Pancreatic cancer is characterized by extensive local invasion and early lymphatic and hematogenous metastasis.1 In fact, only 1–4% of all patients can survive 5 years after diagnosis as the carcinoma of pancreas.2 Although surgical operations represent the only curative treatment available for the patients with pancreatic cancer, 5-year survival rates are <20% even after surgery,3, 4, 5 indicating that effective adjuvant treatments are required. Among anticancer agents used in the treatment of pancreatic cancer, 5-fluorouracil (5-FU) has been known as a standard drug for the effective treatment of pancreatic cancer, but the response rate of 5-FU was quite low.6, 7, 8
During the last few years, a novel nucleoside analogue gemcitabine (2′, 2′-difluoro-deoxycytidine) has been reported to be an effective agent in the treatment of pancreatic cancer. Gemcitabine exhibited in vitro and in vivo the growth suppression of the human pancreatic cell lines that had previously shown insensitivity to multi-drugs such as 5-FU, doxorubicin or cisplatin.9 Furthermore, gemcitabine improved the survival and clinical benefit responses compared to 5-FU.10 Gemcitabine is now a standard first-line treatment for the patients with pancreatic cancer.10, 11
Gemcitabine is metabolized sequentially to nucleoside monophosphate, diphosphate, and triphosphate by deoxycytidine kinase (dCK) after the entering cell. The difluoro-deoxycytidine triphosphate is incorporated into DNA, resulting in chain termination. Because gemcitabine is often effective in multi-drug resistant cells,9 mechanisms of gemcitabine sensitivity have been investigated from several aspects. Buchler et al. reported that alterations of apoptosis-regulating genes, such as bcl-2, bcl-xL and bax, regulate the sensitivity to gemcitabine.12, 13 Veronique et al.14 identified the dCK deficiency as the genes responsible for the gemcitabine resistance. Indeed, the mechanisms of gemcitabine resistance are still controversial, although many studies were examined. Additionally, although there were reported individual gene expressions, any comprehensive expression change of the genes has not been investigated for gemcitabine sensitivity.
The cDNA microarray technique is one of methods that allow measurement of temporal changes in thousands of gene expression profiles during the development of anti-cancer drug resistance. Indeed, this methodology has been used to analyze alterations in the gene expression responsible for doxorubicin sensitivity15 or paclitaxel sensitivity.16 We established a human pancreatic cancer cell line showing decreased sensitivity to gemcitabine. Furthermore, cDNA microarray was utilized to monitor mRNA expression and to find a molecule, selenoprotein P, that might contribute to the gemcitabine-resistance. We showed that selenoprotein P suppressed the intracellular free radicals, with a potent cytotoxicity, induced by gemcitabine. Our study suggests a new mechanism for gemcitabine sensitivity in human pancreatic cancer cells.
MATERIAL AND METHODS
Human KLM1, PK-45P and MIA Paca2 pancreas cancer cell lines were cultured in RPMI 1640, supplemented with 10% heat-inactivated FBS, streptomycin and penicillin at 37°C and 5% CO2. A gemcitabine-resistant KLM1 cell line was established by exposure to gemcitabine as described previously.17 The KLM1 cells cultured at an initial density of 1 × 106cells on 6-well flat-bottomed plates containing 2 ml medium for 1 day were treated with 10 μg/ml gemcitabine for 1 week. Cells were then cultured in a gemcitabine-free medium for 2 weeks to recover cell density. After repeating the above treatment 4×, we established a cell line that exhibited stable characteristics with respect to growth rate, morphology and drug resistance.
The viability of cells after exposure to gemcitabine was determined by MTT assay as follow. The cells were seeded on 96-well flat-bottom plates. At 24 hr after seeding, the indicated amount of gemcitabine was added to the wells, and incubation was continued for another 72 hr. The colorimetric reaction was initiated by adding 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl [2H]-tetrazolium bromide (MTT) at a concentration of 5 mg/ml. The formazan crystals were dissolved in 0.04 N acid-isopropanol, and the absorbance at 570 nm was quantified using a microtiter plate spectrophotometer. The 50% inhibitory concentration (IC50) was estimated from individual inhibition curves and represents the concentration of drug that inhibits cell proliferation by 50%.
The viability of cells after exposure to gemcitabine with selenoprotein P was determined by MTT assay as follow. The cells were seeded on 96-well flat-bottom plates. At 12 hr after seeding, 1 μg/ml selenoprotein P was added to the wells. After incubation for another 24 hr, the indicated amount of gemcitabine was added to the wells, and incubation was continued for another 72 hr. The numbers of viable cells was counted by MTT assay.
The viability of cells after exposure to H2O2 was determined as follows. Cells were seeded on 96-well and incubated for 24 hr. After incubation, the indicated amount of H2O2 was added to the wells, and incubation was continued for another 24 hr. The numbers of viable cells was counted by MTT assay.
The sensitivity of cells to 5FU, 1-β-D-arabinofuranosylcytosine, adriamycin, bleomycin, cisplatin, etoposide, paclitaxel and docetaxel was determined by MTT assays. The cells were seeded on 96-well and incubated for 24 hr. After incubation, each anticancer drug was added to the wells and incubation was continued for another 24 hr. The numbers of viable cells was counted by MTT assay.
KLM1-R cells were incubated for 24 hr with medium containing indicated amount of IFN-γ. After incubation, Total RNA was extracted from the cells using RNeasy mini kit (Qiagen, Valencia, CA) according to the manufacture's protocol. The sensitivity of cells to gemcitabine after IFN-γ treatment was determined as follows. The cells were seeded on 96-well and incubated for 24 hr. After incubation, 200 ng/ml IFN-γ was added to the wells. After an additional 24 hr incubation, the indicated amount of gemcitabine was added to the cells. The viability of the cells was determined using MTT assay.
cDNA microarray analysis
Human 1 cDNA Microarray Kit (Agilent Technologies, Palo Alto, CA) spotted 12,814 genes was used to analyze the different gene expression in the KLM1 and KLM1-R cells. Total RNA was extracted from the cells using RNeasy mini kit (Qiagen, Valencia, CA) according to the manufacture's protocol. The quality of RNA was assessed by running aliquots on agarose gels. Twenty micrograms of RNA were reverse transcribed for 1 hr at 42°C with Cyanine 3-dUTP or Cyanine 5-dUTP using Fluorescent Direct Labeling cDNA Synthesis Kit (Agilent Technologies). Hybridization experiments were carried out as a dye-swapped pair; i.e., hybridization to one microarray with sample X labeled with cyanine 3 and sample Y labeled with cyanine 5, and to second microarray with sample X labeled with cyanine 5 and sample Y labeled with cyanine 3. Averaging the expression measurements obtained from the 2 dye-swapped hybridization should have minimized the impact of any dye-specific biases, enabling an accurate measurement of the differential expression levels. A cyanine3-/cyanine 5-labeled cDNA sample was resuspended in 7.5 μL of nuclease-free water. Cot-1 DNA was added in this cDNA sample for minimizing background fluorescence and incubates 98°C for 2 min to denature cDNA. After the sample was pipetted onto each microarray, the slide is placed in a hybridization chamber and incubated at 65°C for 17 hr. After incubation at 65°C, the slide was washed with wash solution 1 (0.5× SSC, 0.01% SDS) and wash solution 2 (0.06× SSC) and centrifuged to dry. The intensity of each hybridization signal was scanned in both Cy 3 and Cy 5 channels with Scan Array 4000 (GSI Lumonics, Billerica, MA) with a 10 μm resolution. The signal was converted into 16-bits-per-pixel resolution, yielding a 65,536-count dynamic range. A Quant Array soft (GSI Lumonics) was used for image analysis. The elements were determined by a gridding and region-detection algorithm. The area surrounding each element image was used to calculate a local background, which was subtracted from the total element signal. Background-subtracted element signals were used to calculate Cy3:Cy5 ratios. The Cy3:Cy5 ratio for each sample was calculated by global normalization.
Single stranded DNA was synthesized in a reaction mixture containing 120 pmol random primers and Moloney murine leukemia virus transcriptase. PCR reaction mixture contained 10 pmol of forward and reverse primers and 2 U of TaqMan DNA polymerase (Takara, Japan). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the internal control. The amplification and primer was shown as follows. Conditions were: initial denaturation step for 4 min at 94°C then 30 sec at 94°C, 30 sec at each annealing temperature, and 30 sec at 72°C followed by an elongation step for 5 min at 72°C. Selenoprotein P (sense: 5′-AAC AGA GAG CCA GGA CCA AA-3′ and antisense 5′-AAA GTT AGG AAG GAA AAA GG-3′, annealing temperature 53.7°C for 28 cycles), cytosolic glutathione peroxidase (sense: 5′-AGG AGA ACG CCA AGA ACG AA-3′ and antisense 5′-GGG ATC AAC AGG ACC AGC AC-3′, annealing temperature 59.4°C for 25 cycles), glutathione peroxidase-GI (sense: 5′-GTG AGG TGA ATG GGC AGA AC-3′ and antisense 5′-GGC AGA GGG GAA AGG CAA GG-3′, annealing temperature 57.0°C for 28 cycles), thioredoxin reductase1 (sense: 5′- TCG AAA TTA TGG ATG GAA AG-3′ and antisense 5′-CAG TAA GGC AAG GAG AAA AG-3′, annealing temperature 51.4°C for 25 cycles), selenoprotein W (sense: 5′-AAG AAG AAA GGC GAT GGC TAC-3′ and antisense 5′-AGG AGG GTG GGG TGG TGT GG -3′, annealing temperature 58.8°C for 28 cycles), dCK (sense: 5′-GCA TGA ATG AGA CAG AGT GG-3′ and antisense 5′-AGA TAA TCG AAG TTG GTT TT-3′, annealing temperature 49.1°C for 28 cycles), GAPDH (sense: 5′-GTC AAC GGA TTT GGT CGT ATT-3′ and antisense 5′-AGT CTT CTG CGT GGC ACT CAT-3′, annealing temperature 56.0°C for 25 cycles).
Dichlorofluorescein diacetate assay
One hundred micromoles (final concentration) dichlorofluorescein diacetate (DCF-DA) was added to wells 2 hr before cell harvest. At the time of the assay, the cells in each well were washed once and suspended in 5 ml of PBS. Then 0.5 ml of cell suspension from each well was diluted with 2.5 ml of PBS in a cuvette and the emission fluorescence intensity (FI) was taken at 520 nm using an excitation wavelength of 488 nm.
Purification of Selenoprotein P
Selenoprotein P was purified from human plasma using conventional chromatographic methods as described previously.18
Statistical evaluations of numerical variables in viability of cells and DCF assay were carried out using Mann-Whitney's U-test. Significance was defined as p < 0.05.
Establishment of a pancreatic cell line showing insusceptibility
Resistance to gemcitabine was induced in a human pancreatic cancer cell line KLM1 by continuous exposure of 10 μg/ml gemcitabine (Fig. 1). The first variant was established after 4 weeks and termed KLM1-R. After repeated exposure to 10 μg/ml gemcitabine for 4 months, we established a cell line that exhibited stable characteristics with respect to growth rate, morphology and drug resistance. The KLM1-R cell line exhibited the viability to 3.5-fold in 10 μg/ml gemcitabine compared to KLM1 cells (p < 0.05). Based on the IC50 measurements, defined as the concentration of gemcitabine causing 50% growth inhibition, KLM1-R was found to be 20.06-fold resistant to gemcitabine compared to KLM1. Microscopically the parent cell line KLM1 and the resistant KLM1-R variant were not different. Cross-resistance was observed only for bleomycin, known as an agent generating intracellular reactive oxygen species (ROS),19 although no cross-resistance was observed for other agents (Table I).
Table I. IC50 of Several Anticancer Drugs as Determined in KLM1 and KLM1-R Cells1
cDNA microarray to determine alteration of gene expression
The expression of alteration of 12,814 clones was analyzed by cDNA microarray between the KLM1 and KLM1-R cells (Fig. 2a). Table II demonstrates 25 genes that displayed the altered fluorescence ratios of >5-fold. Among the 25 differentially expressed genes, 5 genes were upregulated and 20 were downregulated in KLM1-R cells. It is noteworthy that the most of the upregulated gene was selenoprotein P (9.699-fold) in KLM1-R cells on the examined microarray. Selenoprotein P is one of the major selenium-rich extracellular proteins in human plasma.18 The family of selenoproteins has been reported to play critical roles in anti-oxidative reaction.20, 21, 22, 23 Our microarray analysis clarified no alteration of the expression of bcl-2 related genes (bcl-2, bfl-1, bag-1, bad, bak, bcl-xL) or multi-drug resistance genes (MDR1, MDR3, MRP5) in the gemcitabine resistant KLM1-R cells (data not shown).
Ratio means the expression in KLM1-R/that of KLM1.
Selenoprotein P, plasma 1
TNF ligand superfamily, member 10
Prostaglandin-endoperoxidase synthase 2
Protein tyrosinephosphatase, non-receptoe type 2
Insulin-like growth factor binding protein 7
Melanoma antigen, family A, 9
Aldehyde dehydrogenase 3
Aldehyde dehydrogenase 6
S100 calcium-binding protein A2
Four and a half LIM domains 1
Regulator of G-protein signaling 3
Calponin 3, acidic
Coagulation factor III
Dickkopf (Xenopus laevis) homolog 1
Lymphocyte antigen 6 complex, locus D
Fructose- 1,6-bisphosphatase 1
Cystein-rich protein 1 (intestinal)
Expression of selenoprotein P in pancreatic cancer cells
To evaluate expression of selenoproteins mRNA, RT-PCR analysis was assessed on KLM1 and KLM1-R cells. As shown in Figure 2b, the mRNA of selenoprotein P was clearly expressed in KLM1-R cells, but not in KLM1 cells. These results were compatible with the microarray results. There was no difference of mRNA expression in other selenoproteins (cellular glutathione peroxidase, glutathione peroxidase-GI, selenoprotein W and thioredoxin reductase 1) between KLM1 and KLM1-R cells. It should be mentioned that no different expression of dCK, which is the converting enzyme for gemcitabine and one of genes contributing to the gemcitabine resistance.14 These data indicate that selenoprotein P expression might be associated with the insusceptibility to gemcitabine in the pancreatic cancer cells.
Selenoprotein P improves the cell viability exposed to gemcitabine
To assess whether selenoprotein P protects the pancreatic cell lines from the toxicity of gemcitabine, cell survival was determined after exposure to gemcitabine. As demonstrated in Figure 3, the addition of 1 μg/ml selenoprotein P to the medium recovered the viability of pancreatic cancer cells after exposure of gemcitabine for 72 hr. These findings demonstrate important evidence that the selenoprotein P downregulates the susceptibility to gemcitabine in the pancreatic cancer cells.
Selenoprotein P suppresses gemcitabine-induced intracellular ROS
Because selenoprotein P is known as an ROS scavenger,18, 23, 24, 25 the ROS generation was then analyzed in the pancreatic cancer cells after gemcitabine exposure. The intracellular ROS was measured in KLM1 cells by DCF-DA assay. Figure 4 showed that the ROS levels were increased 2.91-fold by 72-hr exposure to gemcitabine compared to the control (p = 0.02). In contrast, the administration of selenoprotein P induced only 1.89-fold ROS by gemcitabine exposure (p = 0.08). The addition of selenoprotein P suppressed the gemcitabine-induced intracellular free radicals that have extensive cytotoxic effects even on cancer.26
Susceptibility to oxidative stress in pancreatic cancer cells
To assess the susceptibility to oxidative stress in KLM1-R cell, survivals were determined after exposure to H2O2. Because H2O2 is readily converted to hydroxyl radicals, it is considered that a cytotoxic agent causes damage to many cellular components. After exposure to H2O2, cell survivals were determined. As shown in Figure 5, the H2O2-mediated cytotoxicity was suppressed in KLM1-R. After exposure to 500 μM H2O2, the number of viable KLM1-R cells was 14.9-fold compared to that of KLM1 cells (p < 0.01). These findings indicate that the KLM1-R, which we established as a cell line less sensitive to gemcitabine, acquired potent resistance to oxidative stress.
IFN-γ suppresses mRNA expression and activity of selenoprotein P
Because negative regulation of selenoprotein P promoter by cytokines including IFN-γ was known,27 IFN-γ was used to increase the cytotoxicity by gemcitabine through suppression of selenoprotein P. After treatment of IFN-γ for 24 hr, the mRNA expression of selenoprotein P was suppressed (Fig. 6a). Densitometric analysis of band intensities showed suppression in selenoprotein P-mRNA with 200 ng/ml IFN-γ (Fig. 6b). Although IFN-γ has no cytotoxicity (Fig. 6c), the cytotoxicity of gemcitabine to KLM1-R was increased after treatment of 200 ng/ml IFN-γ for 24 hr (Fig. 6d). Furthermore, the intracellular ROS was measured in KLM1-R after IFN-γ treatment. The administration of IFN-γ increased the intracellular ROS levels in KLM1-R (Fig. 6e). These findings indicated that IFN-γ suppressed the expression of selenoprotein P and increased the intracellular ROS levels, leading to improvement of the sensitivity to gemcitabine.
The novel nucleoside analogue gemcitabine is well known as one of the most effective anticancer agents, however, the mechanism of sensitivity has not been clearly understood. In our study, cDNA microarray analyses showed that the selenoprotein P gene was overexpressed in a gemcitabine-resistant pancreatic cancer cell line. Selenoprotein P is a member of the anti-oxidative selenoprotein family that is composed of more than 15 selenoproteins in mammalians, i.e., glutathione peroxidase, thioredoxin reductase, and selenoprotein W. The selenoprotein family has been recognized as playing a survival role in suppression of free radicals.20, 21, 22 In particular, selenoprotein P accounts for about 50% of total plasma selenium in humans, and differs from all other selenoproteins identified so far by its higher selenium content. Human selenoprotein P is predicted to contain 10 selenocysteine residues whereas other selenoproteins contain only 1 selenocysteine residue per subunit.
Previous studies demonstrated a role of selenoprotein P as a protective agent against the oxidation and nitration reactions mediated by peroxynitrite, a potent oxidant generated in vivo.28 The function of selenoprotein P was reported not only as an antioxidative enzyme but also as a selenium supplier to the cell.29 Selenoprotein P-knockout mice exhibited the alteration of selenium distribution in organs that reflected activities of selenium-dependent enzyme.30, 31 The selenium levels in the brain and testis were completely reduced, and the activity of other selenoproteins was severely decreased in the selenoprotein P-knockout mice.30, 31 These reports suggested that selenoprotein P was an antioxidative enzyme in itself and protected tissues against oxidative stress. Furthermore, selenoprotein P activates the other selenoproteins though the delivery selenium to the organs. Selenoprotein P is an important molecule protecting tissues and organs against oxidative stress, cooperative with other selenoproteins.
The apoptosis-regulating genes of the bcl-2 family or multi-drug resistance genes were reported to have an important role in resistance of anti-cancer agents. Although these genes are thought to have a relation to gemcitabine resistance in our study, the result of our microarray clarified no alteration of the expression of these genes in the gemcitabine resistant KLM1-R. This was consistent with the result of no cross-resistance in other anticancer agents except bleomycin, known as an agent generating intracellular ROS in KLM1-R.
In our study, the exposure to gemcitabine induced approximately 3-fold intracellular free radicals that were inhibited by selenoprotein P. Intracellular free radicals occur in response to cytotoxic agents and this is related to the loss of mitochondrial inner membrane potential (MMP).26 The loss of MMP releases mitochondrial mediators of apoptosis such as cytochrome c and induces apoptosis.26 Intracellular free radicals have been noted as playing a contributing role in the cytotoxicity of anti-tumor agents including adriamycin,32 bleomycin19 and tumor necrosis factor.33 Sylvia et al.26 reported that gemcitabine-induced free radicals contributed to lethality through increasing the population with low MMP. Together with our results, the generation of free radicals is considered as one of the major cytotoxic mechanisms of gemcitabine. The 5′-flanking region of human selenoprotein P gene was shown to contain IFN-γ responsive elements and a negative regulation of selenoprotein P promoter using pro-inflammatory cytokines such as interleukin-1β and TNF-α was found.27 Expression of selenoprotein P mRNA and protein in human hepatoma cells HepG2 is efficiently inhibited on a transcriptional level by the anti-inflammatory cytokine transforming growth factor (TGF)-β 1.34 In our study, we examined the effect of IFN-γ to selenoprotein P, because IFN-γ is very common in the clinical environment. IFN-γ suppressed the expression of selenoprotein P mRNA and improved the cytotoxicity of gemcitabine in the human pancreatic cancer cell line. The intracellular ROS level was increased by the administration of IFN-γ. This combination therapy, gemcitabine and IFN-γ, may provide a new approach to the treatment of pancreatic cancer.
Our findings showed selenoprotein P as a novel molecule responsible for gemcitabine resistance in human pancreatic cancer cells, through the suppression of free radical cytotoxicity. In addition, we established a gemcitabine-resistant cell line that suppressed the cytotoxicity induced not only by gemcitabine-mediated endogenous free radicals but also by H2O2-mediated exogenous free radicals. Because free radicals can act directly on intracellular apoptosis signaling, further studies on this mechanism are required before clinical application.
We thank the Hokkaido Red Cross Blood Center for providing human plasma.