) and blocks cells in mitosis. Upon release into normal medium, the intracellular concentration of Taxol is gradually reduced, allowing reentry into the cell cycle. However, whereas the intracellular concentration of Taxol is still rather high (∼4 mm according to Ref. 5), it is clear that most cells begin to divide. Knowing this, we are bound to think that nuclear envelope aberrations may develop because partially stabilized microtubules, contrary to their normal counterparts, fail to “unload” important components of the nuclear envelope and the nuclear matrix that transiently dock on the mitotic spindle during metaphase and anaphase (16, 20, 21, 39, 40). The timely release of these factors from the microtubules is probably necessary to assemble anew the nucleus of the daughter cells in the end of the division cycle.

The observations presented here substantiate the idea that microtubule-acting agents impair nuclear import. Looking critically at our import data, we should note that nuclear transport in Taxol-treated cells is universally inhibited under in vitro conditions because nuclear accumulation of the karyophilic substrate (BSA-NLS) is abolished in all types of digitonin ghosts. However, in an in vivo setting, cells with micronuclei and lobulated nuclei are primarily affected, whereas those that possess normal-looking nuclei are influenced much less from a functional point of view. From this, we suspect that NPC abnormalities develop in all cells but are perhaps more extensive and more easily identifiable in those with an aberrant nuclear morphology. It could also be that NPC malfunction is much more pronounced in vitro because auxiliary or repairing factors have been removed, and the system is challenged to perform under significantly more demanding conditions.

Finally, at the conclusion of this communication, we ought to discuss the effects of Taxol on NFκB function. Recent studies have established that NFκB is a major antiapoptotic factor in mammalian cells. Liver cell apoptosis followed by embryonic lethality is observed in RelA (NFκB65)-deficient mice, whereas fibroblasts from RelA-null animals are extremely sensitive to apoptotic stimuli such as TNF-α treatment, ionizing radiation, and treatment with daunorubicin (41, 42). NFκB induces expression of the TRAF1, TRAF2, c-IAP1, and c-IAP2 genes, which produce potent suppressors of caspase-8 (43). Knowing these observations, we would like to propose that inhibition of NFκB translocation by antimicrotubule drugs removes a major line of antiapoptotic defense and renders the cells prone to programmed death. This postulate is fully supported by recent studies on mice showing an additive or supra-additive effect of Taxol and TNF-α treatment (44). From the sum of these data, it can be predicted that combined therapy schemes that exploit the proapototic properties of TNF-α (45), the NFκB-inactivating potential of corticosteroids or salicylates (46, 47, 48), and the import-inhibiting ability of taxanes (this report) might enhance the efficacy of anticancer treatment.

We thank A. Demakopoulos (The University of Crete, Heraklion, Greece) for help in experimental work and C. Maison, J-B. Sibarita, and J. Salamero (Institut Curie, Paris, France) for advice and materials. We are also indebted to R. Hartig and P. Traub (Max Planck Institute, Ladenburg, Germany) for allowing the use of their confocal microscope.