Preparative scale cell-free expression systems: new tools for the large scale preparation of integral membrane proteins for functional and structural studies.

Abstract

Cell-free expression techniques have emerged as promising tools for the production of membrane proteins for structural and functional analysis. Elimination of toxic effects and a variety of options to stabilize the synthesized proteins enable the synthesis of otherwise difficult to obtain proteins. Modifications in the reaction design result in preparative scale production rates of cell-free reactions and yield in milligram amounts of membrane proteins per one millilitre of reaction volume. A diverse selection of detergents can be supplied into the reaction system without inhibitory effects to the translation machinery. This offers the unique opportunity to produce a membrane protein directly into micelles of a detergent of choice. We present detailed protocols for the cell-free production of membrane proteins in different modes and we summarize the current knowledge of this technique. A special emphasize will be on the production of soluble and functionally folded membrane proteins in presence of suitable detergents. In addition, we will highlight the advantages of cell-free expression for the structural analysis of membrane proteins especially by liquid state nuclear magnetic resonance spectroscopy and we will discuss new strategies for structural approaches.

Full text

Page 1

Methods 41 (2007) 355–369
www.elsevier.com/locate/ymeth
1046-2023/$ - see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.ymeth.2006.07.001
Preparative scale cell-free expression systems: New tools
for the large scale preparation of integral membrane
proteins for functional and structural studies
Daniel Schwarz, Christian Klammt, Alexander Koglin, Frank Löhr,
Birgit Schneider, Volker Dötsch, Frank Bernhard¤
Centre for Biomolecular Magnetic Resonance, University of Frankfurt/Main, Institute for Biophysical Chemistry,
Max-von-Laue-Str. 9, D-60438 Frankfurt/Main, Germany
Accepted 13 July 2006
Abstract
Cell-free expression techniques have emerged as promising tools for the production of membrane proteins for structural and func-
tional analysis. Elimination of toxic eVects and a variety of options to stabilize the synthesized proteins enable the synthesis of otherwise
diYcult to obtain proteins. ModiWcations in the reaction design result in preparative scale production rates of cell-free reactions and yield
in milligram amounts of membrane proteins per one millilitre of reaction volume. A diverse selection of detergents can be supplied into
the reaction system without inhibitory eVects to the translation machinery. This oVers the unique opportunity to produce a membrane
protein directly into micelles of a detergent of choice. We present detailed protocols for the cell-free production of membrane proteins in
diVerent modes and we summarize the current knowledge of this technique. A special emphasize will be on the production of soluble and
functionally folded membrane proteins in presence of suitable detergents. In addition, we will highlight the advantages of cell-free expres-
sion for the structural analysis of membrane proteins especially by liquid state nuclear magnetic resonance spectroscopy and we will dis-
cuss new strategies for structural approaches.
© 2006 Elsevier Inc. All rights reserved.
Keywords: Integral membrane proteins; Structural analysis; Solubilization; Reconstitution; Detergent; NMR spectroscopy; G-protein coupled receptors;
?-barrel proteins; Multidrug resistance; Cell-free expression; Detergent micelles; Transporter; Liposomes; Labelling of proteins; Stable isotopes
1. Introduction
Membrane proteins (MPs) deWne the link between phys-
iological pathways in the cytoplasm and the extracellular
environment. Essential processes like perception and trans-
duction of external signals, import or export of substances
through the membrane or the generation of energy is asso-
ciated with MPs. Many modern drugs are directed against
MPs and this class of proteins is therefore an important tar-
get for medical and pharmaceutical research. However,
high resolution structures of MPs are still the very excep-
tion. A major bottleneck for structural analysis is the lim-
ited availability of suYcient amounts of protein samples.
The bacterial Escherichia coli expression system is most fre-
quently used for the production of recombinant proteins.
Its simplicity, low costs, the wealth of elaborated protocols,
the fast growth rates and often high productivity makes this
system highly competitive against most other expression
systems. However, in vivo expression systems based on pro-
karyotic or eukaryotic cells do not work for a wide range of
MPs, toxins or other problematic targets [1]. MPs often
aVect the physiology of the cell by insertion into the cellular
membranes or by blocking protein traYcking systems. Low
expression rates, aggregation or unfolding of the recombinant
*Corresponding author. Fax: +49 69 798 29632.
E-mail address: fbern@bpc.uni-frankfurt.de (F. Bernhard).

Page 2

356
D. Schwarz et al. / Methods 41 (2007) 355–369
MPs and even toxic eVects to the host cells upon overpro-
duction are therefore frequent problems when MPs have to
be produced [2].
Cell-free (CF)1 protein synthesis provides a recently
developed and powerful alternative tool for protein pro-
duction [3–5]. A unique advantage of CF systems is the
open access to the reaction at any time of the experiment.
This enables the addition of beneWcial compounds at
deWned concentrations and at any stage of the protein syn-
thesis. No membranes have to be penetrated and no selec-
tion of substances occurs by speciWc transport systems.
Metabolic conversion or even degradation of added sub-
stances is furthermore reduced due to the restricted enzy-
matic activity of the CF extracts. The only limitation is that
the supplemented compounds must not aVect the transcrip-
tion and translation machinery of the expression system.
Degradation of proteins or nucleic acids could be prevented
by addition of corresponding inhibitor cocktails. Addition
of chaperones like the GroEL/ES or DnaK/J-GrpE systems
could facilitate the folding eYciency of heterologous pro-
teins. SpeciWc cofactors, substrates or inhibitors might help
to stabilize the synthesized recombinant proteins. The addi-
tion of detergents or lipids enables the direct translation of
MPs into deWned hydrophobic environments. DisulWde
bridges that are essential for the functional folding of many
eukaryotic proteins are likely to become formed due to the
easy access of oxygen to the CF reaction [6–8].
The elimination of cytotoxic eVects is presumably one of
the major reasons for the rapidly increasing number of
diverse MPs of prokaryotic and eukaryotic origin that can
be produced in CF systems [9–14]. While still limited, the
reported examples of CF produced MPs already comprises
?-helical and ?-barrel type MPs of prokaryotic as well as of
eukaryotic origins. One common characteristic is their
almost complete membrane integrated topology with only
small proposed external loop regions. Especially MPs that
will not or only at minor levels be synthesized in living E.
coli cells might therefore become targets for CF expression.
This feature highlights CF expression as a promising future
technique for the high level production of otherwise diY-
cult to obtain MPs. Moreover, additional beneWcial charac-
teristics like various options to protect and to stabilize
recombinant proteins, the possibility to translate MPs into
preformed micelles and the considerable advantages upon
speciWc labelling approaches of proteins make CF expres-
sion as one of the currently most versatile techniques for
the preparative scale production of proteins.
This review summarizes the preparative scale production
of MPs in CF systems based on E. coli extracts. We will
emphasize rather on the set-up of individual CF expression
systems than on commercially available systems. To our
knowledge, there are currently no reported examples of the
preparative scale production of MPs in CF systems based
on wheat germ extracts. However, these systems might play
an important role in the near future, especially for the syn-
thesis of functionally folded eukaryotic MPs. Therefore, a
short overview about the key steps in wheat germ extract
preparation and in the reaction design will be provided.
2. Preparative scale cell-free expression systems
2.1. ConWguration and productivity of CF-systems
First generation CF expression systems have been batch-
formatted reactions containing all compounds in one com-
partment. The rapid depletion of precursors in combination
with the accumulation of inhibitory breakdown products
resulted in short reaction times of less than 1h and conse-
quently in only low product yields of often not more than
several micrograms of recombinant protein per one ml of
reaction [15]. CF expression systems have therefore been
used for a long time only for the analytical scale production
of proteins. The splitting of the CF system into a reaction
mixture (RM) holding all high molecular weight com-
pounds and into a feeding mixture (FM) containing the low
molecular precursors provided the basis for several new
reaction designs with considerably improved eYciencies
[16]. A common characteristic of preparative scale CF
expression systems is the extended supply of fresh precur-
sors combined with the continuous removal of deleterious
reaction by-products like pyrophosphate. The reaction
times are extended up to approx. 20h and allow the synthe-
sis of several milligram amounts of protein per 1ml of RM
[17,18]. A frequently used reaction design for the high level
production of proteins is the continuous-exchange cell-free
(CECF) system [16]. The Wxed volume compartments of
1Abbreviations: AcP, acetyl phosphate; Brij35, polyoxyethylene-(23)-
lauryl-ether; Brij56, polyoxyethylene-(10)-cetyl-ether; Brij58, polyoxyeth-
ylene-(20)-cetyl-ether; Brij72, polyoxyethylene-(2)-stearyl-ether; Brij78,
polyoxyethylene-(20)-stearyl-ether; Brij97, polyoxyethylene-(10)-oleyl-
ether; Brij98, polyoxyethylene-(20)-oleyl-ether; CECF, continuous-ex-
change cell-free; CF, cell-free; CHAPS, 3-[(3-cholamidopropyl)dimethyl
ammonio]-1-propansulfonat; CMC, critical micellar concentration; C12E8,
polyoxyethylene-(8)-lauryl-ether; DHPC, 1,2-diheptanoyl-sn-glycero-3-
phosphocholine; diC6PC, 1,2-dihexanoyl-sn-glycero-3-phosphocholine;
diC8PC, 1,2-dioctanoyl-sn-glycero-3-phosphocholine; DDM, n-dodecyl-?-
D-maltoside; DM, n-decyl-?-D-maltoside; DMPC, di-myristoyl-phosphati-
dyl-choline; DPC, dodecyl-phosphocholine; FID, free induction decay;
FM, feeding mixture; Genapol C 100, polyoxyethylene-(10)-dodecyl-ether;
Genapol X 100, polyoxyethylene-(10)-isotridecyl-ether; GPCR, G-protein
coupled receptor; HECAMEG, (6-O-(N-heptylcarbamoyl)-methyl-?-D-
glucopyranoside); HSQC, heteronuclear single quantum correlation;
LMPG, 1-myristoyl-2-hydroxy-sn-glycerol-3-[phosphor-rac-(1-glycerol)];
LPPG, 1-palmitoyl-2-hydroxy-sn-glycerol-3-[phosphor-rac-(1-glycerol)];
MM, master mixture; MP, membrane protein; MWCO, molecular weight
cut-oV; NG, n-nonyl-?-D-glucoside; NMR, nuclear magnetic resonance;
NP40, nonylphenyl-polyethylene-glycol; NTP, nucleotide triphosphate; ?-
OG, n-octyl-?-D-glucopyranoside; OMP, outer membrane protein; PCR,
polymerase chain reaction; PEP, phosphoenol pyruvate; RM, reaction
mixture; SDS, sodium-dodecyl-sulfate; TB, terriWc broth; Thesit, polyeth-
ylene-glycol 400 dodecylether; TMS, transmembrane segment; Triton
X-100, polyethylene-glycol P-1,1,3,3-tetra-methyl-butylphenyl-ether;
TROSY, transverse relaxation optimized spectroscopy; Tween20, poly-
oxyethylene-sorbitan-monolaurate 20; Tween40, polyoxyethylene-sorbi-
tan-monopalmitate 20; Tween60, polyoxyethylene-sorbitan-monostearate
20; Tween80, polyoxyethylene-sorbitan-monoleate 20; UTR, untranslated
region.

Page 3

D. Schwarz et al. / Methods 41 (2007) 355–369
357
RM and FM are separated by a semipermeable membrane
with molecular weight cut-oVs (MWCOs) between 10 and
50kDa that ensures an eYcient exchange of compounds
(Fig. 1). The individual components of the translation
machinery obviously stick together in a macromolecular
complex and despite the relatively high MWCO of the
membrane, no signiWcant leakage of translation factors is
noticed. The reaction is incubated with intensive agitation
like stirring, shaking or rotating in order to provide an opti-
mal exchange between the two compartments. Commercial
CECF systems (Rapid Translation System (RTS) Roche
Diagnostics, Penzberg, Germany) as well as individually
prepared systems are highly productive [12,13,17]. How-
ever, it should be considered that several parameters of the
reaction like ion concentrations, the composition of the
energy system and the amino acid pool or even the buVer
system can be subject of intensive optimization steps before
the high level expression of a new protein target is achieved.
CF lysates are mostly prepared from E. coli cells, wheat
germs and to a lesser extent from rabbit reticulocytes
[4,19,20]. Eukaryotic backgrounds might be preferred for
the expression of eukaryotic proteins to provide the opti-
mal environment for their functional folding and to enable
posttranslational modiWcations. Low levels of endogenous
mRNAs in wheat germ extracts allow their use in pure
translation systems with added mRNA as template for
translation [20]. Extended reaction times up to 60h can
yield in 1–4mg of recombinant protein per 1ml RM [21].
Most popular is the S30 extract of E. coli that contains the
soluble fraction of cell lysates after centrifugation at
30,000g and including all enzymes necessary for transcrip-
tion and translation [4]. CF expression systems based on E.
coli extracts are almost exclusively used as coupled tran-
scription/translation systems by providing DNA as tem-
plate in combination with the highly speciWc and eYcient
T7-RNA polymerase [22].
For speciWc purposes, the E. coli translation machinery
could alternatively be reconstituted almost completely
in vitro by combining the individually puriWed protein com-
ponents to isolated ribosomes [23]. All aminoacyl-tRNA-
synthetases and translation factors can be overproduced
separately in standard E. coli expression systems, puriWed
by virtue of terminal poly(His)6-tags and added to E. coli
S100 extracts containing the relatively pure ribosome frac-
tion. This PURE system (protein synthesis using recombi-
nant elements) enables the CF protein synthesis under
deWned conditions and allows detailed studies of folding
pathways or translation kinetics.
Fig. 1. Schematic conWguration of a coupled transcription/translation reaction in a CECF system. The CF reaction can be carried out in simple dialysis
tubes placed into suitable plastic vials that hold the FM. The complete set-up is incubated e.g., on a turning device that ensures continuous agitation of the
reaction and substance exchange between the two compartments.

Page 4

358
D. Schwarz et al. / Methods 41 (2007) 355–369
2.2. Preparation of cell-free lysates
CF lysates provide all the high molecular weight compo-
nents of the translation machinery. Endogenous low molec-
ular weight substances like amino acids and salts will be
removed by extensive dialysis during the preparation pro-
cedure. A “run-oV” step is furthermore implemented in
order to eliminate endogenous cellular mRNA. High salt
concentrations cause the dissociation of the ribosomes
from endogenous mRNA that will then subsequently
become degraded due to the high RNAse content of the
extract.
Escherichia coli extracts are relatively easy and fast to
prepare and the individual steps of standard protocols
include cell fermentation, cell disruption, run-oV procedure
and buVer exchange by dialysis [4,24]. Common sources for
CF extracts are E. coli BL21 derivatives or strains devoid of
major endogenous RNAses like A19 (Table 1). The cells
have to be grown with good aeration until mid-log phase at
37°C in rich medium like terriWc broth (TB), chilled down
rapidly and harvested by centrifugation. The time of har-
vest is somehow crucial and corresponds in TB medium to
an OD595 of approximately 3.5. Exceeding the optimal time
point of harvest can drastically reduce the eYciency of the
Wnal CF extract. Rapid chilling of the culture down to
below 10°C upon harvesting stalls further growth and con-
serves the active state of the cellular physiology. A 10L fer-
menter with TB medium should yield 50–70g wet-weight of
bacterial cells. The cell pellet is resuspended and washed
three times in ice cold S30-A buVer and it is Wnally sus-
pended in S30-B buVer pre-cooled at 4°C (Table 1). The
cells should be disrupted by passing through a pre-cooled
French-Press and not by soniWcation, as this treatment
could cause the disintegration of ribosomes. Cell-debris is
removed by centrifugation at 30,000g at 4°C for 30min and
the upper 2/3rd of the supernatant are transferred into a
fresh vial. The centrifugation step and transfer of superna-
tant is repeated once. For the “run oV” step the lysate is
adjusted to a Wnal concentration of 400mM NaCl and
incubated at 42°C for 45min in a water bath [12]. Besides
the elimination of endogenous mRNA, this treatment
causes a considerable precipitate. The turbid solution is
Wlled into a dialysis tube (MWCO 14kDa) and dialyzed at
4°C against 60 volumes of S30-C buVer with gentle stirring.
After one further exchange of the dialysis buVer the E. coli
S30-extract is harvested by centrifugation at 30,000g at 4°C
for 30min. The clear supernatant is transferred in suitable
aliquots into plastic tubes and frozen in liquid nitrogen.
The Wnal protein concentration in the extract should be
between 30–50mg/ml and could be adjusted by ultraWltra-
tion. The complete protocol should yield some 50ml of CF
extract out of a 10L fermentation. We recommend to Wnish
Table 1
Materials, buVers and substances for the cell-free expression of membrane proteins
Source for S30 E. coli lysates:
A19 [rna19 gdh A2 his95 relA1 spoT1 metB1] E. coli Genetic Stock Center (E. coli Genetic Stock Center, New Haven, USA, CGSC No. 5997)
BL21 star [F-ompT hsdS B (rB–mB-) gal dcm rne131] Invitrogen, Karlsruhe, Germany, C6010-03
Materials for S30 extract preparation:
Fermenter
French Press cell disruption device
Dialysis tubes (MWCO 14kDa)
Devices for protein concentration by ultraWltration
CF reaction container:
Microdialysers (MWCO 15–25kDa) (Spectrum Labs, Rancho Dominguez, USA)
Dispodialysers (MWCO 15–25 kDa) (Spectrum Labs, Rancho Dominguez, USA)
Shaking, rolling or stirring device; e.g. Universal Turning Device (Vivascience, Göttingen, Germany; Cat. No. IV-76001061) placed in an incubator
with temperature control
Standard glass or plastic vials
Chemicals for S30 extract preparation and set-up for cell-free expression:
Adenosine-5?-triphosphate disodium salt, acetyl-phosphate, amino acids (Sigma–Aldrich, Taufkirchen, Germany), complete mini protease inhibitor
(Roche Diagnostics GmbH, Mannheim, Germany), cytidine 5?-triphosphate disodium salt, dithiothreitol, ethylenediamine-tetraaceticacid, folinic acid,
guanosine 5?-triphosphate disodium salt, HEPES, KCl, KOAc, Mg(OAc)2, liquid N2, NaN3, NaCl, ?-mercaptoethanol, phenylmethane-
sulfonylXuoride, phosphoenol-pyruvate, polyethyleneglykol 8000, pyruvate kinase, RNAsin (Amersham Biosciences, Freiburg, Germany), total E. coli
tRNA (Roche Diagnostics GmbH, Mannheim, Germany), T7-RNA polymerase, uridine 5?-triphosphate trisodium salt
Selected detergents:
?-[4-(1,1,3,3-tetramethylbutyl)-phenyl]-?-hydroxy-poly(oxy-1,2-ethandiyl) (Triton X-100), digitonin, polyethylenglycododecylether (Brij35),
polyoxyethylene-(20)-cetyl-ether (Brij58); polyoxyethylene-(20)-stearyl-ether (Brij78); polyoxyethylene-(20)-oleyl-ether (Brij98); (all Sigma–Aldrich,
Taufkirchen, Germany); n-dodecylphosphocholine (DPC), 1-myristoyl-2-hydroxy-sn-glycero-3-[phosphor-rac-(1-glycerol)] (LMPG), (Avanti-Lipids,
Alabaster, USA); n-dodecyl-?-D-maltoside (DDM); (Glycon Biochemicals, Luckenwalde, Germany)
BuVers/solutions:
S30-A buVer: 10mM Tris–acetate, pH 8.2, 14mM Mg(OAc)2, 0.6mM KCl, 6mM ?-mercaptoethanol.
S30-B buVer: 10mM Tris–acetate, pH 8.2, 14mM Mg(OAc)2, 0.6mM KCl, 1mM DTT, 0.1mM phenylmethane-sulfonylXuoride.
S30-C buVer: 10mM Tris–acetate, pH 8.2, 14mM Mg(OAc)2, 0.6mM KOAc, 0.5mM DTT. TB-medium (per litre): 24g yeast extract, 12g tryptone,
4ml 100% glycerol, 100mM potassium phosphate buVer, pH 7.4

Page 5

D. Schwarz et al. / Methods 41 (2007) 355–369
359
the complete process consecutively but the interruption
after cell harvesting is also possible.
Expression platforms based on eukaryotic cell extracts
provide a higher stability of mRNA with reaction times of
several days [21]. However, Wnal yields of recombinant pro-
tein are still similar if compared to CF systems based on E.
coli extracts. A major disadvantage is the relatively compli-
cated and time-consuming extract preparation procedure
and high variations in the quality of diVerent extract
batches. Only a short overview of key steps in wheat germ
extract preparation is given below and more extended pro-
tocols are available in the literature [21,25].
Winter wheat is preferred as source for extract prepara-
tion. Fractions containing wheat embryos are isolated from
ground seeds by Xotation in a mixture of carbon tetrachlo-
ride and cyclohexane. MagniWer lenses have to be used to
remove damaged embryos and distinct parts of endosperm
that will inhibit translation by presence of various inhibi-
tory proteins like e.g., ricin [26]. This step is highly critical
and it is the most laborious part. The puriWed embryos are
ground to a Wne powder in liquid nitrogen, resolubilized in
extraction buVer (40mM Hepes–KOH, pH 7.6, 100mM
KOAc 5mM Mg(OAc)2, 2mM CaCl2, 4mM DTT, 0.3mM
of each of the 20 amino acids) and pelleted by centrifuga-
tion. The supernatant is applied on a PD-10 column pre-
equilibrated with extraction buVer to remove inhibitory low
molecular weight substances from the extract. The extract is
Wnally concentrated to an A280 of at least 200/ml and stored
in aliquots at ¡80°C [27].
2.3. Design of DNA templates for cell-free expression
The transcription in E. coli coupled transcription/trans-
lation CF systems is operated by the phage T7-RNA poly-
merase. The puriWed enzyme has to be added into the RM
at relatively high Wnal concentrations between 4–10U/?l
and detailed protocols for the overproduction and puriWca-
tion of the T7-RNA polymerase have been published [28].
The promotor elements of the target gene have to meet the
speciWc requirements of the T7-RNA polymerase. The ribo-
somal binding site has to be present in optimal distance to
the translational start codon and a transcriptional termina-
tor should be placed 3? to the reading frame to prevent
excessive consumption of NTP precursors. Some suitable
commercial vector series are pIVEX (Roche Diagnostics,
Penzberg, Germany), pDEST (Invitrogen, Carlsbad, USA)
or pET (MerckBioscience, Darmstadt, Germany). These
vector systems oVer furthermore the option to add a variety
of terminal tags to the target protein that might increase the
protein expression or that could facilitate puriWcation and
detection strategies.
The translation eYciency of mRNA templates in wheat
germ systems strongly depends on the 5?- and 3?-untrans-
lated regions (UTRs) [29]. Eukaryotic mRNAs are
modiWed after translation with a 5? cap (5?7 mGpppG) and
a 3?-polyadenylated tail that prevents degradation and
results in a better protein expression. The supply of pre-
modiWed mRNAs to CF reactions is useless because of the
eYcient deadenylation and decapping activity of extract
enzymes and higher concentrations of pre-modiWed
mRNAs are even inhibitory for the translation. This prob-
lem could be addressed by several strategies like the use of
phosphothionate mRNA [30], the modiWcation of the 3?-
end of mRNA with adaptor DNA [31] and the immobiliza-
tion of mRNA on latex beads [32]. Alternatively, diVerent
UTRs could be used. Viral UTRs are generally good trans-
lation enhancers in the case of the cap-independent initia-
tion. EYcient 5?-UTRs usually contain an increased A/T
content including motifs such as (AAC)n, (AAAC)n or
(AAAAC)n. Most popular is currently the OMEGA leader
derived from the tobacco mosaic virus [33]. The 3?-UTRs
stabilize mRNA by the formation of complex secondary
structures and also viral 3?-UTRs are frequently used [34].
2.4. Linear DNA as a template for cell-free expression
The possibility to use linear templates generated by PCR
in the CF-system eliminates time consuming cloning/sub-
cloning steps and allows the rapid screening of a variety of
expression constructs [5]. PCR products can furthermore be
directly used for CF expression without prior puriWcation
[35]. Multiple-step PCR protocols or the split primer PCR-
technique have to be employed to add the required relatively
long regulatory elements like T7-promoter and terminator
to the coding sequence. Several strategies have been estab-
lished in order to overcome the degradation of linear DNA
templates by endogenous nucleases. The Lambda phage
Gam protein is an inhibitor of exonuclease V (recBCD) and
its addition to the CF reaction stabilized linear DNA [36].
Extract preparation from an engineered E. coli strain devoid
of the endA gene and containing a modiWed recBCD operon
resulted in protein yields comparable to that obtained with
plasmid templates in batch systems [37]. Stem loop struc-
tures at the 3?-end of mRNAs also help to reduce exonucle-
ase mediated degradation [38]. CF expression is predestined
for high throughput (HT) applications especially by usage
of PCR generated DNA templates [38–40].
2.5. Reaction conditions of E. coli cell-free expression
systems
CF expression can be performed in small analytical scale
reactions with approximately 50–100?l RM for optimiza-
tion reactions and in larger preparative scale reactions of
1–2ml RM for the production of protein. It should be con-
sidered that for almost each new protein target distinct
parameters like the concentrations of critical compounds or
the generation of an optimal expression construct might
need to be optimized in order to Wnd the best conditions for
high level expression. The CECF reaction has a very well
deWned optimum for Mg2+ and K+ ions usually between
13–15 and 280–300mM, respectively. For each new protein
target and also for each new batch of extract at least the
concentrations of these two compounds should be

Page 6

360
D. Schwarz et al. / Methods 41 (2007) 355–369
optimized for best expression rates. A RM/FM ratio
between 1:10 and 1:20 for analytical scale reactions and
commercially available microdialysers (Spectrum Labora-
tories, Rancho Dominguez, USA) with a MWCO between
15 and 25kDa as a reaction device can be recommended.
Expression levels of several milligrams of protein per 1ml
RM can only be obtained if all system components are in
optimal conditions. The results can be scaled up into pre-
parative reactions of 1–2ml RM without signiWcant loss of
eYciency. Dispodialyser (Spectrum Laboratories, Rancho
Dominguez, USA) or even simple dialysis tubes are recom-
mended as preparative scale reaction containers (Fig. 1).
All stock solutions (Table 2) should be mixed carefully
after thawing and the enzymes, tRNA and the S30 extract
should be kept on ice. A master-mix (MM) including all
shared components of FM and RM should be prepared by
Wrst pipetting the higher volume components (e.g. KOAc;
MgO(Ac)2; PEG8000; NaN3). First the FM is completed
and pre-incubated in a water-bath at 30°C. Then the RM is
completed, mixed gently and kept on ice. The appropriate
volume of FM is Wlled into standard plastic or glass vials
that can be used as FM compartments. The RM is trans-
ferred into suitable dispodialyser or washed dialysis tubes
(Fig. 1). Air-bubbles that might restrict an eYcient
exchange between the two compartments should be
avoided. The reaction has to be incubated with intensive
agitation either on rolling or shaking devices or by mag-
netic stirring. The incubation temperature is usually
between 20°C and 30°C and protein synthesis continues up
to 20h (Fig. 2). A RM/FM ratio of 1:17 can already result
in the production of several milligrams of protein per 1ml
RM, but the Wnal yields of recombinant protein per reac-
tion can easily be increased by using higher ratios or by
refreshing the FM after certain times of incubation.
Due to their low internal RNAse content, CF reactions
based on wheat germ extracts can be operated as translation
system with supplied mRNA as a template. Translation in a
Table 2
Standard protocol for an individual continuous exchange cell-free reaction
Approx. 140mM K+ are provided from other components (e.g. PEP, KOH), approx. 5mM Mg2+ are provided from other components (e.g. extract) and
Wnal concentrations might be adjusted according to the amino acid composition of the target protein.
SubstanceStock solutionFinal concentration Notes, references or suggested supplier
RM
E. coli tRNA
Pyruvate kinase
T7-RNA-polymerase
RNAguard porcine
Plasmid vector
40mg/ml
10mg/ml
40U/?l
40U/?l
0.15mg/ml
0.5mg/ml
0.04mg/ml
6U/?l
0.3U/?l
0.015mg/ml
In distilled water; (Roche Diagnostics, Mannheim, Germany; No. 109550)
(Roche Diagnostics GmbH, Mannheim, Germany; No. 109550)
overproduction in E. coli [28]
(Amersham Biosciences, Freiburg, Germany; No. AP27-0816-01)
T7 promoter regulatory region (e.g. pET vector series; Merck Biosciences,
Darmstadt, Germany)
(modiWed after: [4,12])
E. coli S30 extract 100%35%
RM+FM
NTP-Mix ATP: 360mM
CTP, GTP, UTP:
240mM each
500mM
10mg/ml
40%
10%
4mM each
ATP: 1.2mM
CTP, GTP, UTP:
0.8mM each
2mM
0.1mg/ml
2%
0.05%
1mM each
pH 7.0 with NaOH
DTT
Folinic acid
PEG 8.000
Sodium azide
20 amino acid mix
Depends on desired conditions
Ca2+ salt
Dissolved at 30°C in distilled water
In distilled water
Made from individual 100mM stocks in distilled water, tyrosine as 20mM
stock in distilled water, tryptophan as 100mM stock in 100mM Hepes, pH
8.0 (remains turbid).
Arginine, cysteine, tryptophan, methionine, aspartic acid and glutamic
acid are limiting due to instability [15,45]
In distilled water (Roche Diagnostics, Mannheim, Germany; No. 1836153)
K+ salt in distilled water, pH 7.0, with KOH
K+ salt in distilled water, pH 7.0, with KOH
pH 8.0, with KOH
Subject of optimization
Subject of optimization
RCWMDE-Mix16.7mM each1.0mM each
Complete protease inhibitor
Acetyl phosphate
PEP
Hepes buVer
KOAc
Mg(OAc)2
50-fold
1M
1M
2.5M
4M
1M
1-fold
20mM
20mM
100mM
280–300mM
13–15mM
Fig. 2. Kinetics of cell-free production of the nucleoside transporter Tsx.
The diagram shows typical kinetics of protein production in a preparative
scale CECF system. The nucleoside transporter Tsx was synthesized in a
1ml CECF standard reaction without detergent over a time period of 21h
at 30°C.

Page 7

D. Schwarz et al. / Methods 41 (2007) 355–369
361
CECF system with wheat germ extract can continue up to
several days. In a standard protocol, the Wnal concentration
of the components in the RM are: Hepes–KOH (pH 8.0)
40mM, amino acids 0.2mM each, Mg(OAc)2 3.0mM, glyc-
erine 2%, KOAc 80mM, ATP 1mM, GTP/CTP/UTP
0.8mM each, spermidin 0.15mM, NaN3 0.03%, creatine-
phosphate 16mM, mRNA 250pmol/ml, creatinephosphate
kinase 0.1mg/ml, RNAse inhibitor 0.5U/ml, wheat germ
extract at 30% of the RM volume [41]. The composition of
the FM is identical with the exception that the high-molec-
ular weight compounds mRNA, creatinephosphate kinase,
RNAse inhibitor and wheat germ extract are omitted. The
reaction is incubated at 25°C with intensive agitation. The
Wnal concentrations of Mg2+ and K+ ions are subject of
optimization and they may vary in the range between 1.5–
3.5 and 60–120mM, respectively. The optimal concentra-
tion of mRNA furthermore depends on the speciWc UTR’s
and coding sequences [42]. Short expression times in wheat
germ systems are mainly attributed to an increased hydro-
lysis of NTP precursors. Supplying Cu(OAc)2 helps to
reduce endogenous ATPase activity and can successfully
prolong the expression period [43,44].
2.6. Perspectives for the optimization of cell-free expression
systems
The set-up of an eYcient CECF reaction is relatively
complicated and requires some experience. One focus of
further improvements is therefore an increased eYciency of
the easier to handle batch format CF systems. The exact
adjustment of critical ion concentrations and the increased
supply of some rapidly degraded amino acids (e.g. arginine,
cysteine, tryptophan, methionine, aspartic acid and glu-
tamic acid) considerably improved the translation eYciency
and extended reaction times [45]. More eYcient NTP regen-
eration systems can yield almost milligram amounts of pro-
tein per 1ml reaction in bacterial and wheat germ batch
systems [38,46]. Typical energy sources for the regeneration
of ATP in CECF systems are phosphoenol pyruvate (PEP)
[4], creatine phosphate [47] and acetyl phosphate (AcP) [48]
together with the corresponding enzymes pyruvate kinase,
creatine kinase and acetate kinase. These energy sources are
problematic in batch systems due to the rapid accumula-
tion of inorganic phosphate that inhibits protein synthesis
[49]. Few alternative energy systems have therefore been
established. The presence of pyruvate oxidase recycles inor-
ganic phosphate by condensation with added pyruvate into
acetyl phosphate, which then can be used again as energy
source for protein synthesis [49]. Another approach could
be the replacement of the energy source PEP by 3-phospho-
glycerate. This modiWcation extends batch reactions up to
2h if compared to 30–45min with PEP as conventional
energy source and results in higher yields [36]. In addition,
glucose-6-phosphate has been proposed as further option
of a secondary energy source superior to PEP or pyruvate
[49]. An economical improvement for CF protein synthesis
was recently proposed by using glucose as energy source
and nucleotide monophosphates instead of NTPs as pre-
cursors [50]. The cost of reagents could thus be lowered by
over 75% at similar protein production yields.
The amino acid supply during CF-protein synthesis has a
crucial impact on the expression yields. Instability and degra-
dation can produce a rapid bias in the amino acid concentra-
tion. Recombinant protein of 500?g per ml could be obtained
in batch reactions by increasing the initial amino acid concen-
tration from 0.5 to 2mM [49,51]. Repeated addition of amino
acids during the reaction also increases protein yields [45].
Genetic engineering of the genome of E. coli strains used as
extract sources signiWcantly decreased the degradation of
arginine, tryptophan and serine in the reaction [52]. ModiWca-
tions of the reaction conditions provide further potentials for
optimization. Omitting PEG and HEPES as well as the addi-
tion of the polycations spermidine and putrescine were pro-
posed as improvements of batch systems [53]. Optimized
batch systems are already eYcient enough to produced suY-
cient amounts of protein for structural analysis [54].
3. Cell-free preparation of membrane proteins
3.1. SpeciWc characteristics for the CF expression of MPs
The current variety of MPs produced on preparative
scales in CF systems comprises several bacterial multidrug
transporters [10,12,54,55], a bacterial light harvesting pro-
tein [9], the mechanosensitive channel MscL [11], several
eukaryotic GPCRs [13,14] and the ?-barrel nucleoside
transporter Tsx [14]. CF expression allows the production
of MPs at levels of several mg/ml RM in two very diVerent
modes (Fig. 3). First, MP precipitates are produced in stan-
dard CF-systems due to the lack of a hydrophobic environ-
ment. Second, hydrophobic media like detergents or
probably even lipids could be provided resulting in the pro-
duction of already solubilized MPs. The CF production of
precipitated MPs resembles the formation of inclusion
bodies that can frequently be observed upon in vivo produc-
tion of proteins in E. coli. However, CF generated MP pre-
cipitates might be structurally diVerent from inclusion
bodies as they usually do not need strong denaturants like
SDS to become solubilized. Mild detergents like alkyl-gly-
cosides, phosphocholines or alkyl-phosphoglycerols could
already be suYcient for the quantitative solubilization of
CF generated MP precipitates [10–12]. The harvested MP
precipitates should be washed for several times in an appro-
priate buVer (e.g. 15mM sodium phosphate, pH 6.8, 1mM
DTT) followed by centrifugation (5min, 5000g). Precipi-
tated impurities in the pellet could be removed by washing
with a detergent that has poor solubilization properties for
the MP (e.g. 3% n-octyl-?-glucopyranoside (?-OG)). Finally
the MP is solubilized in buVer containing the detergent of
choice (e.g. 2% 1-myristoyl-2-hydroxy-sn-glycero-3-[phos-
pho-rac-(1-glycerol)] (LMPG)) in a volume identical to the
volume of the RM. Incubation on a shaker at 30°C for one
hour is usually suYcient for the quantitative solubilization.
Residual pellet can be removed by centrifugation.

Page 8

362
D. Schwarz et al. / Methods 41 (2007) 355–369
The eYciency of solubilization certainly depends on the
speciWc recombinant MP as well as on the type of detergent.
Precipitates from the small ?-helical multidrug transporter
EmrE can be solubilized in a variety of diVerent detergents
while the quantitative solubilization of precipitates from
the nucleoside transporter Tsx or from the mechanosensi-
tive channel MscL was restricted to only a small selection
of detergents like polyethylene-glycol P-1,1,3,3-tetramethyl-
butylphenyl-ether (Triton X-100) or LMPG. Long-chain
phosphoglycerols like LMPG and the closely related
detergent 1-palmitoyl-2-hydroxy-sn-glycero-3-[phospho-
rac- (1-glycerol)] (LPPG) appear to be most versatile for
the solubilization of CF produced precipitates of structur-
ally diverse MPs. LMPG was highly eYcient in the solubili-
zation of bacterial ?-helical and ?-barrel type MPs and it
proved to be the only detergent suitable for the quantitative
solubilization of several eukaryotic G-protein coupled
receptors (GPCRs) [14].
3.2. CF expression of MPs in presence of detergents
The open nature of the CF expression set-up enables the
addition of deWned amounts of detergents directly into
the reaction [12–14]. The freshly translated proteins have thus
the opportunity to become embedded immediately into pre-
formed detergent micelles. This option to produce MPs in a
soluble form associated with a detergent of choice is a unique
characteristic for CF expression systems. Proteomicelles could
be puriWed directly out of the RM and critical steps like the
destabilization and isolation of MPs from membranes are
eliminated. An indispensable prerequisite is that the supplied
detergent is tolerated by the CF system even at concentrations
Fig. 3. Two modes of the cell-free production of membrane proteins. Both expression modes result in MPs solubilized in detergent micelles that can be
reconstituted into proteoliposomes for further functional and structural studies.

Page 9

D. Schwarz et al. / Methods 41 (2007) 355–369
363
exceeding several times the speciWc critical micellar concentra-
tion (CMC) that deWnes the minimal required eVective con-
centration for the solubilization of proteins. Only few
detergent types like dodecyl-phosphocholine (DPC) or ?-OG
severely inhibit the CF protein production already at low con-
centrations at or only slightly above their speciWc CMCs. For-
tunately, many other commonly used detergents are tolerated
by CF expression systems and the optimal concentration
Table 3
Cell-free expression of membrane proteins in the presence of detergents
S, soluble fraction; P, insoluble precipitated fraction; n.a., not available; n.r., not reported, 0, no detectable expression; 0+, less than 10?g/ml; +, 10–100?g/
ml; ++, 101–500?g/ml; +++, 501–1000?g/ml; ++++, 1001–2000?g/ml; +++++, more than 2001?g/ml.
Detergent name Optimal concentrationProtein References cited
Short name [%][mM] [x CMC]SP
Alkyl-glucosides
n-Dodecyl-?-D-maltosideDDM 0.08
0.1
<1
0.066
0.2
0.75
<1
0.4
0.877
<1
(2.30)
(2.87)
(28.69)
(1.89)
(4.14)
(25.65)
(34.2)
(13.68)
(30)
(29.82)
12.1
15.1
151
10
2.3
1.3
1.8
0.7
1.6
1.8
EmrE
Tsx
?2AR
MscL
EmrE
EmrE
M2
EmrE
MscL
M2
+++++
++++
+
n.r.
++++
+
0+
n.r.
n.r.
+
0
++++
+
n.r.
++
0
0+
n.r.
n.r.
+
[10]
[14]
[13]
[11]
[14]
[14]
[13]
[10]
[11]
[10]
n-decyl-?-D-maltoside
n-octyl-?-D-glucopyranoside
DM
?-OG
(6-O-(N-heptylcarbamoyl)-methyl-?-
D-glucopyranoside)
Steroid-derivatives
Digitonin
HECAMEG
Digitonin0.4
<1
0.75
<1
2.46
(3.25)
(8.13)
(12.20)
(16.26)
(40.01)
4.5
11.1
1.5
2
5
Tsx
?2AR
Tsx
NTR
MscL
++++
++
+
+
n.r.
+++
++
++
+
n.r.
[14]
[13]
[14]
[13]
[11]
3-[(3-cholamidopropyl)dimethly-
ammonio]-1-propansulfonat
CHAPS
Long chain-phosphoglycerols
1-myristoyl-2-hydroxy-sn-glycero-3-
[phospho-rac-(1-glycerol)]
1-palmitoyl-2-hydroxy-sn-glycero-3-
[phospho-rac-(1-glycerol)]
Mono-/Bi-chain-phosphocholines:
1,2-dioctanoyl-sn-glycero-3-phosphocholine
1,2-diheptanoyl-sn-glycero-3-phosphocholine
1,2-dihexanoyl-sn-glycero-3-phosphocholine
Dodecyl-phosphocholine
Polyoxyethylene-alkyl-ether:
polyoxyethylene-(8)-lauryl-ether, (C12/8)
polyoxyethylene-(23)-lauryl-ether, (C12/23)
LMPG0.01(0.21)4.2 Tsx++ ++[14]
LPPG 0.025(0.49) n.a.V2R++++++ [14]
diC8PC
DHPC
diC6PC
DPC
0.1
0.2
0.75
0.1
(1.96)
(4.15)
(16.54)
(2.84)
8.9
3
1.2
1.5
EmrE
Tsx
Tsx
EmrE
+++
+++
++
0+
+++++
++
++
+++
[14]
[14]
[14]
[14]
C12E8
Brij-35
<1
0.1
<1
0.1
0.1
0.01
1.5
0.84
0.2
1
0.2
0.2
<1
0.1
0.072
<1
<1
<1
(18.52)
(0.83)
(8.34)
(1.59)
(1.56)
(0.15)
(13.36)
(0.75)
(5.57)
(8.68)
(2.82)
(1.74)
(8.14)
(0.81)
(0.59)
(7.79)
(7.62)
(7.63)
260.8
10.4
104.2
21.3
10.4
4.2
178.1
10
n.a.
188.8
13
69.6
138
13.8
10
288.5
304.9
636.1
NTR
EmrE
?2AR
EmrE
Tsx
V2R
V2R
MscL
V2R
V2R
EmrE
V2R
NTR
Tsx
MscL
NTR
M2
NTR
+
+++++
++
+
+++
+
+++++
n.r.
+
+++++
0
+++++
+
0
n.r.
+
+
+
++
+
++
++
++++
0+
+
n.r.
+
+
+++++
+
++
+++++
n.r.
++
+
++
[13]
[14]
[13]
[14]
polyoxyethylene-(10)-dodecyl-ether, (C12/10)
polyoxyethylene-(10)-isotridecyl-ether, (C13/10)
polyoxyethylene-(10)-cetyl-ether, (C16/10)
polyoxyethylene-(20)-cetyl-ether, (C16/20)
GPC-100
GPX-100
Brij-56
Brij-58
[14]
[14]
[11]
[14]
[14]
[14]
[14]
[13]
[14]
[11]
[13]
[13]
[13]
polyoxyethylene-(2)-stearyl-ether, (C18/2)
polyoxyethylene-(20)-stearyl-ether, (C18/20)
polyoxyethylene-(10)-oleyl-ether, (C18-1/10)
polyoxyethylene-(20)-oleyl-ether, (C18-1/20)
polyoxyethylene-sorbitan-monolaurate 20
Brij-72
Brij-78
Brij-97
Brij-98
Tween 20
polyoxyethylene-sorbitan-monopalmitate 40
polyoxyethylene-sorbitan-monostearate 60
polyoxyethylene-sorbitan monooleate 80
Polyethylene-glycol derivatives:
polyethylene-glycol P-1,1,3,3-tetramethyl-
Butylphenyl-ether
Tween 40
Tween 60
Tween 80
TX-100 0.2
0.1
<1
0.1
0.1
<1
(3.09)
(1.55)
(15.46)
(1.72)
(1.66)
(16.58)
13.4
6.7
67.2
17.2
9.8
97.6
MscL
Tsx
NTR
Tsx
EmrE
?2AR
+++++
++++
+
++
+
+
n.r.
++++
++
+++
++++
++
[11]
[14]
[13]
[14]
[14]
[13]
Polyethylene-glycol 400 dedecyl-ether
Nonylphenyl-polyethylene-glycol
Thesit
NP40

Page 10

364
D. Schwarz et al. / Methods 41 (2007) 355–369
ranges for a relatively large group of detergents have been
determined (Table 3) [13,14]. The kinetics of MP solubiliza-
tion versus the detergent concentration shows two phases. Ini-
tially, the eYciency of MP solubilization is linear to the
amount of added detergent until a certain threshold concen-
tration has been achieved (Fig. 4). Then the yield of solubi-
lized MP remains constant with further increased detergent
concentrations and only additional empty micelles will be
formed until the detergent becomes toxic to the CF expression
system due to the inactivation of essential compounds. The
production kinetics has therefore a plateau-like appearance
and the maximal yield of soluble MP can be obtained over a
distinct concentration range of a speciWc detergent. It should
be considered that the complete solubilization of a MP might
not be possible and still some residual MP precipitate could
remain even far above the threshold concentration (Fig. 4).
Several elements of CF production of soluble MPs can
be subjected to optimization. Basic parameters are concen-
tration, type and chain length of the supplied detergent. For
initial screens it might be most straightforward to start with
detergent concentrations close to the maximal tolerated lev-
els to receive instantly the highest possible amounts of solu-
ble MP. The detergents should be prepared as highly
concentrated stock solutions in water and care should be
taken that organic solvents like chloroform have been com-
pletely removed e.g. by evaporation, in order to prevent an
inhibition of the CF reaction. Soluble protein fractions are
separated from precipitates after the reaction by centrifuga-
tion at 20,000g for 30min at room temperature. The pro-
duction of the MP should be quantiWed in both fractions by
SDS–PAGE analysis or by immunoassays. The Wnal deter-
gent concentration can then be decreased in subsequent
optimization reactions if desired.
The structure of the supplied detergent type can have a
major impact on the eYciency of solubilization as well as on
the functional folding of the synthesized MP. Some proteins
like the ?-helical multidrug transporter EmrE can be
expressed in soluble form with a diverse variety of structur-
ally diVerent detergents like alkyl-glucosides, phosphocho-
lines, polyethylene-glycol derivatives or polyoxyethylenes.
However, the soluble expression in preparative scales of the
majority of the MPs seems to be restricted to a much smaller
selection of detergents (Table 3). The supply of many popular
detergents that have been used in recent times for the
structural analysis of MPs like the alkyl-glucoside n-dodecyl-
?-D-maltoside (DDM) or the polyethylene-glycol derivative
Triton X-100 result in the high level soluble expression of
speciWc MPs. CF expression in presence of DDM yielded
milligram amounts of EmrE and of the nucleoside trans-
porter Tsx, but the detergent was rather ineVective for the
soluble expression of diVerent GPCRs [13,14]. Tsx could fur-
thermore only partially be solubilized in DDM and approx.
50% of the synthesized protein still remained as precipitate.
Clearly outstanding with respect to their ability to eYciently
solubilize structurally diverse MPs are the steroid-derivative
Digitonin and various detergents from the family of
polyoxyethylene-alkyl-ethers like polyoxyethylene-(23)-lau-
ryl-ether (Brij-35), polyoxyethylene-(20)-cetyl-ether (Brij-58),
polyoxyethylene-(20)-stearyl-ether (Brij-78) and polyoxyeth-
ylene-(20)-oleyl-ether (Brij-98) [13,14]. Out of more then 20
detergents, only Brij derivatives have been successful in the
quantitative soluble expression of the human vasopressin
type 2 receptor [14]. Brij derivatives having less than 10 poly-
oxyethylene groups were not eVective. In addition, the solubi-
lization of the vasopressin type 2 receptor was modulated by
the length of the alkyl chain. As a general guideline, the
eYciencies and the optimal concentrations of the most useful
detergents for the solubilization of structurally diVerent MPs
are compiled in Table 3.
In practice it is recommended to initially perform a set of
CF expression reactions in presence of the most promising
detergents with each new MP target (Fig. 5). In case of the
nucleoside transporter Tsx, the yield of soluble protein in
presence of the detergents Brij58, Brij78 and Brij97 is
Fig. 4. Soluble expression of the nucleoside transporter Tsx at increased detergent concentrations. The yield of soluble expressed MP increases with the
Wnal concentration of Brij58 and reaches a plateau at approx. 47-fold CMC. The amount of soluble Tsx then remains constant upon further increased
detergent concentrations up to 178-fold CMC. The total amount of produced Tsx protein (soluble and precipitate) remains constant at each condition.

Page 11

D. Schwarz et al. / Methods 41 (2007) 355–369
365
comparable to that of the Tsx precipitate isolated out of a
standard CF reaction without any detergent (Fig. 5). How-
ever, generally some lower amounts of protein might be
obtained when choosing the soluble mode of expression if
compared to the expression as a precipitate.
After high level CF production strategies have been
established, the functional folding of the synthesized MP
should be analysed [10–14,56,57]. The modes of expres-
sion, the origin of the CF extract, the solubilization proce-
dures of precipitates and the supplied detergent types are
important factors that could inXuence the folding path-
way of a protein. Circular dichroism spectra or two-
dimensional heteronuclear single-quantum correlation
(HSQC) spectra by solution nuclear magnetic resonance
(NMR) spectroscopy can give Wrst evidences of the pres-
ence of structural elements [12]. However, the develop-
ment of functional assays would be a very precious tool to
prove the three-dimensional folding of a protein into an
Fig. 5. Detergent screen for the cell-free production of the nucleoside transporter Tsx. Coomassie-stained SDS–Page of CF produced Tsx protein: M,
marker; s, supernatant of standard reaction; p, pellet of standard reaction. Lanes 1–8 represent 0.8?l of supernatant of CF expression in the presence of
diVerent type detergents: 1, 0.1% DDM; 2, 0.2% DHPC; 3, 0.4% Digitonin; 4, 0.1% Genapol X100; 5, 0.2% Brij58; 6, 1% Brij78; 7, 0.1% Brij97; 8, 0.5%
Brij98. The arrow indicates the overexpressed nucleoside transporter Tsx.
Fig. 6. 1H–15N TROSY-HSQC spectra of cell-free expressed 2H–15N-?TehA. The protein was expressed either in the unsoluble mode as precipitate and
resolubilized in LMPG or expressed in the soluble mode in the presence of Brij78, puriWed and the detergent was exchanged to LMPG. The concentration
of both samples was 0.5mM, dissolved in 25mM sodium phosphate buVer (pH 6.0) containing 3% LMPG. The spectra were measured at 318K on an
Avance 900MHz NMR spectrometer equipped with a cryogenic probe with eight scans per FID and 512 increments in the indirect dimension for the
insoluble expressed ?TehA and with four scans per FID and 300 increments in the indirect dimension for the soluble expressed transporter.

Page 12

366
D. Schwarz et al. / Methods 41 (2007) 355–369

Page 13

D. Schwarz et al. / Methods 41 (2007) 355–369
367
active conformation. The transport of the substrate ethi-
dium by EmrE could be veriWed with samples that have
been either produced in the soluble mode in presence of
DDM [10] or with proteoliposomes reconstituted from in
DDM solubilized precipitates [12]. On the other hand, the
nucleoside transporter Tsx could only be reconstituted in
a highly active form after its CF expression in the soluble
mode in presence of Triton X-100 [14]. While the soluble
expression in presence of Brij35 still resulted in some
residual activity, it was not possible to detect activity
from protein that has been produced as a precipitate and
solubilized in LMPG. This example demonstrates the
importance of initial expression screens that should con-
sider both, the high level production and the functional
folding of the MP. The available information of CF
expressed MPs that have been analysed by functional
assays is still limited to a few examples. However, it is
already evident that structurally very diVerent prokary-
otic as well as eukaryotic MPs can be produced in CF sys-
tems based on lysates of E. coli cells in both expression
modes as functionally active proteins.
4. New perspectives for the structural analysis of cell-free
produced membrane proteins
The CF expression of MPs oVers a high potential for
their structural analysis by NMR spectroscopy as well as
by X-ray crystallography. Protein samples that have been
eYciently labelled with stable isotopes or with non-natural
amino acids like selenomethionine can now be obtained in
less than 24 hours [54,58–60]. The recently presented crystal
structure of the small multidrug transporter EmrE at 3.7Å
resolution was solved by taking advantage of CF expres-
sion. The proposed active dimer was CF expressed directly
into di-myristoyl-phosphatidyl-choline (DMPC) liposomes
and crystals which were used for structure determination
were obtained in the presence of the detergent N-nonyl-?-D-
glucoside (NG) [54].
Most of the few known NMR structures from MPs were
solved from bacterial ?-barrel proteins that have been pro-
duced as inclusion bodies in E. coli followed by refolding
procedures [61,62]. The CF expression technique might
open new avenues for the determination of MP structures
by NMR spectroscopy as well as by X-ray crystallography
already solely because of the possibility to produce high
yields of functionally active MPs that are diYcult to obtain
with other expression systems. In addition, the speediness
of the reaction is highly competitive as the protein samples
are basically generated over night. Moreover, no special
equipment is required and the technique can principally be
established in standard biochemical labs within few days.
The most important feature however is the easiness and
eYciency in the synthesis of labelled protein samples.
The CF extract is completely devoid of amino acids due to
extensive dialysis steps during the preparation procedure.
As the operator has the complete control over the amino
acid pool of the CF reaction, any non-labelled amino acid
type can just be exchanged by its labelled derivative at the
initial set-up of the experiment. This instantly ensures
the 100% label incorporation into the synthesized protein.
The speciWc labelling of any amino acid type and also of
any amino acid combinations of an expressed protein is
thus as eYcient as the production of non-labelled proteins
[63]. Auxotrophic strains and minimal media that have to
be employed in conventional in vivo expression systems and
that often considerably reduce the yield of the recombinant
protein can be avoided. Scrambling problems are further-
more minimized as the metabolic activity of the CF extract
is very low [22]. The target protein is the only synthesized
protein in the system as all endogenous mRNA of the
extract was eliminated during the extract preparation. The
labelled protein could therefore be analysed by NMR spec-
troscopy directly in the RM without prior puriWcation as
no labelled background is present [64].
One major restriction for the determination of MP struc-
tures is the size limitation of proteins for NMR samples.
The rate of rotational tumbling of proteins decelerates with
increasing size, resulting in line broadening and lower reso-
lution. Because MPs need to be analysed in detergent
micelles, the signal resolution will be even more critical due
to the increased size of the protein/detergent complex. Fur-
thermore, MPs are mostly ?-helical proteins that show
rather narrow chemical shift dispersions with an extensive
signal overlap. The hydrophobic transmembrane segments
(TMSs) of MPs often contain cluster of similar amino acids
with equal chemical shifts that additionally contributes to
lower resolutions of corresponding NMR spectra. These
problems in combination with the low production rates of
many MPs in conventional in-vivo expression systems pre-
vented so far in most cases serious attempts for the struc-
tural analysis of MPs by NMR.
Several limitations can now be addressed by virtue of CF
expression. First, the size limitation of the solubilized protein
samples can be approached by using speciWc detergents. In a
systematic screen for liquid-state NMR compatible deter-
gents with regard to long sample lifetimes, maximal signal
detection as well as high spectral resolution, the class of lyso-
phosphoglycerols including LMPG and LPPG have been
Fig. 7. Combinatorial labelling scheme of membrane proteins by cell-free expression. Example of an amino acid selective combinatorial labelling scheme
with three diVerentially [15N] and [13C]labelled samples of the tellurite transporter TehA. The labelling scheme is shown on the left side. The crossed dotted
lines indicate the peak to become identiWed. In the [15N, 1H]-TROSY spectra, backbone amide protons of the [15N]-labelled amino acids are visible. The
indicated peak at position 121.3/8.07 is identiWed as an alanine as it is present in all three spectra and only the amino acid alanine has been [15N]-labelled in
all three samples. The corresponding HNCO spectra show the amide crosspeaks after carbonyl transfer and indicate that the preceding residue of this ala-
nine must be a leucine because a HNCO peak is only observed for samples 2 and 3 containing 13C-leucine, but not for sample 1 without 13C-leucine. The
corresponding alanine (peak HN:8.07; N:121.3) can now be localized in a Leu–Ala pair in the primary sequence of the protein which in that example was
identiWed as Ala206 of TehA. All spectra were recorded at an Avance 600MHz spectrometer.

Page 14

368
D. Schwarz et al. / Methods 41 (2007) 355–369
found to provide outstanding beneWts [65,66]. LMPG
micelles obviously do not restrict the tumbling rates of the
inserted solubilized MP and therefore do not result in
increased line broadenings. LMPG and its derivatives were
also found to be highly suitable for the solubilization of CF
produced MP precipitates [14]. This detergent type might be
therefore one of the Wrst choices for the structural analysis of
larger MPs by liquid-state NMR spectroscopy. A CF pro-
duced truncated 24kDa fragment of the putative tellurite
and multidrug transporter TehA of E. coli containing seven
TMSs was solubilized in LMPG and analysed by NMR. The
amide protons of the protein backbone could be almost com-
pletely assigned by using a rationally designed combinatorial
labelling approach [63]. The spectral quality of TehA further
depends on the mode of CF expression. 1H–15N-HSQC spec-
tra of TehA samples produced in the soluble mode with 1%
Brij-78 followed by a buVer exchange against 3% LMPG
showed a signiWcantly better resolution if compared with
samples that have been prepared as CF precipitates and re-
solubilized in LMPG (Fig. 6).
The spectral overlap due to the ?-helical structure of
many MPs could be approached by a mixture of amino acid
speciWc and combinatorial labelling strategies [63]. A simulta-
neous labelling of MPs with selected [15N]-labelled amino
acids in combination with distinct [13C]-labelled amino acids
can easily performed by CF expression (Fig. 7). The applica-
tion of two dimensional versions of HNCO experiments
helps to identify only those [15N]-labeled amino acids that
where N-terminally proceeded by a [13C]-labeled amino acid
type [63]. This strategy enables the unambiguous identiWca-
tion of consecutive amino acid pairs that can be subsequently
used as anchor points for further backbone resonance assign-
ments (Fig. 7). The TehA protein represents one of the larg-
est ?-helical proteins currently analysed by NMR and its
structural approach demonstrates the powerful synergy of
liquid-state NMR and CF expression.
References
[1] P. Renesto, D. Raoult, Ann. N.Y. Acad. Sci. 990 (2003) 642–652.
[2] M. Bannwarth, G.E. Schulz, Biochim. Biophys. Acta 1610 (2003)
37–45.
[3] J.M. Betton, Curr. Protein Pept. Sci. 4 (2003) 73–80.
[4] G. Zubay, Annu. Rev. Genet. 7 (1973) 267–287.
[5] A.S. Spirin, Cell-Free Translation Systems, Springer, New York, 2002.
[6] L.A. Ryabova, D. Desplancq, A.S. Spirin, A. Pluckthun, Nat. Biotech-
nol. 15 (1997) 79–84.
[7] G. Yin, J.R. Swartz, Biotechnol. Bioeng. 86 (2004) 188–195.
[8] M. Ali, H. Suzuki, T. Fukuba, X. Jiang, H. Nakano, T. Yamane, J.
Biosci. Bioeng. 99 (2005) 181–186.
[9] Y. Shimada, Z. Y Wang, Y. Mochizuki, M. Kobayashi, T. Nozawa,
Biosci. Biotechnol. Biochem. 68 (2004) 1942–1948.
[10] Y. Elbaz, S. Steiner-Mordoch, T. Danieli, S. Schuldiner, Proc. Natl.
Acad. Sci. USA 101 (2004) 1519–1524.
[11] C. Berrier, K. Park, S. Abes, A. Bibonne, J.M. Betton, A. Ghazi, Bio-
chemistry 43 (2004) 12585–12591.
[12] C. Klammt, F. Löhr, B. Schäfer, W. Haase, V. Dötsch, H. Rüterjans,
C. Glaubitz, F. Bernhard, Eur. J. Biochem. 271 (2004) 568–580.
[13] G. Ishihara, M. Goto, M. Saeki, K. Ito, T. Hori, T. Kigawa, M. Shir-
ouzu, S. Yokoyama, Prot. Express. Purif. 41 (2005) 27–37.
[14] C. Klammt, D. Schwarz, K. Fendler, W. Haase, V. Dötsch, F. Bern-
hard, FEBS Journal 272 (2005) 6024–6038.
[15] D.M. Kim, J.R. Swartz, Biotechnol. Bioeng. 66 (1999) 180–188.
[16] A.S. Spirin, V.I. Baranov, L.A. Ryabova, S.Y. Ovodov, Y.B. Alakov,
Science 242 (1988) 1162–1164.
[17] Y.B. Alakov, V.I. Baranov, S.J. Ovodov, L.A. Ryabova, A.S. Spirin,
I.J. Morozov, Method of preparing polypeptides in cell-free transla-
tion system U.S. Patent no. 5, 478, 73 (1995).
[18] D.M. Kim, C.Y. Choi, Biotechnol. Prog. 12 (1996) 1–8.
[19] Y. Endo, T. Sawasaki, Biotechnol. Adv. 21 (2003) 695–713.
[20] B.E. Roberts, B.M. Paterson, Proc. Natl. Acad. Sci. U S A 70 (1973)
2330–2334.
[21] K. Madin, T. Sawasaki, T. Ogasawara, Y. Endo, Proc. Natl. Acad. Sci.
USA 97 (2000) 559–564.
[22] K. Ozawa, N.E. Dixon, G. Otting, IUBMB Life 57 (2005) 615–622.
[23] Y. Shimizu, A. Inoue, Y. Tomari, T. Suzuki, T. Yokogawa, K. Nishik-
awa, T. Ueda, Nat. Biotechnol. 19 (2001) 751–755.
[24] T. Yamane, Y. Ikeda, T. Nagasaka, H. Nakano, Biotechnol. Prog. 21
(2005) 608–613.
[25] A.H. Erickson, G. Blobel, Biochemistry 22 (1983) 5201–5205.
[26] T. Ogasawara, T. Sawasaki, R. Morishita, A. Ozawa, K. Madin, Y.
Endo, EMBO J. 18 (1999) 6522–6531.
[27] T. Sawasaki, M.D. Gouda, T. Kawasaki, T. Tsuboi, Y. Tozawa, K.
Takai, Y. Endo, Methods Mol. Biol. 310 (2005) 131–144.
[28] Y. Li, E. Wang, Y. Wang, Protein Expr. Purif. 16 (1999) 355–358.
[29] F. Mignone, C. Gissi, S. Liuni, G. Pesole, Genome Biol. 3 (2002)
4.1–4.10.
[30] T. Ueda, H. Tohda, N. Chikazumi, F. Eckstein, K. Watanabe, Nucleic
Acids Res. 19 (1991) 547–552.
[31] I. Hirao, S. Yoshizawa, K. Miura, Nucleic Acids Symp. Ser. 22 (1990)
75–76.
[32] E. Kobatake, A. Ebisawa, O. Asaka, Y. Yanagida, Y. Ikariyama, M.
Aizawa, Apl. Biochem. Biotechnol. 76 (1999) 217–227.
[33] D.R. Gallie, D.E. Sleat, J.W. Watts, P.C. Turner, T.M. Wilson, Nucleic
Acids Res. 15 (1987) 3257–3273.
[34] V. Leathers, R. Tanguay, M. Kobayashi, D.R. Gallie, Mol. Cell Biol.
13 (1993) 5331–5347.
[35] T. Sawasaki, T. Ogasawara, R. Morishita, Y. Endo, Proc. Natl. Acad.
Sci. USA 99 (2002) 14652–14657.
[36] K. Sitaraman, D. Esposito, G. Klarmann, S.F. Le Grice, J.L. Hartley,
D.K. Chatterjee, J. Biotechnol. 110 (2004) 257–263.
[37] J.H. Ahn, H.S. Chu, T.W. Kim, I.S. Oh, C.Y. Choi, G.H. Hahn, C.G.
Park, D.M. Kim, Biochem. Biophys. Res. Commun. 338 (2005)
1346–1352.
[38] Y. Endo, T. Sawasaki, J. Struct. Funct. Genomics 5 (2004) 45–57.
[39] T. Sawasaki, Y. Hasegawa, M. Tsuchimochi, N. Kamura, T. Ogasa-
wara, T. Kuroita, Y. Endo, FEBS Lett. 514 (2002) 102–105.
[40] Y. Endo, T. Sawasaki, Methods Mol. Biol. 310 (2005) 145–167.
[41] L.A. Shaloiko, A. Gorokhovatskii, E.E. Maksimov, I.B. Alakhov, Bio-
org. Chim. 24 (1998) 539–543.
[42] V.A. Shirokov, P.N. Simonenko, S.V. Biryukov, A.S. Spirin, Cell-Free
Translation Systems, in: A.S. Spirin (Ed.), Springer, 2002, 227-234.
[43] Y. Kawarasaki, H. Nakano, T. Yamane, Plant Sci. 119 (1996) 67–77.
[44] Y. Kawarasaki, H. Nakano, T. Yamane, J. Biotech. 61 (1998) 199–208.
[45] D.M. Kim, J.R. Swartz, Biotechnol. Prog. 16 (2000) 385–390.
[46] M.C. Jewett, J.R. Swartz, Biotechnol. Bioeng. 86 (2004) 19–26.
[47] T. Kigawa, T. Yabuki, S. Yokoyama, FEBS Lett. 442 (1999) 15–19.
[48] L. Ryabova, L. Vinocurov, E. Shekhovtsova, Y. Alakhov, A. Spirin,
Anal. Biochem. 226 (1995) 184–186.
[49] D.M. Kim, J.R. Swartz, Biotechnol. Bioeng. 74 (2001) 309–316.
[50] K.A. Calhoun, J.R. Swartz, Biotechnol. Prog. 21 (2005) 1146–1153.
[51] M.C. Jewett, A. Voloshin, J. Swartz, M.P. Weiner, Q. Lu (Eds.), Gene
Cloning and Expression Technologies, Springer, 2002, 391-411.
[52] N. Michel-Reydellet, K. Calhoun, J. Swartz, Metab. Eng. 6 (2004)
197–203.
[53] M.C. Jewett, J.R. Swartz, Biotechnol. Bioeng. 87 (2004) 465–472.
[54] O. Pornillos, Y.J. Chen, A.P. Chen, G. Chang, Science 310 (2005)
950–1953.

Page 15

D. Schwarz et al. / Methods 41 (2007) 355–369
369
[55] M. Lorch, I. Lehner, A. Siarheyeva, D. Basting, N. PXeger, T. Manoli-
kas, C. Glaubitz, Biochem. Soc. Trans. 33 (2005) 873–877.
[56] S. Sonar, N. Patel, W. Fischer, K.J. Rothschild, Biochemistry 32
(1993) 13777–13781.
[57] L.K. Lyford, R.L. Rosenberg, J. Biol. Chem. 274 (1999) 25675–25681.
[58] T.L. Hendrickson, V. de Crécy-Lagard, P. Schimmel, Annu. Rev. Bio-
chem. 73 (2004) 147–176.
[59] T. Kigawa, E. Yamaguchi-Nunokawa, K. Kodama, T. Matsuda, T.
Yabuki, N. Matsuda, R. Ishitani, O. Nureki, S. Yokoyama, J. Struct.
Funct. Genomics. 2 (2002) 29–35.
[60] K. Ozawa, M.J. Headlam, D. Mouradov, S.J. Watt, J.L. Beck, K.J.
Rodgers, R.T. Dean, T. Hubert, G. Otting, N.E. Dixon, FEBS J. 272
(2005) 3162–3171.
[61] C. Fernandez, C. Hilty, G. Wider, P. Güntert, K. Wüthrich, J. Mol.
Biol. 336 (2004) 1211.
[62] C. Fernandez, K. Adeishvili, K. Wüthrich, Proc. Natl. Acad. Sci. USA
98 (2001) 2358–2363.
[63] N. Trbovic, C. Klammt, A. Koglin, F. Lohr, F. Bernhard, V. Dotsch, J.
Am. Chem. Soc. 127 (2005) 13504–13505.
[64] L. Guignard, K. Ozawa, S.E. Pursglove, G. Otting, N.E. Dixon, FEBS
Lett. 524 (2002) 159–162.
[65] T.P. Roosild, J. Greenwald, M. Vega, S. Castronovo, R. Riek, S. Chos,
Science 307 (2005) 1317–1321.
[66] R.D. Krueger-Koplin, P.L. Sorgen, S.T. Krueger-Koplin, I.O. Rivera-
Torres, S.M. Cahill, D.B. Hicks, L. Grinius, T.A. Krulwich, M.E. Gir-
vin, J. Biomol. NMR 17 (2004) 43–57.