Chemosensitization in non-small cell lung cancer cells by IKK inhibitor occurs via NF-kappaB and mitochondrial cytochrome c cascade.

In this study, we demonstrated with mechanistic evidence that parthenolide, a sesquiterpene lactone, could antagonize paclitaxel-mediated NF-kappaB nuclear translocation and activation by selectively targeting I-kappaB kinase (IKK) activity. We also found that parthenolide could target IKK activity and then inhibit NF-kappaB; this promoted cytochrome c release and activation of caspases 3 and 9. Inhibition of caspase activity blocked the activation of caspase cascade, implying that the observed synergy was related to caspases 3 and 9 activation of parthenolide. In contrast, paclitaxel individually induced apoptosis via a pathway independent of the mitochondrial cytochrome c cascade. Finally, exposure to parthenolide resulted in the inhibition of several NF-kappaB transcript anti-apoptotic proteins such as c-IAP1 and Bcl-xl. These data strengthen the rationale for using parthenolide to decrease the apoptotic threshold via caspase-dependent processes for treatment of non-small cell lung cancer with paclitaxel chemoresistance.

Introduction

Systemic treatment, especially chemotherapy, is an indispensable step for curing or increasing survival in patients with lung cancer. However, resistance to chemotherapy is frequently observed; this can result in recurrence and metastasis [1, 2]. Therefore, exploration of molecular mechanisms involved in chemoresistance and further development of drugs and strategies that inhibit chemoresistance are a high priority for lung cancer therapy [3, 4].

The transcription factor NF-κB is known to be involved in the transcriptional regulation of various genes involved in cell proliferation, angiogenesis, metastasis and apoptosis [5, 6]. Anti-apoptotic signalling by NF-κB mediates resistance to chemotherapeutic drugs [7, 8] in ovarian cancer [9-11], pancreatic cancer [12], breast cancer [13, 14] and prostate cancer [15], implying that inhibition of NF-κB might be an effective treatment strategy for many resistant cancers. However, this has been challenged by evidence in other cell types; in neuroblastoma cells, NF-κB activation induced p53-mediated apoptotic signalling in response to the chemotherapy agent Dox [16]. It is presently unclear which signalling events ultimately determine whether NF-κB activation results in a pro- or anti-apoptotic response. The role of NF-κB might depend on the conditions of other oncogenes such as Myc, P53 and Bcl2.

Specifically, aberrant expression of NF-κB has been related to lung cancer development and progression. Our recent work revealed that nuclear RelA and cytoplasmic pI-κBα were significantly associated with a poor prognosis [1]. Furthermore, activation of NF-κB by chemotherapy can blunt the efficacy of the chemotherapy itself to induce death in lung cancer cells [17, 18]. The finding that NF-κB is a key player in lung cancer survival has prompted researchers to explore drugs with NF-κB suppression activity in order to directly kill the cancer or render it more vulnerable to chemotherapeutic agents [19-21]. Additionally, the use of drug combinations with non-overlapping toxicity profiles could allow decreased doses of the individual agents, thereby reducing the severity of undesired side effects.

NF-κB proteins are sequestered in the cytoplasm in an inactive state and in a complex with the inhibitor I-κB. Degradation of I-κB by I-κB kinase (IKK)-mediated phosphorylation can release NF-κB to translocate to the nucleus and promote target gene expression. NF-κB can be blocked by targeting elements of its various signalling cascades such as the IKK complex, the I-κB inhibitory protein, the p65 subunit of the transcriptionally active heterodimer and the proteasome. Thus, any molecular agents that inhibit NF-κB are potential therapeutic targets for cancers whose tumorigenicity depends on NF-κB activity [7, 22, 23]. A great number of such compounds have already been tested to suppress the growth of cancer cells [24-26]. The active compound of feverfew (Chrysanthemum parthenium), parthenolide, has shown potential as an anticancer agent [25, 27]. Parthenolide shows significant cancer suppression activity in vitro and in vivo [28, 29] through inhibition of NF-κB by direct inhibition of IKK [27], which can regulate the phosphorylation of I-κB at serine residues 32 and 36. Although the inhibitory effect of parthenolide on NF-κB activity has been reported previously in several cancer cell lines [28-31], its significance, detailed mechanism and potential use in treatment and as an adjuvant with chemotherapy have not yet been fully investigated in human non-small cell lung cancer (NSCLC).

The aims of this study are to provide further information about the role of NF-κB in lung cancer cell biology and to further determine whether parthenolide could elicit anti-proliferative activity via inhibition of NF-κB activation, thereby increasing the efficacy of paclitaxel in human NSCLC cell lines. To our knowledge, the present study is the first to combine molecular and functional analyses of NF-κB and its related molecules in order to determine adjuvant effects of parthenolide and the synergistic combination of parthenolide and paclitaxel.

Materials and methods

Materials and antibodies

Paclitaxel was purchased from Calbiochem (La Jolla, CA, USA) and dissolved in 100% dimethyl sulfoxide to make a stock solution of 1.0 mmol/l. The I-κB phosphorylation inhibitor (BAY 11–7082, Cat. #EI278–0010) and pancaspase inhibitor Z-Val-Ala-Asp-fluoromethyl ketone (Z-VAD-fmk) (Cat. #P416–0001) were purchased from Biomol (BIOMOL International L.P., PA, USA). Parthenolide was obtained from Alexis Biochemicals (San Diego, CA, USA). These agents were dissolved in DMSO at a concentration of 10 or 20 mmol/l and stored in the dark at −80°C. The entire stock solution was diluted to obtain the desired concentrations with growth medium. The concentration of dimethyl sulfoxide did not exceed 0.2% (vol/vol). Z-VAD-fmk was incubated at a desired concentration of 20 mol/l for 30 min. prior to induction of apoptosis (paclitaxel treatment). During prolonged experiments, supernumerary Z-VAD-fmk was further added for 12 to 16 hrs. All cell culture reagents, DNA-modifying enzymes and Lipofectamine reagent were purchased from Invitrogen (Carlsbad, CA, USA).

Cell cultures

The human lung adenocarcinoma cell line A549, squamous cell line NCI-H446 and A549 taxol-resistant derivative A549-T24 were selected for resistance to taxol in a stepwise manner and maintained in a final concentration of 24 nmol/l taxol. These cells were obtained from the cell bank of the Chinese Academy of Sciences and maintained at 37°C in a 5% CO2 atmosphere in air, in RPMI 1640 medium (Gibco, Gaithersburg, MD, USA) supplemented with 10% (vol/vol) foetal bovine serum and 100 units/ml each of penicillin and streptomycin.

Small interfering RNA

Small interfering RNA (siRNA) was applied in order to achieve gene silencing. High-purity SMARTpool™ siRNAs targeting IKK-α (Cat. # 60–055, accession no. NM_001278), IKK-β (Cat. # 60–025, accession no. NM_001556), p65 (Cat. # 60–076, accession no. NM_021975) and control siRNA were purchased from Upstate (Charlottesville, VA, USA). Lipid-encapsulated SMARTpool™ and control were used for transfection. At the time of transfection, the cell density was ∼70–90% confluent, or approximately 1 × 105 cells/ml. We transfected the cells using the siIMPORTER™ Transfection Reagent (Cat. # 64–101SP, Upstate) according to the manufacturer’s protocol after testing and optimizing the conditions best suited for each cell line or culture. To improve gene silencing, at 24 hrs after transfection the same siRNA reagents were added to the media for a second 24 hrs. At 24 hrs after transfection, the cells received paclitaxel as described above. Gene silencing effects were evaluated by Western blot analysis.

Cell viability determination (MTT assay)

Cells were exposed to different treatments in a medium containing 0.5% foetal bovine serum for 72 hrs, then 10 μl of 3-(4,5-dimethylthiazol-2-yl) 2,5-diphenyl-tetrazolium bromide (MTT) solution (10 mg/ml in PBS) were added followed by 100 μl of 10% sodium dodecyl sulphate (SDS) for formazan crystal formation. The absorption of the samples was determined using an ELISA reader (Anthos Mikrosysteme GmBH, Germany) at a wavelength of 570 nm. The optical density of untreated control cells was considered to represent 100% viability for calculation of the viability of treated cells.

Terminal deoxynucleotidyltransferase-mediated dUTP nick end labelling (TUNEL)

Control and treated cells collected at various time-points following treatment were subjected to terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick-end labelling (TUNEL) assay. The commercially available APO-BrdU™ TUNEL assay kit (with Alexa Fluor® 488 anti-BrdU, 60 assays; Cat. # A23210, Invitrogen) was utilized to label DNA fragmentation with terminal deoxynucleotide transferase following the instructions of the manufacturer. Quantification of the apoptotic cells by FACS analysis (Calibur, BD Biosciences, CA, USA) was achieved using the geo mean function in CellQuest Software (FACScan, BD Biosciences, CA, USA). All flow cytometry analyses were repeated three times.

Nuclear extracts and electrophoretic mobility shift assay (EMSA)

Nuclear proteins of the cultured control and treated cells were extracted using an NE-PER nuclear and cytoplasmic extract kit (Pierce Biotechnology, IL, USA), following the manufacturer’s instructions. EMSA was performed using a gel shift assay system kit (Promega, WI, USA) as described previously [1]. Briefly, double-stranded oligonucleotide probes targeting NF-κB (5′-AGT TGA GGG GAC TTT CCC AGG C-3′) and Oct-1 (5′-ATGCAAAT-3′) were labelled using T4 polynucleotide kinase and γ-32P ATP (Yahui Biotech Inc., Beijing, China). Ten micrograms of protein extract were incubated with 2 × 104 cpm of oligonucleotide probe in addition to 1 μg poly(dI-dC) (Sigma, St. Louis, MO, USA) in binding buffer for 15 min. at room temperature. Equal loading of nuclear extracts was monitored by Oct-1 binding. The resulting protein-DNA complexes were analysed on 4% non-denaturing polyacrylamide gels in 0.5× Tris-borate-ethylenediaminetetraacetic acid buffer and then subsequently dried and analysed by autoradiography.

Western blot analysis

Differently treated cells were lysed using protein extraction buffer and homogenized in 250 μl of lysing buffer. Cytoplasmic extracts were prepared to measure the activation of caspases 3 and 9. Ten micrograms of total protein were electrophoresed through SDS-PAGE on 4–20% gradient gels (Shanghai Sangon Biotech, Shanghai, China) and transferred to nylon membranes (Shanghai Sangon Biotech). The blots were probed with primary antibodies to recognize respective proteins; these antibodies included anti-RelA against the p65 (RelA) unit (p65; 1:150; sc-109, Santa Cruz Biotechnology, CA, USA), anti-phosphorylated I-κBα that recognizes the Ser32 epitope (1:40; sc-8404, Santa Cruz Biotechnology), anti-IκBα (1:100; sc-1643, Santa Cruz Biotechnology), anti-Phospho-IKK-α (Ser180)/IKK-β (Ser181) (1:100; #2681, Cell Signaling Technology, Beverly, MA, USA), anti-IKK-β (1:100; #2684, Cell Signaling Technology), anti-cleaved caspase-3 (Asp175) (1:100; #9661), anti-cleaved caspase-9 (Asp330) (1:100; #9501, Cell Signaling Technology), anti–X-linked inhibitor of apoptosis protein (XIAP) (1:100; #2404, Cell Signaling Technology), anti–Bcl-XL(1:100; #2762, Cell Signaling Technology), anti–c-IAP1 (1:100; #4952, Cell Signaling Technology), anti–c-IAP2 (1:100; #3130, Cell Signaling Technology), anti–Trail-R1 (DR4) (1:100; # sc-6823, Santa Cruz Biotechnology), anti–Trail-R2 (DR-5) (1:100; # sc-65314, Santa Cruz Biotechnology), anti-Fas (1:100; # sc-52395, Santa Cruz Biotechnology) and anti-β-actin (1:100; # sc-47778, Santa Cruz Biotechnology). The secondary antibody, horseradish peroxidase-coupled mouse anti-rabbit immunoglobulin (Jingmei Biotech Co., Ltd. Shenzhen, China), was then incubated for 1 hr at room temperature. The subsequent analysis was performed with chemiluminescence using an ECL Western blotting kit (Amersham Biosciences, IL, USA) following the manufacturer’s recommendations.

Immunofluorescence assays

For immunostaining, cells were fixed in 3.8% paraformaldehyde for 5 min. at room temperature, permeabilized in 0.1% saponin for 5 min., and stained with anti-NF-κB (p65) (1:200; Santa Cruz Biotechnology) and anti-cytochrome c (1:200; #4272, Cell Signaling Technology, MA, USA) antibodies for 30 min. at room temperature. After washing with PBS, cells were incubated with affinity-purified, rhodamine-conjugated mouse anti-rabbit IgG (1:4000; Jackson Immuno Research, West Grove, PA, USA). The stained cells were analysed using a confocal microscope (Leica Microsystems Heidelberg GmbH, Heidelberg, Germany) with excitation at 488 nm and emission at 525 nm.

Cytochrome c release assay

Isolated tumour cells (5 × 107) were collected by centrifugation at 600 ×g for 5 min. at 4°C and washed with ice-cold PBS. The cells were assayed with the Cytochrome c apoptosis assay kit (Cat. #K257–100, Biovision, CA, USA). Briefly, cells were homogenized with the cytosol extraction buffer provided in the kit and then centrifuged at 700 ×g for 10 min. at 4°C to remove debris. The supernatant was then centrifuged at 10,000 ×g for 30 min. at 4°C. The pellet contained the mitochondrial fraction, and the supernatant was collected as the cytosolic fraction. These fractions were analysed for cytochrome c content by Western blotting using the cytochrome c antibody provided in the kit.

Statistical analysis

Statistical analysis was carried out using one-way ANOVA followed by Fisher’s least significant difference test, and the level of significance was set at P < 0.05. Data are expressed as the mean ± S.E.M. Statistical comparisons were carried out using SPSS software for Windows (SPSS, Inc., Chicago, IL, USA).

Results

Paclitaxel treatment induces NF-κB activation and up-regulates its regulatory target Bcl-xl

We first compared NF-κB DNA-binding activities among the human lung cancer cell lines A549, NCI-H446 and A549-T24, which had demonstrated resistance to taxol treatment, in order to detect possible effects of paclitaxel on NF-κB activity. As shown in Fig. 1, the specificity of NF-κB was first evaluated by performing EMSA gel supershift and competition assays (Fig. 1A). Basal NF-κB activity in A549 and NCI-H446 cells was not detectable. After exposure to 100 nmol/l paclitaxel for 12, 24 and 48 hrs, NF-κB activity in A549 and NCI-H446 cells increased notably in comparison with the basal level. Although the basal level of NF-κB activity in A549-T24 cells was higher than in A549 and NCI-H446 cells, NF-κB activity in A549-T24 cells also increased after 24 hrs of paclitaxel treatment, albeit to a lesser degree (Fig. 1B).

Figure 1

Effect of paclitaxel on NF-κB activation in A549, NCI-H446 and A549-T24 cell lines. (A) Supershift analysis and competitive study were performed to confirm the specificity of NF-κB DNA binding in A549-T24 cells stimulated with 100 nmol/l paclitaxel for 48 hrs, with antibodies specific for RelA (p65) (recognizes RelA/p50) (line 3), NF-κB1 (p50) (recognizes p50/p50 and RelA/p50) (line 4) and a 50-fold excess of unlabelled oligonucleotide (diminished both RelA/p50 and p50/p50) (line 5) and another oligonucleotide AP-1 (line 6). Data from one representative experiment are shown. (B) The three cell lines were exposed to 100 nmol/l paclitaxel for sequential durations. Nuclear extracts were prepared and examined for NF-κB DNA binding activities (RelA/p50) by electrophoretic mobility shift assay. Oct-1 was selected as a constitutively expressed control. NF-κB activation-involved molecules such as pI-κB, I-κB and Bcl-xl (cytoplastic extracts) were subjected to Western blotting with β-actin as a loading control. Representative figures of at least three independent experiments with similar results are shown at the indicated hours after treatment. The positions of protein weight markers are noted on the right.

As expected, paclitaxel treatment transiently increased the amount of phosphorylated I-κBα in A549 and NCI-H446 cells as detected by Western blotting. This was linked to a change in NF-κB DNA binding activity; however the total I-κBα was maintained at a high level similar to that without treatment, suggesting that paclitaxel did not affect the degradation of I-κBα in A549 or NCI-H446 cells. Consistent with the NF-κB DNA binding activity, the anti-apoptotic gene Bcl-xl was also up-regulated in response to paclitaxel treatment. Similar effects were also observed in the A549-T24 cells.

IKK kinase complex mediates paclitaxel-induced NF-κB activation

To determine the involvement of IKK in the response to paclitaxel, phospho-IKK-α /β was detected by immunoblotting with specific antibodies. Levels of phosphorylated (active) IKK increased 24 hrs after exposure to paclitaxel. Specifically, IKK-β and not IKK-α was phosphorylated in response to paclitaxel treatment, consistent with the changes in NF-κB and I-kB (Fig. 2A).

Figure 2

BAY 11–7082-mediated inhibition of IKK resulted in reduced induction of NF-κB by paclitaxel. A549 cells with or without pre-treatment of 5 μmol/l BAY 11–7082 were treated with 100 nmol/l paclitaxel for 24 hrs. (A) NF-κB DNA binding activities in the nuclear extract and p-IKK-α/β, IKK-β, phospho-I-κBα, I-κBα, Bcl-xl in the cytosolic extract were compared. The experiment was repeated with identical outcomes for at least three times, and a representative result is shown. B. TUNEL analysis with flow cytometry was carried out to evaluate induction of apoptosis. Columns, average values of at least three independent experiments performed in triplicate; bars, ±S.E.M. #, P < 0.01 versus paclitaxel + BAY 11–7082.

Next, we tested whether inhibition of IKK using an IKK inhibitor (BAY 11–7082) was sufficient to block paclitaxel-induced NF-κB activation. IKK activity (indicated as phospho-IKK-α/β) was induced in human NSCLC cell lines by paclitaxel treatment and inhibited by BAY 11–7082, whereas levels of IKK proteins (indicated as IKK-β) remained at the same level. As shown in Fig. 2A, paclitaxel-induced NF-κB activity measured with EMSA was abrogated by BAY 11–7082. Paclitaxel-induced Bcl-xl expression was also reduced by BAY 11–7082 (Fig. 2A), whereas the amount of total I-kBα was not increased by paclitaxel treatment (Fig. 2A).

We further performed a TUNEL assay to examine whether apoptosis could account for the cell growth inhibition in this system. As seen in Fig. 2B, paclitaxel treatment alone resulted in an apoptosis rate of 25%. BAY 11–7082 at a concentration of 5 μmol/l did not show significant growth inhibition after 24 hrs treatment in NCI-H446 cell lines, and still less apoptotic induction in A549 cell lines. Co-treatment with paclitaxel and BAY 11–7082, resulted in a further 20% and 30% enhancement of the apoptotic response in A549 and NCI-H446 cells, respectively (Fig. 2B), suggesting that interference with NF-κB transcriptional activity could sensitize the paclitaxel response.

Parthenolide inhibits paclitaxel-mediated activation of IKK

Earlier studies reported that parthenolide could inhibit activation of IKK in pancreatic carcinoma cell lines [30]. Here we examined if it also had an effect in human NSCLC lines. As expected, after exposure to parthenolide prior to paclitaxel stimulation, paclitaxel-induced NF-κB activation was potently inhibited in A549 cells as measured by EMSA (Fig. 3A). Incubation with 5 μmol/l parthenolide for 24 hrs completely inhibited paclitaxel–induced activation of IKK activity (Fig. 3B). The activation of IKK was concurrent with degradation of I-κB that showed similar kinetics in both cells types and was prevented by increasing the time of incubation with parthenolide (data not shown).

Figure 3

Regulation of NF-κB activation by parthenolide occurs through IKK inhibition. (A) A549 cells were pre-treated with 5 μmol/l parthenolide for various times (left) and for 24 hrs at various concentrations (right) courses, then incubated with 100 nmol/l paclitaxel for 24 hrs. Equal amounts of nuclear and cytosolic extracts were subjected to electrophoretic mobility shift assay (EMSA) for NF-κB binding and Western blotting for Bcl-xl, respectively. (B) A549 cells were alternatively pre-treated with either 5 μmol/l parthenolide or 5 μmol/l BAY 11–7082 for 24 hrs and then incubated with paclitaxel for 24 hrs. EMSA for NF-κB binding and Western blotting for p-IKK-α/β and Bcl-xl were performed. (C) Nuclear translocation of p65 was determined by indirect immunofluorescence. Representative fields show the exclusive cytoplasmic location of NF-κB in untreated control (a) and parthenolide-treated cells (b). Nuclear accumulation of p65 was observed in paclitaxel-treated cells (c) but not in paclitaxel + BAY 11–7082 pre-treated (d) or paclitaxel + parthenolide pre-treated cells (e). All the experiments shown in this figure were carried out two to four times, producing similar outcomes, and one representative result is shown.

Next, we performed immunoblotting and EMSAs to determine whether parthenolide inhibited the nuclear translocation of NF-κB and/or NF-κB binding activity. As shown in Fig. 3C, strong p65 translocation to the nucleus was observed in A549 cells stimulated with paclitaxel. This increased NF-κB translocation was completely inhibited when cells were incubated with 10 μmol/l parthenolide prior to paclitaxel stimulation; the effect was similar to or stronger than that seen with BAY 11–7082. This effect was also concentration and time dependent, and concentrations that did not affect apoptosis induction also inhibited the DNA binding activity of NF-κB (5 μmol/l). Moreover, NF-κB DNA binding activity in nuclear extracts of A549 cells stimulated with paclitaxel were completely abolished by 10 μmol/l parthenolide (Fig. 3A).

Parthenolide sensitizes to paclitaxel-induced apoptosis

We further examined the involvement of parthenolide in sensitization to paclitaxel-induced apoptosis. Parthenolide induced cell growth inhibition in a concentration- and time-dependent manner, but at a concentration of 5 μmol/l for 12 hrs, parthenolide did not significantly inhibit growth (Fig. 4A). Treatment with paclitaxel for 24 hrs induced cell apoptosis by about 30%; co-treatment with paclitaxel and parthenolide further enhanced apoptosis by almost 55%. The inhibitory effect of parthenolide and paclitaxel administered together was more than that of BAY 11–7082 and paclitaxel (Fig. 4B). Similarly, a DNA fragmentation assay demonstrated that cell death induced by paclitaxel and parthenolide occurred via apoptotis (data not shown).

Figure 4

Inhibition of the IKK/NF-κB pathway enhanced paclitaxel-mediated cytotoxicity. (A) Induction of apoptosis by parthenolide in concentrations (1, 3, 5,15 20 μmol/l) (a) and times (0, 12, 24, 48h) (b) courses in A549 (left), NCI-H446(middle) and A549-T24 (right) cells; TUNEL was assessed using flow cytometry, all of the columns depicted represent a mean of at least three independent experiments in triplicate; bars, ±S.E.M. (B) The apoptotic rate was analysed via TUNEL assay in A549 cells that were pre-treated with or without 5 μmol/l parthenolide for 24 hrs, then incubated with 100 nmol/l paclitaxel for another 24 hrs; bars, ±S.E.M. Significant differences are indicated by asterisks. #, P < 0.01 versus paclitaxel + parthenolide incubation with the same time. (C) Western blots of IKK-α and IKK-β protein levels in A549 cells 48 hrs after transfection with the indicated siRNAs (a). The experiments of Western blot were carried out with similar outcomes for at least three times, and a representative result is shown. A549 cells were transfected with the indicated siRNAs for 48 hrs then treated with 5 μmol/l parthenolide and/or 100 nmol/l paclitaxel in succession; the cell extracts were then subjected to electrophoretic mobility shift assay (EMSA) (NF-κB) (b) and TUNEL analysis (c). EMSA analyses were performed with identical outcomes for three times, and a representative result is shown. TUNEL analysis was carried out two independent siRNA experiments in triplicate, columns, mean of two independent experiments; bars, ±S.E.M. Significant differences are indicated as *, P < 0.01 versus the same cells treated with paclitaxel plus parthenolide and #, P < 0.01 versus cells treated with IKK-α or IKK-β siRNA as well as parthenolide.

Although our results suggest that parthenolide potentiated paclitaxel-mediated apoptosis by inhibiting IKK/NF-κB, it remains possible that parthenolide might sensitize cells to pro-apoptotic processes via IKK-independent pathways. To further investigate this, we transfected cells with siRNA-targeting IKK-α or IKK-β to evaluate whether inhibition of IKK could abolish the effect of parthenolide. As seen in Fig. 4Cb, siRNA-IKK-α and siRNA-IKK-β both blocked the IKK/NF-κB signalling cascade as measured by the DNA binding activity of NF-κB. Blockade with the expression of IKK-α or IKK-β in A549/siRNA-IKK-α/β cells resulted in enhancement of apoptosis induced by paclitaxel. Interestingly, parthenolide treatment in this system did not further enhance paclitaxel-induced apoptosis (Fig. 4Cc). These results are consistent with the EMSA data shown in Fig. 4Cb, and with phosphorylated (active) IKK immunoblotting (data not shown).

Biological NF-κB blockade contributes to parthenolide sensitization to paclitaxel

To determine whether IKK-mediated parthenolide sensitization to paclitaxel relies on NF-κB regulation, gene silencing targeting the p65 RelA subunit was carried out, and cell lines were then co-treated with paclitaxel and parthenolide or BAY 11 7082. Compared to A549 cells, the expression of Bcl-xl was abolished in cells transfected with p65-targeting siRNA. This was accompanied by the decreased RelA DNA binding activity in the nucleus (Fig 5A), confirming that the siRNA worked. As expected, inhibition of NF-kB significantly augmented the response of A549 cells to paclitaxel as assessed by MTT assay, whereas it abolished the inhibitory effects of parthenolide or BAY 11 7082 even when used at high concentrations (Fig. 5B).

Figure 5

Biological NF-κB blockade with p65 RelA siRNA enhanced paclitaxel-induced cytotoxicity. A549 cells were transfected with p65 RelA siRNA for 24 hrs and exposed to paclitaxel (100 nmol/l) for another 24 hrs. The scramble sequence siRNA control showed no effects on cell viability or NF-κB DNA binding activity (data not shown here). (A) NF-κB DNA binding activity and Bcl-xl expression were measured with electrophoretic mobility shift assay and Western blot analysis respectively at least three times with similar outcomes. (B) Cell viability was assessed by the MTT reduction assay with increasing concentrations of paclitaxel (0–200 nmol/l). All of the results represent the mean of at least three independent experiments performed in triplicate; bars, ±S.E.M. Significant differences are indicated by asterisks. #, P < 0.01 versus p65 siRNA-transfected cells treated with the same doses of paclitaxel. (C) TUNEL analysis with flow cytometry was carried out to evaluate induction of apoptosis. Untreated cells (none) served as the control (100%). The experiment was carried out with similar outcomes for at least three replications, and a representative result is shown. The apoptosis rates were 24.8% for A549 cells and 76.89% for A549/p65siRNA as indicated below the box. (D) A549 cells were transfected with p65 RelA siRNA and exposed to parthenolide (5 μmol/l)-paclitaxel (100 nmol/l) and BAY 11 7082 (5 μmol/l)-paclitaxel (100 nmol/l) combination treatments. TUNEL analysis with flow cytometry was carried out to evaluate induction of apoptosis. The data represent the average of two independent experiments performed in triplicate; Bars, ±S.E.M. #, P < 0.01 versus p65 siRNA-transfected cells treated with the same concentrations of paclitaxel.

We next used the TUNEL assay to examine whether induction of apoptosis could account for the cell growth inhibition observed in p65-silenced A549 cells. These cells showed strong induction of DNA fragmentation, indicating apoptosis (Fig. 5C). These results suggest that specific inhibition of NF-κB activity, which regulates expression of the anti-apoptotic gene Bcl-xl, resulted in the induction of apoptosis. In order to further verify that parthenolide could sensitize the effects of paclitaxel via NF-κB regulation, cells genetically silenced for NF-κB were treated with parthenolide and paclitaxel, or with BAY 11 7082 and paclitaxel. As expected, blockade of p65 resulted in the enhancement of paclitaxel-mediated apoptosis comparable to the one obtained with parthenolide or BAY 7082 pre-treatments. But also these pre-treatments do not further enhance paclitaxel-induced apoptosis when p65 is already blocked by siRNA (Fig. 5D).

Effect of IKK inhibition on NF-κB transcript target genes

To determine whether NF-κB downstream target genes are up-regulated by parthenolide, we evaluated the expression of some endogenous genes known to be regulated by NF-κB. These included pro-apoptotic genes such as like Fas, DR4 and DR5 as well as anti-apoptotic genes such as c-IAP1, c-IAP2, XIAP and Bcl-xl. As indicated in Fig. 6, incubation with parthenolide somewhat increased Fas, DR4 and DR5 protein levels. In contrast to the previous finding that exposure to paclitaxel induced up-regulation of the anti-apoptotic molecule Bcl-xl, incubation of A549 cells with parthenolide resulted in a marked decrease of Bcl-xl, c-IAP1 and c-IAP2 expression but XIAP expression had no obvious change (Fig. 6). The inhibition of anti-apoptotic molecules in A549 cells by paclitaxel was also abolished by transfection with IKK-α or IKK-β siRNA, single or in combination (Fig. 6). These results indicate that parthenolide can antagonize paclitaxel to diminish NF-κB-activated anti-apoptotic molecule expression in A549 cells.

Figure 6

Effect of parthenolide and/or silencing of IKK-β on apoptosis-related gene expression. IKK-β siRNA and control scrambled siRNA (data not shown)-transfected cells were treated with parthenolide (5 μmol/l) alone or in combination with paclitaxel (100 nmol/l). Equal amounts (50 μg/lane) of cytoplastic extracts were subjected to Western blotting at least three times with similar outcomes. The positions of molecular weight markers are noted on the right.

Mitochondrial apoptosis signalling cascade and effects of parthenolide

It has been reported that mitochondria play a minor role in paclitaxel-induced cell death of lung cancer cells [32, 33]. However, mitochondrial cytochrome c release during NF-κB inhibition was described previously [34]. Therefore, we next evaluated the possible contribution of cytochrome c release and caspase activation to the chemosensitization effect of parthenolide in A549 cells. A549 cells treated with paclitaxel only displayed a punctate staining pattern for cytochrome c (Fig. 7A); this pattern became diffuse upon parthenolide pre-stimulation, indicating release from the mitochondria. This was further verified by measurement of cytochrome c in the cytoplasm. A concentration of at least 5 μmol/l of parthenolide was necessary to induce cytochrome c release; this mirrored the concentration required to inhibit IKK and suggested a link between IKK inhibition and cytochrome c release.

Figure 7

The IKK inhibitor parthenolide induced mitochondrial cytochrome c release, which precedes cell death. (A) A549 cells treated with 100 nmol/l paclitaxel and 5 μmol/l parthenolide either single or in combination were stained with monoclonal antibodies against cytochrome c and analysed using a confocal microscope (upper panel). The cytosolic extract was prepared and subjected to Western blotting with a cytochrome c antibody (lower panel). (B) A549 cells were incubated with Z-VAD-fmk, and subsequently either 100 nmol/l paclitaxel or 5 μmol/l parthenolide were added. Cytosolic proteins were extracted, separated and incubated with antibodies recognizing active caspases 3 and 9 as indicated. (C) A total of 106 cells were treated with paclitaxel or paclitaxel + parthenolide in the absence or presence of 20 μmol/l Z-VAD-fmk. Cells were labelled with a fluorescein TUNEL kit and analysed by flow cytometry. The percentage of apoptotic cells was 28% (a, paclitaxel), 17% (b, paclitaxel + Z-VAD-fmk), 48% (c, paclitaxel + parthenolide), and 32% (d, paclitaxel + parthenolide + Z-VAD-fmk). This experiment was performed at least five times with similar outcomes. All the experiments shown in this figure were performed two to four times, producing identical outcomes, and one representative result is shown.

We next assessed by immunoblotting whether parthenolide induction of cytosol cytochrome c could lead to the activation of caspases 9 and 3. Activity of both caspases was influenced by parthenolide induction of cytochrome c leakage and associated with parthenolide-induced apoptosis. These effects were completely prevented in the presence of the pan-caspase inhibitor Z-VAD-fmk (Fig. 7B and C), further confirming that parthenolide regulates paclitaxel sensitivity by attenuating NF-κB activation and its downstream mitochondrial apoptosis signalling cascade.

Discussion

In the present study, activation of NF-κB was significantly induced by paclitaxel in chemotherapy-resistant cancer cell lines, supporting a role of NF-κB activation in the cellular mechanism of apoptosis resistance. This is consistent with previous reports in ovarian cancer, prostate cancer, pancreatic cancer and breast cancer. Our results also indicate that the phosphorylation of IKK-β was strongly increased by paclitaxel treatment, whereas p-IKK-α was essentially not affected. This result is in agreement with reports that IKK-β, and not IKK-α, is responsible for cytokineinduced activation of NF-κB [35, 36]. We also found that I-κB phosphorylation, which is catalysed by IKK (including IKK-α and IKK-β) [37, 38], was induced by paclitaxel treatment in parallel with NF-κB activation. Therefore, it is conceivable that a drug that inhibits NF-kB could inhibit or circumvent resistance to chemotherapeutic agents such as paclitaxel.

We further investigated an adjuvant approach to increase the efficacy of paclitaxel through inhibition of NF-κB. We demonstrated that parthenolide, a promising new multifunctional anti-cancer drug previously proven to inhibit NF-κB activity [39, 40], could contribute to the sensitization of A549 cells to paclitaxel-induced apoptosis. We observed that the cytotoxic effects of paclitaxel were potentiated by pre-treatment with 5 μmol/l parthenolide. At this concentration, we did not see noticeable induction of apoptosis by parthenolide alone. These data are consistent with studies that have inhibited NF-κB signalling by NF-κB antisense oligonucleotides [48], ectopically expressed I-κBαM [49-51] or used the proteasome inhibitor PS-341[52, 53] or I-κB inhibitor BAY 11–7085 [54, 55] to block I-κB degradation in different cancer cell lines. Targeting of the NF-κB pathway by proteasome inhibitors and I-κB kinase inhibitors has been used for both preclinical models and clinical trials involving patients with various types of tumours [56-58]. It has been suggested that pre-treatment is a prerequisite for enhanced cell sensitization to chemotherapy agents [46, 47]. Similarly, we found that pre-treatment, but not simultaneous administration, had a synergic effect on paclitaxel cell sensitivity.

Many researchers have shown that parthenolide can reversibly bind IKK and block its activity in vitro and in vivo[41, 42]. We demonstrated that parthenolide inhibited constitutive and paclitaxel-induced RelA/NF-κB activity by interfering with the effect of paclitaxel on IKK-β, thereby functionally inhibiting phosphorylation of I-κBα and subsequently leading to inhibition of NF-κB activity. However, whether paclitaxel-induced up-regulation and parthenolide-induced down-regulation of p65 are only regulated through IKK-β remains unclear. Some reports indicate that paclitaxel can directly induce phosphorylation of I-κBα[43]. Several other studies have also shown that paclitaxel directly activates survival pathway-related proteins such as Bcl-2, Akt, Cox-2 and mitogen-activated protein kinase independent of NF-κB [44, 45]. In the present study, transfection with siRNA-targeting IKK-α or IKK-β abolished parthenolide activity, indicating that when co-treated with paclitaxel, parthenolide might affect the NF-κB/I-κB cascade via IKK-β inhibition; this has not previously been indicated.

It has been verified that this synergetic effect of parthenolide and chemotherapeutic agents is specific to tubulin-modifying agents and cannot be observed with a variety of other cytotoxic agents [46]. Mechanistically, paclitaxel is tubulin-modifying agent that is different from DNA-damaging agents such as cisplatin, doxorubicin or camptothecin. Although apoptosis triggered by DNA-damaging agents is largely dependent on mitochondrial pathways, activation of caspases and mitochondria appears to be a secondary effect of paclitaxel treatment of NSCLC cells [33, 59], even though paclitaxel-induced NF-κB activity mediates inhibition of caspases [60, 61]. Our results also indicate that mitochondrial apoptosis pathways did not play an essential role in paclitaxel-induced cell death because inhibition of these pathways by Z-VAD-fmk resulted only in a temporary protection. Because parthenolide-induced apoptosis has been shown to be dependent on caspase activity [27-30], we next explored the possibility that the synergistic effect of paclitaxel and parthenolide might involve both cascade-dependent (parthenolide-induced) and -independent (paclitaxel-induced) mechanisms. If so, the induction of NF-kB and subsequent apoptotic resistance should be abolished by parthenolide.

The cytotoxic effect of paclitaxel was affected by Z-VAD-fmk treatment. However, caspase activation induced paclitaxel plus parthenolide at a concentration of 5 μmol/l was blocked by the caspase inhibitor Z-VAD-fmk, indicating that the effects of parthenolide occur mainly through a mitochondrial pathway. Treatment with paclitaxel plus parthenolide significantly increased cytochrome c release. Furthermore, the level of phosphorylated I-κBα, which was also reduced by paclitaxel administered in combination with parthenolide, showed an identical time course to the level of cytochrome c. These effects were blocked by siRNA-targeting IKK-β, indicating that cytochrome c and NF-κB might act in concert in response to the combined treatment. This result is consistent with previous studies in other systems [62, 63]. NF-κB activation can result in the transcription of some anti-apoptotic genes such as Bcl-xl [5, 60, 61, 64]. In the present study expression of Bcl-xl was reduced by parthenolide administered in combination with paclitaxel through down-regulation of NF-κB signalling; these effects paralleled the observed synergistic toxicity. Collectively, these observations indicate that the synergistic effect of paclitaxel and parthenolide is derived from reduced Bcl-xl induction via inhibition of NF-κB activity by parthenolide; thus synergistic cellular toxicity occurs at the level of NF-κB and downstream anti-apoptotic factors.

Several ongoing clinical trials are currently evaluating whether NF-κB inhibitors can enhance or restore the response to a wide range of therapeutics against both solid and haematological tumours [65-68]. Parthenolide, which can activate the IKK pathway and subsequently inhibit NF-κB and its downstream anti-apoptotic proteins, was also tested in cell lines derived from solid tumours [65, 68, 69]. The present study therefore suggests that parthenolide might be useful in combination therapy with paclitaxel; such a combination might limit adverse side effects.