• Open Access

Disrupted plasma membrane localization of equilibrative nucleoside transporter 2 in the chemoresistance of human pancreatic cells to gemcitabine (dFdCyd)


3To whom correspondence should be addressed. E-mail: gshiota@med.tottori-u.ac.jp


Although the nucleoside pyrimidine analogue gemcitabine is the most effective single agent in the palliation of advanced pancreatic cancer, cellular resistance to gemcitabine treatment is a major problem in the clinical scene. To clarify the molecular mechanisms responsible for chemoresistance to gemcitabine, mRNA expression of the key enzymes including cytidine deaminase (CDA), deoxycytidine kinase (dCK), 5′-nucleotidase (NT5), equilibrative nucleoside transporter 1 and 2 (ENT1 and ENT2), dCMP deaminase (dCMPK), ribonucleotide reductase M1 and M2 (RRM1 and RRM2), thymidylate synthase (TS) and CTP synthase (CTPS) was examined. The interacellular uptake of gemcitabine was greatly impaired in the chemoresistant cell lines due to dysfunction of ENT1 and ENT2. Protein expression of ENT1 and ENT2 and their protein coding sequences were not altered. Immunohistochemical and western blot analyses revealed that localization of ENT2 on the plasma membrane was disrupted. These data suggest that the disrupted localization of ENT2 is one of causes of the impaired uptake of gemcitabine, resulting in a gain of chemoresistance to gemcitabine. (Cancer Sci 2011; 102: 622–629)

Deaths from pancreatic cancer rank fourth among cancer-related deaths in the United States and 5th in Japan.(1,2) An estimated 34 000 patients in the United States and 25 000 patients in Japan have died from the disease in one year. With a total 5-year survival rate reported to be lower than 5%, the disease is associated with an extremely poor prognosis.(3) Surgical resection with radical intent is still the only potentially curative treatment option that, together with the adjuvant chemotherapy treatment regimens recommended by recent prospective randomized clinical trials, has resulted in reported 5-year survival rates of 20%.(4) However, despite improved diagnostic tools, the often unspecified and vague symptoms compliance and delay of diagnosis render the majority of patients irresectable. In patients with unresectable pancreatic cancer, the previous mean survival time from diagnosis has been a maximum of 4 months.(1) Palliative treatment with gemcitabine has prolonged survival to approximately 6–7 months, and has also improved quality of life.(5,6) However, there are still no evidence-based treatment options for gemcitabine-refractory advanced pancreatic cancer, and effective additional agents to first-line gemcitabine treatment are yet to be found.(7)

Gemcitabine still represents the only established chemotherapeutic agent in the treatment of patients with pancreatic ductal adenocarcinoma. However, the problem is that a substantial number of patients develop gemcitabine chemoresistance.(7) Potential strategies to overcome chemoresistance are required. After intracellular uptake of gemcitabine, which depends on nucleoside transporters such as equilibrate nucleoside transporter 1 (ENT1) and equilibrate nucleoside transporter 2 (ENT2), gemcitabine is phosphorylated by deoxytidine kinase (dCK) and subsequently converted to difluorodeoxycytidine (dFdCTP), the major metabolite, which is associated with cytotoxicity, and is a substrate for both DNA and RNA polymerase. The diphosphate (dFdCDP) is a potent inhibitor of ribonucleotide reductase (RR) leading to a decrease in dCTP, the natural substrate for DNA polymerase. Gemcitabine can be deaminated to difluorodeoxyuridine (dFdU) by cytidine deaminase (CDA), exerting considerable cytotoxicity. The monophosphate of gemcitabine can also be deaminated by dCMP deaminase (dCMPK) to dFdUMP, an inhibitor of thymidylate synthase (TS). In summary, the mechanism of action of gemcitabine is characterized by a unique set of self-potentiating mechanisms: (i) inhibition of RR, leading to a decrease in dCTP, which (ii) will facilitate more incorporation of dFdCTP into DNA, and (iii) will lead to a decreased feedback inhibition of dCK; and (iv) dFdCTP can also inhibit dCMP deaminase; all of these mechanisms will lead to increased synthesis and retention of dFdCTP.(7,8)

In the present study, we carried out gene expression profiling and examined gemicitabine metabolism-related genes by developing gemcitabine-resistant cell lines. We identified a novel mechanism of gemcitabine resistance, in which intracellular uptake of gemcitabine was severely impaired, possibly due to the disruption of plasma membrane localization of ENT2. Our data revealed a possible resistant mechanism to gemcitabine cytotoxicity in pancreatic cancer.

Materials and Methods

Reagents.  Gemcitabine hydrochloride was kindly provided by Eli Lilly Japan K.K. (Tokyo, Japan). Hoechst33258 and S-(4-Nitrobenzyl)-6-thioinosine (NMBPR) were purchased from Sigma (St. Louis, MO, USA).

Establishment of gemcitabine-resistant cells.  The MIAPaCa-2 pancreatic cancer cell line (RCB2094) was provided by RIKEN BRC, Tsukuba, Japan. The cells were cultured in DMEM (Nissui Pharmaceutical, Tokyo, Japan) supplemented with 10% fetal bovine serum (MBL, Nagoya, Japan), l-glutamine and glucose in a humidified atmosphere of 5% CO2 at 37°C. The gemcitabine-resistant cell lines were established by exposing MIA PaCa-2 cells to stepwise increasing concentrations such as 20, 100, 500 and 1000 ng/mL of gemcitabine. Finally, five colonies grown in the presence of 1000 ng/mL were subcloned and expanded. The gemcitabine-resistant cell lines were named MIA-1000A, MIA-1000B, MIA-1000C, MIA-1000D and MIA-1000E, and were maintained in the presence of 1000 ng/mL of gemcitabine.

Cell proliferation assay.  The number of cells was determined by trypan blue exclusion or by TetraColor One (Seikagaku Co., Tokyo, Japan).(9) In the latter system, cell viability was calculated by measuring absorbance at 450 and 600 nm.(9) The IC50 of other chemotherapeutic agents such as 5-FU, bleomycin, mitomycinC, nedaplatin and doxorubicin in MIAPaCa-2, MIA1000-A, MIA1000-C and MIA1000-E cells was caluculated after incubation with chemotherapeutic agents for 72 h.

RNA extraction and real-time reverse transcription–polymerase chain reaction (real-time RT-PCR).  Total RNA extracted with TRIzol reagent (Invitrogen, Tokyo, Japan) was subjected to reverse-transcription using Superscript II (Invitrogen) and oligo(dT) primers. The mRNA expression level was determined by a LightCycler System (Roche Applied Science, Mannheim, Germany) using gene specific primers (Table S1). The genes included dCMPK, TS, RRM1, RRM2, CTPS, CDA, dCK, NT5, ENT1, ENT2, CNT1, CNT3 and β-actin. The pathways by which gemcitabine is metabolized are shown in Figure 1.

Figure 1.

 Metabolism and resistance mechanisms for gemcitabine. Gemcitabine (dFdC) is transported across the cell membrane by ENT1 and ENT2. After intracellular uptake, gemcitabine is phosphorylated by deoxytidine kinase (dCK) to monophosphate and subsequently by nucleotide kinases (dCMP deaminase [dCMPK], NDPK) to difluorodeoxycytidine (dFdCTP), the major cellular metabolite that is associated with cytotoxicity and is a substrate for both DNA and RNA polymerases. The dFdCTP is also an inhibitor of CTP synthase (CTPS). The monophosphate of gemcitabine is deaminated by dCMPK to dFdUMP, an inhibitor of TS. The dFdCDP is a potent inhibitor of RRM1 and RRM2, leading to a decrease in dCTP, the natural substrate for DNA polymerase. On the other hand, NT5 opposes the action of nucleoside kinase by catalizing the conversion of nucleotides back to nucleosides. Gemcitabine is inactivated mainly by dCDA-mediated conversion to dFdU. Deamination of dFdCMP to dFdUMP by dCMPK and subsequently to dFdU represents another inactivation pathway.

cDNA microarray.  Transcriptome analysis of total RNA extracted from parental and MIA1000-C cells was performed using the Illumina Human-V6 Expression Beadchip, Illumina Inc., San Diego, CA, USA. For clustering analysis, publicly available software programs “Cluster” and “Tree view”, both written by Michael B. Eisen (Stanford University, CA, USA), were used. Datasets were logarithmically transformed, using the Pearson correlation coefficient as the measure of similarity and an “average linkage” clustering algorithm was performed.

Immunofluorscent staining and western blot analysis.  Immunofluorescent staining was done as follows. For permeabilization, the cells were treated with 0.05% saponin for 15 min. Primary antibodies against ENT1 (ab48607; Abcam, Cambridge, UK), ENT2 (ab43822; Sigma) and actin (I-19, sc-1616; Santa Cruz Biotechnology, Santa Cruz, CA, USA) were used for immunofluorescent staining. The ENT2 antibody recognizes the extracellular domain of ENT2. The secondary antibodies used were Alexa Fluoro 594-conjugated goat anti-rabbit immunoglobulin G (Molecular Probes, Leiden, the Netherlands) for immunofluorescent staining, and horse radish peroxidase-conjugated anti-rabbit immunoglobulin G (GE Healthcare, Little Chalfont, UK) for western blotting. To detect ENT2 protein expressed on the cell surface, the cells were incubated with biotin using a cell surface protein isolation kit (Pierce Protein Reseach Product, Rockford, IL, USA). The proteins on the cell surface were applied for western blot with anti-ENT2 antibody.

Gemcitabine uptake assay.  MIAPaCa-2 and gemcitabine-resistant cell lines were seeded in a 24-well plate at a density of 5 × 105 cells/well. Following pre-incubation for 36 h at 37°C, cells were incubated with [3H]gemcitabine (Moravek Biochemicals, Inc., Brea, CA, USA) in the presence or absence of 50 nM NMBPR for 0, 2, 4 and 6 h. The amount of [3H]gemcitabine uptake was determined by liquid scintillation counting.

Sequencing analysis.  The sequencing reaction was performed using BigDye Terminator v3.1 and cycle sequencing kit (Life Technologies, Tokyo, Japan).

Statistical analysis.  Values are expressed as means ± SD. Means were compared using the Mann–Whitney U-test. A P value < 0.05 was considered to be significant.


Growth of the gemcitabine-resistant cells.  After 2 weeks of incubation of the MIA PaCa2 cells in the concentration of IC50 (20 ng/mL), the colonies were picked up and then incubated with an increasing concentration of gemcitabine for approximately 6 months. Finally, the concentration of gemcitabine was increased up to 1 μg/mL, and then five colonies were picked up. The five colonies were designated MIA1000-A, B, C, D and E, respectively. The cell numbers of these cells were assessed with increasing concentrations of gemcitabine from 0 to 4000 ng/mL (Fig. 2A). The MIA1000-A, B, C, D and E cells can grow as fast as MIA PaCa-2 cells up to 2 μg/mL. Although the cell numbers of gemcitabine-resistant cells decreased at 4 μg/mL, the IC50 of resistant cells increased 200-fold in comparison with that of the parent cells. The gemcitabine-resistant cells grew day by day, even in the presence of 1 μg/mL gemcitabine (Fig. 2B). However, the gemcitabine-resistant cells in gemcitabine-free conditions grew faster than those in 1 μg/mL of gemcitabine (Fig. 2C).

Figure 2.

 Growth analysis of MIAPaCa-2 cells and gemcitabine-resistant cells. (A) Cell viability of cells at several doses of gemcitabine. The dose varied from 0 to 4000 ng/mL. (B) Cell numbers at 1 μg/mL gemcitabine from day 0 to day 3. (C) Cell numbers without gemcitabine from 0 to day 3.

cDNA microarray.  To clarify the mechanisms for gemcitabine chemoresistance, cDNA microarray was performed (Table 1). The number of upregulated genes by over twofold was 1107, and that of downregulated genes by less than a half was 964. The category includes signal transduction, transcription, cell cycle, cell proliferation, apoptosis, cell–cell signaling, cell growth, protein modification, nucleotide metabolism, DNA repair and others. With respect to the genes of nucleotide metabolism, the fold increase in the expression level of CDA, AL5, CHL1 and OAS1 was 42.28, 2.54, 2.41 and 2.33, respectively (Data S1), and that of CTPS, AK2, BPNT1, dGK, OAS2, dCK, GDA and SLC28A2 was 0.47, 0.46, 0.43, 0.41, 0.40, 0.24, 0.20 and 0.005, respectively (Data S1). Because these 12 genes include CTP and dCK, important genes of gemcitabine metabolism, we focused on the gemcitabine metabolism-related genes as causes of gemcitabine chemoresistance.

Table 1.   Category of the genes analysed by cDNA microarray
CategoryNumber of upregulated genesNumber of downregulated genes
  1. Upregulated genes are the genes in which the expression level in MIA1000-C cells is over twofold higher than that in MIA PaCa-2 cells, and downregulated genes are the genes in which the expression level in MIA1000-C is less than half of MIA PaCa-2 cells.

Signal transduction8453
Cell cycle3230
Cell proliferation3018
Cell–cell signaling178
Cell growth134
Protein modification43
Nucleotide metabolism48
DNA repair48

The IC50 of 5-FU in MIA1000-C, MIA1000-C and MIA1000-E cells was greater than MIAPaCa-2 cells (P < 0.01, P < 0.05 and P < 0.01, respectively); however the differences in IC50 among the cells were not so large (Table 2). The IC50 of bleomycin, mitomycin C, nedaplatin and doxorubicin in gemcitabine-resistant cells were similar to their parent cells, respectively. These data suggest that chemoresistant mechanisms against gemcitabine do not include general mechanisms but gemecitabine-specific mechanisms.

Table 2.   IC50 of gemcitabine-resistant cell lines against other chemotherapeutic agents
Cell linesIC50 (μM)
  1. *P < 0.05; **P < 0.01, compared with MIAPaCa-2 cells.

MIA PaCa-21.2801.6701.6331.5211.573

Expression of gemcitabine metabolism-related genes.  Expression of CDA was greatly enhanced in MIA1000-C cells than MIAPaCa2, and was also increased to a lesser degree in MIA1000-B and MIA1000-E (Fig. 3A). Expression of dCK was downregulated in MIA1000-A, -B, -C, -D and -E cells, compared with the parent cells (Fig. 3B). NT5 was higher in MIA1000-C and MIA1000-E cells than parental cells in the absence of gemcitabine (Fig. 3C). ENT1 expression in MIA1000-B cells was lower than parental cells (Fig. 3D). However, expression of ENT2 was significantly reduced in five gemcitabine-resistant cells, both in the absence and presence of gemcitabine (Fig. 3E). The expression level of dCMPK in MIA1000-A, -B, -C, -D and -E cells was similar to that of MIAPaCa2 in the absence of gemcitabine. The addition of 1 μM gemcitabine increased gemcitabine in parental cells (Fig. 3F). The same tendency was observed in RRM1, RRM2, TS and CTPS (Fig. 3G–J). These data suggest that the expression pattern of gemcitabine metabolism-related genes is almost similar among gemcitabine-resistant cells except CDA expression in MIA1000-C cells.

Figure 3.

 Real-time RT-PCR analysis of gemcitabine metabolism-associated genes in parental cells and gemcitabine-resistant cells. (A) Deoxycitidine deaminase (dCDA), (B) deoxycytidine kinase (dCK), (C) 5′-nucleotidase (NT5), (D) equilibrative nucleoside transporter 1 (ENT1), (E) equilibrative nucleoside transporter 2 (ENT2), (F) dCMP deaminase (dCMPK), (G) ribonucleotide reductase M1 (RRM1), (H) ribonucleotide reductase M2 (RRM2), (I) thymidylate synthase (TS) and (J) CTP synthase (CTPS). + and − represent the presence and absence of gemcitabine, respectively. ++< 0.01; +< 0.05, compared with the absence of gemcitabine of MIAPaCa2. **< 0.01, *< 0.05, compared with MIAPaCa2.

The expression level of ENT1 protein among MIAPaCa2, MIA1000-A, MIA1000-B, MIA1000-C and MIA1000-D was similar; however, expression in MIA1000-E cells seemed to be lower (Fig. 4). Expression of ENT2 protein was also similar among MIAPaCa2 and MIA1000-A, and MIA1000-B and MIA1000-C, whereas it was fainter in MIA1000-D and MIA1000-E. On the other hand, expression of CNT1 and CNT3 was not observed when the cells were treated or they were not (Fig. 4).

Figure 4.

 Protein expression of nucleoside transporters. (A) Expression of ENT1 and ENT2 in MIAPaCa2 (M), MIA1000-A (A), MIA1000-B (B), MIA1000-C (C), MIA1000-D (D) and MIA1000-E (E). (B) Expression of CNT1 and CNT3. Positive controls include human liver tissues.

Activity of ENT1 and ENT2 as uptake ability of gemcitabine was examined in all the cells (Fig. 5). In MIAPaCa2 cells, the radioactivity mediated by ENT1 and ENT2 was increased with the incubation time, and the activity of ENT2 was higher than that of ENT1. However, surprisingly, no activities of uptake of gemcitabine were observed in gemcitabine-resistant cells. These data suggest that some differences in the expression level of ENT1 or ENT2 are not the major causes of gemcitabine resistance in these cells.

Figure 5.

 Uptake of gemcitabine by parental cells and gemcitabine-resistant cells. MIAPaCa-2 and gemcitabine-resistant cell lines were seeded in a 24-well plate at a density of 5 × 105 cells/well. Following pre-incubation for 36 h at 37°C, cells were incubated with [3H]-gemcitabine (Moravek Biochemicals, Inc., Brea, CA, USA) in the presence or absence of 50 nM S-(4-Nitrobenzyl)-6-thioinosine (NMBPR) for 0, 2, 4 and 6 h. Human equilibrative NA+-independent nucleotide transport is mediated by membrane proteins sensitive (system es, ENT1) or insensitive (system ei, ENT2) to NBMPR.(17) The amount of [3H]-gemcitabine uptake was determined by liquid scintillation counting.

The coding sequences of ENT1 and ENT2 were examined; however, no substitutions of nucleotides were present in the protein coding sequences of ENT1 and ENT2 (data not shown).

In order to clarify why the uptake of gemcitabine was so impaired in the gemcitabine-resistant cells, we examined whether ENT1 or ENT2 was localized to the plasma membrane (Fig. 6). Expression pattern of ENT1 in MIA1000-A cells was almost the same with that in MIAPaCa-2 cells in both treatment with or without permealization, which are almost the same patterns with MIAPaCa-2 (Fig. 6A–D). On the other hand, ENT2 in MIAPaCA-2 cells was expressed on the plasma membrane with and without permealization (Fig. 6E,G). However, expression of ENT2 in MIA1000-A cells was not observed after permealization (Fig. 6F), while it was expressed without peramealization (Fig. 6H). However, expression of ENT2 in MIA1000-A cells was observed after permealization (Fig. 6H). These data suggest that ENT2 was not localized to the plasma membrane in MIA1000-A cells. Furthermore, ENT2 protein was strongly expressed on the cell surface of MIAPaCa2 cells, but very faintly in MIA1000-A cells using the anti-ENT antibody, which recognizes the extracellular domain of ENT2 (Fig. 7).

Figure 6.

 Localization of ENT1 and ENT2 proteins in parental cells and gemcitabine-resistant cells. (A,E) MIAPaCa-2 without permealization, (B,F) MIA1000-A without permealization, (C,G) MIAPaCa-2 permealized, and (D,H) MIA1000-A permealized. (A–D) Include ENT1, and (E–H) include ENT2.

Figure 7.

 Western blot analysis of ENT2 protein on the cell surface. The cells were incubated with biotin, and then the cells labeled with biotin were separated with avidin beads. The cell surface proteins were applied to western blot with anti-ENT2 antibody, which recognizes the extracellular domain of ENT2. A, MIA1000-A cells; M, MIAOaCA2 cells.


The present study suggests that gemcitabine resistance due to impaired uptake of gemcitabine results from disrupted localization of ENT2, but not from the expression level in pancreatic cancer cells, because the expression levels of ENT2 in gemcitabine-resistant cells were lower than in parental cells by approximately 50%; however, the uptake of gemcitabine was almost completely lost. These data were supported by the report that decreased nucleotide transporter activity was closely associated with increased resistance to gemcitabine.(10) In recent papers, it has been reported that the ENT1 protein expression level was associated with increased overall survival and disease-free survival in pancreatic cancer patients who received gemcitabine, but not in those who received 5-FU.(11,12) Taken together, these data stress the importance of nucleotide transporter activity in gemcitabine-resistant cells.

In the chemoresistant cells against gemcitabine, the activities of dCK and ribonucleotide reductase increased, leading to a decrease in the pool of gemcitabine and its active metabolites.(13) dCK, essential for the phosphorylation of gemcitabine, was correlated with gemcitabine sensitivity in human pancreatic tumors treated with gemcitabine, but expression of CDA, which catalyzes the degradation of gemcitabine, varied.(14) In addition, gemcitabine resistance is associated with RRM1,(15) and survival of ovarian cancer patients treated with gemcitabine was related to increased RRM2.(16) In the present study, expression levels of dCK, RRM1 and RRM2 in gemcitabine-resistant cells were decreased in comparison with those in parental cells. Taken together, these data suggest that the causes of chemoresistance against gemcitabine vary between cells and types of cancer.

The cellular uptake of nucleotides, including adenosine, is mediated primarily by a family of ENT.(17,18) ENT1 is widely expressed and can transport both purine and pyrimidine nucleotides. ENT2 is less widely distributed, but has been shown to be highly expressed in human skeletal muscle.(17) ENT2 has a similar nucleoside substrate profile to ENT1, but it also transports nucleobases such as adenine and hypoxanthine. It is generally assumed that ENT2 can transport substrates in both directions, into and out of cells. Cellular localization and functional characterization of equilibrative nucleotide transporters in the chemosensitivity of cells has been reported.(19) When the TMK-1 cells were transfected with hENT1 and hENT2, chemosensitivity to antitumor nucleosides increased.(19) However, when the cells were transfected with hENT3 and hENT4, it did not increase because hENT1 and hENT2 were localized to the plasma membrane; however, hENT3 and hENT4 were not localized on the plasma membrane.(19) The importance of a RXXV motif for hENT1 and a dileucine repeat for hENT2 in the COOH-terminal as targeting motifs implicated the presence of hENT1 and hENT2 on the basolateral membrane.(20) In addition, a 156-residue COOH-terminal truncation variant had a plasma membrane distribution similar to wild-type hENT2, but was retained intracellularly. In another report, disruption of a highly conserved triplet (PWN) near the N-terminus, or the last eight C-terminal residues (two hydrophobic triplets separated by a positive arginine) result in loss of plasma membrane localization and/or transport function.(21) In the present study, no nucleotide sequences of ENT1 and ENT2 were changed. However, ENT2 was not expressed on the plasma membrane. These data suggest that severely impaired activity of gemcitabine uptake by ENT2 is due to the disrupted localization of ENT2.

For plasma membrane proteins to localize to the plasma membrane, they must undergo correct processing, targeting and trafficking. These processes are aided by many different interacting partners that regulate folding, endoplasmic reticulum (ER) export and/or incorporation into vesicles.(22) Correct folding of membrane proteins is aided by chaperone proteins, which associate with hydrophobic residues, and in an ATP-dependent manner, guides the alignment of peptides that will later be linked by disulfide bridges formed by protein disulfide isomerase.(21) Besides folding, interacting proteins also play a role in allowing a protein to travel along the secretory pathway to its final destination. These interacting partners allow a protein to be incorporated into vesicles for transport and later incorporation into their target membranes.(23,24) In the present study, although the precise mechanisms for disrupted localization of ENT2 are unknown, the interacting proteins with ENT2 will be changed. In the future, we will focus on these proteins to further investigate the molecular mechanism of the disrupted localization of ENT2.

Disclosure Statement

The authors have no conflict of interest to declare.