ArticlePDF Available

Dynamic Ligand-induced Conformational Rearrangements in P-glycoprotein as Probed by Fluorescence Resonance Energy Transfer Spectroscopy



P-glycoprotein (Pgp), a member of the ATP-binding cassette transporter family, functions as an ATP hydrolysis-driven efflux pump to rid the cell of toxic organic compounds, including a variety of drugs used in anticancer chemotherapy. Here, we used fluorescence resonance energy transfer (FRET) spectroscopy to delineate the structural rearrangements the two nucleotide binding domains (NBDs) are undergoing during the catalytic cycle. Pairs of cysteines were introduced into equivalent regions in the N- and C-terminal NBDs for labeling with fluorescent dyes for ensemble and single-molecule FRET spectroscopy. In the ensemble FRET, a decrease of the donor to acceptor (D/A) ratio was observed upon addition of drug and ATP. Vanadate trapping further decreased the D/A ratio, indicating close association of the two NBDs. One of the cysteine mutants was further analyzed using confocal single-molecule FRET spectroscopy. Single Pgp molecules showed fast fluctuations of the FRET efficiencies, indicating movements of the NBDs on a time scale of 10-100 ms. Populations of low, medium, and high FRET efficiencies were observed during drug-stimulated MgATP hydrolysis, suggesting the presence of at least three major conformations of the NBDs during catalysis. Under conditions of vanadate trapping, most molecules displayed high FRET efficiency states, whereas with cyclosporin, more molecules showed low FRET efficiency. Different dwell times of the FRET states were found for the distinct biochemical conditions, with the fastest movements during active turnover. The FRET spectroscopy observations are discussed in context of a model of the catalytic mechanism of Pgp.
Dynamic Ligand-induced Conformational Rearrangements in
P-glycoprotein as Probed by Fluorescence Resonance Energy
Transfer Spectroscopy
Received for publication, September 5, 2011, and in revised form, November 7, 2011 Published, JBC Papers in Press, November 15, 2011, DOI 10.1074/jbc.M111.301192
Brandy Verhalen
, Stefan Ernst
, Michael Bo¨ rsch
, and Stephan Wilkens
From the
Department of Biochemistry and Molecular Biology, State University of New York Upstate Medical University, Syracuse,
New York 13210 and the
Single Molecule Microscopy Group, Jena University Hospital, Friedrich Schiller University Jena,
07743 Jena, Germany
Background: P-glycoprotein is an ATP-binding cassette transporter involved in multidrug resistance.
Results: The two nucleotide binding domains are found to be in close association during the catalytic cycle as determined by
fluorescence spectroscopy.
Conclusion: Small distance changes were observed during ATP hydrolysis supporting an alternating site mechanism.
Significance: Understanding the mechanism of P-glycoprotein is pertinent for developing inhibitors aimed at overcoming
multidrug resistance.
P-glycoprotein (Pgp), a member of the ATP-binding cassette
transporter family, functions as an ATP hydrolysis-driven efflux
pump to rid the cell of toxic organic compounds, including a variety
of drugs used in anticancer chemotherapy. Here, we used fluores-
cence resonance energy transfer (FRET) spectroscopy to delineate
the structural rearrangements the two nucleotide binding domains
(NBDs) are undergoing during the catalytic cycle. Pairs of cysteines
were introduced into equivalent regions in the N- and C-terminal
NBDs for labeling with fluorescent dyes for ensemble and single-
molecule FRET spectroscopy. In the ensemble FRET, a decrease of
the donor to acceptor (D/A) ratio was observed upon addition of
drug and ATP. Vanadate trapping further decreased the D/A ratio,
indicating close association of the two NBDs. One of the cysteine
mutants was further analyzed using confocal single-molecule
FRET spectroscopy. Single Pgp molecules showed fast fluctuations
of the FRET efficiencies, indicating movements of the NBDs on a
time scale of 10 –100 ms. Populations of low, medium, and high
FRET efficiencies were observed during drug-stimulated MgATP
hydrolysis, suggesting the presence of at least three major confor-
mations of the NB D s during catalysis. Under conditions of vanadate
trapping, most molecules displayed high FRET efficiency states,
whereas with cyclosporin, more molecules showed low FRET effi-
ciency. Different dwell times of the FRET states were found for the
distinct biochemical conditions, with the fastest movements during
active turnover. The FRET spectroscopy observations are discussed in
context of a model of the catalytic mechanism of Pgp.
ATP-binding cassette (ABC)
transporters constitute a
superfamily of membrane proteins that couple the hydrolysis of
MgATP to substrate translocation across a lipid bilayer (1–3).
In bacteria, ABC transporters can function in either import or
export of nutrients and toxic molecules, respectively, whereas
in eukaryotes, transport occurs exclusively in the direction of
export (4). The 48 human ABC transporters are responsible for
the transmembrane transport of structurally diverse com-
pounds such as bile acids, carbohydrates, nucleosides, sterols,
peptides, inorganic ions, and environmental toxins (5). An
important member of the mammalian ABC transporter family
is P-glycoprotein, a 170-kDa plasma membrane protein that is
involved in the export of a large variety of structurally unrelated
organic molecules (6, 7). Human P-glycoprotein (Pgp; ABCB1;
MDR1) is expressed in tissues that function in detoxification
such as liver, placenta, and the blood-brain barrier. In certain
cancers, high level expression of Pgp (along with other ABC
transporters such as MRP1 and ABCG2) in the plasma mem-
brane can result in the failure of chemotherapy by preventing
the mostly hydrophobic anticancer drugs from entering the
cytoplasm or the nucleus, resulting in what is known as multi-
drug resistance (8–10).
Many eukaryotic ABC transporters, including Pgp, are
expressed as single polypeptides organized in four domains as
follows: two 6
-helix transmembrane domains (TMD) and two
cytoplasmic nucleotide binding domains (NBDs). The four
domains of Pgp are arranged in the order N-TMD1-NBD1-
TMD2-NBD2-C with an 60-amino acid-long linker connect-
ing NBD1 and TMD2 (11). To carry out productive MgATP
hydrolysis coupled to drug transport, the two NBDs have to
interact in a head-to-tail fashion, thereby sequestering the
nucleotide(s) at the NBD1-NBD2 interface. In this so-called
“sandwich” configuration, nucleotide is interacting with the
* This work was supported, in whole or in part, by National Institutes of Health
Grants CA100246 and GM058600 (to S. W.).
This article contains supplemental Figs. S1 and S2.
To whom correspondence should be addressed: Dept. of Biochemistry and
Molecular Biology, State University of New York Upstate Medical Univer-
sity, Syracuse, NY 13210. E-mail:
The abbreviations used are: ABC, ATP-binding cassette; cps, counts/s; Pgp,
P-glycoprotein; NBD, nucleotide binding domain; TMD, transmembrane
domain; D/A, donor to acceptor; ATP
S, adenosine 5-O-(3-thio)triphos-
phate; DDM, n-dodecyl-
-D-maltopyranoside; TCEP, tris(2-carboxyethyl)
phosphine; CHAPS, 3-[(3-cholamidopropyl)-dimethylammonio]-1-pro-
pane sulfonate/N,N-dimethyl-3-sulfo-N-[3-[[3
droxy-24-oxocholan-24-yl]amino]propyl]-1-propanaminium hydroxide;
PC, phosphatidylcholine; PA, phosphatidic acid; cpms, counts/ms; smFRET,
single-molecule FRET; AMPPNP, adenosine 5-(
THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 287, NO. 2, pp. 1112–1127, January 6, 2012
© 2012 by The American Society for Biochemistry and Molecular Biology, Inc. Published in the U.S.A.
by guest on October 27, 2015 from
phosphate binding loop (P-loop) of one NBD and the ABC sig-
nature motif (LSGGQ) of the other NBD and vice versa. High
affinity binding of transport substrate (hereafter referred to as
“drug”) occurs in an amphipathic cavity that is formed by the
two TMDs toward the cytoplasmic side of the transporter (11).
The drug-binding cavity can accommodate a large variety of
structurally unrelated molecules and exhibits distinct (but
overlapping) binding regions for different subsets of com-
pounds (e.g. the “H” and “R” sites) (12); simultaneous binding of
two drug molecules has been reported for Pgp (11–14).
Despite ongoing efforts, the mechanism by which transport
of drug (from the inner to the outer leaflet of the membrane) is
coupled to the hydrolysis of one or two molecules of ATP is still
not fully understood. Several models have been proposed based
on a combination of site-directed mutagenesis and biochemical
experiments involving transport assays and trapping of cata-
lytic intermediates using inorganic vanadate (V
) or fluoroalu
minate followed by photoaffinity labeling with 8-azido-nucleo-
tide analogs (15–20). These studies revealed that the two NBDs
act in a cooperative manner in that mutating of a catalytically
essential residue in only one of the two NBDs resulted in virtu-
ally complete loss of drug-stimulated ATPase activity. How-
ever, some aspects of the mechanism such as the number of
ATP molecules hydrolyzed for each transport event and
whether the NBDs have to come apart to enable nucleotide
and/or drug binding remain controversial.
Over the last few years, crystal structures have been deter-
mined for a number of bacterial ABC transporters, and these
structural models have provided a molecular picture of some of
the catalytic intermediates. According to the current mechanis-
tic model for ABC transporter function (11, 21), Pgp can exist in
at least two major conformations during the catalytic cycle as
follows: an “inward”-facing conformation, with the drug bind-
ing pocket exposed to the cytoplasmic side and separated
NBDs, and an “outward”-facing conformation with a low affin-
ity drug-binding site exposed to the extracellular space and
closely interacting NBDs. Conversion of the inward- to out-
ward-facing conformation requires binding of MgATP,
whereas subsequent ATP hydrolysis and/or product release
involving one (or both) NBD resets the transporter to the
ground state (the inward-facing conformation). More recently,
the first crystal structures for Pgp (mouse Mdr3) have been
reported that show the transporter in the inward-facing con-
formation (11). Interestingly, irrespective of whether drug
(inhibitor) was bound or not, the overall structures were virtu-
ally identical, including a 20-Å separation of the NBDs (11).
To characterize the structure of Pgp in its natural environ-
ment, we previously generated two-dimensional crystals of the
mouse and human protein reconstituted in the lipid bilayer.
Electron microscopic projection structures of the two-dimen-
sional crystals showed the two halves of the protein in close
contact in projection (22). In the absence of drugs and/or nucle-
otides, a large cavity was observed in the center of the trans-
porter, consistent with the 3.8-Å resolution crystal structure of
detergent-solubilized apo-Pgp (11). Addition of various nucle-
otides with or without drug and inorganic vanadate to the two-
dimensional crystals resulted in the disappearance of the cen-
tral cavity and a sideways motion of the two halves of the
transporter under some conditions but not others (23). How-
ever, because drugs and/or nucleotides were added after two-
dimensional crystal formation, a clear correlation between sub-
strate combination and structural changes could not be drawn.
Structural changes in Pgp were also investigated by chemical
cross-linking via disulfide bonds involving naturally occurring
cysteine residues in the P-loop or cysteines introduced into the
signature motif by site-directed mutagenesis (24–26). In one
study, disulfide bond formation is observed between the P-loop
cysteines, and in another, the disulfide bonds were formed
between a P-loop cysteine of one and the cysteine introduced
into the signature sequence of the other NBD. In the latter
study, efficiency of the cross-links was affected by the nature of
the transport substrate with inhibitors resulting in little or no
disulfide bond formation. It should be noted that formation of
disulfide bonds under the experimental conditions employed in
these studies is irreversible and that the resulting cross-linked
species may represent conformations that are not found during
the catalytic cycle because the cysteines used for cross-link for-
mation are in the active site, and thus the cross-linked species is
likely inactive. To investigate the movements of the NBDs dur-
ing drug-stimulated ATP hydrolysis, we recently introduced
cysteine residues into the C-terminal regions of the NBDs for
disulfide cross-linking. The data showed that linking the C-ter-
minal regions of the NBDs does not abolish drug-stimulated
ATPase activity, suggesting that a large separation of the two
NBDs at their C-terminal ends is not necessary for drug binding
and ATP hydrolysis (27).
A common limitation of the available structural data is that
they provide a static picture of a dynamic transport protein. To
directly monitor the structural changes Pgp is undergoing dur-
ing ATP hydrolysis-driven drug translocation, we designed a
FRET spectroscopy-based system to monitor structural
changes in real time via fluorescent dyes attached to cysteine
residues introduced into equivalent regions of the N- and
C-terminal NBDs using site-directed mutagenesis. For all dou-
ble cysteine mutants tested, we observed that FRET efficiency
increases upon addition of MgATP and transport substrate to
apo-Pgp, suggesting NBD dimerization (NBD “sandwich” for-
mation) under these conditions. Much smaller changes were
seen in the presence of the nonhydrolyzable nucleotide analog
S, and taken together, the data suggest that hydrolysis of
ATP in at least one of the catalytic sites is required for efficient
NBD sandwich formation. Further analysis of one the double
cysteine mutants by single-molecule FRET spectroscopy
revealed distance fluctuations in the NBDs with different dwell
times. Three major conformations of catalytically active Pgp
were characterized by fluctuations between high to medium
and medium to low FRET efficiencies, respectively. Although
drug-stimulated ATP hydrolysis populated both transitions
about equally, trapping of ADP in the presence of inorganic
vanadate shifted the population toward the higher FRET states,
whereas in presence of the modulator cyclosporin, many mol-
ecules with low FRET efficiency transitions were observed.
Dwell time analyses of the FRET states supported drug-depen-
dent changes in NBD dynamics. The ensemble and single-mol-
ecule FRET spectroscopy data are discussed in context of the
mechanism of drug-stimulated MgATP hydrolysis by Pgp.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
Construct Design and Site-directed Mutagenesis—Site-di-
rected mutagenesis was carried out using a cysteine-less con-
struct of Mus musculus Mdr3 in which the seven native cys-
teines are changed to alanines with a six-histidine tag fused to
the C terminus for affinity purification (pHILmdr3CLHis
DNA kindly provided by Gregory Tombline and Alan Senior,
University of Rochester, Rochester, NY). Site-directed
mutagenesis was performed to change the following residue
pairs: S375C/Q1020C, T492C/S1137C, H514C/D1159C,
S586C/S1231C, and R613C/A1258C. These mutants are here-
after referred to as SQ, TS, HD, SS, and RA, respectively. Con-
structs were confirmed by DNA sequencing (State University of
New York Upstate DNA Sequencing Core Facility). The
plasmid was then digested with NotI and trans
formed into Pichia pastoris using electroporation as described
previously (29).
Protein Expression—Expression of cysteine-less and mutant
mouse Pgp in P. pastoris was carried out essentially as described
(27, 30). Briefly, 1 liter of 50% glycerol supplemented with 4.25
ml/liter Pichia trace metals (Pichia manual, Invitrogen) was fed
to the inoculated fermenter over 24 48 h. Upon consumption
of the glycerol, 100% methanol supplemented with 4.25 ml/liter
Pichia trace metals was fed to induce expression with increas-
ing flow rates over 3648 h. Temperature was maintained at
29 °C. Dissolved oxygen content was monitored and main-
tained 35% with increased agitation and an oxygen feed. The
pH was maintained between 4.7 and 5.0 with NH
OH. Cells
were harvested with a final wet cell weight between 240 and 300
g/liter, resuspended 1:1 in mPIB (0.33
M sucrose, 0.3 M Tris/
HCl, 0.05
M NaCl, 1 mM EDTA, 1 mM EGTA, pH 7.4, at 4 °C),
and frozen at 80 °C in 87 g of cell/87 ml of buffer aliquots.
Protein Purification—Purification of double cysteine
mutants was done as described (27, 30). All purification steps
were carried out at 4 °C unless noted otherwise. 87 g of cells
were thawed overnight at 4 °C. Cells were diluted by adding an
equal volume of mPIB without sucrose in the presence of 1 m
dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride
(PMSF), 2
g/ml leupeptin, 2
g/ml pepstatin, and 0.5
chymostatin. Cells were homogenized with a homemade bead
beater with 0.5-mm zirconia beads (Biospec) with 12 1-min
bursts, keeping the temperature of the cell suspension below
10 °C by cooling with a salt/ice mixture. After adding 1 m
fresh PMSF, the cells were then sonicated at a setting of 20 watts
for 6 min in 1-min intervals on ice followed by the addition of 1
M PMSF. The lysate was centrifuged at 3,500 g for 15 min.
The microsomal fraction was obtained by ultracentrifugation
as described previously (27). Microsomes were resuspended at
a protein concentration of 4 mg/ml in solubilization buffer (30%
glycerol, 0.05
M NaCl, 0.05 M Tris/HCl, 0.01 M imidazole, pH
7.4) and extracted with 1.2% dodecyl maltopyranoside (DDM)
in the presence of protease inhibitors. Solubilized membranes
were incubated on ice for 10 min and centrifuged at 64,000 g
in a type 70Ti rotor. The supernatant was incubated for 2–4 h
with co- nitrilotriacetic acid-agarose, loaded onto an AKTA
FPLC with 20 column volumes of wash (0.01
M imidazole,
M NaCl, 0.05 M Tris/HCl, 30% glycerol, 0.1% DDM, 1 mM
-mercaptoethanol, pH 7.4), and eluted either with an imidaz-
ole gradient or directly with elution buffer (0.3
M imidazole,
M NaCl, 0.05 M Tris/HCl, 20% glycerol, 0.1% DDM, 1 mM
-mercaptoethanol, pH 7.4). The fractions were pooled and
diluted 1:3 in DE52 buffer (0.01
M Tris/HCl, 20% glycerol, 0.1%
DDM, 1 m
-mercaptoethanol, pH 7.8) and applied to a 5-ml
DE52 anion exchange column. The flow-through was collected,
and fractions containing purified Pgp were pooled and concen-
trated. Protein was either used immediately after the last puri-
fication step or stored at 80 °C.
Labeling with Fluorescent Dyes for Ensemble Studies—For
mutants SQ and HD, purified protein was applied to a Superdex
200 column (either 16/50 or 10/30; GE Healthcare) in labeling
buffer (0.01
M MOPS/NaOH, pH 7.0, 0.05 M NaCl, 0.5 mM
EDTA, 10% glycerol, 100
M tris(2-carboxyethyl)phosphine),
0.5% CHAPS (or 0.1% DDM for fluorescence experiments with
detergent solubilized samples). Pgp-containing fractions were
pooled, and protein concentration was determined by A
an extinction coefficient of 109,000
. Protein was
incubated at 20 °C with the donor and acceptor dyes for 1 h;
Alexa dyes (Molecular Probes) were used in a 2.5-fold excess,
and Atto dyes (Atto-Tec) were used in a 15–16-fold excess. The
reaction was quenched with 1 m
M cysteine for at least 10 min at
20 °C. For samples in detergent solution, excess dye was
removed with two sequential Sephadex G-50 spin columns
equilibrated in labeling buffer with 0.1% DDM. Concentrations
of protein and dyes were determined by absorbance for deter-
gent-solubilized samples. For labeling of mutants TS and RA,
protein was dialyzed against labeling buffer with a membrane of
3,500-Da MWCO for at least 20 h and one buffer change. Pro-
tein concentration was determined as above. The proteins were
labeled with 2.5-fold excess of Alexa 488 and 21–30-fold excess
of Atto 610 for4hat20°Cinthedark. The labeling reaction was
stopped with 1 m
M cysteine for 10 min at 20 °C. Labeled protein
was placed over Superdex 200 10/30 (GE Healthcare) in label-
ing buffer with 0.5% CHAPS. Fractions were analyzed by SDS-
PAGE, and Pgp-containing fractions were identified with a
Typhoon Imager. Pooled fractions containing Pgp were ana-
lyzed by UV-visible spectroscopy to determine protein/donor/
acceptor ratios. If necessary, labeled Pgp was concentrated
using a 5,000-Da molecular mass cutoff VivaSpin concentrator.
Labeling with Fluorescent Dyes for Single Molecule Studies
Purified protein was applied to a Superdex 200 (10/30; GE
Healthcare) column and eluted in labeling buffer with 0.5%
CHAPS. The protein was reacted with a 2.5-fold excess of Alexa
488 and a 16-fold excess of Atto 610 for1hat20°C.Before lipid
reconstitution, excess dye was quenched with 1 m
M cysteine for
10 min.
Reconstitution into Proteoliposomes—A mixture of 19:1
phosphatidylcholine to phosphatidic acid (w/w) was dried
under a stream of nitrogen. The dried lipid mixture was resus-
pended to 20 mg/ml in water, and 1-ml aliquots were frozen at
80 °C and lyophilized to remove residual solvent. The lyoph-
ilized lipids were then stored at 20 °C until use. Immediately
preceding reconstitution, a 20-mg aliquot was resuspended
into 0.5–1.0 ml of labeling buffer (no detergent) and sonicated
with a vial tweeter until almost clear. The lipid preparation was
then mixed with labeled protein in CHAPS at a 1:20 protein/
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
lipid ratio (w/w) for ensemble studies and 1:1739 protein/lipid
ratio (w/w) for single molecule observations. The protein/lipid/
detergent mixture was placed over an 80-ml Sephadex G-50
column (46 1.5 cm) at room temperature, and 1-ml fractions
were collected. Dye concentrations were determined by visible
absorbance in the presence of 2% CHAPS. Samples were either
used immediately or stored in aliquots at 80 °C.
Ensemble Fluorescence Spectroscopy—Fluorescence spec-
troscopy experiments were carried out in a FluoroLog3-21
spectrofluorometer (HORIBA Jobin Yvon) in a 3 3-mm
quartz cuvette. Temperature was maintained at 37 °C. Excita-
tion and emission slit settings were between 2 and 3 nm. Exci-
tation for Alexa 488 was 493 nm and for Atto as follows: Atto
610, 615 nm; Atto 565, 563 nm; and Atto 655, 663 nm. Emission
was recorded at each wavelength with 0.5-s integration time.
Spectra were corrected for dilution factors. Samples were incu-
bated at 37 °C for 30 min before a 5-min incubation in the
cuvette, which was maintained at 37 °C. Three emission scans
were collected in the apo condition. Substrate was then added
to the cuvette (200
M verapamil (3.2
l) or 5 mM MgSO
ATP (3.2
l)) and incubated for 5 min, and three individual
emission spectra were recorded. The second substrate was then
added and recorded similarly. Multiple spectra were recorded
for each condition, sequentially, to compose a set. After addi-
tion of 10
M sodium orthovanadate, sequential emission scans
were collected over a period of 20 min until the spectral
changes were minimal between scans (the final emission scan
was used for analysis). Mixing was executed by hand. Each set of
each condition was averaged. At least two individual sets of
experiments were completed. Experiments were not combined
for averaging to prevent additional error from combining two
independent sets of data.
Data Analysis—The donor emission peak intensity, an aver-
age of the peak intensity over 3 nm, in counts/s (cps) was
divided by the acceptor emission peak intensity, an average of
the peak intensity over 3 nm in cps, to give the donor to accep-
tor ratio (D/A). Standard deviation was calculated for each con-
dition, except for the vanadate-trapped data point (only the
final scan after addition of vanadate was used). For comparison
of data from different experiments, the D/A ratio for each
experiment was normalized to the D/A ratio of the apo
Single Molecule FRET—Samples were thawed at 4 °C, applied
to a 1-ml Sephadex G-50 spin column equilibrated in 10 m
MOPS, 50 mM NaCl, 5 mM MgCl
pH 7.0, and diluted as nec
essary in buffer to obtain a maximum of one proteoliposome at
any time in the confocal volume. Samples were then incubated
for 5 min at 37 °C with the conditions as follows: “apo,” 5 m
, no substrate; “verapamil transport,” 5 mM MgSO
ATP, 200
M verapamil; “vanadate-trapped,” 5 mM MgCl
M ATP, 200
M verapamil, 237
M sodium orthovanadate;
“cyclosporin transport,” 5 m
M MgCl
,1mM ATP, 5
cyclosporin A.
The home-built microscope with expanded confocal detec-
tion volume of about 10 fl was described previously (31, 32).
Two lasers in duty cycle-optimized alternating fashion were
used to excite the donor and acceptor (33). Briefly, the donor
was excited with a blue pulsed laser (PicoTa 490, up to 80 MHz
repetition rate, Picoquant) at 488 nm with 150 microwatts. The
acceptor was excited with an orange continuous HeNe laser at
594 nm with 30 microwatts, switched by an acousto-optical
modulator. The alternating laser sequence was set for four blue
laser pulses at a 16-ns interval with a 60-ps pulse duration fol-
lowed by a single 32-ns pulse of the orange laser 16 ns after the
fourth blue pulse. Photons were detected by two avalanche
photo diodes and detected between 497 and 567 nm (HQ 532/
70, Advanced Harmonic Filter) for Alexa 488 and wavelengths
longer than 595 nm for Atto 610 (LP 595, AHF). Recording of
the photons with picosecond time resolution was achieved by
synchronized TCSPC electronics (SPC 153, Becker & Hickl) in
a computer for subsequent fluorescence lifetime analysis, fluo-
rescence correlation spectroscopy, and FRET analysis.
Single Molecule Analysis—Bursts were automatically marked
based on fluorescence intensity thresholds using the software
Burst-Analyzer by N. Zarrabi (34, 35). Briefly, photon bursts
had to exceed 20 counts per ms (cpms) in the acceptor channel
when directly excited with 594 nm, as well as 10 cpms in the
FRET donor channel and 10 cpms in the FRET acceptor chan-
nel upon 488 nm excitation. The duration of a burst was
allowed to be in the range between 20 and 1,000 ms related to
the mean diffusion time of 30 ms for the proteoliposomes.
Background corrections of 11 and 5 cpms were subtracted for
the FRET donor and acceptor intensities, respectively. FRET
states within the photon bursts were identified manually. Each
biochemical condition was compiled in a data file for further
analysis using a series of MATLAB scripts to obtain FRET effi-
ciency histograms, FRET transition density plots, and dwell
time distributions.
Site-directed Mutagenesis—To address domain movement in
Pgp using FRET spectroscopy, pairs of cysteines were placed at
equivalent positions in the N- and C-terminal nucleotide bind-
ing domains of the transporter for covalent labeling with
maleimide-linked fluorescent dyes. Site-directed mutagenesis
was carried out with a cysteine-less version of Pgp (mouse
Mdr3) that had previously been shown to exhibit near wild type
levels of drug-stimulated ATPase activity (28). For structure-
based placement of cysteine residues, we modeled the N- and
C-terminal NBDs of Pgp using the crystal structure of MgADP-
bound Sav1866 (36) as a template
(Fig. 1). The locations of the
cysteine pairs were chosen to get a sampling of the various
subdomains of the NBDs in terms of distance from the mem-
brane and across the putative NBD dimer interface. A further
criterion for the chosen locations was to be able to detect both
translation (y axis in Fig. 1A) and opening-closing (x axis)
motion of the NBDs, motions predicted from a comparison of
available crystal structures of ABC transporters adopting
inward- and outward-facing conformations.
Protein Purification, Maleimide Labeling, and Reconstitution
Based on immunoblots, all five double cysteine mutants were
expressed in P. pastoris following induction with methanol
(data not shown). Although mutant SS was rapidly degraded
We used Sav1866 (Protein Data Bank code 2hyd (36)) as a template because
a crystal structure of Pgp was not available at the time.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
upon extraction from the membrane, mutants TS, HD, RA, and
SQ were stably expressed and could be purified according to
established protocols. The typical yield was 3–4 mg of purified
Pgp per 87 g of cells. Fig. 2A shows SDS-PAGE of purified
mutant TS. Similar results were obtained for mutants SQ, HD,
and RA (data not shown). Typical values for verapamil-stimu-
lated specific ATPase activities for Cys-less (CL), TS, RA, SQ,
and HD Pgp were 1.7, 2.1, 2.0, 1.3, and 2.2
respectively, close to published values for wild type and Cys-less
Pgp expressed in P. pastoris (28).
Mutant proteins were labeled with donor and acceptor dyes
selected based on the predicted distances of the two cysteines in
the Pgp homology model of the outward facing conformation as
described under “Experimental Procedures.” Dyes that stimu-
lated Pgp ATPase activity in a manner seen for transport sub-
strates, and/or bound nonspecifically to cysteine-less Pgp (e.g.
Alexa Fluor 546), were not used for FRET experiments. Several
labeling schemes were tested, including substoichiometric
labeling with one dye followed by labeling with an excess of the
second dye or labeling with mixtures of the dyes in a predeter-
mined ratio, either before or after concentrating the protein.
Most reproducible results were obtained by labeling dilute pro-
tein in the presence of 0.1% DDM with subsequent removal of
excess dye and exchange of DDM by 0.5% CHAPS (for liposome
reconstitution by size exclusion chromatography).
Labeled Pgp mutant proteins were reconstituted into proteo-
liposomes (phosphatidylcholine/phosphatidic acid 19:1) using
gel filtration as described under “Experimental Procedures.”
Turbid fractions were analyzed by SDS-PAGE, and the pres-
ence of donor and acceptor fluorescent dyes in the proteolipo-
some-containing fractions was verified on a fluorescence scan-
ner (Fig. 2B). The average size of the proteoliposomes as
determined by negative stain transmission electron microscopy
was around 50 –150 nm (Fig. 2C). Labeling efficiency and
donor/acceptor stoichiometry was determined by UV-visible
spectroscopy. For mutants RA and TS, protein and dye concen-
trations were determined after excess dye was removed by
Superdex 200 gel filtration, before liposome reconstitution. For
mutants HD and SQ, dye concentrations were measured after
liposome reconstitution. For these samples, protein concentra-
tion could not be determined reliably due to interference of the
CHAPS used to solubilize the liposomes before UV-visible
spectroscopy. The concentrations of the samples used in these
experiments were as follows: RA, 4.1
M protein, 1.98
M Alexa
488, 5.02
M Atto 610; TS, 2.35
M protein, 0.31
M Alexa 488,
M Atto 610; HD 0.25
M Alexa 488, 0.33
M Atto 610; SQ,
M Atto 565, 0.10
M Atto 655. Column fractions contain-
ing liposome-reconstituted, double-labeled protein were
pooled and used for fluorescence spectroscopy.
FIGURE 1. Placement of cysteine mutants in the structural models of Pgp.
A, residue placement in the crystal structure of Pgp in the apo-form (Protein
Data Bank code 3g5u (11)) looking from the cytosol toward the membrane
(top) and parallel to the membrane (bottom). B, same views as in A for the Pgp
NBD homology model generated with Sav1866 (Protein Data Bank code 2hyd
(36)) as template. The transmembrane domains were not modeled due to the
low similarity between Pgp and Sav1866 in this region of the proteins. Resi-
due locations for introducing cysteines were chosen that were not highly
conserved and not part of secondary structure. Mutant pairs are S375C/
Q1020C (SQ, black), T492C/S1137C (TS, yellow), H514C/D1159C (HD, pink),
S586C/S1231C (SS, green), and R613C/A1258C (RA, cyan). Mutant SS, chosen
before the release of the Sav1866 structure, could not be purified. C, pre-
dicted C
distances between cysteines introduced by site-directed mutagen-
esis using the Pgp apo and homology model of the NBDs.
FIGURE 2. Protein purification, labeling, and reconstitution. A, Coomassie-
stained 10% polyacrylamide gel. Lane 1, molecular weight marker; lane 2, 2.6
g of purified mutant TS. Similar yields and purity were obtained for SQ, RA,
and HD mutants (data not shown). B, SDS-PAGE of labeled HD proteolipo-
some fractions from Sephadex G-50 column, imaged with a fluorescence
scanner at wavelengths for Alexa 488 (top) and Atto 610 (bottom). Lane 1,
molecular weight marker, lanes 2– 6, fractions 13–17. C, negative stain trans-
mission electron microscopic image of labeled mutant SQ reconstituted into
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
FRET Spectroscopy of Reconstituted Pgp—FRET spectros-
copy was carried out to investigate the conformational changes
that occur in response to drugs (verapamil), nucleotide, and
vanadate as described under “Experimental Procedures.” The
dyes used for each mutant, excitation wavelength, and excita-
tion and emission slit widths are summarized in Table 1. Fig. 3
shows FRET spectra obtained for bilayer reconstituted mutant
TS labeled with Alexa 488 (donor) and Atto 610 (acceptor),
normalized to the apo-donor peak for comparison. Comparing
the donor to acceptor (D/A) ratio between the spectra is an
established method of estimating relative distance changes
between two fluorophores (37–39). An increase in D/A ratio
would indicate a decrease in FRET efficiency and an increase in
the distance between the donor and acceptor molecules. A
decrease in D/A ratio would indicate a decrease in distance. As
can be seen, addition of the ATPase stimulator verapamil (Fig.
3, green trace) to liposome-bound apo-Pgp (blue trace) does not
lead to a significant decrease of the D/A fluorescence ratio,
suggesting that binding of the drug alone does not induce a
conformational change in the NBDs. A more significant
decrease of the D/A ratio (12%) is observed upon addition of
MgATP (Fig. 3, red trace), suggesting that verapamil-stimu-
lated hydrolysis of MgATP increases the population of mole-
cules with closely interacting NBDs.
In control experiments using a mixture of free dyes (see sup-
plemental material), we found that excitation of the donor dye
(Alexa 488) did not produce significant acceptor (Atto 610)
fluorescence at similar dye concentrations as present in the
protein samples. The observed acceptor fluorescence in dou-
ble-labeled Pgp must therefore be due to energy transfer from
the donor dye. Pgp molecules with one or two donor dyes only
(no acceptor) will produce donor fluorescence signal along with
the donor fluorescence from Pgp molecules having both donor
and acceptor bound. The excess donor fluorescence will result
in a larger D/A ratio than would be expected from uniformly
donor- and acceptor-labeled Pgp. An overall increase in D/A
ratio due to excess donor labeling will therefore only “dilute”
the observable change in the D/A ratio but not change its direc-
tion. Pgp molecules having two acceptor dyes bound will not
contribute to the signal in any way and will thus be invisible in
the experiment.
Vanadate Trapping—Vanadate has been shown to inhibit
drug-stimulated ATPase activity of Pgp by trapping ADP at one
catalytic site (presumably in the “nucleotide sandwich” confor-
mation) (20). Thus, vanadate trapping was carried out as
described under “Experimental Procedures” to investigate
changes in FRET efficiency upon inducing this transition state
conformation. As can be seen, addition of 10
M sodium
orthovanadate (Fig. 3, black trace) resulted in a final overall
drop of the D/A ratio of 20%. The decrease of the D/A ratio
was time-dependent with stable fluorescence emission reached
after about 15–20 min after addition of vanadate.
Similar experiments as the ones described above for mutant
TS were carried out for mutants HD, RA, and SQ, and the sub-
strate-dependent changes in D/A fluorescence for all four
mutants tested are summarized in Fig. 4. Spectra for SQ, HD,
and RA displayed an overall similar appearance with varying
D/A fluorescence intensity ratios and changes in the ratio upon
addition of drugs and/or nucleotides. As can be seen from the
barographs in Fig. 4, mutant HD shows the overall largest
decrease in D/A ratio upon vanadate inhibition ( 24%), closely
followed by TS (20%) and RA (19%). The smallest change
(6%) is seen for mutant SQ, consistent with the prediction
form the comparison of apo- and ADP-bound structures shown
in Fig. 1.
To test if the magnitude of the overall FRET efficiency
change is a function of the order of substrate addition, we per-
formed experiments in which MgATP was added first followed
by verapamil. The result is shown in Fig. 4, C and D, for mutant
RA. As can be seen, the D/A ratio of RA decreased by 17% when
Parameters for ensemble fluorescence spectroscopy data collection
Mutant Donor Acceptor
slit Excitation wavelength
nm nm nm
SQ Atto 565 Atto 655 3 3 563
TS Alexa 488 Atto 610 2.5 2.5 493
RA Alexa 488 Atto 610 2 2 493
HD Alexa 488 Atto 610 2–2.5 2–2.5 493
FIGURE 3. Fluorescence spectroscopy of reconstituted mutant TS. Shown
are fluorescence spectra from one experiment with TS proteoliposomes,
labeled with Alexa 488 and Atto 610. All spectra are taken at 37 °C. Excitation
wavelength is 493 nm. Three sequential spectra are recorded for each condi-
tion. After the apo-spectra were collected, 200
M verapamil was added, and
the cuvette was incubated for at least 5 min; three sequential fluorescence
wavelength scans from 503 to 730 nm were then recorded. The same
sequence was repeated after addition of 1 mM MgATP. After that, 10
sodium orthovanadate was added, and wavelength scans were recorded
until emission was stabilized. Each trace is representative of at least three
emission scans averaged, except the condition for vanadate where the last
scan is shown. For ease of viewing, the spectra have been normalized to the
donor peak recorded for the apo condition.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
MgATP was added before verapamil compared with a 19%
decrease when verapamil was added first (followed by vanadate
trapping under both conditions).
Although the changes in D/A ratios for the different mutants
are not necessarily comparable on a quantitative level due to the
aforementioned differences in labeling ratios and efficiencies,
the observation that the overall D/A ratio changes are similar
and independent of the order of addition of drug and MgATP
suggests that the resulting steady state is likely the same.
S and Verapamil—Previously, it had been shown
that the magnesium-bound form of the nonhydrolyzable ATP
analog, ATP
S, binds Pgp with higher affinity compared with
MgATP (16, 18, 20). We therefore investigated whether bind-
ing of MgATP
S would lead to a similar decrease in D/A ratio
as MgATP. Observation of a similar change in FRET efficiency
upon MgATP
S binding would indicate that the nonhydrolyz-
able ATP analog was able to induce NBD sandwich formation.
As is seen in Fig. 5, the addition of 200
M verapamil followed by
and1mM ATP
S yielded only minor changes in
the D/A ratio, independent of whether verapamil or MgATP
was added first. Generally, there is even less change in the over-
all D/A ratio when MgATP
S is present before addition of vera-
pamil. Using HD as an example, the normalized D/A ratio when
adding verapamil first decreases by 2% followed by essentially
no change with the addition of MgATP
S. However, when
S is added to HD followed by verapamil, the D/A
increases by 1% followed by a 3% decrease. Details of all mutants
in the presence of MgATP
S and verapamil are shown in Fig. 5.
Overall, the MgATP
S binding experiments seem to suggest
that binding of the nonhydrolyzable ATP analog does not pro-
duce a significant population of Pgp molecules with closed
NBDs. This result is surprising considering the higher affinity
of MgATP
S compared with MgATP (16, 18, 20). It is possible
that MgATP
S was not able to induce the occluded state under
the conditions of the FRET experiment, but it cannot be ruled
out that MgATP
S binds differently compared with MgATP.
Single-molecule FRET—The ensemble FRET experiments
described above suggest that in presence of verapamil and
MgATP, the two NBDs (as observed by way of the attached dye
molecules) get on average closer to each other (and even more
so in presence of vanadate). However, due to the nature of the
ensemble experiments, it is not possible to distinguish between
a situation where all the molecules have semi-closed NBDs or
where subpopulations of molecules exist at any given time that
either have tightly associated, semi-open, or completely open
NBDs, a situation as could be expected for active turnover. To
overcome the limitation of ensemble FRET, we therefore
decided to employ single-molecule FRET spectroscopy to fur-
ther analyze the conformational changes associated with drug-
stimulated ATP hydrolysis. Using mutant TS stochastically
labeled with Alexa 488 and Atto 610, fluorescence was recorded
from single Pgp molecules reconstituted into liposomes using a
FIGURE 4. D/A ratios of vanadate-trapped Pgp. A, comparison of the D/A fluorescence emission ratios in response to verapamil, MgATP, and vanadate.
Verapamil (200
M) and MgATP (1 mM) traces are averages of a minimum of three scans, whereas the D/A ratio for vanadate (10
M) is from the final scan of the
experiment. B, normalized D/A ratios from A. The D/A ratio of each mutant in the apo condition was used to normalize the D/A ratio of the substrate conditions.
C, using mutant RA, addition of 1 mM MgATP was followed by 200
M verapamil and then 10
M vanadate to test whether the order of adding drug or
nucleotide had an influence on the final D/A ratio. D, normalized D/A ratio of RA as in B.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
confocal single-molecule FRET (smFRET) setup as described
(33). In the smFRET setup, proteoliposomes diffuse freely
through a confocal excitation volume of about 10 fl (calculated
by fitting the autocorrelation function of rhodamine 110 in
buffer for FCS analysis). Two pulsed lasers were focused on the
confocal volume in a droplet of proteoliposomes containing
buffer to either excite Alexa 488 for the FRET measurement or
Atto 610 to verify the existence of both fluorophores at one
protein, respectively, in an optimized interleaved excitation
scheme (40, 41). Photon counts were binned to 1-ms intervals.
The resulting time trajectories exhibited bursts of photons indi-
cating that a double-labeled Pgp had traversed the confocal
volume. The lasers generated also some background of scat-
tered light. Background corrections between 11 cpms for Alexa
488 and 5 cpms for Atto 610 were subtracted. The proximity
factor P was calculated per data point in the time trace by Equa-
tion 1,
(Eq. 1)
with I
and I
intensity of the acceptor and donor, respectively.
To determine FRET efficiencies, the
correction factor
accounting for detection efficiencies of the instruments and
quantum yields of the donor and acceptor was calculated. Using
quantum yields of 0.6 and 0.7 for Alexa 488 and Atto 610,
respectively, the
correction factor was calculated to be 0.79
(quantum yields as given by the supplier of the dyes).
FRET efficiencies were calculated by Equation 2,
(Eq. 2)
A total of four conditions was tested in the smFRET spectros-
copy setup as follows: (i) apo (5 m
M MgCl
in the buffer but no
nucleotide or drug added); (ii) verapamil transport (1 mM
MgATP 200
M verapamil); (iii) vanadate-trapped (1 mM
MgATP 200
M verapamil 237
M orthovanadate); and
(iv) cyclosporin transport (1 m
M cyclosporin
A). Here, 1 m
M MgATP refers to 5 mM MgCl
(already present
in the buffer) plus 1 mM ATP. Although considered a transport
substrate, cyclosporin A has been shown to activate Pgp
ATPase barely above basal levels (27). We therefore included
the cyclosporin transport condition with the expectation that
the steps of the catalytic cycle that are slowed down compared
with verapamil, for example, would be noticeable in the single
molecule measurements. For each condition, several time
traces of 10 min each were collected. Photon bursts were auto-
matically identified using threshold criteria in the software
“Burst-Analyzer” (34, 35), followed by manually screening for
artifacts and verifying the presence of both donor and acceptor
dyes. Individual FRET steps, i.e. distinguishable different levels
of proximity factors or FRET efficiencies, were marked within
each burst manually. Photon bursts with marked steps were
then analyzed with MATLAB scripts to generate FRET effi-
FIGURE 5. Substrate-induced changes in donor to acceptor ratio. A, comparison of the ratio of donor to acceptor emissions in response to addition of 200
M verapamil, followed by 1 mM MgATP
S(5mM MgSO
,1mM ATP
S). B, D/A ratio for each mutant was normalized to the apo condition. B is the normalized
D/A ratio for A. C, comparison of the donor to acceptor emission in response to MgATP
S followed by verapamil (concentrations as in A). D, normalized D/A
ratios as in B for C.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
ciency histograms, FRET 1–2 transition density plots and dwell
time distributions.
Photon bursts of single Pgp for different biochemical condi-
tions are shown in Fig. 6. In the lower panels of Fig. 6, the
fluorescence intensities of FRET donor (green trace) and accep-
tor (red trace) with 1-ms binning are depicted. In the upper
trace of Fig. 6, the corresponding fluorophore-fluorophore dis-
tances are plotted based on photon probabilities using the soft-
ware “FRET-TRACE” (42). MgATP-driven transport of vera-
pamil (Fig. 6, A–C) resulted in relative intensity changes of
FRET donor and acceptor within a photon burst. To ensure that
both fluorophores were attached, a third fluorescence intensity
trace was recorded using direct excitation of the acceptor with
594 nm (see supplemental Fig. S1). To illustrate the process of
single-molecule FRET data analysis, we will discuss Fig. 6A
briefly. At about 25 ms, both fluorescence intensities increased
above the thresholds indicating that the FRET-labeled Pgp had
entered the confocal detection volume. This photon burst
ended at about 190 ms because the transporter was leaving the
confocal volume, and the intensities drop below the threshold.
Within this photon burst at about 35 ms, the FRET distance
increased from 6.0 to 6.5 nm but switched back after about 20
ms. At about 90 ms, the distance increased again to 6.5 nm,
followed by a decrease to 5.5 nm for 20 ms. At 125 ms, the
fluorophore distance increased to 6.0 nm and switched back to
5.5 nm after 10 ms. Around 150 ms, the distance decreased even
further. Manual assignment of FRET levels in this burst based
on the proximity factor trace is shown in the supplemental Fig.
S1. The other two photon bursts (Fig. 6, B and C) showed also
strong fluctuations of the FRET distances during MgATP-
driven transport conditions.
In contrast, bursts for the apo condition (Fig. 6D) exhibited a
nearly constant large distance between 6.5 and 7 nm through-
out the observation time. In the presence of vanadate (vana-
date-trapped condition), the FRET distances within a photon
burst fluctuated mostly around lower distances between 5.0
and 5.5 nm, as seen in Fig. 6 E. In the presence of MgATP and
cyclosporin A (cyclosporin transport), similar FRET distances
changes were observed as in the case of MgATP-driven vera-
pamil transport. The photon burst in Fig. 6F illustrates the large
changes of the fluorophore distances between 4.5 and 6 nm
with slower transition for the shorter distances.
Of the 80 bursts analyzed for the apo condition, 55% showed
a constant FRET efficiency and 37% three and more distin-
guishable FRET levels. In contrast, for verapamil transport, 74%
of the 216 bursts analyzed exhibited three and more FRET lev-
els with only 7% of the bursts with constant FRET distances. For
the vanadate-trapped, 16% of 249 burst remained in a constant
FRET level, with 61% fluctuating between three and more lev-
els. For cyclosporin transport, 33% from 224 bursts showed one
FIGURE 6. Example bursts from smFRET spectroscopy for mutant TS. A–C are for condition verapamil transport (1 mM ATP, 5 mM MgCl
, and 200
verapamil). D–F are for conditions apo (no substrate, 5 mM MgCl
in the buffer), vanadate-trapped (1 mM ATP, 5 mM MgCl
, 200
M verapamil, and 237
vanadate), and cyclosporin transport (1 mM ATP, 5 mM MgCl
, and 5
M cyclosporin A), respectively. For details, see text.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
constant FRET level and 38% exhibited three and more FRET
To analyze the observed distance distributions, the assigned
FRET efficiency levels were plotted in histograms. Fig. 7 shows
FRET efficiency histograms (with corresponding FRET dis-
tances as the upper axis in the panel) for the four conditions
tested for mutant TS. As can be seen in the apo condition, FRET
efficiencies varied from 0.18 to 0.72 centered near 0.5. In the
verapamil transport condition, the FRET efficiencies ranged
from 0.22 to 0.88 with the center near 0.6. For vanadate-
trapped, FRET efficiencies ranged from 0.18 to 0.88 with the
center shifting to near 0.7. Under the cyclosporin transport
condition, sampled FRET efficiencies ranged from 0.22 to 0.82
centered near 0.5.
A transition from one FRET efficiency to another FRET effi-
ciency within a burst, or a step to step, gives information
regarding the conformational changes sampled by the fluores-
cence burst. This is displayed by plotting FRET distance 1 in a
burst against the subsequent FRET distance 2 in the same burst
(Fig. 8). The FRET 1–2 transition plot for condition apo
revealed only few transitions (out of 119 FRET levels), mostly in
the 5.5- to 6.5-nm range, because most photon bursts did not
show the required distance changes between at least four FRET
levels (see above). A different picture was obtained for vera-
pamil transport, with two dominating transitions between 5.0
and 5.5 nm and back, as well as between 6.0 and 5.5 nm and
back. However, also transitions at larger distances as well as
shorter distances were observed within 809 FRET levels. Under
conditions of vanadate trapping, only the transitions between
5.0 and 5.5 nm were densely populated (based on 621 FRET
levels), consistent with the ensemble experiments that showed
the highest mean FRET efficiency for this condition. Con-
FIGURE 7. Single molecule histograms for mutant TS. The FRET efficiencies of single molecule trajectories of mutant TS are shown. The conditions are as
follows. A, apo (5 mM MgCl
in the buffer); B, verapamil transport (1 mM ATP, 5 mM MgCl
, and 200
M verapamil); C, vanadate-trapped (1 mM ATP, 5 mM MgSO
M verapamil, and 237
M vanadate); and D, cyclosporin transport (1 mM ATP, 5 mM MgCl
, and 5
M cyclosporin A).
FIGURE 8. FRET transition plots for mutant TS. After marking bursts and
steps within the bursts, FRET transition plots were created by calculating the
first FRET level of the burst (x axis) and plotting that distance against the
second FRET level (y axis). If there were more than two FRET levels in each
burst, then level two was plotted against level three, etc. The panels follow
the same layout as Fig. 7.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
versely, the condition cyclosporin transport appeared to shift
the population to several transitions, one at higher FRET effi-
ciency centered at 4.8 nm (that is present but barely detected
under condition verapamil transport), one around intermedi-
ate efficiency (5.0 to 5.5 nm), and one at a lower efficiency FRET
centered on 6 nm (367 FRET levels in total). Cyclosporin A is
one of the larger transport substrates that barely activates ATP
hydrolysis above basal levels in biochemical ensemble measure-
ments, and it is possible that conformations that are too short
lived to be observed in significant numbers under condition
verapamil transport were stabilized with cyclosporin A bound.
Furthermore, information about how long the molecule
spends in each level or conformational state can be gained by
examining the dwell times. This is shown in Fig. 9. The longer
the dwell time, the longer the molecule stays in one conforma-
tion with no change in FRET efficiency. Therefore, photon
bursts had to show at least three FRET levels, because only the
intermediate dwell could be used for the dwell time distribu-
tion. The length of the first and the last FRET level in a burst
remains unknown because they were limited by entering or
leaving the confocal detection volume. In the apo condition, the
average dwell time was 36 ms using a mono-exponential
decay fitting. Under condition verapamil transport, the dwell
time was reduced significantly to around 7.4 ms, although
under condition vanadate-trapped, the average life time was
prolonged and close to 35 ms. Finally, under conditions of
cyclosporin transport, the dwell time was close to 10 ms similar
to the transport condition in presence of verapamil.
Taken together, the smFRET data show that condition apo is
characterized by low FRET efficiencies and long dwell times
with few observed transitions in the FRET 1–2 plot. At least
three major conformations are observed under conditions of
verapamil transport with a much shorter dwell time of 7 to 8 ms.
Vanadate trapping shifts the population to transitions at higher
FRET efficiencies with slower transitions, although cyclosporin
transport populates high, intermediate, and low FRET efficien-
cies with shorter dwell times.
Previously, we have studied the structure- and ligand-in-
duced structural changes of the lipid bilayer-bound Pgp by
transmission electron microscopy of two-dimensional crystals
(22, 23). The data showed that although apo-Pgp displayed a
large central cavity in projection, addition of nucleotide and/or
drug resulted in projections in which the two halves of the
transporter appeared to interact more closely, thereby closing
the central cavity. However, because of the limited resolution
of the projection data and the static nature of the protein in the
two-dimensional crystals, questions remained as to the order
and magnitude of the structural changes induced by addition of
drugs and nucleotides. Consistent with the EM data, a large
central cavity is also seen in the recent crystal structure of apo-
and drug-bound Pgp (11); however, a high resolution structure
of nucleotide-bound Pgp that would provide insight into how
the NBDs are interacting upon nucleotide binding is currently
not available. However, there is now ample evidence that the
FIGURE 9. Dwell time histograms for mutant TS. After marking bursts and steps within the bursts, the first and last step of each burst was discarded since the
amount of time the protein had been in that level before entry into the confocal volume and remains in the final level after exiting the confocal volume is
unknown. The duration that Pgp spent in each “middle” level was then recorded and plotted to give the dwell time. The panels follow the same layout as Figs.
7 and 8.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
catalytic cycle of Pgp includes formation of a so-called ATP
sandwich conformation characterized by dimerization of the
N- and C-terminal NBDs with nucleotide sandwiched by the
P-loop of one and the ABC signature motif of another NBD.
Such a nucleotide sandwich configuration can be seen in a vari-
ety of x-ray crystal structures of either isolated NBDs (43, 44) or
intact bacterial ABC transporters (36, 45, 46), and these struc-
tures have provided a detailed picture of the interface in the
symmetric NBD-NBD dimer.
What is currently not known with certainty is how much the
NBDs have to come apart (dissociate) during the catalytic cycle
to allow release of inorganic phosphate (P
) and exchange of
ADP with ATP as well as opening of the cytoplasmic drug bind-
ing pocket. Considering the recent crystal structure of apo-Pgp
that shows the NBDs separated by up to 30 Å, the authors of
that study suggested that the NBDs will likely have to come
apart during catalysis to allow for binding of especially bulky
drug molecules (11).
From studies with mutant Pgp or vanadate-trapping experi-
ments, it is well established that the two catalytic sites in Pgp act
in a cooperative fashion in that inactivating one catalytic site by
mutation or vanadate binding abolishes drug-stimulated
ATPase activity (15, 47). Furthermore, it has been shown that
nucleotide can be trapped in catalytic sites given that hydrolysis
cannot proceed because essential residues involved in catalysis
have been mutated (47), or by using a nonhydrolyzable nucleo-
tide analog such as ATP
S (16, 17, 20), or by replacing phos-
phate with e.g. orthovanadate or beryllium fluoride to stabilize
the ADP-bound post- or prehydrolysis state (20, 48). Taken
together, the mutagenesis and nucleotide-trapping experi-
ments suggest that drug-stimulated ATP hydrolysis in Pgp
occurs in an alternating fashion with the NBDs staying associ-
ated for at least two hydrolysis events (18).
In this study, we have designed a system in which structural
changes involving the NBDs are detected in real time using
double cysteine mutants labeled with donor and acceptor fluo-
rescent dyes for FRET spectroscopy using bulk samples as well
as single molecules. FRET spectroscopy is a powerful tool for
studying relative distance changes in biological macromole-
cules, and the technique has previously been applied to esti-
mate distances and distance changes in Pgp using fluorescently
labeled lipid molecules and nucleotide analogs (49–51). We
used a combination of secondary structure prediction and
homology modeling to identify pairs of residues in the two
NBDs that were predicted to undergo large distance changes
when going from an outward-facing (such as seen in the ADP-
bound structure of Sav1866 (36)) to the inward-facing confor-
mation (such as seen in the recent crystal structure of mouse
Pgp (11)). Site-directed mutagenesis was conducted using a
Cys-less background of mouse Pgp (Mdr3), and four of the five
mutants that were generated could be expressed in P. pastoris
and purified to homogeneity (double mutant S586C/S1231C
was expressed but could not be purified due to rapid degrada-
tion). For the four mutant proteins that could be purified, the
introduced cysteines did not compromise the drug-stimulated
ATPase activity compared with the Cys-less Pgp background
(see “Results”). Mutants, SQ, TS, HD, and RA were subse-
quently used in FRET spectroscopy.
In the first part, we analyzed FRET efficiency changes in bulk
samples as a function of nucleotide and drug binding. Gener-
ally, the overall magnitude of the FRET signal depends on the
distance of donor and acceptor dyes but also a number of other
factors, including how close the actual distance is to the opti-
mum distance for the dye pair used (R
) as well as the labeling
efficiencies, fluorescence lifetimes, and quantum yields. The
measured FRET efficiency of bulk samples will also depend on
the amplitudes of the various catalytic intermediates under
turnover conditions: in other words, what fraction of one cata-
lytic cycle the NBDs of Pgp spend in close association (in the
ATP sandwich), partly dissociated association (during P
release), or fully dissociated association (as in the crystal struc-
ture of apo-Pgp). For example, if the NBD dimer (or ATP sand-
wich) is only a transient intermediate (short lived), its ampli-
tude during steady state ATP hydrolysis will be small. If,
however, the ATP hydrolysis reaction is the rate-limiting step,
the ATP sandwich will be the predominant species.
Starting with lipid bilayer-reconstituted apo-Pgp, we showed
that addition of drug alone resulted in only a small (2–3%)
decrease of the D/A ratio, consistent with the recent crystal
structures of Pgp, which showed the same separation of the
NBDs in the apo- and drug-bound models (11). A larger
increase in FRET efficiency, however, was seen upon addition of
MgATP, consistent with a decrease of the average distance sep-
arating the donor and acceptor dyes as would be expected for
NBD dimer formation. Although the absolute changes in D/A
ratio for the different mutants cannot be compared in a
straightforward manner because of the above-mentioned vari-
ables, the observation that mutant SQ produced the smallest
increase in FRET efficiency is consistent with the predicted dis-
tance change for this residue pair when going from inward- to
outward-facing conformations as shown in Fig. 1. It should also
be noted that irrespective of whether verapamil or MgATP was
added first, the final D/A ratio was approximately the same (as
shown for mutant RA in Fig. 4), suggesting that binding of drug
and nucleotides are independent and additive events.
It is known that inorganic vanadate traps ADP in one cata-
lytic site with concomitant loss of ATPase activity (20). Further-
more, it has been postulated that the vanadate-inhibited state
requires NBD dimerization so that one catalytic site can be
closed (52). For all four mutants, we observed a further decrease
of the D/A ratio upon addition of orthovanadate (most pro-
nounced for HD followed by TS and RA). This increase in FRET
efficiency upon vanadate inhibition could be either due to a
further decrease of the NBD-NBD distance (with approxi-
mately constant amplitude of the NBD dimer intermediate) or
an increase of the population of Pgp molecules with closely
associated NBDs (compared with the state of drug-stimulated
ATP hydrolysis before vanadate inhibition). Considering the
smFRET analysis discussed below, it appears that the latter pos-
sibility is the more likely explanation.
Previous reports using MgATP
S showed that the nonhy-
drolyzable ATP analog binds Pgp with significantly higher
affinity compared with MgATP (16, 18, 20). Furthermore, it has
recently been shown that MgATP
S can be trapped in one cat-
alytic site, and much like with vanadate trapping, it has been
assumed that the resulting “nucleotide occluded state” requires
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
closing of one of the two catalytic sites, although the other site is
open and can exchange nucleotide freely (16, 18). It was there-
fore surprising to see that addition of MgATP
S to doubly
labeled Pgp did not result in a significant increase in FRET
efficiency beyond the small increase seen with verapamil, irre-
spective of which mutant was used and in which order nucleo-
tide and drug were added. Interestingly, when adding
S followed by MgATP (or vice versa), the same FRET
efficiency change was observed as with MgATP alone (data not
shown). There are conflicting reports whether ATP
S is turned
over by Pgp and whether at least one turnover is required for
reaching the occluded state (16, 18). Given that at least one
turnover is required, it is possible that under the conditions of
the FRET spectroscopy experiment no turnover and therefore
no ATP
S occlusion occurred. However, because there is cur-
rently no high resolution structure showing that the
S-occluded state resembles the ATP sandwich, it can-
not be ruled out that the nonhydrolyzable ATP analog is bound
in a configuration that is different compared with the e.g. ADP-
-trapped state. This finding would be consistent with EM
projections that showed ADP-V
-trapped Pgp without a central
cavity (outward facing) in contrast to MgATP
S-bound Pgp
that had a central cavity between the two halves similar to the
apo-form of the transporter (23).
In summary, we have shown that FRET spectroscopy can be
applied to obtain information regarding the motions the NBDs
are undergoing upon binding of substrates starting from the
apo-form of the transporter. However, in the so-far described
ensemble FRET, catalytic turnover of the Pgp molecules is not
synchronized, which means that, as mentioned above, an aver-
age distance of the dye molecules is measured, and that the
magnitude of the FRET change will not only depend on the
distance between the dye molecules but also on lifetimes and
amplitudes of the reaction intermediates.
To extend the information obtained from analyzing bulk
samples by ensemble FRET, we have monitored real time con-
formational changes in one mutant, TS, by single-molecule
FRET spectroscopy. Mutant TS was diluted and reconstituted
into lipid vesicles at a high lipid/protein ratio to ensure that
each liposome contained on average less than one double-la-
beled transporter. Four conditions were tested by single-mole-
cule FRET as follows: (i) apo (no nucleotide or drug added); (ii)
verapamil transport (MgATP plus verapamil); (iii) vanadate-
trapped (MgATP plus verapamil plus orthovanadate); and (iv)
cyclosporin transport (MgATP plus cyclosporin A). For each
condition, a similar number of fluorescence photon bursts
was collected. However, only relatively few transitions for the
apo condition could be included in the statistics because of the
large fraction of Pgp transporters that remained in a stable non-
fluctuating conformation according to the FRET signal under
this condition. The observed low FRET efficiencies are likely
due to the large distance of the dye molecules (bound to the
NBDs) in the apo condition, consistent with the ensemble
FRET experiments where the lowest mean FRET efficiency was
observed for apo-Pgp. Despite the differences in the number of
observations, all four conditions show a relatively broad distri-
bution in the FRET efficiency histograms, indicating that Pgp is
a highly dynamic protein that is constantly sampling a range of
NBD conformations. As expected from the ensemble measure-
ments, the average FRET efficiency increased in conditions
verapamil transport and vanadate-trapped. Cyclosporin A,
which is considered a transport substrate but activates ATPase
to a much lesser degree than verapamil, showed an intermedi-
ate average of FRET efficiencies centered on 0.5. When looking
at the single-molecule FRET histograms for all four conditions,
it is obvious that Pgp can adopt more than two major confor-
mations (compared with the histograms of a three FRET-level
system like F
-ATP synthase (53, 54)).
Detailed information about the number of distinguishable
conformations is obtained from the two-dimensional FRET
transition density plot (55). The FRET 1–2 transition plot for
condition apo revealed only a few transitions, mostly in the
5.5–6.5-nm range. A different picture was obtained for vera-
pamil transport, with two dominating transitions between 5.0
and 5.5 and 5.5 and 6.0 nm. A large number of forward and
backward fluctuations with small distance changes in the range
of 0.5 nm was observed. As the precision of FRET-based dis-
tance calculation is in a similar range given the number of
detected photons per FRET level, we had to exclude possible
fluorescence anisotropy artifacts. The single molecule anisot-
ropy values measured for FRET donor and acceptor bound to
Pgp were in the range of r 0.1 (see supplemental Fig. S2).
Therefore, an orientational fluorophore dipole moment artifact
seems to be unlikely to cause the small FRET distance changes
(42). The turnover number for ATP hydrolysis by Pgp under the
conditions of the smFRET experiments is about one/s, which
means that the average fluorescence burst (100–250 ms) will
likely capture only substeps of one catalytic cycle. It is therefore
possible that the two transitions centered on 5.2 and 5.8 nm
represent distinct conformational changes and that, due to the
slow turnover, only very few full transport cycles were observed.
Under conditions of vanadate trapping, the transition cen-
tered on 5.2 nm is most populated, again consistent with the
ensemble experiments that showed the highest mean FRET
efficiency for this condition. Considering that the nucleotide-
binding site not occupied by ADP-V
is still able to exchange
nucleotide freely (51, 55), it is possible that the small transitions
between 5 and 5.5 nm under this condition are due to motions
in the helical subdomains of the NBDs (which contain the cys-
teines in mutant TS), motions that may be part of opening and
closing of the unoccupied catalytic site.
The condition cyclosporin transport appears to shift the pop-
ulations to several states, one with high FRET efficiency cen-
tered on 4.8 nm (that is present but barely populated under
condition verapamil transport), one with intermediate effi-
ciency (5.4 nm), and a low efficiency FRET state centered on 6
nm. As mentioned under “Results,” cyclosporin A is a relatively
large transport substrate that activates ATP hydrolysis only a
little above basal levels, and it is possible that conformations
that are too short lived to be observed in significant numbers
under condition verapamil transport are stabilized or pro-
longed with cyclosporin bound and could be identified as sep-
arate FRET levels.
The dwell time histograms in Fig. 9 show a long dwell time or
average lifetime of each step to be 36 ms for the apo condition
based on only a few detectable FRET level transitions. Vera-
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
pamil transport has the shortest lifetime of all four conditions of
7.4 ms, consistent with rapid changes that would occur during
ATP binding, hydrolysis, and drug translocation in a transport-
stimulating condition. However, all FRET transitions were
added into the dwell time histograms without sorting according
to the respective FRET efficiencies. To correlate dwell times
with sequential biochemical states of catalysis, future analysis
will require longer time trajectories of the conformational
changes as well as models to assign FRET levels to biochemical
states, for example using Hidden Markov Models (35, 56). Two
conditions known biochemically to slow down ATP hydrolysis
compared with verapamil, vanadate-trapped and cyclosporin
transport, have longer dwell times, meaning that the transitions
between two different FRET efficiencies are slower under these
conditions compared with verapamil transport. In an ATPase
assay, vanadate binding reduces the activity to around 3% of
drug stimulated activity. Although the dwell time for the vana-
date-trapped enzyme is similar to the apo condition, more fre-
quent FRET transitions occurred with vanadate, possibly due to
rapid closing and opening of the unoccupied catalytic site (see
above). Cyclosporin A exhibited a relatively short dwell time
averaged over all FRET transitions, but these transitions were
less frequent than for the verapamil transport condition.
Taken together, we have shown here that Pgp can sample a
range of conformations that are populated differentially as a
function of substrate conditions. Going from the apo confor-
mation of Pgp as seen in the crystal structure to an outward-
facing conformation as seen in ADP-bound Sav1866, for exam-
ple, we were expecting at least one transition for mutant TS
moving between 7 and 5 nm. However, the lack of such large
conformational transitions (as evidenced by the absence of off-
diagonal transitions in Fig. 8) lead us to believe that the catalytic
cycle of Pgp instead proceeds via a series of relatively small
steps. The small transitions we see support a model where the
NBDs are never completely dissociated from one another dur-
ing steady state catalysis (Fig. 10). However, the model pre-
sented in Fig. 10 is different from alternating site models pro-
posed previously (16, 18, 57) in several aspects as follows.
Recent cross-linking studies from our laboratory and others
(27, 58) showed that both the NBDs and the TMDs can be
cross-linked without significant loss of drug-stimulated
ATPase activity, suggesting that the wide separation of the
NBDs and TMDs (at the cytoplasmic side) as seen in the crystal
structure (11) is a conformation that may not be adopted during
every cycle of steady state turnover. We therefore postulate that
drug and nucleotide bindings result in a state with more closely
associated TMDs and pseudo-symmetric ATP sandwich, in
which one of the catalytic sites is poised for hydrolysis (state A
in Fig. 10). This state is still inward-facing with a high affinity
drug-binding site, but because the TMDs have a different con-
formation in this state compared with the apo-form, it may
explain why photo-affinity labeling points to a reduced affinity
FIGURE 10. Working model of Pgp catalytic mechanism. The model is based on the alternating site mechanism (2, 16, 18, 57), and the data obtained are
presented as FRET spectroscopy study. The cycle starts with addition of drug (verapamil) and MgATP to apo-Pgp (left) resulting in a transporter with two
molecules of ATP sandwiched at the NBD-NBD interface (state A). If catalysis could not continue because of e.g. mutation in one or both catalytic sites, this state
would represent the symmetric nucleotide sandwich seen in crystal structures. We speculate that binding of two MgATPs brings the two TMDs close together
with a high affinity drug-binding site that may or may not be exposed to the cytoplasm or inner leaflet. A more closed TMD domain (compared with what is seen
in the x-ray structure of apo-Pgp (11)) is still compatible with drug-stimulated ATP hydrolysis as shown in a recent cross-linking study (58). The presence of drug
in the (possibly sequestered) TMD-binding site acts to stabilize the subsequent transition state in one catalytic site (here site 1) with tightly bound ATP in site
2. Collapse of the transition state in site 1 (leading to ADP-P
in state B) converts the high affinity drug-binding site to low affinity by means of helix rotation in
the TMDs (21), ultimately resulting in release of drug to the extracellular milieu. Helix rotation may be assisted by the free energy drop of P
release leading to
state C. ADP bound in site 1 is released (one of the rate-limiting steps) and replaced by ATP. Subsequent closing of site 1 allows the tightly bound ATP in site 2
to enter the transitions state to form state A. Once state A is formed, the cycle repeats except that ATP hydrolysis now occurs in site 2. In presence of V
, stage
C or C will result in the vanadate-trapped state. For further details, see text.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
for drugs under this condition (59). In the next step, hydrolysis
of ATP in site 1 followed by collapse of the transition state (and
release) results in helix rotation in the TMDs (21), thus con
verting the high affinity drug-binding site to low affinity with
subsequent dissociation of the drug (Fig. 10, state B). Although
site 1 then opens just enough to allow release of ADP and bind-
ing of ATP, site 2 remains closed with occluded ATP (Fig. 10,
state C). If vanadate is present, state C is converted to state D
with ADP-V
trapped in site 1. Note that as a result of vanadate
trapping, the formerly occluded site 2 needs to convert back to
a site that can exchange nucleotide freely (51, 55) with concom-
itant formation of the high affinity drug-binding site (60). Dur-
ing steady state turnover, ADP release and binding of ATP and
drug converts state C to state A, which is a mirror conforma-
tion of state A and another half-cycle starts leading back to
state A.
The model presented here is supported by the FRET spec-
troscopy data in several ways. From the bulk measurements, the
vanadate-trapped state displays the smallest D/A ratio, consis-
tent with close association of the two NBDs. In the presence of
an excess of MgATP, the nucleotide exchanging site 2 in state D
is likely predominantly in a closed conformation with loosely
bound ATP. Structurally, this state may resemble the symmet-
ric nucleotide sandwich seen in crystal structures most closely.
An intermediate D/A ratio was seen for all mutants under con-
ditions of drug-stimulated ATP hydrolysis, consistent with the
alternating opening and closing of catalytic sites. Note that
mutant pairs TS and HD are located within the helical sub-
domain of the NBDs. The helical subdomain has been shown to
undergo significant motions relative to the RecA-like domain in
response to nucleotide binding in the maltose transporter from
Escherichia coli (61). The ensemble FRET data also support the
notion that drug binding alone does not result in a significant
structural rearrangement that would affect the spatial arrange-
ment of the two NBDs. The single-molecule FRET data support
and extend the information gained from the ensemble mea-
surements. For mutant TS, two major transitions are seen in
distance 1–2 plots for condition verapamil transport, one
between 5.5 and 6 nm and one between 5.5 and 5 nm, whereas
upon vanadate trapping, the predominant transition is between
5 and 5.5 nm. We speculate the following: (i) the small transi-
tion in the presence of vanadate corresponds to small fluctua-
tions of the helical domain of either NBD, similar to what has
been observed in molecular dynamics simulations of AMP-
PNP-bound Sav1866 in which one nucleotide was removed
from one catalytic site before performing the calculations (62);
and (ii) the larger transitions under conditions of verapamil
transport correspond to the collapse of the transition state in
either NBD with concomitant opening of the catalytic site for
nucleotide exchange.
However, it should be noted that the FRET data would also be
consistent with a model in which hydrolysis of the first ATP
results in drug transport, although the free energy gained from
hydrolyzing the second ATP is utilized for an efficient resetting
of the transporter to the ground state (2, 63– 65). Such a model
would still require cooperativity of the two NBDs (for which
there is ample experimental evidence) but at the same time
would allow complete dissociation of the NBDs between some
or all catalytic cycles.
Here, we have shown that ensemble and single-molecule
FRET spectroscopy can be applied to study the catalytic mech-
anism of mammalian P-glycoprotein. We showed that drug and
nucleotide binding makes the NBDs come together, a state fur-
ther promoted by trapping of ADP and vanadate in catalytic
sites. With the vast substrate classes of Pgp, it would be of great
interest to find a molecule that precludes the NBDs from close
association, thereby inhibiting the transporter’s interference
with drug treatment. We have described a system that will allow
future investigations of the dynamics of Pgp during its trans-
port cycle. Further studies, especially in real time, are needed to
visualize all the conformations that Pgp is undergoing during
drug-stimulated ATP hydrolysis (66). Such studies will require
D/A pairs not only located within the NBDs but also in the
TMDs and intracellular loops or “coupling helices” to further
the understanding of the communication between the TMDs,
NBDs, ATP hydrolysis, and drug extrusion. Ultimately, these
studies will require trapping Pgp in the laser focus for pro-
longed observations. Such longer observation times have
recently been achieved with the “Anti-Brownian Electroki-
netic” (ABEL) trap method (67, 68), which is in developmental
stages (69) in our laboratories for future Pgp dynamics analysis.
Acknowledgments—The cysteine-less mouse Mdr3 construct was
kindly provided by Drs. Gregory Tombline and Alan Senior, Univer-
sity of Rochester, Rochester, NY. We thank Drs. Tom Duncan and
Stewart Loh for the use of the fluorimeter and for many helpful
1. Rees, D. C., Johnson, E., and Lewinson, O. (2009) Nat. Rev. Mol. Cell Biol.
10, 218 –227
2. Jones, P. M., O’Mara, M. L., and George, A. M. (2009) Trends Biochem. Sci.
34, 520 –531
3. Kos, V., and Ford, R. C. (2009) Cell. Mol. Life Sci. 66, 3111–3126
4. Holland, I. B., and Blight, M. A. (1999) J. Mol. Biol. 293, 381–399
5. Dean, M., Hamon, Y., and Chimini, G. (2001) J. Lipid Res. 42, 1007–1017
6. Juliano, R. L., and Ling, V. (1976) Biochim. Biophys. Acta 455, 152–162
7. Gottesman, M. M., and Ling, V. (2006) FEBS Lett. 580, 998 –1009
8. Sharom, F. J. (2008) Pharmacogenomics 9, 105–127
9. Gottesman, M. M., Fojo, T., and Bates, S. E. (2002) Nat. Rev. Cancer 2,
10. Higgins, C. F. (2007) Nature 446, 749–757
11. Aller, S. G., Yu, J., Ward, A., Weng, Y., Chittaboina, S., Zhuo, R., Harrell,
P. M., Trinh, Y. T., Zhang, Q., Urbatsch, I. L., and Chang, G. (2009) Science
323, 1718 –1722
12. Dey, S., Ramachandra, M., Pastan, I., Gottesman, M. M., and Ambudkar,
S. V. (1997) Proc. Natl. Acad. Sci. U.S.A. 94, 10594 –10599
13. Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2003) J. Biol. Chem. 278,
14. Lugo, M. R., and Sharom, F. J. (2005) Biochemistry 44, 643– 655
15. Urbatsch, I. L., Tyndall, G. A., Tombline, G., and Senior, A. E. (2003) J. Biol.
Chem. 278, 23171–23179
16. Sauna, Z. E., Kim, I. W., Nandigama, K., Kopp, S., Chiba, P., and Ambud-
kar, S. V. (2007) Biochemistry 46, 13787–13799
17. Al-Shawi, M. K., and Omote, H. (2005) J. Bioenerg. Biomembr. 37,
18. Siarheyeva, A., Liu, R., and Sharom, F. J. (2010) J. Biol. Chem. 285,
19. Delannoy, S., Urbatsch, I. L., Tombline, G., Senior, A. E., and Vogel, P. D.
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
(2005) Biochemistry 44, 14010 –14019
20. Urbatsch, I. L., Sankaran, B., Weber, J., and Senior, A. E. (1995) J. Biol.
Chem. 270, 19383–19390
21. Gutmann, D. A., Ward, A., Urbatsch, I. L., Chang, G., and van Veen, H. W.
(2010) Trends Biochem. Sci. 35, 36–42
22. Lee, J. Y., Urbatsch, I. L., Senior, A. E., and Wilkens, S. (2002) J. Biol. Chem.
277, 40125– 40131
23. Lee, J. Y., Urbatsch, I. L., Senior, A. E., and Wilkens, S. (2008) J. Biol. Chem.
283, 5769 –5779
24. Urbatsch, I. L., Gimi, K., Wilke-Mounts, S., Lerner-Marmarosh, N., Rous-
seau, M. E., Gros, P., and Senior, A. E. (2001) J. Biol. Chem. 276,
25. Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2002) J. Biol. Chem. 277,
26. Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2003) J. Biol. Chem. 278,
27. Verhalen, B., and Wilkens, S. (2011) J. Biol. Chem. 286, 10476 –10482
28. Tombline, G., Urbatsch, I. L., Virk, N., Muharemagic, A., White, L. B., and
Senior, A. E. (2006) Arch. Biochem. Biophys. 445, 124–128
29. Beaudet, L., Urbatsch, I. L., and Gros, P. (1998) Methods Enzymol. 292,
30. Lerner-Marmarosh, N., Gimi, K., Urbatsch, I. L., Gros, P., and Senior, A. E.
(1999) J. Biol. Chem. 274, 34711–34718
31. Du¨ser, M. G., Bi, Y., Zarrabi, N., Dunn, S. D., and Bo¨rsch, M. (2008) J. Biol.
Chem. 283, 33602–33610
32. Johnson, K. M., Swenson, L., Opipari, A. W., Jr., Reuter, R., Zarrabi, N.,
Fierke, C. A., Bo¨rsch, M., and Glick, G. D. (2009) Biopolymers 91, 830 840
33. Ernst, S., Verhalen, B., Wilkens, S., and Bo¨rsch, M. (2011) Proc. SPIE 7902,
34. Zarrabi, N., Zimmermann, B., Diez, M., Gra¨ber, P., Wrachtrup, J., and
Bo¨rsch, M. (2005) Proc. SPIE 5699, 175–188
35. Du¨ser, M. G., Zarrabi, N., Cipriano, D. J., Ernst, S., Glick, G. D., Dunn,
S. D., and Bo¨rsch, M. (2009) EMBO J. 28, 2689 –2696
36. Dawson, R. J., and Locher, K. P. (2006) Nature 443, 180 –185
37. Berney, C., and Danuser, G. (2003) Biophys. J. 84, 3992– 4010
38. Stryer, L., and Haugland, R. P. (1967) Proc. Natl. Acad. Sci. U.S.A. 58,
39. Eidne, K. A., Kroeger, K. M., and Hanyaloglu, A. C. (2002) Trends Endo-
crinol. Metab. 13, 415– 421
40. Kapanidis, A. N., Lee, N. K., Laurence, T. A., Doose, S., Margeat, E., and
Weiss, S. (2004) Proc. Natl. Acad. Sci. U.S.A. 101, 8936 8941
41. Mu¨ller, B. K., Zaychikov, E., Bra¨uchle, C., and Lamb, D. C. (2005) Biophys.
J. 89, 3508 –3522
42. Schro¨der, G. F., and Grubmu¨ller, H. (2003) J. Chem. Phys. 119, 9920 –9924
43. Smith, P. C., Karpowich, N., Millen, L., Moody, J. E., Rosen, J., Thomas,
P. J., and Hunt, J. F. (2002) Mol. Cell 10, 139 –149
44. Chen, J., Lu, G., Lin, J., Davidson, A. L., and Quiocho, F. A. (2003) Mol. Cell
12, 651– 661
45. Ward, A., Reyes, C. L., Yu, J., Roth, C. B., and Chang, G. (2007) Proc. Natl.
Acad. Sci. U.S.A. 104, 19005–19010
46. Oldham, M. L., Khare, D., Quiocho, F. A., Davidson, A. L., and Chen, J.
(2007) Nature 450, 515–521
47. Tombline, G., Bartholomew, L. A., Tyndall, G. A., Gimi, K., Urbatsch, I. L.,
and Senior, A. E. (2004) J. Biol. Chem. 279, 46518 46526
48. Sankaran, B., Bhagat, S., and Senior, A. E. (1997) Biochemistry 36,
49. Liu, R., and Sharom, F. J. (1998) Biochemistry 37, 6503– 6512
50. Liu, R., and Sharom, F. J. (1996) Biochemistry 35, 11865–11973
51. Qu, Q., Russell, P. L., and Sharom, F. J. (2003) Biochemistry 42, 1170 –1177
52. Tombline, G., and Senior, A. E. (2005) J. Bioenerg. Biomembr. 37, 497–500
53. Bo¨rsch, M., Diez, M., Zimmermann, B., Reuter, R., and Gra¨ber, P. (2002)
FEBS Lett. 527, 147–152
54. Zimmermann, B., Diez, M., Zarrabi, N., Gra¨ber, P., and Bo¨rsch, M. (2005)
EMBO J. 24, 2053–2063
55. Russell, P. L., and Sharom, F. J. (2006) Biochem. J. 399, 315–323
56. McKinney, S. A., Joo, C., and Ha, T. (2006) Biophys J. 91, 1941–1951
57. Senior, A. E., al-Shawi, M. K., and Urbatsch, I. L. (1995) FEBS Lett. 377,
58. Loo, T. W., Bartlett, M. C., and Clarke, D. M. (2010) Biochem. Biophys. Res.
Commun. 395, 436 440
59. Ramachandra, M., Ambudkar, S. V., Chen, D., Hrycyna, C. A., Dey, S.,
Gottesman, M. M., and Pastan, I. (1998) Biochemistry 37, 5010 –5019
60. Qu, Q., Chu, J. W., and Sharom, F. J. (2003) Biochemistry 42, 1345–1353
61. Orelle, C., Alvarez, F. J., Oldham, M. L., Orelle, A., Wiley, T. E., Chen, J.,
and Davidson, A. L. (2010) Proc. Natl. Acad. Sci. U.S.A. 107, 20293–20298
62. Jones, P. M., and George, A. M. (2011) Biophys. J. 100, 3025–3034
63. Lu, G., Westbrooks, J. M., Davidson, A. L., and Chen, J. (2005) Proc. Natl.
Acad. Sci. U.S.A. 102, 17969 –17974
64. Oldham, M. L., and Chen, J. (2011) Proc. Natl. Acad. Sci. U.S.A. 108,
65. Sauna, Z. E., and Ambudkar, S. V. (2001) J. Biol. Chem. 276, 11653–11661
66. Senior, A. E. (2011) Proc. Natl. Acad. Sci. U.S.A. 108, 15015–15016
67. Cohen, A. E., and Moerner, W. E. (2005) Proc. SPIE 5699, 296 –305
68. Cohen, A. E., and Moerner, W. E. (2006) Proc. Natl. Acad. Sci. U.S.A. 103,
69. Bo¨rsch, M. (2011) Biol. Chem. 392, 135–142
Dynamics of P-glycoprotein NBDs by FRET Spectroscopy
by guest on October 27, 2015 from
Börsch and Stephan Wilkens
Brandy Verhalen, Stefan Ernst, Michael
Transfer Spectroscopy
Probed by Fluorescence Resonance Energy
Rearrangements in P-glycoprotein as
Dynamic Ligand-induced Conformational
Membrane Biology:
doi: 10.1074/jbc.M111.301192 originally published online November 15, 2011
2012, 287:1112-1127.J. Biol. Chem.
10.1074/jbc.M111.301192Access the most updated version of this article at doi:
.JBC Affinity SitesFind articles, minireviews, Reflections and Classics on similar topics on the
When a correction for this article is posted
When this article is cited
to choose from all of JBC's e-mail alertsClick here
Supplemental material:
This article cites 69 references, 26 of which can be accessed free at
by guest on October 27, 2015 from
... The switch of TMDs and their rearrangement allow a 180 o rotation of lipids. Such a mechanism is very relevant in regard to observations on many ABC exporters (Husada et al. [2018]; Verhalen et al. [2012Verhalen et al. [ , 2017; Cooper and Altenberg [2013]; Zoghbi and Altenberg [2018]). ...
... BmrA purification was done using the histidine tag present at its C-terminal and in detergent DDM at 0, 05% (weight). Su-Jin Paik designed several mutants during her thesis so that we could observe the FRET signal with different dyes positions on the NBDs (see figure 4.1): • M1: C436S, cystein-less mutant • M2: C436S / T444C, the cystein is moved to a close position compared to WT (blue position on figure 4.1), • M3: C436S / T559C, the cystein is placed on a loop at the bottom of the NBD, a way more accessible position (yellow position on figure 4.1).The choice of these mutants was motivated by reported studies of FRET and LRET on BmrA homologs, MsbA and PgP(Cooper and Altenberg [2013];Verhalen et al. [2012];Zoghbi et al. [2017]). The purification and labelling protocols are similar between the WT and the mutants. ...
Cell membranes are composed of a lipid bilayer, crowded with transmembrane proteins that mediate nearly all of the membrane functions, such as detoxification and communication. There is a tight interaction between lipids and transmembrane proteins. Membrane can apply mechanical stress and impact transmembrane proteins shape and function. Reciprocally, the inclusion of a transmembrane protein can bend or stretch the membrane. In particular, this interplay might play a crucial role for proteins that change conformation to mediate cargoes transport. We study BmrA, a bacterial ABC exporter from B.subtillis. ABC (ATP Binding Cassette) transporters represent one of the largest families of membrane proteins. Some ABC exporters lead to a phenotype of multi-drug resistance, such as human P-glycoprotein (P-gp) which transports anti-cancer agents out the cell. BmrA shares high homology with P-gp and is expected to undergo a large conformational change between “open” and “closed” conformations. The objective is to characterize the interplay between the conformation cycle of BmrA and membrane curvature, at the single molecule level. BmrA is purified in detergent, labelled with dyes suitable for FRET experiments, and incorporated in small liposomes. Conformations of BmrA are probed by single-molecule FRET in TIRF microscopy: a low (resp. high) FRET efficiency corresponds to a protein in open (resp. closed) conformation. Three different liposome sizes are used in order to vary the membrane curvature: diameter 140, 60 and 40nm. We have found a different protein behavior in 40nm liposomes as compared to larger ones, indicating an effect of the membrane curvature on BmrA conformations.
... ADP concentration. In agreement with our data, LRET [23], FRET [75] and EPR [73] experiments showed that the NBDs assume larger inter-domain distances in the absence of ATP. ...
Full-text available
Several ABC exporters carry a degenerate nucleotide binding site (NBS) that is unable to hydrolyze ATP at a rate sufficient for sustaining transport activity. A hallmark of a degenerate NBS is the lack of the catalytic glutamate in the Walker B motif in the nucleotide binding domain (NBD). The multidrug resistance transporter ABCB1 (P-glycoprotein) has two canonical NBSs, and mutation of the catalytic glutamate E556 in NBS1 renders ABCB1 transport-incompetent. In contrast, the closely related bile salt export pump ABCB11 (BSEP), which shares 49% sequence identity with ABCB1, naturally contains a methionine in place of the catalytic glutamate. The NBD-NBD interfaces of ABCB1 and ABCB11 differ only in four residues, all within NBS1. Mutation of the catalytic glutamate in ABCB1 results in the occlusion of ATP in NBS1, leading to the arrest of the transport cycle. Here we show that despite the catalytic glutamate mutation (E556M), ABCB1 regains its ATP-dependent transport activity, when three additional diverging residues are also replaced. Molecular dynamics simulations revealed that the rescue of ATPase activity is due to the modified geometry of NBS1, resulting in a weaker interaction with ATP, which allows the quadruple mutant to evade the conformationally locked pre-hydrolytic state to proceed to ATP-driven transport. In summary, we show that ABCB1 can be transformed into an active transporter with only one functional catalytic site by preventing the formation of the ATP-locked pre-hydrolytic state in the non-canonical site.
... These results suggest that verapamil causes conformational changes in P-gp. (14), artificial lipid bilayers (5,7,15,16) or membrane vesicles derived from P-gpoverexpressing cells (17,18). verapamil, but low in the absence of ATP (Fig. 4D, 6B). ...
Full-text available
P-glycoprotein (P-gp; also known as MDR1 or ABCB1) is an ATP-driven multidrug transporter that extrudes various hydrophobic toxic compounds to the extracellular space. P-gp consists of two transmembrane domains (TMDs) that form the substrate translocation pathway and two nucleotide-binding domains (NBDs) that bind and hydrolyze ATP. At least two P-gp states are required for transport. In the inward-facing (pre-drug transport) conformation, the two NBDs are separated, and the two TMDs are open to the intracellular side; in the outward-facing (post-drug transport) conformation, the NBDs are dimerized, and the TMDs are slightly open to the extracellular side. ATP binding and hydrolysis cause conformational changes between the inward-facing and the outward-facing conformations and these changes help translocate substrates across the membrane. However, how ATP hydrolysis is coupled to these conformational changes remains unclear. In this study, we used a new FRET sensor that detects conformational changes in P-gp to investigate the role of ATP binding and hydrolysis during the conformational changes of human P-gp in living HEK293 cells. We show that ATP binding causes the conformational change to the outward-facing state and that ATP hydrolysis and subsequent release of γ-phosphate from both NBDs allow the outward-facing state to return to the original inward-facing state. The findings of our study underscore the utility of using FRET analysis in living cells to elucidate the function of membrane proteins such as multidrug transporters.
In all domains of life, transmembrane proteins from the ATP‐binding cassette (ABC) transporter family drive the translocation of diverse substances across lipid bilayers. In pathogenic fungi, the ABC transporters of the pleiotropic drug resistance (PDR) subfamily confer antibiotic resistance and so are of interest as therapeutic targets. They also drive the quest for understanding how ABC transporters can generally accommodate such a wide range of substrates. The Pdr5 transporter from baker's yeast is representative of the PDR group and, ever since its discovery more than 30 years ago, has been the subject of extensive functional analyses. A new perspective of these studies has been recently provided in the framework of the first electron cryo‐microscopy structures of Pdr5, as well as emergent applications of machine learning in the field. Taken together, the old and the new developments have been used to propose a mechanism for the transport process in PDR proteins. This mechanism involves a “flippase” step that moves the substrates from one leaflet of the bilayer to the other, as a central element of cellular efflux. Asymmetric motions during the ATP hydrolysis cycle catalyzes the Pdr5‐mediated substrate flipping from the inner to the outer leaflet of the plasma membrane.
Full-text available
The human multidrug transporter P‐glycoprotein (P‐gp) or ABCB1 protects tissues from xenobiotics and other toxic compounds by pumping them out of cells. P‐gp is also overexpressed by several types of cancer and has been implicated in altering the bioavailability of chemotherapeutic drugs due to efflux, resulting in the development of multidrug resistance. After decades of research, the modulation of P‐gp‐mediated drug transport to overcome drug resistance in the clinic has not been successful. In this chapter, we focus on progress made in last five years on the biochemical, pharmacological, and structural aspects of P‐gp. We discuss the biochemical and structural basis of the polyspecificity of P‐gp and the mechanism of drug transport. We also review recent efforts to develop non‐toxic natural product and small‐molecule modulators of P‐gp that appear very promising for use in combination with anticancer drugs to improve the efficiency of chemotherapy.
Full-text available
Uncovering the structure and function of biomolecules is a fundamental goal in structural biology. Membrane‐embedded transport proteins are ubiquitous in all kingdoms of life. Despite structural flexibility, their mechanisms are typically studied by ensemble biochemical methods or by static high‐resolution structures, which complicate a detailed understanding of their dynamics. Here, we review the recent progress of single molecule Förster Resonance Energy Transfer (smFRET) in determining mechanisms and timescales of substrate transport across membranes. These studies do not only demonstrate the versatility and suitability of state‐of‐the‐art smFRET tools for studying membrane transport proteins but they also highlight the importance of membrane mimicking environments in preserving the function of these proteins. The current achievements advance our understanding of transport mechanisms and have the potential to facilitate future progress in drug design.
The molecular mechanisms of P-glycoprotein (P-gp; also known as MDR1 or ABCB1) have been mainly investigated using artificial membranes such as lipid-detergent mixed micelles, artificial lipid bilayers, and membrane vesicles derived from cultured cells. Although these in vitro experiments help illustrate details about the molecular mechanisms of P-gp, they do not reflect physiological membrane environments in terms of lateral pressure, curvature, constituent lipid species, etc. The protocol presented here includes a detailed guide for analyzing the conformational change of human P-gp in living HEK293 cells by using intramolecular fluorescence resonance energy transfer (FRET), in which excitation of the donor fluorophore is transferred to the acceptor without emission of a photon when two fluorescent proteins are in close proximity. Combining FRET analysis with membrane permeabilization, the contribution of small molecules such as nucleotides to the conformational change can be evaluated in living cells.
ABC transporters use the energy of ATP binding and hydrolysis to transport substrates across cellular membranes. They comprise two highly conserved nucleotide binding domains and two transmembrane domains that form the transmembrane channel and contain the substrate binding sites. Structural analyses have found a variety of seemingly unrelated folds for the ABC transporter transmembrane domains, and from these a set of diverse mechanistic models has been inferred. Nevertheless, in spite of the explosion in structure determination of ABC transporters in the last decade, advancement in certainty and clarity as to fundamental aspects of their molecular mechanisms remains elusive. With this in mind, here we put and examine the question: Could current ABC structures differ from the physiologic membrane-embedded forms?
The classic conceptualisation of ATP binding cassette (ABC) transporter function is an ATP‐dependent conformational change coupled to transport of a substrate across a biological membrane via the transmembrane domains (TMDs). The binding of two ATP molecules within the transporter’s two nucleotide binding domains (NBDs) induces their dimerization. Despite retaining the ability to bind nucleotides, isolated NBDs frequently fail to dimerise. ABC proteins without a TMD, e.g. ABCE and ABCF, have NBDs tethered via elaborate linkers, further supporting that NBD dimerization does not readily occur for isolated NBDs. Intriguingly, even in full‐length transporters, the NBD‐dimerised, outward‐facing state is not as frequently observed as might be expected. This leads to questions regarding what drives NBD interaction and the role of the TMDs or linkers. Understanding the NBD‐nucleotide interaction and the subsequent NBD dimerization is thus pivotal for understanding ABC transporter activity in general. Here, we hope to provide new insights into ABC protein function by discussing the perplexing issue of (missing) NBD dimerization in isolation and in the context of full‐length ABC proteins.
This chapter discusses cytochrome P450 systems and presents some case studies with analysis for drug metabolism. To enter the circulation from an oral dose, the drugs must pass through the gut, the portal circulation and then into the liver itself. Since the liver's blood flow is not likely to undergo a sustained increase, then the only way such a large effect on drug levels can occur is that the liver is extracting much more of the drug than usual in the presence of the other drug. This acceleration of drug metabolism as a response to the presence of certain drugs is known as 'enzyme induction' and drugs that cause it are often referred to as 'inducers' of drug metabolism. There are several broad groups of drugs and chemicals capable of inducing hepatic metabolism; these include: anticonvulsants, steroids, anti‐androgens, antibacterials/antifungals, anti‐virals, herbal remedies, polycyclic aromatic hydrocarbons, recreational agents, etc.
Full-text available
ATP-binding cassette transporters are powered by ATP, but the mechanism by which these transporters hydrolyze ATP is unclear. In this study, four crystal structures of the full-length wild-type maltose transporter, stabilized by adenosine 5'-(β,γ-imido)triphosphate or ADP in conjunction with phosphate analogs BeF(3)(-), VO(4)(3-), or AIF(4)(-), were determined to 2.2- to 2.4-Å resolution. These structures led to the assignment of two enzymatic states during ATP hydrolysis and demonstrate specific functional roles of highly conserved residues in the nucleotide-binding domain, suggesting that ATP-binding cassette transporters catalyze ATP hydrolysis via a general base mechanism.
Full-text available
P-glycoprotein (Pgp), a member of the ABC transporter family, functions as an ATP hydrolysis-driven efflux pump to rid the cell of toxic organic compounds, including a variety of drugs used in anti-cancer chemotherapy. We have recently obtained EM projection images of lipid-bound Pgp without nucleotide and transport substrate that showed the two halves of the transporter separated by a central cavity (Lee, J. Y., Urbatsch, I. L., Senior, A. E., and Wilkens, S. (2002) J. Biol. Chem. 277, 40125-40131). Addition of nucleotide and/or substrate lead to a close association of the two halves of the transporter, thereby closing the central cavity (Lee, J. Y., Urbatsch, I. L., Senior, A. E., and Wilkens, S. (2008) J. Biol. Chem. 283, 5769-5779). Here, we used cysteine-mediated disulfide cross-linking to further delineate the structural rearrangements of the two nucleotide binding domains (NBD1 and NBD2) that take place during catalysis. Cysteines introduced at or near the C-terminal ends of NBD1 and NBD2 allowed for spontaneous disulfide cross-linking under nonreducing conditions. For mutant A627C/S1276C, disulfide formation was with high efficiency and cross-linked Pgp retained 30-68% drug-stimulated ATPase activity compared with reduced or cysteine-less Pgp. Two other cysteine pairs (K615C/S1276C and A627C/K1260C) also formed a disulfide but to a lesser extent, and the cross-linked form of these two mutants had lower drug-stimulated ATPase activity. The data suggest that the C-terminal ends of the two NBDs of Pgp are not required to undergo significant motion with respect to one another during the catalytic cycle.
Full-text available
FoF1-ATP synthase is the essential membrane enzyme maintaining the cellular level of adenosine triphosphate (ATP) and comprises two rotary motors. We measure subunit rotation in FoF1-ATP synthase by intramolecular Foerster resonance energy transfer (FRET) between two fluorophores at the rotor and at the stator of the enzyme. Confocal FRET measurements of freely diffusing single enzymes in lipid vesicles are limited to hundreds of milliseconds by the transit times through the laser focus. We evaluate two different methods to trap the enzyme inside the confocal volume in order to extend the observation times. Monte Carlo simulations show that optical tweezers with low laser power are not suitable for lipid vesicles with a diameter of 130 nm. A. E. Cohen (Harvard) and W. E. Moerner (Stanford) have recently developed an Anti-Brownian electrokinetic trap (ABELtrap) which is capable to apparently immobilize single molecules, proteins, viruses or vesicles in solution. Trapping of fluorescent particles is achieved by applying a real time, position-dependent feedback to four electrodes in a microfluidic device. The standard deviation from a given target position in the ABELtrap is smaller than 200 nm. We develop a combination of the ABELtrap with confocal FRET measurements to monitor single membrane enzyme dynamics by FRET for more than 10 seconds in solution.
Full-text available
ATP-binding cassette (ABC) transporters are powered by a nucleotide-binding domain dimer that opens and closes during cycles of ATP hydrolysis. These domains consist of a RecA-like subdomain and an α-helical subdomain that is specific to the family. Many studies on isolated domains suggest that the helical subdomain rotates toward the RecA-like subdomain in response to ATP binding, moving the family signature motif into a favorable position to interact with the nucleotide across the dimer interface. Moreover, the transmembrane domains are docked into a cleft at the interface between these subdomains, suggesting a putative role of the rotation in interdomain communication. Electron paramagnetic resonance spectroscopy was used to study the dynamics of this rotation in the intact Escherichia coli maltose transporter MalFGK(2). This importer requires a periplasmic maltose-binding protein (MBP) that activates ATP hydrolysis by promoting the closure of the cassette dimer (MalK(2)). Whereas this rotation occurred during the transport cycle, it required not only trinucleotide, but also MBP, suggesting it is part of a global conformational change in the transporter. Interaction of AMP-PNP-Mg(2+) and a MBP that is locked in a closed conformation induced a transition from open MalK(2) to semiopen MalK(2) without significant subdomain rotation. Inward rotation of the helical subdomain and complete closure of MalK(2) therefore appear to be coupled to the reorientation of transmembrane helices and the opening of MBP, events that promote transfer of maltose into the transporter. After ATP hydrolysis, the helical subdomain rotates out as MalK(2) opens, resetting the transporter in an inward-facing conformation.
The Anti-Brownian ELectrophoretic trap (ABEL trap) is a new device that allows a user to trap and manipulate fluorescent objects as small as 20 nm freely diffusing in solution. We describe in detail how to build an ABEL trap.
For a time, membrane proteins proved intractable to X-ray crystallography. There were difficulties of overexpression and purification, multiple choices of solubilizing detergent that could only be decided by trial and error, unwillingness to crystallize, and often, poor diffraction from resultant crystals. However, the tide now seems to be turning, and in PNAS, Oldham and Chen (1) present an X-ray crystallography study of the entire WT maltose uptake protein from Escherichia coli, in which four different structures are described at commendable 2.2- to 2.4-A resolution. Maltose uptake protein is a well-characterized model (2) for the large ABC transporter family, members of which occur in all living organisms. ABC transporters, whether exporters or importers, display a common architecture consisting of two transmembrane domains (TMDs) involved directly in transport and two nucleotide binding domains (NBDs) that cooperatively hydrolyze ATP (actually MgATP) to power transport (3). Internationally, numerous researchers study ABC transporters because of their wide involvement in human and plant disease and consequent importance in medicine and agriculture. Maltose uptake protein has subunit structure MalGFK2, with GF forming the TMDs and K2 forming the head to tail NBD dimer, with its two ATP binding sites. Bacterial ABC importers use a binding protein that captures the requisite transport substrate from the periplasm and delivers it to the transporter. In the work by Oldham and Chen (1), maltose binding protein is bound to the MalGFK2 complex in its open state, and the just-delivered maltose molecule is clearly seen in the transport pathway at midmembrane.
Very little structural information is available for P-glycoprotein (Pgp), which has been implicated in the multidrug resistance of human tumors because of its ability to act as an ATP-driven efflux pump for hydrophobic compounds. Highly purified Pgp has been labeled on two cysteine residues with the fluorescence probe NBD-Cl (7-chloro-4-nitro-2,1,3-benzoxadiazole). We show that NBD labels the same cysteine residues as MIANS [2-(4-maleimidoanilino)naphthalene-6-sulfonic acid]; they are located within the Walker A motif of the nucleotide binding domain, close to the site where ATP binds. NBD- and MIANS-labeled Pgps were reconstituted by detergent dilution into phospholipid vesicles containing increasing mole fractions of rhodamine- or NBD-labeled phosphatidylethanolamine (PE), respectively. The fluorescence of the NBD-Pgp and MIANS-Pgp donors was quenched in a concentration-dependent manner by the rhodamine-PE and NBD-PE acceptors. Using two different methods to analyze Förster resonance energy transfer, the distance of the Pgp-bound probes from the lipid-water interfacial region of the bilayer was estimated to be 31-35 A. This distance is compatible with the low-resolution structure of Pgp determined by electron microscopy, and indicates that the nucleotide binding domains lie close to the membrane surface. The experimental data fitted very well to theoretical quench curves for a single protein-bound fluor, suggesting that the two nucleotide binding domains are located equidistant from the bilayer. Following the addition of ATP to MIANS-Pgp, the NBD-PE quench curve no longer conformed to the models. These results imply that Pgp interacts differently with PE when it is in the ATP-bound form.
Human P-glycoprotein (Pgp), a plasma membrane protein that confers multidrug resistance, functions as an ATP-dependent drug efflux pump. Pgp contains two ATP binding/utilization sites and exhibits ATPase activity that is stimulated in the presence of substrates and modulating agents. The mechanism of coupling of ATP hydrolysis to drug transport is not known. To understand the role of ATP hydrolysis in drug binding, it is necessary to develop methods for purifying and reconstituting Pgp that retains properties including stimulation of ATPase activity by known substrates to an extent similar to that in the native membrane. In this study, (His)6-tagged Pgp was expressed in Trichoplusia ni (High Five) cells using the recombinant baculovirus system and purified by metal affinity chromatography. Upon reconstitution into phospholipid vesicles, purified Pgp exhibited specific binding to analogues of substrates and ATP in affinity labeling experiments and displayed a high level of drug-stimulated ATPase activity (specific activity ranging from 4.5 to 6.5 micromol min-1 mg-1). The ATPase activity was inhibited by ADP in a competitive manner, and by vanadate and N-ethylmaleimide at low concentrations. Vanadate which is known to inhibit ATPase activity by trapping MgADP at the catalytic site inhibited photoaffinity labeling of Pgp with substrate analogues, [125I]iodoarylazidoprazosin and [3H]azidopine, only under ATP hydrolysis conditions. Because vanadate-trapped Pgp is known to resemble the ADP and phosphate-bound catalytic transition state, our findings indicate that ATP hydrolysis results in a conformation with reduced affinity for substrates. A catalytic transition conformation with reduced affinity would essentially result in substrate dissociation and supports a model for drug transport in which an ATP hydrolysis-induced conformational change leads to drug release toward the extracellular medium.
ATP-binding cassette transporters use the energy of ATP hydrolysis to transport substrates across cellular membranes. They have two transmembrane domains and two cytosolic nucleotide-binding domains. Biochemical studies have characterized an occluded state of the transporter in which nucleotide is tenaciously bound in one active site, whereas the opposite active site is empty or binds nucleotide loosely. Here, we report molecular-dynamics simulations of the bacterial multidrug ATP-binding cassette transporter Sav1866. In two simulations of the ATP/apo state, the empty site opened substantially by way of rotation of the nucleotide-binding domain (NBD) core subdomain, whereas the ATP-bound site remained occluded and intact. We correlate our findings with elastic network and molecular-dynamics simulation analyses of the Sav1866 NBD monomer, and with existing experimental data, to argue that the observed transition is physiological, and that the final structure observed in the ATP/apo simulations corresponds to the tight/loose state of the NBD dimer characterized experimentally.
Conformational changes of proteins can be monitored in real time by fluorescence resonance energy transfer (FRET). Two different fluorophores have to be attached to those protein domains which move during function. Distance fluctuations between the fluorophores are measured by relative fluorescence intensity changes or fluorescence lifetime changes. The rotary mechanics of the two motors of F(o)F(1)-ATP synthase have been studied in vitro by single-molecule FRET. The results are summarized and perspectives for other transport ATPases are discussed.