Abstract
ABCG2 (also known as breast cancer resistance protein) is an ATP-binding cassette (ABC) transporter localized to the plasma membrane where it mediates the efflux of xenobiotics, including potential therapeutics. Studies investigating Abcg2 function at the blood-brain barrier in mouse models are often compared with human ABCG2 function. It is critical to understand the nature of species differences between mouse and human ABCG2, since extrapolations are made from murine data to humans. Two independent drug-selected cell line pairs expressing human or mouse ABCG2 were compared for efflux of fluorescent substrates using flow cytometry. To this end, we developed and characterized a new mouse Abcg2-expressing subline that demonstrated efflux of known fluorescent ABCG2 substrates and increased resistance to mitoxantrone, which is reduced in the presence of the ABCG2 inhibitor Ko143. Our results indicate that the substrate specificity of human and mouse ABCG2 is very similar. We identified a new human and mouse ABCG2 substrate, a porphyrin analog, purpurin-18 (Pp-18), which is not a substrate for P-glycoprotein or multidrug resistance protein 1. The ability of inhibitors to block efflux activity of ABCG2 was assessed using Pp-18. Inhibitors also demonstrated similar effects on human and mouse ABCG2. Chrysin, benzoflavone, and cyclosporin A inhibited Pp-18 efflux in both human and mouse ABCG2. The similarity of the substrate and inhibitor specificity of human and mouse ABCG2 supports interpretation of mouse models in understanding the clinical, pharmacological, and physiologic roles of ABCG2.
Introduction
ABCG2 (also known as breast cancer resistance protein) is an ATP-binding cassette (ABC) transporter localized to the plasma membrane that actively pumps a wide variety of molecules out of cells. ABCG2 was first identified in multidrug-resistant cancer cell lines (Polgar et al., 2008), but it also plays a protective role at the maternal-fetal, blood-testis, and blood-brain barriers by preventing the entry of small molecules (Kannan et al., 2009; Robey et al., 2009). It facilitates absorption in the liver and limits absorption from the small intestine, and it is expressed in the mammary glands, where it is responsible for active secretion of substrates into milk (Jonker et al., 2005). ABCG2 function is responsible for the “side-population” phenomenon observed in flow cytometry due to Hoechst 33342 efflux by hematopoietic cells with stem cell–like characteristics (Natarajan et al., 2012). When Abcg2 is absent in transgenic mice, the primary phenotype is photosensitivity, with phototoxic lesions due to the accumulation of endogenous porphyrin metabolites that would otherwise be excreted (Jonker et al., 2002), and it was recently shown that Abcg2 mediates the transport of sulfate conjugates of phytoestrogens (van de Wetering and Sapthu, 2012). Polymorphic forms of ABCG2 are associated with gout because of decreased excretion of uric acid in the proximal tubule of the kidney (Woodward et al., 2009). It was recently recognized that healthy individuals of the Jr(a−) blood type carry two null alleles of ABCG2 despite the important physiologic role understood for ABCG2 (Saison et al., 2012; Zelinski et al., 2012). The functional consequences of ABCG2 loss are still unknown but are not associated with obvious disease. In light of the roles of ABCG2 in drug resistance and normal physiology, further research on models aimed at understanding ABCG2 function is warranted.
Cell lines expressing human ABCG2 are commonly used to study its function as an efflux pump or to screen for novel inhibitors (Deeken et al., 2009). Fluorescent substrates have proven to be useful tools for measuring transporter function, and in vitro studies using these substrates are sometimes reported alongside pharmacokinetic assessments in Abcg2 knockout mice to study ABCG2 function in vivo (Kannan et al., 2010; Hall and Pike, 2011; Mairinger et al., 2011). Abcg2-deficient mice have also been instrumental in elucidating the normal physiologic roles of ABCG2, such as its role in preventing oral drug absorption or brain penetration of substrates (Vlaming et al., 2009). The mouse ortholog of ABCG2 has 81% protein sequence homology with human ABCG2 (Allen et al., 1999), and a single amino acid mutation can alter the substrate and antagonist specificity in both species (Robey et al., 2003), leading to the assumption that substrate and inhibitor profiles of human and murine ABCG2 are directly comparable. However, in the case of P-glycoprotein (P-gp, ABCB1), significantly different substrate and inhibitor specificities have been observed between species (Pike, 2009; Syvanen et al., 2009). Baltes et al. (2007) demonstrated that phenytoin and levetiracetam were transported by mouse but not human P-gp. Differences in inhibitor efficacy have also been suggested for human and mouse ABCG2 (Zhang et al., 2005). This suggests caution when extrapolating data from mouse models to humans. The comparative specificity of substrates and inhibitors in human versus mouse ABCG2 is still unclear and warrants further investigation.
In this study, the abilities of cell lines expressing human or mouse ABCG2 to efflux fluorescent substrates were compared using flow cytometry. The genomic sequence and protein expression of ABCG2 was assessed in the cell lines. A new fluorescent porphyrin substrate, purpurin-18 (Pp-18) was identified, and demonstrated to be transported by mouse and human ABCG2, but not P-gp or multidrug resistance protein 1 (MRP1). The ability of inhibitors to block efflux activity of human and mouse ABCG2 was assessed using Pp-18 as a substrate.
Materials and Methods
Chemicals.
Pheophorbide a, pyropheophorbide a methyl ester (MPPa), chlorin e6, and Pp-18 were obtained from Frontier Scientific (Logan, UT). Flavopiridol was obtained from the National Cancer Institute In Vitro Anticancer Drug Discovery Screen (Bethesda, MD). BODIPY-prazosin, a conjugate of the ABCG2 substrate with a boron-dipyrromethene fluorescent dye, was obtained from Molecular Probes (Eugene, OR). Ko143 was purchased from Tocris Bioscience (Minneapolis, MN). Nilotinib (AMN107) was purchased from Novartis (Basel, Switzerland). 1-Isatin-4-(4′-methoxyphenyl)thiosemicarbazone (NSC73306) was synthesized as previously reported (Hall et al., 2011). Benzoflavone was purchased from Indofine (Hillsborough, NJ). Vismodegib was purchased from Selleck Chemicals (Houston, TX). Elacridar (GF120918) and tariquidar (XR9576) were purchased from MedKoo Biosciences (Chapel Hill, NC). DCPQ [(2R)-anti-5-{3-[4-(10,11-dichloromethanodibenzo-suber-5-yl)piperazin-1-yl]-2-hydroxypropoxy}quinoline trihydrochloride] was provided by Dr. Victor W. Pike, National Institute of Mental Health (Bethesda, MD). Daunorubicin (daunomycin), rhodamine 123, novobiocin (albamycin), and all other chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless stated otherwise.
Cell Lines.
The parental (control) and resistant (ATP-binding cassette transporter-expressing) cell lines used in this study were (drug selection shown in parentheses): 3T3 and its Abcg2-expressing subline 3T3 MX15 (15 nM mitoxantrone), and its P-gp–expressing subline 3T3 C3M (1 μg/ml colchicine) (Hall et al., 2011); the mouse fibroblast line Ltk-, and its Abcg2-expressing subline Ltk-HoeR415 (1 μM Hoechst 33342) (Smith et al., 2009); the mouse fibroblast MEF3.8, and its Abcg2-expressing subline MEF3.8 M32 (32 nM mitoxantrone) (Allen et al., 1999); the human large-cell lung cancer cell line H460, and its ABCG2-expressing subline H460 MX20 (20 nM mitoxantrone) (Robey et al., 2004); the human breast cancer cell line MCF-7, its ABCG2-expressing subline MCF-7 FLV500 (500 nM flavopiridol) (Robey et al., 2001b) and its MRP1-expressing subline MCF-7 VP-16 (16 nM etoposide) (Schneider et al., 1994); and the human adenocarcinoma cell line KB-3-1, and its P-gp–expressing subline KB-8-5-11 (250 nM colchicine) (Shen et al., 1986). All cells were grown at 37°C in 5% CO2 and were maintained either in RPMI or Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, penicillin, and streptomycin. Medium was removed and cells were grown in the same medium in the absence of drug selection 5–30 days before assay.
The mouse fibroblast 3T3 MX15 subline was selected by exposing NIH 3T3 cells to increasing concentrations of mitoxantrone (1–15 nM). The Ltk- mouse cell line was developed from a subline of the BrdU-resistant strain of the L-M mouse line (Kit et al., 1963). The Ltk- drug-selected subline (Ltk-HoeR415) has been shown to have increased Abcg2 expression and function (Smith et al., 1988, 2009). The Ltk- and Ltk-HoeR415 were provided by Paul J. Smith (Cardiff University). The MEF3.8 cell line, derived from Mdr1a/b−/− Mrp1−/− mice, and the MEF3.8 M32 subline were provided by Alfred H. Schinkel (Netherlands Cancer Institute). The human ABCG2-expressing sublines used (H460 MX20 and MCF7 FLV500) have been previously characterized (Robey et al., 2001b, 2004).
Simple Western.
An automated capillary-based Simple Western system (Simon; ProteinSimple, Santa Clara, CA) was used to quantify the levels of ABCG2 and P-gp. Automation allows more accurate and reproducible assessment of protein levels compared with traditional Western blots (Chen et al., 2013). Samples were analyzed as previously described unless stated otherwise (Kohn et al., 2012). In brief, samples containing the same amount of protein, 57 ng (mouse) or 20 ng (human) for ABCG2 analysis and 20 ng (human and mouse) for P-gp analysis, were used. The target proteins were immunoprobed for ABCG2 using monoclonal BXP-53 antibody (Abcam) and for P-gp expression using monoclonal C219 antibody (Fujirebio Diagnostics, Inc., Malvern, PA) and for tubulin using anti–α-tubulin (Cell Signaling Technologies, Danvers, MA). For quantitative data analysis, the ABCG2 and P-gp signals were normalized against α-tubulin as a loading control.
Cytotoxicity Assay.
Cytotoxicity was measured with a luminescent cell viability assay (CellTiter Glo; Promega, Madison, WI). Stock solutions of cytotoxic drugs were prepared in RPMI medium. Cultured cells were seeded 3000 cells/well and incubated in serial dilutions of the compound for 72 hours before assaying cell growth, as described by (Brimacombe et al., 2009). Cytotoxicity (IC50) was defined as the drug concentration that reduced cell viability to 50% of the untreated control and was calculated from three independent experiments.
Flow Cytometry.
For each condition, 106 cells were suspended in Iscove’s modified Dulbecco’s medium supplemented with 5% fetal bovine serum. After addition of the fluorescent substrate with or without the inhibitor of interest, cells were incubated in the dark for 30 minutes at 37°C. Concentrations used were based on values reported in the literature (as cited elsewhere). In some instances, higher concentrations were employed to ensure inhibitory activity was easily resolved by flow cytometry. They were then washed and allowed to efflux in substrate-free media for 30 minutes under the same conditions, then centrifuged, resuspended in ice-cold phosphate-buffered saline, and kept on ice until analysis (within 1 hour). Cells were gated for forward versus side scatter and the geometric mean of fluorescence intensity (cellular uptake of fluorescent substrate) was recorded for a total of 20,000 cells using an LSR II flow cytometer (BD Biosciences, San Jose, CA) under the excitation emission wavelengths listed in Table 1. Fluorescence-activated cell sorting data were analyzed using FlowJo software (Tree Star, Inc., Ashland, OR).
Statistical Analysis.
Data are expressed as mean ± S.D. from three observations for fluorescence accumulation assays and cytotoxicity assays. After the data were tested for homogeneity of variance, statistical significance was evaluated by the Student’s t test (unpaired, two-tailed, α = 0.05) and by a two-way analysis of variance followed by the Bonferroni post–t test (α = 0.05).
Results
Generation of Mitoxantrone-Resistant Mouse Fibroblast Line Overexpressing ABCG2.
Exposure to mitoxantrone in incremental doses can select for expression of ABCG2 (Robey et al., 2001a). Here, NIH 3T3 fibroblasts were passaged with increasing mitoxantrone concentrations (1–15 nM) over a period of 3 months. Quantitative immunodetection of protein from NIH 3T3 and its selected mitoxantrone-resistant subline 3T3 MX15 demonstrated increased levels of ABCG2 (Fig. 1). While resistance to mitoxantrone may also result from increased expression of P-gp, this was not significantly overexpressed in 3T3 MX15 cells (Supplemental Fig. 1). To confirm that the resistance phenotype was mediated by Abcg2, we performed cytotoxicity assays with mitoxantrone in the presence or absence of the ABCG2 inhibitor Ko143 (5 μM). The 3T3 MX15 cells (IC50 = 134.6 ± 28.2 μM) were 33-fold more resistant to mitoxantrone relative to the parental cells (IC50 = 4.0 ± 2.0 μM). Resistance decreased to 5-fold in the presence of Ko143 (3T3 IC50 1.7 ± 0.2 μM and MX15 IC50 10.0 ± 8.0 μM). Both cell lines were sensitized to mitoxantrone by Ko143, confirming that 3T3 cells express low levels of Abcg2, as previously reported (Allen et al., 2000), and seen here by the electropherogram (Fig. 1). Functional Abcg2 was determined by the ability to efflux fluorescent ABCG2 substrates and analyzed by flow cytometery (Table 1). For example, 3T3 MX15 cells demonstrated a significantly lower accumulation of ABCG2 substrates than 3T3 cells (see Table 1; Supplemental Fig. 2), which increased in the presence of Ko143.
Human and Mouse ABCG2 Demonstrated an Overlap in Efflux of Fluorescent Substrates.
In this study, we examined independent drug-selected mouse and human ABCG2-expressing sublines with their respective parental lines (two pairs from each species; Fig. 1; Supplemental Figs. 1 and 3). A third mouse fibroblast cell line was used when a difference was observed (discussed below). Mutations at amino acid 482 in human and mouse ABCG2 can affect substrate and antagonist specificity (Allen et al., 2002a; Robey et al., 2003). This region has been suggested as a mutation hotspot acquired in highly drug-resistant cells. Therefore the mouse cell lines 3T3, 3T3 MX15, Ltk-, and Ltk-HoeR415 were sequenced to check for this mutation. Sequencing data showed homozygous wild-type sequence (arginine codon AGG) in all cell lines, with no trace of secondary mutant DNA signatures (Supplemental Fig. 4). In the other cell lines used in this study, this region had been previously characterized as wild-type (Allen et al., 2002a; Honjo et al., 2002; Volk et al., 2002).
To evaluate fluorescent substrates of human and mouse ABCG2, we conducted efflux studies using flow cytometry. Cells were incubated with fluorescent substrates identified from the literature (Table 1) with or without inhibitor (10 μM Ko143), washed, and then allowed to efflux in substrate-free media under the same conditions. The compounds studied were mitoxantrone, Hoechst 33342, BODIPY-prazosin, rhodamine 123, daunorubicin, pheophorbide a, MPPa, and chlorin e6 (Polgar et al., 2008). Daunorubicin and rhodamine 123 were assessed because they are transported by the R482 mutant but not wild-type ABCG2 (Robey et al., 2003). Cell fluorescence was normalized to the value derived in ABCG2-expressing cells in the presence of Ko143 (i.e., cells with fully inhibited ABCG2) and the percentage (%) difference with and without the inhibitor is displayed for ABCG2-expressing cells in Table 1. Accumulation levels for both parental and ABCG2 cells for each compound are shown in Supplemental Fig. 2. Mitoxantrone, Hoechst 33342, BODIPY-prazosin, pheophorbide a, MPPa, and chlorin e6 demonstrated significant efflux in human and mouse ABCG2-expressing cells when compared with parental cells (Supplemental Fig. 2), and the accumulation of substrates increased when co-incubated with Ko143 (10 μM) (Table 1). Daunorubicin accumulation did not significantly differ in the human and mouse ABCG2-expressing cells when inhibited, or compared with parental cells. The average percentage efflux of most substrates by human and mouse ABCG2 did not demonstrate a significant difference (Fig. 2).
Rhodamine 123 efflux was observed in mouse (not human) Abcg2-expressing sublines, (Supplemental Figs. 2 and 5), but this was due to P-gp efflux. Accumulation of rhodamine 123 increased ∼2- to 3-fold in the presence of the P-gp inhibitors PSC833 and DCPQ and ∼1.5-fold in the presence of Ko143 at high concentrations (Supplemental Fig. 6). Additionally, rhodamine 123 efflux was not observed in a mouse fibroblast cell line derived from a P-gp knockout mouse (MEF3.2) and its Abcg2-expressing subline (MEF 3.2 M32) (Supplemental Fig. 5B). The accumulation of mitoxantrone, daunorubicin, BODIPY-prazosin, and Hoechst 33342 was tested with and without a specific P-gp inhibitor (1 μM DCPQ) since they are weak substrates of P-gp. Mitoxantrone and daunorubicin accumulation did not significantly increase in the presence of a specific P-gp inhibitor. However, BODIPY-prazosin and Hoechst 33342 accumulation did increase in the presence of a specific P-gp inhibitor (Supplemental Fig. 7), but the difference was smaller than that due to Abcg2 efflux (Supplemental Fig. 2). Therefore, our data suggest that the low baseline P-gp in the mouse cell lines is not a major participant in the transport of mitoxantrone, Hoechst 33342, and BODIPY-prazosin.
Purpurin-18, a Porphyrin Derivative, Is a Specific Substrate of Human and Mouse ABCG2.
In addition to other known porphyrin ABCG2 substrates that are used for photodynamic therapy (Robey et al., 2004, 2005), we investigated Pp-18 (structure shown in Fig. 3A) as a potential new porphyrin substrate of ABCG2 by efflux studies with ABCG2-overexpressing sublines (Fig. 3B; Supplemental Fig. 2). Efflux of Pp-18 by parental cells (Fig. 3B, top row) was compared with efflux of Pp-18 from ABCG2-expressing cells (Fig. 3B, bottom row). In the parental cells, a small but detectable amount of inhibitable Pp-18 efflux was observed when these cells were incubated in Pp-18 without (gray line) or with (black line) Ko143. To evaluate the ABCG2 specificity of Pp-18 among ABC transporters, we conducted efflux studies on sublines overexpressing human P-gp (KB-8-5-11) and MRP1 (VP16), and mouse P-gp (C3M) compared with their parental lines (KB-3-1, MCF7, and 3T3, respectively). Cells were incubated with Pp-18 with or without their respective transport inhibitor (20 μM verapamil for P-gp; 100 μM indomethacin for MRP1). No significant differences were observed in the accumulation of Pp-18 between parental and transporter-expressing lines for human P-gp or MRP1 (Fig. 3C), and mouse P-gp (Fig. 3D), though a small nonspecific effect was evident in all cells. Pp-18 is therefore a specific substrate for human and mouse ABCG2 and can be used to test the effect of inhibitors on ABCG2 efflux.
Inhibitors Demonstrated a Similar Effect on Human and Mouse ABCG2-Mediated Efflux of Pp-18.
We next examined the inhibition of Pp-18 efflux to assess reported inhibitors of ABCG2-mediated drug resistance (Fig. 2B; Supplemental Fig. 8; Table 2). Small molecules reported as inhibitors or competitive substrates of ABCG2 were assessed: flavopiridol, elacridar, tariquidar, nilotinib, quercetin, 17-β-estradiol, novobiocin, cyclosporin A (Polgar et al., 2008), NSC73306 (Wu et al., 2007), chrysin (Zhang et al., 2005), benzoflavone (Zhang et al., 2005), and vismodegib (Zhang et al., 2009). Sulfasalazine is an ABCG2 substrate (van der Heijden et al., 2004) and we tested if sulfasalazine can act as a competitive inhibitor (Ambudkar et al., 1999). Ko143 (Allen et al., 2002b) and fumitremorgin C (FTC) (Rabindran et al., 1998) were included as positive controls. The fold value is defined as the accumulation of Pp-18 in the presence of an inhibitor divided by the accumulation of Pp-18 in the absence of an inhibitor. Elacridar, tariquidar, nilotinib, quercetin, NSC73306, flavopiridol, chrysin, benzoflavone, 17-β-estradiol, novobiocin, and vismodegib all demonstrated a significant increase in Pp-18 accumulation in both human and mouse ABCG2-expressing cell lines (Supplemental Fig. 8; Table 2). A significant difference between the average fold values of mouse and human ABCG2 was not observed (Fig. 2B). Sulfasalazine did not inhibit either human or mouse ABCG2 efflux of Pp-18. Quercetin, sulfasalazine, NSC73306, chrysin, 17-β-estradiol, novobiocin, and vismodegib were examined at lower concentrations (10 μM), and exhibited the same relative inhibition profile against all four cell lines (Supplemental Fig. 9).
Given conflicting reports that cyclosporin A inhibits ABCG2 function (Ejendal and Hrycyna, 2005; Gupta et al., 2006), we tested its ability to inhibit the efflux of Pp-18 in a dose-dependent manner (Fig. 4). We observed a dose-dependent increase in accumulation of Pp-18 when cyclosporin A (1, 5, 10, 25, 50 μM) was added. Incubation with cyclosporin A at 50 μM demonstrated a similar inhibition profile to Ko143 in the Ltk-HoeR415 subline.
Discussion
ABCG2 has a well-established role in limiting the oral bioavailability of substrates as well as limiting brain penetration and facilitating renal excretion of compounds. Abcg2-deficient mouse models have been extensively used to confirm these roles, as well as to test the ability of ABCG2 inhibitors to limit brain penetration or increase oral absorption of drugs. Despite this, few studies have been performed to assess the overlapping substrate specificity of human and murine ABCG2. To compare mouse and human ABCG2 function, we used human and mouse ABCG2-expressing sublines (and their parental lines) that were independently drug selected. To this end, we generated a mouse Abcg2-expressing subline (3T3 MX15) by long-term selection in mitoxantrone. We found that the ABCG2 substrates examined were similarly transported by human and mouse ABCG2. This suggests that human ABCG2 transporter substrate interactions may be extrapolated to mouse Abcg2 function for the tested substrates. Additionally, the fluorescent substrates examined here can be used to confirm that inhibitors of human ABCG2 can be used to inhibit murine Abcg2 when used in mouse models.
We evaluated the inhibition of mouse and human ABCG2 using purpurin-18 as the substrate. Known ABCG2 inhibitors demonstrated a similar pattern of inhibiting Pp-18 efflux in human and mouse ABCG2-expressing cells. Ko143 and FTC are considered to be “general” inhibitors of ABCG2 in that they inhibit adenosine triphosphatase (ATPase) activity of ABCG2. Alternatively, the remaining inhibitors have been demonstrated to be substrates of ABCG2, and as such act as competitive inhibitors. It has previously been reported that chrysin and benzoflavone can inhibit human but not mouse Abcg2 (Zhang et al., 2005). Zhang et al. observed inhibition of topotecan efflux in human ABCG2-expressing cells. However, these inhibitors did not alter mouse topotecan pharmacokinetics in vivo and did not significantly inhibit mouse Abcg2-expressing cells. This was attributed to species differences of ABCG2. We observed a significant fold-increase in Pp-18 accumulation (comparable to inhibition by FTC or Ko143) in mouse and human ABCG2-expressing cell lines when adding the same concentrations of chrysin and benzoflavone used by Zhang et al. Due to the existence of multiple binding sites on ABCG2 (Giri et al., 2009), there may be different transporter-mediated interactions, and competitive substrates may not inhibit the efflux of all transport substrates. As such, caution should be taken that an inhibitor of ABCG2 is dependent on the substrates tested.
Flavoperidol and nilotinib are ABCG2 substrates and can inhibit ABC transporter function at high concentrations as competitive substrates. Sulfasalazine (a known ABCG2 substrate) did not inhibit human or mouse ABCG2, and we also did not observe an increased uptake of other fluorescent substrates by sulfasalazine (data not shown). Some inhibition was noted in the HoeR415 Abcg2-expressing cell line that expressed the least ABCG2 of the cell lines used at 100 μM. The weak effect may be due to the carboxylic acid of sulfasalazine rendering the molecule negative at physiologic pH, reducing its cell-membrane permeability.
There is a longstanding question on the interaction of cyclosporin A with ABCG2 due to conflicting reports. In the presence of cyclosporin A (∼5–10 μM), Ejendal and Hrycyna (2005) did not observe displacement of a photo-affinity analog of prazosin, increased fluorescence accumulation of mitoxantrone or an effect on prazosin adenosine triphosphatase activity by cyclosporin A and concluded it does not inhibit ABCG2. However, Gupta et al. (2006) reported that cyclosporin A (∼1–10 μM) significantly inhibited ABCG2 efflux of BODIPY-prazosin, mitoxantrone, and pheophorbide A and concluded that cyclosporin A can inhibit ABCG2. Our studies indicated that cyclosporin A inhibited human and mouse ABCG2 efflux of Pp-18 in a dose-dependent manner that correlated with the amount of ABCG2 expressed at the cell membrane. The addition of cyclosporin A produced the highest fold-increase in the cells with the lowest expression of ABCG2 (Ltk-HoeR415) and this interaction was masked in the cells with higher ABCG2 expression when lower concentrations of cyclosporin A were used (∼1–5 μM) (Fig. 4), Cyclosporin A (5 μM) also increased the accumulation of BODIPY-prazosin in H460 MX20 (Supplemental Fig. 10), confirming the ability of cyclosporin A to act as an ABCG2 inhibitor.
We identified a new specific human and mouse ABCG2 porphyrin substrate, Pp-18. Porphyrins have proven useful in fluorescence-guided resections and treatment of cancer (Sun et al., 2013) and are known ABCG2 substrates (Jonker et al., 2002). Pp-18 is a natural red-emitting fluorescent product of chlorophyll and is a photosensitizer with cell-targeting utility (via covalent attachment to antibodies due to an anhydride group) (Hoober et al., 1988) (Sharma et al., 2006). Far-red emitting markers are more useful than those emitting blue-green light due to low background autofluorescence and also provide maximal tissue penetration (Edward, 2012). The expression of the ABCG2 transporter plays a significant role in the efflux of porphyrins from cells, as seen with Pp-18 efflux by ABCG2. Understanding the role of ABCG2 and its inhibition to increase the accumulation of porphyrins to improve therapeutic efficiency has great clinical implications for malignant gliomas (Sun et al., 2013). Despite new treatment modalities, patients with aggressive glioblastoma have a median survival rate of less than 15 months, and multiple resections are often necessary to prolong survival (Chaichana et al., 2013). One reason for this poor prognosis may be due to the function of ABCG2, which enables resistance to treatment as well as escape from fluorescence-guided resections. Additionally, ABCG2 is commonly found in the “side-population” of cells implicated in gliomagenesis (Bleau et al., 2009) and may play a role in the recurrence of glioblastoma. Given that Pp-18 is extruded by mouse and human ABCG2 and it is not a substrate for human P-gp or MRP1 or mouse P-gp (Fig. 3) it may be useful for preclinical mouse models to study ABCG2 function.
The ABCG2 protein plays an important role in a variety of physiologic and pathologic functions. It can protect the body from xenobiotics but it also confers multidrug resistance in cancer cells. ABCG2−/− [Jr(a−)] individuals exist, and the functional consequence on normal physiology drug pharmacodynamics is still not known. Therefore, the substrate and inhibitor overlap of mouse and human ABCG2 substantiates the use of mouse models with potential clinical implications.
Acknowledgments
The authors thank Karen M. Wolcott for flow cytometry assistance, Dominic Esposito and Vanessa Wall for gene sequencing assistance, and George Leiman for editorial assistance.
Authorship Contributions
Participated in research design: Bakhsheshian, Hall, Robey, Bates, Gottesman.
Conducted experiments: Bakhsheshian, Herrmann, Chen.
Performed data analysis: Bakhsheshian, Hall, Robey, Herrmann, Chen.
Wrote or contributed to the writing of the manuscript: Bakhsheshian, Hall, Robey, Bates, Gottesman.
Footnotes
- Received May 30, 2013.
- Accepted July 18, 2013.
This research was supported by the Intramural Research Program of the National Institutes of Health [National Cancer Institute]. J.B. is a National Institutes of Health Medical Research Scholar. The authors declare no conflicts of interest.
↵This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- ABCG2
- ATP-binding cassette subfamily G member 2
- DCPQ
- (2R)-anti-5-{3-[4-(10,11-dichloromethanodibenzo-suber-5-yl)piperazin-1-yl]-2-hydroxypropoxy}quinoline trihydrochloride
- FTC
- fumitremorgin C
- MPPa
- pyropheophorbide a methyl ester
- MRP1
- multidrug resistance protein 1
- NSC73306
- 1-isatin-4-(4′-methoxyphenyl)thiosemicarbazone
- P-gp
- P-glycoprotein
- Pp-18
- purpurin-18
- U.S. Government work not protected by U.S. copyright