Involvement of ribonucleotide reductase M1 subunit overexpression in gemcitabine resistance of human pancreatic cancer



Pancreatic cancer is the most lethal of all solid tumors partially because of its chemoresistance. Although gemcitabine is widely used as a first selected agent for the treatment of this disease despite low response rate, molecular mechanisms of gemcitabine resistance in pancreatic cancer still remain obscure. The aim of this study is to elucidate the mechanisms of gemcitabine resistance. The 81-fold gemcitabine resistant variant MiaPaCa2-RG was selected from pancreatic cancer cell line MiaPaCa2. By microarray analysis between MiaPaCa2 and MiaPaCa2-RG, 43 genes (0.04%) were altered expression of more than 2-fold. The most upregulated gene in MiaPaCa2-RG was ribonucleotide reductase M1 subunit (RRM1) with 4.5-fold up-regulation. Transfection with RRM1-specific RNAi suppressed more than 90% of RRM1 mRNA and protein expression. After RRM1-specific RNAi transfection, gemcitabine chemoresistance of MiaPaCa2-RG was reduced to the same level of MiaPaCa2. The 18 recurrent pancreatic cancer patients treated by gemcitabine were divided into 2 groups by RRM1 levels. There was a significant association between gemcitabine response and RRM1 expression (p = 0.018). Patients with high RRM1 levels had poor survival after gemcitabine treatment than those with low RRM1 levels (p = 0.016). RRM1 should be a key molecule in gemcitabine resistance in human pancreatic cancer through both in vitro and clinical models. RRM1 may have the potential as predictor and modulator of gemcitabine treatment. © 2006 Wiley-Liss, Inc.

Pancreatic cancer remains one of the most malignant cancers. Although surgery is the only curative treatment currently available, over 80% of patients have advanced regional disease or distant metastasis at the time of diagnosis and less than 20% of the patients are candidates for resection.1 Therefore, chemotherapy, radiation or a combination of these therapies most commonly plays an important role in pancreatic cancer treatment. They have not had a significant impact on survival rates in recent decades, however, despite many clinical trials.1

Gemcitabine (2′,2′-difluorodeoxycytidine, dFdC, Gemzar) has been recognized as the standard first-line chemotherapeutic agent used in patients with pancreatic cancer, since it was shown to have some meaningful impact on either survival or disease-related symptoms when compared with 5-fluorouracil (5-FU) in randomized trials.2 However, not more than 25% patients with pancreatic cancer will benefit from gemcitabine, a proportion that is slightly less than in patients with other cancers.2 Although gemcitabine in combination with other various cytotoxic agents is being investigated, no randomized phase III trial has yet established any survival benefit for combination therapy when compared with gemcitabine alone.1, 3 The major cause of this relative treatment failure is thought to be tumor cell resistance to chemotherapy, whether it is inherent or acquired.4

A variety of attempts have recently been undertaken in vitro to detect the molecular markers of gemcitabine resistance. Alterations involved in cell cycle regulation, proliferation or apoptosis, such as mutated p53,5 Bcl-xl,6 c-Src,7 focal adhesion kinase8 and BNIP3,9 have been described in a variety of cancers including pancreatic cancer. Nucleotide transporters were also described as molecules related to the intracellular transport of extracellular gemcitabine from outside.10, 11 The ribonucleotide reductase M1 subunit (RRM1),12, 13 ribonucleotide reductase M2 subunit (RRM2),14, 15 deoxycytidine kinase (dCK)16 and cytidine deaminase (CDA)17 are supposed to play a role in gemcitabine resistance of the variety of cancer as metabolic enzymes of the drug. However, they remain still controversial because of the lack of direct evidence based on either in vitro gene transfer model systems or clinical data from patients with pancreatic cancer.

Variant cells with characteristics resistant to chemotherapeutic agents have widely contributed to the investigation of molecular mechanisms in chemoresistance.7, 12, 13, 16, 18, 19 These chemoresistant variants are traditionally established by continuous drug exposure and gradually increased drug concentration. Although drug resistance can occur at many levels, including increased drug efflux, drug inactivation, alterations in drug target, processing of drug-induced damage and evasion of apoptosis,20 the advent of recently established analytical technologies such as microarray and protein array systems has opened up feasible opportunities to identify molecules involved in drug resistance. Indeed, microarray analysis has become a key tool for characterizing gene expression in a variety of experimental systems with chemoresistant variants and has succeeded in identifying the molecules associated with gemcitabine resistance using an oligonucleotide microarray system in vitro and in vivo.12, 13, 21

In this study, we developed gemcitabine-resistant cells from the human pancreatic cancer cell lines and attempted to identify novel genes involved in gemcitabine chemoresistance using an oligonucleotide microarray system covering 30,000 human oligonucleotides. Furthermore, the detected candidate gene was also revealed to be responsible for gemcitabine resistance by an RNAi assay and by clinical analysis of the patients treated with gemcitabine.

Material and methods

Pancreatic cancer cell lines and selection of gemcitabine resistant cells

Five types of human pancreatic carcinoma cell lines were used in the present study. MiaPaCa-2 and PSN1 cell lines were obtained from the Japanese Collection of Research Bioresources (JCRB, Tokyo, Japan). The BxPC3 and Panc1 cell lines were obtained from the American Type Culture Collection (ATCC, Rockville, MD). The PCI6 cell line was a gift from Dr. H. Ishikawa (Hokkaido University, Sapporo, Japan). All cell lines were cultured at 37°C under 5% CO2 in DMEM (Sigma Chemical Co., St. Louis, MO) supplemented with 10% FBS (Hyclone Laboratories, Inc., Rockville, MD) and 100 units/ml each of penicillin and streptomycin. Relative gemcitabine-sensitive cell lines, BxPC3, PSN1 and MiaPaCa2 were used for the establishment of chemoresistant variants. Gemcitabine-resistant cells were generated by exposure to gradually increasing concentrations of the drug for 2 months as described previously.12, 22 The starting concentration was 1 ng/ml gemcitabine. When cells adapted to the drug, the gemcitabine concentration was increased. The final concentrations were 10 ng/ml gemcitabine for PSN1 and 20 ng/ml gemcitabine for BxPC3 and MiaPaCa-2.


Gemcitabine was kindly provided by Eli Lilly Pharmaceuticals (Indianapolis, IN). 5-Fluorouracil (5-FU) was purchased from Sigma Chemical Co. Gemcitabine and 5-FU were dissolved in distilled water and applied to cells at a concentration of less than 0.1% of the medium volume.

Cytotoxicity assay

Cell growth was assessed by the 3-(4-, 5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromide (MTT) (Sigma Chemical Co.) method.23 Briefly, 3 × 104 cells were seeded to a 96-well plate in 100 μl of medium and left overnight to adhere. Several concentrations of the test drugs in 100 μl volumes were added, and the cells were incubated for 48 hr. After treatment, 10 μl of MTT solution (5 mg/ml) was added to each well and incubated for another 4 hr at 37°C. Then, 100 μl of acid-isopropanol was added, and after 24 hr at 4°C, reduced MTT was measured spectrophotomechanically in a dual beam microtiter plate reader at 570 nm with a 650 nm reference. Resulting absorbencies were converted to percent survival by comparing treated with untreated (100% survival) cells. 50% inhibitory concentrations (IC50s) are defined as the concentrations of drug that result in 50% cell survival when compared with untreated cells.

Growth curve

Cells (1 × 104) were seeded to a 24-well plate in 1 ml of medium and left overnight to adhere. The medium was replaced daily with 1 ml of fresh medium with or without gemcitabine at the dose of parental IC50. Cell numbers were counted with an automatic cell counter (Celltec MEK-5103, Nihon Kohden, Tokyo, Japan) after being treated with trypsin.

Animals and in vivo antitumor experiments

Four-week-old female BALB/c nu/nu mice were purchased from Japan Clea (Tokyo, Japan) and maintained in specific pathogen-free conditions. Human pancreatic tumor xenografts were prepared by subcutaneous implantation (5 × 106 cells; total volume 100 μl) of MiaPaCa2 and MiaPaCa2-RG, resistant variant established from MiaPaCa2, into the right back of 10 nude mice each. The animals were monitored for activity, physical condition, determination of body weight and measurement of tumor volume [1/2 × (the major axis) × (the minor axis)2] every other day. When the tumors reached a volume between 100 and 200 mm3, mice were divided into the following 4 groups of 5 mice each: parental cell with no treatment, parental cell with weekly intraperitoneal injections of gemcitabine, resistant cell with no treatment, resistant cell with weekly intraperitoneal injections of gemcitabine. Gemcitabine was injected weekly into the peritoneal cavity at the dose of 240 mg/kg as described.24

[3H] gemcitabine cellular uptake assay

Cells were seeded to a flat-bottomed 24-well microplate (1 × 104/well) and incubated for 24 hr. The medium was replaced by 1 ml of fresh medium by an additional 48 hr of culture. The cells were then exposed to [3H] gemcitabine (Moravek Biochemicals, Inc. Brea, CA) at a concentration of 23.9 ng/ml (1.0 μCi/ml). After 1-hr exposure, the cells were washed 3 times in 1 ml of ice cold phosphate-buffered saline (PBS). The cells were then dissolved in 0.5 ml of 0.5% Triton X-100, and 0.4 ml aliquots were sampled for radioactivity counting. Aliquots of 20 μl were also sampled for protein determination. The uptake level of [3H] gemcitabine was expressed as radioactivity levels divided by protein concentrations measured by the Bradford method (Bio-Rad Laboratories, Hercules, CA).

Oligonucleotide microarray

RNA extraction was carried out with TRIzol reagent (Invitrogen, Carlsbad, CA) using a single-step method,25 and RNA quality was checked with an RNA 6000 Nano LabChip kit (Agilent Technologies, Waldbronn, Germany) according to the manufacturer's protocols. An oligo-microarray covering 30,000 human oligonucleotides (AceGene human 30K; DNA Chip Research Inc. and Hitachi Software Engineering Co., Ltd., Yokohama, Japan) was used in this study.26 Sample preparation, hybridization and wash were carried out according to the manufacturer's protocols ( A sample and the reference were labeled with Cy5-dUTP and Cy3-dUTP (Amersham Pharmacia Biotech, Piscataway, NJ), respectively, mixed, and hybridized on a microarray. The hybridized array was scanned using ScanArray 4000 (GSI Lumonics) at wavelengths corresponding to each probe's unique fluorescence (635 and 532 nm for Cy5 and Cy3, respectively). The signal intensity of each spot (16 bit tiff image) was converted into text format by DNASISArray software (Hitachi software Inc., Tokyo, Japan). Data processing was performed through background subtraction using the average blank spot intensity in each block. If the signal was higher than the background and the signal levels of Cy3 and Cy5 were higher than 1,000, these data were used for further analysis. At this stage, 10,517 genes remained. The Cy3/Cy5 ratio values of each spot were log-transformed and normalized so that the median Cy3/Cy5 ratio of whole genes was 1.0.27

Reverse transcription-polymerase chain reaction

RNA extraction was carried out with TRIzol reagent (Invitrogen, Carlsbad, CA), and cDNA was generated with avian myeloblastosis virus reverse transcriptase (Promega, Madison, WI), as described previously.25 In this assay, porphobilinogen deaminase (PBGD) mRNA was used as an internal control.28 PCR was performed in a 25-μl reaction mixture containing 2 μl of cDNA template, 1× Perkin-Elmer PCR buffer, 1.5 mM MgCl2, 0.8 mM deoxynucleotide triphosphates, 0.2 μM each primer and 1 U of Taq DNA polymerase (AmpliTaq Gold, Roche Molecular System, Inc.). The primers for PBGD were synthesized as described previously.28 The PCR primers used for the detection were as follows: dCK (forward primer, 5′-TGCAGGGAAGTCAACATT-3′; reverse primer, 5′-TCCCACCATTTTTCTGAG-3′), CTP synthetase (forward primer, 5′-CTCATATCACAGATGCAATC-3′; reverse primer, 5′-GATCATATCTGTCAGCCATCTC-3′), CDA (forward primer, 5′-GGAGGCCAAGAAGTCAG-3′; reverse primer, 5′-GACGGCCTTCTGGATAG-3′), DCTD (forward primer, 5′-GTGCAGTGATGACGTGTGTTGC-3′; reverse primer, 5′-CATGTAGATTCCATGTGAC-3′), RRM1 (forward primer, 5′-GAAGACTGGGATGTATTATTTAAG-3′; reverse primer, 5′-CAGAATAACCTATAGGAC-3′), RRM2 (forward primer, 5′-ATGAAAACTTGGTGGAGCGATT-3′; reverse primer, 5′-TGGCAATTTGGAAGCCATAGA-3′), p53R2 (forward primer, 5′-CCAGTTGGCCTCATTGGAAT-3′; reverse primer, 5′-TAGAGTTTTAAAACGAGAGG-3′), ENT1 (forward primer, 5′-GCTTGAAGGACCCGGGGAGC-3′; reverse primer, 5′-TGGAGAAGGCAAAGGCAGCCA-3′). PCR was performed with cycling conditions of 95°C for 10 min, followed by 35 cycles of denaturation at 95°C for 30 sec, annealing at 62°C (RRM1: 57°C) for 30 sec and extension at 72°C for 60 sec, and the products were run on 2% agarose gels and visualized by ethidium bromide staining. A quantitative gene expression assay was performed using LightCycler (Idaho Technology, Salt Lake City, UT), as described previously.29 PCR was performed with cycling conditions of 95°C for 10 min, followed by 40 cycles of denaturation at 95°C for 10 sec, annealing at 62°C (RRM1: 57°C) for 10 sec and extension at 72°C for 20 sec. Quantification data from each sample were analyzed using the LightCycler analysis software (Roche Diagnostics, Mannheim, Germany) as recommended by the manufacturer. Relative gene expression levels are expressed as quantified gene expression divided by quantified PBGD levels.

Western blotting

Cells grown to subconfluence in 90-mm dishes were lysed in protease inhibitor (1 mM PMSF, 40 μM leupeptin) containing PBS. After sonication, aliquots containing 50 μg of total protein were size-fractionated by SDS-PAGE (5–20% gradient gels), and the proteins were transferred to polyvinylidine difluoride membranes (Immobilon, Millipore, Bedford, MA) as described previously.25 The membranes were blocked with 5% skim milk and incubated for 1 hr at room temperature with mouse monoclonal anti-human RRM1 (Chemicon international, Inc., Temecula, CA) or rabbit polyclonal anti-human actin (Sigma). After 3 washings with 0.1% Tween 20 in TBS, the membranes were incubated for 30 min at room temperature with the horseradish peroxidase-conjugated secondary antibody. After further 5 washings peroxidase was detected with an enhanced chemiluminescence system from Amersham (Arlington Heights, IL).

RNAi treatment

RRM1 and control RNAis were purchased from Invitrogen (Stealth RNAi; Invitrogen, Carlsbad, CA). The RRM1-specific RNAi designed by BLOCK-iT RNAi Designer (Invitrogen, Carlsbad, CA) was as follows: Sense 5′-GGAUAUUGUUCUGGCCAAUAAAGAU-3′; Anti-sense 5′-AUCUUUAUUGGCCAGAACAAUAUCC-3′. A stealth RNAi negative control with medium GC duplex was used as a control. RNAis were dissolved in DEPC-treated water to make a 20 μM working stock. One day before transfection, 2 × 105 cells were plated into 35 mm, 6-well trays and allowed to adhere. Transfection was performed using Lipofectamine 2000 transfection regent (Invitrogen, Carlsbad, CA) following Invitrogen's protocols. The ability of the RNAi molecules to knock down RRM1 expression was analyzed by mRNA and protein detection, and the final dilution volume of RNAi was 50 pmol in 500 μl OptiMEM medium per well.

Patients and tissue samples

Eighteen recurrent pancreatic cancer patients in Osaka University Hospital were recruited. All patients had undertaken curative resection at Osaka University Hospital between September 1999 and February 2004 and were followed-up without any adjuvant treatment until recurrence. The tumor tissues had been collected and stored at −80°C until use under informed written consent. Each tumor was confirmed histopathologically to be advanced stage cancer. All patients had a measurable recurrent lesion and were treated with only gemcitabine after recurrence. Response to gemcitabine was defined as follows: complete response (CR), partial response (PR), stable disease (SD) and progressive disease (PD). This classification was based on New Guidelines to Evaluate the Response to Treatment in Solid Tumors (RECIST guidelines).30 Patients were divided into 2 groups based on the chemotherapeutic response. Responders were defined as CR, PR and SD. Nonresponders were PD. A total of 18 tumor samples resected at the primary curative operation were analyzed to determine RRM1 mRNA expression levels. Total RNA was isolated from the homogenate tumor samples using TRIzol method27 for quantitative reverse transcription-polymerase chain reaction (RT-PCR) using same conditions as already described. Total RNA of MiaPaCa2-RG was used for analytical curve.

Statistical analysis

Statistical analyses were performed using the SPSS 11.5J software (SPSS Inc., Chicago, IL). All data were expressed as mean ± SD. Differences between groups were examined for statistical significance using the Student's t test. In the clinical study, associations between the candidate molecule expression and gemcitabine response were assessed by Fisher's exact test. Overall, survival probabilities were estimated using the Kaplan-Meier method, and the log-rank test was used to determine the level of significance between the survival curves. A p value less than 0.05 denoted the presence of a statistically significant difference.


Establishment of gemcitabine-resistant pancreatic cancer cell lines

Three types of pancreatic cancer cell lines, BxPC3, PSN1 and MiaPaCa2, were cultured in the medium containing gemcitabine for 2 months. After selection, we established 3 variant cells resistant to gemcitabine with different degrees with the MTT assay (Table I). The selected cell lines were called BxPC3-RG, MiaPaCa2-RG or PSN1-RG, based on the names of their parental cell line. On the basis of the IC50 measurement, BxPC3-RG, MiaPaCa2-RG and PSN1-RG were 11-fold, 81-fold and 986-fold more resistant than parental cells to the cytotoxic effects of gemcitabine, respectively. BxPC3-RG and PSN1-RG were also cross-resistant to 5-FU, although MiaPaCa2-RG represented no significant cross-resistance (Table I). By the growth curve analysis, MiaPaCa2-RG and PSN1-RG showed significant resistant to gemcitabine, although BxPC3-RG did not show any resistance to gemcitabine. In the absence of gemcitabine, BxPC3-RG and MiaPaCa2-RG demonstrated almost the same growth curves when compared with parental cells, although PSN1-RG's growth rate was 10-fold slower than PSN1 (Fig. 1a). BxPC3-RG and MiaPaCa2-RG preserved the cell morphology of parental cells regardless of these chemoresistant alterations, and PSN1-RG showed significant difference in the cell morphology (Fig. 1b). MiaPaCa2-RG remained gemcitabine-resistant after 1 month culture in the medium without gemcitabine. Furthermore, MiaPaCa2-RG showed significant gemcitabine-resistance when compared with MiaPaCa2 in an in vivo xenograft model (Fig. 2). The level of [3H] gemcitabine cellular uptake in MiaPaCa2-RG (25.0 ± 3.2 pg GEM/μg protein) is half of that in MiaPaCa2 (49.9 ± 5.8 pg GEM/μg protein). These data suggest that MiaPaCa2-RG should be the most suitable for identifying genetic alterations relating to gemcitabine resistance among the 3 types of gemcitabine-selected variants. We chose MiaPaCa2-RG for further analysis to identify molecules associated with gemcitabine resistance.

Figure 1.

(a) Growth curves of pancreatic cancer cell lines. In the absence of gemcitabine (open circle, open triangle) and in continuous exposure to gemcitabine at the dose of parental IC50s (closed circle and closed triangle). Points, mean; bars, SD (n = 3). *p < 0.01. (b) Morphology of pancreatic cancer cell lines, BxPC3, MiaPaCa2, PSN1 and gemcitabine-resistant variants, BxPC3-RG, MiaPaCa2-RG, PSN1-RG. (Original magnification, ×40).

Figure 2.

In vivo gemcitabine sensitivity of MiaPaCa2 and MiaPaCa2-RG in subcutaneous xenograft model of nude mice. Gemcitabine was injected weekly into the peritoneal cavity at the dose of 240 mg/kg (closed circle and closed triangle) and control (open circle, open triangle). Points, mean; bars, SD (n = 5). *p < 0.01.

Table I. IC50 of Gemcitabine (GEM) and 5-Fluorouracil (5-FU) in Gemcitabine-Selected Pancreatic Cancer Cell Lines
Cell linesDrugIC50 (mean ± SD)Fold resistance
Parental cellResistant cell
BxPC3GEM (ng/ml)50.5 ± 7.1556.6 ± 7611-fold
5-FU (μg/ml)0.6 ± 0.237.5 ± 4.160-fold
MiaPaCa2GEM (ng/ml)44 ± 5.33592.1 ± 17081-fold
5-FU (μg/ml)2.4 ± 0.33.07 ± 0.141.3-fold
PSN1GEM (ng/ml)3.4 ± 0.33392 ± 44986-fold
5-FU (μg/ml)1.4 ± 0.145 ± 2.632-fold

Microarray analysis

To investigate the candidate genes involved in gemcitabine resistance, oligo-microarray experiments were carried out with MiaPaCa2 and MiaPaCa2-RG cells. Out of the 30,000 spotted genes, 10,517 genes were used for further analysis (See Material and methods). Scatter plotting showed that 99.6% genes (10,474 genes out of 10,517 genes) had altered expressions of less than 2-fold, and 43 genes were up- or downregulated more than 2-fold in MiaPaCa2-RG cells when compared with MiaPaCa2 cells (Fig. 3). Among the 43 genes in which 12 upregulated genes and 31 downregulated genes were identified, the RRM1 was the most upregulated with 4.5-fold (Table II). This gene is the subunit of ribonucleotide reductase (RR) considered as an enzyme associated with gemcitabine metabolism.4 This upregulation was validated by both quantitative RT-PCR and Western blotting (data not shown). Other subunits of RR, RRM2 and p53R2, and other enzymes involved in gemcitabine metabolism such as CDA, dCK, CTP synthetase and dCMP deaminase and nucleotide transporters were not chosen in the microarray analysis because of their low expression levels or failure to show any altered expression between MiaPaCa2 and MiaPaCa2-RG with quantitative RT-PCR. The functions of other up- and downregulated genes were not considered to have any association with gemcitabine sensitivity. Therefore, we focused on the RRM1 gene for further functional analysis.

Figure 3.

Representative scatterplots showing hybridizations of cDNA from MiaPaCa2 cells labeled Cy3 and MiaPaCa2-RG cells labeled Cy5. The lines show 2-fold difference expression in both channels. RRM1; arrow.

Table II. Genes up- and Downregulated by More than 2-Fold in MiaPaCa2-RG with MiaPaCa2 as a Reference
 FoldGene nameSymbolAccession no.
14.46Ribonucleotide reductase m1 polypeptideRRM1NM_001033
22.63ensembl genscan prediction AL050329
32.29kiaa0101 gene productKIAA0101NM_014736
42.27Hypothetical proteinATP5SNM_015684
52.20Inosine monophosphate dehydrogenase 1IMPDH1XM_004627
62.19Hypothetical protein flj20558FLJ20558NM_017880
72.09Suppression of tumorigenicity 7ST7NM_018412
82.07Hypothetical protein xp_040263LOC91732XM_040263
92.06Suppressor of g2 allele of skp1SUGT1NM_006704
102.02Unknown (protein for image:3456579)FUBP3BC001325
112.00ba196n14.4.1 (pro1085 protein, isoform 1) AL354776
122.00Hypothetical protein xp_039528LOC91613XM_039528
15.32ensembl genscan prediction AC063943
33.49ensembl genscan prediction AC068601
43.36Adenylate cyclase 6, isoform bADCY6NM_020983
53.31ensembl genscan prediction AC009294
63.01Activator of s phase kinaseASKNM_006716
72.78udp glycosyltransferase 2 family, polypeptide b4UGT2B4NM_021139
82.69ensembl genscan prediction AF131216
92.66ensembl genscan prediction AF277315
112.48Potassium voltage-gated channel, shal-related subfamily, member 3KCND3NM_004980
122.34ensembl genscan prediction AC005034
132.29ensembl genscan prediction AL356751
142.26ensembl genscan prediction AL135978
152.23ensembl genscan prediction AC021883
162.21Adaptor-related protein complex 2, mu 1 subunitAP2M1NM_004068
172.21Phosphoserine phosphatasePSPHNM_004577
182.18Transaldolase-related proteinTALDO1AF010400
192.16Hypothetical protein xp_016148LOC95556XM_016148
202.15Transcription elongation factor a (sii), 1TCEA1NM_006756
212.14Hepatitis a virus cellular receptor 1HAVCR1NM_012206
222.13Protein phosphatase 1, regulatory (inhibitor) subunit 2PPP1R2NM_006241
232.12Ring finger protein 22, isoform betaTRIM3NM_033278
242.11Glycine cleavage system protein hGCSHNM_004483
252.07Hypothetical protein nuf2rCDCA1BC008489
262.06Dj1093g12.6 (a novel protein)C20orf93AL121751
272.05Inosine-5′-monophosphate dehydrogenaseIMPDH2J04208
282.02Hypothetical protein xp_052919LOC112547XM_052919
292.02ensembl genscan prediction AL132801
302.01ensembl genscan prediction AC010553
312.00Heme-regulated initiation factor 2-alpha kinaseHRINM_014413

Chemosensitivity after RRM1-specific RNAi transfection

To verify that RRM1 should be involved in gemcitabine resistance, RNAi experiments were carried out on MiaPaCa2 and MiaPaCa2-RG. The ability of RRM1-specific RNAi to suppress RRM1 expression was confirmed by both RT-PCR (Figs. 4a and 4b) and Western blotting (Fig. 4c). After transfection with RRM1-specific RNAi, more than 90% suppression of RRM1 was observed (Fig. 4a). Other subunits of ribonucleotide reductase, RRM2 and p53R2, did not have any significant mRNA expression change. RRM1-specific RNAi transfection did not bring about any major effect on cell viability. After RRM1-specific RNAi transfection, the gemcitabine chemoresistance of MiaPaCa2-RG was significantly reduced to same level as that of MiaPaCa2, and gemcitabine response of MiaPaCa2 also became more sensitive (Fig. 5).

Figure 4.

(a) Quantitative RT-PCR, (b) RT-PCR and (c) Western blotting analyses of RRM1 expression after RRM1-specific RNAi transfection in MiaPaCa2 and MiaPaCa2-RG cells. Columns, mean; bars, SD (n = 3).

Figure 5.

Dose–response curves for gemcitabine in MiaPaCa2 (circle) and MiaPaCa2-RG (triangle) after RNAi transfection. Open circles and triangles indicate mock RNAi transfectant. Closed circles and triangles indicate RRM1-specific RNAi transfectant. Points, mean; bars, SD (n = 3).

RRM1 expression and gemcitabine response in human pancreatic cancer cells and patients with pancreatic cancer

To investigate that the increased expression also should be involved in intrinsic resistance to gemcitabine, we examined the association between RRM1 mRNA expression levels and gemcitabine sensitivity of 5 human pancreatic cancer cell lines at first. RRM1 mRNA expression levels are significantly associated with gemcitabine sensitivity in 5 pancreatic cancer cell lines (Fig. 6), although increased expression of RRM1 was not likely correlated with the increase of cellular resistance to gemcitabine between acquired gemcitabine resistant MiaPaCa2-RG cells and PSN1-RG cells. Next, we examined the correlation of RRM1 mRNA expression levels with clinical course of 18 patients with recurrent pancreatic cancer. Seven patients developed liver metastasis, 4 developed local recurrence, 3 developed lymph node metastasis, 2 developed lung metastasis and 2 developed multi site recurrence (1 patient had local recurrence and liver metastasis and the other had liver, lung, bone and lymph node metastasis). The response to gemcitabine were CR (n = 0), PR (n = 2), SD (n = 6) and PD (n = 10). We classified 8 patients as responders (PR and SD) and 10 patients as nonresponders (PD). On the other hand, the median RRM1 mRNA expression relative to the housekeeping gene PBGD was 1.3 × 10−2 (minimum expression, 0.0 × 10−2; maximum expression, 132.0 × 10−2) in 18 pancreatic tissue samples (Fig. 7). According to a cut-off value of 1.3 × 10−2, 9 patients (50%) were classified into the low RRM1 expression group, and 9 patients (50%) into the high RRM1 expression group. There was a significant association between gemcitabine response and RRM1 expression (p = 0.018) (Table III). Furthermore, patients with high RRM1 levels had poor survival times after gemcitabine treatment than those with low RRM1 levels (Fig. 8; p = 0.016). Median survival times after gemcitabine treatment was 6.0 months for patients with high RRM1 levels and 14.6 months for patients with low levels.

Figure 6.

Correlation of gemcitabine sensitivity with RRM1 expression in 5 human pancreatic cancer cell lines (n = 3). R = 0.99, p < 0.001.

Figure 7.

RRM1 expression levels classified by gemcitabine response in the human pancreatic cancer tissues. Dotted bar: RRM1 cut-off value of 1.3 × 10−2.

Figure 8.

Overall survival after gemcitabine treatment of 18 recurrent pancreatic cancer patients for RRM1 mRNA expression levels. Solid line: low RRM1 expression group (n = 9). Dotted line: high RRM1 expression group (n = 9). Log-rank test = 5.78, p =0.016.

Table III. Association between Gemcitabine Response and RRM1 mRNA Expression Levels
 RRM1 level1
  • Fisher's exact test, p = 0.018.

  • 1

    Cut-off value is median RRM1 expression relative to PBGD (1.3 × 10−2).

Responder (PR, SD)178
Nonresponder (PD)8210


The present study have demonstrated that RRM1, which is a subunit of ribonucleotide reductase (one of the key enzymes in gemcitabine metabolism), should be clearly involved in gemcitabine resistance in human pancreatic cancer. First, oligonucleotide microarray analysis covering 30,000 human oligonucleotides between human pancreatic cancer cells resistant to gemcitabine and parental cells demonstrated that the most upregulated gene in the gemcitabine-resistant variant MiaPaCa2-RG cells was the RRM1 gene. RRM1 expression in the resistant cells was 4.5-fold higher than parental cells. This up-regulation was validated by quantitative RT-PCR and Western blotting. Furthermore, there was no difference between the expression levels of the other subunits of ribonucleotide reductase or the other molecules in gemcitabine metabolism including dCK, CTP synthetase, dCMP deaminase and nucleotide transporters. Second, by RRM1-specific RNAi transfection, RRM1 expression in both mRNA and protein levels were significantly decreased and the gemcitabine chemoresistance of MiaPaCa2-RG was significantly reduced to same level as that of MiaPaCa2. Third, the most important point was confirmation by the clinical analysis. Increased RRM1 expression was significantly associated with antitumor effects and with poor survival after treatment with gemcitabine in pancreatic cancer patients (p = 0.018 and 0.016, respectively). Therefore, RRM1 could be the targeted molecule to regulate gemcitabine resistance. Furthermore, its expression levels could be a useful indicator of gemcitabine resistance.

Ribonucleotide reductase (RR) acts as the rate-limiting enzyme in de novo DNA synthesis, because it is the only known enzyme that converts ribonucleotides to deoxyribonucleotides, which step is mandatory for DNA polymerization and repair.31, 32 In the cell, a deoxycytidine analogue, gemcitabine, is phosphorylated to monophosphate, diphosphate, and triphosphate before incorporation into DNA, which is required for its growth inhibiting activity. The diphosphorylated form of gemcitabine acts as a RR inhibitor, and some of gemcitabine cytotoxic activity is due to this inhibition.33 Ribonucleotide reductase increases the deoxynucleoside triphosphate (dNTP) pool in the cells, which could lead to decreased incorporation of dNTP analogues such as triphosphorylated gemcitabine into DNA and might reduce the antitumor effect of gemcitabine.22 In fact, MiaPaCa2-RG, higher expresser of RRM1 mRNA, showed lower gemcitabine uptake than lower RRM1 expresser MiaPaCa2.

Recent results have shown that there are 3 human ribonucleotide reductase subunits: RRM1, RRM2 and p53R2. RRM1 is a large peptide chain (α), and RRM2 and p53R2 are small protein subunits of RR (β). The catalytically active form of eukaryotic ribonucleotide reductase is proposed to be a α2β2 heterotetrameter made up of 2 large subunits and 2 small subunits.34, 35 Although ribonucleotide reductase enzymatic activity is modulated by levels of RRM236 and p53R2,37 RRM1 could play a key role among the 3 subunits in the course of gemcitabine treatment. RRM1 controls substrate specificity and global on/off enzyme activity.36, 37 As suggested by Davidson et al., RRM1 could act as a “molecular sink” for gemcitabine, in which RRM1 binds irreversibly to the drug and inactivates it, while increased RRM1 expression did not alter ribonucleotide reductase activity in the gemcitabine resistant variant human lung cancer cells.12 RRM1 was upregulated in the 2 selected gemcitabine resistant human lung cancer cell lines, where RRM1 expression levels were correlated with gemcitabine concentration for cell selection.12 A recent microarray analysis has suggested that in vivo induction of resistance to gemcitabine should result in increased expression of RRM1.13 These data are consistent with the present findings, although they only suggested an association of gemcitabine resistance with higher RRM1 expression. More important was the clear demonstration in the present study of the direct association of RRM1 with gemcitabine resistance through RRM1-specific RNAi treatment. However, the precise mechanisms how the increased expression of RRM1 acts in gemcitabine resistance still remain obscure. In the in vitro study with acquired gemcitabine resistant MiaPaCa2-RG and PSN1-RG cells, increase expression of RRM1 was not likely correlated with the increase of cellular resistance to gemcitabine. Although gemcitabine-resistance of PSN1-RG cells was almost equal to that of MiaPaCa2-RG cells, RRM1 expression level was much higher in MiaPaCa2-RG than in PSN1-RG cells. Other molecules involving gemcitabine or 5-FU metabolism or molecules such as p838 may be participated in gemcitabine resistance. Further studies are needed to clarify these points.

Although association of the increased expression of RRM1 gene with gemcitabine resistance has been reported based on in vitro and in vivo acquired gemcitabine resistant tumor cells12, 13 as in the present study, the mechanisms under RRM1 upregulation of resistant cells have not been fully elucidated. Polymorphisms39 and amplified gene copy number40 in the RRM1 gene are supposed to be related to the gemcitabine chemoresistance of tumor cells. Gene mutation or epigenetic mechanism such as methylation may influence the expression of RRM1 in resistant cells. Our preliminary experiments, however, did not show any mutational or polymorphic changes in the RRM1 gene between parental and gemcitabine resistant selected cells. Demethylation agents such as 5-aza-2′-deoxycytidine did not change RRM1expression in gemcitabine resistant cells. Future studies for the regulation of RRM1expression could therefore be helpful to obtain modulation of gemcitabine sensitivity in pancreatic cancer cells.

Because the present findings on RRM1 as a factor in gemcitabine resistance are based on an in vitro acquired gemcitabine resistant model as shown in the previous studies,12 it is still unclear whether or not RRM1 should be one of the key molecules involved in the intrinsic resistance to gemcitabine. However, in in vitro analysis with human pancreatic cancer cell lines, RRM1 mRNA expression levels are significantly associated with gemcitabine sensitivity in 5 pancreatic cancer cell lines, while increased expression of RRM1 was not likely correlated with the increase of cellular resistance to gemcitabine between acquired gemcitabine resistant MiaPaCa2-RG cells and PSN1-RG cells. Furthermore, clinical data from patients treated with gemcitabine may indicate that RRM1 should play an important role in the intrinsic resistance to gemcitabine. Gemcitabine was more effective to recurrent tumors in those patients with low RRM1 mRNA expression in the tumor obtained at surgery, although expression levels of recurrent tumors were supposed to reflect those of primary tumors. Therefore, patients with low RRM1 mRNA expression might have a significantly longer survival than those with a high expression as previously reported in lung cancer patients41, 42 even though the survival of recurrent pancreatic cancer patients is generally poor. Although our data do not rule out that other molecules of gemcitabine resistance determine intrinsic or acquired sensitivity to gemcitabine in vivo as reported in the recent study, the clinical results should be the most feasible for further investigations.

In conclusion, we have demonstrated in the present study that RRM1 should be a key molecule in gemcitabine resistance in pancreatic cancer through both in vitro and clinical models. In the continuous struggle to overcome the chemoresistance of pancreatic cancer, RRM1 may have the potential to play the role of a predictor of gemcitabine resistance and modulator of gemcitabine treatment.