Abstract
The ATP-binding cassette (ABC) transporter multidrug resistance protein 1 (MRP1/ABCC1) protects cells from arsenic (a proven human carcinogen) through the cellular efflux of arsenic triglutathione [As(GS)3] and the diglutathione conjugate of monomethylarsonous acid [MMA(GS)2]. Previously, differences in MRP1 phosphorylation (at Y920/S921) and N-glycosylation (at N19/N23) were associated with marked differences in As(GS)3 transport kinetics between HEK293 and HeLa cell lines. In the current study, cell line differences in MRP1-mediated cellular protection and transport of other arsenic metabolites were explored. MRP1 expressed in HEK293 cells reduced the toxicity of the major urinary arsenic metabolite dimethylarsinic acid (DMAV), and HEK-WT-MRP1-enriched vesicles transported DMAV with high apparent affinity and capacity (Km 0.19 µM, Vmax 342 pmol⋅mg−1protein⋅min−1). This is the first report that MRP1 is capable of exporting DMAV, critical for preventing highly toxic dimethylarsinous acid formation. In contrast, DMAV transport was not detected using HeLa-WT-MRP1 membrane vesicles. MMA(GS)2 transport by HeLa-WT-MRP1 vesicles had a greater than threefold higher Vmax compared with HEK-WT-MRP1 vesicles. Cell line differences in DMAV and MMA(GS)2 transport were not explained by differences in phosphorylation at Y920/S921. DMAV did not inhibit, whereas MMA(GS)2 was an uncompetitive inhibitor of As(GS)3 transport, suggesting that DMAV and MMA(GS)2 have nonidentical binding sites to As(GS)3 on MRP1. Efflux of different arsenic metabolites by MRP1 is likely influenced by multiple factors, including cell and tissue type. This could have implications for the impact of MRP1 on both tissue-specific susceptibility to arsenic-induced disease and tumor sensitivity to arsenic-based therapeutics.
Introduction
Arsenic is a proven human carcinogen, causing lung, skin, and bladder tumors (IARC, 2012). Chronic arsenic exposure is associated with increased incidences of kidney and liver tumors and a myriad of noncancerous adverse health effects (Platanias, 2009; IARC, 2012; Naujokas et al., 2013). Millions of people worldwide are exposed to levels of arsenic above the World Health Organization acceptable level of 10 µg/l, predominantly through the consumption of groundwater naturally contaminated with inorganic arsenic [arsenite (AsIII) and arsenate (AsV)] (Rahman et al., 2009). In addition to environmental exposures, arsenic trioxide is used clinically in the treatment of acute promyelocytic leukemia and is in clinical trials for the treatment of other hematologic and solid tumors (Kritharis et al., 2013; Ally et al., 2016; Cicconi and Lo-Coco, 2016; Falchi et al., 2016). Other arsenic compounds are also in clinical trials for the treatment of various cancers (Khairul et al., 2017). Thus, understanding the cellular handling of arsenic is critical for the development of therapeutics to treat chronic arsenic exposure and to maximize the clinical effectiveness of arsenic based drugs.
Cellular uptake of arsenic has been recently reviewed (Mukhopadhyay et al., 2014; Roggenbeck et al., 2016). Once inside most mammalian cells, arsenic undergoes extensive methylation (Vahter, 1999; Drobna et al., 2010). In humans, the four major methylation products are monomethylarsonic acid (MMAV), monomethylarsonous acid (MMAIII), dimethylarsinic acid (DMAV), and dimethylarsinous acid (DMAIII) (Thomas et al., 2007). Arsenic methylation has a significant impact on its toxicity, tissue distribution, and retention (Thomas et al., 2004, 2007; Wang et al., 2015). Although arsenic methylation results in an increased rate of arsenic whole body clearance (Drobna et al., 2009, 2010; Hughes et al., 2010) and reduces susceptibility to acute arsenic toxicity (Yokohira et al., 2010, 2011), trivalent methylated forms of arsenic (MMAIII and DMAIII) are considered bioactivation products because they are more reactive metabolites than AsIII (Petrick et al., 2000; Styblo et al., 2000; Mass et al., 2001; Kligerman et al., 2003; Moe et al., 2016).
In addition to methylation, arsenic can be conjugated with reduced glutathione (GSH/GS) (Leslie, 2012). Arsenic triglutathione [As(GS)3] and the diglutathione conjugate of the highly toxic MMAIII [MMA(GS)2] have been isolated from rat bile and mouse urine, thus these two As-GSH complexes are formed physiologically and account for a major fraction of arsenic in urine and bile (Kala et al., 2000, 2004; Suzuki et al., 2001; Cui et al., 2004; Bu et al., 2011). The ATP-binding cassette (ABC) transporter multidrug resistance protein 1 (MRP1, gene symbol ABCC1), along with the related MRP2 and MRP4 (ABCC2 and ABCC4, respectively) mediate the cellular export of multiple methylated and/or glutathionylated metabolites of arsenic (Kala et al., 2000, 2004; Leslie et al., 2004; Carew and Leslie, 2010; Carew et al., 2011; Banerjee et al., 2014; Shukalek et al., 2016).
MRP1 is a 190-kDa phosphoglycoprotein with three polytopic membrane spanning domains (MSDs) and two nucleotide binding domains (NBDs) arranged as MSD0-MSD1-NBD1-MSD2-NBD2 (Cole, 2014). MRP1 confers resistance to a chemically diverse array of anti-cancer drugs and is involved in the cellular export of physiologic compounds including GSH, glutathione disulfide, 17β-estradiol 17-(β-d-glucuronide), and leukotriene C4 (Cole, 2014). Furthermore, MRP1 transports a variety of xenobiotics often conjugated to GSH, glucuronate, or sulfate (Jedlitschky et al., 1996; Loe et al., 1996a,b; Leslie et al., 2005). Included in this list are the arsenic metabolites As(GS)3 and MMA(GS)2, the glutathionylated forms of inorganic arsenic (AsIII and AsV), and MMAIII, respectively (Leslie et al., 2004; Carew et al., 2011).
Interestingly, As(GS)3 is transported by MRP1 expressed in HEK293 cells with markedly different kinetics than by MRP1 expressed in HeLa cells (Shukalek et al., 2016). Further investigation revealed that MRP1 affinity and capacity for As(GS)3 was associated with the phosphorylation status of two residues in the linker region between NBD1 and MSD2 (Y920 and S921). Furthermore, the glycosylation status of two residues in the amino terminus (N19 and N23) influenced the stability of Y920 and/or S921 phosphorylation (Shukalek et al., 2016). Given this cell line difference, the first objective of the current study was to determine differences in arsenical cytotoxicity between the HEK-MRP1 and HeLa-MRP1 cell lines. The second objective was to use MRP1-enriched membrane vesicles isolated from HEK and HeLa cells to determine the cell line differences in MRP1 transport function. The third objective was to investigate the influence of Y920/S921 phosphorylation on the ability of HEK-MRP1-enriched membrane vesicles to transport arsenic metabolites in addition to As(GS)3.
Materials and Methods
Materials.
Carrier-free 73AsV (158 Ci/mol) was purchased from Los Alamos Meson Production Facility (Los Alamos, NM). Tris base, GSH, ATP, AMP, sucrose, DMAV (>99% purity), MgCl2, creatine kinase, GSH reductase, creatine phosphate, NADPH, sodium (meta)-arsenite [Na2AsO2] (>99% pure), AsV (>98% purity), sodium metabisulfite [Na2S2O5], and sodium thiosulfate [Na2S2O3] were from Sigma-Aldrich (Oakville, Canada). Protease inhibitor cocktail tablets (Complete, Mini, EDTA-free) and PhosSTOP phosphatase inhibitor cocktail (PIC) tablets were purchased from Roche Applied Science (Laval, Canada). Phenylmethylsulfonyl fluoride was from Bioshop Canada Inc. (Burlington, Canada). Suprapur nitric acid was purchased from Merck (Darmstadt, Germany). MMAIII and DMAIII in the form diiodomethylarsine (CH3AsI2) and iododimethylarsine ([CH3]2AsI), respectively, were synthesized as previously described (Cullen et al., 2016) and were at least 99% pure as confirmed by NMR analysis. The rat monoclonal antibody (mAb) MRPr1 was from Novus Biologicals (Littleton, CO), while rabbit anti-Na+/K+ATPase (H-300) was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA).
Cell Lines and Stable Transfection.
HEK293T cells were obtained from the American Type Culture Collection (Rockville, MD) and maintained in Dulbecco’s modified Eagle’s medium with 7.5% fetal bovine serum. HeLa and HEK293 cell lines stably expressing the empty pcDNA3.1(−) vector (HeLa-vector) and/or pcDNA3.1(−)-MRP1 (HeLa-MRP1 and HEK-MRP1) were gifts from Dr. Susan P.C. Cole (Queen’s University, Kingston, Canada) and generated as described previously (Ito et al., 2001; Conseil and Cole, 2013). The HEK-vector expressing cell line (HEK-V4) was generated as described previously (Banerjee et al., 2014). HeLa stables were maintained in Roswell Park Memorial Institute (RPMI)-1640 medium with 5% calf serum and 600 μg/ml G418 (Geneticin). HEK293 stables were maintained in Dulbecco’s modified Eagle’s medium with 7.5% fetal bovine serum and 600 μg/ml G418. Stable cell lines were checked for the proportion of cells expressing MRP1 by flow cytometry (BD FACS Calibur, Cross Cancer Institute) using the MRP1-specific MAb MRPr1, as described previously (Hipfner et al., 1994; Leslie et al., 2003). Populations of less than 80% were not used in experiments.
Cytotoxicity Testing.
The cytotoxicity of five arsenic species was measured using HEK-vector, HEK-MRP1, HeLa-vector, and HeLa-MRP1 as previously described (Carew et al., 2011). Briefly, cells were seeded in 96-well plates at 1 × 104 cells/well and grown for 24 hours. In quadruplicate, cells were treated with AsIII (0.01–300 μM), AsV (0.05–5000 µM), MMAIII (0.03–30 μM), DMAIII (0.01–300 μM), or DMAV (0.01–30 mM) for 72 hours. These doses were experimentally determined to range from nontoxic to causing complete loss of viability over 72 hours. The pH of DMAV was adjusted to pH 7.4 prior to treating cells. Cytotoxicity was determined using the tetrazolium-based MTS assay [CellTiter 96 Aqueous Non-Radioactive Cell Proliferation Assay (Promega, Madison, WI)], according to the manufacturer’s instructions. The IC50, defined as the concentration of arsenical that resulted in half the maximal toxic effect, for each arsenical was determined for HEK-MRP1 and HEK-vector cells using the sigmoidal dose-response equation in GraphPad Prism (GraphPad Software, La Jolla, CA). Relative resistance values (equivalent to relative protection), defined as the ratio of the IC50 value in HEK-MRP1 to that in HEK-vector, were determined for each arsenical tested.
Site-Directed Mutagenesis.
Mutants of MRP1 (S905A-, S915A-, S916A-, S917A-, Y920F-, S921A-, Y920F/S921A-, Y920E/S921E-, and T931A-MRP1) were generated previously (Shukalek et al., 2016). In addition, several new potential phosphorylation mutants (S916E-, S918A-, S919A-, and S919E-MRP1) were generated using the QuikChange II XL site-directed mutagenesis kit (Stratagene; Agilent Technologies, Santa Clara, CA). pcDNA3.1(−)-MRP1 was used as the PCR template, and mutagenesis was carried out according to the manufacturer’s instructions using mutagenic primers from Integrated DNA Technology (Coralville, IA); sequences are available upon request. The incorporation of desired mutations was confirmed by DNA sequencing (Molecular Biology Servicing Unit, University of Alberta, Edmonton, Canada).
Expression of Wild-Type and Mutant Forms of MRP1 in HEK293T Cells, Preparation of Membrane Vesicles, and Immunoblots.
HEK293T cells were transfected using the calcium-phosphate method as described previously (Carew and Leslie, 2010) and incubated for 48- to 72-hours posttransfection. For membrane vesicle preparations, WT and MRP1 mutant-transfected cells were collected by centrifugation, layered with Tris sucrose buffer (50 mM Tris, pH 7.4, 250 mM sucrose) containing CaCl2 (0.25 mM), EDTA-free protease inhibitors, and where indicated, phosSTOP PIC, and cell pellets stored at −80°C until plasma membrane-enriched vesicles were isolated as described previously (Carew and Leslie, 2010). Expression of WT and MRP1 mutants were confirmed by immunoblotting as described previously (Shukalek et al., 2016), using the MRPr1 antibody (1:10,000 dilution). Blots were also probed for Na+/K+-ATPase as a loading control using the Na+/K+-ATPase-specific antibody H-300 (1:10,000 dilution), except in the case of comparison of MRP1 levels between the two different cell lines (the Na+/K+-ATPase level was found to be different). Thus, blots were stained with Coomassie blue and normalized to total protein levels in each lane. Relative levels of MRP1 were quantified using ImageJ Software (National Institutes of Health, Bethesda, MD) and/or ImageLab Software (Bio-Rad, Hercules, CA).
MMA(GS)2, DMAV, and As(GS)3 Vesicular Transport Assays.
Reduction of arsenate [73AsV] into arsenite [73AsIII] and subsequent synthesis of 73As(GS)3 from 73AsIII and GSH were carried out as described previously (Reay and Asher, 1977; Shukalek et al., 2016). MMA(GS)2 was synthesized from MMAIII and GSH as described previously (Carew et al., 2011; Banerjee et al., 2014). Vesicular transport was carried out in triplicate for each substrate, employing previously described methods (Carew et al., 2011; Banerjee et al., 2014; Shukalek et al., 2016). The amount of 73As(GS)3 transported was quantified using liquid scintillation counting, as described previously (Leslie et al., 2004; Shukalek et al., 2016). The amount of MMA(GS)2 and DMAV transported was quantified by inductively coupled plasma mass spectrometry using the standard addition method, as described previously (Banerjee et al., 2014). ATP-dependent transport was calculated by subtracting transport in the presence of AMP from transport in the presence of ATP, and data are expressed as picomoles or nanomoles As(GS)3, MMA(GS)2, or DMAV transported per milligram protein per minute.
The linear range of DMAV transport was determined by incubating the HEK-WT-MRP1 or HEK-vector membrane vesicles with DMAV (1 µM) in transport buffer at 37°C for the indicated time points. Kinetic parameters were determined by measuring the initial rate of DMAV and MMA(GS)2 transport at eight different substrate concentrations [0.1–2.5 μM for DMAV and 1–200 μM for MMA(GS)2] at 20 seconds and 1 minute, respectively. Curve fitting was done by nonlinear regression analysis using GraphPad Prism 6 software.
The ability of DMAV and MMA(GS)2 to inhibit 73As(GS)3 transport was first characterized by using a fixed concentration of 73As(GS)3 (1 µM, 40 nCi) in the presence of increasing concentrations of DMAV (0.04 and 1000 µM) or MMA(GS)2 (0.01–300 µM). The conditions for the synthesis of As(GS)3 resulted in the presence of 3 mM GSH in transport reactions (Leslie et al., 2004). Thus, As(GS)3 inhibition experiments were completed under plus GSH conditions. This is important because the in vitro inhibition of MRP1 by certain compounds can be enhanced by physiologic concentrations of GSH (Cole and Deeley, 2006). IC50 values were calculated for MMA(GS)2 inhibition using GraphPad Prism 6 Software [nonlinear regression log(inhibitor) vs. response variable slope (4 parameters)]. To determine the mode of inhibition of 73As(GS)3 by MMA(GS)2, the Ki of MMA(GS)2 was determined by performing 73As(GS)3 (0.1–20 µM, 40–100 nCi) transport in the presence of three different MMA(GS)2 concentrations (5, 10, and 15 µM), as described previously (Leslie et al., 2004).
All transport data were normalized as needed to correct for any difference in level of HeLa-WT-MRP1 or mutant MRP1 expressed in HEK cells relative to HEK-WT-MRP1 as determined by immunoblotting of each membrane vesicle preparation. Positive control transport experiments using MMA(GS)2 or As(GS)3 were run for each vesicle preparation, as described above and previously (Leslie et al., 2004; Carew et al., 2011).
Results
MRP1 Expressed in HEK293 Cells Decreases the Cytotoxicity of AsIII, AsV, MMAIII, and DMAV.
Previously we showed that MRP1 stably expressed in HeLa cells reduced the toxicity of AsIII, AsV, and MMAIII, but not MMAV, DMAIII, or DMAV, relative to HeLa cells expressing empty vector alone (Carew et al., 2011). Given that we recently reported substantial differences in the transport of As(GS)3 between MRP1-enriched membrane vesicles isolated from HEK293 and HeLa cells (Shukalek et al., 2016), differences in arsenical cytotoxicity between the two cell lines were investigated. Thus, the cytotoxicity of five of these arsenicals (AsIII, AsV, MMAIII, DMAIII, and DMAV) in HEK-MRP1 and HEK-vector cell lines was determined in parallel with HeLa-MRP1 and HeLa-vector for comparison [Fig. 1; Table 1, and (Carew et al., 2011)].
The arsenic compounds that HEK-MRP1 cells conferred resistance to (relative to HEK-vector) were the same as the HeLa cell line pair (Fig. 1, A–C; Table 1), except that HEK-MRP1 conferred a significantly increased level of resistance against DMAV (relative resistance value of 1.4, P < 0.05) [Fig. 1D; Table 1, and (Carew et al., 2011)]. In addition, HEK-MRP1 cells had a relative resistance of 9.2 against AsV (Fig. 1B; Table 1), whereas HeLa-MRP1 cells had a relative resistance of only 2.1 [Table 1 and (Carew et al., 2011)]. These results led us to characterize the cell line differences for DMAV and AsV transport using MRP1-enriched membrane vesicles. Furthermore, although resistance to MMAIII was conferred by both HEK-MRP1 and HeLa-MRP1 cell lines [Fig. 1C; Table 1 and (Carew et al., 2011)], differences in transport characteristics between membrane vesicles prepared from these cell lines were also investigated for MMA(GS)2.
Transport of DMAV by MRP1-Enriched Membrane Vesicles.
To test if differences in cytotoxicity were due at least in part to differences in ATP-dependent cellular efflux, DMAV transport by HEK- and HeLa-WT-MRP1 membrane vesicles was measured at 0.05 and 1 μM of DMAV (Fig. 2A). MRP1-enriched vesicles prepared from transiently transfected HEK293T cells transported DMAV (101 ± 15 pmol⋅ mg−1 protein⋅min−1 at 0.05 μM and 264 ± 42 pmol⋅mg−1 protein⋅min−1 at 1 μM). In contrast, DMAV transport was not detected for HeLa-WT-MRP1 vesicles (Fig. 2A), despite the fact that the same vesicle preparations were functional for MMA(GS)2 transport (Fig. 4A), and comparable levels of MRP1 were present (Supplemental Fig. 1). For certain compounds transported by MRP1, GSH can either be required for or enhance transport (Cole, 2014). To determine whether this was the case for DMAV, transport assays were completed in the presence of 3 mM GSH and this had no effect on HEK- or HeLa-WT-MRP1 DMAV transport activity (data not shown). Thus, we report for the first time that MRP1 is capable of transporting this important methylated arsenic metabolite, at least under certain conditions.
Kinetic Analysis of MRP1-Mediated DMAV Transport.
Transport of DMAV by HEK-WT-MRP1 membrane vesicles was then further characterized. The linear range of DMAV (1 µM) transport versus time was measured for HEK-WT-MRP1 and HEK-vector control membrane vesicles (Fig. 2B). Transport by HEK-WT-MRP1 was linear for up to 30 seconds and reached a maximal activity of 138 ± 24 pmol⋅mg−1 protein at 30 seconds. ATP-dependent transport of DMAV by HEK-vector was very low and similar to transport observed in the presence of AMP. HEK-WT-MRP1 membrane vesicle transport of DMAV was characterized kinetically by determining the initial rates of transport over several concentrations of DMAV (Fig. 2C; Table 2). HEK-MRP1 was found to transport DMAV with high apparent affinity and capacity (Km of 0.19 ± 0.06 µM and Vmax of 342 ± 37 pmol⋅ mg−1 protein⋅min−1).
Analysis of DMAV Transport by HEK-MRP1 Phosphorylation Mutants.
Due to the complete lack of DMAV transport by HeLa-WT-MRP1 and substantial transport by HEK-WT-MRP1 membrane vesicles, this cell line difference was further explored. We previously showed that HEK293 and HeLa cell line differences in As(GS)3 transport by MRP1 were associated with differences in phosphorylation at Y920/S921 (Shukalek et al., 2016). To determine if phosphorylation differences at these sites were responsible for the difference in DMAV transport by MRP1 between the two cell lines, DMAV transport by MRP1-Y920/S921 dephosphorylation- and phosphorylation-mimicking mutants was investigated.
Dephosphorylation-mimicking HEK-Y920F/S921A-MRP1 membrane vesicles exhibited a complete loss of DMAV transport (Fig. 3A), suggesting these sites are critical for the cell line difference. However, the phosphorylation-mimicking HEK-Y920E/S921E-MRP1, which we expected would restore DMAV transport, also completely lacked DMAV transport (Fig. 3A). Furthermore, individual dephosphorylation mimicking mutants Y920F-MRP1 and S921A-MRP1 also completely lost function (data not shown). Mutant membrane vesicle preparations had MRP1 levels similar to HEK-WT-MRP1 (Supplemental Fig. 1) and were functional for MMA(GS)2 (Fig. 4B; Table 2) and/or As(GS)3 (Supplemental Fig. 2A) transport.
Analysis of DMAV Transport by HeLa-WT-MRP1 in the Presence of Phosphatase Inhibitors.
The possible modulation of DMAV transport by MRP1 through phosphorylation was also investigated by studying the transport in MRP1-enriched vesicles prepared from HeLa cells in the presence or absence of a phosphatase inhibitor cocktail (PIC) at two different substrate concentrations (0.05 and 1 µM) (Fig. 3B). We previously showed for As(GS)3 that HeLa-WT-MRP1 membrane vesicles prepared in the presence of a PIC have a 19- and 12-fold increase in Km and Vmax, respectively, compared with HeLa-WT-MRP1 prepared in the absence of a PIC (Shukalek et al., 2016). In contrast, the inclusion of a PIC did not influence DMAV transport by HeLa-WT-MRP1 at either concentration (Fig. 3B). To ensure that the HeLa-WT-MRP1 ± PIC membrane vesicles used in the DMAV transport assays were functional, and the PIC active, As(GS)3 transport experiments were completed on the same vesicle preparations (Fig. 3C). Consistent with our previous report (Shukalek et al., 2016), As(GS)3 transport by HeLa-WT-MRP1 vesicles, prepared in the presence of a PIC, was significantly higher than for HeLa-WT-MRP1 vesicles prepared in the absence of a PIC. These results could suggest that mechanisms other than phosphorylation are likely responsible for the cell line difference observed in DMAV transport by HEK-WT-MRP1 compared with HeLa-WT-MRP1. Alternatively, a differential phosphorylation site that remains stable in the absence of PIC could be involved.
The lack of DMAV transport by all mutants of Y920/S921 could also be due to the disruption of phosphorylation/dephosphorylation of neighboring sites that are critical for DMAV transport. We screened the influence of seven additional putative phosphorylation sites in the linker region (S905-, S915-, S916-, S917-, S918-, S919-, T931-MRP1) by mutating them to phospho-mimicking and/or dephospho-mimicking residues in full-length MRP1, expressing them in HEK293 cells, preparing membrane vesicles, and measuring DMAV transport. Surprisingly, all mutants lacked DMAV transport activity, despite the fact that they were comparable to HEK-WT-MRP1 for As(GS)3 transport and MRP1 level (Supplemental Fig. 2).
Inhibition of As(GS)3 Transport by DMAV.
To further characterize the differences in interaction of As(GS)3 and DMAV with MRP1, transport of As(GS)3 (1 µM) by HEK-WT-MRP1 vesicles was measured in the presence of DMAV. DMAV did not inhibit As(GS)3 transport even at a concentration of 1 mM (data not shown), four orders of magnitude greater than the apparent Km of DMAV for MRP1 (Table 2).
MRP1 Does Not Transport AsV.
Given the striking difference in AsV relative resistance levels between the HeLa and HEK293 cell line pairs [Fig. 1B; Table 1 and (Carew et al., 2011)], the ability of HEK-MRP1 and HeLa-MRP1 membrane vesicles to transport AsV was investigated. AsV (1 and 10 μM) was not transported by MRP1-enriched membrane vesicles prepared from HEK293 cells in the presence or absence of 3 mM GSH (data not shown). Similar results were obtained for membrane vesicles prepared from HeLa-MRP1 cells (data not shown). These results are consistent with a lack of detectable AsV transport (in the presence and absence of GSH) by MRP1-enriched vesicles prepared from the H69AR small cell lung cancer cell line and our previous conclusion that inorganic arsenic (AsIII and AsV) is metabolized to As(GS)3 before transport by MRP1 (Leslie et al., 2004).
Transport of MMA(GS)2 by MRP1-Enriched Membrane Vesicles.
We previously showed that MMAIII is transported by HEK-MRP1 in the form MMA(GS)2 and that HeLa-MRP1 cells confer protection against MMAIII relative to HeLa-vector cells (Carew et al., 2011). In the current study, HEK-MRP1 cells were also found to confer protection against MMAIII relative to HEK-vector (Table 1). To determine if differences in transport characteristics existed between cell lines, MMA(GS)2 transport by HEK- and HeLa-WT-MRP1-enriched vesicles was compared. Under the conditions tested, MMA(GS)2 (1 μM) transport by HeLa-WT-MRP1 vesicles was 1.6-fold higher (P < 0.01) than HEK-WT-MRP1 vesicles (Fig. 4A).
Kinetic Analysis of MRP1-Mediated MMA(GS)2 Transport.
To determine if the increased transport of MMA(GS)2 was due to changes in Km and/or Vmax, MRP1-mediated transport of MMA(GS)2 was measured at an initial rate over eight different concentrations of MMA(GS)2 (Fig. 4B; Table 2). HEK- and HeLa-WT-MRP1 membrane vesicles were found to have similar apparent affinity for MMA(GS)2 (Km of 23 ± 2.2 and 33 ± 24 µM, respectively), similar to what we previously reported for HEK-WT-MRP1 (Carew et al., 2011). In contrast, the Vmax was 3.4-fold higher for MMA(GS)2 transport by HeLa-WT-MRP1 (Vmax of 17 ± 5.3 nmol⋅mg−1⋅min−1) than HEK-WT-MRP1 (Vmax of 4.9 ± 0.5 nmol⋅mg−1⋅min−1) membrane vesicles (Fig. 4B; Table 2). In addition, kinetic characterization of MMA(GS)2 transport by the double dephosphorylation-mimicking mutant HEK-Y920F/S921A-MRP1 previously demonstrated to have substantially reduced apparent Km and Vmax values for As(GS)3 relative to HEK-WT-MRP1(Shukalek et al., 2016) was completed. Interestingly, HEK-Y920F/S921A-MRP1 had very similar apparent affinity (Km of 24 ± 5.5 µM) and capacity (Vmax of 5.3 ± 2.2 nmol⋅mg−1⋅min−1) to that of HEK-WT-MRP1 (Fig. 4B; Table 2). Consistent with this, the phosphorylation-mimicking mutant HEK-Y920E/S921E-MRP1 also had similar apparent affinity and capacity to HEK-WT-MRP1 (mean Km of 24 µM, Vmax of 5.5 nmol⋅mg−1 protein⋅min−1, n = 2) (Table 2). These data suggest that, in contrast with As(GS)3 transport, phosphorylation of Y920/S921 has little influence on the transport of MMA(GS)2 by MRP1.
Inhibition of As(GS)3 Transport by MMA(GS)2.
Although the Km of As(GS)3 for MRP1 was >10-fold higher for HEK-WT-MRP1 compared with HeLa-WT-MRP1 and HEK-Y920F/S921A-MRP1 (Shukalek et al., 2016), no difference between the apparent Km value for MMA(GS)2 transport was observed for these membrane vesicles (Fig. 4; Table 2). This result suggested that MMA(GS)2 and As(GS)3 interact at nonidentical binding sites. To begin to characterize the differences in interaction of As(GS)3 and MMA(GS)2 with MRP1, transport of As(GS)3 (1 µM) by HEK-WT-MRP1 vesicles was measured in the presence of increasing concentrations of MMA(GS)2 (Fig. 5A). MMA(GS)2 was found to potently inhibit As(GS)3 transport with an IC50 value of 11 ± 1.5 µM.
The inhibition of As(GS)3 transport by MMA(GS)2 was further characterized by measuring the effect of MMA(GS)2 (5, 10, and 15 µM) on As(GS)3 (0.1–20 µM) transport (15 µM shown in Fig. 5B). Michaelis-Menten analysis showed that MMA(GS)2 at each concentration tested reduced both the apparent Km and Vmax values, suggesting an uncompetitive mode of inhibition with an average Ki of 7.3 ± 5.1 µM (±S.D., n = 3). These data are consistent with MMA(GS)2 and As(GS)3 interacting at nonidentical binding sites.
Discussion
The proven human carcinogen arsenic naturally contaminates the drinking water of hundreds of millions of people worldwide. One of the most affected countries is Bangladesh, where the arsenic contamination has been referred to as “the largest mass poisoning of a population in history” (Smith et al., 2000). Understanding the cellular handling of arsenic, including efflux pathways, is critical for the prevention and treatment of arsenic-induced disease.
We investigated the ability of MRP1 to confer resistance to and/or transport important methylated arsenic metabolites when expressed in HEK293 cells compared with HeLa cells. The cellular resistance conferred by MRP1 against different arsenic species is useful information; however, resistance levels can be influenced by cellular metabolism and uptake efficiency. To draw conclusions about MRP1-mediated transport of specific arsenic compounds, it was critical to measure their transport directly using MRP1-enriched membrane vesicles. The population of membrane vesicles that is accumulating MRP1 substrates are inside-out, allowing the measurement of ATP-dependent transport with minimal influence of metabolism and cellular uptake. This allows the MRP1 contribution to cellular export of specific arsenic compounds to be evaluated and allows the accurate determination of kinetic parameters.
The most pronounced difference between cell lines was for DMAV, which HEK-MRP1 cells conferred resistance to and HEK-WT-MRP1 membrane vesicles transported with high apparent affinity and capacity (Fig. 2; Table 1). This is the first report that MRP1 is capable of transporting this important arsenic metabolite. In contrast, HeLa-MRP1 cells did not confer resistance to DMAV relative to HeLa-vector cells (Carew et al., 2011), and HeLa-WT-MRP1 membrane vesicles did not have detectable DMAV transport activity (Fig. 2A).
The relative resistance conferred by MRP1 expressed in HEK293 cells was small, but significant (1.4-fold, P < 0.05, Fig. 1C; Table 1). This marginal resistance was in contrast with the high-affinity and high-capacity transport of DMAV observed with HEK-WT-MRP1-enriched membrane vesicles (Fig. 2C; Table 2). A likely explanation for the difference in results between the two assays is that DMAV is poorly taken up by cells (Delnomdedieu et al., 1995; Dopp et al., 2004, 2005; Naranmandura et al., 2007, 2011), including HEK293 cells (Banerjee et al., 2014). Thus, it is likely that the data generated with the inside-out MRP1-enriched membrane vesicles, with no requirement for cell entry, more accurately reflects what is occurring after formation of DMAV within the cell. Humans are predominantly exposed to AsIII and AsV in drinking water, which are taken up by cells efficiently (Roggenbeck et al., 2016) and then converted to methylated products (e.g., DMAV).
As a starting point for determining a mechanism for cell line differences in MRP1-mediated DMAV transport, we investigated the potential contribution of differential phosphorylation. Mutation of two phosphorylation sites (Y920/S921-MRP1), previously reported as responsible for cell line differences in the transport of As(GS)3, to both phospho- or dephospho-mimicking amino acids, surprisingly resulted in a complete loss of DMAV transport. Individual mutant HEK-Y920F-MRP1 and HEK-S921A-MRP1 membrane vesicles, shown previously to transport As(GS)3 to a similar extent as HEK-WT-MRP1, also completely lacked DMAV transport. The inclusion of a PIC during the preparation of HeLa-WT-MRP1 membrane vesicles did not result in a gain of DMAV transport activity (although MRP1-mediated As(GS)3 transport was increased), suggesting that either phosphorylation was not important or a stable phosphorylation site (not influenced by the PIC) was involved. Mutation of multiple other putative phosphorylation sites in the linker region also resulted in a complete loss of HEK-MRP1 DMAV transport. Our data suggest that DMAV transport by HEK-WT-MRP1 membrane vesicles is extremely sensitive to alterations in the linker region between NBD1 and MSD2. The reasons for this are currently not understood and require further investigation. Differences in posttranslational modifications and/or protein:protein interactions that alter the structure of this region could potentially explain the cell line differences in MRP1-mediated DMAV transport. Indeed, there is some suggestion in the literature that this linker region is important for protein:protein interactions and that such interactions may be modulated by phosphorylation (Yang et al., 2012; Ambadipudi and Georges, 2017).
MRP1 transport of DMAV is the second DMAV efflux pathway to be identified. Previously, we reported that the related MRP4 transports DMAV with similar affinity (K0.5 0.22 ± 0.15 µM for MRP4 vs. Km 0.19 ± 0.06 µM for MRP1), but through a cooperative mechanism (Hill coefficient 2.9 ± 1.2) and assuming equal protein levels, lower capacity (Vmax 32 ± 3 pmol⋅mg−1 protein⋅min−1 for MRP4 vs. Vmax 342 ± 37 pmol⋅mg−1 protein⋅min−1 for MRP1) [(Banerjee et al., 2014, 2016) and Table 2]. The tissue expression and cellular localization of MRP4 likely make it critical for urinary elimination of hepatic metabolites (Banerjee et al., 2014). The localization of MRP1 to the basolateral surface of epithelial cells and expression in specific cell types of most tissues (undetectable protein levels in human hepatocytes), likely makes MRP1 important for cellular/tissue protection rather than playing a role in arsenic elimination.
DMAV is an arsenic compound with low toxicity relative to trivalent arsenic species (Moe et al., 2016). The efflux of DMAV from the cell is critical to prevent the reduction of DMAV to the highly reactive DMAIII (Nemeti and Gregus, 2013). Furthermore, export of DMAV would likely prevent product inhibition of arsenic (+3 oxidation state) methyltransferase, allowing the formation and cellular export of more DMAV. The reducing intracellular environment might suggest that DMAIII is the predominant form of dimethylated arsenic within the cell; however, this has been difficult to prove and DMAV has been detected in human cell lines and mouse liver homogenate (Currier et al., 2011). The highly reactive DMAIII is highly protein bound and unlikely to be available for cellular export (Hippler et al., 2011; Shen et al., 2013). An equilibrium between DMAIII and DMAV will exist within the cell, and the high-affinity high-capacity export of DMAV by MRP1 would provide a good mechanism for cellular detoxification.
Out of the five arsenic compounds tested HEK-MRP1 conferred the highest level of resistance against AsV (9-fold) followed by AsIII (5-fold). Consistent with previous studies using MRP1-enriched membrane vesicles isolated from H69AR cells (Leslie et al., 2004), HEK- and HeLa MRP1-enriched vesicles did not transport AsV in the presence or absence of GSH. Once inside the cell, AsV is reduced to AsIII and then enters the methylation pathway (Cullen, 2014). Thus, our results are consistent with AsV and AsIII being converted to As(GS)3 before efflux by MRP1, as we showed previously (Leslie et al., 2004). The reason why HEK-MRP1 cells confer higher levels of resistance to AsV than AsIII is not understood. AsV enters cells more slowly (through Na+-dependent phosphate transporters) than AsIII (through aquaglyceroporins) (Mukhopadhyay et al., 2014; Roggenbeck et al., 2016), and this could influence the methylation and glutathionylation of arsenic and alter the metabolites available for MRP1 export.
HEK-MRP1 and HeLa-MRP1 cell lines both conferred significantly higher levels of resistance to MMAIII than their respective vector controls [(Carew et al., 2011) and Table 1]. MMA(GS)2 was transported with comparable apparent affinity by HeLa-WT-MRP1 and HEK-WT-MRP1 membrane vesicles; however, the Vmax was 3.4-fold higher for HeLa-MRP1 membrane vesicles. Kinetic parameters for MMA(GS)2 transport were not significantly different between HEK-WT-MRP1 and HEK-Y920F/S921A-MRP1 or HEK-Y920E/S921E-MRP1, suggesting that differential phosphorylation at these sites is not responsible for the cell line differences in Vmax. Consistent with these phosphorylation sites being important for the interaction between MRP1 and As(GS)3, but not MRP1 and MMA(GS)2, we found that MMA(GS)2 was an uncompetitive inhibitor of As(GS)3 transport. Thus, increasing concentrations of As(GS)3 did not overcome MMA(GS)2 inhibition (Fig. 5B), providing support for the idea that As(GS)3 and MMA(GS)2 do not share identical binding sites.
Arsenic was previously reported to activate kinase and inhibit phosphatase pathways (Rehman et al., 2012; Beauchamp et al., 2015), and we had postulated cellular exposure to arsenic would result in a shift to a prophosphorylation state of Y920/S921-MRP1 (Shukalek et al., 2016). This in turn would result in the switch of MRP1 from a high-affinity, low-capacity transporter of As(GS)3 to a more efficient low-affinity, high-capacity As(GS)3 transporter (Shukalek et al., 2016). Phosphorylation of these residues appear to be important specifically for As(GS)3, but not for MMA(GS)2 or DMAV (this study) or as previously reported for methotrexate, leukotriene C4, or 17β-estradiol 17-(β-d-glucuronide) (Loe et al., 1996b; Stride et al., 1997; Shukalek et al., 2016). Why phosphorylation of MRP1 at Y920/S921 has an impact on As(GS)3, but not other arsenic metabolites is unknown. Potentially, MRP1 exports As(GS)3 over a broad concentration range to reduce AsIII availability for the formation of more toxic trivalent methylated forms. The Km values for As(GS)3 (Km range ∼0.3–4 µM) (Leslie et al., 2004; Shukalek et al., 2016) and DMAV (Km 0.19 µM) (Table 2) are much lower than MMA(GS)2 (Km range 11–33 µM) [Table 2 and (Carew et al., 2011)]. At low levels of arsenic exposure, MRP1 is potentially important for the export of As(GS)3, and any DMAV that is formed (preventing the formation of the highly toxic DMAIII). During higher cellular arsenic exposure, MRP1 phosphorylation allows it to still export As(GS)3 efficiently; MMA(GS)2 accumulation might start to occur and MRP1 would be able to export this, and potentially DMAV. It is worth noting that the transport of DMAV by MRP1 is remarkably more efficient than reported for any other transporter and arsenical combination (Roggenbeck et al., 2016), providing support for MRP1 being an important transport pathway for DMAV under environmentally relevant exposure conditions.
Differences in MRP1 transport of DMAV (this study) and As(GS)3 (Shukalek et al., 2016) by membrane vesicles isolated from different cells raises the possibility that MRP1 could have a distinct role in arsenic efflux, depending in which tissue and/or cell type it is expressed. MRP1 has been indirectly implicated in the protection of specific tissues from arsenic toxicity, including kidney and brain (Kimura et al., 2005, 2006; Dringen et al., 2016; Wang et al., 2016). Furthermore, MRP1 could play a role in resistance to arsenic-based therapies and this could be modified depending upon the tumor type. This study lays the groundwork for further investigation into how the cellular environment influences the function of MRP1, particularly for the cellular detoxification of important arsenic metabolites.
Acknowledgments
Diane Swanlund is thanked for outstanding technical assistance. Xiufen Lu is gratefully acknowledged for assistance with inductively coupled plasma mass spectrometry. Dr. Susan P.C. Cole (Queen’s University) is thanked for providing the HEK/HeLa-MRP1 and HeLa-vector stable cell lines and the pcDNA3.1(−)MRP1. The 73-arsenic used in this research was supplied by the United States Department of Energy Office of Science by the Isotope Program in the Office of Nuclear Physics.
Authorship Contributions
Participated in research design: Banerjee, Kaur, Whitlock, Carew, Le, Leslie.
Conducted experiments: Banerjee, Kaur, Whitlock, Carew.
Contributed new reagents or analytic tools: Le.
Performed data analysis: Banerjee, Kaur, Whitlock, Carew, Leslie.
Wrote or contributed to the writing of the manuscript: Banerjee, Kaur, Whitlock, Leslie.
Footnotes
- Received November 21, 2017.
- Accepted May 9, 2018.
↵1 Current affiliation: Department of Pharmacology and Toxicology, University of Louisville, Kentucky.
This work was supported by the Canadian Institutes of Health Research [CIHR Grant MOP-272075]. M.B. was supported by an Alberta Cancer Foundation Cancer Research Postdoctoral Fellowship award. X.C.L. holds the Canada Research Chair in Bioanalytical Technology and Environmental Health. G.K. holds a Queen Elizabeth II doctoral scholarship and a Medical Sciences Graduate Program Scholarship from the University of Alberta. E.M.L. was an Alberta Innovates Health Solutions Scholar and a CIHR New Investigator.
↵This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- ABC
- ATP-binding cassette
- AsIII
- arsenite
- AsV
- arsenate
- As(GS)3
- arsenic triglutathione
- DMAIII
- dimethylarsinous acid
- DMAV
- dimethylarsinic acid
- GSH
- glutathione
- HEK293
- human embryonic kidney 293
- IC50
- half-maximal inhibitory concentration
- MMAIII
- monomethylarsonous acid
- MMAV
- monomethylarsonic acid
- MMA(GS)2
- monomethylarsenic diglutathione
- MRP
- multidrug resistance protein
- MSD
- membrane spanning domain
- NBD
- nucleotide binding domain
- PIC
- phosphatase inhibitor cocktail
- WT
- wild type
- Copyright © 2018 by The American Society for Pharmacology and Experimental Therapeutics