Human concentrative nucleoside transporter 1-mediated uptake of 5-azacytidine enhances DNA demethylation

The 5-azanucleosides 5-azacytidine (5-azaCyd) and 5-aza-2′-deoxycytidine (5-aza-2′-dCyd; decitabine) exhibit significant activity in the treatment of leukemias and solid tumors (1, 2). It has been shown that both 5-azaCyd and 5-aza-2′-dCyd are able to inhibit DNA methyltransferases in patients. This effect has been shown to cause pronounced DNA demethylation and, in some cases, epigenetic reactivation of tumor suppressor genes. As such, 5-azaCyd and 5-aza-2′-dCyd represent prototypical drugs for epigenetic cancer therapies (3).

Clinical response rates to 5-azanucleosides are typically in the range of 30% to 40%. Also at the epigenetic level, a wide range of responses has been described. For example, a previous analysis of bone marrow samples from patients undergoing 5-aza-2′-dCyd therapy showed no detectable demethylation in 2 of 7 patients, whereas the remaining 5 patients showed demethylation levels ranging from 10% to 70% (4). Similar results were also obtained in solid tumors (2) and with 5-azaCyd (5, 6). These findings strongly indicate that responses to 5-azaCyd and 5-aza-2′-dCyd can be modified by patient-specific factors. However, these factors have remained unknown.

Cellular uptake mechanisms are critical modulators of drug responses and can thus play a role in the therapy outcome with 5-azaCyd and 5-aza-2′-dCyd. The members of the human equilibrative nucleoside transporter (hENT; SLC29A) and human concentrative nucleoside transporter (hCNT; SLC28A) families have been previously implicated in the cellular uptake of several chemotherapeutic nucleoside analogues and are considered the principal nucleoside transporters under physiologic and pathophysiologic conditions (7). Studies based on gene expression profiles suggested that nucleoside transporters may mediate transport of 5-azaCyd and 5-aza-2′-dCyd (8); however, these studies provided no direct demonstration of the molecular mechanism for the transport of 5-azaCyd and 5-aza-2′-dCyd across the plasma membrane. The hENT proteins facilitate transport of nucleosides in a bidirectional manner with broad selectivities but relatively low affinities (compared with hCNT proteins) and the hCNT proteins mediate uptake of nucleosides against the concentration gradient, in a Na⁺-dependent manner, and with higher affinity compared with the hENTs (9, 10). In addition, both types of transporters are widely expressed (11). Lastly, hCNT proteins are involved in the regulation of nucleoside salvage, whereas the hENTs contribute to the nucleoside salvage in a more constitutive manner (12). Thus, we focused our study on hCNT proteins and, more specifically, on hCNT1.

Madin-Darby canine kidney (MDCK) cells stably expressing specific human transport proteins are a well-established cell system for the identification and functional characterization of drug transport mechanisms independent of the tissue-specific expression of the transport proteins (13–16). In this study, we show that expression of hCNT1 in MDCK cells is required for the uptake of 5-azaCyd. Importantly, expression of hCNT1 also sensitized MDCK cells to the cytotoxic effects of the drug, indicating a functional role of the protein in mediating drug responses. Consistent with this notion, hCNT1 also facilitated the degradation of cellular DNA methyltransferase 1 (DNMT1) protein and strongly sensitized the cells for the demethylating effects of 5-azaCyd. hCNT1-mediated sensitization of DNA demethylation was also observed for 5-aza-2′-dCyd, which showed that hCNT1 is functionally important for mediating the epigenetic effects of both 5-azaCyd and 5-aza-2′-dCyd.

[2-¹⁴C]Cytidine (1.9 MBq/mmol), [5-³H]uridine (0.6 TBq/mmol), and [6-³H]5-fluoro-5′-deoxyuridine (5-F-5′-dUrd; 0.1 TBq/mmol) were purchased from Moravek Biochemicals. The synthesis of [6-¹⁴C]5-azaCyd was done as published previously (17), with the following modification: the synthesis started from commercially available [¹⁴C]dimethylformamide (50-62 mCi/mmol; GE Healthcare), thus obviating the conversion of ¹⁴C-labeled sodium formate to [¹⁴C]dimethylformamide. The specific activity of the final product was determined by dividing the radioactivity amount determined in a calibrated β-well-counter by the weight of the final crystalline compound. The purity of [¹⁴C]5-azaCyd was 98% as determined by radio-TLC and radio high-performance liquid chromatography. Unlabeled cytidine, 5-azaCyd, 5-aza-2′-dCyd, and uridine were obtained from Sigma and 5-F-5′-dUrd was from Sequoia Research Products. All other chemicals were of analytic grade and purchased from Sigma, Merck, or AppliChem.

The peptide corresponding to the 23-amino-proximal amino acids of the hCNT1 sequence (AEIRSSWSEAAPKPFSRWRNLQP; National Center for Biotechnology Information accession no. NP_004204) was synthesized and coupled to keyhole limpet hemocyanin (Peptide Specialty Laboratories). Rabbits were immunized with this conjugate to raise the AEI antiserum. The AEI antiserum was purified on an affinity column with the immobilized AEI peptide (Peptide Specialty Laboratories). The horseradish peroxidase-conjugated goat anti-rabbit IgG was from Bio-Rad and the Alexa Fluor 488-conjugated goat anti-rabbit IgG was from Molecular Probes.

MDCK cells strain II were cultured as described previously (16). MDCK cells were transfected with the pcDNA3.1(+)-hCNT1 cDNA construct, which was kindly provided by Marçal Pastor-Anglada (University of Barcelona; ref. 18) or vector only using Metafectene transfection agent (Biontex). Stable transfectants were selected using medium containing G418 (0.5 mg/mL). Resistant clones were induced with 10 mmol/L sodium butyrate for 24 h to enhance the expression of the recombinant protein (19) and screened by immunoblot analysis and immunofluorescence microscopy for hCNT1 expression.

Crude membranes were prepared from transfected MDCK cells in the presence of proteinase inhibitors and analyzed by immunoblotting as described (19). The AEI antiserum was diluted 1:80,000 in PBS containing 0.05% Tween 20 and the horseradish peroxidase-conjugated goat anti-rabbit antiserum was used at a dilution of 1:20,000. Deglycosylation was done as described (19).

MDCK cells were grown on ThinCert membrane inserts as described previously (16). MDCK cells were fixed for 30 min with 2% paraformaldehyde in PBS and permeabilized for 30 min in 1% Triton X-100 in PBS. Cells were incubated with the affinity-purified AEI antiserum for 1 h at room temperature, washed in PBS, incubated with Alexa Fluor 488-conjugated goat anti-rabbit IgG for 1 h at room temperature, and then washed again in PBS. The membrane inserts were mounted onto glass slides using 50% glycerol in PBS. Confocal laser scanning micrographs of the cells were taken under the same conditions on a LSM510 apparatus (Carl Zeiss).

MDCK cells were seeded in 6-well plates at a density of 2 × 10⁶ per well and cultured for 24 h at confluence. Cells were induced with 10 mmol/L sodium butyrate for additional 24 h. Uptake studies examining the Na⁺-dependent uptake of nucleosides were done at 37°C and using Na⁺-containing buffer [100 mmol/L NaCl, 2 mmol/L KCl, 1 mmol/L CaCl₂, 1 mmol/L MgCl₂, and 10 mmol/L HEPES (pH 7.5)] and Na⁺-free buffer in which NaCl was replaced by equimolar concentrations of choline chloride. Cells were first washed three times with Na⁺-free buffer and then incubated with 1 mL Na⁺-containing buffer containing the labeled substrate at the indicated concentration. After incubation at 37°C, cells were washed three times with ice-cold Na⁺-free buffer. Intracellular radioactivity was determined by liquid scintillation counting after cell lysis with 0.2% SDS. For determination of kinetic constants, transport rates of the nucleosides were measured at substrate concentrations ranging from 75 μmol/L to 1 mmol/L and after 20 s incubation of the substrate with the cells.

The sensitivity of MDCK cells to 5-azaCyd and 5-aza-2′-dCyd was assessed with AlamarBlue assays (Biosource). MDCK cells were seeded (1 × 10⁴ per well) in 96-well plates and incubated for 24 h before exposure to graded concentration of each drug for 72 h (13). The IC₅₀ value was defined as the drug concentration required to reduce cell survival, as determined by the relative absorbance of reduced AlamarBlue, to 50%.

Cells were treated with 50 or 100 nmol/L 5-azaCyd for 6 or 24 h as indicated. Soluble protein extracts were prepared by freeze-thaw lysis as described previously (20) and further analyzed by immunoblotting using antibodies against DNMT1 and β-actin (Santa Cruz Biotechnology).

Genomic DNA was purified using the DNeasy Blood and Tissue Kit (Qiagen). Global methylation levels were determined by capillary electrophoresis as described previously (21).

For statistical analysis, the Student's t test was used. P < 0.01 was considered significant, and P < 0.001 was considered highly significant.

A stable clonal cell line (MDCK-CNT1) was generated to investigate the localization of hCNT1 in polarized cells and to functionally characterize hCNT1. Expression of hCNT1 was analyzed by immunoblotting (Fig. 1A). The polyclonal AEI antiserum specifically detected a strong, broadband, characteristic for glycosylated proteins at 66 to 80 kDa in crude membranes from MDCK-CNT1 cells (Fig. 1A). The band at 66 to 80 kDa disappeared specifically in MDCK-CNT1 crude membranes when the AEI antiserum was preincubated with the synthetic AEI peptide that had been used to generate the AEI antiserum (data not shown). Deglycosylation of the hCNT1 protein using peptide N-glycosidase F shifted the 66 to 80 kDa band to a band with a molecular mass of ∼60 kDa (Fig. 1B).

Confocal laser scanning microscopy using the affinity-purified AEI antiserum was done to study the localization of hCNT1 in stably transfected MDCK cells (Fig. 1C-F). Optical vertical sections showed intense green fluorescence for hCNT1 in the apical membrane of MDCK-CNT1 cells (Fig. 1D). No plasma membrane staining was observed in vector-transfected MDCK-Co cells (Fig. 1E and F). Thus, the affinity-purified AEI antiserum localized hCNT1 to the apical plasma membrane of polarized MDCK cells.

To experimentally address the possibility that hCNT1 might mediate the transport of 5-azaCyd across the plasma membrane, we synthesized a ¹⁴C-labeled tracer molecule (see Materials and Methods). Uptake of [¹⁴C]5-azaCyd was measured in MDCK-Co and MDCK-CNT1 cells using Na⁺-containing and Na⁺-free buffers to determine the Na⁺-dependent uptake (Fig. 2A). The hCNT1-mediated transport of 5-azaCyd was unidirectional, Na⁺- and time-dependent, and saturable with a K_m value of 63 μmol/L (Fig. 2A; Table 1). The intracellular accumulation of [¹⁴C]5-azaCyd at a concentration of 1 μmol/L amounted to 175.2 ± 9.8 and 4.1 ± 0.6 pmol/mg protein after 1 min in Na⁺ and choline medium, respectively, in MDCK-CNT1 cells. In contrast, the MDCK-Co cells did not mediate Na⁺-dependent uptake of 5-azaCyd (Fig. 2A; 4.8 ± 0.8 and 4.5 ± 0.5 pmol/mg protein after 1 min in Na⁺ and choline medium, respectively).

Several established hCNT1 substrates were used to further characterize the MDCK transfectants. The MDCK-CNT1 cells were functionally active in nucleoside transport as evidenced by the time- and Na⁺-dependent transport of cytidine, uridine, and 5-F-5′-dUrd (Fig. 2B-D), which amounted to 213, 257, and 141 pmol/mg protein/min in MDCK-CNT1 cells, respectively (Fig. 2B-D). hCNT1 showed K_m values for cytidine, uridine, and 5-F-5′-dUrd of 123, 467, and 403 μmol/L, respectively (Table 1).

The effect of 5-azaCyd into cells via hCNT1 was studied in cell viability assays (Fig. 3A). 5-azaCyd was added for 72 h to the MDCK-Co and MDCK-CNT1 cells growing on 96-well plates, cell viability was determined by the AlamarBlue assay, and IC₅₀ values of 54 ± 5 and 1 ± 0.1 μmol/L were obtained for MDCK-Co and MDCK-CNT1 cells, respectively. The relative sensitivity of MDCK-CNT1 cells increased 54-fold compared with MDCK-Co cells (P < 0.001). Increased uptake of 5-azaCyd by hCNT1 expression therefore enhanced the relative sensitivity of the cells to the drug. Similarly, MDCK-CNT1 cells also showed increased sensitivity to 5-aza-2′-dCyd, with IC₅₀ values 72 ± 10 μmol/L for MDCK-CNT1 cells and 331 ± 48 μmol/L for MDCK-Co cells.

Incorporation of 5-azanucleosides into DNA results in covalent trapping of DNA methyltransferase protein, which can be visualized by the depletion of DNMT1 from protein extracts (22). DNMT1 is highly conserved among vertebrates and antibodies against human DNMT1 can be used to specifically detect the canine orthologue in MDCK cells (Fig. 3B). To test if the hCNT1-mediated import of 5-azaCyd would facilitate the depletion of DNMT1, we determined DNMT1 protein levels by immunoblotting. Although DNMT1 was still detectable in control cells after 6 h treatment with 50 nmol/L 5-azaCyd, the protein was undetectable in MDCK-CNT1 cells (Fig. 3B). Similar results were obtained in parallel experiments with a higher concentration of 5-azaCyd (100 nmol/L; Fig. 3B). These results showed that overexpression of hCNT1 facilitates 5-azaCyd-dependent depletion of DNMT1 and further support the notion that hCNT1 plays a functional role in mediating the effects of this drug.

To determine whether the increased depletion of DNMT1 caused a more pronounced demethylation by 5-azaCyd, we analyzed the global methylation levels with capillary electrophoresis. The results showed that control MDCK cells had a methylation level of ∼3.0%, which is very similar to the global methylation level of human cell lines and tissues (23, 24). After 24 h treatment with 50 nmol/L 5-azaCyd, cytosine methylation levels of MDCK-CNT1 cells were significantly reduced when compared with controls (1.7% versus 3.2%; P < 0.001; Fig. 3C). Similar results were obtained in parallel experiments with a higher concentration of 5-azaCyd (100 nmol/L; Fig. 3C). These results corroborate our findings with the DNMT1 trapping assay (Fig. 3B) and further confirm a functional role of hCNT1 in mediating the effects of 5-azaCyd.

We also sought to determine whether hCNT1 plays a similar role in the cellular uptake of the closely related DNA methyltransferase inhibitor 5-aza-2′-dCyd. To this end, we incubated MDCK-CNT1 and MDCK-Co cells with 50 nmol/L 5-aza-2′-dCyd and isolated genomic DNA after 24 h. Capillary electrophoretic analysis revealed 2.9% cytosine methylation in control cells but only 1.5% methylation in hCNT1-expressing cells (P < 0.001; Fig. 3D). These data strongly indicate that hCNT1 also mediates the cellular uptake of 5-aza-2′-dCyd.

The effective cellular uptake of physiologic nucleosides and therapeutic nucleoside analogues across the plasma membrane needs to be mediated by transport proteins. Membrane transport of nucleosides and nucleoside analogues therefore represents an essential and often rate-determining step in many cell types that lack the ability to synthesize these molecules de novo and for the treatment of tumor cells with analogues (7, 25). The two nucleoside analogues, 5-azaCyd and 5-aza-2′-dCyd, are two emerging drugs that have found increasing use for the treatment of myeloid leukemias and other human cancers. However, the extended mode of action of these drugs is still largely unknown and the cellular uptake mechanisms remained to be identified (26). The members of the hCNT/SLC28 and hENT/SLC29 families have been described to mediate nucleoside and nucleoside analogue transport (7) and therefore were considered candidates for mediating cellular uptake of 5-azanucleosides. The ability of concentrative nucleoside transporters to mediate nucleoside uptake in a unidirectional manner, combined with previous reports (27, 28), established hCNT1 as a particularly attractive candidate for mediating the cellular uptake of 5-azaCyd. Consequently, the principal aim of our study was to establish a cellular model expressing hCNT1 for the study of the uptake of nucleosides and nucleoside analogues and thus to test the functional role of hCNT1 in mediating the drug effects of 5-azaCyd.

hCNT1 was expressed only in the MDCK-CNT1 transfectants and at the expected apparent molecular mass. The affinity-purified AEI antiserum localized this nucleoside transporter to the apical plasma membrane of polarized MDCK cells. Until now, the localization of hCNT1 in polarized cells has only been studied in cell lines transfected with the hCNT1-XFP fusion constructs (14) or through the characterization of membrane vesicles from tissue epithelial cells (29). To our knowledge, the present stable MDCK-CNT1 transfectants therefore represent the first polarized cell line that express the recombinant untagged hCNT1 and sorted hCNT1 to the apical plasma membrane. In addition, the AEI antiserum represents a valuable tool for the detection of hCNT1 protein.

Using MDCK-CNT1 transfectants, we identified hCNT1 as a novel and thus far the first known nucleoside transporter for 5-azacytdine. The hCNT1-mediated transport of [¹⁴C]5-azaCyd was unidirectional, Na⁺- and time-dependent, and saturable. hCNT1 showed high affinity for 5-azaCyd with a K_m value of 63 μmol/L. The MDCK-CNT1 cells were also functionally active in nucleoside transport as evidenced by the time-dependent transport of cytidine, uridine, and 5-F-5′-dUrd and their saturable kinetics with K_m values. The K_m values obtained for these pyrimidine nucleosides are in the micromolar range as those reported previously (18, 30–32). Comparably minor differences may, in part, be explained by the expression of recombinant hCNT1 in different cell systems. Moreover, our results are in line with previous reports that suggested a common nucleoside transport system for 5-azaCyd as well as cytidine and uridine with similar efficiency (33). Further studies are required to determine whether additional hCNT proteins, such as hCNT3 (34), can also mediate 5-azaCyd transport.

For the characterization of DNA methyltransferase inhibitors, such as 5-azaCyd and 5-aza-2′-dCyd, human cancer cell lines have been used as experimental systems (6, 35). However, the cellular uptake systems for these drugs are not well defined. The present data show that the specific expression of hCNT1 in MDCK cells strongly enhances DNA demethylation by 5-azaCyd and by 5-aza-2′-dCyd. Of note, we also observed a detectable decrease in DNMT1 protein levels in control transfected MDCK cells treated with 5-azaCyd, thus suggesting the presence of a basal 5-azaCyd transport system in these cells. However, limited depletion of DNMT1 did not result in detectable DNA demethylation, which is in agreement with recent reports showing that low levels of DNMT1 protein are sufficient for the maintenance of comparably normal genomic DNA methylation in human cancer cell lines (36, 37). Complete depletion of DNMT1 could only be detected in hCNT1-transfected cells, which further underscores the significance of this transporter for the demethylation response to 5-azaCyd.

Other nucleoside transporters might also be involved in mediating cellular uptake of 5-azaCyd. Further studies are required to assess the effect of efflux transporters on intracellular nucleoside concentrations in addition to uptake transporters. It will be important to correlate the expression and subcellular localization of transport proteins such as hCNT1 to clinical and molecular response variables from patients undergoing demethylation therapy with 5-azanucleosides and to evaluate hCNT1 expression as a potential biomarker for 5-azaCyd and 5-aza-2′-dCyd responses.

No potential conflicts of interest were disclosed.

We thank Marçal Pastor-Anglada (University of Barcelona) for providing the hCNT1 cDNA and Ralf Schirrmacher (McGill University) and Peter Bartenstein (University of Mainz) for support.