Journal list menu

Volume 276, Issue 24 p. 7237-7252
Free Access

Disruption of N-linked glycosylation enhances ubiquitin-mediated proteasomal degradation of the human ATP-binding cassette transporter ABCG2

Hiroshi Nakagawa

Hiroshi Nakagawa

Department of Biomolecular Engineering, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Japan

Search for more papers by this author
Kanako Wakabayashi-Nakao

Kanako Wakabayashi-Nakao

Department of Biomolecular Engineering, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Japan

Search for more papers by this author
Ai Tamura

Ai Tamura

Department of Biomolecular Engineering, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Japan

Search for more papers by this author
Yu Toyoda

Yu Toyoda

Department of Biomolecular Engineering, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Japan

Search for more papers by this author
Shoko Koshiba

Shoko Koshiba

Department of Biomolecular Engineering, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Japan

Search for more papers by this author
Toshihisa Ishikawa

Toshihisa Ishikawa

Department of Biomolecular Engineering, Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Midori-ku, Yokohama, Japan

Search for more papers by this author
First published: 27 November 2009
Citations: 75
Toshihisa Ishikawa, Omics Science Center, RIKEN Yokohama Institute, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama 230-0045, Japan
Fax: +81 45 503 9216
Tel: +81 45 503 9222 or 9179
E-mail: [email protected]

Abstract

The human ATP-binding cassette (ABC) transporter, ABCG2 (BCRP/MXR/ABCP), is a plasma membrane protein containing intramolecular and intermolecular disulfide bonds and an N-linked glycan at Asn596. We have recently reported that the intramolecular disulfide bond is a critical checkpoint for determining the degradation fates of ABCG2. In the present study, we aimed to analyze quantitatively the impact of the N-linked glycan on the protein stability of ABCG2. For this purpose, we incorporated one single copy of ABCG2 cDNA into a designated site of genomic DNA in Flp-In-293 cells to stably express ABCG2 or its variant proteins. When ABCG2 wild type-expressing cells were incubated with various N-linked glycosylation inhibitors, tunicamycin profoundly suppressed the protein expression level of ABCG2 and, accordingly, reduced the ABCG2-mediated cellular resistance to the cancer chemotherapeutic SN-38. When Asn596 was converted to Gln596, the resulting variant protein was not glycosylated, and its protein level was about one-third of the wild type level in Flp-In-293 cells. Treatment with MG132, a proteasome inhibitor, increased the level of the variant protein. Immunoblotting with anti-ubiquitin IgG1k after immunoprecipitation of ABCG2 revealed that the N596Q protein was ubiquitinated at levels that were significantly enhanced by treatment with MG132. Immunofluorescence microscopy demonstrated that treatment with MG132 increased the level of ABCG2 N596Q protein both in intracellular compartments and in the plasma membrane. In conclusion, we propose that the N-linked glycan at Asn596 is important for stabilizing de novo-synthesized ABCG2 and that disruption of this linkage results in protein destabilization and enhanced ubiquitin-mediated proteasomal degradation.

Structured digital abstract

Abbreviations

  • ABC
  • ATP-binding cassette
  • ABCG2 N596Q
  • ABCG2 variant in which Asn596 was substituted by Gln596
  • ABCP
  • placenta-specific ABC transporter
  • BCRP
  • breast cancer resistance protein
  • BMA
  • bafilomycin A1
  • DMEM
  • Dulbecco’s modified Eagle’s medium
  • Endo H
  • Endoglycosidase H
  • ER
  • endoplasmic reticulum
  • ERAD
  • endoplasmic reticulum-associated degradation
  • Flp-In-293/ABCG2 N596Q cells
  • Flp-In-293 cells expressing ABCG2 N596Q variant
  • Flp-In-293/ABCG2 WT cells
  • Flp-In-293 cells expressing ABCG2 WT
  • Flp-In-293/Mock cells
  • Flp-In-293 cells transfected with pcDNA/FRT Mock vector
  • FRT
  • Flp recombination target
  • GAPDH
  • glyceraldehyde-3-phosphate dehydrogenase
  • HRP
  • horseradish peroxidase
  • MTT
  • 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide
  • MXR
  • mitoxantrone resistance-associated protein
  • NaCl/Pi
  • phosphate-buffered saline
  • PNGase F
  • N-glycosidase F
  • TM5
  • transmembrane domain 5
  • TM6
  • transmembrane domain 6
  • TTBS
  • Tris-buffered saline containing 0.05% (v/v) Tween 20
  • Introduction

    Human ABCG2 (BCRP/MXR/ABCP) is a member of the ATP-binding cassette (ABC) transporter family. This ABC transporter was originally discovered as a multidrug resistance protein in cancer cells in vitro [1,2]. As ABCG2 extrudes a variety of anticancer drugs, modulation of its function is of great interest in cancer chemotherapy [3]. In fact, overexpression of ABCG2 is reportedly a cause of cancer cell resistance to SN-38 [4], an active metabolite of irinotecan (CPT-11) that is widely used for the chemotherapeutic treatment of lung, stomach, colon, breast, ovary and uterine cancers. In addition to cancer cells, ABCG2 has been found to be expressed in many different types of normal cells, such as placenta, small intestine, liver, kidney, lactating mammary gland, the blood–brain barrier and the blood–testis barrier, and in stem cells, including undifferentiated human embryonic stem cells [5–10]. The physiological function of ABCG2 in normal tissues was elucidated by using Abcg2−/− mice [11]. Recent studies have revealed important contributions of ABCG2 to the blood–brain, blood–testis and blood–fetal barriers, indicating a primary biological role of ABCG2 in protecting those organs from a wide range of xenobiotics. Furthermore, other physiological functions elucidated from studies in Abcg2−/− mice include the extrusion of porphyrins and/or porphyrin conjugates from hematopoietic cells, liver and harderian gland, as well as the secretion of vitamin B2 (riboflavin) and possibly other vitamins (biotin, vitamin K) into breast milk [11–20].

    Sequencing of the ABCG2 gene from human samples has revealed over 80 different, naturally occurring sequence variations [21]. Clinical relevance is implicated between the genetic polymorphisms of the ABCG2 gene and individual differences in drug response [22]. We have recently demonstrated that certain nonsynonymous single nucleotide polymorphisms (SNPs) affect not only the substrate specificity of ABCG2 but also its protein expression levels [23,24], indicating that individual differences in the function of ABCG2 are partly caused by nonsynonymous SNPs. In fact, the protein expression levels of ABCG2 SNP variants (Q141K, F208S and S441N) were significantly lower than that of the wild-type (WT) ABCG2 [23]. We have recently provided evidence that those SNP variants were degraded via the ubiquitin-proteasome proteolytic pathway [25,26].

    The ubiquitin-proteasome proteolytic process is known as a major pathway of endoplasmic reticulum-associated degradation (ERAD). The endoplasmic reticulum (ER) is the site where newly synthesized secretory and membrane proteins are folded and assembled under a stringent quality-control system. The accumulation of misfolded proteins in the ER would detrimentally affect cellular functions. ABCG2 proteins lacking one intramolecular disulfide bond are recognized as misfolded proteins in the ER and are readily degraded via the ERAD pathway [27]. The intramolecular disulfide bond between Cys592 and Cys608, which resides in the extracellular loop, has been shown to be important for stability of the protein [27].

    In previous studies, we proposed that the ER has at least two checkpoints to monitor the quality of the human ABCG2 protein [25,27]: one checkpoint is the intramolecular disulfide bond between Cys592 and Cys608, and the other is the N-linked glycan at Asn596. At present, however, the involvement of the N-linked glycan of ABCG2 in the checkpoint, and its physiological roles, are not well understood. Previous studies have indicated that ABCG2 is glycosylated at Asn596 and that this post-translational modification was not essential for expression, transport activity, or localization to the plasma membrane [28,29]. We re-examined the role of the N-linked glycan of ABCG2, however, in terms of the protein stability and ERAD. For this purpose, we used the Flp recombinase system to integrate one single ABCG2 cDNA sequence into chromosomal DNA for quantitative analysis [30]. The present study provides evidence that disruption of N-linked glycosylation enhances the ubiquitin-mediated proteasomal degradation of ABCG2. The N-linked glycan at Asn596 is considered to be important for stabilization of the nascent ABCG2 protein synthesized in the ER. Furthermore, we propose that N-linked glycosylation facilitates dimer formation of ABCG2 proteins through a cysteinyl disulfide bond in the ER.

    Results

    Influence of N-linked glycosylation inhibitors on the expression level of human ABCG2 protein

    We attempted to examine the importance of N-linked glycosylation in the stability and level of expression of human ABCG2 protein. For this purpose, we first used six different types of N-linked glycosylation inhibitors (tunicamycin, castanospermine, deoxynojirimycin, brefeldin A, deoxymannojirimycin and swainsonine). Figure 1A illustrates schematically the sites of action of those inhibitors in the pathway, including dolicol glycoside biosynthesis, N-linked glycosylation of newly synthesized proteins and the processing of N-linked glycans.

    Details are in the caption following the image

    Effects of glycosylation inhibitors on human ABCG2 expressed in Flp-In-293/ABCG2 WT cells. (A) Schematic model of N-linked glycosylation on a protein and its inhibitors, and schematic illustration of human ABCG2. E-1, α1,2-glucosidase I; E-2, α1,3-glucosidase II; E-3, golgi α1,2-mannosidase I; E-4, golgi α1,2-mannosidase II; Dol, dolichol; P, inorganic phosphate. (B) Glycosylation status and protein expression level of cells treated with glycosylation inhibitors. Flp-In-293/ABCG2 WT cells were cultured, for 48 h, in the presence or absence of 10 μg·mL−1 of tunicamycin(TNM), 1 mm castanospermine (CSP) and 1 mm deoxynojirimycin (DNJ). Cell lysate samples (20 μg of protein) were subjected to SDS/PAGE after treatment with or without PNGase F. Human ABCG2 proteins in the resulting samples were analyzed by immunoblotting with the BXP-21 antibody. The intensity of the bands recognized by BXP-21 was determined and expressed as a ratio of the bands in untreated cells (None). Similar results were obtained in more than two other experiments. The apparent relative molecular mass values of mature (G) and nonglycosylated (non-G) ABCG2 were 81 000 and 72 000, respectively. (C) The SN-38 resistance activity of Flp-In-293/ABCG2 WT cells (closed symbols) and Flp-In-293/Mock cells (open circles). Flp-In-293/ABCG2 WT cells (closed symbols) and Flp-In-293/Mock cells (open circles) were precultured in the presence of 10 μg·mL−1 of tunicamycin (closed triangles), 1 mm castanospermine (closed squares), or vehicle (closed and open circles) for 48 h. Then, the cells were cultured for 72 h with different concentrations of SN-38. After culture, the cell viability was determined using the MTT assay. Data are expressed as the mean values ± SD of triplicate determinations. *, P < 0.01 compared with the control.

    Figure 1B depicts the effects of those inhibitors on the expression levels of the ABCG2 protein in Flp-In-293 cells expressing ABCG2 WT (Flp-In-293/ABCG2 WT cells). To analyze, quantitatively, their effects on the protein-expression levels of WT human ABCG2, we treated all samples with N-glycosidase F (PNGase F) to remove N-linked glycans. After immunoblotting with the ABCG2-specific mAb, BXP-21, one single band was detected (Fig. 1B). Incubation with 10 μg·mL−1 of tunicamycin or 1 mm castanospermine reduced the level of expression of ABCG2 protein in Flp-In-293/ABCG2 WT cells, whereas 2 μm brefeldin A, 1 mm deoxymannojirimycin or 200 μm swainsonine did not reduce ABCG2 protein levels (data not shown).

    To examine, in more detail, the impact of those inhibitors on the N-linked glycosylation status of ABCG2, the same samples were analysed by immunoblotting without PNGase F treatment. In these experiments, additional bands with lower molecular weights (indicated by asterisks) were detected in the cells that had been incubated with tunicamycin or brefeldin A (Fig. 1B upper panel). These additional bands were suggested to be nonglycosylated or immature forms of ABCG2 because their molecular weights were lower than that of the N-glycosylated mature form of ABCG2.

    As the levels of ABCG2 protein were significantly decreased after treatment with tunicamycin or castanospermine, we evaluated the drug-resistance profile of those cells. Flp-In-293/ABCG2 WT cells were pretreated with tunicamycin or castanospermine for 24 h and then incubated with different concentrations (0–100 nm) of SN-38, an active metabolite of irinotecan, for 72 h. The cell viability was determined using the MTT assay. Flp-In-293/ABCG2 WT cells were more resistant to SN-38 than were Flp-In-293 cells transfected with pcDNA/FRT Mock vector (Flp-In-293/Mock cells) (Fig. 1C). It is noteworthy, however, that the tunicamycin treatment partially restored the sensitivity of Flp-In-293/ABCG2 WT cells to SN-38 (Fig. 1C). This reversal effect corresponded to a decreased ABCG2 protein level in the tunicamycin-treated cells. By contrast, the reversal effect of castanospermine (1 mm) on cellular resistance was smaller (Fig. 1C), which is in accordance with smaller changes in the ABCG2 protein level (Fig. 1B).

    Effect of MG132 on the expression of human ABCG2 protein in tunicamycin-treated cells

    The ABCG2 protein level in Flp-In-293/ABCG2 WT cells decreased in a time-dependent manner during incubation with tunicamycin (10 μg·mL−1) (Fig. 2A). Immunoblotting following treatment with PNGase F revealed that the ABCG2 protein expression level decreased by approximately 50% after incubation with tunicamycin for 72 h (Fig. 2A). In the absence of PNGase F treatment, both glycosylated and nonglycosylated forms of ABCG2 were detected by immunoblotting the samples obtained 24, 48 and 72 h after incubation with tunicamycin (Fig. 2A). Such effects on the protein expression level and on the glycosylation status of ABCG2 appeared to be more prominent when cells were treated with tunicamycin at concentrations higher than 1 μg·mL−1 (Fig. 2B).

    Details are in the caption following the image

    Effects of tunicamycin on the protein expression status of human ABCG2 expressed in Flp-In-293/ABCG2 WT cells. (A and B) Cell lysate samples (20 μg of protein) from Flp-In-293/ABCG2 WT cells cultured with tunicamycin (10 μg·mL−1) for different periods of time (0, 24, 48 and 72 h) (A), or cultured with different concentrations of tunicamycin (0, 0.1, 1 and 10 μg·mL−1) for 48 h (B), were treated with PNGase F (B) or not treated with PNGase F (A and B). Human ABCG2 proteins in the resulting samples were analyzed by immunoblotting with the BXP-21 antibody. The intensity of the bands recognized by BXP-21 was determined and expressed as a ratio of the bands in untreated cells (0 h in A and 0 μg·mL−1 in B). Similar results were obtained in more than two other experiments. (C) Cell lysate samples (20 μg of protein) from Flp-In-293/ABCG2 WT cells cultured for 1 h, with or without 1 μm MG132 (pretreatment), and subsequently for 48 h with or without tunicamycin (10 μg·mL−1) and in the presence of absence of 1 μm MG132, were treated or not treated with PNGase F. Human ABCG2 proteins in the resulting samples were analyzed as described above. Data are expressed as the mean values ± SD of triplicate determinations. *, P < 0.05 between the groups indicated. (D) Cell lysate samples (20 μg of protein) from Flp-In-293/ABCG2 WT cells cultured for 48 h with tunicamycin (10 μg·mL−1) in the presence or absence of 1 μm MG132 were subjected to SDS/PAGE under nonreducing conditions. Human ABCG2 proteins in the resulting samples were analyzed as described above. Similar results were obtained in more than two other experiments.

    Tunicamycin is known to inhibit N-linked glycosylation and to activate indirectly the ubiquitin-proteasome proteolytic pathway [31]. We therefore examined the involvement of the ubiquitin-proteasome proteolytic pathway in the decrease of ABCG2 protein expression levels in Flp-In-293/ABCG2 WT cells treated with tunicamycin. To do so, we cultured Flp-In-293/ABCG2 cells in the presence or absence of 1 μm MG132, an inhibitor of proteasomal proteolysis, for 1 h (pretreatment) and subsequently with tunicamycin (10 μg·mL−1) for 48 h in the presence or absence of 1 μm MG132. As shown in Fig. 2C, the pretreatment with MG132 partly suppressed the effect of tunicamycin that reduced ABCG2 protein levels. Similar effects of tunicamycin and MG132 were also observed in human breast adenocarcinoma cell line MCF-7 cells that endogenously express ABCG2 (data not shown).

    Furthermore, as shown in Fig. 2D, immunoblot analysis under nonreducing conditions revealed that the levels of ABCG2 monomer and homodimer were higher in cell lysate obtained from cells treated with MG132 than in cell lysate obtained from the control (i.e. without MG132 treatment). Importantly, the nonglycosylated monomeric form of ABCG2 (indicated by an asterisk) was detected in the cells treated with MG132 (Fig. 2D). These results suggest that inhibition of N-linked glycosylation partly impairs the formation of homodimers, and that the nonglycosylated monomer form of ABCG2 is subjected to ubiquitin-mediated proteasomal degradation.

    Characterization of human ABCG2 N596Q expressed in Flp-In-293 cells

    To examine in more detail the potential role of N-linked glycosylation in the stability and degradation of ABCG2 protein, we established the N596Q variant of human ABCG2, in which Asn596, the amino acid for N-linked glycosylation, was substituted by Gln596 and therefore N-linked glycosylation was predicted not to occur at all. By employing the Flp recombinase system, we integrated one single copy of the N596Q variant cDNA into the designated Flp recombination target (FRT) site in chromosome 12, as previously reported [23,32,33]. Immunoblot analysis (Fig. 3A left panel) demonstrated that the relative molecular mass of human ABCG2 N596Q (72 000) was smaller than that of WT ABCG2 (81 000). The relative molecular mass of the N596Q variant was changed by neither Endoglycosidase H (Endo H) nor PNGase F treatments (Fig. 3A right panel). These results confirm that human ABCG2 N596Q completely loses the N-linked glycan and that there are no other N-linked glycosylation sites in this protein. Immunoblot analysis revealed that the protein expression level of human ABCG2 N596Q in Flp-In-293 cells was about one-third of the expression level of ABCG2 WT protein (Fig. 3B left panel), whereas the mRNA expression levels of both ABCG2 WT and ABCG2 N596Q were almost identical (Fig. 3B right panel).

    Details are in the caption following the image

    Comparison between human ABCG2 WT and N596Q expressed in Flp-In-293 cells. (A) Apparent molecular weights (left panel) and sensitivities to glycosidases (right panel) of human ABCG2 WT and human ABCG2 N596Q. Cell lysate samples (20 μg of protein) were treated with Endo H or PNGase F, as described in the Materials and Methods. ABCG2 WT and ABCG2 N596Q proteins in the resulting samples were analyzed as described above. The relative molecular mass value of human ABCG2 N596Q was the same as that of nonglycosylated human ABCG2 WT (72 000). (B) Protein and mRNA levels of human ABCG2 WT and N596Q expressed in Flp-In-293 cells. The mRNA levels were analyzed by RT-PCR and by quantitative PCR with total RNA extracted from Flp-In-293 cells expressing ABCG2 WT or ABCG2 N596Q. For comparison of the protein levels, the cell lysate of each cell population was analyzed by immunoblotting with the ABCG2-specific mAb (BXP-21) or with the GAPDH-specific antibody, after treatment with PNGase F. The signal intensity ratio (ABCG2/GAPDH) was normalized to the WT level. *, P < 0.01 compared with Flp-In-293 cells expressing ABCG2 WT. (C) The effect of cycloheximide on the protein stability of human ABCG2 WT and ABCG2 N596Q expressed in Flp-In-293 cells. Cells expressing ABCG2 WT (circles) or ABCG2 N596Q (closed triangles) were incubated with 10 μm cycloheximide for 0, 2, 4 and 8 h. Cell lysate samples were prepared and subsequently subjected to treatment with PNGase F and immunoblotting analyses as described in the Materials and Methods. ABCG2 and GAPDH were detected using specific antibodies. The protein levels of ABCG2 WT or ABCG2 N596Q were calculated from the corresponding signal intensities and then normalized to the values observed at t = 0 h. Data are expressed as mean values ± SD in quadruplicate experiments. *, P < 0.01 compared with the normalized values (t = 0 h). (D) SN-38 resistance activity of the cells. Flp-In-293 cells expressing ABCG2 WT (circles) or ABCG2 N596Q (closed triangles) and Flp-In-293/Mock (squares) cells were cultured with different concentrations of SN-38 for 72 h. After the culture, the cell viability was determined using the MTT assay. Data are expressed as the mean values ± SD of triplicate determinations. *, P < 0.01 compared with Flp-In-293 cells expressing ABCG2 WT.

    To examine the protein stability of the N596Q variant, we treated Flp-In-293 cells with 10 μm cycloheximide to inhibit de novo protein synthesis. Immunoblotting (Fig. 3C) revealed that the amount of ABCG2 N596Q protein decreased faster than that of the ABCG2 WT protein after treatment with cycloheximide. These results indicate that disruption of N-linked glycosylation enhances the degradation of the ABCG2 protein.

    Fig. 3D demonstrates the SN-38 resistance profiles of Flp-In-293/Mock, Flp-In-293/ABCG2 WT and Flp-In-293/Flp-In-293 cells expressing ABCG2 N596Q variant (ABCG2 N596Q cells). The experiments were performed in the same way as described for Fig. 1C. Flp-In-293/ABCG2 N596Q cells were 2.5-fold more sensitive to SN-38 than were Flp-In-293/ABCG2 cells. The half-maximal inhibitory concentration (IC50) values were 8 and 20 nm for Flp-In-293/ABCG2 N596Q cells and Flp-In-293/ABCG2 cells, respectively (Fig. 3D).

    Effects of bafilomycin A1 and MG132 on the expression level of human ABCG2 N596Q protein

    Protein degradation occurs in two major sites, namely lysosomes and proteasomes. Lysosomal degradation is inhibited by bafilomycin A1 (BMA), whereas proteasomal degradation is inhibited by MG132. To elucidate the contribution of those pathways to the degradation and turnover of ABCG2 WT and N596Q proteins, we incubated Flp-In-293 cells expressing either WT protein or the variant lacking N-linked glycan in the presence of those inhibitors (2 μm MG132 or 10 nm BMA). The concentrations of those inhibitors were nontoxic to Flp-In-293 cells but potently inhibited protein degradation, as previously reported [25,27]. Figure 4A demonstrates the effect of MG132 and BMA on the protein levels of ABCG2 WT and N596Q. In these experiments, the samples were treated with PNGase F, and SDS/PAGE was performed under reducing conditions (i.e. with 2-mercaptoethanol). A marked difference was observed between human ABCG2 WT and ABCG2 N596Q with respect to the effects of MG132 and BMA. In the case of ABCG2 WT, its protein level was enhanced more than 2.5-fold when the cells were treated with BMA, whereas there was little effect with MG132. In the case of ABCG2 N596Q lacking N-linked glycan, however, the protein level was increased about twofold, not only by BMA but also by treatment with MG132. These results suggest that lysosomal degradation is the major pathway for the degradation of ABCG2 WT protein. By contrast, the lack of N-linked glycosylation appears to shift the pathway towards proteasomal degradation.

    Details are in the caption following the image

    Effect of MG132 on the protein levels of human ABCG2 WT and ABCG2 N596Q. Cell lysate samples (20 μg of protein) from Flp-In-293/ABCG2 WT cells and Flp-In-293/ABCG2 N596Q cells cultured with 10 nm BMA (A) or 2 μm MG132 (MG) (A and B) for 24 h were subjected to SDS/PAGE after treatment with (A) or without (B) 2-mercaptoethanol. Analysis of human ABCG2 proteins in the resulting samples was carried out as described above. The protein expression levels are represented as ratios relative to the control (None) level in drug-untreated cells (A). Data are expressed as mean values ± SD in triplicate experiments (*, P < 0.05) (A). Similar results were obtained in more than two other experiments.

    To examine the effects of MG132 on homodimer formation of ABCG2 WT and N596Q, SDS/PAGE was performed under nonreducing conditions (i.e. without 2-mercaptoethanol). As demonstrated in Fig. 4B, the ABCG2 N596Q variant forms homodimers via a cysteinyl disulfide bond under standard conditions. It is important to note that MG132 enhanced the levels of ABCG2 WT and ABCG2 N596Q monomers, where the level of ABCG2 N596Q protein was much higher than the nonglycosylated monomer form of ABCG2 (Fig. 4B). Thus, MG132 enhanced the levels of ABCG2 WT and ABCG2 N596Q monomers, which suggests that the proteasomal degradation pathway prefers monomeric forms of ABCG2.

    Effect of MG132 on the cellular localization of ABCG2 WT and ABCG2 N596Q

    It was of great interest to establish how the inhibition of proteasomal protein degradation by MG132 affects the cellular localization of ABCG2 WT and N596Q proteins. Figure 5A depicts the immunofluorescence images of Flp-In-293 cells expressing ABCG2 WT or N596Q that were incubated with or without 2 μm MG132 for 24 h. The ABCG2 protein was probed with either BXP-21 or 5D3 antibody. With the BXP-21 antibody [which recognizes an epitope (amino acids 271-396) in the intracellular loop of ABCG2], the localization of ABCG2 was detected both at the plasma membrane and within intracellular compartments. By contrast, with the 5D3 antibody, which recognizes an epitope in the extracellular loop of the ABCG2 protein [34,35], we observed the localization of ABCG2 solely at the plasma membrane.

    Details are in the caption following the image

    Immunocytochemical staining of Flp-In-293 cells expressing ABCG2 WT and N596Q proteins. Cells were incubated with (b, d, f, and h) or without (a, c, e, and g) 2 μm MG132 for 24 h. ABCG2 proteins were detected immunologically with an ABCG2-specific mAb, either BXP-21 (a, b, e, and f) or 5D3 (c, d, g, and h), and Alexa Fluor 488-conjugated antibody (green). As antibody BXP21 recognizes an intracellular loop of ABCG2, the cells in panels a, b, e and f were permeabilized with 0.02% (v/v) Triton X-100. By contrast, as antibody 5D3 recognizes an extracellular loop of ABCG2, the cells in panels c, d, g and h were not permeabilized; the fluorescence reflected the cell-surface signal, but no intracellular signal. Cellular nuclei were stained with Hoechst 33342 (blue), as described in the Materials and Methods. The horizontal white bars correspond to 10 μm. (B) The intensity of immunofluorescence was measured for a total of 300 cells, and the average intensity per cell was calculated based on the immunofluorescence data. Data are expressed as mean values ± SD (*, P < 0.005).

    As demonstrated in Fig. 5A, the ABCG2 protein was labeled with green fluorescence dye (Alexa Fluor 488), whereas DNA in the nuclei was stained with Hoechst 33342 (blue fluorescence). Immunofluorescence of the N596Q variant was relatively weaker at the plasma membrane as well as within intracellular compartments (Fig. 5A, panels e and g). After treatment with MG132, the amount of ABCG2 variant protein in the intracellular compartments increased (Fig. 5A, panel f). Furthermore, immunofluorescence with the 5D3 antibody revealed that the plasma membrane localization of ABCG2 N596Q was clearly enhanced by treatment with MG132 (Fig. 5A, compare panels g and h). In the case of ABCG2 WT, by contrast, strong green fluorescence was observed at the plasma membrane and within intracellular compartments in Flp-In-293 cells expressing ABCG2 WT (Fig. 5A, panels a and c). The localization and intensity of the WT protein were not greatly affected by the treatment with MG132 (Fig. 5A, panels b and d).

    Figure 5B demonstrates the changes in the average intensity of ABCG2-positive dot clusters per cell in the presence and absence of MG132 treatment. Statistically significant (P < 0.005) differences are indicated by use of an asterisk. In accordance with the immunoblotting data shown in Fig. 4A, the inhibition of proteasomal protein degradation by MG132 significantly increased the levels of the ABCG2 N596Q protein within each Flp-In-293 cell.

    Effect of MG132 on the ubiquitination states of human ABCG2 N596Q

    To investigate the effect of MG132 on the ubiquitination states of human ABCG2 N596Q, the Flp-In-293 cells expressing human ABCG2 N596Q were incubated in the presence or absence of 2 μm MG132 for 24 h. As shown in Fig. 6, a significant increase in the ubiquitinated form (arrowheads) of human ABCG2 N596Q was detected by immunoblotting with the anti-ubiquitin IgG1k after immunoprecipitation with the anti-ABCG2 IgG2a, BXP-21. Similar results were obtained by immunoblotting with BXP-21 after immunoprecipitation with the anti-ubiquitin IgG1k. By contrast, the ubiquitination state of human ABCG2 WT was little affected by the same treatment with MG132.

    Details are in the caption following the image

    Effect of MG132 on ubiquitination of ABCG2 WT and N596Q. After incubation with or without 2 μm MG132 for 24 h, cell lysate samples were prepared from Flp-In-293/ABCG2 WT cells and Flp-In-293/N596Q cells. The samples (1 mg of protein for each) were then immunoprecipitated with the ABCG2-specific mAb (BXP-21) then immunoblotted with the anti-ubiquitin mAb raised in mouse hybridoma cells (A). In those panels, the corresponding results are indicated by immunoprecipitation (ABCG2) and immunoblotting (ubiquitin). The same cell lysate samples were also subjected to immunoprecipitation with the anti-ubiquitin mAb then immunoblotted with the ABCG2-specific mAb (BXP-21) (B). In those panels, the corresponding results are indicated by immunoprecipitation (ubiquitin) and immunoblotting (ABCG2).

    Discussion

    Impact of N-linked glycosylation on the stability of human ABCG2 protein

    In eukaryotes, N-linked glycosylation occurs on Asn residues within the consensus sequence of Asn-X-Thr/Ser, where X can be any amino acid except proline. N-linked glycosylation is known to play a role in protein stability, sorting to designated sites and biological activities [36–38]. ABCG2 is an N-linked glycosylated protein localized in the apical domain of plasma membranes. This N-linked glycosylation site resides in the extracellular loop between transmembrane domain 5 (TM5) and transmembrane domain 6 (TM6). Amino acid sequences in the extracellular loop exhibit similarities among the ABCG2 orthologues in different species hitherto reported. As shown in Table 1, ABCG2 orthologues have one or two putative N-linked glycosylation sites in their extracellular loop between TM5 and TM6. In particular, the N-linked glycosylation sites corresponding to Asn596 in human ABCG2 are highly conserved.

    Table 1. Partial amino acid sequences of the extracellular loop between TM5 and TM6 in human ABCG2 and the corresponding sequences of its orthologues in mammals.
    Species Homologya Sequences Positions Access no.
    Human LGQNFCPGL N ATGNNPC NYATCTGE 587–611 NM004827
    Chimpanzee 75.70 LGQNFCPGL N ATGNNPC NYATCTGE 792–816 XM526633
    Monkey 93.50 LGQNFCPGL N ATV N NTC NYATCTGE 606–630 AY841878
    Pig 84.45 LGQNFCPGL N VTT N NTC SYAICTGE 588–612 NM214010
    Cow 83.92 LGQNFCPGL N VTT N NTC SYAICTGE 590–614 NM001037478
    Goat 84.98 LGQNFCPGL N VTA N NTC SYAICTGE 590–614 DQ904356
    Sheep 84.98 LGQNFCPGL N VTA N NTC SYAICTGE 590–614 NM001078657
    Dog 82.83 LGQNFCPGV N VTT N NTC SYAICTGE 587–611 NM001048021
    Rat 81.00 LGQEFCPGL N VTM N STC VN SYTICTGN 587–613 NM181381
    Mouse 81.61 LGQEFCPGF N VTD N STC VN SYAICTGN 587–613 NM011920
    • Multiple alignment and homology calculation were carried out using the genetyx-win Ver. 5.1 program (Software Development Co., Ltd., Tokyo, Japan). Putative N-linked glycosylation sites are underlined. a Homology to human ABCG2 (%).

    In the extracellular loop between TM5 and TM6 of human ABCG2, Cys603 appears to be involved in the formation of human ABCG2 dimers, whereas two cysteine residues close to Asn596 (i.e. Cys592 and Cys608) are important for the stability, as well as for the plasma membrane-targeting, of the ABCG2 protein [33,39]. ABCG2 WT is degraded in lysosomes, whereas misfolded ABCG2 proteins lacking the intramolecular disulfide bond undergo ubiquitin-mediated protein degradation in proteasomes [27]. The formation of an intramolecular disulfide bond between Cys592 and Cys608 in the extracellular loop would be a critical quality-control checkpoint for the human ABCG2 protein in the ER [27]. Based on those findings, N-linked glycosylation at Asn596 was also considered as an important step in the quality control of the ABCG2 protein.

    Mohrmann et al. (2005), as well as Diop and Hrycyna (2005), have reported that N-linked glycosylation on Asn596 is not essential for trafficking to the plasma membrane, protein expression or transport activity, of ABCG2. It is important to note that in those studies, ABCG2 WT, the R482 acquired mutant form, or their N596Q variants were expressed in mammalian cells (CHO9, MDCKII or HeLa) by transient transfection methods using the pcDNA3 vector or the vTF 7-3 vaccinia virus [28,29]. In general, by using transient transfection methods, it is difficult to evaluate quantitatively the impact of N-glycosylation at Asn596 on the protein stability and other biological activities, as the number of cDNA molecules in each cell vary among transiently transfected cells.

    In the present study, to examine the role of N-linked glycosylation at Asn596 on the protein stability of ABCG2, we used the Flp-In method to integrate cDNA of ABCG2 WT or cDNA of the ABCG2 N596Q variant into genomic DNA. The Flp recombinase system allowed one single copy of cDNA to be introduced into the FRT site at the telomeric region of one chromosome 12 in Flp-In-293 cells [23,32]. The Flp-In-293 cells used in the present study showed equal mRNA levels for both human ABCG2 WT and the ABCG2 N596Q variant (Fig. 3B). This was a fundamentally important requirement to study the impact of N-linked glycosylation at Asn596 on the ABCG2 protein expression levels. We report here that the N-linked glycan is not prerequisite for plasma membrane localization (Fig. 5A), supporting previous reports [28,29]. However, the present study provides the first direct evidence that disruption of N-linked glycosylation results in a reduced expression level of human ABCG2 protein (Fig. 3B). This has not been reported previously.

    Inhibition of N-linked glycosylation by tunicamycin

    In the ER, N-linked glycans are added to proteins as the ‘core oligosaccharide’ (Glc3Man9GlcNAc2), whereas the repertoire of oligosaccharide structures is still rather small. These glycans are therefore subjected to extensive modification as glycoproteins mature and move through the ER via the Golgi apparatus to their final destination, for example, the plasma membrane. In the present study, six inhibitors were used to inhibit N-linked glycosylation and the processing of N-linked glycans (Fig. 1A). Among them, tunicamycin was found to strongly reduce the protein level of human ABCG2 WT expressed in Flp-In-293 cells (Fig. 1B). Accordingly, after the treatment with tunicamycin, Flp-In-293 cells expressing ABCG2 WT became more sensitive to SN-38, compared with the untreated cells (Fig. 1C). Those observations are in accordance with the effect of disruption of N-linked glycosylation on the protein level of ABCG2 N596Q and the cellular resistance to SN-38 (Figs. 3B and 3D).

    During the synthesis of N-linked glycans in mammalian cells, the core oligosaccharide unit (Glc3Man9GlcNAc2) is assembled as a membrane-bound dolichol pyrophosphate precursor by enzymes located on both sides of the ER membrane. Tunicamycin inhibits the early steps of the synthesis of the dolichol pyrophosphate precursor (Fig. 1A). The nonglycosylated form of the ABCG2 protein was observed when Flp-In-293 cells were treated with tunicamycin (1, 2). In addition, after treatment with MG132, a monomeric form of ABCG2 was observed in Flp-In-293 cells treated with tunicamycin (Fig. 2D). Likewise, nonglycosylated and monomeric forms were observed for ABCG2 N596Q, when the cells were treated with MG132. It is suggested that N-linked glycosylation occurring at Asn596 facilitates, in part, dimer formation of ABCG2 protein through a cysteinyl disulfide bond at Cys603.

    Disruption of N-linked glycosylation enhances ubiquitin-mediated proteasomal degradation of human ABCG2

    The ER is the subcellular site where de novo-synthesized nascent proteins acquire their proper tertiary structures. The N-linked glycan appears to be important for maximal stability of the ABCG2 protein in the ER. We recently demonstrated that amino acid substitution in human ABCG2 could result in its ubiquitination and degradation [25,27], indicating that the ubiquitin-proteasome proteolytic pathway is involved in the degradation of human ABCG2. The N-glycosylated WT protein of ABCG2 is degraded in lysosomes, whereas misfolded mutant proteins undergo ubiquitin-mediated protein degradation in proteasomes. Furthermore, in the previous studies, we proposed that the ER has at least two checkpoints to monitor the quality of the human ABCG2 protein [25,27]. Namely, one checkpoint is the intramolecular disulfide bond between Cys592 and Cys608, and the other is the N-linked glycan at Asn596.

    The present study conveys experimental data demonstrating that the protein level of ABCG2 N596Q expressed in Flp-In-293 cells was significantly lower than that of ABCG2 WT (Fig. 3B), and that the N596Q variant protein was less stable than the WT protein when de novo protein synthesis was inhibited with cycloheximide (Fig. 3C). Interestingly, the proteasome inhibitor, MG132, increased the expression level of human ABCG2 N596Q protein in Flp-In-293 cells, whereas it had little effect on the expression level of ABCG2 WT protein (Fig. 4A). These observations indicate that the disruption of N-linked glycosylation leads to an increase of misfolded ABCG2 proteins and enhances their susceptibility to the ubiquitin-proteasome proteolytic pathway. Because the expression level of human ABCG2 N596Q protein was increased by treatments with BMA and MG132 (Fig. 4A), the variant protein appears to be degraded via both the lysosomal and ubiquitin-mediated proteasomal proteolytic pathways. Based on the results shown in Fig. 4, we assume that about half of the de novo synthesized N596Q variant proteins are sorted to the plasma membrane through the Golgi apparatus and then degraded in lysosomes, whereas the other half undergo ERAD (i.e. ubiquitin-mediated proteasomal proteolysis).

    Protein folding in cells is facilitated by several chaperone proteins that improve folding efficacy and minimize the occurrence of misfolded proteins in the crowded macromolecular environment of the ER. Many chaperone proteins recognize misfolded proteins and facilitate their degradation through the ubiquitin-proteasome system. There may be competing pathways for protein maturation and degradation intersecting in the ER [40]. In this regard, Wang et al. (2006) and Younger et al. (2006) recently provided new insights into the complexity of protein networks that govern the fate of the cystic fibrosis transmembrane conductance regulator, an apical membrane ABC transporter. Misfolding of the cystic fibrosis transmembrane conductance regulator is implicated in cystic fibrosis [41,42]. It is increasingly important to identify and characterize multiple chaperone proteins that control the folding, misfolding and degradation of membrane proteins.

    Concluding remarks

    In the present study, we provided evidence that N-linked glycosylation is crucial for regulating the stability of human ABCG2 protein and that the ER has a checkpoint for the state of N-linked glycosylation of human ABCG2. Disruption of the N-linked glycan at Asn596 enhanced the susceptibility of the ABCG2 protein to ubiquitin-mediated proteasomal proteolysis. Analysis of the molecular mechanism underlying the quality control of human ABCG2 protein will be the next important step that facilitates our understanding of the ERAD system. Because human ABCG2 is a membrane protein that forms intramolecular and intermolecular disulfide bonds and carries an N-linked glycan, it is a good probe with which to investigate the post-translational modification of ABC transporter proteins under pathological conditions, including cancer.

    Materials and methods

    Cells and biochemicals

    Flp-In-293 cells, antibiotic/antimycotic (100× concentrated), hygromycin B and Zeocin™ were purchased from Invitrogen Co. (Carlsbad, CA, USA). Dulbecco’s modified Eagle’s medium (4.5 g·L−1 of glucose) (DMEM) and protease inhibitor cocktail for general use were from Nacalai Tesque, Inc. (Kyoto, Japan). Tunicamycin from Streptomyces sp., MG132 and 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT reagent) were purchased from Sigma-Aldrich Co. (St Louis, MO, USA). Castanospermine and 1-deoxynojirimycin were from EMD Biosciences, Inc. (Darmstadt, Germany). Endo H and PNGase F were obtained from New England Biolabs, Inc. (Ipswich, MA, USA). BMA and SN-38, an active metabolite of irinotecan, were generously provided by Prof. Keiji Hasumi (Tokyo Noko University) and the Yakult Central Institute (Kunitachi, Tokyo, Japan), respectively. All other chemicals used were of analytical grade.

    Cell culture

    In the present study, we used Flp-In-293 cells (Invitrogen Co.), which have the FRT site at the telomeric region of only one of the pair of chromosomes 12 [23,32], to compare the characteristics between more than two cell lines (e.g. WT and mutant). For this purpose, we prepared Flp-In-293/ABCG2 N596Q cells, in addition to the Flp-In-293/ABCG2 WT and Flp-In-293/Mock cells that had been established in our previous studies [33]. Parental Flp-In-293 cells were maintained in DMEM supplemented with 10% (v/v) heat-inactivated fetal bovine serum (ICN Biomedicals, Inc. Aurora, OH, USA), 2 mm l-glutamine, 100 U·mL−1 of penicillin, 100 μg·mL−1 of streptomycin, 250 ng·mL−1 of amphotericin B and 100 μg·mL−1 of Zeocin™ at 37 °C in a humidified atmosphere of 5% (v/v) CO2 in air. By contrast, Flp-In-293/ABCG2 WT, Flp-In-293/ABCG2 N596Q and Flp-In-293/Mock cells were maintained in DMEM supplemented as described above, except that 100 μg·mL−1 of hygromycin B was used instead of Zeocin™.

    Generation of the N596Q variant form of human ABCG2

    The ABCG2-pcDNA5/FRT vector, constructed in our previous study [33], was used as the template for generation of the N596Q variant form of human ABCG2. The Asn596-encoding codon, AAT, was converted to CAA by using the QuikChange Site-directed Mutagensis Kit (Stratagene Co., La Jolla, CA, USA) according to the manufacturer’s protocol. The PCR reaction consisted of 98 °C for 10 s and 16 cycles of 98 °C for 10 s, 55 °C for 5 s, and 72 °C for 10 min, where Pfu Turbo DNA polymerase (Stratagene Co.) and the following PCR primers were used: 5′-TTCTGCCCAGGACTCCAAGCAACAGGAAACAATCCT-3′ and 5′-AGGATTGTTTCCTGTTGCTTGGAGTCCTGGGCAGAA-3′. The resulting sequence was examined to confirm the generation of the ABCG2 N596Q-pcDNA5/FRT vector.

    The cells expressing human ABCG2 N596Q (Flp-In-293/ABCG2 N596Q cells) were prepared as previously reported [33] by using ABCG2 N596Q-pcDNA5/FRT. After the transfection, single colonies resistant to 100 μg·mL−1 of hygromycin B were picked and subcultured.

    Preparation of cell lysates

    For immunoblot analysis, Flp-In-293 cells expressing human ABCG2 (1 × 106 cells) were placed into each well of a six-well culture plate (Becton Dickinson and Company, Franklin Lakes, NJ, USA) and precultured under the above-mentioned conditions for 24 h. The cells were then cultured in the presence or absence of compounds at different concentrations, and for designated periods of incubation time, as described in the figure legends. After the incubation, cells were harvested with culture medium by centrifugation and washed with ice-cold phosphate-buffered saline (NaCl/Pi). For analysis under reducing or nonreducing conditions, the cells were treated with lysis buffer A [50 mm Tris/HCl (pH 7.4), 1 mm dithiothreitol, 1% (v/v) Triton X-100, and protease inhibitor cocktail for general use (Nacalai Tesque, Inc., Kyoto, Japan)] or lysis buffer A [50 mm Tris/HCl (pH 7.4), 1% (v/v) Triton X-100, 20 mmN-ethylmaleimide and protease inhibitor cocktail for general use (Nacalai Tesque, Inc.)], respectively. The cell lysate samples were homogenized by passage, 10 times, through a 27-gauge needle. The homogenate was centrifuged (800 × g, 4 °C, 10 min) and the protein concentration of the resulting supernatant was determined using the Quick Start™ Bradford Dye Reagent (Bio-Rad Laboratories, Inc., Hercules, CA, USA) with BSA as a standard. For glycosidase treatments, the cell lysate sample containing 20 μg of protein was incubated with 20 U of Endo H or PNGase F at 37 °C for 10 min.

    For immunoprecipitation, Flp-In-293 cells expressing human ABCG2 (3 × 106 cells) were placed in a 100-mm culture dish, precultured for 24 h and then incubated in the presence or absence of 2 μm MG132 for 24 h. After the incubations, cells were harvested with culture medium by centrifugation, washed with ice-cold PBS and lysed in the immunoprecipitation buffer [20 mm Tris/HCl (pH 7.4), 150 mm NaCl, 1% (v/v) Triton X-100, 100 μm MG132, and protease inhibitor cocktail for general use (Nacalai Tesque, Inc.)]. The cell lysate samples were homogenized by passage, 20 times, through a 27-gauge needle. After centrifugation of the homogenate (800 × g, 4 °C, 10 min), the resulting supernatant was taken as a cell lysate sample and its protein concentration was determined (using the method outlined above).

    Immunoprecipitation

    Immunoprecipitation experiments were performed as described previously [25,27]. A sample of the cell lysate, containing 1 mg of protein, was taken and the volume of the incubation medium was adjusted to 500 μL with immunoprecipitation buffer (see above). The mixture was pretreated with 40 μL of antibody-free Preclearing Matrix E (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) at 4 °C for 2 h with gentle agitation. After the pretreatment, the mixture was centrifuged (17 800 × g, 4 °C, 1 min). The resulting supernatant was used for immunoprecipitation. The human ABCG2-specific mAb to human ABCG2 BXP-21 (Alexis Co., Lausen, Switzerland) (1 : 125 dilution), or anti-ubiquitin mAb (Millipore Corp., Billerica, MA, USA) (1 : 500 dilution), were incubated with Preclearing Matrix E in Tris-buffered saline (NaCl/Tris), containing 0.1% (w/v) BSA, at 4 °C for 1 h with gentle agitation, to form the antibody–matrix complex. After the incubation, the antibody–matrix complex and the matrix-treated cell-lysate sample were mixed and incubated at 4 °C overnight with gentle agitation. Then, the resulting immune complexes were washed four times with 20 mm Tris/HCl (pH 7.4), 150 mm NaCl, 0.05% (v/v) Tween 20 (TTBS). The immunoprecipitated protein was eluted by treatment with 45 μL of the SDS/PAGE sample buffer solution, containing 10% (v/v) 2-mercaptoethanol, at 4 °C for 20 min. The resulting elution was subjected to SDS/PAGE and immunoblot analysis.

    Immunoblot analysis

    For immunoblot analysis, cell lysate samples were first treated with the SDS/PAGE sample buffer solution, with or without 10% (v/v) 2-mercaptoethanol. Thereafter, sample proteins were electrophoretically separated on 7.5% (v/v) polyacrylamide gels and then electroblotted onto Hybond-ECL® (enhanced chemiluminescence) nitrocellulose membranes (GE Healthcare UK Ltd., Bucks., UK). The membrane was incubated in TTBS, containing 5% (w/v) skim milk, at 4 °C overnight.

    We used the human ABCG2-specific mAb BXP-21 (Alexis Co.) (1 : 200 dilution), mAb against rabbit muscle glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (American Research Products, Inc., Belmont, MA, USA) (1 : 1000 dilution), or anti-ubiquitin mAb (Millipore Corp.) (1 : 1000 dilution) as the first antibody, depending on the specific purpose of the immunoblot analyses. For the second antibody, we used anti-mouse IgG horseradish peroxidase (HRP)-linked antibody (Cell Signaling Technology, Inc., Beverly, MA) at a dilution of 1 : 3000. For immunoprecipitated samples, mouse ExactaCruz™ E-HRP (Santa Cruz Biotechnology, Inc.) was specifically used as a second antibody at a dilution of 1 : 1000.

    HRP-dependent luminescence was developed with Western Lightning Chemiluminescent Reagent Plus (PerkinElmer Life and Analytical Sciences, Inc., Boston, MA, USA) and detected using a Lumino Imaging Analyzer FAS-1000 (Toyobo, Osaka, Japan). Intensity of chemiluminescence was determined using Gel-Pro Analyzer Version 3.1.00.00 (NIPPON ROPER Co., Ltd., Tokyo, Japan). The protein expression levels of human ABCG2 were calculated from a standard curve, which was obtained as previously reported [43]. Data are expressed as mean values ± SD of three independent samples.

    MTT assay

    The MTT assay was performed to evaluate cellular resistance to SN-38. In brief, cells were seeded at 2 × 103 cells per well into 96-well culture plates and precultured for 24 h. To examine the effect of tunicamycin and castanospermine on the cellular resistance to SN-38, cells that had been precultured with 10 μg·mL−1 of tunicamycin or 1 mm castanospermine for 48 h at 1 × 106 cells per well in six-well culture plates were then seeded at 1 × 104 cells per well into 96-well culture plates. After 72 h of incubation with SN-38, the cells were treated with 500 μg·mL−1 of MTT reagent for 4 h and then 10% (w/v) SDS in NaCl/Pi (100 μL per well) was added to solubilize the MTT-formazan. After overnight incubation, the absorbances at 570 and 630 nm were measured as test and reference wavelengths, respectively. Data are expressed as mean values ± SD of quadruplicate cultures from one of three experiments.

    Detection of mRNA by RT-PCR

    Total RNA was extracted using NucleoSpin® RNA II (Macherey-Nagel GmbH & Co. KG, Duren, Germany) from Flp-In-293/ABCG2 WT and Flp-In-293/ABCG2 N596Q cells, according to the manufacturer’s instructions. First-strand cDNA was prepared from the extracted total RNA in a reverse transcriptase reaction with a High Capacity cDNA Archive Kit (Applied Biosystems, Lincoln Centre Drive, Foster City, CA, USA) and random hexamers as a primer, according to the manufacturer’s protocol. The first-strand cDNA from the mRNA of human ABCG2 and GAPDH were amplified by PCR in an iCycler™ thermal cycler (Bio-Rad Laboratories Inc., Hercules, CA, USA) using the following specific primer sets: human ABCG2 (5′-GATCTCTCACCCTGGGGCTTGTGGA-3′; 5′-TGTGCAACAGTGTGATGGCAAGGGA-3′); and GAPDH (5′-ACTGCCAACGTGTCAGTGGTGGACCTGA-3′; 5′-GGCTGGTGGTCCAGGGGTCTTACTCCTT-3′). The PCR reaction consisted of hot-start incubation at 94 °C for 2 min, and 30 cycles at 94 °C for 30 s, 59 °C for 30 s and 72 °C for 30 s. After the PCR, the resulting amplicons were separated by 1% (w/v) agarose-gel electrophoresis and detected with ethidium bromide under UV light.

    Quantitative real-time PCR

    The RNA levels of ABCG2 and GAPDH in Flp-In-293/ABCG2 WT cells and Flp-In-293/ABCG2 N596Q cells were determined in a 7500 Fast Real Time-PCR System (Applied Biosystems) using TaqMan® Fast Universal Master Mix (Applied Biosystems) and TaqMan® probes (ABCG2, Hs00184979_m1; GAPDH, Hs99999905_m1) (Applied Biosystems). The expression levels of ABCG2 were normalized against those of GAPDH.

    Immunofluorescence analysis for expression levels and distribution of intracellular ABCG2

    ABCG2-expressing Flp-In-293 cells (2 × 104 cells) were seeded into 96-well SensoPlates™ (Greiner Bio-One Co., Ltd., Tokyo, Japan) in which the bottom of each well had been coated with 1.6 μg per well of mouse collagen type IV (Becton Dickinson and Company). These cells were precultured for 24 h under the above-mentioned culture conditions and were then incubated for 24 h in the presence or absence of 2 μm MG132. Thereafter, cells were fixed with 4% paraformaldehyde in NaCl/Pi at room temperature for 20 min. To block free aldehyde groups in the formaldehyde, cells were treated with NaCl/Pi containing glycine (10 mg per mL) at room temperature for 10 min followed by a further incubation with NaCl/Pi containing 0.5% (w/v) BSA at room temperature for 1 h. To detect the ABCG2 protein, cells were treated initially with the ABCG2-specific mAb BXP-21 (1 : 1000 dilution; Alexis Co.,), which recognizes an epitope (amino acids 271-396) in the intracellular loop of ABCG2, or with 5D3 (1 : 1000 dilution; R&D Systems, Inc., Minneapolis, MN), which recognizes an epitope in the extracellular loop of ABCG2 [34,35], as the first antibody and subsequently with Alexa Fluor 488-conjugated anti-mouse IgG (1 : 1000 dilution; Invitrogen). In the same preparations, nuclear DNA was stained with Hoechst 33342 (1 μg·mL−1; Invitrogen) in NaCl/Pi containing 0.5% (w/v) BSA.

    To detect the whole ABCG2 protein expressed in these cells, cell membranes were permeabilized by incubation with NaCl/Pi containing 0.02% (v/v) Triton X-100 at room temperature for 5 min before treatment with glycine. To selectively detect the ABCG2 protein localized on the plasma membrane of these cells, we did not treat the cell membranes with 0.02% (v/v) Triton X-100 (in order to keep the plasma membrane impermeable to the antibody).

    Immunofluorescence of the Flp-In-293 cells was detected using CellVoyager (Yokogawa Electric Corp., Kanazawa, Japan), a newly developed system of confocal fluorescence microscopy. The fluorescence signals of Alexa Fluor 488 and Hoechst 33342 were observed with an excitation laser light at 488 and 405 nm, respectively. To analyze, quantitatively, the levels of ABCG2 protein expressed in cells, an immunofluorescent cluster (Alexa Fluor 488), with a diameter of 0.5 to 10 μm, was counted as one ABCG2-positive dot and the intensity of each dot was then measured. Such digital counting of ABCG2-positive dots and their intensity measurements were performed for a total of 300 cells for each immunofluorescence preparation, as previously described [25,26]. The resulting data were accumulated and then statistically analyzed. The average fluorescence intensity per cell was calculated as the mean values of the signal intensities accumulated over the ABCG2-positive dots in a total of 300 cells.

    Statistical analysis

    Statistical analyses were performed using microsoft excel 2003 software (Microsoft Co., Redmond, WA, USA). The statistical significance of differences was determined according to the Student’s t-test. P-values of < 0.05 were considered statistically significant.

    Acknowledgements

    We thank Drs Kenta Mikuriya, Takayoshi Matsubara, and Satoshi Kometani (Yokogawa Electric Corporation) for generous support in fluorescence-microscopic observations. In addition, our thanks go to Dr Masako Osumi (Integrated Imaging Research Support), Dr Naoyuki Taniguchi (Osaka University), Dr Tadashi Suzuki (Osaka University) and Dr Katsuko Yamashita (Tokyo Institute of Technology) for their fruitful discussions. This study was supported by the Japan Science and Technology Agency (JST) project ‘Development of the world’s fastest SNP detection system’ as well as Grant-in-Aid for Scientific Research (A) (no. 18201041), Grant-in-Aid for Exploratory Research (no. 19659136) from the Japanese Society for the Promotion of Science (JSPS), and Grant-in-Aid for Young Scientists (B) (no. 19791361) from the Ministry of Education, Culture, Sports, Science and Technology. Kanako Wakabayashi-Nakao, Ai Tamura, Yu Toyoda, and Shoko Koshiba are JSPS research fellows.