Metabolites of ginsenosides as novel BCRP inhibitors

www.elsevier.com/locate/ybbrc

Biochemical and Biophysical Research Communications 345 (2006) 1308–1314

BBRC

Metabolites of ginsenosides as novel BCRP inhibitors

Jing Jin a,*,1, Sanjay Shahi a,2, Hee Kyoung Kang b, Hendrik W. van Veen a, Tai-Ping Fan a,*

a Department of Pharmacology, University of Cambridge, Cambridge, CB2 1PD, UKb Department of Medicine, Cheju National University, Ara-1, Jeju 690-756, South Korea

Received 23 April 2006Available online 4 May 2006

Abstract

We have previously shown ginsenosides derived from Panax ginseng exert opposing effects on angiogenesis. Here, we examined pro-topanaxadiol-containing ginsenosides (Rg3, Rh2, and PPD) and protopanaxatriol-containing ginsenosides (Rg1, Rh1, and PPT) aspotential inhibitors of breast cancer resistance protein (BCRP). Among these ginsenosides, metabolites Rh2, PPD, and PPT significantlyenhanced the cytotoxicity of mitoxantrone (MX) to human breast carcinoma MCF-7/MX cells which overexpress BCRP. PPD was themost potent followed by Rh2 and PPT. This effect was not seen in sensitive MCF-7 cells. Rg3, Rg1, and Rh1 were ineffective in eitherMCF-7 or MCF-7/MX cells. PPD, Rh2, and PPT were able to inhibit MX efflux in MCF-7/MX cells. PPD and Rh2 also increased MXuptake. In inside out membrane vesicles from Lactococcus lactis cells expressing BCRP, only PPD was found to significantly inhibitBCRP-associated vanadate sensitive ATPase activity. These results indicate that metabolites PPD, Rh2, and PPT were inhibitors ofBCRP.� 2006 Elsevier Inc. All rights reserved.

Keywords: Breast cancer resistance protein; P-glycoprotein; Ginsenosides; PPD; Rh2

The multidrug resistance phenotype is often associatedwith overexpression of members of the ATP-binding cas-sette (ABC) transporter family. They are plasma mem-brane proteins that can actively extrude a wide variety ofstructurally diverse anticancer agents, thereby reducingintracellular drug concentrations.

P-gp and multidrug resistance-associated protein (MRP)are two of the most extensively studied ABC transporters.An additional member of the ABC transporter family,found recently, is BCRP, also known as MXR, ABCP, orABCG2, a 655-amino-acid protein encoded by the abcg2

gene located on chromosome 4q22 [1]. Overexpression ofBCRP was shown to confer resistance against variousclinically relevant compounds, e.g., mitoxantrone, metho-

0006-291X/$ - see front matter � 2006 Elsevier Inc. All rights reserved.

doi:10.1016/j.bbrc.2006.04.152

* Corresponding authors. Fax: +44 1223 334 100.E-mail address: [email protected] (T.-P. Fan).

1 Current address: Welsh School of Pharmacy, Cardiff University,Cardiff, CF10 3XF, UK.

2 Current address: School of Biomedical and Molecular Sciences,University of Surrey, Guildford, GU2 7XH, UK.

trexate, topotecan, SN38, and flavopiridol. In addition,several cytological dyes such as rhodamine 123, Lysotrac-ker Green, and BBR3390, as well as the fluorescentconjugate BODIPY-prazosin, demonstrated decreasedaccumulation in BCRP-overexpressing cells. Thus, BCRPmight be an important addition to potential sources of clin-ical multidrug resistance. It is interesting that BCRP isphysiologically expressed in the canalicular membrane ofthe liver, in the epithelia of small intestine, colon, lung, kid-ney, adrenal, and sweat glands, as well as in the endotheliaof veins and capillaries. The tissue distribution of BCRPdemonstrated very extensive overlap with that of P-gp,which suggests BCRP might have an important role inthe pharmacological handling of drugs in animals andhumans. Indeed, inhibition of BCRP in mice markedlyincreases the oral availability to topotecan, slows its elimi-nation from the body, and increases its penetration of theplacental barrier to the foetus [2]. Inhibitors of BCRPtherefore have the potential to improve such aspects of che-motherapy by modulating the pharmacokinetic behaviourof drugs.

J. Jin et al. / Biochemical and Biophysical Research Communications 345 (2006) 1308–1314 1309

Ginsenosides are among the active ingredients of Gin-seng, the root of which has been used in traditional herb-al remedies/medicine in Eastern Asia for over 2000 years.To date, over 30 ginsenosides have been isolated, eachwith a different set of properties. Ginsenosides belongto the steroidal saponin family [3]. They have a four-trans-ring rigid steroid structure with a modified side-chain at C20, and most ginsenosides also have a sugarmoiety at C3. It is the type, site of attachment (C3, C6or C20), and number of sugars that give structural vari-ation between the ginsenosides. Within the ginsenosidefamily, there are the protopanaxadiol-containing (e.g.,Rg3, Rh2, and PPD) and the protopanaxatriol-contain-ing ginsenosides (e.g., Rg1, Rh1, and PPT) (Fig. 1). Gin-senoside Rg3 has been reported to reverse P-gp-mediatedmultidrug resistance [4]. Protopanaxatriol ginsenosidesshowed reversal activity of multidrug resistance in dau-norubicin- and doxorubicin-resistant acute myelogenousleukaemia AML-2 cells [5]. A very useful finding wasthe demonstration by de Bruin et al. [6] thatGF120918, a highly efficient P-gp inhibitor that couldreadily be used in animals and patients, was also quiteeffective in inhibiting BCRP. Moreover, reserpine couldeffectively reverse drug resistance and increase cellulardrug accumulation in BCRP-expressing cells [7]. So itis interesting to investigate if ginsenosides could inhibitBCRP. This paper aims to determine whether ginseno-sides can modulate BCRP-mediated multidrug resistancein MCF-7 cells, and whether there is a difference betweenthe efficacy of protopanaxadiol-containing (Rg3, Rh2,and PPD) and protopanaxatriol-containing (Rg1, Rh1,and PPT) ginsenosides.

Fig. 1. The structures of protopanaxadiol ginsenosides Rg3, Rh

Materials and methods

Cell lines and ginsenosides. MCF-7 breast carcinomas cells and BCRP-overexpressed subline MCF-7/MX [8] were generously provided by Dr.Margery Barrand in University of Cambridge. MCF-7/Adr cells over-expressing P-gp were kindly gifted by Dr. Timothy Grant from Universityof Leicester [9]. These cell lines were incubated in RPMI 1640 mediumsupplemented with 10% foetal bovine serum, 100 U/mL penicillin, and100 lg/mL streptomycin in 5% CO2 at 37 �C. The resistant MCF-7/MXand MCF-7/Adr cells were cultured in the presence of 80 nmol/Lmitoxantrone (MX) and 0.5 lmol/L doxorubicin (Dox), respectively, untilat least 3 days before starting of the experiments. All ginsenosides used inthis study were obtained from Canfo Natural Products (Chengdu, China).

RNA extraction and RT-PCR. Total cellular RNA was extracted usingTRIzol reagent (Sigma) according to the manufacturer’s protocol andquantified spectrophotometrically. First-strand cDNA synthesis was car-ried out with 1 lg RNA by using SuperScript II reverse transcriptase(Invitrogen). PCR were performed in 25 ll volumes containing 1 ll RTcDNA product, 1.5 mmol/L MgCl2, 0.8 mmol/L dNTP, 0.3 lmol/L ofeach primer, 800 lmol/L dNTPs, and 5 U Taq polymerase (Invitrogen).Reactions were amplified for a total of 33 cycles on the MBS PCR system(Thermo, Milford, MA) using 94 �C for denaturation (30 s), 55 and 57 �Cfor annealing for BCRP and mdr1, respectively (30 s), and 72 �C forextension (2 min), followed by a further 10 min extension. The primerpairs used were 5 0-TTA TCC GTG GTG TGT CTG GA-3 0 and 5 0-CCTGCT TGG AAG GCT CTA TG-3 0 for BCRP (amplify a fragment of429 bp), 5 0-GCT CCT GAC TAT GCC AAA GC-3 0 and 5 0-GCT GCCCTC ACA ATC TCT TC-3 0 for mdr1 (amplify a fragment of 464 bp), or5 0-ACG TTA TGG ATG ATG ATA TCG-30 and 5 0-CTT AAT GTCACG CAC GAT TTC-3 0 for b-actin (amplify a fragment of 644 bp) as aninternal control. After amplification, 40% of the product was separated ona 2% (w/v) agarose gel, stained with 0.5 lg/mL ethidium bromide, andphotographed under UV light.

MTT assay. The in vitro chemosensitivity was measured by dimethylthiazolyl-2,5-diphenyltetrazolium bromide (MTT) assay. Briefly, tumourcells were inoculated into 96-well plates (Coster) in 100 ll medium per welland allowed to attach and grow for 24 h. To determine the antiprolifer-ative effect of ginsenosides, various concentrations of ginsenosides diluted

2, PPD and protopanaxatriol ginsenosides Rg1, Rh1, PPT.

Fig. 2. mRNA level of mdr1 and BCRP in three MCF-7 cell lines. Totalcellular RNA was extracted using TRIzol reagent and the mRNAexpression was determined using RT-PCR as described in Materials andmethods.

Table 1Effects of ginsenosides on the chemosensitivities of MCF-7 and MCF-7/MX cells to MX

Concentration (lM) IC50 of MX (lM)

MCF-7 MCF-7/MX

Control 0.32 ± 0.04 92.79 ± 10.70Rg3 25 0.22 ± 0.02 (1.4) 104.62 ± 62.30 (0.9)

100 0.41 ± 0.28 (0.8) 37.15 ± 6.38 (2.5)Rh2 5 0.22 ± 0.11 (1.4) 13.58 ± 9.19 (6.8)

20 0.16 ± 0.02 (2.0) 1.62 ± 0.57 (57.3)PPD 5 0.18 ± 0.09 (1.8) 13.59 ± 3.23 (6.8)

20 0.31 ± 0.12 (1.0) 1.13 ± 0.15 (82.1)Rg1 100 0.31 ± 0.12 (1.0) 49.86 ± 12.83 (1.9)Rh1 50 0.25 ± 0.08 (1.3) 74.89 ± 13.93 (1.2)PPT 5 0.30 ± 0.09 (1.1) 25.24 ± 2.20 (3.7)

20 0.23 ± 0.10 (1.4) 3.50 ± 1.59 (26.5)Res 5 0.17 ± 0.01 (1.9) 1.95 ± 0.39 (47.6)

Exponentially growing cells were treated with MX in the presence orabsence of the indicated concentrations of ginsenosides for 72 h. Cellsurvival rate was analysed by MTT assay. IC50s were calculated by themedian effect software and results are presented as means ± SD of threedifferent experiments. The data in parentheses are CFs, which were cal-culated by IC50 MX divided by IC50 MX + modulator. Reserpine (Res) 5 lmol/L was used as a positive control.

1310 J. Jin et al. / Biochemical and Biophysical Research Communications 345 (2006) 1308–1314

with the medium were added into the wells. To determine the reversal effectof ginsenosides, graded concentrations of mitoxantrone or doxorubicinwith or without ginsenosides were added. Then the cells were exposed todrugs for 3 days at 37 �C, after which, MTT (Sigma, 0.5 mg/mL diluted inRPIM1640) was added to each well and cultured for an additional 4 h. Theformed formazan was dissolved in 150 ll of dimethyl sulphoxide afteraspirating the culture medium. The plates were shaken mechanically for1 min and the optical density of each well was immediately read on amicroplate reader (multiskan model 354, Labsystems, Finland) at a wave-length of 540 nm. Results are expressed as IC50 which is analysed byTWODRUGS program based on a median–effect plot. The reversal activityof ginsenosides on MX resistance was expressed as the chemosensitizingfolds (CFs), which were calculated according to the following equation:CFs = IC50 anticancer drug alone/IC50 anticancer drug + modulator.

Flow cytometry. To determine drug efflux, cells were incubated with5 lmol/L MX and 1 lmol/L Dox in serum-free medium for 30 and150 min, respectively. Cells were then washed with ice-cold PBS to removeMX/Dox-containing medium, after which cells were incubated with gin-senoside or vehicle (0.1% DMSO) for 90 min. Efflux was stopped by pel-leting the cells and resuspending them in ice-cold PBS. The cells were kepton ice until analysis. To determine drug uptake, the cells were preincu-bated in culture medium containing ginsenosides or vehicle (0.1% DMSO)for 30 min at 37 �C. Then 5 lmol/L MX and 1 lmol/L Dox were addedfor a further 40 and 120 min incubation, respectively, after which cellswere washed twice with ice-cold PBS and kept on ice until analysis.Measurements of cellular MX fluorescence were made on a FACScan flowcytometer (Becton–Dickinson, Mountain View, CA) with excitation at488 nm and the emission recorded via 670 nm long pass (FL3) filter.Intracellular accumulation of Dox was measured with excitation at488 nm and emission at 585 nm (FL2 filter). The mean fluorescencemeasurements were analysed with the Summit Offline software programand used as a quantitative measure of intracellular MX and Dox contentin MCF-7 cells. Resperine and nifidepine were used as positive controls forMX and Dox accumulation, respectively.

ATPase assay. Because of well-well variation and cell clumping it wasdifficult to compare the ATPase values for different cell lines. Instead wedetermined the ATPase values from Lactococcus lactis cells expressingBCRP. BCRP expression and characterization including ATPase assayhave been previously described [10,11]. Lactococcal inside out membranevesicles containing BCRP R482 were prepared as described previously[10,11]. A typical reaction of ATPase assay contained 5 lg protein,5 mmol/L ATP; in a total volume of 100 ll made up with 50 mmol/LHepes containing 5 mmol/L MgSO4 (pH 7). All the components wereadded and mixed while keeping the reaction tube on ice. The ATPasereactions were carried out at 30 �C for 5 min in the presence/absence ofdifferent concentrations of ginsenosides followed by freezing the samplesimmediately in ethanol-dry ice. Appropriate controls were included totest the Pi released from the buffer and ATP solution. The samples werediluted 1:3 with Hepes buffers from which 30 ll aliquoted into 96-wellplate kept on ice. Then 150 ll activated malachite green solution wasadded, which was prepared by mixing a solution of malachite green(17 mg/3.75 ml water) and ammonium molybdate (0.525 g/12.5 ml of 4 NHCl) to get the final volume to 50 ml with water and activated by adding1:100 of 10% Triton X-100. The plate was removed from the ice andincubated for 5 min at room temperature. Finally 75 ll of 34% (w/v)citric acid was added in each well, and the plate was further incubatedfor 45 min at 30 �C for colour development. The plate was read at600 nm. A solution of Pi containing 0–5 nmole was used as standard forcomparing and calculating the Pi released by the enzymatic reaction ofthe proteins.

Results

Expression of MDR-related proteins

The mRNA level of BCRP and mdr1 was determined byRT-PCR. MCF-7/MX cells overexpressed BCRP and

MCF-7/Adr cells overexpressed mdr1 as compared to theparental MCF-7 cells (Fig. 2). A very faint band of BCRPwas also seen in MCF-7/Adr cells. When MCF-7/MX andMCF-7/Adr cells were treated with ginsenosides, no chang-es in mRNA level of BCRP or mdr1 took place (data notshown). In addition, ginsenosides seemed not to affect theexpression of BCRP or P-gp at the protein level determinedby Western blot (data not shown).

Modulation of drug resistance

As shown in Table 1, Rg3 at 100 lmol/L had a littleeffect on the IC50 of MX in both MCF-7 and MCF-7/MX cells. Neither Rg1 nor Rh1 enhanced the sensitivityof MCF-7/MX or MCF-7 cells to MX. However, PPD20 lmol/L significantly enhanced the antiproliferationaction of MX to MCF-7/MX cells, whereas this effectwas not seen in sensitive MCF-7 cells. Rh2 and PPT also

J. Jin et al. / Biochemical and Biophysical Research Communications 345 (2006) 1308–1314 1311

partly reversed the resistance of MCF-7/MX cells to MX,but not as powerfully as PPD. All the concentrations usedwere non-toxic. When MCF-7/Adr cells were tested, thechemosensitizing effect on Dox resistance was not obviousin the treatments of ginsenosides at their maximum non-toxic concentrations (Fig. 3). Only PPD at 20 lmol/Lshowed mild activity to reverse Dox resistance.

Inhibition of MX efflux by ginsenosides

In order to determine whether the reversal activities ofginsenosides in MCF-7/MX cells might be associated withdecreased drug transport by BCRP, MX efflux was exam-

Fig. 3. Effects of ginsenosides on the growth inhibition activity of Dox in MCabsence of Rg3, Rh2, PPD (A) and Rg1, Rh1, PPT (B) for 72 h. Cell surviva

Fig. 4. MX accumulation in MCF-7 cells. (A) Effect of ginsenosides on MX efflMX for 30 min, washed with PBS, and then resuspended in the medium contaifor 90 min. The retention of MX was measured by flow cytometry at FL3. *P <ginsenosides on the intracellular uptake of MX in MCF-7 and MCF-7/MX cell(in lmol/L) for 30 min then 5 lmol/L MX was added. After the incubation f*P < 0.05, **P < 0.01 vs DMSO vehicle control of respective cell line.

ined in both sensitive and resistant MCF-7 cells by measur-ing MX retention using flow cytometry. MX retention wasremarkably lower in MCF-7/MX cells than in their paren-tal MCF-7 cells (Fig. 4A), indicating the ability of MCF-7/MX cells to extrude MX. Ginsenosides PPD, Rh2, andPPT potently increased MX retention in resistant MCF-7/MX cells, in which the potency order at 20 lmol/L wasPPD > Rh2 > PPT, in accordance with their reversal activ-ities determined by MTT assay. Rg3 at 100 lmol/L alsoshowed a similar potency as PPT. However, these effectswere not seen in sensitive MCF-7 cells. Moreover, MXretention did not increase in the presence of Rg1 or Rh1in either MCF-7 cells or MCF-7/MX cells.

F-7/Adr cells. MCF-7/Adr cells were treated with Dox in the presence orl rate was analysed by MTT assay.

ux in MCF-7 and MCF-7/MX cells. Cells were incubated with 5 lmol/Lning ginsenosides (the concentration as indicated, respectively, in lmol/L)

0.05, ***P < 0.001 vs DMSO treatment of MCF-7/MX cells. (B) Effect ofs. The cells were preincubated with ginsenosides at indicated concentrationor 40 min, intracellular content of MX was measured by flow cytometry.

Fig. 6. Effect of PPD, Rh2, and PPT on ATPase activity of BCRP.*P < 0.05, **P < 0.01 compared with basal activity of BCRP. Inside outmembrane vesicles with BCRP from L. lactis were used and the reaction ofATPase assay based on Pi release was performed as described in Materialsand methods. A solution of Pi containing 0–5 nmole was used as standardfor comparing and calculating the Pi released by the enzymatic reaction ofthe proteins.

1312 J. Jin et al. / Biochemical and Biophysical Research Communications 345 (2006) 1308–1314

Enhancement in MX uptake by ginsenosides

The uptake rate for MX is similar between MCF-7 andMCF-7/MX cells. Significant increased cellular accumula-tion of MX was observed in the presence of PPD andRh2 in MCF-7/MX cells, in which PPD was still morepotent, but not in MCF-7 cells (Fig. 4B). Rg1 and Rh1showed marginal effects on MX uptake in MCF-7 orMCF-7/MX cells. The uptake of MX did not change obvi-ously in PPT treatment. However, it is surprising that Rg3was the most efficacious to increase the uptake of MX byresistant MCF-7/MX cells, which was also significant insensitive MCF-7 cells.

Dox accumulation in MCF-7/Adr cells

Compared with sensitive MCF-7 cells, MCF-7/Adr cellshave a deficiency in the retention of Dox. But none of theginsenosides inhibited Dox efflux in MCF-7/Adr cells (datanot shown). In the assay to determine Dox uptake, PPT20 lmol/L augmented the ability of MCF-7/Adr cells touptake Dox, so did PPD 20 lM but to a less extent(Fig. 5). Other ginsenosides showed a marginal effect onthe entry of Dox into MCF-7/Adr cells.

Effect of PPD, Rh2, and PPT on BCRP ATPase activity

BCRP transports its substrates by using the energy fromATP binding and hydrolysis [12]. Since PPD, Rh2, and

Fig. 5. Effects of ginsenosides on Dox uptake in MCF-7/Adr cells. Cellswere preincubated with ginsenosides at the concentrations indicated inlmol/L for 30 min, after which 1 lmol/L Dox was added for a further120 min incubation. Measurements of cellular Dox fluorescence wereperformed on a FACScan flow cytometer at FL2 filter. Nifedipine at50 lmol/L was used as a positive control. **P < 0.01, ***P < 0.001 vsDMSO vehicle control.

PPT inhibited the efflux of MX in MCF-7/MX cells with-out significant effect on the efflux of Dox in MCF-7/Adrcells, the effect of ginsenosides on ATPase activity of BCRPwas investigated. ATPase activities of human BCRPexpressed in L. lactis were measured in presence of theseginsenosides over a concentration range of 0–50 lmol/L.The BCRP-associated vanadate sensitive ATPase activityonly in presence of 20 and 50 lM PPD was significantlydifferent from the basal activity (P < 0.05 and P < 0.01,respectively, n = 4) (Fig. 6). However, the ATPase activityat 5–10 lM did not alter with a statistical significance.

Discussion

Ginsenosides Rg3, Rh2, and PPD used in this studybelong to the protopanaxadiol family with sugar moietiesattached on C-3. It has been reported that Rg3 is metabo-lized to Rh2 and ultimately PPD by intestinal bacteria [13].Ginsenosides Rg1, Rh1, and PPT are members of proto-panaxatriol family. Rg1 is converted into PPT via Rh1 ina similar metabolic way to Rg3 [14]. We have previouslyshown ginsenosides exert opposing effect on angiogenesis[15]. In this study, we characterized ginsenosides PPD,Rh2, and PPT that are potent inhibitors of BCRP, with arank order of potency PPD > Rh2 > PPT. All of themshowed little effect on the activity of P-gp. Neither Rg1nor Rh1 was observed to inhibit BCRP or P-gp. Rg3 wasdemonstrated to be a mild inhibitor of BCRP and P-gp.

All ginsenosides possess a four-trans-ring rigid steroidskeleton, so called steroidal saponins, which share structur-al features with those of classical steroid hormones. Thesite of attachment of hydroxyl groups and sugar moietyhas been shown to influence their biological activities.

J. Jin et al. / Biochemical and Biophysical Research Communications 345 (2006) 1308–1314 1313

For example, ginsenoside aglycone PPD and PPT differonly by the binding site of hydroxyl group, C-6 for PPTbut C-3 for PPD. Rh1 and Rh2 are structurally similar,except for the binding site of the b-D-glucopyranosyl groupat C-6 and C-3, respectively. The substitution at C-6 of gin-senosides seems to confer antitumour activities. It has beenreported that much higher concentration of PPT is neededto induce apoptosis compared to PPD in several cancer celllines [3]. Odashima et al. found that Rh2, but not Rh1,decreased the growth of B16-Bl6 melanoma cells and stim-ulated cell-to-cell adhesion [16]. In our study, PPD reversedBCRP-mediated resistance to MX and inhibited MX effluxin MCF-7/MX cells, whereas PPT showed a weak action.Rh2 was efficacious to inhibit BCRP, however, Rh1 wasnot. These results suggest that C-6 substitution was alsoimportant to the potency of ginsenosides to inhibit BCRP.Any substitution would abolish or at least weaken theBCRP-inhibitory activity of ginsenosides. In contrast, aglucose replacement at C-3 position weakens the inhibitoryeffect of ginsenosides not only on BCRP in this study butalso on the proliferation of tumour cells from other studies.Two glucose substitutions on C-3 position such as Rg3nearly abolished its effect to inhibit BCRP. A much higherconcentration of Rg3 was needed to achieve just a mildinteraction with BCRP.

The decrease of intracellular accumulation of cytotoxicdrugs, apart from the increased drug efflux by MDR-relat-ed proteins, could be dependent on a slower drug uptakethrough the plasma membrane [17]. Hydrophobic interac-tions, membrane fluidity, and drug lipophilicity are decisivefor efficient intracellular uptake of anticancer drugs. Sever-al studies revealed reduced fluidity of membrane for multi-drug resistant cells compared to parental lines [17,18].When MCF-7/MX cells were incubated with Rg3, MXuptake by cells was prominent; whereas MCF-7/Adr cellsaccumulate the most Dox when exposed to PPT. PPDaffected both MX uptake by MCF-7/MX cells and Doxuptake by MCF-7/Adr cells. It is believed that the steroidskeleton of ginsenoside endows the whole molecule witha favoured structure to allow, for example, insertion intomembranes leading to changes in membrane fluidity. Inaddition, the different lipophilicity of ginsenosides due tothe various substitutions and hydrophobic interactionswith membrane lipid probably affected MX and Doxinflux. Rg3 and PPT induced distinct effects on drug uptakein MCF-7/MX cells and MCF-7/Adr cells indicating thepossible difference in membrane properties of the two resis-tant MCF-7 cells.

Cholesterol is an intrinsic membrane lipid, which sharesthe steroid backbone and amphipathic nature of ginseno-sides. Membrane proteins are found to be localized selec-tively in cholesterol-rich domains or in cholesterol-poordomains. Cholesterol enrichment has an inhibitory effecton many membrane ATPases, and it may directly interactwith the boundary lipids of ATPase and alter the intermo-lecular hydrogen bonds of the protein [19]. In order to ruleout the possible involvement of membrane alterations

induced by PPD, Rh2, and PPT on ATPase activities ofBCRP, inside out membrane vesicles were prepared andchosen to investigate the effects of ginsenosides on ATPaseactivity. It was observed that PPD stimulated the BCRP-associated ATPase activity in a dose-dependent manner,but this was not the case for Rh2 or PPT. A number of ste-rol hormones, e.g., estrone and 17b-estradiol as well asestrogen antagonists such as tamoxifen and toremifeneinhibited the BCRP-mediated drug efflux and overcamedrug resistance [20,21]. Similar to our previous observa-tion, the BCRP-associated ATPase activity was significant-ly stimulated by steroids including cholesterol andestradiol, natural steroids such as progesterone and testos-terone as well as tamoxifen [10,11]. These findings suggestBCRP may transport sterols, which competitively inhibitsthe transport of anticancer drugs such as MX by BCRP.Among the three sterol ginsenosides which showedBCRP-inhibiting activity in this study, only PPD is proba-bly transported by BCRP suggested by its stimulation ofBCRP-associated ATPase activity. Based on the fact thatBCRP transports its substrate in an ATP-dependent man-ner, neither Rh2 nor PPT could be transported by BCRPsince no stimulation of the BCRP-ATPase activity wasobserved. However, interpretation of ATPase activity datahas to be done with care as stimulation of the BCRP-asso-ciated ATPase activity could not be directly correlated withactual transport of certain substrates [22,23]. Direct proofthus is needed to confirm this, for example, by using iso-tope-labelled ginsenosides.

Although very few clinical data are available up to now,the widespread expression of BCRP suggests it could con-tribute to either innate or acquired resistance of tumours toanticancer drugs in a manner analogous to P-gp. A fewclinical studies have demonstrated the presence of a smallpopulation of cells within solid tumours biopsies that havestem cell characteristics. The percentage of this populationof cells correlates with the aggressiveness of the tumourphenotype [24]. The eradication of non-cancer stem cellsmay result in tumour remission, and the disease will relapseif the cancer stem cells are not eliminated. In view of therole for BCRP as a phenotypic marker and functional reg-ulator of stem cells, inhibitors of BCRP, such as PPD, canbe beneficial in the treatment of cancer, apart from the cir-cumvention of drug resistance.

In conclusion, we have shown here that ginsenosidesinteracted directly with BCRP, but not P-gp in MCF-7cells. Central structures of ginsenosides, either PPD orPPT, reversed MX resistance and inhibited MX efflux med-iated by BCRP, and enhanced uptake of MX, in particularPPD showed the potency to a much more extent and addi-tionally stimulated ATPase activity. Rh2 with a sugar moi-ety attached to C-3 position still showed BCRP-inhibitingactivity. However, more sugars, e.g., Rg3 and sugar moietyon C-6 position (e.g., Rg1 and Rh1) abolished any interac-tion with BCRP. Hydroxyl group and sugar moietiesattaching to different positions result in different roles inBCRP reversal activity. These findings should assist the

1314 J. Jin et al. / Biochemical and Biophysical Research Communications 345 (2006) 1308–1314

development of ginsenosides as chemosensitizing agentsand contribute to the design of more effective and saferagents to reverse multidrug resistance.

Acknowledgment

This work was supported by Taiwan to NuLiv Science,Taiwan and the Association for International CancerResearch (AICR).

References

[1] L.A. Doyle, D.D. Ross, Multidrug resistance mediated by the breastcancer resistance protein BCRP (ABCG2), Oncogene 22 (2003) 7340–7358.

[2] J.W. Jonker, J.W. Smit, R.F. Brinkhuis, M. Maliepaard, J.H.Beijnen, J.H.M. Schellens, A.H. Schinkel, Role of breast cancerresistance protein in the bioavailability and fetal penetration oftopotecan, J. Natl. Cancer. Inst. 92 (2000) 1651–1656.

[3] D.G. Popovich, D.D. Kitts, Structure–function relationship exists forginsenosides in reducing cell proliferation and inducing apoptosis inthe human leukemia (THP-1) cell line, Arch. Biochem. Biophys. 406(2002) 1–8.

[4] S.W. Kim, H. Kwon, D.W. Chi, J.H. Shim, J.D. Park, Y.H. Lee, S.Pyo, D.K. Rhee, Reversal of P-glycoprotein-mediated multidrugresistance by ginsenoside Rg(3), Biochem. Pharmacol. 63 (2003) 75–82.

[5] C.H. Choi, G. Kang, Y.D. Min, Reversal of P-glycoprotein-mediatedmultidrug resistance by protopanaxatriol ginsenosides from Koreanred ginseng, Planta Med. 69 (2003) 235–240.

[6] M. De Bruin, K. Miyake, T. Kitman, R. Robey, S.E. Bates, Reversalof drug resistance by GF120918 in cell lines expressing the ABC half-transporter, MXR, Cancer Lett. 146 (1999) 117–126.

[7] S. Zhou, J.D. Schuetz, K.D. Bunting, A.M. Colapietro, J. Sampath,J.J. Morris, I. Lagutina, G.C. Grosveld, M. Osawa, H. Nakauchi,B.P. Sorrentino, The ABC transporter Bcrp1/ABCG2 is expressed ina wide variety of stem cells and is a molecular determinant of the side-population phenotype, Nat. Med. 7 (2001) 1028–1034.

[8] H.C. Cooray, T. Janvilisri, H.W. van Veen, S.B. Hladky, M.A.Barrand, Interaction of the breast cancer resistance protein with plantpolyphenols, Biochem. Biophys. Res. Commun. 317 (2004) 269–275.

[9] P.K. Gill, A. Gescher, T.W. Gant, Regulation of MDR1 promoteractivity in human breast carcinoma cells by protein kinase C isozymesa and h, Eur. J. Biochem. 268 (2001) 4151–4157.

[10] T. Janvilisri, H. Venter, S. Shahi, G. Reuter, L. Balakrishnan, H.W.van Veen, Sterol transport by the human breast cancer resistanceprotein (ABCG2) expressed in Lactococcus lactis, J. Biol. Chem. 278(2003) 20645–20651.

[11] T. Janvilisri, S. Shahi, H. Venter, L. Balakrishnan, H.W. vanVeen, Arginine-482 is not essential for transport of antibiotics,primary bile acids and unconjugated sterols by the human breastcancer resistance protein (ABCG2), Biochem. J. 385 (Pt 2) (2005)419–426.

[12] J.D. Allen, A.H. Schinkel, Multidrug resistance and pharmacologicalprotection mediated by the breast cancer resistance protein (BCRP/ABCG2), Mol. Cancer Ther. 1 (2002) 427–434.

[13] E.Y. Bae, M.J. Han, M.K. Choo, S.Y. Park, D.H. Kim, Metabolismof 20(S)- and 20(R)-ginsenoside Rg3 by human intestinal bacteria andits relation to in vitro biological activities, Biol. Pharm. Bull. 25(2002) 58–63.

[14] S. Shibata, Chemistry and cancer preventing activities of ginsengsaponins and some related triterpenoid compounds, J. Korean Med.Sci. 16 (Suppl) (2001) S28–S37.

[15] S. Sengupta, S.A. Toh, L.A. Sellers, J.N. Skepper, P. Koolwijk, H.W.Leung, H.W. Yeung, R.N. Wong, R. Sasisekharan, T.P. Fan,Modulating angiogenesis: the yin and the yang in ginseng, Circulation110 (2004) 1219–1225.

[16] S. Odashima, T. Ota, H. Kohno, T. Matsuda, I. Kitagawa, H. Abe, S.Arichi, Control of phenotypic expression of cultured B16 melanomacells by plant glycosides, Cancer Res. 45 (1985) 2781–2784.

[17] A.B. Hendrich, K. Michalak, Lipids as a target for drugs modulatingmultidrug resistance of cancer cells, Curr. Drug Targets 4 (2003)23–30.

[18] M. Jedrzejczak, A. Koceva-Chyla, K. Gwozdzinski, Z. Jozwiak,Changes in plasma membrane fluidity of immortal rodent cellsinduced by anticancer drugs doxorubicin, aclarubicin and mitoxan-trone, Cell Biol. Int. 23 (1999) 497–506.

[19] A.S. Attele, J.A. Wu, C.S. Yuan, Ginseng pharmacology: multipleconstituents and multiple actions, Biochem. Pharmacol. 58 (1999)1685–1693.

[20] Y. Imai, S. Tsukahara, E. Ishikawa, T. Tsuruo, Y. Sugimoto,Estrone and 17beta-estradiol reverse breast cancer resistanceprotein-mediated multidrug resistance, Jpn. J. Cancer Res. 93(2002) 231–235.

[21] Y. Sugimoto, S. Tsukahara, Y. Imai, Y. Sugimoto, K. Ueda, T.Tsuruo, Reversal of breast cancer resistance protein-mediated drugresistance by estrogen antagonists and agonists, Mol. Cancer Ther. 2(2003) 105–112.

[22] C. Ozvegy, A. Varadi, B. Sarkadi, Characterization of drug transport,ATP hydrolysis, and nucleotide trapping by the human ABCG2multidrug transporter. Modulation of substrate specificity by a pointmutation, J. Biol. Chem. 277 (2002) 47980–47990.

[23] Y. Imai, S. Asada, S. Tsukahara, E. Ishikawa, T. Tsuruo, Y.Sugimoto, Breast cancer resistance protein exports sulfated estrogensbut not free estrogens, Mol. Pharmacol. 64 (2003) 610–618.

[24] R.Y. Tsai, A molecular view of stem cell and cancer cell self-renewal,Int. J. Biochem. Cell Biol. 36 (2004) 684–694.