The survival of cells submitted to genotoxic stress induced by anticancer chemotherapy is largely determined by the cellular response, especially DNA repair, cell-cycle checkpoints and apoptosis. Hence, loss of cell-cycle checkpoints increases the cytotoxicity of cisplatin,1 topoisomerase I-mediated DNA damages,2 or tubulin stabilizing agents.3 Increasing evidence indicates that intracellular reactive oxygen species (ROS) play a critical role in cellular homeostasis. A ROS elevation is necessary for induction of permanent growth arrest.4 Induction of p21CIP1/WAF1 and generation of ROS were shown to regulate differentiation and apoptosis of human leukemia cells induced by the histone deacetylase inhibitor MS-275.5 Furthermore, the p53 tumor suppressor protein influences cell commitment into apoptosis or cell cycle arrest via the upregulation of downstream target genes. Hence, the generation of superoxide anion is induced by p53 upon 5-fluorouracil treatment.6, 7 The effects of pharmacologic inhibitors of ROS indicate that the generation of oxidative stresses in mitochondria contributes to p53-mediated apoptosis.7
Several anticancer agents can lead to ROS accumulation and apoptosis through p53-independent pathways. Cisplatin and γ-irradiation promote ROS production which, in turn, contributes to Fas receptor aggregation and cell death in Fas-expressing cells.8 Elevated ROS levels potentiate c-jun N-terminal kinase and p38 mitogen-activated kinases and account for the sensitization of cells to cisplatin.9 Conversely, scavenging intracellular toxic oxidant molecules increases cellular resistance to cisplatin.10 Hence, ROS might participate to p53-dependent or p53-independent apoptosis.11, 12, 13
We studied the role of ROS in the cytotoxicity of a distinct family of anticancer agents, the taxanes. Taxanes promote the polymerization of purified tubulin in vitro. At high concentrations, they enhance the fraction of polymerized tubulin in cells and have thus been referred to as “microtubule stabilizing agents.”14 The most potent mechanism of paclitaxel cytotoxicity, the kinetic stabilization of spindle microtubule dynamics, occurs at nanomolar concentrations, inhibits mitosis and induces apoptosis.15, 16 At these low concentrations, taxane-induced cell death occurs after an aberrant mitosis resulting in an aneuploid G1 cell population, whereas, at higher concentrations, cell death is the result of mitotic arrest.17, 18 Failure of paclitaxel to induce microtubule stabilization is one major mechanism of cellular resistance.19 However, the molecular pathways downstream from spindle microtubule stabilization leading to cell death remain poorly understood.
To determine whether ROS production is essential to paclitaxel cytotoxicity, we studied the levels of ROS in A549 human lung cancer cells exposed in vitro to paclitaxel and investigated the consequences of its inhibition both in vitro and in vivo. We found that H2O2 accumulation is an early and crucial step in paclitaxel cytotoxicity. Moreover, we have shown that tailoring ROS signaling pathways are a potential, relevant therapeutic approach used for increasing the antitumor activity of taxanes in vivo.
All chemicals were from Sigma (Saint Quentin Fallavier, France) except for Paclitaxel (Taxol®, Bristol Myers Squibb, Rueil Malmaison, France). A549 (human lung carcinoma, ATCC n°CL-185) and LLC-1 (mouse Lewis lung carcinoma, ATCC n°CRL-1642) were from the American Type Culture Collection (Manassas, VA). Cells were cultured in DMEM/Glutamax-I supplemented with 10% heat-inactivated fetal calf serum and antibiotics (Life Technologies, Cergy Pontoise, France) at 37°C in humidified atmosphere with 5% CO2. Cell lines were passaged every 3 days and routinely tested for mycoplasma infection.
The number of adherent cells was evaluated by the crystal violet assay. Cells were fixed and stained in 0.5% crystal violet and 30% ethanol in PBS for 30 min at room temperature. After washing twice in PBS, the stain was dissolved in 50% ethanol, and absorbance measured at 560 nm on an ELISA multiwell reader (Victor2, Perkin Elmer, Paris, France). Results are expressed either as percentage of cell number ± SEM vs. cells in culture medium alone or as numbers of adherent cells calculated for each sample using a standard curve. The number of adherent cells evaluated by crystal violet assay was correlated to the number of viable cells evaluated by MTT assay. MTT assay was not used in this study because of potential interference with antioxidant drug, such as N-acetylcysteine (NAC). The concentration of cytotoxic reagent reducing the number of adherent cells by 50% (IC50) was determined after plotting cell number as a function of concentration. Cell death was evaluated spectrofluorimetrically (Victor2, Perkin Elmer, Paris, France) by Yopro-1 assay.20 Adherent cells seeded on 96-well tissue culture plates (Costar, Corning, NY) were incubated with 1 μM Yopro-1 in PBS for 20 min at room temperature. Excitation and emission wavelengths used were 490 nm and 510 nm, respectively.
Intracellular ROS measurement
Levels of intracellular O°2− and H2O2 were assessed spectrofluorimetrically by oxidation of specific probes: dihydroethidium (DHE, Molecular Probes, Leiden, The Netherlands) and 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA, Molecular Probes, Leiden, The Netherlands), respectively. Briefly, cells seeded on 96-well tissue culture plates were washed once with PBS, then incubated with either 250 μM DHE or 200 μM H2DCFDA in PBS. Excitation and emission wavelengths used were 500 nm and 605 nm for DHE and 490 nm and 535 nm for H2DCFDA, respectively. To test early ROS accumulation, paclitaxel diluted in PBS and DHE or H2DCFDA were added concomitantly to A549 cells (2 × 104 cells/well previously incubated for 24 hr). Fluorescence intensity was measured immediately after paclitaxel addition (T0), then every 15 min for 90 min. To test the ROS accumulation after 24 hr exposure to paclitaxel, fluorescence intensity was measured after incubation for 1 hr at 37°C with either DHE or H2DCFDA. The levels of O°2− or H2O2 were calculated in each sample as follows: ROS level (arbitrary units/min/106 cells) = (Fluorescence intensity (arbitrary units) at T60 min – Fluorescence intensity (arbitrary units) at T0)/60 min/numbers of adherent cells measured by crystal violet assay.
Intracellular GSH measurement
Levels of intracellular reduced glutathione (GSH) was assessed spectrofluorimetrically by monochlorobimane staining.21 Briefly, cells seeded on 96-well plates were washed once with PBS and were incubated, with 50 μM monochlorobimane diluted in PBS. The fluorescence intensity was measured after 15 min at 37°C. Excitation and emission wavelengths were 380 nm and 485 nm, respectively. Intracellular GSH level was expressed as arbitrary units of fluorescence intensity.
Effects of oxidative stress modulators on paclitaxel-induced cytotoxicity and ROS accumulation
Cells (5 × 104 cells/well) were seeded on 96-well plates and incubated at 37°C for 24 hr, with 5, 10 or 20 μM paclitaxel alone or with either 1.6 mM NAC, or 1.6 mM GSH, or 0.4 mM Buthionine Sulfoximine (BSO), or 0.2 mM aminotriazole (ATZ). At the end of the incubation period, medium was removed and cells were washed twice with PBS and analyzed for cell number using crystal violet or intracellular ROS level by oxidation of DHE or H2DCFDA or intracellular GSH level as described earlier.
Evaluation of subcellular location of O°2− production
Cells (2 × 104 cells/well) were seeded on 96-well plates and incubated at 37°C for 24 hr before the medium was removed. Cells were stained with 250 μM DHE and then treated with 20 μM paclitaxel with 10, 20, 40 μM rotenone, or 10, 20, 40 μM antimycin, or 10, 20, 40 μM diphenylene iodonium (DPI), or 10, 20, 40 μM allopurinol for 45 min in PBS. Fluorescence intensity was measured immediately after paclitaxel exposure and after 45 min.
Effect of caspase 3 and 8 inhibitors on paclitaxel-induced H2O2 production and cell death
Cells were pretreated for 30 min with either 100 or 200 μM of the caspase 3 inhibitor DEVD-CHO, or with 100 or 200 μM of caspase 8 inhibitor, IETD-CHO in PBS, or in DMEM complete medium. Then, 5 μM paclitaxel was added and intracellular H2O2 level assessed immediately, and then after 15 and 30 min. Cell number was assessed by crystal violet assay after 24 hr.
In vivo antitumor activity of antioxidant treatments
C57/BL6 female mice of 6–8 weeks of age were used (Iffa Credo, L'Arbresles, France). Animals received humane care in compliance with institutional guidelines. Two million LLC1 cells diluted in PBS were injected subcutaneously into the back of the neck of each mouse. When the tumor reached a mean size of 200–500 mm3 (day 0), the animals received three i.p. injections of either 20 mg/kg paclitaxel (day 0, 2 and 4) or of vehicle alone. The dose of paclitaxel was chosen on the basis of a 50% tumor volume reduction at day 15 without any limiting toxicity. At the same time, mice were injected i.p. or not with 150 mg/kg NAC (3 injections/week for 15 days), as previously described.22 In a separate experiment, one group of 10 mice received NAC daily in the drinking water at a dose of 1 g/kg body weight, starting at day 0.23 One group of mice injected with tumor cells remained untreated. Fifteen mice were treated in each group. Tumor size was measured with a vernier calliper every 3 days from day 0. Tumor volume was calculated as follows : TV (mm3) = (L × W2)/2, where L is the longest and W the shortest radius of the tumor in millimeters. Results were expressed as means of tumor volumes ± SEM.
All values are averages of 3 independent experiments made in triplicates, except when specified. Error bars shown in the figures represent standard error of the mean (SEM) and are shown when larger than the symbol. Statistical comparisons were carried out using two-tailed Student's t test for comparison of means. p values <0.05 were accepted as statistically significant.
Paclitaxel induces an early concentration-dependent accumulation of ROS in A549 cells
As assessed by H2DCFDA fluorescence, H2O2 levels were significantly increased (p = 0.01) in A549 cells exposed for 15 min to increasing concentration of paclitaxel from 100 nM. H2O2 accumulation was concentration- and time-dependent (Fig. 1a). The maximal relative accumulation of H2O2 was reached when cells were exposed to 5 μM of paclitaxel for 1 hr. DHE fluorescence, reflecting O°2− accumulation, also increased during paclitaxel exposure (Fig. 1b). Paclitaxel (100 nM) induced a 20% increase in DHE fluorescence after 75 min of exposure (p < 0.0001).
As we used pharmaceutical formulation of paclitaxel diluted in cremophor EL, we tested the ability of pure cremophor EL, in a dilution equivalent to 20 μM paclitaxel, to increase DHE and H2DFCDA fluorescence in A549 cells. Cremophor did not induce any ROS production in A549 cells.
As assessed by monochlorobimane staining, there was no significative modification of the reduced GSH pool after exposure to paclitaxel for 1 hr (data not shown).
N-acetylcysteine and reduced GSH protects A549 cells against the cytotoxicity of paclitaxel
To study the role of ROS accumulation in paclitaxel cytotoxicity toward A549 cells, we evaluated the effects of co-incubation of paclitaxel with antioxidant molecules. As measured by the crystal violet assay, paclitaxel alone had an IC50 at 24 hr of 5 μM. The addition of 1.6 mM NAC or 1.6 mM GSH induced a 4-fold increase in the IC50 value (Fig. 2a). When given alone, GSH had no effect on cell number (percent vs. untreated cells : 116 ± 13%, p = 0.42), whereas 1.6 mM NAC tended to increase cell number (percent vs. untreated cells survival: 131 ± 12%, p = 0.07). However, 0.4 mM NAC did not increase cell number (percent vs. untreated cells : 97 ± 5%), while it retained significant protective effect against paclitaxel cytotoxicity (percent vs. untreated cells after 10 μM paclitaxel exposure for 24 hr: 36 ± 3% vs. 18 ± 4% for NAC treated- and untreated-cells, respectively).
The protective effect of NAC against paclitaxel cytotoxicity was also assessed by measuring cell death using Yopro-1 (Fig. 2b). In A549 cells exposed to 5 μM paclitaxel for 24 hr, 1.6 mM NAC decreased cell death by 30%.
The co-incubation of NAC with paclitaxel had to be simultaneous to be protective since to delay by 4 hr the incubation of NAC resulted in a 3-fold reduction in cell protection (Fig. 2c).
When a very low initial cell density (104 ml−1) and a 72 hr exposure to paclitaxel were used as previously described,24 the paclitaxel IC50 was 5 nM. Under these conditions, the cytoprotective effect of NAC was in the same range as with micromolar concentrations of paclitaxel (data not shown). We used short exposures of high concentrations of paclitaxel, because high cell density was preferable for the ROS measurement by spectrofluorometry.
The paclitaxel-induced accumulation of H2O2 was reduced by both NAC and GSH (Fig. 3a). However, concomitantly, O°2− accumulation was not modified (Fig. 3b). The reduced GSH pool was depleted by 40% in cells exposed to 5 μM paclitaxel for 24 hr. According to its GSH reductase-like activity, NAC partially reversed the depletion of GSH (Fig. 3c). These results indicate that the cytoprotection conferred by NAC and GSH was associated with a decrease in H2O2 but not in O°2− levels in cells exposed to paclitaxel.
BSO is an inhibitor of GSH synthesis. Given alone, 0.4 mM BSO had no effect on number of adherent A549 cells. In contrast, BSO significantly increased the cytotoxicity of paclitaxel after a 24 hr co-incubation (Fig. 4a). The addition of BSO to 5 μM paclitaxel induced a three-fold increase in H2O2 levels and a 80% depletion of the reduced GSH pool (Figs. 4b and 4d). However, BSO did not change O°2− level (Fig. 4c).
We also explored the effect of ATZ, a specific catalase inhibitor on paclitaxel cytotoxicity. ATZ (0.2 mM), while inhibiting catalase activity within the cells (data not shown), did not enhance paclitaxel cytotoxicity (Fig. 4a) and increased by 50% H2O2 levels (Fig. 4b).
Paclitaxel-induced O°2− production is associated in part to increase activity of NADPH oxidase
The primary source of H2O2 is the conversion of O°2− by superoxide dismutase. To determine the subcellular location of early paclitaxel-induced O°2− production, A549 cells were incubated with inhibitors of NADPH oxidase (diphenylene iodonium, DPI), of xanthine oxidase (allopurinol) and of the mitochondrial respiratory chain (rotenone and antimycin). The optimal conditions for measuring O°2− level were after 45 min exposure to 20 μM paclitaxel. In A549 cells not exposed to paclitaxel, basal O°2− production was not altered by DPI, while it was increased by 50% by antimycin (40 μM) that blocks mitochondrial electron transport at complexes III22 (Fig. 5a). This result indicates that mitochondria but not NAPDH oxidase is a significative source of O°2− in untreated A549 cells. In contrast, DPI partially inhibited paclitaxel-induced O°2− production in a concentration-dependent manner (Fig. 5b): production of O°2− induced by paclitaxel was reduced by 17 ± 4% in presence of 40 μM of DPI (p < 0.0005). Thus, paclitaxel-induced O°2− production seems related in part to increased activity of NADPH oxidase (Nox). Antimycin increased O°2− production to a lesser extend in paclitaxel-treated A549 cells than in untreated cells, by 50% and 15%, respectively (Figs. 5a and 5b).
Caspase 3 and 8 inhibitors partially protect A549 cells from paclitaxel cytotoxicity but do not alter early H2O2 accumulation
In cells exposed to 5 μM paclitaxel for 24 hr, specific caspase 3 and 8 inhibitors, DEVD-CHO and IETD-CHO increased relative cells number in a concentration-dependent manner (Fig. 6a). The early paclitaxel-induced accumulation of H2O2 was unchanged by DEVD-CHO or IETD-CHO at concentrations up to 200 μM (Fig. 6b). In contrast, 1.6 mM NAC increased cell number after a 24 hr paclitaxel exposure, and significantly decreased the early H2O2 accumulation induced by 5 μM paclitaxel (Fig. 6b).
N-acetyl cysteine abrogates paclitaxel-antitumor effects in mice
Animal bearing LLC1 tumors were treated with either paclitaxel or NAC or the combination of two (Fig. 7). Mice treated by i.p. NAC alone, three times a week, developed larger tumors than the untreated mice from day 7 (1.38 ± 0.12 vs. 0.98 ± 0.07 cm3, p = 0.003) (Fig. 7a). Paclitaxel was associated with smaller tumors from day 10 (1.44 ± 0.11 vs. 1.85 ± 0.11 cm3, p = 0.04). Adding i.p. NAC, three times a week, to paclitaxel abrogated the antitumor effect of paclitaxel. Tumor sizes were significantly higher from day 10 in mice treated with NAC + paclitaxel than in mice treated with paclitaxel alone (2.31 ± 0.25 vs. 1.44 ± 0.11 mm3, p = 0.02). A similar effect of NAC was observed when administered daily along with the drinking water (Fig. 7b). At day 10, tumor sizes were 2.50 ± 0.41 in mice treated with NAC and paclitaxel and 1.09 ± 0.18 mm3 after paclitaxel alone (p = 0.01).
In this study, we have shown that paclitaxel induces an early accumulation of H2O2 and O°2− in A549 cells. The antioxidants NAC and reduced GSH partially decrease H2O2 accumulation and protect A549 cells against paclitaxel cytotoxicity. In contrast, BSO, which indirectly decreases hydrogen peroxide metabolism by inhibiting GSH synthesis, strongly enhances paclitaxel cytotoxicity. It seems very unlikely that paclitaxel can be directly inactivated by GSH or NAC. Indeed, in contrast to alkylating agents, paclitaxel cannot bind convalently to GSH or NAC, since it lacks electrophilic radical.25 While circulating thiols inactivate platinum agents, no thiol-paclitaxel conjugates have been detected in paclitaxel-treated patients.26 Taken together, these results demonstrate that H2O2 but not O°2− accumulation is crucial for paclitaxel cytotoxicity in vitro. The protective effect of NAC is lost when NAC exposure is delayed, emphasizing the role of early production of ROS in paclitaxel cytotoxicity.
The early ROS accumulation induced by paclitaxel can be related to an increased intracellular production of O°2− and H2O2 or, alternatively, to a decrease in detoxification of ROS. H2O2 is in part converted into H2O by GSH–peroxidase that uses GSH as co-enzyme.27 Since GSH depletion has not been observed during the first hour of paclitaxel exposure, but only after 24 hr, it cannot explain the early ROS accumulation.
Thus, the early accumulation of ROS is probably related to the overproduction of O°2− and to its rapid conversion into H2O2. Our data suggest that paclitaxel-induced O°2− production is, in part, associated with an increased activity of Nox, a multimeric cytosolic enzyme, which catalyzes the production of O°2− from oxygen and NADPH.28 Paclitaxel-induced microtubule stabilization could activate Nox by a Rac-dependent pathway, since Nox is regulated by the activation of Rac GTPase, a member of Rho GTPase family closely interacting with microtubule.29, 30
However, because DPI inhibits paclitaxel-induced O°2− production only partially, Nox is probably not the only source of paclitaxel-induced O°2− production. Early mitochondrial O°2− production induced by paclitaxel has also been reported both in cell-free system and intact cells, and appears to occur upstream of caspase activation.31, 32 Possible interaction of paclitaxel with mitochondrial tubulin can account for the direct effects on mitochondria.33 Thus, the role of mitochondrial respiration in taxol-induced ROS generation is not totally conclusive based on the available data, and requires further studies.
The cytotoxicity of paclitaxel involves the activation of caspases 3 and 8, since their inhibition increases the survival of A549 cells after 24 hr exposure to paclitaxel. This observation is not only consistent with the action of paclitaxel on mitochondria, but also with previous studies showing that the activation of both caspases 3 and 8 is required for paclitaxel-induced apoptosis in a death receptor independent pathway.34, 35 However, the early H2O2 accumulation induced by paclitaxel is not decreased by caspase 3 and 8 inhibitors, suggesting that H2O2 accumulation occurs before the commitment of the cells to undergo apoptosis.
An array of recent data merge with our own finding that the overproduction of H2O2 is a key-event in the cytotoxicity of paclitaxel: H2O2 is required for activation of the c-Jun N-terminal kinase/Stress Activating Protein Kinase (JNK/SAPK) signaling pathway, which induces apoptosis by the intrinsic pathway.9, 36 On the other hand, the activation of the JNK/SAPK signaling pathway is an early consequence of the interaction of paclitaxel with microtubules.37 The activation of this pathway is independent of cell cycle and is required for paclitaxel-induced early apoptosis.37, 38, 39 In addition, paclitaxel-induced SAPK activation is dependent on functional Rac protein identified as a critical component of Nox.39 Taken together, these data suggest that paclitaxel activates Nox and disrupts the mitochondrial respiratory chain via tubulin interaction. The superoxide anions generated are converted in H2O2, which itself activates the SAPK pathway and induces apoptosis. The role of ROS in paclitaxel cytotoxicity may be dependent on the cellular model. Hence, it has been recently shown that, in T lymphoblastic leukaemia HSB-2 cells, ROS were not involved in paclitaxel-induced apoptosis.40 This discrepancy could be explained by the fact that apoptosis of type I cells such as T cells is independent of mitochondrial signaling.41 Therefore, mitochondrial production of ROS is probably not essential to paclitaxel-induced apoptosis in T cells.
Most anticancer agents are able to enhance ROS production, but the mechanisms of ROS production differ from one agent to another. For instance, a specific mechanism of ROS production involving p53 activation has been described in anticancer agents such as 5-FU and anthracyclins.7 On the other hand, anthracyclins, unlike paclitaxel, are unable to induce direct production of ROS by mitochondria.32
We then tested the in vivo relevance of our in vitro findings using NAC, a molecule that restores the GSH pool and is endowed with catalase-like properties.42 NAC by itself triggers the growth of LLC1 tumors implanted in mice. Since some tumor cells are subject to a detrimental oxidative stress, NAC can act by lowering the level of H2O2 in LLC1 cells, thus increasing cell proliferation or decreasing the rate of cell death.43, 44 However, this promoting effect of NAC is probably dependent on basal intracellular oxidative stress. The report that NAC prevents the transformation of normal cells into cancer cells by tobacco smoke is not really conflicting with our data.45 Indeed, NAC is able to prevent the DNA damages resulting from ROS generated by tobacco smoke oxidants.44, 45
We have shown that NAC abrogates the antitumor effect of paclitaxel in mice bearing subcutaneously implanted LLC1 tumors. This result is fully in line with the previous reports, showing that NAC decreases the activity of ROS-dependent anticancer agents such as arsenic trioxide and adriamycin.46, 47 On the other hand, De Flora and co-workers reported that NAC acts synergically with adrimycin to prevent tumorigenicity and metastasis of B16 melanoma cells injected intravenously in mice.48 It was proposed that NAC could prevent invasion of tumor cells by inhibiting type IV collagenase.49 The discrepancy between these results and ours could be explained by differences in tumor cell types, route of tumor cell administration and time of NAC administration.
Our observations emphasize the critical role of intracellular H2O2 levels in the sensitivity of tumor cells to paclitaxel. Therefore, cellular antioxidant enzymes may influence the sensitivity of tumor cells to paclitaxel, as previously shown for cisplatin.50, 51 This hypothesis is strengthened by the recent finding that resistance to docetaxel in breast cancer can be mediated by the activation of several genes controlling the cellular redox environment.52
This study opens several clinical perspectives: first, it suggests that NAC should be used with caution in cancer patients considering the risk of increased tumor growth. Second, an increase in paclitaxel antitumoral activity could be achieved by modulating the oxidant–antioxidant status of tumor cells. Lastly, intracellular antioxidant enzymes can be seen as new targets participating to pleiotropic resistance.
In conclusion, ROS production is involved in paclitaxel cytotoxicity. The accumulation of H2O2 is an early and crucial step for paclitaxel-induced tumor cell death. Modulation of ROS production might be a promising approach to increase paclitaxel and other anticancer agents cytotoxicity.