Insertion of Membrane Proteins into Discoidal Membranes Using a
Cell-Free Protein Expression Approach
Federico Katzen,‡Julia E. Fletcher,‡Jian-Ping Yang,‡Douglas Kang,‡Todd C. Peterson,‡
Jenny A. Cappuccio,†Craig D. Blanchette,†Todd Sulchek,†Brett A. Chromy,†
Paul D. Hoeprich,†Matthew A. Coleman,†and Wieslaw Kudlicki*,‡
Invitrogen Corporation, 5791 Van Allen Way, Carlsbad, California 92008, and Lawrence Livermore National
Laboratory, 7000 East Avenue, Livermore, California 94551
Received April 9, 2008
We report a cell-free approach for expressing and inserting integral membrane proteins into water-
soluble particles composed of discoidal apolipoprotein-lipid bilayers. Proteins are inserted into the
particles, circumventing the need of extracting and reconstituting the product into membrane vesicles.
Moreover, the planar nature of the membrane support makes the protein freely accessible from both
sides of the lipid bilayer. Complexes are successfully purified by means of the apoplipoprotein
component or by the carrier protein. The method significantly enhances the solubility of a variety of
membrane proteins with different functional roles and topologies. Analytical assays for a subset of
model membrane proteins indicate that proteins are correctly folded and active. The approach provides
a platform amenable to high-throughput structural and functional characterization of a variety of
traditionally intractable drug targets.
Keywords: Membrane protein • nanodisc • proteomics • cell-free • protein expression
Integral membrane proteins (MPs) comprise nearly 30% of
any given proteome, play fundamental roles in transport,
signaling, bioenergetics, and cell-growth processes, and ac-
count for over 50% of all human drug targets. However,
biochemical and structural characterization of membrane
bound proteins is currently cumbersome, often intractable, and
lags far behind the study of water soluble proteins. Overex-
pression of MPs in vivo frequently results in cell toxicity, protein
aggregation, misfolding, and low yield. Membrane protein
purification involves detergent extraction, detergent refolding,
or reconstitution into liposomes, processes that inexorably lead
to constraints on protein accessibility, surface immobilization,
and activity assays.1
A novel method for displaying active MPs in nanoscale
particles made of lipoprotein encircled membrane bilayers has
been described.2These particles are referred to as nanodiscs3
or nanolipoprotein particles (NLPs).4NLPs are self-assembled
discoidal particles composed of a planar phospholipid mem-
brane bilayer surrounded by an apolipoprotein ring (scaffold
protein). Advantages of NLPs include water-solubility, mono-
dispersity within preparations, consistency between prepara-
tions, flexible lipid composition, and more importantly, unre-
stricted and simultaneous access to both faces of the membrane
leaflet.5It has been demonstrated that certain MPs can be
incorporated into NLPs by adding detergent solubilized MPs
to the NLP assembly process.6Despite the demonstrated
capture of MPs in NLPs, the current method to produce NLP-
associated MPs is laborious and requires successful expression,
purification, and solubilization of MPs, many of which exhibit
detrimental idiosyncratic behaviors in the course of one or
more of these procedures, prior to assembly with NLPs. Here,
we report a simple approach to assemble MPs directly into
NLPs by introducing preformed empty NLPs into cell-free
membrane protein expression reactions.
Plasmids and Clones. Unless otherwise specified, genes
encoding MPs (Gateway compatible Ultimate ORF collection,
Invitrogen, Carlsbad, CA) were recombined into the plasmid
pEXP4-DEST (Invitrogen, Carlsbad, CA) following the manu-
facturer’s directions. The gene encoding EmrE from Escherichia
coli (GenBank acc. no. Z11877) was PCR-amplified using
TCATGAACCCTTATATTTATC-3′ and 5′-GGGGACCACTTTGTA-
CAAGAAAGCTGGGTCTTAATGTGGTGTGCTTCG-3′ and recom-
bined via BP and LR Gateway reactions (Invitrogen, Carlsbad,
CA) into pEXP6-DEST, a modified pEXP3-DEST plasmid (In-
vitrogen, Carlsbad, CA) devoid of the histidine tag coding
sequence. Also, the exact EmrE coding sequence (GenBank acc.
no. Z11877) was PCR-amplified and cloned into pEXP5-NT/
TOPO (Invitrogen, Carlsbad, CA). Human apoA1 (GenBank acc.
no. NM_000039) was PCR-amplified using primers 5′-CACGTG-
GATGAACCACCACAAAG-3′ and 5′-CCTAGGCTATTGAGTGT-
* To whom correspondence should be addressed. Dr. Wieslaw Kudlick,
Invitrogen Corp., 5791 Van Allen Way, Carlsbad, CA 92008. E-mail:
†Lawrence Livermore National Laboratory.
10.1021/pr800265f CCC: $40.75
Published on Web 06/17/2008
XXXX American Chemical Society
Journal of Proteome Research XXXX, xxx, 000 A
His (Invitrogen, Carslbad, CA) between the PmlI and AvrII
restriction sites. The bacteriorhodopsin (bR) gene sequence was
PCR-amplified from plasmid p72bop7using primers 5′-
CAAAAAAAACGGGCC-3′. The resulting PCR product was cloned
into pIVEX 2.4b (Roche Applied Science, Indianapolis, IN) using
the NdeI and BamHI restriction enzyme sites (pIVEX 2.4b-bR).
The plasmid (pET32-E4NT) and expression conditions to
produce the N-terminal 22 kDa fragment of human apolipo-
protein E4 (apoE422K) were described earlier.8The sequence
of the GFP gene (acc. no. U62637) was PCR-amplified and
cloned into pENTR/D-TOPO (Invitrogen, Carlsbad, CA). The
resulting plasmid was digested with NheI and PvuII, and the
released DNA fragment was cloned into a pUC18-derivative
between the T7 promoter and T7 terminator elements (pF-
NLP Samples and Lipids. Phospholipids were from Avanti
Polar Lipids (Alabaster, AL). NLPs composed of mature human
apoA1 and 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC)
were made using an established protocol with minor modifica-
tions.3Briefly, DMPC, cholate, and apoA1 were mixed in a
molar ratio of 140:280:1 and subjected to 3 temperature shift
cycles (RT for 10 min and 30 °C for 10 min). The mix was
incubated further for 90 min and the detergent was removed
with Bio-Beads SM-2 nonpolar polystyrene adsorbents (Bio-
Rad, Hercules, CA) following the manufacturer’s directions.
Monodisperse NLPs were recovered using size exclusion chro-
matography (Superdex 200 10/300 GL, GE Healthcare, Uppsala,
Sweden). NLPs composed of his-apoE422K and DMPC were
made as described.4Briefly, the apolipoprotein and lipids were
combined in a mass ratio of 4:1 in TBS buffer and subjected to
3 temperature shift cycles (30 and 20 °C) with a final overnight
incubation at 23.8 °C. NLPs were purified with size exclusion
chromatography (Superdex 200 10/300 GL, GE Healthcare,
Cell-Free Protein Expression Reactions. E. coli-based cell-
free protein expression reactions were set up using Expressway
Maxi Cell-Free E. coli Expression System (Invitrogen, Carlsbad,
CA) following the manufacturer’s directions. NLPs were added
to the reactions at a final concentration of 1.35 mg/mL. Where
indicated, [35S]Met (135 mCi/mmol final) (Perkin-Elmer, Walth-
am, MA) was added and yield was calculated by TCA precipita-
tion following the manufacturer’s directions. The soluble
fraction from the reactions was obtained by centrifugation at
14 000g for 5 min. Autoradiograms were generated by overnight
exposures. Where indicated, all trans-retinal (Sigma, St. Louis,
MO) was added at a final concentration of 5 µM. Eukaryotic-
based cell-free protein expression reactions were set up using
TNT-coupled rabbit reticulocyte lysate (Promega, Madison, WI)
and RTS 100 Wheat Germ CECF (Roche Applied Science,
Mannheim, Germany) kits following the manufacturer’s direc-
tions. NLPs were added to the reactions at a final concentration
of 0.27 mg/mL. Final specific activity for [35S]Met was 81600
and 91 mCi/mmol for the rabbit reticulocyte and wheat germ
Protein Analysis. Proteins were purified using the Ni-NTA
Purification System (Invitrogen, Carlsbad, CA) following the
manufacturer’s directions. In-gel visualization of proteins in
denaturing gels was performed by Coomassie-blue staining or
by using the Lumio Green Detection Kit (Invitrogen, Carlsbad,
CA) and subjecting the gel to laser scanning (Typhoon 8600
Variable Mode Imager, GE Healthcare, Piscataway, NJ). Native
gel electrophoresis was performed on 4-12% Tris-glycine
polyacrylamide gels (Invitrogen, Carlsbad, CA) using NativeP-
AGE running buffer (Invitrogen, Carlsbad, CA). Proteins were
visualized with Sypro Ruby protein gel stain (Bio-Rad, Hercules,
CA). Absorbance spectra of bR were collected in a Ultrospec
5300 pro UV/Visible spectrophotometer (Amersham Bioscienc-
es). Protein and NLP concentration were determined using the
BCA Protein Assay Kit (Pierce, Rockford, IL).
EmrE Activity Assays. EmrE activity was assayed using a
tetraphenylphosphonium (TPP+)-binding assay.9Briefly, EmrE
was expressed in vitro as described above and immobilized on
Ni2+-nitrilotriacetic acid beads (Probond Protein Purification
System; Invitrogen, Carlsbad, CA). The beads were then washed
with binding buffer containing 150 mM NaCl, 10 mM imida-
zole, and 15 mM Tris·Cl, pH 7.5, and the protein content was
estimated by gel densitometry. One tenth of a microgram of
EmrE was added to the binding buffer containing 0.125-320
nM [3H]TPP+(28 Ci/mmol; GE Healthcare), and incubated for
1 h at room temperature. Nonspecific binding was determined
by competition with 20 µM cold TPP+(Sigma-Aldrich, St. Louis,
MO). Data points were fitted to a saturation binding curve by
nonlinear regression using Prism (GraphPad Software, San
Atomic Force Microscopy (AFM). Atomically flat mica disks
were glued to metal substrates to secure them to the scanner
of a stand-alone MFP-3D AFM instrument (Asylum Research,
Santa Barbara, CA). The AFM has a closed loop in the x, y, and
z axes. The topographical images were obtained with Biolevers
(Olympus, Tokyo, Japan) with a spring constant of 0.03 N/m.
Images were taken in alternate contact (AC) mode in liquid,
with very low amplitudes at the primary resonance frequency
that was obtained from thermal analysis of the cantilever in
solution. Height, amplitude, and phase images were recorded.
Heights of contiguous particles in images were determined by
cross-sectional analysis in the slow scan direction, using IgorPro
software routines (WaveMetrics, Portland, OR). The maximum
particle height was recorded and the results were displayed by
histogram analysis. Experiments were carried out in a temper-
ature controlled room at 23 ( 1 °C. The thermal spectrum of
the cantilevers was obtained both in air and liquid, and the
stiffness was estimated by fitting with the thermal noise theory
and compared to the Sader method for the normal spring
constant of a rectangular cantilever.10The error in calculating
the spring constant is estimated to be <20%.
Size Exclusion Chromatography (SEC). SEC was conducted
using a Shimadzu VP HPLC device (Columbia, MD) using a
Superdex 200 10/300 GL column (GE Healthcare, Uppsala,
Sweden), in TBS at a flow rate of 0.5 mL/min. The column was
calibrated with four protein standards (thyroglobulin, ferritin,
aldolase, and chymotrypsinogen) of known molecular weight
and stokes diameter that span the separation range of the
column and the NLP samples. The void volume was established
with blue dextran. When indicated, samples were previously
concentrated through a Vivaspin 100 kDa polyethersulfone
membrane (Sartorious, Edgewood, NY).
Two Strategies for Complex Formation and Purification.
In initial experiments, we used the E. coli multidrug transporter
EmrE (GenBank acc no. Z11877), a protein that has been
extensively studied and is known to express at relatively high
rates in cell-free protein expression systems.11For these
experiments, the scaffold proteins (apoA1 or apoE422K) in the
preformed NLPs were affinity-tagged for purification, whereas
Katzen et al.
BJournal of Proteome Research • Vol. xxx, No. xx, XXXX
EmrE was expressed with a lumio recognition sequence (Cys-
Cys-Pro-Gly-Cys-Cys) but devoid of a poly-histidine tag. The
opposite strategy, where the affinity-tagged protein was EmrE,
was also followed. Under our conditions, EmrE expressed at
levels up to 0.4 mg/mL of reaction. In all these cases, results
showed that EmrE produced by the cell-free methodology
readily copurified with the NLPs, suggesting that complexes
were formed between NLPs and EmrE (Figure 1a,b). In addi-
tion, a strong correlation between complex formation and
protein solubility was found (compare panels a and c of Figure
1). Finally, activity assays showed that the EmrE substrate
[3H]tetraphenylphosphonim (TPP+) binds specifically to EmrE
only when NLPs are included in the reaction (Figure 1d, inset).
Earlier kinetics measurements of EmrE based on [3H]TPP+
binding were conducted in the presence of a variety of
detergents such as N-dodecyl-b-D-maltoside, N-octyl-?-D-glu-
copyranoside and SDS.12–15In those circumstances, Kdvalues
ranging from 2.6 to 45 nM were reported.12–15[3H]TPP+binding
using EmrE preparations reconstituted into unilamellar vesicles
could not be determined.13In our case, we determined the
binding constants of EmrE complexed with NLPs in the absence
of detergents finding that the binding to [3H]TPP+is saturable
at a value of 15.1 pmol ligand/µg EmrE with a Kdof 19.17 nM
(Figure 1d). Overall, our results indicate that EmrE is active,
suggesting that it is correctly folded, therefore, correctly
inserted into the NLPs’ lipid bilayer.
Analytical Characterization of the Complexes. To further
study the biochemical and biophysical properties of MP-NLP
complexes, Halobacterium salinarum bacteriorhodopsin (bR)
(GenBank acc. no. J02755) was used as a second model system.
Twice as large as EmrE, bR allowed us to readily distinguish
between empty and loaded NLPs when using a variety of
biochemical and biophysical methods (see below).
The bR protein is a 7 transmembrane, R-helix, light-driven
proton pump with a cofactor pocket in which a retinal molecule
is covalently linked. Although its mechanism does not involve
Figure 1. In situ complex formation between NLPs and EmrE. (a) E. coli EmrE was expressed in vitro either in the absence or presence
of NLPs. NLPs were composed of histidine-tagged apoA1 (his-apoA1) as a scaffold and DMPC, or histidine-tagged apoE422K (his-
apoE422K) and DMPC. Proteins were separated by SDS electrophoresis and stained by Coomassie blue or visualized by laser scan
imaging. EmrE was fused to a Lumio recognition sequence (Cys-Cys-Pro-Gly-Cys-Cys)37and lacked a histidine tag. (b) Reactions were
set up and processed as in panel a. NLPs were composed of apoA1 devoid of any additional sequence and DMPC. EmrE was fused to
a histidine tag. (c) Proteins from the total crude extract and from the soluble fraction in panel a were separated by SDS-electrophoresis
and visualized by laser scan imaging. (d) [3H]TPP+binding analyses were performed in triplicate as described9(filled circles). Unspecific
binding (filled squares) was determined by assessing [3H]TPP+bound in the presence of 20 µM nonradioactive competitor. The DNA
source for EmrE was pEXP5-NT-EmrE. In the inset, [3H]TPP+binding was performed in the absence (empty bars) or presence (filled
bars) of nonradioactive TPP+. Binding reactions were carried out with EmrE synthesized in the presence (+) or absence (-) of NLPs.
NLPs were composed of apoA1 devoid of any additional sequence and DMPC. C, crude extract; FT, flow-through fraction; W, wash
fractions; E, imidazole-eluted fractions, T, total crude extract; S, supernatant.
Insertion of Membrane Proteins into Discoidal Membranes
Journal of Proteome Research • Vol. xxx, No. xx, XXXX
the activation of G proteins, bR has been used as structural
model for rhodopsin and other GPCR family members.16A
purple color and characteristic absorption peaks at 558 and
568 nm (trimer) or at 546 and 553 nm (monomer) indicate a
correctly folded, functional bR,17and that the retinal is bound
to a Lys residue at position 216 via Schiff base formation.18,19
Early reports showed that bR was inactive when synthesized
in vitro, but could be functionally refolded using halobacterial
lipids.19,20More recently, it was shown that active bR could be
produced in vitro; however, a limited fraction of the translated
material could be incorporated into the membranes.21
With in vitro synthesis of bR in the presence of preformed,
naked NLPs, the characteristic purple color could be observed
as early as 5 min after initiation of the reaction (Figure 2a).
This effect was accompanied by a dramatic increase in protein
solubility (nearly 100%) compared to the case where no NLPs
were added and could be observed using NLPs of different size
and composition (see Further Validation of the Approach
below). Under our conditions, bR expressed at levels up to 0.9
mg/mL of the reaction. The assays were performed using NLPs
composed of nontagged apoE422K, which are significantly
larger in size than NLPs composed of apoA1. The resulting bR-
NLP complexes were purified by nickel affinity using the
N-terminal histidine tag attached to bR. Spectral analyses of
the corresponding purified complexes, either exposed to light
or incubated in the dark, exhibited the characteristic bR peaks
yielding a 5 nm shift, which indicates that the majority of the
bR was in a monomeric form17(Figure 2b). This result is in
agreement with other studies that used preformed apolipo-
protein scaffolds to solubilize native forms of bR.22
To determine the molecular weight (MW) and stokes diam-
eter of the bR-NLP complexes, affinity-purified samples were
fractionated using SEC (Table 1, Figure 2c). Size and width
differences found between empty and NLPs thought to contain
bR are consistent with models where one to three molecules
of bR are inserted into each NLP. Note that three bR molecules
Figure 2. Formation and characterization of NLP-bR complexes. (a) Photograph of the reaction tubes where bR was expressed in the
presence (+) or absence (-) of NLPs. NLPs were composed of histidine-tagged apoE422K as a scaffold and DMPC. (b) NLP-bR complexes
were purified over a nickel column and samples were light-adapted (white light for 15 min) or kept in the dark for 4 h. (c) To remove
small particles, purified complexes were concentrated through a 100 kDa cutoff membrane and run over a precalibrated Superdex 200
10/300 GL column. Elution times and biophysical properties of the markers are indicated. Photograph of the tubes of relevant fractions
are depicted. Aliquots of those fractions were run on a native gel and results are shown. MW markers are indicated on the left. (d)
Panels show 350 × 350 nm topographical AFM images of empty (-) and bR-loaded (+) his-apoE422K-NLP complexes obtained in panel
c. Scale bar, 50 nm. A color bar scale identifies NLP height. A section line trace below the two panels shows a cross section of the
height of the particles identified in the insets above. The average height for empty and bR-loaded NLPs is 4.9 ( 0.2 and 6.5 ( 0.3 nm,
respectively. The bottom panels represent scatter plots of NLP diameter and height of the corresponding panels above. The purpose
of AFM studies is to assess particle height only, as the diameter size obtained by AFM is known to show a low x, y resolution due to
tip convolution effects.
Katzen et al.
D Journal of Proteome Research • Vol. xxx, No. xx, XXXX
incorporated into a single NLP complex do not necessarily need
to be grouped in a trimeric form. The extent of the size variation
appears to be governed by the type of NLPs used and might
be influenced by lipid displacement (not studied). Calculated
MW of apoA1-based particles (Table 1) is in agreement with
earlier results where controlled in vitro assembly of apoA1-
based bR-NLPs complexes were reported4,22and correlate well
with the apparent MW estimated from native protein acryla-
mide gel electrophoresis (Figure 2c).
In addition to the bR-NLP complex, the SEC profile also
showed a large peak that coeluted with the column’s void
volume (Figure 2c). This peak, which was also observed in
similar experiments where no DNA and no NLPs were included,
exhibited no apparent protein content as judged by mass
spectrometry analyses (not shown). The above observations
suggest that these fractions are composed primarily of non-
proteinaceous components that might adsorb to the column’s
nickel matrix and coelute with the complexes during the affinity
Height and dispersity of bR-NLP complexes were character-
ized by AFM, which is a well-established technology previously
used to image soluble protein-lipid complexes.23Results
showed that 80-90% of the particles had a height of ap-
proximately 6 nm, which was 1.6 nm larger than empty NLPs
and in agreement with the calculated 5 nm height of bR derived
from the X-ray structure.24These results further support the
notion that the complexes are the product of a bona fide
membrane protein insertion process rather than a consequence
of a membrane association or adsorption artifact.
NLPs Must Be Present during the Translation Process. To
start to investigate the insertion process mechanism, we
performed cell-free reaction assays where NLPs were added
before or after peptide chain elongation was inhibited with
chloramphenicol (cam). Complex formation and protein activ-
ity was observed only when NLPs were added before translation
inhibition. Assays where NLPs were added after the addition
of cam resulted in inactive bR that did not copurify with NLPs
(Figure 3). These observations, together with the fact that under
normal reaction conditions the purple color can be observed
5 min after the reaction has started (see above), suggest that
the insertion of the MP into the NLP lipid bilayers is part of a
Further Validation of the Approach. We further validated
the method by studying the in vitro expression and solubility
of a host of MPs of different topologies, sizes, origins, and
proposed roles (Figure 4a,b). Sixty-four percent of the proteins
analyzed expressed at a level >0.1 mg/mL of the reaction, and
without exception, the proteins significantly increased their
solubility in the presence of NLPs. For the analyzed data set,
the overall solubility increased from 17.3 ( 2.2% to 78.8 ( 3.4%.
Remarkably, all the G protein-coupled receptors (GPCRs),
despite being notoriously difficult to express in heterologous
systems,1showed a similar incrrease in solubility when the
NLPs were added to the reaction. Recent attempts to express
GPCRs in vitro relied on reconstitution into liposomes once
the protein was synthesized and purified.25
In vitro expression in the presence of NLPs produced many
MPs at elevated levels that could be readily visualized on a
coomassie-stained gel. For example, bR expressed at a rate of
0.9 mg/mL and exhibited virtually 100% solubility when NLPs
were added to the reaction (Figure 4c). When DMPC liposomes
were included into the reaction, less than 20% of bR remained
in the supernatant (not shown), consistent with a previous
Table 1. Physical Characterization of Complexes
type of NLP
his-apoE422K empty NLPs
his-apoA1 empty NLPs
aParameters estimated by SEC.
Figure 3. Temporal dependency of membrane protein insertion. (a) bR was expressed in vitro in the presence (+) or absence (-) of
NLPs. All trans-retinal was added in all cases. Reactions were extended for 4 h. Cam (200 µg/mL) and NLPs were added at the indicated
times. When the same time points for both reagents are indicated, cam was added first. NLPs were composed of his-apoA1 as a
scaffold and DMPC. Reactions were trace-labeled using [35S]Met. Complexes were purified by nickel-immobilized affinity chromatography,
separated by SDS electrophoresis, and visualized by autoradiography. C, crude extract; F, flow-through fraction; W, wash fractions; E,
imidazole-eluted fractions. (b) Photograph of the reaction vessels described in panel a. The slightly lighter purple color of the third
sample is due to the use of a shorter translation time (1 h) that results in a somewhat reduced amount of bR synthesized.
Insertion of Membrane Proteins into Discoidal Membranes
Journal of Proteome Research • Vol. xxx, No. xx, XXXX
Figure 4. Cell-free expression and analysis of membrane proteins. (a) Thirty-two membrane proteins of human, mouse, and bacterial
origin were expressed in the presence (filled bars) or absence (empty bars) of NLPs. NLPs were composed of his-apoA1 as a scaffold
and DMPC. Native (nontagged) proteins were expressed from plasmid pEXP4-DEST, with the exception of the product of Z11877 that
was expressed from pEXP6-DEST and was tagged with the Lumio epitope. Reactions were trace-labeled using [35S]Met. Annotated
gene definitions or gene symbols, GenBank accession numbers, and number of predicted transmembrane segments (TMSs) of their
protein products are indicated. Black, red, and green annotations identify the gene source as human, murine, and bacterial, respectively.
Protein yield ranged from 0.05 to 0.4 mg/mL of the reaction. (b) Cell-free protein expression samples of a subset of proteins shown in
panel a (with the exception of the product of Z11877 that was expressed from pEXP4-DEST) were run on an SDS gel and exposed to
an X-ray film. GenBank accession number, predicted MW, predicted number of transmembrane segments, and reaction yield are
indicated. (c) Cell-free protein expression samples of bR (GenBank acc. no. J02755) expressed from plasmid pIVEX2.4b in the absence
(-) or presence (+) of NLPs, trace-labeled with [35S]Met, run on an SDS gel, stained with coomassie blue (left panel), and exposed to
an X-ray film (right panel). bR expressed from this plasmid expresses at a significantly higher level (0.9 mg/mL) than from pEXP4-
DEST in panel A (0.4 mg/mL). NLPs were composed of his-apoA1 and DMPC. Corresponding total crude extract (T) and soluble (S)
fractions were run on a gel, stained with Coomassie blue, and photographed, and an autoradiogram was prepared.
Katzen et al.
F Journal of Proteome Research • Vol. xxx, No. xx, XXXX
report where a small fraction of the translated material where
incorporated into the liposomes.21
We also evaluated the compatibility of eukaryotic in vitro
protein expression systems for use with NLPs. Membrane
proteins (but not water-soluble proteins such as GFP), readily
copurified with histidine-tagged NLP-supplemented rabbit
reticulocyte and wheat germ cell-free protein expression reac-
tions (Figure 5), confirming that our approach works not only
with prokaryotic, but also with eukaryotic in vitro extracts.
Although the E. coli lysate rendered significantly higher protein
yields compared with other lysates (up to 1 mg/mL of the
reaction volume), we found that for particularly large eukaryotic
membrane proteins (>100 kDa) the bacterial extract produced
a small portion of truncated products (for an example, see
Supplementary Figure 1).
As opposed to pancreatic dog microsomes, a membrane
protein insertion platform developed for the rabbit reticulocyte
cell-free protein expression system,26NLPs appear to exhibit
no detrimental effects on the lysate’s translation efficiency
(Supplementary Figure 2). Similarly, the use of preparations
of enriched E. coli membrane vesicles in E. coli based cell-free
protein synthesis systems has been reported.27,28Although this
approach demonstrates successful membrane insertion and
folding, it relies on a relatively polydisperse and undefined
sample and results in products with limited accessibility.
We developed a method for the expression and insertion of
integral membrane proteins into water soluble discoidal lipo-
protein particles. Notable attributes of our approach include
(i) unlike conventional membrane protein expression methods,
proteins are directly inserted into the final membrane support,
avoiding extraction and reconstitution steps; (ii) the process is
rapid (from gene to complex under 6 h), flexible (NLP com-
position can be tailored), high-throughput amenable, compat-
ible with a variety of cell-free lysates, and yields enough product
for functional and structural studies; (iii) the protein may be
expressed in their native state without the addition of any
extraneous sequences; and (iv) the final product is freely
accessible from both sides of the lipid bilayer.
Protein insertion into membranes lacking a translocon has
been reported previously and the insertion mechanism remains
unclear. For example, a number of proteins have been reported
to spontaneously integrate in vitro into naked liposomes and
vesicles.29–32Despite significant progress in demonstrating
protein insertion and export into and across membranes, the
mechanism has been poorly studied at the molecular level. Our
novel in vitro approach opens new alternatives to study the
mechanism of membrane protein insertion and translocation.
For example, the strategy sets the basis for recreating in vitro
the entire membrane protein insertion apparatus by combining
a cell-free protein expression system reconstructed from puri-
fied components33with the recently described bacterial SecYEG
translocon assembled into individual disks.34
There remain some limitations to our method. First, some
membrane proteins appear to need a transmembrane potential
for efficient insertion, which in the absence of compartmen-
talization may not insert appropriately.35Second, the intrinsic
structure of the discs may exhibit limited flexibility to accom-
modate particularly large membrane protein complexes. And
last, because of their intrinsic folding pathways, some proteins
may still need to be subjected to a refolding process once
complexed with the NLPs.
Lipid composition of membranes has a profound effect on
MP topology and activity.36MPs originating from different
organisms, tissues, organelles, or even different membrane
Figure 5. NLP-MP complex formation in eukaryotic cell-free protein expression systems. GFP (a and c) and the human serotonin receptor
(5HT1A) (b and d) were expressed and trace-labeled using rabbit reticulocyte (a and b) and wheat germ-based (c and d) cell-free
protein expression systems in the presence (+) or absence (-) of NLPs. Proteins, devoid of a histidine tag, were expressed from pFKI032
(GFP) and pEXP4-5HT1A. NLPs were composed of his-apoA1 and DMPC. Complexes were purified by nickel-immobilized affinity
chromatography. Proteins were separated by SDS-PAGE and exposed to a film. C, crude extract; FT, flow through fraction; W, wash
fractions; E, imidazole-eluted fractions.
Insertion of Membrane Proteins into Discoidal Membranes
Journal of Proteome Research • Vol. xxx, No. xx, XXXX
domains may require distinct specific environments for proper Download full-text
folding and biological activity. Our system’s flexibility in
permitting the interchange of NLPs of different chemistries
allows the fine-tuning of expression and solubilization condi-
tions for specific MPs to provide near-native context circum-
venting the biophysical constrains of traditional approaches.
Finally, the method is of special relevance to membrane
protein targeting, a developing area of fundamental importance
to multiple disciplines including neuroscience, cancer, and
Acknowledgment. We thank Kenneth Rothschild for
kindly providing p72bop, Karl Weisgraber for pET32-E4NT,
Brent Segelke for providing the E422K protein, Edward Kuhn
for making E422K NLPs, Joe Beechem for his insights,
Sanjay Vasu for reviewing the manuscript, and Rob Bennett
and Claude Benchimol for supporting this project.
Supporting Information Available: Supplementary
Figure 1, expression of hERG in E. coli and rabbit reticulocyte
cell-free protein expression systems; Supplementary Figure 2,
effect of NLPs on protein product yield and solubility. This
material is available free of charge via the Internet at http://
(1) McCusker, E. C.; Bane, S. E.; O’Malley, M. A.; Robinson, A. S.
Heterologous GPCR expression: a bottleneck to obtaining crystal
structures. Biotechnol. Prog. 2007, 23 (3), 40–47.
(2) Bayburt, T. H.; Carlson, J. W.; Sligar, S. G. Reconstitution and
imaging of a membrane protein in a nanometer-size phospholipid
bilayer. J. Struct. Biol. 1998, 123 (1), 37–44.
(3) Bayburt, T. H.; Grinkova, Y. V.; Sligar, S. G. Self-assembly of
discoidal phospholipid bilayer nanoparticles with membrane
scaffold proteins. Nano Lett. 2002, 2 (8), 853–856.
(4) Chromy, B. A.; Arroyo, E.; Blanchette, C. D.; Bench, G.; Benner,
H.; Cappuccio, J. A.; Coleman, M. A.; Henderson, P. T.; Hinz, A. K.;
Kuhn, E. A.; Pesavento, J. B.; Segelke, B. W.; Sulchek, T. A.; Tarasow,
T.; Walsworth, V. L.; Hoeprich, P. D. Different apolipoproteins
impact nanolipoprotein particle formation. J. Am. Chem. Soc. 2007,
129 (46), 14348–14354.
(5) Service, R. F. Materials Research Society meeting. Sushi-like discs
give inside view of elusive membrane proteins. Science 2004, 304
(6) Leitz,A. J.; Bayburt, T. H.; Barnakov, A. N.; Springer, B. A.; Sligar,
S. G. Functional reconstitution of Beta2-adrenergic receptors
utilizing self-assembling Nanodisc technology. BioTechniques
2006, 40 (5), 601-602, 604, 606, passim.
(7) Sonar, S.; Patel, N.; Fischer, W.; Rothschild, K. J. Cell-free synthesis,
functional refolding, and spectroscopic characterization of bac-
teriorhodopsin, an integral membrane protein. Biochemistry 1993,
32 (50), 13777–13781.
(8) Morrow, J. A.; Arnold, K. S.; Weisgraber, K. H. Functional charac-
terization of apolipoprotein E isoforms overexpressed in Escheri-
chia coli. Protein Expression Purif. 1999, 16 (2), 224–230.
(9) Yerushalmi, H.; Mordoch, S. S.; Schuldiner, S. A single carboxyl
mutant of the multidrug transporter EmrE is fully functional.
J. Biol. Chem. 2001, 276 (16), 12744–12748.
(10) Sader, J. E.; Larson, I.; Mulvaney, P.; White, L. R. Method for the
calibration of atomic force microscope cantilevers. Rev. Sci.
Instrum. 1995, 66 (7), 3789–3798.
(11) Elbaz, Y.; Steiner-Mordoch, S.; Danieli, T.; Schuldiner, S. In vitro
synthesis of fully functional EmrE, a multidrug transporter, and
study of its oligomeric state. Proc. Natl. Acad. Sci. U.S.A. 2004, 101
(12) Soskine, M.; Mark, S.; Tayer, N.; Mizrachi, R.; Schuldiner, S. On
parallel and antiparallel topology of a homodimeric multidrug
transporter. J. Biol. Chem. 2006, 281 (47), 36205–36212.
(13) Sikora, C. W.; Turner, R. J. Investigation of ligand binding to the
multidrug resistance protein EmrE by isothermal titration calo-
rimetry. Biophys. J. 2005, 88 (1), 475–482.
(14) Muth, T. R.; Schuldiner, S. A membrane-embedded glutamate is
required for ligand binding to the multidrug transporter EmrE.
EMBO J. 2000, 19 (2), 234–240.
(15) Tate, C. G.; Ubarretxena-Belandia, I.; Baldwin, J. M. Conformational
changes in the multidrug transporter EmrE associated with
substrate binding. J. Mol. Biol. 2003, 332 (1), 229–242.
(16) Taylor, E. W.; Agarwal, A. Sequence homology between bacterio-
rhodopsin and G-protein coupled receptors: exon shuffling or
evolution by duplication. FEBS Lett. 1993, 325 (3), 161–166.
(17) Wang, J.; Link, S.; Heyes, C. D.; El-Sayed, M. A. Comparison of the
dynamics of the primary events of bacteriorhodopsin in its trimeric
and monomeric states. Biophys. J. 2002, 83 (3), 1557–1566.
(18) Mukai, Y.; Kamo, N.; Mitaku, S. Light-induced denaturation of
bacteriorhodopsin solubilized by octyl-beta-glucoside. Protein Eng.
1999, 12 (9), 755–759.
(19) Sonar, S.; Lee, C. P.; Coleman, M.; Patel, N.; Liu, X.; Marti, T.;
Khorana, H. G.; RajBhandary, U. L.; Rothschild, K. J. Site-directed
isotope labelling and FTIR spectroscopy of bacteriorhodopsin. Nat.
Struct. Biol. 1994, 1 (8), 512–517.
(20) Popot, J. L.; Gerchman, S. E.; Engelman, D. M. Refolding of
bacteriorhodopsin in lipid bilayers. A thermodynamically con-
trolled two-stage process. J. Mol. Biol. 1987, 198 (4), 655–676.
(21) Kalmbach, R.; Chizhov, I.; Schumacher, M. C.; Friedrich, T.;
Bamberg, E.; Engelhard, M. Functional cell-free synthesis of a
seven helix membrane protein: in situ insertion of bacteriorho-
dopsin into liposomes. J. Mol. Biol. 2007, 371 (3), 639–648.
(22) Bayburt, T. H.; Grinkova, Y. V.; Sligar, S. G. Assembly of single
bacteriorhodopsin trimers in bilayer nanodiscs. Arch. Biochem.
Biophys. 2006, 450 (2), 215–222.
(23) Carlson, J. W.; Jonas, A.; Sligar, S. G. Imaging and manipulation
of high-density lipoproteins. Biophys. J. 1997, 73 (3), 1184–1189.
(24) Luecke, H.; Schobert, B.; Richter, H. T.; Cartailler, J. P.; Lanyi, J. K.
Structural changes in bacteriorhodopsin during ion transport at
2 angstrom resolution. Science 1999, 286 (5438), 255–261.
(25) Klammt, C.; Schwarz, D.; Eifler, N.; Engel, A.; Piehler, J.; Haase,
W.; Hahn, S.; Dotsch, V.; Bernhard, F. Cell-free production of G
protein-coupled receptors for functional and structural studies. J.
Struct. Biol. 2007, 158 (3), 482–493.
(26) Walter, P.; Blobel, G. Preparation of microsomal membranes for
cotranslational protein translocation. Methods Enzymol. 1983, 96,
(27) Kuruma, Y.; Nishiyama, K.; Shimizu, Y.; Muller, M.; Ueda, T.
Development of a minimal cell-free translation system for the
synthesis of presecretory and integral membrane proteins. Bio-
technol. Prog. 2005, 21 (4), 1243–1251.
(28) Wuu, J. J; Swartz, J. R. High yield cell-free production of integral
membrane proteins without refolding or detergents. Biochim.
Biophys. Acta 2008, 1778, 1237–1250.
(29) Geller, B. L.; Wickner, W. M13 procoat inserts into liposomes in
the absence of other membrane proteins. J. Biol. Chem. 1985, 260
(30) Klammt, C.; Schwarz, D.; Lohr, F.; Schneider, B.; Dotsch, V.;
Bernhard, F. Cell-free expression as an emerging technique for
the large scale production of integral membrane protein. FEBS J.
2006, 273 (18), 4141–4153.
(31) Shimizu, Y.; Kuruma, Y.; Ying, B. W.; Umekage, S.; Ueda, T. Cell-
free translation systems for protein engineering. FEBS J. 2006, 273
(32) Noireaux, V.; Libchaber, A. A vesicle bioreactor as a step toward
an artificial cell assembly. Proc. Natl. Acad. Sci. U.S.A. 2004, 101
(33) Shimizu, Y.; Inoue, A.; Tomari, Y.; Suzuki, T.; Yokogawa, T.;
Nishikawa, K.; Ueda, T. Cell-free translation reconstituted with
purified components. Nat. Biotechnol. 2001, 19 (8), 751–755.
(34) Alami, M.; Dalal, K.; Lelj-Garolla, B.; Sligar, S. G.; Duong, F.
Nanodiscs unravel the interaction between the SecYEG channel
and its cytosolic partner SecA. EMBO J. 2007, 26 (8), 1995–2004.
(35) Andersson, H.; von Heijne, G. Membrane protein topology: effects
of delta mu H+ on the translocation of charged residues explain
the ’positive inside′ rule. EMBO J. 1994, 13 (10), 2267–2272.
(36) Bogdanov, M.; Heacock, P. N.; Dowhan, W. A polytopic membrane
protein displays a reversible topology dependent on membrane
lipid composition. Embo J 2002, 21 (9), 2107–2116.
(37) Feldman, G.; Bogoev, R.; Shevirov, J.; Sartiel, A.; Margalit, I.
Detection of tetracysteine-tagged proteins using a biarsenical
fluorescein derivative through dry microplate array gel electro-
phoresis. Electrophoresis 2004, 25 (15), 2447–2451.
Katzen et al.
HJournal of Proteome Research • Vol. xxx, No. xx, XXXX