Proc. Natl. Acad. Sci. USA
Vol. 95, pp. 15665–15670, December 1998
A multidrug resistance transporter from human MCF-7 breast
L. AUSTIN DOYLE*†, WEIDONG YANG*, LYNNE V. ABRUZZO*‡, TAMMY KROGMANN*‡, YONGMING GAO*,
ARUN K. RISHI*, AND DOUGLAS D. ROSS*†§¶
*Greenebaum Cancer Center of the University of Maryland, Baltimore MD 21201;†Department of Medicine, Division of Hematology?Oncology, University of
Maryland School of Medicine, Baltimore, MD 21201;‡Department of Pathology, University of Maryland School of Medicine, Baltimore, MD 21201; and
§Baltimore Veterans Medical Center, Department of Veterans Affairs, Baltimore, MD, 21201
Communicated by David E. Housman, Massachusetts Institute of Technology, Cambridge, MA, October 12, 1998 (received for review
March 17, 1998)
man breast cancer subline that displays an ATP-dependent
reduction in the intracellular accumulation of anthracycline
anticancer drugs in the absence of overexpression of known
multidrug resistance transporters such as P glycoprotein or
the multidrug resistance protein. RNA fingerprinting led to
the identification of a 2.4-kb mRNA that is overexpressed in
MCF-7?AdrVp cells relative to parental MCF-7 cells. The
mRNA encodes a 663-aa member of the ATP-binding cassette
superfamily of transporters that we term breast cancer resis-
tance protein (BCRP). Enforced expression of the full-length
BCRP cDNA in MCF-7 breast cancer cells confers resistance
to mitoxantrone, doxorubicin, and daunorubicin, reduces
daunorubicin accumulation and retention, and causes an
ATP-dependent enhancement of the efflux of rhodamine 123
in the cloned transfected cells. BCRP is a xenobiotic trans-
porter that appears to play a major role in the multidrug
resistance phenotype of MCF-7?AdrVp human breast cancer
MCF-7?AdrVp is a multidrug-resistant hu-
The development of resistance to multiple chemotherapeutic
drugs occurs frequently during the treatment of advanced
carcinoma of the breast. Two transmembrane xenobiotic trans-
porter proteins, P glycoprotein (Pgp) and the multidrug resis-
tance protein (MRP), cause multidrug resistance when trans-
fected into drug-sensitive cells in culture (1–5). Despite these
findings, the role these transporters play in clinical drug
resistance exhibited by human breast cancer is unclear; hence,
alternate or additional drug resistance mechanisms operative
in this disease have been sought. To address this problem,
Chen et al. (6) selected human breast carcinoma MCF-7 cells
for resistance to the anthracycline doxorubicin in the presence
of verapamil, an inhibitor of Pgp. The resultant multidrug-
resistant subline, MCF-7?AdrVp, exhibits marked crossresis-
tance to certain other anthracyclines [daunorubicin and 3?-
deamino-3?-(3-cyano-4-morpholinyl) doxorubicin] and to the
anthracenedione mitoxantrone but remains sensitive to vinca
alkaloids, paclitaxel (6, 7), and cis-platin. MCF-7?AdrVp cells
do not overexpress Pgp or MRP, despite a marked reduction
in the intracellular accumulation of the anthracycline dauno-
rubicin and the fluorescent dye rhodamine 123 compared with
MCF-7 cells (6, 7). MCF-7?AdrVp cells do not display the
altered subcellular distribution of drug (7) seen in certain cells
that overexpress MRP. Although the decreased accumulation
of daunorubicin is not reversed by the classical Pgp antagonist
cyclosporin A, depletion of ATP results in complete abroga-
tion of the enhanced efflux of both daunorubicin and rhoda-
mine (7). These findings led to the hypothesis that an ATP-
dependent xenobiotic transporter may contribute significantly
to the multidrug-resistance phenotype of MCF-7?AdrVp cells.
To test this hypothesis, we sought mRNA species that were
differentially overexpressed in MCF-7?AdrVp cells compared
with MCF-7 cells by using the technique of RNA fingerprint-
MATERIALS AND METHODS
Cell Lines. MCF-7 human breast carcinoma cells, their
drug-resistant subline MCF-7?AdrVp, and a partial-revertant
subline (MCF-7?AdrVpPR), cultured in the absence of adria-
mycin, were obtained from Antonio Fojo (Medicine Branch,
National Cancer Institute). The cells were maintained in
culture as described (8). The MCF-7?AdrVp subline was
continuously maintained in the presence of 1 ?g?ml doxoru-
RNA Fingerprinting. RNA Fingerprinting was performed
by using the protocol in the Delta RNA Fingerprinting kit
(CLONTECH), a modification of the differential-display tech-
nique (9, 10). The sequence of the P6 arbitrary primer was
5?-ATTAACCCTCACTAAATGCTGGGTG-3?; the se-
quence of the T9 oligo(dT) primer was 5?-CATTATGCT-
GAGTGATATCTTTTTTTTTGG-3?. PCR products that
represented differentially expressed cDNAs were excised from
the differential-display gels, eluted by boiling in 40 ?l of
distilled, deionized H2O for 5 min, amplified by PCR for 20
cycles using the original primers, and separated on 2% aga-
rose?ethidium bromide gels to confirm the size of the ream-
plified product. The reamplified PCR products were ligated
into the multiple cloning site of TA cloning vector pCR2.1
(Invitrogen) according to the manufacturer’s protocol; after
ligation, the vector was transfected into the TOP10F strain of
Escherichia coli. Individual bacterial colonies were picked, and
plasmid DNA was isolated (Wizard miniprep; Promega). To
confirm that the cloned PCR products isolated by TA cloning
were overexpressed in the RNA fingerprinting reaction mix-
ture from the drug resistant cells, a ‘‘reverse’’ Northern
analysis was performed. Plasmid DNA isolated from 12 dif-
ferent colonies of transfected E. coli was fixed in duplicate to
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked ‘‘advertisement’’ in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
© 1998 by The National Academy of Sciences 0027-8424?98?9515665-6$2.00?0
PNAS is available online at www.pnas.org.
Abbreviations: RT-PCR, reverse transcription–PCR; Pgp, P-
glycoprotein; MRP, multidrug resistance protein; ABC, ATP-binding
cassette; BCRP, breast cancer resistance protein; w, Drosophila white
Data deposition: The sequence reported in this paper has been
deposited in the GenBank database (accession no. AF09895).
¶To whom reprint requests should be addressed at: Greenebaum
Cancer Center of the University of Maryland, Room 9–015 Bressler
Research Building, 655 West Baltimore Street, Baltimore, MD 21201.
Zeta Probe-GT (Bio-Rad, Richmond, CA) membranes in a
slot-blot apparatus. One of the duplicate membranes was
probed with the
MCF-7 cDNA by using the original ‘‘P’’ and ‘‘T’’ primers in the
the original33P-labeled parallel PCR reaction mixture that
standard Northern blot conditions of hybridization, after
which the binding of probe was assessed by using autoradiog-
Construction of cDNA Library. A cDNA library was con-
structed from MCF-7?AdrVp RNA by using the CapFinder
PCR cDNA library construction kit (CLONTECH) according
to the manufacturer’s protocol. The CapFinder technique is
designed specifically to produce full-length double-stranded
cDNA. The library was screened with the RNA Fingerprinting
PCR product of interest by using the manufacturer’s recom-
mended protocol. Positive clones were isolated and subjected
to secondary and tertiary screening, with additional testing by
MCF-7?AdrVp, and MCF-7?AdrVpPR cells. Multiple clones
had 2.4-kb inserts, the approximate size of the BCRP mRNA
suggested by Northern blotting. Four 2.4-kb inserts were
ligated into the pCR2.1 plasmid (see above); sequencing of the
2.4-kb cDNA insert was performed by using an automated
DNA sequencer (Perkin–Elmer). All DNA sequences were
confirmed by sequencing in the reverse direction.
Data Analysis. Analyses of cDNA and deduced protein
sequences were accomplished using protein and nucleotide-
sequence databases that were accessed by using the Wisconsin
sequence analysis package, Version 8 (Genetics Computer
Group, Madison, WI) which are available through the Fred-
erick Cancer Research Center’s Supercomputing Facility
(Frederick, MD). Statistical analyses were accomplished with
the MINITAB statistical software (MINITAB release 8 extended;
Minitab, State College, PA).
Reverse Transcription–PCR (RT-PCR). The program
OLIGO (Version 5.0; National Biosciences, Plymouth, MN) was
used to help determine suitable primers for detection of the
human homologue of the Drosophila white gene (w) by RT-
PCR. The upper primer began at 5? position 2,136 of human
w mRNA and had the sequence 5?-CGACCGACGACA-
CAGA-3?; the lower primer began at 3? position 2,590 and had
the sequence 5?-CTTAAAATGAATGCGATTGAT-3?. The
expected PCR product was 475 bp in length. Random hexam-
ers were used to prime the reverse transcription reaction,
which was followed by 25 cycles of PCR. An RT-PCR assay for
?-actin was also performed; reaction conditions for this assay
have been reported (11).
Transfection and Enforced Expression of BCRP in MCF-7
cells. The full-length breast cancer resistance protein (BCRP)
cDNA was inserted into the multiple cloning site of expression
vector pcDNA3 (Invitrogen). After the pcDNA3–BCRP con-
struct was sublconed, DNA sequence analysis was performed
to confirm the insert of the selected clone was in a sense
orientation to the cytomegalovirus (CMV) promoter of the
pcDNA3 vector. MCF-7 cells were transfected with pcDNA3–
BCRP by using the calcium phosphate precipitation method
(12), selected by culture with geneticin (G418, 1 mg?ml), and
subcloned by limiting dilution in 96-well flat-bottomed culture
plates (Sarstedt, Newton, NC). Subclones were tested for
expression of BCRP mRNA by using Northern blot analysis.
As a control, MCF-7 cells were also transfected with the empty
pcDNA3 vector and selected by growth in medium containing
1 mg?ml G418.
Pharmacokinetics of Intracellular Drugs and Effect of ATP
Depletion. The intracellular accumulation and retention of
daunorubicin in MCF-7 cells was determined by using flow
cytometry as described (8). Cells cultured in 25-cm2flasks
(Corning Costar) were exposed to 1 ?g?ml daunorubicin for
33P-labeled PCR mixture that amplified
up to 180 min (accumulation phase) or exposed to daunoru-
bicin for 180 min, washed free of drug with ice-cold saline
solution, and resuspended in prewarmed culture medium in
the absence of drug (retention phase). At the time intervals
indicated in the figure, aliquots of cells were trypsinized off of
the plates, and intracellular daunorubicin content was mea-
sured (8). Controls for binding of anthracycline to plasma
membrane were accomplished by incubating cells with dauno-
rubicin on ice for an appropriate period of time, whereupon
cellular daunorubicin content was determined by using flow
cytometry. The value for membrane binding was subtracted
from the value obtained for intracellular drug content of cells
incubated with drug at 37°C. Intracellular drug content is
expressed in fluorescence units (FU) per cell. Fluorescence
units are arbitrary numbers between 1 and 10,000. Cell vol-
umes were measured by using a Coulter Channelyzer as
MCF-7 cells were depleted of ATP by incubation in glucose-
free DMEM containing 50 mM 2-deoxy-D-glucose and 15 mM
sodium azide for 20 min (37°C). Rhodamine 123 was added
(0.5 ?g?ml final concentration) for an additional 30 min. The
RNA was treated with DNase, reverse-transcribed into cDNA, and
amplified by PCR using upstream and downstream primers and
radiolabeled [33P]dATP. The figure depicts a portion of the autora-
diograph of a 5% polyacrylamide gel electrophoresis of the PCR
mixture by using the primer pair P6 and T9. Lanes 1, 3, and 5 are
reaction mixtures in which cDNA diluted 1:10 was added; lanes 2, 4,
and 6 represent reaction mixtures in which cDNA diluted 1:40 was
added. Lanes 7 and 8 are H2O controls, in which sterile water was
added to the PCR mixture in place of cDNA. Lanes 9, 10, and 11 are
RNA controls, in which 0.02 ?g of cellular RNA from MCF-7?W,
MCF-7?AdrVp, or MCF-7?AdrVpPR cells was added instead of
cDNA. These RNA controls served to indicate contamination of the
RNA with genomic DNA. The arrow indicates a PCR product
representing an mRNA species overexpressed in MCF-7?AdrVp cells,
compared with MCF-7?W or MCF-7?AdrVpPR cells. This product
was excised from the gel for further amplification by PCR. (B)
Northern blot hybridization (Upper) of mRNA from MCF-7?W,
MCF-7?AdrVp, or MCF-7?AdrVpPR cells by using the 795-bp PCR
product obtained from RNA fingerprinting studies and isolated by TA
cloning as a probe after labeling with [32P]dCTP (Prime-a-Gene
labeling kit, Promega). To control for equivalence in sample loading,
the blot was stripped and rehybridized with a radiolabeled probe for
18S RNA (Lower).
(A) RNA fingerprinting of MCF-7 cells. Total cellular
15666 Medical Sciences: Doyle et al. Proc. Natl. Acad. Sci. USA 95 (1998)
cells were placed on ice, washed free of rhodamine, and
incubated under ATP-depleting conditions for an additional
30 min, and rhodamine retention was determined by flow
cytometry (excitation 488 nm, emission 520 nm).
To test the hypothesis outlined in the Introduction, we em-
ployed the technique of RNA fingerprinting, a modification of
the method of differential display that uses PCR and degen-
erate-primer pairs to amplify cellular mRNA. For the PCR
reactions, cDNA was prepared from MCF-7, MCF-7?AdrVp,
and MCF-7?AdrVpPR cells, the latter of which have partially
reverted to drug sensitivity by culture in the absence of
selecting agents. When RNA fingerprinting was performed
with the primers P6 and T9, a PCR product was found
reproducibly in excess in reactions that used cDNA from
MCF-7?AdrVp cells as template, compared with those reac-
tions with cDNA from MCF-7 or MCF-7?AdrVpPR cells (Fig.
1A). The PCR product overexpressed in MCF-7?AdrVp cells
was excised from the dried gel, purified, and ligated into a TA
cloning vector. Screening of the TA vector clones by the
reverse Northern blot method led to the isolation of a single
clone whose PCR product insert identified, by standard North-
ern analysis, a 2.4-kb mRNA species that was markedly
overexpressed in MCF-7?AdrVp cells, compared with MCF-7
(Fig. 1B, Top). The partially revertant MCF-7?AdrVpPR
subline had intermediate expression of the 2.4-kb mRNA
species (Fig. 1B).
The differentially expressed PCR product in the TA clone
was sequenced and found to be a 795-bp cDNA. Protein
database searches of the deduced amino acid sequence re-
vealed a high degree of homology to members of the ATP-
binding cassette (ABC) family of transport proteins. The
795-bp cDNA fragment was radiolabeled and used as a probe
to screen a cDNA library prepared from MCF-7?AdrVp cells.
A clone of ? bacteriophage containing a 2.4-kb cDNA frag-
ment was identified and isolated. The cDNA insert was
completely sequenced and found to be 2,418 bp in length.
Analysis of the cDNA for ORFs using the program FRAMES
contained in the GCG software package indicated the pres-
ence of a long ORF that began at position 239 and ended with
the stop codon TAA at position 2,204–2,206. The deduced
amino acid sequence of this ORF is shown in Fig. 2A. The
protein encoded by this sequence has been designated breast
cancer resistance protein (BCRP). A comparison of the BCRP
amino acid sequence using the GCG software FASTA revealed
a high degree of homology to at least 50 ABC transport
proteins. The highest match was PIR2:G02068, the human
homologue of the Drosophila white (w) gene, which has 638
amino acids and is 29.3% identical to BCRP. The w gene in
Drosophila functions in the cellular transport of guanine and
tryptophan, which are retinal-pigment precursors (13–16). The
human homologue of w (17) is not overexpressed in MCF-7?
AdrVp cells compared with MCF-7 cells as detected by an
RT-PCR assay (data not shown).
Analysis of the BCRP peptide sequence with the GCG
program MOTIFS demonstrated a single Walker ‘‘A’’ ATP?
GTP binding region (18) at amino acids 88–95 and a phos-
phopantetheine attachment site at amino acids 221–236 (Fig.
2A). As the prosthetic group of acyl carrier proteins in some
multienzyme complexes, phosphopantetheine serves in the
attachment of activated fatty acid and amino acid groups (19).
Examination of BCRP structure with GCG programs PEP-
PLOT and PLOTSTRUCTURE revealed a relatively hydrophilic
of motifs. The deduced sequence was obtained from the nucleotide
sequence of full-length BCRP cDNA by applying the program TRANS-
LATE, which is part of the GCG software package. TM 1, TM 2, and
TM 3 refer to potential transmembrane regions; Glyc refers to
potential sites of N-glycosylation. (B) Phylogram of the evolution of
of the ABC family of transport proteins. The phylogram was created
by using the multiple-sequence alignment program PILEUP, which is
part of the GCG software package, using a gap weight of 3.0 and a gap
length weight of 0.1. The GCG program DISTANCES was then applied
to create pairwise evolutionary distances between the aligned se-
quences, by using the Kimura protein-distance correction method. A
graphical phylogenetic tree was then created from the distance matrix
with the GCG program GROWTREE.
(A) Deduced amino acid sequence of BCRP with display
performed by using commercially prepared human multiple-tissue
Northern blots (MTN Blot and MTN Blot II, CLONTECH) and the
manufacturer’s recommended procedure. The 795-bp PCR product
obtained from RNA fingerprinting was used as probe after labeling
with [32P]dCTP. Tissues tested were: heart (lane 1), brain (lane 2),
placenta (lane 3), lung (lane 4), liver (lane 5), skeletal muscle (lane 6),
kidney (lane 7), pancreas (lane 8), spleen (lane 9), thymus (lane 10),
14), colon (lane 15), and peripheral-blood leukocytes (lane 16).
Tissue distribution of BCRP mRNA. This analysis was
Medical Sciences: Doyle et al.Proc. Natl. Acad. Sci. USA 95 (1998) 15667
amino-terminal domain (amino acids 1–400) that contains the
ATP-binding sequence and a relatively hydrophobic carboxyl-
terminal domain (amino acids 401–663) containing at least
three putative transmembrane domains and four potential
N-glycosylation sites (Fig. 2A). The transmembrane domains
were estimated with a program to predict helices in integral
membrane proteins (20). Analysis of the BCRP sequence by
the GCG program DOTPLOT demonstrated that the peptide is
homologous with one-half of the duplicated Pgp or MRP
molecule, except that Pgp or MRP have the configuration
NH2-[transmembrane domains]-[ATP binding 1]-[transmem-
brane domains]-[ATP binding 2]-COOH, whereas BCRP is
NH2-[ATP binding]-[transmembrane domains]-COOH. The
phylogenetic relationship of BCRP to other members of the
ABC transporter superfamily was determined by using the
GCG programs PILEUP, DISTANCES, and GROWTREE. This
The BCRP cDNA was used as a probe in Northern blots to
examine the expression of BCRP mRNA in selected normal
human tissues (Fig. 3). The greatest expression was seen in
placental tissue with considerably lower levels of expression in
brain, prostate, small intestine, testis, ovary, colon, and liver.
BCRP transcripts were below the level of detection in heart,
lung, skeletal muscle, kidney, pancreas, spleen, thymus, and
To evaluate BCRP function in vitro, the full-length cDNA
was inserted into the multiple cloning site of the expression
vector pcDNA3. MCF-7 cells were transfected with the vector
containing the full-length BCRP cDNA (pcDNA3–BCRP) or
with the empty vector as a control. After selection with
geneticin (G418), the pcDNA3–BCRP-transfected cells were
pcDNA3-BCRP or in MCF-7?W cells transfected with empty vector pcDNA3 (uncloned-vector control) after selection in medium containing G418.
The subclones were isolated by plating the transfected cells by limiting dilution in 96-well flat-bottomed culture flasks. The 795-bp PCR product
originally obtained from the differential-display studies was radiolabeled and used as probe. Ethidium bromide stains of 1% agarose gel
electrophoresis of the total cellular RNA used for the Northern blot demonstrated approximate equivalency of sample loading (results not shown).
(B) Daunorubicin (DNR) accumulation and retention of MCF-7 cells transfected with empty pcDNA3 vector (vector control) or with clone 6 or
clone 8 isolated from MCF-7?W cells after transfection with pcDNA3-BCRP. The data points are the mean of duplicate determinations; the vertical
bars represent the upper or lower range for that data point. The cell volumes, measured by Coulter Channelyzer are 2,515 ? 56, 3,074 ? 112, and
2,459 ? 56 ?m3for MCF-7?BCRP-clone 6, MCF-7?BCRP-clone 8, and MCF-7?pcDNA3 vector control cells, respectively. These values are
comparable to our previous measurements of MCF-7 cell volumes (8). (C) Effects of ATP depletion on the retention of rhodamine 123 by
transfectant MCF-7?pcDNA3 (empty vector control) or MCF-7?BCRP clone 8 cells. The cells were incubated in complete medium or under
ATP-depleting conditions (see Materials and Methods) for 20 min, and rhodamine 123 (0.5 ?g?ml final concentration) was added for an additional
30 min. The cells were washed free of rhodamine and returned to culture either in complete medium or under ATP-depleting conditions for an
488 nm, emission 520 nm). The vertical lines over each bar represent the SD of four replicate determinations. Viability (trypan blue dye exclusion)
of the cells in each treatment group was ?90% at the completion of the study. (D) Representative sulforhodamine-B cytotoxicity (21) studies for
mitoxantrone, daunorubicin, doxorubicin, cis-platin, paclitaxel, or vincristine against MCF-7?W (?) or MCF-7?pcDNA3-BCRP clone 8 (I) cells.
These data are typical of those used to obtain LC50 values that comprise the data displayed in Table 1. The vertical bars for each data point
represent ? SD of six replicate determinations.
(A) Northern analysis of the expression of BCRP mRNA in subclones of MCF-7?W cells stably transfected with expression vector
15668 Medical Sciences: Doyle et al. Proc. Natl. Acad. Sci. USA 95 (1998)
cloned by using limiting dilution; multiple clones were tested
for BCRP expression by using Northern blot hybridization
(Fig. 4A). Two BCRP-overexpressing clones (clones 6 and 8)
and one clone that did not overexpress BCRP (clone 19) were
selected for studies of BCRP function in comparison to cells
transfected only with empty vector and to parental MCF-7
cells. The expression of BCRP mRNA in clone 6 was less than
in clone 8 (Fig. 4A).
Daunorubicin accumulation and retention were examined in
transfected cells by using flow cytometry. The BCRP-
overexpressing clones 6 and 8 reproducibly displayed dimin-
ished accumulation and retention of daunorubicin when com-
pared with the vector-transfected controls (Fig. 4B). The
intracellular steady-state concentrations of daunorubicin in
clones 8 and 6, respectively, were approximately 30% or 50%
of that attained in the vector control cells. This was not the
result of differences in cell volume, because the volumes of the
BCRP-overexpressing sublines tested were not lower than the
empty vector-transfected control cells. In previous studies (8),
the intracellular steady-state accumulation of daunorubicin in
MCF-7?AdrVp cells was approximately 25% of that in MCF-
7?W cells, which is comparable to the accumulation of dauno-
rubicin that we observed in transfectant clone 8 in the current
The transport function of BCRP appears to depend on ATP
(Fig. 4C). The intracellular retention of rhodamine 123 in
BCRP-overexpressing clone-8 cells was diminished to 45% of
that of vector control cells. Depletion of ATP increased
rhodamine retention in clone 8 cells to levels comparable to
those of vector control cells cultured in complete medium but
had no effect on the rhodamine retention in the vector control
cells (Fig. 4C).
The sensitivities of the various transfected sublines to che-
motherapeutic agents were tested by the sulforhodamine-B
cytotoxicity assay (21). The BCRP-overexpressing clones 6 and
8 displayed resistance to mitoxantrone, daunorubicin, and
doxorubicin compared with non-BCRP-overexpressing clone
19 cells, MCF-7 cells, or the empty vector-transfected controls
(Table 1; Fig. 4D). Like MCF-7?AdrVp cells, the MCF-7?
BCRP transfectant clones 6 and 8 displayed the greatest
degree of resistance to mitoxantrone. The pattern of crossre-
sistance displayed by the BCRP-overexpressing transfected
cells is very similar to the pattern displayed by MCF-7?AdrVp
cells; however, MCF-7?AdrVp cells have a greater relative
resistance to all cytotoxic drugs within the phenotype. The
BCRP-overexpressing clones 6 and 8 are sensitive to cis-platin,
paclitaxel, and vincristine, as are MCF-7?AdrVp cells (Table
1). BCRP transfectant clone 19, which did not overexpress
BCRP, met the minimal statistical criterion (P ? 0.0497) for
significance of a low degree of resistance to cis-platin (resis-
tance factor ? 2.7) in comparison to MCF-7?W but not in
comparison to cells transfected with the empty vector (MCF-
7?pcDNA3; see Table 1). With this exception, the sensitivity
of clone 19 was not statistically different from MCF-7?W or
MCF-7?pcDNA3 for the drugs tested.
The data presented here strongly support the conclusion that
the novel ABC-transport protein family member BCRP is an
ATP-dependent xenobiotic transporter that plays a major role
in the drug-resistance phenotype of MCF-7?AdrVp cells. The
overexpression of BCRP mRNA in MCF-7?AdrVp cells,
which is diminished in MCF-7?AdrVpPR, suggests an impor-
tant role for BCRP in resistance to cytotoxic agents. Further-
more, the enforced overexpression of BCRP in MCF-7 cells
diminished daunorubicin cellular accumulation and imparted
a pattern of drug crossresistance to the transfected cells that
was virtually identical to that of MCF-7?AdrVp cells. The
degree of BCRP overexpression in transfectant clones 6 and 8
Table 1. Effects of chemotherapeutic drugs on BCRP-transfected MCF-7 cells
BCRPclone19MCF-7?BCRP clone6 MCF-7?BCRP clone8 MCF-7?AdrVp
[Range, nM] N RF
[Range, nM] N RF
[Range, nM] N RF
[Range, nM] N RF
[Range, nM] N RF
Mitoxantrone41 7 1.0 406 1.054
49.0 5 32160,0002 3,902
Daunorubicin7 1.05 1.6 1 1.6
44.35 8.33 41
Doxorubicin 5 1.04 0.94 1.343.0
Cisplatin 4 1.0 3 1.53 2.73 273 1.52 1.1
Paclitaxel4 1.02 2.7 3 0.53 1.03 1.641.0
Vincristine3 1.02 1.53 2.43 1.43 1.133.5
(nM), the number of experimental determinations of LC50that were performed (N), and the resistance factor (RF). The RF was calculated by
dividing the median LC50for a given drug against a transfected cell line by the median LC50of that drug against nontransfected MCF-7?W cells.
For each drug tested, the LC50for the BCRP-transfected cells was examined for statistically significant difference from the LC50of MCF-7?W or
MCF-7?pcDNA3 by the Mann–Whitney test, using MINITAB statistical software MINITAB release 8 extended, Minitab, State College, PA) and a 95%
confidence interval. The values of P for the statistically significant differences are as follows Mitoxantrone: MCF-7?W vs. MCF-7?BCRPclone6,
P ? 0.0107, MCF-7?W vs. MCF-7?BCRPclone8, P ? 0.0058, MCF-7?pcDNA3 vs. MCF-71BCRPclone6, P ? 0.0142, MCF-7?pcDNA3 vs.
MCF-7?BCRPclone8, P ? 0.0081; daunorubicin: MCF-7?W vs. MCF-7?BCRPclone6, P ? 0.0107, MCF-7?W vs. MCF-7?BCRPclone8, P ?
0.0058, MCF-7?pcDNA3 vs. MCF-7?BCRPclone6, P ? 0.0195, MCF-7?pcDNA3 vs. MCF-7?BCRPclone8, P ? 0.0163; doxorubicin: MCF-7?W
vs. MCF-7?BCRPclone6, P ? 0.0373, MCF-7?W vs. MCF-7?BCRPclone8, P ? 0.02, MCF-7?pcDNA3 vs. MCF-7?BCRPclone8, P ? 0.0304;
cis-platin: MCF-7?W vs. MCF-7?BCRPclone19, P ? 0.0497.
*Differs significantly from MCF-7?W, P ? 0.05 (Mann–Whitney test).
†Differs significantly from MCF-7?pcDNA3 (empty vector control, P ? 0.05, Mann–Whitney U test).
Medical Sciences: Doyle et al.Proc. Natl. Acad. Sci. USA 95 (1998)15669
correlates with the alterations in the intracellular steady-state
level of daunorubicin and their degree of resistance to mitox-
antrone, daunorubicin, and doxorubicin.
A major difference between the BCRP-overexpressing
transfectant clones and the original MCF-7?AdrVp subline is
that the degree of drug resistance in the latter is greater than
in the transfected cells; however, the steady-state BCRP
mRNA levels in the transfectants are comparable to those of
MCF-7?AdrVp cells (Fig. 4A). A number of possibilities may
contribute to this difference. Alterations in protein stability,
localization, or both may contribute to the full drug-resistant
phenotype, or the expression of other proteins may be re-
quired. Recently, we reported members of the carcinoembry-
onic antigen (CEA) family, primarily the nonspecific cross-
reacting antigen and CEA itself, are markedly overexpressed
on the cell surface of MCF-7?AdrVp and MCF-7?AdrVpPR
cells compared with drug-sensitive MCF-7 cells (22). A high
density of these acidic glycoproteins on the cell surface may
protonate drugs such as mitoxantrone, daunorubicin, or doxo-
rubicin, which will prevent entry into the cell. Indeed, Kawa-
harata et al. (23) reported that the enforced expression of CEA
in transfected NIH 3T3 cells resulted in both diminished
accumulation of and resistance to doxorubicin in the trans-
fected cells. Hence, the relative overexpression of CEA family
members on the MCF-7?AdrVp cell surface could act in
concert with BCRP to cause greater resistance to mitox-
antrone, doxorubicin, and daunorubicin than could BCRP
alone. This hypothesis could be tested by cotransfecting the
MCF-7?BCRP-clone 8 subline with an expression vector con-
taining the nonspecific cross-reacting antigen or CEA.
Another possible explanation for the greater degree of
resistance of MCF-7?AdrVp cells compared with the trans-
fectants is that BCRP may participate in a multiprotein
transporter complex. The translocation pathway of typical
ABC transporters consists of two ATP-binding domains and
two highly hydrophobic domains that contain membrane-
spanning regions. This configuration can be accomplished in a
single large molecule (e.g., MRP or Pgp). Alternatively, the
active complex of certain ABC transporters can be formed by
the heterodimerization of two nonidentical proteins, each
containing a single ATP-binding and hydrophobic region. The
w and brown (b) proteins of Drosophila and the Tap-1 and
Tap-2 proteins that transport major histocompatibility class I
peptides are examples of ABC-transport protein family mem-
bers that exhibit such a cooperative interaction. The presence
of the phosphopantetheine attachment site on BCRP suggests
that BCRP may be part of a multiprotein complex. Possibly
BCRP has one or more protein cofactors that function as a
much more efficient transporter in a heteromeric state. The
activation or overexpression of this cofactor in MCF-7?AdrVp
relative to MCF-7 cells could explain the greater drug resis-
tance in the MCF-7?AdrVp subline relative to the BCRP
Note added in Proof. A BLAST-N search of the nonredundant database
of GenBank human expressed sequence tag (EST) entries for homol-
ogy to the BCRP cDNA sequence revealed 46 ‘‘blast hits.’’ The highest
score was clone EST57481, which was predicted to represent 1 of 21
new genes of the human ABC family, as reported previously (24).
EST157481 is almost identical to bases 1,187–2,016 of the BCRP
cDNA sequence as recorded in the GenBank database.
We thank Drs. Susan Bates and Antonio Fojo of the Medicine
Branch, National Cancer Institute, National Institutes of Health, for
the gift of MCF-7?AdrVp sublines. We also thank Dr. Gary Smythers
of the Supercomputing Facility of the Frederick Cancer Research
Center, National Cancer Institute, National Institutes of Health for
help with the GCG software and identifying potential transmembrane
domains of BCRP. We are grateful to Dr. De-Qi Xu for assistance in
preparing the cDNA library from MCF-7?AdrVp cells. We thank Drs.
Raji Sridhara and Julie Eiseman for assistance with the statistical
analysis of data. This work was supported in part by Public Health
Service Grant CA52178 to D.D.R. from the National Cancer Institute,
National Institutes of Health, Department of Health and Human
Services, and by a Veteran’s Affairs merit review grant to D.D.R.
1. Kessel, D., Botterill, V. & Wodinsky, I. (1968) Cancer Res. 28,
Biedler, J. L. & Riehm, H. (1970) Cancer Res. 30, 1174–1184.
Ling, V. & Thompson, L. H. (1974) J. Cell. Physiol 83, 103–116.
Gros, P., Ben Neryah, Y. B., Croop, J. M. & Housman, D. E.
(1986) Nature (London) 323, 728–731.
Cole, S. P. C., Bhardwaj, G., Gerlach, J. H., Mackie, J. E., Grant,
C. E., Almquist, K. C., Stewart, A. J., Kurz, E. U., Duncan,
A. M. V., et al. (1992) Science 258, 1650–1654.
Chen, Y.-N., Mickley, L. A., Schwartz, A. M., Acton, E. M.,
Hwang, J. & Fojo, A. T. (1990) J. Biol. Chem. 265, 10073–10080.
Lee, J. S., Scala, S., Matsumoto, Y., Dickstein, B., Robey, R.,
Zhan, Z., Altenberg, G. & Bates, S. E. (1997) J. Cell. Biochem.
Doyle, L. A., Ross, D. D., Sridhara, R., Fojo, A. T., Kaufmann,
S. H., Lee, E. J. & Schiffer, C. A. (1995) Br. J. Cancer 71, 52–58.
Liang, P. & Pardee, A. (1992) Science 257, 967–971.
Liang, P., Averboukh, L., Keyomarsi, K., Sager, R. & Pardee, A.
(1992) Cancer Res. 52, 6966–6968.
Ross, D. D., Doyle, L. A., Schiffer, C. A., Lee, E. J., Grant, C. E.,
Cole, S. P. C., Deeley, R. G., Yang, W. & Tong, Y. (1996)
Leukemia 10, 48–55.
Ausubel, F., Kingston, R., Moore, D., Seidman, J., Smith, J. &
Struhl, K., eds. (1988) Current Protocols in Molecular Biology
(Wiley, New York), Vol. 1, Chap. 9.
Morgan, T. H. (1910) Science 32, 120–122.
Bingham, P. M., Levis, R. & Rubin, G. M. (1981) Cell 25,
O’Hare, K., Murphy, C., Levis, R. & Rubin, G. M. (1984) J. Mol.
Biol. 180, 437–455.
Pepling, M. & Mount, S. M. (1990) Nucleic Acids Res. 18, 1633.
Chen, H., Rossier, C., Lalioti, M. D., Lynn, A., Chakravarti, A.,
Perrin, G. & Antonarkis, S. E. (1996) Am. J. Hum. Genet. 59,
Walker, J. E., Saraste, M., Runswick, M. J. & Gay, N. J. (1982)
EMBO J. 1, 945–951.
Pugh, E. L. & Wakil, S. J. (1965) J. Biol. Chem. 240, 4727–4733.
Rao, J. K. M. & Garos, P. (1986) Biochim. Biophys. Acta 899,
Skehan, P., Storeng, R., Scudiero, D., Monks, A., McMahon, J.,
Vistica, D., Warren, J. T., Bokesch, H., Kenny, S. & Boyd, M. R.
(1990) J. Natl. Cancer Inst. 82, 1107–1112.
Ross, D. D., Gao, Y., Yang, Y., Leszyk, J., Shively, J. & Doyle,
L. A. (1997) Cancer Res. 57, 5460–5464.
Kawaharata, H., Hinoda, Y., Itoh, F., Endo, T., Oikawa, S.,
Nakazato, H. & Imai, K. (1997) Int. J. Cancer 72, 377–382.
Allikmets, R., Gerrard, B., Hutchinson, A. & Dean, M. (1996)
Hum. Mol. Genet. 5, 1649–1655.
15670 Medical Sciences: Doyle et al.Proc. Natl. Acad. Sci. USA 95 (1998)
Cell Biology. In the article “ADP-ribosylation factor and
coatomer-coated vesicles” by Mark Stamnes, Giampietro
Schiavo, Gudrun Stenbeck, Thomas H. So ¨llner, and James E.
Rothman, which appeared in number 23, November 10, 1998,
of Proc. Natl. Acad. Sci. USA (95, 13676–13680), the authors
wish to note the following correction. The label, described
correctly in the legend to Fig. 1, incorrectly indicated that
coatomer was included in stage 1 of the two-stage reactions. A
corrected figure and its legend are shown below.
Medical Sciences. In the article “Gadolinium(III) texaphyrin:
A tumor selective radiation sensitizer that is detectable by
MRI” by Stuart W. Young, Fan Qing, Anthony Harriman,
Jonathan L. Sessler, William C. Dow, Tarak D. Mody, Gregory
W. Hemmi, Yunpeng Hao, and Richard A. Miller, which
appeared in number 13, June 25, 1996, of Proc. Natl. Acad. Sci.
USA (93, 6610–6615), the following correction should be
noted. It has come to our attention that the radiation sensi-
tivity of the HT29 control cell line reported in Fig. 2 on page
6611 is inconsistent with that reported in the cited literature
(33). Consequently, the in vitro HT29 radiation sensitization
experiments have been repeated with Gd-tex2?(compound 1).
Radiation enhancement comparable to our original findings
was observed at doses between 8 and 20 Gy. The results
indicate that the Gy scale reported on the x-axis of Fig. 2 is
incorrect. We apologize for this error. The conclusions
reached in the article remain unchanged.
Medical Sciences. In the article “Production of ?-defensins by
human airway epithelia” by Pradeep K. Singh, Hong Peng Jia,
Kerry Wiles, Jay Hesselberth, Lide Liu, Barbara-Ann D.
Conway, Everett P. Greenberg, Erika V. Valore, Michael J.
Welsh, Tomas Ganz, Brian F. Tack, and Paul B. McCray, Jr.,
which appeared in number 25, December 8, 1998, of Proc. Natl.
Acad. Sci. USA (95, 14961–14966), due to a printer’s error, the
following change should be noted: the symbol for Brian F.
Tack should be ‡, to indicate that he is affiliated with the
Department of Microbiology of the University of Iowa College
Medical Sciences. In the article “A multidrug resistance
transporter from human MCF-7 breast cancer cells” by L.
Austin Doyle, Weidong Yang, Lynne V. Abruzzo, Tammy
Krogmann, Yongming Gao, Arun K. Rishi, and Douglas D.
Ross, which appeared in number 26, December 22, 1998, of
Proc. Natl. Acad. Sci. USA (95, 15665–15670), the following
corrections should be noted. In the abstract and text, later
amino acids as stated in the article. The first 8 amino acids
displayed in Fig. 2A on page 15667 should be removed, making
the initial sequence of the peptide MSSSNVEVFI. . . .
On page 15665 in the data deposition footnote, the Gen-
Bank database accession number for BCRP is incorrect. The
correct accession number is AF098951.
On page 15670 in the “Note Added in Proof,” the accession
number for the human EST clone that was homologous to
BCRP is incorrect. The correct number is HUEST157481.
tions. The amount of membrane-bound ARF and coatomer (?-COP)
determined by Western blot analysis of binding reactions. Lanes 1 and
2 show one-stage reactions in which ARF and coatomer were incu-
bated together. Lanes 3–6 show the results from two-stage reactions
in which the membranes were first incubated with ARF but not
coatomer, reisolated, and then incubated in a second stage with
coatomer but not ARF. As controls, membranes (lane 2) or ARF (lane
3) were excluded from stage 1, or coatomer (lane 4) was excluded from
both stages. All incubations were carried out at 37°C except lane 5,
which was carried out at 0°C.
ARF and coatomer binding in one- and two-stage reac-
Corrections Proc. Natl. Acad. Sci. USA 96 (1999) 2569