• Open Access

Cellular localization and functional characterization of the equilibrative nucleoside transporters of antitumor nucleosides


To whom correspondence should be addressed. E-mail: yendo@kenroku.kanazawa-u.ac.jp


Nucleoside transporters play an important role in the disposition of nucleosides and their analogs. To elucidate the relationship between chemosensitivity to antitumor nucleosides and the functional expression of equilibrative nucleoside transporters (ENT), we established stable cell lines of human fibrosarcoma HT-1080 and gastric carcinoma TMK-1 that constitutively overexpressed green fluorescent protein-tagged hENT1, hENT2, hENT3 and hENT4. Both hENT1 and hENT2 were predictably localized to the plasma membrane, whereas hENT3 and hENT4 were localized to the intracellular organelles. The chemosensitivity of TMK-1 cells expressing hENT1 and hENT2 to cytarabine and 1-(3-C-ethynyl-β-d-ribopentofuranosyl) cytosine increased markedly in comparison to that of mock cells. However, no remarkable changes in sensitivity to antitumor nucleosides were observed in cell lines that expressed both hENT3 and hENT4. These data suggest that hENT3 and hENT4, which are mainly located in the intracellular organelles, are not prominent nucleoside transporters like hENT1 and hENT2, which are responsible for antitumor nucleoside uptake. (Cancer Sci 2007; 98: 1633–1637)

Abbreviations: Ara-C



cyan fluorescent protein


concentrative nucleoside transporter

dCyd kinase

deoxycytidine kinase




1-(3-C-ethynyl-β-d-ribo-pentofuranosyl) cytosine


equilibrative nucleoside transporter


green fluorescent protein


Madin–Darby canine kidney


3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide




nucleoside transporter


phosphate-buffered saline


uridine–cytidine kinase-2


yellow fluorescent protein.

Nucleoside analogs are administered therapeutically to patients with malignant tumors and viral infections. These drugs, including Ara-C, gemcitabine and fludarabine, are phosphorylated metabolically by dCyd kinase and are thus finally converted to active triphosphate forms.(1) These triphosphates mainly inhibit DNA polymerases. Therefore, the antitumor action of nucleoside analogs is closely related to the intracellular concentration of their phosphates; this concentration is controlled by cellular phosphorylation enzymes such as dCyd kinase. In particular, dCyd kinase is frequently targeted for the acquisition of resistance in tumor cells.(2–7) Another antitumor nucleoside, ECyd, is a potent inhibitor of RNA polymerases.(8–13) ECyd requires the activity of UCK2 to produce the corresponding active metabolite (ECyd triphosphate).(14) UCK2 is an exclusive target for the acquisition of resistance to the 3′-ethynyl nucleosides ECyd and 1-(3-C-ethynyl-β-d-ribo-pentofuranosyl) uracil.(15)

However, to exhibit antitumor activity, the antitumor nucleosides need to be either transported into the cytosol or diffused into the cells. Antitumor nucleosides are well known to be good substrates of ENT, and the expression of these transporters has been shown to influence the sensitivity to antitumor nucleosides.(16–22) Most antitumor nucleosides are hydrophilic and thus require facilitated transport into the cell via specific NT. In mammalian cells, there are two major gene families of NT: the CNT (SLC28) and the ENT (SLC29).(20,21,23–25) The former are able to cotransport many nucleosides with Na+ to the inwardly directed Na+ gradient. Based on functional studies using cells, five Na+-dependent NT subtypes (N1–N5) that have different substrate specificities were reported.(26) In the past decade, three human CNT have been cloned; they are predicted to posses a common 13-transmembrane structure and are also mainly localized to the apical membrane in polarized cells.(25,27) Four human ENT (hENT1–hENT4) composed of an 11-transmembrane structure have been cloned. Among them, hENT1 and hENT2 have been known for a long time, and many studies on their function and organ distribution have been carried out. The hENT1 (es-type) transporter but not hENT2 (ei-type) is NBMPR-sensitive (nanomolar range). However, nucleoside transportation via both hENT is inhibited by DPM. hENT are thus considered to be a major and important pathway responsible for chemosensitivity to antitumor nucleosides as these hENT are expressed in many tumors and tissues. Furthermore, hENT1 and hENT2 were localized to the basolateral membrane when expressed as fusion proteins with CFP or YFP in MDCK cells.(27,28) Although remarkable progress has been achieved recently in both genome research and analytic database technology, the biological function of each transporter gene remains to be elucidated. hENT3 and hENT4 were identified by an expressed sequence tag and genome database search. hENT3 has been shown to be a lysosomal transporter in the heart and functions as a pH-dependent NT.(16,29) Recently, hENT4 was reported to function as a plasma membrane monoamine transporter that can transport organic cations and serotonin in the brain and heart. Although hENT3 was predominantly located in intracellular organelles, forced expression of hENT4 was observed in the plasma membrane. However, endogenous hENT4 was distributed in a punctate pattern in intracellular organelles and partially in the plasma membrane. It also exhibited pH-dependent (at acidic pH) adenosine transport activity similar to hENT3.(20,30,31)

It is apparent that the transport activity of hENT1 and hENT2 is one of the factors that determines chemosensitivity to antitumor nucleosides. It is still unclear, however, whether hENT3 and hENT4 affect chemosensitvity to antitumor nucleosides. We therefore aimed to clarify the types of NT that are actually related to chemosensitivity to such antitumor nucleosides. In the present paper, we aimed to elucidate whether hENT3 and hENT4 affect the chemosensitivity of tumor cells to Ara-C and ECyd and their site of localization in the cell.

Materials and Methods

Cell culture and materials.  ECyd was synthesized as described previously.(8,32) Ara-C was purchased from Sigma (St Louis, MO, USA). Human fibrosarcoma tumor HT-1080 cells, human colon tumor HT-29 cells and human gastric tumor TMK-1 cells were obtained from the American Type Culture Collection (Rockville, MD, USA) and the Japanese Cancer Research Resources Bank (Tokyo, Japan). These cells were cultured in RPMI-1640 medium (Nissui Pharmaceutical, Osaka, Japan) containing 10% fetal bovine serum (Gibco, Grand Island, NY, USA) and 50 µg/mL kanamycin at 37°C in a humidified atmosphere.

Gene construction and stable expression of NT.  The full coding sequences of hENT1 (HSU81375), hENT2 (AF034102), hENT3 (NM018344) and hENT4 (BK000627) were amplified from human tumor cell lines using the following primer sets: hENT1, sense primer (ATGACAACCAGTCACCAGCCTCAGGACAGA) and antisense primer (GTCACACAATTGCCCGGAACAGGAAGGAGA); hENT2, sense primer (ATGGCGCGAGGAGACGCCCCGCGGGACAGC) and antisense primer (GCCACTTCAGAGCAGCGCCTTGAAGAGGAA); hENT3, sense primer (ATGGCCGTTGTCTCAGAGGACGACTTTCAG) and antisense primer (GTCCTCCCTTCTAGATGAGGTGCACCAGGA); hENT4, sense primer (ATGGGCTCCGTGGGGAGCCAGCGC) and antisense primer (GCTCAGAGGCCTGCGAGGATGGAACCATTG). The amplified fragments were subcloned into the mammalian GFP expression vector pcDNA3.1/NT-GFP-TOPO (Invitrogen, Carlsbad, CA, USA). The resultant constructs GFP-hENT1, GFP-hENT2, GFP-hENT3 and GFP-hENT4 with the GFP moiety fused to the N-terminus of each hENT were then used to produce stable cell lines. All of the constructs were confirmed by cycle sequencing using the Gene Rapid sequencer (Amersham Biosciences, Piscataway, NJ, USA). To generate stable hENT1, hENT2, hENT3 and hENT4 transfectants, 2 × 105 cells/well were seeded in a six-well plate 1 day before transfection. The expression vectors were transfected into cells using FuGENE 6 transfection reagent (Roche Diagnostics Corporation, Indianapolis, IN, USA), according to the manufacturer's instructions. Stable cell lines expressing each GFP-hENT fusion protein were then selected in culture medium containing 1 mg/mL G418 (Invitrogen).

Cellular localization using confocal microscopy.  The stable cell lines were seeded on a cover glass and cultured for 24 h. The GFP-hENT fusion proteins were visualized using an LSM 710 confocal microscope (Carl Zeiss, Jena, Germany). Alternatively, GFP-hENT fusion proteins were detected by an immunofluorescence assay using an anti-GFP rabbit polyclonal antibody (Medical & Biological Laboratories Co., Nagoya, Japan) and a fluoroscein-5-isothiocyanate (FITC)-conjugated sheep antibody to rabbit IgG (Silenus, Melbourne, Victoria, Australia) as the secondary antibody. Mitochondria and lysosomes were stained by incubation with Mitotracker Red CMXRos or LysoTracker Red DND-99 (Molecular Probes, Eugene, OR, USA), according to the manufacturer's instructions. Golgi complexes, early endosomes and late endosomes were identified immunologically by specific antibodies, mouse antihuman Golgi-zone monoclonal antibody (MAB1271; Chemicon International, Temecula, CA, USA), early endosome antigen 1 (BD Biosciences, San Jose, CA, USA) and mouse antihuman CD63 monoclonal antibody (Chemicon International). The cells cultured on a cover glass were fixed for 15 min in 4% paraformaldehyde in PBS, pretreated with 0.1% Triton X-100 for 15 min, and blocked using 1% bovine serum albumin for 30 min. Next, the cells were incubated for 1 h at room temperature with organelle-specific antibodies. The cells were washed with PBS and reacted with Alexa Fluor 546 goat antimouse IgG (Molecular Probes) at room temperature for 30 min.

Cellular uptake examination.  The intracellular nucleoside accumulation was measured using [3H]-Ara-C (Amersham Biosciences) as a typical antitumor nucleoside transported via NT. Briefly, 1 × 105 cells were preincubated in a four-well plate at 37°C for 24 h. The medium was replaced by a culture medium containing 37 kBq of 400 nM [3H]-Ara-C with or without NT inhibitors (20 µM NBMPR or 2 µM DPM). After incubation for 15 min, the plate was placed on ice and each well was immediately washed three times with ice-cold PBS. The radioactivity was measured by lysing the cells, adding a scintillation cocktail (ACS II; Amersham Biosciences) and counting in an Aloka scintillation counter (Tokyo, Japan).

Chemosensitivity. In vitro cytotoxicity was examined using a modified tetrazolium-based semiautomatic colorimetric assay with MTT reagent (Sigma), as reported previously.(33,34) Briefly, exponentially growing cells (2 × 103 cells) were incubated with various concentrations of antitumor nucleoside in a 96-well microplate for 72 h. MTT reagent was then added to each well to form MTT formazan. The formazan was dissolved in dimethyl sulfoxide and the absorbance at 540 nm was measured using a microplate reader. More than three clones of each stably transfected cell line were used in this experiment. The percentage of cell growth inhibition was calculated by the following formula:

% of cell growth inhibition = T/C × 100,

where C is the mean A540 of the control group and T is that of the treated group. The IC50 value was measured graphically from the dose–response curve with at least three drug concentration points. The average and standard deviation were estimated by more than three independent experiments.


Expression of hENT mRNA.  The human colon tumor HT-29 cells expressed the mRNA of all hENT and hCNT. The HT-1080 fibrosarcoma cells showed a higher expression level of hENT1 mRNA and a lower expression level of hENT3 mRNA than the HT-29 cells; however, the HT-1080 cells did not express hENT4 or any hCNT at the mRNA level. Therefore, it appeared that the HT-1080 cells incorporated antitumor nucleosides such as Ara-C and ECyd predominantly via ENT1 (Fig. 1). However, the TMK-1 human gastric tumor cells had only a very low expression level of hENT1 and hENT3 (Fig. 1).

Figure 1.

The mRNA expression of nucleoside transporters was determined by reverse transcription–polymerase chain reaction analysis. Total RNA from TMK-1, HT-1080 and HT-29 cells was used to analyze the mRNA expression of nucleoside transporters.

Localization of hENT.  At first, we established cell lines stably expressing the GFP-hENT fusion proteins HT-1080/hENT1, HT-1080/hENT2, HT-1080/hENT3, and HT-1080/hENT4. No remarkable morphological changes were observed in any of the stably expressed cell lines. To examine the cellular localization of GFP-hENT in HT-1080, the GFP fusion proteins were visualized using a confocal laser scanning microscope. As expected, the GFP-hENT1 and GFP-hENT2 proteins were localized in the plasma membrane of HT-1080 cells (Fig. 2). Although the GFP-hENT3 protein was expressed weakly in the cells, it appeared as speckles. It was also localized in the intracellular organelles but not in the plasma membrane. Similarly, the GFP-hENT4 protein was localized mainly in the intracellular organelles and it was expressed more strongly than GFP-hENT3; however, both GFP-hENT3 and GFP-hENT4 appeared to have different localization profiles in the cellular organelles. A small fraction of GFP-hENT4 was detected in the plasma membrane. To assess the localization of GFP-hENT3 and GFP-hENT4 in detail, we used organelle-specific dyes or antibodies. We found that the intracellular localization of hENT3-GFP overlapped with the early endosomes and the lysosomes but not with the mitochondria and Golgi complex (Fig. 2). The intracellular localization of GFP-hENT4 did not overlap with the early endosomes, mitochondria or Golgi complex; GFP-hENT4 was partly colocalized with the lysosomes (Fig. 2). In other cell lines (TMK-1/hENT1, TMK-1/hENT2, TMK-1/hENT3 and TMK-1/hENT4), the localization of each GFP-hENT was the same as that in the HT-1080 cells (data not shown).

Figure 2.

The localization of human equilibrative nucleoside transporters (hENT) (green fluorescent protein [GFP]-hENT1–4) in HT-1080 cells that stably overexpressed the GFP-hENT fusion protein (green). (a) Lysosomes (red), (b) mitochondria (red) and (f) nuclei (red) were stained by organelle-selective dyes. (c) Golgi complexes (red), (d) early endosomes (red) and (e) late endosomes (red) were identified immunologically by specific antibodies.

Effect of hENT localization on chemosensitivity.  hENT3 and hENT4 were localized in intracellular organelles; we therefore hypothesized that they may not affect chemosensitivity to antitumor nucleosides. To test this hypothesis, we investigated the chemosensitivity of cell lines that expressed GFP-hENT to the antitumor nucleosides Ara-C and ECyd. However, the chemosensitivity of all stable clones of HT-1080 cells to Ara-C remained unchanged (Table 1). In HT-1080/hENT3 and HT-1080/hENT4, the chemosensitivity to ECyd shifted slightly toward insensitivity. In contrast, the IC50 values in TMK-1/hENT1 and TMK-1/hENT2 cells decreased; in particular, the chemosensitivity to ECyd in TMK-1/hENT2 was enhanced 3.9 times that observed in the TMK-1 mock cells (Table 1). The changes in chemosensitivity in the TMK-1/hENT3 and TMK-1/hENT4 cells to Ara-C and ECyd tended to be fewer than those in the TMK-1/hENT1 and TMK-1/hENT2 cells (Table 1).

Table 1. Growth inhibitory effect of cytarabine (Ara-C) and 1-(3-C-ethynyl-β-d-ribo-pentofuranosyl) cytosine (ECyd) on stable transfectant cells that overexpressed the green fluorescent protein–equilibrative nucleoside transporter fusion protein
CellIC50 (µM)
  1. IC50 values are expressed as the mean ± SD. The statistical analysis against mock control was carried out using Student's t-test. *P < 0.05, **P < 0.01.

HT-1080/mock0.14 ± 0.040.041 ± 0.032
HT-1080/hENT10.10 ± 0.050.053 ± 0.045
HT-1080/hENT20.13 ± 0.040.077 ± 0.013
HT-1080/hENT30.12 ± 0.040.085 ± 0.072
HT-1080/hENT40.11 ± 0.100.091 ± 0.075
TMK-1/mock0.20 ± 0.070.70 ± 0.06
TMK-1/hENT10.14 ± 0.050.25 ± 0.02**
TMK-1/hENT20.10 ± 0.03*0.18 ± 0.07*
TMK-1/hENT30.18 ± 0.080.53 ± 0.08
TMK-1/hENT40.20 ± 0.070.54 ± 0.09

Transport function of hENT.  We examined the cellular uptake of [3H]-Ara-C with or without the NT inhibitors NBMPR and DPM. NBMPR in nanomolar concentrations has been used as a specific inhibitor for the ENT1 (es-type) transporter. Furthermore, DPM is inhibited by both es- and ei-type ENT (ENT1 and ENT2). Thus, the analysis of transporter function using these inhibitors is a useful experiment for the distinction of transporter types. The HT-1080/hENT1 cells showed almost the same transport activity as the HT-1080/mock cell lines, even in the presence of NT inhibitors (Fig. 3). In the HT-1080/hENT2 cells, NBMPR was not able to inhibit the transport activity, and the function of NBMPR-insensitive ENT2 (ei-type NT) was markedly facilitated instead of that of ENT1 (es-type NT) in the HT-1080/mock cell line. The sensitivity of the HT-1080/hENT2 cells to DPM decreased considerably, whereas the activity of the DPM-insensitive transporter was enhanced. The total uptake of [3H]-Ara-C in the HT-1080/hENT3 and HT-1080/hENT4 cells increased slightly, but a change in the NT function was not recognized compared with that in the HT-1080/mock cell line. In contrast, NT function in the TMK-1/mock cell line was different from that in the HT-1080/mock cell line (Fig. 3). In the TMK-1/mock cell line, both NT (es- and ei-type, 59 and 37%, respectively) were activated, but the HT-1080/mock cell line showed mainly es-type NT (96%). In TMK-1/hENT1 and TMK-1/hENT2, the total uptake of [3H]-Ara-C increased in comparison with that in the TMK-1/mock cell line; this increase was the result of the increased expression of hENT1 and hENT2. In particular, the activity of the ei-type NT in the TMK-1/hENT2 cell line was remarkably increased (3.4-fold), whereas the activity of the DPM-insensitive transporter was also enhanced. Less change was observed in the transporter activity in the TMK-1/hENT3 and TMK-1/hENT4 cells than in the TMK-1/hENT1 and TMK-1/hENT2 cells.

Figure 3.

The transport function of human equilibrative nucleoside transporters (hENT) analyzed as the intracellular uptake of cytarabine (Ara-C). The uptake of [3H]-Ara-C into HT-1080 and TMK-1 cells that stably overexpressed the green fluorescent protein (GFP)-hENT fusion protein. Total cellular accumulation of [3H]-Ara-C was measured at 37°C for 15 min (open column). To evaluate the functional transporter for [3H]-Ara-C, nucleoside transporter inhibitors such as nitrobenzylthioinosine (NBMPR) (2 µM, hatched column) or dipyridamole (DPM) (20 µM, closed column) were added to each well. Statistical analysis against the mock control was carried out using Student's t-test. *P < 0.05, **P < 0.01.


Ara-C and ECyd that were used in the present experiment were phosphorylated enzymatically by dCyd kinase or UCK2. These monophosphate kinases are critical key enzymes and the catalytic reaction is a rate-limiting step for antitumor activity. However, the phosphorylation step is an event that occurs after the transportation of antitumor nucleosides into the tumor cell. Recent reports have shown NT to be a critical and important factor regulating chemosensitivity to antitumor nucleosides. In the present study, we demonstrated that hENT1 and hENT2 affect chemosensitivity to antitumor nucleosides to a greater extent than hENT3 and hENT4. Furthermore, our results showed that GFP-hENT3 and GFP-hENT4 are localized in the intracellular organelles, whereas GFP-hENT1 and GFP-hENT2 are localized mainly in the plasma membrane. Previous reports have shown that hENT1 and hENT2 were localized to the basolateral membrane when they were expressed as fusion proteins with CFP or YFP in MDCK cells.(28) However, hENT3 was reported to localize to the lysosomes due to the presence of typical [DE]XXXL[LI] endosomal–lysosomal targeting motifs.(29) Our data correlate with the findings in the above report; the endosomal or lysosomal localization may cause the weak expression of GFP-hENT3. It is certain that hENT4 is located in the intracellular organelles rather than in the plasma membrane, although we could not clearly determine the intracellular localization of GFP-hENT4 in this study. Previous reports have indicated that the YFP-hENT4 fusion protein in MDCK cells is located in the plasma membrane, but endogenous hENT4 is located in intracellular organelles and partially in the plasma membrane.(30,31) Almost all of our findings coincided with the results obtained in these previous studies, although the localization of hENT4 differed from that of YFP-hENT4 expressed in MDCK cells because of the difference in the fluorescent tag used. In the cells expressing native ENT4, Barnes et al. have confirmed by specific antibody that the localization of ENT4 is an intracellular organelle.(30) The ENT4 localization of the GFP tag used in the present paper should be more similar to the native ENT4 localization than YFP-tagged ENT4.

We sequentially investigated whether the forced expression of GFP-fusion hENT, including hENT1 and hENT2, affects chemosensitivity to antitumor nucleosides. Contrary to our expectation, the cytotoxicity of Ara-C and ECyd in the HT-1080 cell lines with forced expression of GFP-hENT was almost unchanged. However, TMK-1/hENT1 and TMK-1/hENT2 were more sensitive to Ara-C and ECyd than the TMK-1/mock cells, whereas the changes in chemosensitivity in TMK-1/hENT3 and TMK-1/hENT4 cells to Ara-C and ECyd tended to be fewer than those in the TMK-1/hENT1 and TMK-1/hENT2 cells. We consider that the forced expression of GFP-hENT in the HT-1080 cells does not affect the antitumor efficacy of nucleoside analogs; this is because HT-1080 cells originally and constitutively express hENT1 mRNA, and the level of endogenous hENT1 mRNA in HT-1080 cells is sufficient to enable the expression of the hENT1 protein on the cellular plasma membrane for the intracellular transport of nucleoside analogs. In contrast, the endogenous mRNA expression levels of hENT and hCNT in the TMK-1 cells were obviously lower than those in the HT-1080 cells. Thus, in the TMK-1 cells, the forced expression of GFP-hENT1 and GFP-hENT2 might strongly affect the NT activity and, further, the chemosensitivity to nucleoside analogs. The effect of DPM on [3H]-Ara-C uptake in both the TMK-1/hENT2 and HT-1080/hENT2 cells was considerably less than that in other GFP-hENT-expressing clones. The decreased sensitivity to DPM might have resulted from the changes in the affinity of hENT2 together with the GFP protein; however, the mechanism responsible for the decreased sensitivity and the relationship between the sensitivity to DPM and the expression of hENT2 have not yet been determined. Furthermore, a slight stimulation of NT activity was observed in the clones expressing GFP-hENT3 and GFP-hENT4. It is quite likely that the GFP-hENT3 and GFP-hENT4 proteins cause increased accumulation of nucleoside analogs in intracellular organelles because of their cellular localization. It appears that the antitumor nucleosides are trapped in intracellular organelles. In these circumstances, nucleoside analogs may not be able to exert sufficient antitumor action. In fact, the chemosensitivity in the clones expressing GFP-hENT3 and GFP-hENT4 decreased slightly. It is clear that chemosensitivity to antitumor nucleosides is influenced by the enzymes responsible for the metabolism of the nucleosides, such as dCyd kinase, UCK2 and cytidine deaminase. In all of the transfected clones used in this experiment, there appears to be no change in the kinase and deaminase activity because the mRNA levels of these enzymes were not affected (data not shown). However, we could not find any obvious relationship between the expression of hENT3 and hENT4 and the chemosensitivity to antitumor nucleoside analogs in tumor cells. The cellular functions of hENT3 and hENT4 are unknown but their important cellular functions may exist. Our present data suggest that the functional expression levels of hENT1 and hENT2 are a more useful marker than those of hENT3 and hENT4 for predicting the chemosensitivity to antitumor nucleosides.


This research was partially supported by the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Young Scientists (B), 2004 (no. 15790071).