Hypoxia inducible factor‐1 influences sensitivity to paclitaxel of human lung cancer cell lines under normoxic conditions

Abstract

Paclitaxel (PTX) is an anticancer drug that is effective against a wide range of solid tumors. The effect of PTX on two human lung cancer cell lines, PC14PE6 and NCI‐H441 cells, was examined in an orthotopically transplanted animal model with an in vivo imaging devise. Although PTX effectively suppressed tumor growth and improved survival rate in NCI‐H441, it did not influence these in PC14PE6. In vitro experiments confirmed that PC14PE6 cells are resistant to PTX under normoxic conditions and that both cell lines were resistant to PTX under hypoxic conditions. It was found that the expression level of endogenous hypoxia inducible factor (HIF)‐1α in PC14PE6 is much higher than that in NCI‐H441 cells under normoxic conditions. Furthermore, sensitivity to PTX in these cell lines was reversed when HIF‐1α expression was decreased by siRNA specific to HIF‐1α in PC14PE6 and increased by overexpression of the exogenous HIF‐1α gene in NCI‐H441. These results suggest that HIF‐1 influences the PTX sensitivity of these cells. The authors further examined β‐tubulin, a target molecule of PTX, with western blotting and immunohistochemical analysis in these cells. The expression level of β‐tubulin was comparable in these cells under both normoxic and hypoxic conditions while the distribution of β‐tubulin and cell morphology were changed according to HIF‐1α expression levels, suggesting that HIF‐1 influences the conformation and dynamics of microtubules. These data support the potential development of HIF‐1 targeted approaches in combination with PTX, where drug resistance tends to contribute to treatment failure. (Cancer Sci 2007; 98: 1394–1401)

Abbreviations:  

ATCC

American Type Culture collection

EGTA

ethylene glycol bis(βaminoethyl ether)‐N,N,N′,N′‐tetra‐acetic acid

FACS

fluorescence‐activated cell sorting

FBS

fetal bovine serum

GAPDH

glyceraldehyde‐3‐phosphate dehydrogenase

HIF

hypoxia inducible factor

HP

horseradish peroxidase

MDR

multidrug resistance

NP‐40

Nonidet P‐40

PBS

phosphate‐buffered saline

PCR

polymerase chain reaction

P‐gp

P‐glycoprotein

PTX

paclitaxel

RPMI

Roswell Park Memorial Institute

SDS

sodium dodecyl sulfate

siRNA

small inhibitory ribonucleic acid

SDC

sodium dodecyl chlorate

VEGF

vascular endothelial growth factor.

PTX is one of the most important agents for first‐line chemotherapy in clinical cancer management. It has shown promising response and improved the survival time in cancer patients with non‐small cell lung cancer, breast cancer, ovarian cancer, head and neck cancer, and so on.^{( 1 )} The cellular target for PTX has been identified as the microtubule system, which plays a significant role in mitosis, intracellular transport, cell motility and maintenance of cell shape.^{( 2 )} PTX arrests the cell cycle in cancer cells via inhibition of microtubule dynamics, in which PTX promotes the assembly of stable microtubules from α‐ and β‐tubulin.^{( 3 )}

Clinical and experimental studies have shown dramatic heterogeneities in the response of cancer cells to tubulin‐binding agents, including PTX. Approximately 20–50% of the human cancers are sensitive at the first treatment of PTX or other microtubule stability agents.^{( 4} , ⁵ , ^{6 )} This means that more than half of cancer cells are resistant to tubulin‐binding agents. It has been reported that microtubule‐stabilizing and microtubule‐destabilizing drugs inhibit HIF‐1α accumulation, which leads to suppression of HIF‐1 activity and subsequent tumor progression and angiogenesis.^{( 7} , ^{8 )}

Hypoxia is one of the major resistance factors in conventional chemotherapy. Within solid tumors, hypoxic cell chemoresistance was originally attributed to poor drug delivery and to the contention that hypoxic tumor cells are predominantly quiescent. More recently, the contributory role of HIF‐1 has been revealed.^{( 9} , ¹⁰ , ^{11 )} HIF‐1 is a heterodimeric transcription factor composed of α and β subunits and is activated in a wide range of cancer cells.^{( 12 )} HIF‐1 increases metabolic adaptation to O₂ deprivation by inducing several decad downstream genes, which in part play a role in tumor progression, including erythropoietin, transferrin, endothelin‐1, inducible nitric oxide synthetase, hemeoxygenase 1, and VEGF.^{( 13 )} The α subunit of HIF‐1 has been reported to be an important predictor of tumor progression for several types of solid tumors^{( 14} , ¹⁵ , ¹⁶ , ^{17 )} and to be associated with resistance to chemotherapy.^{( 18} , ^{19 )} Although Huang et al. have reported that induced HIF‐1α expression by CoCl₂ could increase the resistance to PTX in an ovarian cell line,^{( 20 )} a direct correlation between endogenous HIF‐1 expression levels and PTX sensitivity has not been described.

In the present paper the authors investigated the direct correlation between the resistance to PTX under normoxic conditions and HIF‐1α expression in two human lung cancer cell lines, PC14PE6 and NCI‐H441, which are PTX‐resistant and ‐sensitive, respectively, in an in vivo orthotopic lung cancer model. Although both cell lines were resistant to PTX under hypoxic conditions in vitro, only NCI‐H441 showed high sensitivity to PTX under normoxic conditions. The sensitivity of these cell lines to PTX was reversed by changing their HIF‐1α expression level using forced expression of the exogenous gene or suppressed expression with siRNA. HIF‐1α expression did not influence the protein level of tubulin but did significantly affect cell morphology and tubulin distribution. These data suggest that HIF‐1 influences cancer cell sensitivity to PTX by affecting tubulin.

Materials and Methods

Cell cultures and hypoxic treatment.  The NCI‐H441 cell line was obtained from the ATCC. The PC14PE6 human adenocarcinoma cells were selected from the parental PC14 cell line for the ability to form pleural effusions in the nude mouse.^{( 21 )} PC14PE6 and NCI‐H441 cell lines were maintained in RPMI 1640 (Gibco, Invitrogen Corporation, Carlsbad, CA, USA) supplemented with 10% FBS (Biosource, USA) supplemented with penicillin (100 U/mL) and streptomycin (100 µg/mL). To create a hypoxic condition of <0.02% of oxygen tension, the cells were treated in an hypoxic chamber, Baclite‐1 (Sheldon Manufacturing, Cornelius, OR, USA).

Plasmid constructions.  The human HIF‐1α was amplified using PCR from HepG2 cDNA. The primer sets used were as follows: forward primer KHIFATG: 5′‐AAAAGGTACCATGGAGGGCGCCGGC‐3′, reverse primer HIFTGAN: 5′‐CCCGCTAGCTCAGTTAACTTGATCCAAAGC‐3′. PCR products were digested with Kpn I and Nhe I, and subcloned into the Kpn I‐ and Nhe I‐digested pcDNA3.1/v5‐His expression plasmid vector (Invitrogen) (designated as pcDNA‐HIF‐1α). Annealed oligonucleotides for FLAG (5′‐CGCCACCATGGACTACAAAGACGATGACGACAAGGGTAC‐3′ and 5′‐CCTTGTCGTCATCGTCTTTGTAGTCCATGGTGGCGGTAC‐3′) were cloned into Kpn I‐digested pcDNA‐HIF‐1α (designated as pcDNA‐FLAG‐HIF‐1α). A construct was confirmed by sequence analyses using T7 and BGH primers with Big Dye Terminators^® and ABI Prism 310 Genetic Analyzer^® (Applied Biosystems, Foster City, CA, USA).

Gene silencing was performed using the pSuperRetro plasmid purchased from Oligoengine (Seattle, WA, USA). The corresponding sequence of the HIF‐1α oligonucleotide 5′‐GACAGTACAGGATGCTTGC‐3′^{( 22 )} was inserted to the plasmid to construct pSR‐ND‐HIF‐1α.

DNA transfection.   To establish PC14PE6/EF‐Luc and NCI‐H441/EF‐Luc cell lines, stable DNA transfection was carried out using a calcium phosphate method as described previously.^{( 23 )} Briefly, PC14PE6 and NCI‐H441 cells (10⁶/100‐mm dish) were transfected with 20 µg pEF/Luc or 5 HRE‐hCMVmp/luc plasmid, respectively. The cells were then trypsinized 24 h after transfection and cultured in the selection medium containing 5 µg/mL blasticidin‐S or 500 µg/mL G418 (Nacalai Tesque, Kyoto, Japan), respectively, for 10 days. The G418‐resistant colonies were isolated and used for in vitro cell proliferation assays and mouse xenograft assays.

For transient DNA transfection, cells (1 × 10⁶) were seeded in a 100‐mm dish 24 h before transfection. 10 µg of pSR‐ND‐HIF‐1α, pcDNA‐FLAG‐HIF‐1α or an empty vector was cotransfected with 1 µg pFGBH, which encodes EGFP‐BSD (blastcidin‐S‐resistant gene product) fusion protein to check the transfection efficiency, according to the manufacturer's instructions (Biorad, TransFectin^TM Lipid Reagent). The medium was replaced with the fresh selection medium containing 500 µg/mL G418 and 5 µg/mL blasticidin‐S 6 h after transfection. After incubation for approximately 24 h, the cells were trypsinized and reseeded to several dishes with the selection medium for 2–3 days. They were then examined for luciferase activity, HIF‐1α protein expression and sensitivity to PTX. They were incubated for approximately 10–12 h before either hypoxic or normoxic incubation for 24 h. The experiments were done in triplicate and each experiment was repeated at least three times.

Luciferase activity assay.  10⁵ cells were seeded onto 24‐well plaques in 1 mL 10% FBS RPMI 1640 medium, and cultured overnight, then further cultured under the normoxic or hypoxic condition for 24 h. The cells were then washed with PBS and lyzed with 100 µL passive lysis buffer (Promega, Madison, WI, USA) under normoxic or hypoxic conditions. Luciferase activity was measured using a luminometer (Lumat LB 9507; Berthold, Bad Wildbad, Germany) after the addition of 20 µL of a substrate reagent (luciferin; Promega) to 10 µL of cell lysate.

Western blotting analysis.  Cells were harvested and lyzed with 100 µL of western lysis buffer (50 mM Tris [pH 7.5], 1% NP‐40, 0.25% SDC, 150 mM NaCl, 1 mM EGTA). The cells cultured under hypoxic conditions were treated in the hypoxic chamber. The cell lysates were sonicated for 5 min and then centrifuged at 10 000g for 30 min. The supernatant was assayed to determine protein concentration and 30 µg per lane was applied to SDS‐polyacrylamide electrophoresis (10% for HIF‐1α and 12.5% for β‐tubulin and GAPDH). The protein was then transferred to a nitrocellulose filter (Hybond‐ECL; GE Healthcare Biosciences, Piscataway, NJ, USA) and the resultant filter was blocked with 5% non‐fat milk in PBS. For detection of HIF‐1α and GAPDH, the filters were probed with monoclonal anti‐HIF‐1α antibody (1:500, BD Bioscience Pharmingen, San Diego, CA, USA), mouse monoclonal anti‐β‐tubulin antibody (1:1000, Convance, CA, USA) and polyclonal anti‐GAPDH antibody (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA, USA), and incubated overnight at 4°C. Then HP‐linked secondary antibodies against antimouse and antirabbit IgG (GE Healthcare Biosciences) were incubated for 1 h at room temperature. Detection was carried out with a chemiluminescence‐based method using the ECL Plus western blotting detection system (GE Healthcare Biosciences).

FACS analysis of PTX sensitivity.  Cells (3 × 10⁵/300 µL) were seeded into 24‐well plates. The following day, in case of hypoxic conditions, the cells were preincubated in a hypoxic chamber for 6 h, various concentration of PTX were added to the culture medium and incubated for another 48 h. Then the cells were harvested, gently suspended in PBS and mixed with an equal volume of 2 × hypotonic fluorochrome solution (100 µg/mL propidium iodide in 0.2% sodium citrate–0.2% Triton X‐100) immediately before the analysis with flow cytometry using CELLQuest (BD Biosciences, Franklin Lakes, NJ, USA). The population of the cells with degraded smaller genomic DNA (subG1 fraction) was determined using CELLQuest Analysis (Becton Dickinson) and indicated as a percentage of dead cells.

Animals and orthotopic transplantation.  Cultured PC14PE6/EF‐Luc cells and NCI‐H441/EF‐Luc cells were harvested by exposure to 0.25% trypsin–0.02% EDTA solution. The cells were washed twice in PBS and resuspended in growth factor‐reduced Matrigel (0.5 mg/mL) in PBS. Cell viability was determined using the trypan blue exclusion test and only single‐cell suspensions of >90% viability were used for in vivo studies. 7–8‐week old male athymic nude mice (BALB/c nu/nu; Japan SLC Inc., Hamamatsu, Japan) were anesthetized using sodium pentobarbital (50 mg/kg body weight) and placed into the right lateral decubitus position. The skin over the left chest wall was cut between the 4th and 5th intercostal spaces at the midaxillar line and lung cancer cells (PC14PE6/EF‐Luc cell: 1 × 10⁶ in 50 µL; NCI‐H441/EF‐Luc cell: 5.0 × 10⁴ in 50 µL) were injected into the left lung through the left chest wall using a 30‐gauge needle following the method of Onn et al.^{( 24 )} After injection the skin was closed with wound clips.

In vivo imaging and PTX treatment.  For the in vivo imaging of bioluminescence, the tumor‐bearing mice were intraperitoneally injected with 200 µL of D‐luciferin solution (10 mg/mL in PBS; Xenogen, Alameda, CA, USA) 20 min before imaging. The animals were then applied to the IVIS^TM 200 Imaging System (Xenogen) to monitor the cells in solid tumor. The luciferase activity was calculated as the externally detected photon count, using Living Image Software 2.20 (Xenogen). When tumor formation was confirmed by imaging, the mice were randomized into four groups (n = 8) as follows: saline‐treated control (PC14PE6 and NCI‐H441 xenograft); PTX‐alone treatment. PTX (10 mg/kg) was given by intraperitoneal injection every 5 days for 3 or 4 times, beginning after the confirmation of tumor uptake by imaging. Body weights and photon counts were monitored every other day after the initial treatment. The ethical committee of the Kyoto University Institute of Laboratory Animals approved the study.

Immunofluorescence analysis.  After transient transfection and culture for 48 h with the selection medium, cells were seeded on Laboratory‐Teck chamber slides (Nalge Nunc International, Naperville, IL, USA) and incubated at 37°C with 5% CO₂ overnight. Cells were treated in the annormoxic chamber for 24 h and then they were fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.2% Triton X‐100/PBS for 4 min. Immunostaining was performed by incubating the slides with mouse monoclonal antibody polymerized β‐tubulin (1:1000; Convance), followed by incubation with Alexa Fluor 488 goat antimouse fluorescent antibody (1:100; Invitrogen) for 1 h at room temperature. As a negative control, slides were incubated with normal goat serum instead of the primary antibodies. Photos were taken under inverted fluorescent microscope (Biozero, Keyence, Japan).

Statistical analysis.  Data are expressed as means ± SEM. Statistical significance of differences was determined using a paired two‐tailed Student's t‐test and a log‐rank test for the survival curve. Differences were considered statistically significant at P < 0.05.

Results

In vivo imaging of PTX effects on the PC14PE6 and NCI‐H441 orthotopic lung cancer models.  To analyze tumor progression and antitumor effects of PTX in orthotopically transplanted human lung cancer cell lines, PC14PE6 and NCI‐H441, it was first established that the transfectants were stably expressing firefly luciferase. This enabled non‐invasive monitoring of the orthotopically transplanted xenografts by detecting bioluminescence from the outside with an in vivo imaging device. The transfectants grew with a similar speed to their corresponding parental cell lines (data not shown). The cells were injected into the left lung through the left chest wall and bioluminescence externally detected from the xenografts was monitored daily. The typical time course of tumor progression in both NCI‐H441 and PC14PE6 lung cancer was the following: the images were detected in the left lung 1 week after transplantation; the image spread to the right lung during the second week; the mice died during the third week (Fig. 1, control groups). When photon counts of tumors were approximately 5 × 10⁴ (6, 7 days after transplantation), the mice were randomized to start the PTX treatment. PTX was intraperitoneally injected every 5 days for 3 or 4 times. Body weights and photon counts of bioluminescence from the tumors were monitored every 2 days after the initial treatment. The image appearance and the photon counts during the experimental period were not significantly different between the untreated and PTX‐treated groups in PC14PE6 xenografts (1, 2). In addition, there was no significant difference in the mean survival time between untreated and PTX‐treated groups in PC14PE6 xenografts (Fig. 2C and Table 1). In contrast, the increases in images and photon counts were significantly delayed in the PTX‐treated group compared to the untreated group in NCI‐H441 xenografts (1, 2). Furthermore, the mean survival of PTX‐treated NCI‐H441‐tumor bearing mice was prolonged to 22 days compared with 12–13 days in untreated NCI‐H441‐tumor bearing mice and in both untreated and PTX‐treated PC14PE6‐tumor bearing mice (Fig. 2C,D and Table 1). There was no significant body‐weight loss associated with the treatments (data not shown). These data indicate that NCI‐H441 cells are sensitive to PTX in vivo but PC14PE6 cells are not.

Figure 1.

Optical imaging of PC14PE6 and NCI‐H441 cells in orthotopic lung cancer models. (A) PC14PE6/EF‐Luc and (B) NCI‐H441/pEF‐Luc cells were injected into the left lung. When tumor formation was confirmed by bioluminescent images, the mice were randomized into paclitaxel (PTX)‐treated (PTX) and untreated (control) groups. The images were taken on the indicated day, in which day 1 is the first treated day. The mice in the figure are representative of each group.

Figure 2.

Effects of paclitaxel (PTX) on PC14PE6 and NCI‐H441 tumor xenografts. (A, B) The bioluminescence intensity in (A) PC14PE6/EF‐Luc and (B) NCI‐H441/EF‐Luc xenografts treated with PTX (PTX) and untreated (control) were monitored every other day and quantified as photons/s/ROI. Each group consisted of at least 8 mice and the average photon counts ± SD of each group are shown. *P < 0.05 versus control group of NCI‐H441 xenografts. (C, D) Survival rate of (C) PC14PE6/EF‐Luc and (D) NCI‐H441/EF‐Luc orthotopic xenografts treated with PTX (PTX) and untreated (control) are shown. **P < 0.01 versus control group of NCI‐H441 xenografts. The median survival time of each treatment group is shown in Table 1.

Table 1.

Median survival time of PC14PE6/EF‐Luc and NCI‐H441/EF‐Luc orthotopic lung cancer models

Cell type	Treatment group	No. of mice	Median survival time (days ± SD)
PC14PE6/EF‐Luc	Control (saline)	9	12 ± 1.49
	Paclitaxel	9	13 ± 1.49
NCI‐H441/EF‐Luc	Control (saline)	13	12 ± 4.39
	Paclitaxel	11	22 ± 9.34 †

^† P < 0.05 versus control group.

Effects of PTX on the PC14PE6 and NCI‐H441 cells in vitro.  To investigate the effects of PTX on PC14PE6 and NCI‐H441 cells in vitro, they were treated with various concentrations of PTX. The effect of PTX on cell death was examined using FACS analysis, in which cells with degraded genome DNA corresponds to the sub‐G1 fraction (Fig. 3A). PTX induces polymerization of tubulin and thus blocking of the cell cycle in the M phase,^{( 25 )} It was found that PTX did not induce cell death to PC14PE6 cells under both hypoxic and normoxic conditions, even at 100‐nM PTX concentration (Fig. 3A). At this concentration, PC14PE6 significantly increased G2/M population (Fig. 3C). In contrast to PC14PE6, NCI‐H441 cells showed significant sensitivity to PTX under normoxic conditions. Even 0.01 nM PTX induced NCI‐H441 to cell death and increased G2/M population under normoxic conditions (Fig. 3B,C). Under hypoxic conditions, however, both PC14PE6 and NCI‐H441 cells showed significant resistance to PTX (Fig. 3B,C).

Figure 3.

Effects of paclitaxel (PTX) on PC14PE6 and NCI‐H441 cell lines in vitro. (A) PC14PE6/EF‐Luc and (B) NCI‐H441/EF‐Luc cells were seeded into 24‐well plates and the following day various concentrations of PTX (as indicated) were added to the culture medium and incubated for another 48 h under normoxic or hypoxic conditions. The percentage of dead cells with degraded genomic DNA was analyzed using fluorescence‐activated cell sorting (FACS) after staining genomic DNA with propidium iodide. Each experiment was done in triplicate and the average dead cell percentage in each group is shown. Experiments were repeated more than three times and representative data is shown. Error bars indicate SEM. (C) Representative FACS data of each group.

Suppression of HIF‐1α expression with siRNA increases the sensitivity to PTX of PC14PE6 cells.  Because HIF‐1 is the major regulator in hypoxia, the authors investigated whether HIF‐1 activity was related to the resistance to PTX. The HIF‐1α protein expression level in PC14PE6 and NCI‐H441 cells was examined first and it was found that PC14PE6 expressed a significantly higher level of HIF‐1α than NCI‐H441 cells even under normoxic conditions (Fig. 4).

Figure 4.

Western blot analysis of hypoxia inducible factor (HIF)‐1α protein expression in the PC14PE6 and NCI‐H441 cells. The cells were cultured under normoxic (N) and hypoxic (H) conditions for 24 h and then cells were harvested and lyzed with western lysis buffer under normoxic or hypoxic conditions. The lysates were subjected to an immunoblotting analysis with anti‐HIF‐1α and anti‐glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) antibodies as described in Materials and Methods.

To further analyze the correlation between HIF‐1 and PTX sensitivity in PC14PE6 and NCI‐H441 cells, the influence of HIF‐1α expression levels on PTX sensitivity of these cells was examined next.

HIF‐1α expression in PC14PE6 cells was suppressed with an siRNA specific to HIF‐1α. When a plasmid expressing siRNA specific to HIF‐1α was transfected to PC14PE6 cells, the HIF‐1α protein levels were significantly reduced (Fig. 5A). FACS analysis revealed that PC14PE6 cells with reduced HIF‐1α expression significantly increased the sensitivity to PTX compared to the cells transfected with an empty vector (Fig. 5B, P < 0.05).

Figure 5.

Suppression of hypoxia inducible factor (HIF)‐1α expression with siRNA increases the sensitivity to paclitaxel (PTX) of PC14PE6. (A) The cells were transfected with plasmids encoding enhanced green fluorescent protein (EGFP) and siRNA specific to HIF‐1α (+) or empty vector (–) and cultured under normoxic (N) or hypoxic (H) conditions for 48 h with selection medium. The cells were harvested at the end of the culture period and lyzed with western blot buffer under normoxic or hypoxic conditions, respectively. Western blotting was done as described in the legend for Fig. 4. (B) The cells prepared as for (A) were further treated with PTX for 24 h at the indicated concentration. The cells were harvested at the end of the culture period and the DNA contents of EGFP‐positive cells were analyzed using fluorescence‐activated cell sorting (FACS) as described in the legend for Fig. 3AB. *P < 0.05 versus control group. Experiments were repeated in triplicate more than three times and representative data is shown. Error bars indicate SEM.

Forced expression of HIF‐1α reduced the sensitivity to PTX of NCI‐H441 cells.  When a plasmid encoding HIF‐1α was transfected to NCI‐H441 cells, the HIF‐1α protein levels were significantly increased (Fig. 6A). The HIF‐1α overexpressing NCI‐H441 cells were significantly more resistant to PTX than the cells transfected with a control plasmid (Fig. 6A, P < 0.05). Together, these results suggest that HIF‐1 activity influences the sensitivity to PTX.

Figure 6.

Forced expression of hypoxia inducible factor (HIF)‐1α reduced the sensitivity to paclitaxel (PTX) of NCI‐H441 cells. (A) The cells were transfected with plasmids encoding enhanced green fluorescent protein (EGFP) and HIF‐1α (+) or empty vector (–) and cultured under normoxic (N) or hypoxic (H) conditions for 48 h with selection medium. The cells were harvested at the end of the culture period and lyzed with western blot buffer under normoxic or hypoxic conditions. Western blotting was done as described in the legend for Fig. 4. (B) The cells prepared as for (A) were further treated with PTX for 24 h at the indicated concentration. The cells were harvested at the end of the culture period and the DNA contents of EGFP‐positive cells were analyzed using fluorescence‐activated cell sorting (FACS) as described in the legend for Fig. 3AB. Experiments were repeated in triplicate more than three times and representative data is shown. Error bars indicate SEM.

HIF‐1 influences the conformation of β‐tubulin.  To investigate the mechanism by which HIF‐1 influenced the PTX sensitivity in these cells, their β‐tubulin expression was first examined. The β‐tubulin protein level was not significantly different between PC14PE6 and NCI‐H441 cells under both normoxic and hypoxic conditions (Fig. 7A). Then the β‐tubulin distribution in the cells was examined using immunohistochemical analysis (Fig. 7B). In PC14PE6 cells, reduced expression of HIF‐1α resulted in flatter morphology and less‐assembled β‐tubulin under normoxic conditions (Fig. 7, upper left panels). In contrast, forced expression of HIF‐1α in NCI‐H441 cells resulted in more round morphology and more bundled β‐tubulin under normoxic conditions (Fig. 7, upper right panels). All the cells appeared flatter under hypoxic conditions compared to the corresponding cells under normoxic conditions (Fig. 7, lower panels).

Figure 7.

Influence of hypoxia inducible factor (HIF)‐1α expression on β‐tubulin conformation. (A) β‐tubulin protein expression in the PC14PE6 and NCI‐H441 cells. The cells were transfected with plasmids encoding enhanced green fluorescent protein (EGFP) and siRNA specific to HIF‐1α (+) or empty vector (–) to PC14PE6 or HIF‐1α (+) or empty vector (–) to NCI‐H441 and cultured under normoxic (N) or hypoxic (H) conditions for 48 h with selection medium. Cells were harvested and lyzed with western lysis buffer under normoxic or hypoxia conditions, then subjected to immunoblotting with anti‐β‐tubulin and anti‐glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH) as described in Materials and Methods. (B) Immunofluorescence analysis of polymerized β‐tubulin protein expression in PC14PE6 and NCI‐H441 cell lines. After transient transfection and culture for 48 h with the selection medium, cells were seeded on chamber slides and incubated overnight. Cells were treated in the annormoxic chamber for 24 h and immunofluorescence analysis was carried out on human lung cancer cell lines, PC14PE6 and NCI‐H441, with the combination of polymerized β‐tubulin antibody as described in Materials and Methods. Representative results from three independent experiments are shown. Scar bar = 10 µm.

Discussion

The authors demonstrate here that the HIF‐1α expression level influences the sensitivity of human cancer cell lines, PC14PE6 and NCI‐H441, to PTX. The endogenous HIF‐1α protein level in PC14PE6 is significantly higher than that of NCI‐H441 (Fig. 4), which corresponds to their PTX sensitivity, that is, PC14PE6 is more resistant to PTX than NCI‐H441 in vivo (Fig. 1) as well as in vitro (Fig. 3). Furthermore, when HIF‐1α expression was suppressed by siRNA, PC14PE6 increased sensitivity to PTX (Fig. 5), while increased expression of HIF‐1α in NCI‐H441 resulted in increased resistance to PTX (Fig. 6). The authors have repeatedly observed that the suppression of HIF‐1α expression by siRNA slightly decreased the viability of PC14PE6 cells under normoxic conditions (Fig. 6B). In a pancreatic cell line, AsPC‐1, the knockdown of HIF‐1α also induced minor cell death even under normoxic conditions (M. Inoue et al., unpublished data, 2007).

Although reduced HIF‐1α expression by itself influenced the viability of cancer cells, it significantly enhanced the cell death of PTX‐treated PC14PE6 cells (Fig. 5B). These results suggest that HIF‐1 is one of the factors influencing PTX sensitivity of cancer cells under normoxic conditions. HIF‐1α expression levels did not influence the expression level of tubulin protein (Fig. 7A), but did influence the cellular distribution of tubulin and consequently cell morphology (Fig. 7B), suggesting that HIF‐1 influenced the conformation of tubulin in these cells. All the cells appeared flatter under hypoxic conditions. These phenomena have been observed not only in cancer cells but also in other types of cells and have been explained by hypoxia‐induced alterations in cytoskeletal fiber, actin and tubulin, and extracellular components such as metalloproteinase.^{( 26} , ^{27 )} A recent study strongly supported the role of hypoxia in cytoskeleton changes, as it showed that hypoxia increased tubulin stabilization and changed vesicle trafficking.^{( 27 )}

Several studies support the notion that the cell cycle plays also a critical role in chemosensitivity for combination chemotherapy. PTX‐induced apoptosis can occur either directly after a mitotic arrest^{( 28 )} or following an aberrant mitotic exit into a G1‐like ‘multinucleate state.’^{( 25} , ²⁸ , ^{29 )} Under normoxic conditions, both NCI‐H441 and PC14PE6 cells increased G2/M population when treated with PTX (Fig. 3C), while their destinies were different. Under hypoxic conditions, however, they were arrested in neither G1 nor G2/M, suggesting that the effects of PTX under hypoxic conditions might be completely different from those under normoxic conditions. Substantial evidence indicates that the G2/M arrest of the cell cycle is not the only mechanism for PTX‐induced apoptosis.^{( 30} , ³¹ , ³² , ^{33 )} Additional signal transduction pathways are also involved in inducing apoptosis.^{( 34} , ^{35 )} Further study should be done to clarify the mechanisms resistant to PTX under hypoxic conditions.

Although it is not known how HIF‐1 influences the conformation of tubulin, previous reports have suggested some possible mechanisms for the association between HIF‐1α and PTX resistance. At present, the best‐described mechanism of resistance to tubulin‐binding agents is the MDR pump model. In a number of cases,^{( 36} , ³⁷ , ^{38 )} development of cell lines resistant to paclitaxel has been shown to be associated with the expression of MDR1. But little is known concerning the significance of the MDR phenotype in the emergence of resistant tumors in patients treated with tubulin‐binding agents. Clinical trials that have aimed to sensitize MDR‐positive tumors to agents such as vinblastine with P‐gp modulators have been disappointing.^{( 4 )}

Comerford et al. found that HIF‐1 induced MDR1 gene expression and concomitant functional P‐gp expression in both transformed epithelia and primary cultured endothelia and suggested that P‐gp expression represented a pathway for resistance of some tumors to chemotherapeutics.^{( 9 )} Therefore, the authors examined P‐gp expression in PC14PE6 and NCI‐H441 under both normoxic and hypoxic conditions. However, P‐gp protein was not detected in either cell lines under both conditions, although other cancer cell lines such as HepG2 expressed a significant amount of P‐gp protein under both conditions (data not shown). In these human cancer cell lines, P‐gp seems not to contribute to PTX resistance.

Another explanation includes the impacts of HIF‐1α‐induced downstream genes on the cell cycle and apoptosis‐associated pathways. HIF‐1α can influence the product of the tumor suppressor gene p53,^{( 39} , ^{40 )} which can induce programmed cell death by regulating proteins such as Bax, or cause growth arrest that is mediated by p21. It has been shown that HIF‐1α directly binds to the p53 ubiquitin ligase mdm2 both in vivo and in vitro, thereby stabilizing p53.^{( 41 )} However, there has also been a report that showed direct binding of p53 to the oxygen‐dependent degradation domain of HIF‐1α.^{( 42 )} HIF‐1α interacts with wild‐type p53 but not with tumor‐derived mutant p53.^{( 42 )} This suggests that the contribution of differing p53 status of cells might affect their response to stress factors such as PTX treatment for cancer cells. Thus the status of the p53 gene might play a role in the difference in PTX resistance in PC14PE6 and NCI‐H441 because PC14PE6 is wild‐type p53 and NCI‐H441 has a mutant‐type p53.^{( 43 )} Whether the difference in p53 status in PC14PE6 and NCI‐H441 plays a role or not in the association between PTX and HIF‐1α in the present experiments has not been revealed. Further study should be done to clarify this issue.

PTX resistance is also explained based on the dynamics of microtubules.^{( 44 )} Cabral described a PTX resistance model of ‘hypostable’ and ‘hyperstable’ microtubules.^{( 45 )} Hypostable microtubules tend toward depolymerization spontaneously, and hyperstable microtubules are relatively resistant to depolymerization. Cells with hypostable microtubules are hypersensitive to the depolymerizing agents such as vinca alkaloids while resistant to microtubule stabilizing agents such as PTX. Conversely, cells containing hyperstable microtubules are resistant to the vinca alkaloids but relatively sensitive to PTX. Jordan and Wilson suggested another model.^{( 46 )} They analyzed the dynamic behavior of individual microtubules and found that PTX at low concentrations reduces microtubule dynamics without significant alteration in microtubule length. According to this explanation, the cells with highly dynamic microtubules are more sensitive to the microtubule‐stabilizing agents such as PTX than the cells with less dynamic microtubules.

Although the authors do not know the mechanism by which HIF‐1 changes the stability and dynamics of microtubules, it is speculated that unknown factors induced by HIF‐1 may influence the conformation and/or dynamics of microtubules. Therefore, it is expected that HIF‐1 targeted approaches in combination with PTX may contribute to improve outcome of PTX treatment.