Novel Extracellular Vesicles Mediate an ABCG2-Dependent ...Novel Extracellular Vesicles Mediate an...

Novel Extracellular Vesicles Mediate an ABCG2-Dependent

Anticancer Drug Sequestration and Resistance

Ilan Ifergan,1George L. Scheffer,

2and Yehuda G. Assaraf

1

1Department of Biology, Technion-Israel Institute of Technology, Haifa, Israel and 2Department of Pathology, VU Medical Center,MB Amsterdam, the Netherlands

Abstract

Overexpression of the multidrug efflux transporter ABCG2 inthe plasma membrane of cancer cells confers resistance tovarious anticancer drugs, including mitoxantrone. Here, weexplored the mechanism underlying drug resistance in theMCF-7 breast cancer sublines MCF-7/MR and MCF-7/FLV1000cells in which wild-type (R482) ABCG2 overexpression is highlyconfined to cell-cell attachment zones. The latter comprised themembrane of novel extracellular vesicles in which mitoxan-trone was rapidly and dramatically sequestered. After 12 hoursof incubation with mitoxantrone, the estimated intravesiculardrug concentration wasf1,000-fold higher than in the culturemedium. This drug compartmentalization was prevented bythe specific and potent ABCG2 transport inhibitors Ko143 andfumitremorgin C, thereby resulting in restoration of drugsensitivity. Consistently, this intravesicular drug concentrationwas abrogated by energy deprivation and was restored uponprovision of energy substrates. Fine-structure studies corrob-orated the presence of numerous large extracellular vesiclesthat were highly confined to cell-cell attachment zones betweenneighbor cells. Furthermore, high-resolution electron micros-copy revealed that the membrane of these extracellular vesiclescontained microvilli-like invaginations protruding into theintravesicular lumen. It is likely that these microvilli-likeprojections increase the vesicular membrane surface, therebyallowing for a more efficient ABCG2-dependent intravesicularanticancer drug concentration. Hence, these novel extracellu-lar vesicles mediate the ABCG2-dependent extraction ofintracellular drug, thereby serving as cytotoxic drug disposalchambers shared by multiple neighbor cancer cells. Thisconstitutes a novel modality of anticancer drug resistance.(Cancer Res 2005; 65(23): 10952-8)

Introduction

Chemotherapeutic agents constitute a key component of thetreatment of various human malignancies. However, the frequentemergence of cancer cells with a simultaneous resistance tomultipleanticancer drugs, a trait known as multidrug resistance, poses amajor obstacle towards curative cancer chemotherapy (1–6). Hence,the elucidation of the molecular mechanisms underlying inherentand acquired multidrug resistance is a prerequisite to the over-coming of anticancer drug resistance. Several transporters of theATP-binding cassette (ABC) superfamily have the facility to activelytranslocate an extraordinarily diverse array of structurally dissimilar

endogenous and exogenous substrates and their metabolites acrosscell membranes (reviewed in refs. 1, 2). Among these are threeimportant anticancer drug efflux transporters, including P-glyco-protein (Pgp/ABCB1; reviewed in refs. 1, 2, 5, 6), the multidrug-resistant protein 1 (MRP1/ABCC1; ref. 3), and breast cancerresistance protein (BCRP/ABCG2; ref. 4). Overexpression of thesemultidrug-resistant efflux transporters in the plasma membrane ofvarious malignant cells results in an ATP-dependent decrease incellular drug accumulation (1, 2, 4–6) and acquisition of multidrugresistance to a multitude of anticancer drugs. However, little isknown about the role of the overexpression of multidrug-resistanttransporters in alternative localizations that are distinct from theplasma membrane (7). As a step toward this end, we here exploredthe mechanism underlying mitoxantrone resistance in breast cancercells in which ABCG2 overexpression is highly confined to cell-cell attachment zones. We report on the identification of a novelmechanism of drug resistance that is based upon a dramatic ABCG2-dependent drug concentration in novel extracellular vesicles sharedby neighbor breast cancer cells.

Materials and Methods

Chemicals. Mitoxantrone hydrochloride was from Cyanamid of Great

Britain Ltd. (Gosport, Hampshire, United Kingdom). Ko143 was generously

provided by Dr. A.H. Schinkel (The Netherlands Cancer Institute, Amsterdam,the Netherlands), whereas fumitremorgin C and flavopiridol were kindly

provided by Dr. S.E. Bates (National Cancer Institute, Bethesda, MD). Sodium

azide and carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP)

were from Sigma (St. Louis, MO).Tissue culture and growth inhibition with mitoxantrone. MCF-7

human breast cancer cells and their mitoxantrone-resistant MCF-7/MR (8)

and flavopiridol-resistant MCF-7/FLV1000 (9) sublines kindly provided by Dr.

S.E. Bates were grown as monolayers in RPMI 1640 as previously described(7, 10). It should be emphasized that in a previous study we found that MCF-

7/MR cells overexpressed the wild-type R482 ABCG2; thus, no ABCG2

mutations were present in MCF-7/MR (10) and MCF-7/FLV1000 cells (11).

Specifically, MCF-7/MR cells were pulsed with 100 nmol/L mitoxantroneevery 2 weeks for 3 days, whereas MCF-7/FLV1000 cells were continuously

grown in the presence of 1 Amol/L flavopiridol. All subsequent experiments

were initiated after 4 days of incubation with mitoxantrone-free orflavopiridol-free medium. Cells (5 � 103 per well) were seeded in 24-well

plates in growth medium (2 mL/well) and incubated for 48 hours at 37jC.Then, the medium of MCF-7 and MCF-7/MR cells was replaced with a

fresh one lacking or containing Ko143 (0.4 Amol/L). After 2 hours ofincubation, mitoxantrone was added at various concentrations. Then, the

cells were exposed to this drug for 5 hours at 37jC, following which the

drug-containing medium was aspirated, and three successive washes (each

of 10 minutes) in RPMI 1640 containing 10% dialyzed fetal bovine serum(FBS) were done at 37jC. Drug-free medium was then added (2 mL/well),

and cultures were incubated for four additional days at 37jC. Finally, cellswere detached by trypsinization, and the number of viable cells wasdetermined microscopically using trypan blue exclusion.

Western blot analysis of ABCG2. ABCG2 levels as normalized to

h-tubulin were determined by Western blots using monoclonal antibodies

BXP-53 and clone 2-28-33, respectively, as previously described (7, 10).

Requests for reprints: Yehuda G. Assaraf, Department of Biology, Technion-IsraelInstitute of Technology, Haifa 32000, Israel. Phone: 972-4-829-3744; E-mail:[email protected].

I2005 American Association for Cancer Research.doi:10.1158/0008-5472.CAN-05-2021

Cancer Res 2005; 65: (23). December 1, 2005 10952 www.aacrjournals.org

Research Article

Research. on May 17, 2020. © 2005 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

Mitoxantrone accumulation and immunohistochemical localizationof ABCG2 in specific colonies of MCF-7/MR and MCF-7/FLV1000 cells.Cells (5 � 104) were seeded in 25-mm tissue culture flasks (5-mL medium)

and incubated for 4 days. Then, the growth medium was replaced with a

fresh one containing 20 Amol/L mitoxantrone. After 12 hours of incubationat 37jC, monolayer cells were washed thrice with medium containing 10%

dialyzed FBS. Then, 1 mL of medium was added to each flask, and random

colonies were examined using a LEICA microscope at a bright-field mode.

For immunohistochemical staining of ABCG2, monolayer cells were thenprocessed as previously described (7, 10). The immunolocalization results

obtained with BXP-53 were fully corroborated with other monoclonal

antibodies to ABCG2, including BXP-21 and BXP-43.

Determination of the number of light-refracting extracellularvesicles. MCF-7 and MCF-7/MR cells (5 � 104) were seeded in 25-mm

tissue culture flasks (5-mL medium per flask) and incubated for 4 days at

37jC, following which the medium was replaced with a fresh one (1 mL/flask). Random colonies were examined for visible, light-refracting

extracellular vesicles using a LEICA microscope at a bright-field mode.

Three independent experiments were done using f200 cells in each

determination for each cell line.Inhibition of mitoxantrone accumulation with ABCG2 transport

inhibitors and ATP-depleting agents. Cells were seeded in 24-well plates

(104 per well) and incubated for 4 days at 37jC. Then, themediumof a control

well was replaced with a fresh one. In contrast, the medium of the three

remaining wells was replaced with one containing either 0.4 Amol/L Ko143,

5 Amol/L fumitremorgin C, or the combination of the metabolic energy

inhibitors 5 Amol/L FCCP and 5 mmol/L sodium azide. Following 1 hour of

incubation at 37jC, mitoxantrone was added to a final concentration of 20

Amol/L. After 6 hours of incubation at 37jC, the growth medium was

aspirated and a wash step with medium containing 10% dialyzed FBS was

done. Then, fresh medium was added (0.3 mL/well), and random colonies

were rapidly examined for their mitoxantrone blue staining using a LEICA

microscope at a bright-field mode. To remove the ABCG2 transport and

metabolic energy inhibitors, the cells were incubated twice (each for 7

minutes) in fresh growth medium at 37jC followed by aspiration of the

medium. Freshmedium containing 20 Amol/Lmitoxantrone was then added;

cells previously incubated with FCCP and azide were also supplemented with

the ATP-restoring agents sodium pyruvate (1 mmol/L; Life Technologies

Bethesda Research Laboratories, Gaithersburg, MD) and D-glucose (5 mmol/

L; Sigma). After 6 hours of incubation at 37jC, the medium was discarded,

and cells were washed once withmedium containing 10% dialyzed FBS. Then,

fresh mediumwas added (0.3 mL/well), and random colonies were examined

microscopically for mitoxantrone accumulation.Estimation of the intravesicular concentration of mitoxantrone. To

explore the time dependence of mitoxantrone accumulation in the intra-

vesicular lumen, we incubated cells with 20 Amol/L mitoxantrone for 3 to 12

hours at 37jC. Photographs of random colonies were taken using a LEICA

microscope at a bright-field mode. To generate a calibration curve, 10-ALaliquots of standard solutions containing increasing mitoxantrone concen-

trations were dispensed onto glass slides, which were then covered by glass

coverslips. Photographs were taken at random locations using a bright-field

mode. Photographs were then transformed to a gray-scale format and

analyzed individually by scanning densitometry using the program ‘‘TINA’’

(version 2.10g). The densitometric background levels of the calibration curve

(i.e., at a zero mitoxantrone concentration) and the monolayer cell culture

with unstained extracellular vesicles (t = 0) were numerically normalized. The

experiment was done thrice using f150 extracellular vesicles in each

experiment for each incubation time with mitoxantrone.

Autofluorescence detection with viable cells. Cells (4 � 103) wereseeded in 24-well plates (2-mL phenol red–free medium per well) for 4 days

at 37jC in the presence or absence of 0.4 Amol/L Ko143. Then, 1.5 mL of the

growth medium was removed, and random colonies were examined fortheir autofluorescence using a fluorescence microscope at an FITC-like

mode; a bright-field mode examination was also used here.

Confocal microscopy of ABCG2 confinement to cell-cell attachmentzones. Cells (4 � 103) were seeded in 24-well plates (1-mL medium per well)on sterile glass coverslips and incubated for 4 days at 37jC. Cells were then

washed, reacted with the ABCG2-specific monoclonal antibody BXP-53followed by a secondary goat anti-rat IgG and a third FITC-conjugated

rabbit-anti-goat IgG as described recently (7). The fluorescent cells were

then examined for cross-section and perpendicular section using a Bio-Rad

(Richmond, CA) MRC1024 confocal microscope.Confocal microscopy studies of the accessibility of the culture

medium to the extracellular vesicles. MCF-F/MR cells (4 � 103 per well)

were seeded in 24-well plates (2-mL medium per well) on sterile glass

coverslips and incubated for 3 days at 37jC. The growth medium wasremoved, andmonolayer cells were washed twice with a fresh medium. Then,

a PBS solution containing 10% FCS, 15% tetramethylrhodamine isothiocya-

nate (TRITC)/goat anti-rabbit IgG (Sigma, T5268) was placed between the

coverslips and the glass slides. The slides were then examined using a Bio-RadMRC1024 confocal microscope for cross-section and perpendicular section.

The FITC-like mode was used to follow the green autofluorescence of the

vesicles and the red fluorescence of the culturemediumwas detected by a Kr/Ar laser (excitation at 568 nm and emission at 585 nm).

Electron microscopy studies. The presence of extracellular vesicles andtheir fine structure were studied by first seeding MCF-7/MR cells on glass

slides mounted on eight-well tissue culture chambers (Lab-Tek, Nunc,Naperville, IL). The cells were grown for 4 days until confluence was achieved;

an examination under a light microscope revealed the presence of numerous

large vesicles. Slides containingmonolayer cells were fixed using an overnight

incubation in 2% glutardialdehyde in phosphate buffer and an additional 30-minute incubation in osmium tetroxide/collidine (2:1) buffer for 30 minutes.

Slides were dehydrated using solutions containing increasing concentrations

of ethanol (70-100%). Then, the chambers were discarded, and the slidescontaining monolayer cells were impregnated for 1.5 hours in a solution of

Epon/propylene oxide/DMP-30 (1:1:0.02). Cells were embedded by placing on

top of them open-end capsules that were filled with embedding fluid (Epon/

DMP-30, 1:0.015), following which polymerization was allowed overnight at70jC. After polymerization, the glass slides were removed by snap-freezing in

liquid nitrogen and thawing, thereby resulting in the entrapment of the

monolayer cells in the polymerized resin. Then, ultrathin sections (60-70 nm)

were cut using a Diatome diamant knife and an LKB Ultrotome III andcollected on support film-coated (1.5% Formvar in dichloroethane) Cu grids.

The sections were then counterstained with 2% uranyl acetate for 20 minutes

and lead nitrate/sodium tricitrate for 20 minutes and then examined with aJeol 1200EX electron microscope. Photographs were finally printed using a

Leitz Focomat IIc.

Statistical analysis. We used a Student’s t test to examine the

significance of the difference between two populations for a certainvariable. A difference between the averages of two populations was

considered significant if the obtained P < 0.05.

Results

Overexpression and immunolocalization of ABCG2 toextracellular vesicles in mitoxantrone-resistant MCF-7/MRcells. MCF-7/MR breast cancer cells overexpress ABCG2 (7, 8)and consequently display 20-fold resistance to the establishedABCG2 transport substrate mitoxantrone, relative to their parentalMCF-7 cells (Fig. 1). Sensitivity to this anticancer drug wasrestored with Ko143, a potent and specific ABCG2 transportinhibitor (12). Previously, we have shown that ABCG2 was highlyconfined to cell-cell attachment zones in monolayers of MCF-7/MRcells (7). Microscopic examination of immunohistochemicalstaining with anti-BCRP as well as after hematoxylin staining ofmonolayers of MCF-7 (Fig. 2A-C) and MCF-7/MR cells (Fig. 2D-F)revealed numerous extracellular vesicle-like structures that werehighly confined to cell-cell attachment zones between multipleneighbor cells. Immunohistochemical analysis of MCF-7/MRcells with a monoclonal antibody to ABCG2 (BXP-53) revealedthat ABCG2 was highly confined to the vesicular membrane con-tacting the surrounding cells (Fig. 2D-E, continuous arrows) as well

Mitoxantrone Concentration in Extracellular Vesicles

www.aacrjournals.org 10953 Cancer Res 2005; 65: (23). December 1, 2005



as to cell-cell attachment zones (Fig. 2E, dashed arrow); no stainingwas observed in the absence of BXP-53 antibodies (Fig. 2F). Incontrast, parental MCF-7 cells which poorly express ABCG2 didnot show any detectable staining of the vesicular membranewhether the BXP-53 antibody was present (Fig. 2A-B) or absent(Fig. 2C). These immunolocalization results with the BXP-53antibody were recapitulated with additional monoclonal anti-bodies to ABCG2, including BXP-21 and BXP-34 (data not shown).Fine-structure studies corroborated the presence of multiple

extracellular vesicles in MCF-7/MR cells (Fig. 3A-B, continuousarrows), particularly betweenmultiple neighbor cells (Fig. 3B, dashedarrows). High-resolution electron microscopy revealed that thevesicular membrane had a typical lipid bilayer structure (Fig. 3C,continuous arrow). Moreover, these extracellular vesicles containedmicrovilli-like invaginations protruding into the vesicular lumen(Fig. 3C, dashed arrow).

Intravesicular concentration of mitoxantrone in an ATP-dependent and ABCG2-dependent manner. We explored themechanism underlying mitoxantrone resistance in breast cancerMCF-7/MR cells in which ABCG2 overexpression is highlyconfined to these extracellular vesicles in cell-cell attachmentzones. The intense blue color of mitoxantrone rendered drugaccumulation readily discernible by light microscopy. Pulseexposure of MCF-7/MR breast cancer cells to 20 Amol/Lmitoxantrone for 6 hours resulted in a dramatic sequestration ofthis blue drug in the lumen of these extracellular vesicles thatwere confined to cell-cell attachment zones (Fig. 4A). Theseextracellular vesicles residing in between neighbor cells refractedlight; this was used to estimate the number of extracellularvesicles. The number of light-refracting extracellular vesicles per100 cells was estimated to be 23.3 F 2.5 and 3.2 F 0.5 in drug-resistant MCF-7/MR and parental MCF-7 cells, respectively.

Figure 2. Immunohistochemical localization of ABCG2 in parental MCF-7 cells and their MCF-7/MR subline. MCF-7 (A-C ) and MCF-7/MR cells (D-F ) grown in24-well culture plates were fixed with formaldehyde and reacted with (A, B, D , and E ) or without (C and F ) the anti-BCRP monoclonal antibody BXP-53 followedby the addition of horseradish peroxidase–conjugated rabbit anti-mouse IgG as the second antibody. Color development was carried out using the chromogen3,3V-diaminobenzidine (brown). Cells were then counterstained with hematoxylin (violet ) and examined with a light microscope at a �200 magnification. Continuousarrows, extracellular vesicles (A-F ). Note that ABCG2 precisely localizes to the membrane surrounding the extracellular vesicles (D and E, continuous arrows ) and tocell-cell attachment zones in MCF-7/MR cells (E, dashed arrow ). These immunolocalization results with BXP-53 were also corroborated with additional monoclonalantibodies to ABCG2, including BXP-21 and BXP-34 (data not shown).

Figure 1. Cellular growth inhibition with mitoxantrone.Parental MCF-7 and MCF-7/MR cells were exposed tovarious concentrations of mitoxantrone for 5 hours in theabsence or presence of the ABCG2 inhibitor Ko143(0.4 Amol/L). After 4 days of incubation in drug-freemedium, the number of viable cells was determined.Points, means of three independent experiments; bars,SD. Inset, Western blot analysis of ABCG2 expressionas normalized to h-tubulin.

Cancer Research




Furthermore, the number of mitoxantrone-concentrating extracel-lular compartments was 44.1 F 6.5 per 100 MCF-7/MR cells andnone in parental cells. In agreement with the above results,immunohistochemical analysis revealed that ABCG2 stainingformed a circumferential ring in the membrane of the extracellularvesicles (Fig. 4B), thereby establishing that ABCG2 is highly confinedto the vesicular membrane contacting the surrounding cells. Incontrast, ABCG2 was barely detectable in the apical and basalmembranes of these vesicles (Fig. 4B). The intense blue color of thesequestered mitoxantrone in these extracellular vesicles allowed forthe quantification of the intravesicular concentration of the drug.Based on a calibration curve of knownmitoxantrone concentrations(Fig. 4C), the intravesicular concentration of the drug was estimatedto be as high as 12.8 F 3.5 mmol/L after 6 hours of incubation with20 Amol/L mitoxantrone and further increased in a time-dependentmanner to f20 mmol/L after 12 hours (Fig. 4D). Hence, theintravesicular concentration of mitoxantrone after 12 hours ofincubation with this drug wasf1,000-fold higher than in the culture

medium (P = 5.5� 10�12). Similarly, the intravesicular concentrationof mitoxantrone was also explored in flavopiridol-resistant MCF-7/FLV1000 cells with ABCG2 overexpression. Consistently, MCF-7/FLV1000 cells, which also contained extracellular vesicles, albeit at alower frequency than MCF-7/MR cells, displayed a robust intra-vesicular concentration of mitoxantrone (data not shown). Thislatter result suggests that the extracellular vesicles and their abilityto sequester mitoxantrone in an ABCG2-dependent manner is notlimited to mitoxantrone-resistant MCF-7/MR cells.The intravesicular concentration of mitoxantrone in MCF-7/MR

cells (Fig. 5A) was prevented by the specific and potent ABCG2 drugefflux inhibitors Ko143 (Fig. 5B) and fumitremorgin C (Fig. 5C) aswellas by energy deprivation achieved by treatment with the respirationinhibitor sodium azide and the uncoupler FCCP (Fig. 5D). To con-firm that the high concentration of mitoxantrone did not impairthe ability of the vesicles to concentrate this drug, we did anexperiment in which the ABCG2 transport inhibitor and metabolicenergy inhibitors were first washed out followed by mitoxantrone

Figure 3. Transmission electron microscopyanalysis of the extracellular vesicles in monolayersof MCF-7/MR cells. MCF-7 MR cells were grownon glass slides until confluence was achieved.Slides containing monolayer cells were then fixed,dehydrated with ethanol, embedded in anEpon/DMP-30 resin, cut with an ultrotome, andanalyzed with an electron microscope as inMaterials and Methods. A and C, continuousarrows, membrane of the extracellular vesicles.Magnification, �4500 (A) and �18,000 (C ).B, dashed arrows, plasma membrane of neighborcells surrounding the vesicle. Magnification,�9,000. C, furthermore, high-resolution electronmicroscopy revealed that these vesicles containeda lipid bilayer membrane (continuous arrow ) withmultiple microvilli-like invaginations protruding intothe intravesicular lumen (dashed arrow ).

Figure 4. Mitoxantrone accumulationin extracellular vesicles andimmunohistochemical localization ofABCG2 in MCF-7/MR cells. Cells wereincubated in growth medium containing20 Amol/L mitoxantrone for 12 hours at37jC. Then, cells were washed thrice, freshmedium was added, and random colonieswere examined by light microscopy at abright-field mode. A, note that mitoxantroneaccumulated in extracellular vesicles(blue extracellular vesicles ). B, cells werethen processed for immunohistochemicallocalization of ABCG2 (brown ); colonieswere examined by light microscopy aftercounterstaining of nuclei with hematoxylin(violet ). Note that ABCG2 preciselylocalizes to the membrane surrounding theextracellular vesicles that have previouslyaccumulated mitoxantrone (A). Theintravesicular concentration ofmitoxantrone was estimated using amitoxantrone calibration curve (C)revealing drug concentrations of f20mmol/L after 12 hours of incubation (D ).Columns, means of three independentexperiments using f150 extracellularvesicles in each experiment for eachincubation time; bars, SD. The estimatedintravesicular concentration ofmitoxantrone at 6 and 12 hours ofincubation was significantly higher thanthat at 3 hours (P = 4.9 � 10�16 and5.5 � 10�12, respectively).





reaccumulation. Hence, washing out the drug efflux inhibitorsfollowed by further incubation with mitoxantrone restored theintravesicular concentration of this drug (Fig. 5F-G) at a level thatwas comparable with untreated cells (Fig. 5E). Likewise, washing outthe metabolic energy inhibitors followed by provision of the energysubstrates glucose and pyruvate in the presence of mitoxantronerestored the intravesicular sequestration of the drug (Fig. 5H).Intravesicular concentration of an endogenous green

fluorescent chromophore. Under normal growth in mitoxan-trone-free medium, the extracellular vesicles were easily identifi-able by an intense endogenous green fluorescence (Fig. 6A-C). Thisautofluorescence was retained in phenol red–free medium, therebyexcluding the possibility that this common pH indicator is theendogenous fluorescent compound. However, this intravesicularfluorescence was completely lost upon cellular growth in thepresence of Ko143 (Fig. 6D-F). This result indicated that ABCG2mediated the intravesicular concentration of some endogenousfluorescent compound(s). Hence, ABCG2 mediates the intra-vesicular concentration of both mitoxantrone and the endogenousfluorescent compound(s). Taking advantage of this autofluores-cence, the structural and functional characteristics of these vesicleswere explored. The accessibility of the culture medium to theextracellular vesicles was first examined; a cell-impermeable red-fluorescence TRITC-IgG conjugate was used to label the extracel-lular milieu of MCF-7/MR monolayers (Fig. 6G and J). Whereas theextracellular vesicles were readily discernible by their endogenousgreen fluorescence (Fig. 6H and K), confocal laser microscopy ofcross-sections of monolayer MCF-7/MR cells incubated in mediumcontaining TRITC-IgG revealed that this red fluorescence chromo-phore was inaccessible to the green fluorescent extracellular vesiclefrom the cytosol (Fig. 6I). In contrast, a section perpendicularto the monolayer plane showed that the apical side of the extra-cellular vesicle was the only surface accessible to the TRITC-IgG-containing culture medium (Fig. 6J-L). The confinement of ABCG2to the vesicular membrane of this cylindrical extracellular com-partment was also corroborated by confocal laser microscopy afterstaining with a green fluorescent–labeled antibody to ABCG2(Fig. 6M-O). Consistent with the above immunohistochemistryresults (Fig. 2D-E and Fig. 4B), confocal analysis of a cross-sectionrevealed that ABCG2 staining formed a circumferential ring

(Fig. 6M). This further confirmed that ABCG2 was highly confinedto the vesicular membrane contacting the surrounding cells butwas barely detectable in the intravesicular lumen. Consistentwith the cross-section, a section perpendicular to the plane of themonolayer established the confinement of ABCG2 to the wallslining the extracellular vesicle (Fig. 6N and O). In contrast, theapical membrane of the extracellular vesicle that faces the culturemedium was devoid of ABCG2 (Fig. 6N). These results are in accordwith the immunohistochemistry findings, which show that ABCG2was barely detectable in the apical membrane of the extracellularvesicle. These analyses (Fig. 6) allowed for the estimation of theaverage volume of the cylindrical extracellular vesicle, which wasfound to be 190 F 64 fL. These results establish that ABCG2, whichis highly confined to the membrane walls lining the extracellularvesicles, mediates the ATP-driven transport of mitoxantrone fromthe cytosol into the intravesicular lumen of these extracellularcompartments.

Discussion

Several lines of evidence establish that the intravesicularconcentration of mitoxantrone is mediated by ABCG2. First,inhibition of ABCG2 transport activity by Ko143 and fumitremorginC prevented the intravesicular concentration of mitoxantrone.Furthermore, washing out these drug transport inhibitors resultedin restoration of mitoxantrone compartmentalization. Theseresults are in accord with the finding that Ko143 induced a near-complete reversal of mitoxantrone resistance in MCF-7/MR cells.Second, depletion of cellular ATP pools by the respiration inhibitorsodium azide (13) and the uncoupler FCCP (14) prevented theintravesicular concentration of mitoxantrone. Consistently, remov-al of metabolic energy inhibitors followed by restoration of cellularenergy resources by provision of glucose and pyruvate in thepresence of mitoxantrone resulted in the resumption of intra-vesicular drug concentration. These findings are in agreement withthe tight coupling of ABCG2 drug transport to ATP hydrolysis andconsequent intravesicular drug concentration.The concentration of mitoxantrone in these extracellular vesicles

by ABCG2 was energy and time dependent and reached a f20mmol/L concentration after 12 hours of incubation with 20 Amol/Ldrug. This 1,000-fold concentrative ability of ABCG2 gained further

Figure 5. Prevention of intravesicular mitoxantrone accumulation by ABCG2 transport inhibitors and metabolic energy deprivation: MCF-7/MR cells were incubatedfor 1 hour at 37jC in medium (A) lacking or containing either (B ) 0.4 Amol/L Ko143, (C ) 5 Amol/L fumitremorgin C, or a (D ) combination of the metabolic energyinhibitors FCCP (5 Amol/L) and azide (5 mmol/L). Mitoxantrone was added at 20 Amol/L, and cells were incubated for six additional hours at 37jC. Random colonieswere then rapidly examined for the intravesicular accumulation of mitoxantrone. After ridding off the various ABCG2 and metabolic energy inhibitors (F-H ), freshmedium containing 20 Amol/L mitoxantrone was added, and cells that were previously incubated with FCCP and azide were supplemented with the ATP-restoringsubstrates sodium pyruvate (1 mmol/L) and D-glucose (5 mmol/L) along with mitoxantrone. After 6 hours of incubation at 37jC, the growth medium was removed, andcells were washed once with medium containing 10% dialyzed FBS. Random colonies were finally examined for the intravesicular accumulation of mitoxantrone usinga light microscope at a bright-field mode (E-H ).

Cancer Research




support by the dramatic intravesicular sequestration of a greenfluorescent compound(s). Whereas this chromophore(s) was notfluorescently visible in the cytosol of neighbor cells surrounding theextracellular vesicle, it was intensely fluorescent in the lumen of this

compartment in both MCF-7/MR and MCF-7/FLV1000 cells withABCG2 overexpression. This intravesicular fluorescence was com-pletely absent after 4 days of cellular growth in the presence ofthe specific ABCG2 transport inhibitor Ko143. Furthermore, thepresence of the light-refracting extracellular vesicles in parentalMCF-7 cells was less frequent. When present, these extracellularvesicles in parental MCF-7 cells with low-level R482 ABCG2expression were completely devoid of the green autofluorescencethat is characteristic of the vesicles in drug-resistant MCF-7/MR andMCF-7/FLV1000 cells, which overexpress the wild-type R482 ABCG2.This lack of intravesicular autofluorescence in drug-sensitive cellsthat poorly express ABCG2 is consistent with the finding that theintravesicular concentration of the green fluorescent compound(s)in MCF-7/MR and MCF-7/FLV1000 cells is mediated by ABCG2.Hence, the energy and time dependence and the f1,000-foldconcentrative capacity of mitoxantrone and that of the autofluor-escent compound(s) by ABCG2 are consistent with the highlyconcentrative transport of various ABC transporters. First, lysosom-al and vacuolar membranes contain V-type ATP-driven protonpumps that maintain a >100-fold proton gradient across the acidiclumen of the lysosome (pH f4.5-5) and the neutral cytosol (pHf7.0; ref. 15). Second, because an increase in the concentration ofCa2+ ions in the cytosol of mammalian cells (e.g., erythrocytes) is animportant regulatory signal, the plasma membrane P-class Ca2+

ATPase efficiently transports Ca2+ out of the cell; consequently, theextracellular (i.e., blood) concentration of Ca2+ is as high as 3,600-fold (1.8 mmol/L) than in the cytosol of the erythrocyte (0.5 Amol/L;ref. 16). Similarly, Ca2+ ATPase from the sarcoplasmic reticulum ofmuscle cells efficiently pumps Ca2+ ions from the cytosol (0.1-1Amol/L) into the lumen of the sarcoplasmic reticulum (10 mmol/L),thereby resulting in at least 10,000-fold concentrative transport (17).The third example concerns the H+,K+ ATPase present in the plasmamembrane of acid-secreting parietal gastric cells. This P-type H+,K+

ATPase maintains an extremely acidic pH in the gastric fluid,whereas the cytosolic pH of these cells is neutral (pH 7.0). Thus, thisH+,K+ ATPase concentrates protons by a factor of 100,000 (18).Upon cross-section confocal microscopy experiments with a cell-

impermeable TRITC-IgG conjugate, there was no accessibility of theculture medium containing this red chromophore to the extracel-lular vesicles. Whereas, a section that is perpendicular to the plane ofthe monolayer revealed that the only contact of these vesicles withthe fluorochrome-labeled medium was from the apical side of thisextracellular compartment. Furthermore, confocal microscopy andimmunohistochemistry revealed that ABCG2 was primarily local-ized at the circumference of the extracellular vesicle but was absentfrom its apical side that faces the culture medium; ABCG2 wastherefore localized at the walls lining this vesicle but was absentfrom its apical side. Thus, the ATP-binding fold and the substrate-binding site of ABCG2 must face the cytoplasm of the cellssurrounding this extracellular vesicle. As such, ABCG2 presumablyextracts mitoxantrone from the cytosol of the surrounding cells andhighly concentrates it in the lumen of these extracellular vesicles.Fine-structure studies corroborated the presence of numerous

large extracellular vesicles emerging from cell-cell attachment zones.Furthermore, high-resolution electronmicroscopy revealed that thesevesicles contained a lipid bilayer membrane with multiple microvilli-like invaginations protruding into the intravesicular lumen. Hence,these fine-structure projections are reminiscent of the microvilliinvaginations of both the gastrointestinal mucosa and the placentalepithelium. Given the ATP-driven ABCG2-dependent trans-vesiculartransport of mitoxantrone into the intravesicular lumen, it is likely

Figure 6. Detection of intravesicular green autofluorescence in viableMCF-7/MR cells. MCF-7/MR monolayer cells were cultured in medium lacking(A-C ) or containing 0.4 Amol/L Ko143 (D-F). Then, random colonies wereexamined with a fluorescence microscope for their green autofluorescence(B and E ); microscopic examination using a bright-field mode revealed theintercellular localization of the extracellular vesicles (A and D, black arrows ).The green autofluorescence merged perfectly with the extracellular vesicles (C ),whereas this autofluorescence was absent from the extracellular vesicles inmonolayer cells growing in the presence of Ko143 (E and F ). The accessibilityof the growth medium to the extracellular vesicles was examined by confocalmicroscopy (G-L). Cells grown on sterile glass coverslips were incubated in abuffer solution containing a TRITC-IgG conjugate (red fluorescence). Confocalanalysis of cross-section (G-I ) and perpendicular section (J-L ) of the green andred fluorescence was done with viable monolayer cells; the FITC-like mode wasused to detect the extracellular vesicles’ green autofluorescence (H and K),whereas the culture medium red fluorescence of the cell-impermeableTRITC-IgG conjugate was detected using a Kr/Ar laser (G and J ). Mergingthe green fluorescence of the extracellular vesicles and the red fluorescenceis shown for both the cross-section (I ) and perpendicular section (L) analyses.Note that the accessibility of the extracellular vesicles (green fluorescence) tothe culture medium (red fluorescence) and not to the cytosol is restricted toits apical side (I and L). The confinement of ABCG2 to the circumferentialmembrane of the extracellular vesicles was revealed by cross-section (M ) andperpendicular section (N ) confocal microscopy after immunofluorescent stainingwith anti-ABCG2 antibodies. Cells grown on glass coverslips were fixed withmethanol and reacted with a monoclonal antibody to ABCG2 followed bya second FITC-conjugated antibody. The ABCG2 green fluorescence wasconfined to the circumferential membrane of the extracellular vesicles upona cross-section (M ) and to the membrane walls lining the extracellular vesiclesupon a perpendicular section (N ). Merging the cross-section and perpendicularsection is shown as well (O ).





that these microvilli-like invaginations increase the vesicularmembrane surface, thereby allowing for a more efficient intra-vesicular drug concentration. Similarly, the surface of the syncytialtrophoblast of the human placenta is covered by a microvillus (i.e.,brush) border that is in direct contact with maternal blood; thislocation is the site of a variety of transport and receptor activities. Forexample, endocytosis of gold-labeled LDL into primary humanplacental cells involveduncoatedplasmalemmal invaginations, whichsubsequently became clathrin-uncoated endosome vesicles (19).The encouraging results of the current study with anticancer

drug-resistant breast cancer cell lines warrant further clinicalevaluation of the presence of such drug-sequestering extracellularvesicles in tumor-derived specimens. The possible future finding ofsuch extracellular vesicles, which could efficiently concentrateanticancer drugs, may have potential implications for cancerchemotherapy. First, inclusion of specific, potent, and nontoxicABCG2 transport inhibitors, such as Ko143 (12) and GF120918 (20),which reverse anticancer drug resistance, may potentially proveeffective in the prevention of drug compartmentalization in tumors,thereby resulting in reversal of drug resistance. Moreover, variousapproved cytotoxic drugs were recently found to be efficientinhibitors of ABCG2 efflux activity, including Iressa (ZD1839,Gefitinib), a selective epidermal growth factor receptor tyrosinekinase inhibitor (21, 22); Imatinib mesylate (Gleevec, STI571), atyrosine kinase inhibitor selective for Bcr-Abl (23); and CI-1033, aHER tyrosine kinase inhibitor (24). In addition, phytoestrogens andflavonoids were also shown to efficiently reverse drug resistancemediated by ABCG2 (25). Clearly, these ABCG2 efflux inhibitors mayprove effective reversal agents of drug resistance mediated byABCG2 overexpression, including when the latter is highly confinedto the vesicular membrane. Second, compounds that may interferewith the formation of these novel extracellular vesicles and/or with

the sorting of ABCG2 to the vesicular membrane should render cellssensitive to anticancer drugs like mitoxantrone. For example, arecent article (26) reported on the rapid translocation of ABCG2from the plasma membrane to the cytoplasmic compartment(endoplasmic reticulum-Golgi) in freshly derived hematopoieticstem cells known as side population; in this study, it was shown thata brief treatment (1.5 hours) of freshly derived mouse bone marrowcells with LY294002, an inhibitor of the Akt effector proteinphosphatidylinositol-3-kinase (PI3K), resulted in the rapid translo-cation of ABCG2 from the plasma membrane to the cytoplasmiccompartment. The authors therefore suggested that the PI3K-Aktsignaling axis is an important regulator of ABCG2 expression andsubcellular localization of the bone marrow–derived side popula-tion stem cell phenotype. Another example involves a recent studyfrom our laboratory showing that short-term deprivation of folicacid from the growth medium of ABCG2-overexepressing MCF-7/MR cells resulted in the cytoplasmic confinement of this ABCG2multidrug-resistant efflux transporter (7). Hence, it is possible thatsuch agents and treatment strategies, which block protein sortingof ABCG2 from the cytoplasmic compartment to the plasmamembrane and vesicular membrane, may be used to reverseanticancer drug resistance mediated by these novel ABCG2-richextracellular vesicles.

Acknowledgments

Received 6/9/2005; revised 8/28/2005; accepted 9/15/2005.Grant support: Israel Cancer Association and Star Foundation (Y.G. Assaraf).The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.

We thank J.M. Fritz for his skillful assistance with the electron microscopy studies,Dr. A.H. Schinkel for the generous gift of Ko143, and Dr. S.E. Bates for the cell linesMCF-7/MR and MCF-7/MR1000 and for providing us with fumitremorgin C andflavopiridol.

Cancer Research


References

1. Borst P, Elferink RO. Mammalian ABC transportersin health and disease. Annu Rev Biochem 2002;71:537–92.

2. Haimeur A, Conseil G, Deeley RG, Cole SP. The MRP-related and BCRP/ABCG2 multidrug resistance proteins:biology, substrate specificity and regulation. Curr DrugMetab 2004;5:21–53.

3. Cole SP, Bhardwaj G, Gerlach JH, et al. Overexpressionof a transporter gene in a multidrug-resistant humanlung cancer cell line. Science 1992;258:1650–4.

4. Sarkadi B, Ozvegy-Laczka O, Nemet K, Varadi A.ABCG2: a transporter for all seasons. FEBS Lett 2004;567:116–20.

5. Ambudkar SV, Kimchi-Sarfaty C, Sauna ZE, GottesmanMM. P-glycoprotein: from genomics to mechanism.Oncogene 2003;22:7468–85.

6. Gottesman MM, Fojo T, Bates SE. Multidrug resistancein cancer: role of ATP-dependent transporters. Nat RevCancer 2002;2:48–58.

7. Ifergan I, Jansen G, Assaraf YG. Cytoplasmic confine-ment of breast cancer resistance protein (BCRP/ABCG2)as a novel mechanism of adaptation to short-term folatedeprivation. Mol Pharmacol 2005;4:1–11.

8. Taylor CW, Dalton WS, Parrish PR, et al. Differentmechanisms of decreased drug accumulation in doxo-rubicin and mitoxantrone resistant variants of the MCF-7 human breast cancer cell line. Br J Cancer 1991;63:923–9.

9. Robey RW, Medina-Perez WY, Nishiyama K, et al.Overexpression of the ATP-binding cassette half-trans-porter, ABCG2 (MXR/BCRP/ABCP1), in flavopiridol-resistant human breast cancer cells. Clin Cancer Res2001;7:145–52.

10. Ifergan I, Shafran A, Jansen G, Hooijberg JH, SchefferGL, Assaraf YG. Folate deprivation results in the loss ofbreast cancer resistance protein (ABCG2) expression: arole for BCRP in cellular folate homeostasis. J Biol Chem2004;279:25527–34.

11. Honjo Y, Hrycna CA, Yan QW, et al. Acquiredmutations in the MXR/BCRP/ABCP gene alter substratespecificity in MXR/BCRP/ABCP-overexpressing cells.Cancer Res 2001;61:6635–9.

12. Allen JD, van Loevezijn A, Lakhai JM, et al. Potentand specific inhibition of breast cancer resistanceprotein multidrug transporter in vitro and in mouseintestine by a novel analogue of fumitremorgin C. MolCancer Ther 2002;1:417–25.

13. Mechetner EB, Schott B, Morse BS, et al. P-glycoprotein function involves conformational transi-tions detectable by differential immunoreactivity. ProcNatl Acad Sci U S A 1997;94:12908–13.

14. Stuhlsatz-Krouper SM, Bennett NE, Schaffer JE.Substitution of alanine for serine 250 in the murinefatty acid transport protein inhibits long chain fatty acidtransport. J Biol Chem 1998;273:28642–50.

15. Forgac M. Structure and function of vacuolar class ofATP-driven proton pumps. Physiol Rev 1989;69:765–96.

16. James P, Vorherr T, Krebs J, et al. Modulation oferythrocyte Ca2+-ATPase by selective calpain cleavage ofthe calmodulin-binding domain. J Biol Chem 1989;264:8289–96.

17. Clarke DM, Loo TW, Inesi G, Maclennan DM.Location of high affinity Ca2+-binding sites within thepredicted transmembrane domain of the sarcoplasmicreticulum Ca2+-ATPase. Nature 1989;339:2914–8.

18. Rabon ER, McFall TL, Sachs G. The gastric[H K]ATPase: H+/ATP stoichiometry. J Biol Chem1982;257:6296–9.

19. Malassine A, Besse C, Roche A, et al. Ultrastructuralvisualization of the internalization of low densitylipoprotein by human placental cells. Histochemistry1987;87:457–64.

20. de Bruin M, Miyake K, Litman T, Robey R, Bates SE.Reversal of resistance by GF120918 in cell linesexpressing the ABC half-transporter, MXR. Cancer Lett1999;146:117–26.

21. Yanase K, Tsukahara S, Asada S, Ishikawa E, Imai Y,Sugimoto Y. Gefitinib reverses breast cancer resistanceprotein-mediated drug resistance. Mol Cancer Ther2004;3:1119–25.

22. Elkind NB, Szenpetery Z, Apati A, et al. Multidrugtransporter ABCG2 prevents tumor cell death in-duced by the epidermal growth factor receptorinhibitor Iressa (ZD1839, Gefitinib). Cancer Res2005;65:1770–7.

23. Houghton PJ, Germain GS, Harwood FC, et al.Imatinib mesylate is a potent inhibitor of theABCG2 (BCRP) transporter and reverses resistanceto topotecan and SN-38 in vitro . Cancer Res 2004;64:2333–7.

24. Erlichman C, Boerner SA, Hallgren CG, et al.The HER tyrosine kinase inhibitor CI-1033enhances cytotoxicity of 7-ethyl-10-hydroxycampto-thecin and topotecan by inhibiting breast cancerresistance protein drug efflux. Cancer Res 2001;61:739–48.

25. Imai Y, Tsukahara S, Asada S, Sugimoto Y. Phytoes-trogens/flavonoids reverse breast cancer resistanceprotein/ABCG2-mediated multidrug resistance. CancerRes 2004;64:4346–52.

26. Mogi M, Yang J, Lambert JF, et al. Akt signalingregulates side population cell phenotype via Bcrp1translocation. J Biol Chem 2003;278:39068–75.



2005;65:10952-10958. Cancer Res Ilan Ifergan, George L. Scheffer and Yehuda G. Assaraf Anticancer Drug Sequestration and ResistanceNovel Extracellular Vesicles Mediate an ABCG2-Dependent

Updated version

http://cancerres.aacrjournals.org/content/65/23/10952

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://cancerres.aacrjournals.org/content/65/23/10952To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Novel Extracellular Vesicles Mediate an ABCG2-Dependent ...Novel Extracellular Vesicles Mediate an...

Documents

Transcript of Novel Extracellular Vesicles Mediate an ABCG2-Dependent ...Novel Extracellular Vesicles Mediate an...