Chemicals.

Poly-d-lysine, 6MP, TG, thioxanthine (TX), 6-thioguanosine (TGrib), 6-mercaptopurine riboside (MPrib), MeMP, 6-methyl thioguanine, MeMPrib, and PRPP were obtained from Sigma (Zwijndrecht, The Netherlands). tIMP, tGMP, and tXMP were synthesized from 6MP, TG, TX, and PRPP using HGPRT from Saccharomyces cerevisiae (Sigma), as described previously (Wijnholds et al., 2000). MetIMP and 6-methylated tGMP were made from tGMP or tIMP (1 mM) by overnight incubation at room temperature in an aqueous solution containing 70 mM methyl bromide (Sigma) and 3.75% ammonia; thereafter, excess ammonia and methyl bromide were evaporated under vacuum, essentially as described previously (Keuzenkamp-Jansen et al., 1995). TXrib was generated from tXMP by 5′-nucleotidase, and thio-GDP, tGTP, thio-IDP, and thio-ITP were synthesized as described previously (Breter and Mertes, 1990).

Cell Lines.

HEK293 parental cells and HEK293/5I and HEK293/5E cells transduced with MRP5 cDNA have been described previously (Wijnholds et al., 2000). The subclone HEK293/5GE was isolated at the same time as HEK293/5I and HEK293/5E but has not been described before. HEK293/5I and HEK293/5E have high MRP5 overexpression, whereas HEK293/5GE has a moderate increase in MRP5 relative to the parental cells (see Results). MRP4-overexpressing cells HEK293/4.3 and HEK293/4.63 were made by transducing HEK293 cells with pMSCV-IRES-EGFP virus containing full-length MRP4 cDNA, the cloning of which has been described elsewhere (Adachi et al., 2002). All cells were grown in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal calf serum, 100 IU/ml penicillin, and 100 μg/ml streptomycin (Invitrogen) at 37°C under 5% CO₂/humidified air. The cells were routinely checked for mycoplasma and MRP4 and MRP5 expression levels.

Monoclonal Antibodies.

Anti-MRP5 monoclonal antibody (mAb) NKI-12C5 was generated by cloning a fragment of the mouse Mrp5 cDNA, encoding amino acids 1 to 38, into the pMAL-c vector (New England Biolabs, Beverly, MA) to produce a fusion protein consisting ofEscherichia coli maltose-binding protein fused to the N terminus of mouse Mrp5. The affinity-purified protein was injected into rats, and hybridoma cells producing MRP5-specific mAbs were generated, essentially as described previously (Harlow and Lane, 1988). NKI-12C5 specifically recognizes the full-length human MRP5 and mouse Mrp5 protein. The MRP4-specific antibody NKI-12C4 was generated in a similar way using a maltose-binding protein fusion protein containing an internal epitope (amino acids 372–432) from human MRP4. A more complete characterization of these mAbs will be presented elsewhere.

Western Blot Analysis.

Protein from a total cell lysate was quantified using the Bio-Rad protein assay (Bio-Rad, Hercules, CA), fractionated on a 8% polyacrylamide slab gel and transferred to a nitrocellulose membrane by electroblotting, essentially as described previously (Kool et al., 1997). After blocking for 1 h in phosphate-buffered saline (PBS) containing 1% nonfat dry milk, 1% bovine serum albumin, and 0.05% Tween 20, the membrane was incubated for 1 h at room temperature with the first antibody. As secondary antibody, horseradish peroxidase-labeled goat anti-rat antibody, preabsorbed for both mouse and human IgGs, was used at a dilution of 1:2000 (Santa Cruz Technology, Santa Cruz, CA). Enhanced chemiluminescence (Amersham Biosciences, Little Chalfont, Buckinghamshire, UK) was used for detection.

Immunohistochemistry.

To detect MRP4 and MRP5 by immunofluorescence, cells were grown overnight on poly(d-lysine)-coated microscope slides. The next day, the cells were fixed with methanol (30 s, −20°C) and immunostained with NKI-12C4 and NKI-12C5 (undiluted hybridoma culture medium). The first antibody was visualized using a goat anti-rat-Alexa488-conjugated fluorescent antibody (1:500; Molecular Probes, Leiden, the Netherlands). Cells were mounted using Vectashield (Vector Laboratories, Burlingame, CA) containing 40 μg/ml ethidium bromide to visualize nuclear DNA. Slides were analyzed by confocal laser-scanning microscopy (Leica, Heidelberg, Germany), using a 488-nm laser for excitation and a 530-nm band-pass filter for detection of Alexa488 and a 560-nm high-pass filter for detection of ethidium bromide-stained nuclear DNA.

Cytotoxicity.

Relative growth of cells in the presence of thiopurines was tested essentially as described before (Wijnholds et al., 2000). In short, cells were plated in triplicate in 96-well plates (1500 cells/well, 100 μl/well) in conditioned medium. The next day, drug was added at the appropriate dilutions in a volume of 25 μl. After an additional 4 days, cell growth was determined using the CyQuant cell proliferation assay kit (Molecular Probes). The fluorescence in each well was determined using a CytoFluor 4000 plate reader (Applied Biosystems, Foster City, CA) and the IC₅₀ value was defined as the concentration of drug that inhibited growth of the cells by 50%. The relative resistance was then determined as the ratio of the IC₅₀ values of the transduced and parental cells.

Transport Experiments.

Cells were seeded at a density of 2 × 10⁶ per well in poly(d-lysine)-coated 12-well plates and were grown overnight. The next day, the cells were washed with PBS and then incubated at 37°C with various thiopurines in Hanks' buffered salt solution (Invitrogen). At the time points indicated, both incubation medium and the cells were collected. The incubation medium was centrifuged and the supernatant was used for analysis. For the analysis of the intracellular metabolites, the cells were washed with ice-cold PBS and extracted using 70% methanol/water, and insoluble material was removed by centrifugation. After complete evaporation of the methanol/water phase at 40°C, the residue was solubilized in 200 μl of HPLC elution buffer A (see below) and analyzed by HPLC.

HPLC Analysis.

Cell extracts and incubation media were analyzed by reversed-phase ion-pairing HPLC analysis using a Luna-C18 column (284 Luna, 5 μm, 250 × 4.6 mm; Phenomenex, Cheshire, UK) and a two-buffer gradient system to separate the different nucleotides. The following pump conditions were used: 0–5 min, 100% buffer A; 5–25 min, linear to 70% buffer B; 25–28 min, 70% buffer B; 28–30 min, linear to 100% buffer A; 30–45 min, 100% buffer A. The composition of buffer A was 100 mM NaH₂PO₄, 5.9 mM tetrabutylammonium hydrogen sulfate, 0.34 mM EDTA, and 1% (v/v) acetonitrile. Buffer B was the same as buffer A but contained 25% (v/v) acetonitrile. The flow was kept at 1 ml/min. The complete UV-visible absorbance spectrum (200–800 nm) of the effluent was monitored using a Waters 966 photodiode array detector (Waters Chromatograpy B.V., Etten-Leur, the Netherlands). The metabolites were identified by their specific retention times and UV absorbance maxima (UV_max; see Table 2). Thiopurine metabolites were quantified at their UV_max by relating the peak area of the metabolites detected with the peak area measured for a standard amount of either 6MP, TG, TX, or MeMPrib. Intracellular ATP was monitored to ensure that the cells were metabolically active, and the absence of extracellular ATP confirmed that cells had remained intact during the experiment.

MRP4 and MRP5 Levels in Transfected HEK293 Cells.

The levels of MRP4 and MRP5 in the HEK293 cell lines transduced with MRP4 or MRP5 cDNA, respectively, were quantified using two new mAbs, NKI-12C4 (MRP4) and NKI-12C5 (MRP5), as shown in Fig. 1. Recombinant MRP4 protein migrates as a broad band of approximately 220 to 230 kDa; MRP4-transduced HEK293/4.3 and HEK293/4.63 cells contain similar levels of MRP4 (Fig. 1A). NKI-12C4 recognizes an endogenous cross-reacting protein of ∼90 kDa in all HEK293 variants, in addition to a band of ∼70 kDa, which increases in intensity with increasing MRP4 expression. The latter may represent an MRP4 breakdown product. MRP5 migrates as a single band of 220 to 230 kDa (Fig. 1B). Overexpression of MRP5 is high in HEK293/5I and HEK293/5E and intermediate in HEK293/5GE cells. Two smaller bands of ∼95 and ∼60 kDa may represent breakdown products. In contrast to our previous anti-MRP5 antibody (Wijnholds et al., 2000), the new antibody NKI-12C5 clearly detected endogenous expression of MRP5 in the parental HEK293 cells, whereas the NKI-12C4 mAb detects endogenous MRP4 (Fig. 1, A and B). We see no effect on endogenous levels after introduction of exogenous MRP4 or MRP5 (Fig. 1, A and B). Figure 1C shows the location of MRP4 and MRP5 in HEK293 cells with immunofluorescence. Although we detect weak signals for endogenous MRP4 and MRP5 in the parental line, the intensity was too low to determine the subcellular localization (Fig. 1C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Western blot analysis and immunofluorescent detection of MRP4 and MRP5. MRP4 and MRP5 expression in the three cell lines used in this study is shown. For the Western blot, 15 μg of total protein was loaded per lane. MRP4 was detected by rat anti-human MRP4 monoclonal antibody NKI-12C4 (A) and MRP5 by rat anti-mouse MRP5 monoclonal antibody NKI-12C5 (B). Detection of MRP4 and MRP5 by immunofluorescence (C) was done on methanol-fixed cells using the indicated antibodies as outlined under Materials and Methods. MRP4- and MRP5-specific staining is shown in green, and the nuclear DNA stained by ethidium bromide is shown in red. *, ∼90 kDa band recognized by NKI-12C4 found in all cell lines; thus, it is most probably an endogenous protein. **, ∼75 kDa band in A, two bands of ∼95 and ∼60 kDa in B, of which the intensity increases with increasing MRP4 or MRP5 expression. The shorter film exposures demonstrate the relative levels of the introduced protein in the MRP4- and MRP5-overexpressing cells.

Overexpression of MRP5 in HEK293 cells, as in HEK293/5I cells, results in a 2- to 3-fold relative resistance against 6MP and TG (Table1). Analogous results were obtained for HEK293 cells overexpressing MRP4, as shown in Table 1.

HPLC Analysis.

To analyze the thiopurine metabolites formed by the HEK293 cells, we set up a reversed-phase ion-pairing HPLC system in combination with spectral UV-visible detection. This system is stable and allows the separation of the metabolites of both physiological and thiopurine nucleosides within 30 min (Fig.2). The combination of elution time and UV-visible absorbance spectra (Fig. 2, insets) makes it possible to discriminate between metabolites. Table 2provides an overview of UV_max and retention times for the metabolites analyzed in this study.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

HPLC analysis of reference compounds. Chromatograms of mixtures of nucleobases, nucleosides and nucleotides separated by reversed-phase ion-pairing HPLC. A, chromatogram showing the separation of guanosine (1), inosine (2), IMP (3), GMP (4), adenosine (5), AMP (6), and ATP (7), detected at 260 nm. B, chromatogram showing the separation of 6MP (8), TX (9), MPrib (10), tIMP (11), TXrib (12), and tXMP (13), detected at 330 nm. *, peaks that were derived from the Hanks' balanced salt solution in which thiopurines were dissolved. UV absorbance is given in arbitrary units. The insets show examples of absorbance spectra of indicated peaks, which were used to identify different metabolites.

6-Mercaptopurine Uptake by HEK293 Cells.

In previous experiments with HEK293/5I cells (Wijnholds et al., 2000), cells were loaded with radiolabeled 6MP under ATP-depletion conditions and efflux was followed after restoration of cellular ATP. This discontinuous incubation is somewhat artificial and results in 10-fold lower 6MP uptake than in the presence of normal cellular ATP levels. We therefore switched to a protocol in which the cells are continuously exposed to 6MP under normal conditions. 6MP is lipophilic and is assumed to rapidly equilibrate across the plasma membrane. Accumulation of 6MP is therefore mainly determined by its intracellular ribophosphorylation, a process that is dependent on HGPRT activity and the intracellular PRPP concentration. Uptake of 6MP by HEK293 cells over 4 h increased with increasing 6MP (Fig. 3A). However, increasing the 6MP concentration 5-fold from 2.5 to 12.5 μM yielded only a 2-fold increase in uptake. This suggests that the uptake of 6MP was no longer linear (Fig. 3B), presumably because the cells were unable to maintain the initial rate of conversion of 6MP into tIMP. HEK293, HEK293/5I, and HEK293/4.3 cells took up similar levels of 6MP from the medium (Fig. 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Uptake of 6MP. A, concentration-dependent uptake of 6MP was determined after 4 h for the HEK293 (■), HEK293/5I (○), and HEK293/MRP4.3 (▵) cells. Initial 6MP concentrations in the incubation medium were 0.5, 2.5, and 12.5 μM. After 4 h of incubation at 37°C, the 6MP concentration in the medium was determined. Uptake was determined by the difference in 6MP concentration in the medium at t = 0 andt = 4 h. B, time-dependent uptake of 2.5 μM 6MP, with symbols as in A. The extracellular 6MP concentration is plotted as picomoles per 10⁶ cells calculated using the volume of incubation medium (1 ml) and number of cells (2 × 10⁶). Data are the mean ± S.E.M. of three independent determinations.

Thiopurine Efflux Kinetics for HEK293 Cells in the Continuous Presence of 6-Mercaptopurine.

Five thiopurine metabolites were present in the chromatograms from HPLC analysis of the medium from HEK293, HEK293/4.3, and HEK293/5I cells after incubation with 2.5 μM 6MP. These were identified as MPrib, tIMP, TX, TXrib, and tXMP. Figure4 shows the excretion of the four major metabolites, MPrib, tIMP, TXrib and tXMP over time. Table3 gives the excretion rates for the different thiopurines and shows that the excretion of tIMP was greatly increased in cells overexpressing MRP4 or MRP5. The excretion of tXMP was detected only with HEK293/5I cells. Using the other MRP4 and MRP5 overexpressing HEK293 clones (Fig. 1), we obtained similar results (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

6MP metabolites excreted by HEK293 cells. Metabolites excreted from the HEK293 (■), HEK293/5I (○), and HEK293/4.3 (▵) during continuous incubation with 6MP (initial concentration 2.5 μM). The levels of tIMP (A), tXMP (B), MPrib (C), and TXrib (D) were quantified at their UV_max wavelength, using chromatograms similar to those shown in Fig. 2. The amounts of the various thioinosine and thioxanthosine metabolites were calculated by relating the peak area of the detected metabolites with the peak area obtained with standard amounts of either 6MP or TX. The data are the mean ± S.E.M of three independent determinations. The indicated data points were significantly different (*, p < 0.05) from the parental cells in a Student's t test.

To exclude the possibility that an increase in intracellular formation of tIMP and tXMP was responsible for the release of these compounds from HEK293/4.3 and HEK293/5I cells, the intracellular thiopurine metabolites were determined. The time course of metabolite formation in HEK293 cells is shown in Fig. 5. After incubation with 2.5 μM 6MP, intracellular tIMP levels rose rapidly (Fig. 5A), followed by a rise in MPrib. This suggests that the MPrib is made by dephosphorylation of tIMP by a phosphatase or nucleotidase (Skladanowski et al., 1996; Hunsucker et al., 2001), followed by rapid excretion of the nucleoside via a nucleoside transporter (Baldwin et al., 1999). As expected, intracellular tXMP and TXrib concentrations follow the rise in tIMP in the cells with some delay (Fig. 5), and these compounds also appear later in the medium (Fig. 4). The decrease in 6MP uptake after 1 h (Fig. 3B) is accompanied by a strong decrease in intracellular tIMP and MPrib (Fig. 5). This suggests that conversion of 6MP decreases sharply after 1 h of incubation. Eventually, excretion of tIMP also slows down after 2 h (Fig. 4A). Extracellular MPrib even tends to decrease after 2 h (Fig. 4C), which may be caused by re-entry of MPrib into the cells and conversion into 6MP. Indeed, when HEK293 cells were incubated with MPrib, substantial amounts of 6MP were formed (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Intracellular 6MP metabolites. Metabolites found in the intracellular fractions from the HEK293 (A), HEK293/4.3 (B), and HEK293/5I (C) during continuous incubation with 6MP (initial concentration, 2.5 μM). Levels of tIMP (▵), MPrib (○), tXMP (▴), and TXrib (●) were quantified by relating the peak areas of detected metabolites with the peak area obtained with a 6MP and TX standards. TX levels were below the detection limit (< 1 pmol/10⁶ cells). The data are the mean ± S.E.M. of three independent determinations.

The bulk of the 6MP taken up by the cells was found back in the medium as nucleoside in the form of MPrib and to a lesser extent TXrib (Fig.4). To exclude the possibility that these nucleosides were produced by extracellular dephosphorylation of exported nucleotides, we incubated cells in medium with tIMP (2.5 μM): only 1 to 1.5% tIMP/10⁶ cells/h was dephosphorylated (results not shown), suggesting that the vast majority of MPrib observed was formed intracellularly.

The total tIMP accumulated after a 1-h incubation was 60, 50, and 60 pmol/10⁶ cells for HEK293, HEK293/4.3, and HEK293/5I (Fig. 5). Although comparable intracellular tIMP levels were achieved, the rate of tIMP efflux (calculated by simple linear regression) was markedly different: 0.09, 0.56 and 0.64 pmol/10⁶ cells/min for HEK293, HEK293/4.3, and HEK293/5I cells, respectively (Table 3). Only the HEK293/5I seemed to efflux tXMP (Fig. 4B and Table 3) at a rate comparable with the rate of tIMP excretion (Table 3). The levels of tIMP and tXMP accumulated after 2 h were decreased in the resistant cells (Fig. 5), showing that as tIMP production slows, the pumps are able to decrease tNMP levels in the cells. With HEK293/4.63 cells, we obtained results similar to those of the HEK293/4.3 cells, and the HEK293/5I results were essentially duplicated with HEK293/5E and HEK293/5GE cells (data not shown). The tNMP transport activity of the HEK293/5GE cells was lower, however, than that of the HEK293/5I cells, in line with their lower expression (Fig. 1). Both HEK293/4.3 (Fig. 4D) and HEK293/4.63 cells (not shown) excreted more TXrib than the HEK293 parental and MRP5-overexpressing cells. This was not caused by an increased excretion of thionucleosides in general, because the MPrib excretion was lower in these cells (Fig.4C).

In our 6MP incubation experiments, we detected neither thioguanine metabolites nor di- or triphosphorylated thiopurine nucleosides. Methylated thioinosines (UV_max ∼286 nm) or methylated thioguanosines (UV_max ∼310 nm) were not detected in cells either. Absence of thioguanine nucleotides and methylated thiopurines might be attributable to the low rates of TPMT and IMP dehydrogenase under our conditions. Small amounts of di- and triphosphates are probably formed but remain undetectable because our cell extracts contain high background absorbance in the area where these compounds elute from the HPLC. It is unlikely, however, that we miss a substantial fraction of the metabolites formed, because we obtain nearly 100% recovery of the 6MP taken up in the form of 6MP metabolites in the cells plus extracellular medium. The calculated values for the 4-h time points were 94 ± 3, 96 ± 14, and 103 ± 25% (mean ± S.E.M.) for HEK293, HEK293/4.3, and HEK293/5I cells, respectively.

Thiopurine Metabolism and Transport by HEK293 Cells Incubated with 6-Thioguanine.

Because we were unable to detect TG metabolites in cells incubated with 6MP, we tested TG itself, a compound to which the transfectants are resistant (Table 1). The HEK293 cells took up TG at about the same rate as 6MP (results not shown). The only metabolites found consistently were tGMP and TGrib. The excretion of tGMP was increased in both MRP4 and MRP5 transfectants; the increase was most pronounced in the MRP4 cells (Fig. 6A and Table 3). The efflux rates were 0.24, 0.77, and 1.22 pmol/10⁶ cells/min for HEK293, HEK293/5I, and HEK293/4.3 cells, respectively. Excretion of TGrib was also substantial and did not differ for the three cell lines (Table 3 and Fig. 6B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Continuous incubation with TG. Thiopurine metabolites found in the extracellular (left, A and B) and intracellular (right, C and D) fractions from the HEK293 (■), HEK293/5I (○), and HEK293/4.3 (▵) cells during continuous incubation with TG (initial concentration 2.5 μM). Thioguanine metabolite levels were quantified by relating the peak areas of detected metabolites with the peak area obtained with a TG standard. The data are the mean ± S.E.M. of three independent determinations. The indicated data points were significantly different (*, p < 0.05) from the parental cells in a Student's t test.

In contrast to the intracellular MPrib levels formed during exposure to 6MP, intracellular TGrib levels stayed relatively low and did not show a high peak after 1 h (Fig. 6D). In the analysis of the intracellular fractions, di- and triphosphorylated thioguanines were detected in only one of three experiments. In that experiment, we found a peak after 1 h with a retention time of 29 min and UV_max of ∼342 nm, which is most probably tGTP. This was present at levels of 51, 33, and 39 pmol/10⁶ cells for the HEK293, HEK293/4.3, and HEK293/5I cells, respectively, comparable with the tGMP levels of these cells. The overall recovery of TG taken up by the cells for the 4 h time point was 86 ± 19, 109 ± 11, and 106 ± 9% (mean ± S.E.M.). We found no thioxanthosine metabolites, showing that the HEK293 cells do not deaminate tGMP and TGrib at measurable rates under our conditions.

Thiopurine Transport from Cells Incubated with 6-Methyl-mercaptopurine Riboside.

In patients treated with 6MP, 6-methylated thiopurines are formed by TPMT. 6MP can be methylated to MeMP, which is not a substrate of HGPRT. In this case, methylation leads to reduction of the effective 6MP dose and decreased cytoxicity (Lennard et al., 1987). However, when tIMP is methylated, this may lead to a build up of MetIMP, which is an inhibitor of purine de novo synthesis and thereby contributes to cytotoxicity (Vogt et al., 1993). To examine whether MRP4 or MRP5 also transport MetIMP, we incubated cells with MeMPrib, which is converted to MetIMP by adenosine kinase. Uptake of MeMPrib was rapid and MetIMP was the only metabolite detected by HPLC analysis in the intracellular and extracellular fractions (Fig.7A). We did not detect any 6-methylated guanosines or di- or triphosphorylated methyl thiopurines. MetIMP accumulated to high concentrations in the cells (Fig. 7B) and was excreted at substantial rates (Table 3). Average MetIMP efflux rates (determined by linear regression) were 1.3, 1.9, and 3.2 pmol/10⁶ cells/min for HEK293, HEK293/5I, and HEK293/4.3 cells, respectively. The relatively high rate of MetIMP secretion by the HEK293/4.3 cells is mirrored by a decreased intracellular concentration at the 4 h time point in Fig. 7B. Recovery of MeMPrib was 77 ± 23, 85 ± 15, and 79 ± 21% (mean ± S.E.M.) for the HEK293, HEK293/4.3, and HEK293/5I cells. It is possible, therefore, that we may have missed MeMPrib metabolites. The analysis of these metabolites is complicated by their relatively low absorption maximum (Table 2), which results in interference by the standard nucleotides present in cell extracts.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Continuous incubation with MeMPrib. Thiopurine metabolites found in the extracellular (A) and intracellular (B) fractions from the HEK293 (squares), HEK293/5I (circles), and HEK293/4.3 (triangles) cells during continuous incubation with MeMPrib (initial concentration, 2.5 μM). Metabolite levels were quantified by relating the peak areas of detected metabolites with the peak area obtained with a standard amount of MeMPrib. The data are the mean ± S.E.M. of three independent determinations. The indicated data points were significantly different (*, p < 0.05) from the parental cells in a Student's t test.