Báo cáo khoa học: Membrane targeting of a folded and cofactor-containing protein potx

Membrane targeting of a folded and cofactor-containing protein
Thomas Bru¨ ser
1
, Takahiro Yano
2
, Daniel C. Brune
3
and Fevzi Daldal
1,2
1
Department of Biology, University of Pennsylvania, Philadelphia, PA 19104-6018, USA;
2
Johnson Research Foundation,
Department of Biochemistry and Biophysics, University of Pennsylvania, Philadelphia, PA 19104-6059, USA;
3
Department of Chemistry and Biochemistry, Arizona State University, Tempe, AZ 85287-1604, USA
Targeting of proteins to and translocation across the
membranes is a fundamental biological process in all
organisms. In bacteria, the twin arginine translocation
(Tat) system can transport folded proteins. Here, we
demonstrate in vivo that the high potential iron-sulfur
protein (HiPIP) from Allochromatium vinosum is trans-
located into the periplasmic space by the Tat system of
Escherichia coli. In vitro, reconstituted HiPIP precursor
(preHoloHiPIP) was targeted to inverted membrane
vesicles from E. coli by a process requiring ATP when
the Tat substrate was properly folded. During membrane
targeting, the protein retained its cofactor, indicating that
it was targeted in a folded state. Membrane targeting did
not require a twin arginine motif and known Tat system
components. On the basis of these findings, we propose

that a pathway exists for the insertion of folded cofactor-
containing proteins such as HiPIP into the bacterial
cytoplasmic membrane.
Keywords: ATP dependence; high potential iron–sulfur
protein (HiPIP); in vitro folding; membrane targeting; twin
arginine translocation.
Bacteria translocate proteins across the cytoplasmic mem-
brane by two main pathways, the general secretory (Sec)
and the twin arginine translocation (Tat) systems [1,2]. In
the past, most studies on protein targeting have focused on
translocation or membrane integration of unfolded protein
substrates by the Sec machinery, and many components
have been identified that play specific roles in Sec-dependent
targeting pathways [3]. On the other hand, the Tat system
has been shown to translocate folded proteins powered by
the transmembrane proton gradient [4]. So far, only four
components, TatA, TatB, TatC and TatE, have been
identified in Escherichia coli. Three of the corresponding
genes, tatABC, are organized in an operon together with
tatD, which encodes a nuclease that is probably unrelated to
the Tat system [5]. TatE is a structural and functional
homolog of TatA and encoded at a different locus [6]. TatA
and TatB together can form a double-layered ring structure
and are suggested to constitute the translocation pore [7].
The precise role of TatC remains to be determined, but it is
already known that this component can form a functional
unit with TatB [8]. It is currently believed that TatA or
TatE, together with TatB and TatC, can carry out most of
the required functions, such as binding of Tat substrates,
recognition of the folded state, formation of a translocation

pore, usage of the DpH for translocation, or prevention of
ion leakage [2].
Protein substrates for both the Sec and Tat systems are
synthesized with similar N-terminal signal peptides, com-
posed of a hydrophilic and positively charged n-region,
followed by a hydrophobic h-region and then often by a
c-region which determines a cleavage site (Fig. 1). Sub-
strates of the bacterial Tat system contain longer signal
peptides which include a conserved (S/T)RRXFLK motif in
their n-region [9,10] and a significantly less hydropho-
bic h-region [11]. In addition, Tat signal peptides often
contain charged amino-acid residues in their c-region, which
are not common in Sec-typical signal peptides [2,12,13].
Folding of Tat substrates before their Tat-dependent
translocation in E. coli has been demonstrated in vivo in
several cases, including the green fluorescent protein and
hydrogenase [14–16]. Furthermore, cytoplasmic matur-
ation systems that induce protein folding such as iron-
sulfur cluster assembly pathways can also act in
conjunction with translocation [17]. Moreover, natural
Tat substrates that acquire a folded and often cofactor-
containing state before their translocation appear not be
secreted by the Sec system [2]. On the other hand,
Sanders et al. [18] have shown that typical Sec substrates
such as c-type cytochromes can be translocated via the
Tat system only when they are synthesized with Tat
signal sequences and if they receive in the cytoplasm their
heme cofactor allowing their folding.
Recently, a functional in vitro Tat system has been
established using in vitro translated and cofactor-free Tat

Correspondence to T. Bru
¨
ser, Institut fu
¨
r Mikrobiologie,
Universita
¨
t Halle, Kurt-Mothes-Str. 3, 06120 Halle, Germany.
Fax: + 49 345 5527010, Tel.: + 49 345 5526360,
E-mail:
Abbreviations: Tat, twin arginine translocation; HiPIP, high potential
iron-sulfur protein; INV, inverted membrane vesicle; MalE,
maltose-binding protein; DDM, dodecyl maltoside.
Note: The prefixes ÔHoloÕ, ÔMalÕ and ÔApoÕ are used herein solely for the
description of HiPIP, the cofactor content and folded state of which
has been investigated in vitro. On the other hand, the prefixes ÔpreÕ
and ÔmatÕ are used to distinguish precursor and mature proteins.
For example, the precursor of the HiPIP holoprotein is termed
preHoloHiPIP, whereas a HiPIP precursor of unknown folded state
is termed preHiPIP.
(Received 10 October 2002, revised 19 January 2003,
accepted 27 January 2003)
Eur. J. Biochem. 270, 1211–1221 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03481.x
substrates [19,20]. However, the mechanism of targeting and
translocation of folded redox proteins by the Tat system
remains largely unknown. In this study, we opted to use a
fully folded and cofactor-containing Tat substrate. As the
folded state is of importance for the Tat system in vivo,we
expected that this approach might lead to new insights. We
have chosen as a model substrate the high-potential iron-

sulfur protein (HiPIP) from Allochromatium vinosum,which
is a monomeric 9-kDa periplasmic protein containing one
[4Fe)4S]
2+/3+
cluster bound via four cysteines [21,22]. As
biosynthesis of iron-sulfur clusters is thought to take place
in the cytoplasm, it is thought that HiPIP is folded and
loaded with an iron-sulfur cluster before its translocation
into the periplasmic space [17]. The signal peptide of HiPIP
has all the characteristics of Tat substrates [23] (Fig. 1), and
its structure and folding properties are well characterized
[24].
In this work, we first demonstrated that the hetero-
logously expressed HiPIP is translocated into the periplas-
mic space in E. coli, and that its translocation requires the
tatABC genes as well as the RR motif of the signal sequence.
Next, we successfully reconstituted the [4Fe)4S] cluster into
the HiPIP precursor (preHoloHiPIP) in vitro andusedthis
folded protein for in vitro targeting experiments. We found
that freshly prepared preHoloHiPIP can be targeted
efficiently to inverted cytoplasmic membrane vesicles
(INVs) from E. coli, that this process requires ATP
hydrolysis and an ATP-regeneration system, and that the
membrane-associated preHoloHiPIP undergoes a conform-
ational change without losing its [4Fe)4S] cofactor. Fur-
thermore, this in vitro targeting reaction requires neither the
known Tat components (TatABCE) nor the RR motif in
the signal peptide. We also observed that, on extended
storage, preHoloHiPIP could be converted into a form
capable of integrating into the membrane without ATP,

probably because of conformational changes induced by the
loss of its cofactor. The overall findings suggest that a
pathway exists for the membrane insertion of folded and
cofactor-containing proteins. Such a pathway may play a
role in the biogenesis of membrane-bound redox proteins,
or it may precede recognition and translocation by the Tat
system in vivo.
Materials and methods
Genetic methods
TatABC genes (including the tatA promoter region)
were amplified with Pfu polymerase (Stratagene) from
genomic DNA from E. coli MC4100 using the primers
5¢-AGTCGTGGATCCAAGATCAGGTCGGTATT-3¢ and
5¢-TGCGCGGCGAGCTCAATAATCGCTTC-3¢.ThePCR
product was cleaved with BamHI and SacI at primer-
generated cleavage sites and cloned into the corresponding
sites of pRK415, resulting in pRK-tatABC.TheXbaI–
BamHI fragment of pCVH1 [23], which contains hip and its
promoter region, was cloned into the corresponding sites of
pRK-tatABC, resulting in pRK-tatABC-hip. For construc-
tion of the hip expression vector pRK-hip,thetatABC-
containing fragment from pRK-tatABC-hip was removed
by restriction with SacI–BamHI and self-ligation of the
remaining vector. The RR in the signal peptide of preHiPIP
was mutated to KK using the primer couples 5¢-AAGAGC
AAGAAAGACGCTGTCAAAGTGATG-3¢/5¢-TCCGG
ATATAGTTCCTCCT-3¢ and 5¢-ACGTTACTGGTTTC
ACATTC-3¢/5¢-AGCGTCTTTCTTGCTCTTGCTGATT
GGCTT-3¢ to generate two overlapping PCR fragments
with pEXH5 as template [13]. These two PCR fragments

were subjected to a second round of PCR to generate the
RRfiKK-mutant fragment, which was then cleaved by
NdeIandHindIII and cloned into the corresponding sites of
pET22b+ (Novagen), resulting in pEXH15 used for inclu-
sion body formation of the mutant protein. For the in vivo
analyses, the RRfiKK exchange was achieved using the
primers 5¢-CATCACTTTGACAGCGTCTTTCTTGCTC
TTGCTGATTGGCTTATCG-3¢ and 5¢-CGATAAGCCA
ATCAGCAAGAGCAAGAAAGACGCTGTCAAAGTG
ATG-3¢ using the QuikChange kit (Stratagene) with
pCVH1 as a template. The RRfiKK exchanges were
confirmed by sequencing.
In vivo
analysis of preHiPIP translocation
E. coli strain MC4100 and its derivatives B1LK0 (DtatC)
[25] and DADE (DtatABCDE) [5] were generously provided
by T. Palmer (University of East Anglia, Norwich, UK) and
grown on Luria–Bertani medium under aerobic conditions,
or on Luria–Bertani medium supplemented with 0.4%
NaNO
3
and 0.5% glycerol under anaerobic growth condi-
tions. hip was expressed from its own promoter using either
pRK-hip or pRK-tatABC-hip, which also contains tatABC
under the control of the tat promoter, and which was used
for complementation of HiPIP translocation in Tat mu-
tants. Periplasmic fractions were prepared using 50 mL
anaerobically grown cell cultures, as described elsewhere
[26]. Immunoblot analysis was carried out as described
previously [13].

Preparation of fully folded precursor HiPIP
(preHoloHiPIP)
For preparation of preApoHiPIP inclusion bodies, a 1-L
E. coli BL21 DE3 culture carrying pEXH5 [13] was
grown in Luria–Bertani medium with high aeration, and
hip expression was induced for 3 h with 1 m
M
isopropyl
Fig. 1. The signal peptide of A. vinosum HiPIP contains all known
determinants specific of the Tat translocation pathway. The Tat signal
peptide of A. vinosum HiPIP (upper sequence) is compared with the
Sec-typical signal peptide sequence of the outer membrane protein A
(OmpA) from E. coli. Note that, compared with Sec signal peptide
sequences, the Tat signal peptide sequence is generally longer, it has a
twin arginine motif (underlined bold) within a conserved pattern
(bold), an extended hydrophilic N-terminus (n-region), a longer
uncharged region with moderate hydrophobicity (h-region), and often
a charged residue near the cleavage site (c-region followed by an
arrow).
1212 T. Bru
¨
ser et al.(Eur. J. Biochem. 270) Ó FEBS 2003
thio-b-
D
-galactoside at D
600
¼ 1. Harvested cells were
washed once in 50 mL 100 m
M
Tris/HCl, pH 8.0, resus-

pended in 30 mL of the same buffer, and broken by two
passages through a French pressure cell operating at
138 MPa. Inclusion bodies and cell debris were sedimented
and washed twice by centrifugation (20 min, 25 000 g,
4 °C), dissolved in 20 mL ice-cold 50 m
M
Tris/HCl
(pH 8.0)/2 m
M
dithiothreitol/8
M
urea, and cell debris was
separated by centrifugation (30 min, 30 000 g,4°C). The
supernatant (inclusion body solution) was shock-frozen in
liquid nitrogen in 1 mL aliquots and stored at )80 °C.
For in vitro folding, preApoHiPIP was first allowed to
assemble iron at room temperature in a reaction mixture
containing 43 l
M
preApoHiPIP, 220 l
M
Fe(NH
4
)SO
4
,
2m
M
dithiothreitol and 5
M

urea in a total volume of
15 mL. After 5 min of incubation, 1.25 m
M
Na
2
Swas
added and folding was continued for 20 min. The solution
was then applied to a 2-mL DEAE-Sephacryl (Pharmacia)
column equilibrated with 20 m
M
Tris/HCl, pH 9.0. Folded
and cofactor-containing preHoloHiPIP passed through the
column and was subsequently dialyzed against STM buffer
(250 m
M
sucrose, 5 m
M
Tris/HCl pH 8.0, 5 m
M
MgSO
4
).
PreHoloHiPIP prepared in this way was stable for about
1 week on ice. Iron content of HiPIP was determined using
the bathophenanthrolinedisulfonate method [27]. UV/vis
spectra were recorded using an Hitachi U3210 spectro-
photometer.
Preparation of inverted cytoplasmic membrane vesicles
Cells (6 g wet weight) were resuspended in 40 mL 10 m
M

Tris/acetate, pH 7.6, containing 20% sucrose, 0.1 m
M
EDTA, and 1 m
M
dithiothreitol, incubated for 10 min at
room temperature, and sedimented at 5000 g for 20 min at
4 °C. The pellet was resuspended in 40 mL ice-cold 5 m
M
MgSO
4
and incubated on ice for 20 min, followed by
centrifugation at 5000 g for 20 min at 4 °C. The pellet was
resuspended in 50 m
M
Tris/acetate, pH 7.6, containing
250 m
M
sucrose, 1 m
M
dithiothreitol, and 50 lgÆmL
)1
DNase I, and passed through a French pressure cell
operating at 27.6 MPa to produce INVs [28]. The solution
was then centrifuged for 10 min at 5000 g,andthe
supernatant was centrifuged again at 150 000 g for 2 h at
4 °C. The membrane pellet was resuspended in 2 mL STM
buffer supplemented with 1 m
M
dithiothreitol. Aggregated
material was removed by a final centrifugation at 15 000 g,

and the clear supernatant was divided into aliquots and
frozen in liquid nitrogen.
Cofactor tracing and membrane-targeting assays
The [4Fe)4S]
2+/3+
cofactor was radioactively labeled by
including 50 lCi
55
FeCl
3
in the iron-assembly step of the
folding protocol described above. PreHoloHiPIP was
targeted to INVs in a mixture containing  600 pmol
preHoloHiPIP ( 40 000 c.p.m.), 26 lg INV protein,
250 m
M
sucrose, 5 m
M
MgSO
4
,5m
M
ATP, 60 m
M
phos-
phocreatine, 100 lgÆmL
)1
creatine kinase, 5 mgÆmL
)1
BSA,

1m
M
dithiothreitol, and 15 m
M
Tris/HCl, pH 7.5. The
reaction was started after 1 min preincubation by addition
of INVs, carried out for the indicated amounts of time at
37 °C, and terminated by rapid vacuum filtration through
0.22-lm pore size GV-type membranes (Millipore). The
filtered INVs were immediately washed with 3 mL STM/
200 m
M
NaCl/50 m
M
MgSO
4
,and
55
Fe bound to INVs was
monitored by determination of the radioactivity thus
retained by liquid scintillation counting. For immunoblots,
filter membranes were extensively washed with 100 lL
SDS/PAGE sample buffer, and 10 lL were used for SDS/
PAGE and blotting as described elsewhere [13]. [
35
S]Met-
labeled maltose-binding protein (MalE) was produced by
in vitro translation with rabbit reticulocyte lysate (Promega
protocol) from RNA obtained by in vitro transcription of
HindIII-digested pBAR107N [29]. The MalE used herein is

a C-terminally truncated form which cannot fold and
therefore has been found to be more suitable for Sec-
dependent in vitro translocation than full-length MalE [29].
MalE was targeted for 40 min at 37 °CtoINVsina
mixture containing 1 lLMalEand50lg INV protein
under conditions identical with the targeting assay of
preHoloHiPIP described above. The assay mixture was then
incubated on ice with or without thermolysin (200 lgÆmL
)1
,
1 h), followed by trichloroacetate precipitation, SDS/PAGE
analysis, and analysis of the radioactive protein bands by
use of the phosphoimager system and the quantification
program
IMAGEQUANT
(Molecular Dynamics). When
indicated, the targeting reaction was carried out in the
presence of gramicidine (10 l
M
) or cyanide m-chlorophenyl-
hydrazone (CCCP, 100 l
M
), or ATP was replaced by
NADH.
Other biochemical methods
Mature HiPIP (matHoloHiPIP) was purified from photo-
heterotrophically grown A. vinosum. Cells (10 g) were
broken in 20 m
M
Tris/HCl, pH 8.5, by two passages

through a French pressure cell operating at 138 MPa. After
low-speed and ultracentrifugation steps, the supernatant
containing the soluble proteins was loaded onto a 100-mL
DEAE-Sephacel column equilibrated with the same buffer.
After washing of the column, matHoloHiPIP was eluted by
changing the buffer to 20 m
M
Tris/HCl, pH 7.0. HiPIP-
containing fractions were dialyzed against 20 m
M
Tris/HCl,
pH 8.5, and further purified using a 1-mL Mono Q FPLC
column and a 40-mL gradient of 0–200 m
M
NaCl in 20 m
M
Tris/HCl, pH 8.5. MatHoloHiPIP was homogeneous as
judged by Coomassie-stained SDS/PAGE gels.
SDS/PAGE analysis was carried out with 15% T
Laemmli gels, and protein was determined by the Lowry
method [30,31]. For N-terminal amino-acid sequence
determination of Coomassie-stained, Immobilon filter blot-
ted proteins, Edman degradation was carried out using a
Proton 2090E gas-phase protein sequencer (Beckman,
Fullerton, CA, USA) equipped with an online Hewlett–
Packard 1090L HPLC [32]. For affinity purification,
150 nmol preHoloHiPIP was coupled to a 2-mL Aminolink
column matrix (Pierce, Rockford, IL, USA). Membranes
from 6 g cells of the E. coli strain DADE lacking TatA-
BCDE were resuspended in STM buffer, and solubilized for

1 h by stirring at 4 °C and addition of dodecyl maltoside
(DDM) to a final concentration of 1%. Solubilized mem-
branes were centrifuged (145 000 g, for 1 h) and the
supernatant was diluted with STM buffer to 0.2% DDM.
The Aminolink-preHiPIP column (operated by gravity
Ó FEBS 2003 Membrane targeting of HiPIP (Eur. J. Biochem. 270) 1213
flow) was equilibrated to STM/0.2% DDM, and solubilized
membranes were loaded, followed by a wash with 10
column volumes (20 mL) of STM/0.2% DDM and elution
with 10 mL STM/0.2% DDM/2 m
M
ATP. The column
wasthenwashedwith10mLSTM/0.2%DDM/200m
M
NaCl, followed by a final wash with 10 mL STM/0.2%
DDM/500 m
M
NaCl in order to detect any additional
preHoloHiPIP-binding protein not eluted by ATP.
EPR measurements
X-band (9.4 GHz) EPR spectra were recorded by a Bruker
ESP 300E spectrometer using an Oxford Instruments
ESR-9 helium flow cryostat to control desired sample
temperature. HiPIP preparations were oxidized with 5 m
M
ferricyanide. Final EPR spectra were obtained after
subtracting a spectrum of the buffer containing 5 m
M
ferricyanide measured under the same conditions. EPR
conditions used are described in detail in the legends to the

individual figures.
Chemicals
55
FeCl
3
and [
35
S]Met were obtained from Perkin–Elmer
Life Sciences. DNA-modifying enzymes were from Bio-
Labs, and in vitro transcription and translation kits from
Promega. All other chemicals and enzymes were from
Sigma or from Fisher Scientific and were of the highest
available purity.
Results
A. vinosum
HiPIP is a Tat substrate in
E. coli
The gene hip from A. vinosum encoding HiPIP was
expressed from its own promoter in various E. coli strains
and their subcellular fractions were prepared. Western-blot
analyses indicated that HiPIP was translocated into the
periplasm under anaerobic growth conditions (Fig. 2). In
wild-type cells carrying the plasmid pRK-hip, all processed
HiPIP was detected in the periplasmic fraction (Fig. 2A,
lanes 1–3). In contrast, no HiPIP could be detected in the
periplasm of tat mutants deficient in TatC or TatABCDE,
indicating that translocation of HiPIP to the periplasm does
not occur (Fig. 2A, lanes 4–9). Consequently, HiPIP
precursor accumulated in the cytoplasm fraction of the
mutant strains. Translocation of HiPIP could be restored in

the tatABCDE mutant by providing only the tatABC genes
in trans (Fig. 2A, lanes 10 and 11), demonstrating that the
lack of translocation of HiPIP was due to the absence of the
tatABC genes. To further establish that HiPIP is indeed a
substrate of the Tat translocon in E. coli, the twin arginine
residues in its signal peptide were exchanged with lysines.
This kind of substitution has been reported to block Tat-
dependent translocation of other Tat substrates in E. coli
[9]. When the RRfiKK signal sequence mutant of HiPIP was
analyzed, we observed that the translocation was blocked,
and that the precursor as well as a degradation product with
the size of mature HiPIP accumulated inside the cytoplasm
in large amounts (Fig. 2B). These findings confirmed that
the twin arginine motif in the signal peptide of A. vinosum
preHiPIP is required for its Tat-dependent translocation in
E. coli, and established A. vinosum HiPIP as a bona fide Tat
substrate.
HiPIP precursor can be folded
in vitro
to native
conformation
Highly purified HiPIP precursor apoprotein (preApo-
HiPIP) was obtained from inclusion bodies and folded to
its native conformation in vitro as described in Materials
and Methods. Reconstituted HiPIP precursor (preHolo-
HiPIP) contained 3.9 Fe atoms per protein and showed the
typical optical absorption spectrum of purified mature
HiPIP (matHoloHiPIP) with maxima at 283 nm and
388 nm and an absorbance ratio A
282

/A
388
of 2.6 ± 0.1
[13,33]. Further, EPR spectra of reconstituted preHolo-
HiPIP exhibited the well-characterized matHoloHiPIP
signature with g
x
, g
y
and g
z
tensor values of 2.037, 2.045
and 2.122 and a g
av
value of 2.068 (Fig. 3). The data indicate
that preHoloHiPIP was folded fully under the conditions
used, as shown previously for mature HiPIP [24]. As the
dimerization of HiPIP induces well-characterized hetero-
geneities in the g
z
region [34], a close examination of our
data indicated that in vitro folded preHoloHiPIP is mono-
meric in solution. Treatments of preHoloHiPIP with
proteinase K, thermolysin, trypsin or chymotrypsin as well
as with combinations of these proteases generated a mature
Fig. 2. Translocation of A. vinosum HiPIP depends on the Tat system
in E. coli. Detection of A. vinosum HiPIP precursor (pre) and mature
HiPIP (mat) by Western blotting in cell fractions from E. coli wild-type
(MC4100) and Tat-deficient strains expressing hip.Strainsandplas-
mids are indicated above corresponding lanes, as follows: wt, wild type

E. coli MC4100 in (A), or XL1-BLUE in (B) and (C); DtatC,B1LK0;
DtatABCDE, DADE strains. The plasmid pRK-tatABC-hip contained
both the E. coli tatABC and A. vinosum hip genes, expressed inde-
pendently (see Materials and methods for description of this plasmid).
In each lane, protein corresponding to  375 lL bacterial culture were
loaded, and the applied cell fractions are indicated: c, cytoplasm; m,
membrane; p, periplasm. (A) Restoration of HiPIP translocation in a
DtatABCDE deletion strain complemented with the tatABC genes
carried by pRK-tatABC-hip; (B) effect of replacing the conserved
arginine residues R(10)R(11) with K(10)K(11) residues of hip on the
translocation of HiPIP protein. (C) Coomassie-stained gel from a
representative cell fractionation. Molecular mass markers are indicated
on the left.
1214 T. Bru
¨
ser et al.(Eur. J. Biochem. 270) Ó FEBS 2003
form of HiPIP, suggesting that only its 4-kDa signal peptide
could be truncated by protease treatment without losing or
perturbing the EPR characteristics of its [4Fe)4S] cluster
(data not shown), and indicated that the iron-sulfur cluster-
binding domain of preHoloHiPIP is highly resistant to
proteases. The cleavage site of protease-treated preHolo-
HiPIP was determined by N-terminal amino-acid sequen-
cing. Proteinase K cleaved at position )3, )2, )1, 0 and +1
relative to the natural signal peptide cleavage site, with 80%
cleavage at position )2. Thermolysin treatment on the other
hand resulted in a clean single cut of the signal peptide at
position )3. This indicated that these proteases indeed
cleaved off the signal peptide from in vitro folded pre-
HoloHiPIP without affecting the remainder of the protein.

Therefore, preHoloHiPIP treated with proteinase K or
thermolysin was regarded as mature matHoloHiPIP.
ATP-dependent targeting of HiPIP precursor into
inverted membrane vesicles from
E. coli
In vitro folded and [4Fe)4S] cluster-containing preHolo-
HiPIP was next incubated with INVs under the conditions
described in Materials and Methods. In the presence of ATP
and an ATP-regenerating system consisting of phosphocre-
atine and creatine kinase, we observed that a large amount
of preHoloHiPIP accumulated in the membranes (Fig. 4A).
To quantify this membrane-targeting reaction, and also
to monitor the fate of the cofactor during this process,
preHoloHiPIP was radiolabeled by including
55
FeCl
3
into
the folding procedure, and time-dependent
55
Fe accumula-
tion was monitored. The data indicated that only a small
amount of
55
Fe could be associated with the membrane in
the absence of ATP, and addition of ATP enhanced it by
about 5- to 20-fold (Figs 4B and 5). From
55
Fe label tracing
Fig. 3. preHoloHiPIP reconstituted in vitro has native spectroscopic

properties. EPR spectra of in vitro reconstituted preHoloHiPIP are
compared with those of mature HiPIP as purified from A. vinosum.The
EPR spectroscopy conditions were as follows: modulation frequency,
100 kHz; modulation amplitude, 10.145 G; time constant, 163.84 ms;
conversion time, 163.84 ms (see Materials and methods for more
details). Note that the reconstituted preHoloHiPIP exhibits a symmet-
ricalpeakintheg
z
region (seeinsert), indicatingits monomeric state [34].
Fig. 4. ATP-dependent targeting of preHoloHiPIP to INVs from
E. coli. (A) Immunoblot analysis using anti-HiPIP serum after tar-
geting. INVs from 15 min targeting assays were filtered and resus-
pended in 100 lL SDS/PAGE sample buffer for analysis, and samples
were separated by SDS/PAGE (15% T) and blotted on nitrocellulose
for HiPIP detection using polyclonal antibodies. Lane 1, HiPIP pre-
cursor standard (solubilized inclusion bodies,  0.25 lg); lanes 2–5,
HiPIP after targeting to INVs in the absence (lanes 2 and 3) or presence
(lanes 4 and 5) of ATP; lane 6, mature HiPIP purified from A. vinosum
( 0.25 lg, see Materials and methods), used as a control. The posi-
tions of precursor (pre) and mature (mat) HiPIP bands are indicated.
Note that without INVs no significant preHoloHiPIP was retained on
the filter, and that ATP had no effect when INVs were absent (not
shown). (B)
55
Fe-labeled preHoloHiPIP was targeted to E. coli INVs.
Targeting was terminated using a rapid filtration assay, and filter
retained
55
Fe radioactivity was monitored by liquid scintillation
counting (see Materials and methods). Assay mixtures contained

besides the standard mix (see Materials and methods) 5 m
M
ATP/
10 l
M
gramicidin (r), 5 m
M
ATP (j); no ATP (negative control) (e);
5m
M
NADH (h). Each data point is the mean of three independent
incubation/filtration assays, and error bars show the standard devia-
tions observed between samples. The presence of a DpH is indicated
and generated by either addition of ATP or NADH. The formation of
a DpH under the assay conditions used was confirmed by fluorescence
quenching assays performed separately (data not shown).
Ó FEBS 2003 Membrane targeting of HiPIP (Eur. J. Biochem. 270) 1215
kinetics, the specific activity of ATP-dependent membrane
insertionwasestimatedtobe33±10pmoltargeted
preHiPIPÆmin
)1
Æ(mg INV protein)
)1
.Upto18pmol
( 0.3 lg) preHoloHiPIP could be targeted to 26 lgmem-
branes, indicating thatthe membranes were efficiently loaded
with preHoloHiPIP. An ATP-regenerating system was
required for targeting, indicating that ATP is hydrolyzed in
the assay. Additional experiments indicated that hydrolysis
of ATP is required for this process, as neither the non-

hydrolyzable ATP analogue p[NH]ppA nor AMP could
substitute for ATP (S. Trautmann and T. Bru
¨
ser, unpub-
lished results). The existence of a transmembrane proton
gradient, as generated by either reverse action of ATP
synthase or NADH-dependent electron transport, was not
required for membrane targeting, and it rather affected
negatively the kinetics (Fig. 4B). Moreover, with a similar
assay, matHoloHiPIP could not be targeted to INVs,
indicating that the signal peptide is required for the targeting
process (Fig. 5A).
Next, to probe whether the ATP-dependent targeting of
preHoloHiPIP to the membrane required the twin arginine
motif in the signal peptide or the known Tat components,
membrane targeting with an in vitro folded R(10)R(11)fi
K(10)K(11) signal sequence mutant (preHoloHiPIP-KK)
and targeting to INVs derived from various Tat-deficient
mutant strains were tested. The data indicated that
preHoloHiPIP-KK was accepted as an efficient substrate
for ATP-dependent membrane insertion, and that deletion
of tatC or tatABCDE did not significantly affect the ATP-
dependent targeting of preHoloHiPIP to INVs (Fig 5B,C).
Therefore, the ATP-dependent membrane-targeting reac-
tion does not appear to require the twin arginine motif
in vitro, nor does it depend on any of the as yet known Tat
components. To examine the possibility that the ATP
dependence results from an involvement of SecA in the
targeting process, we compared the azide sensitivity of
preHoloHiPIP targeting with that of the targeting of the

Fig. 5. Requirements for membrane targeting of HiPIP. (A)
55
Fe
tracing data obtained using standard targeting assays (see Materials
and methods) with matHoloHiPIP and preHoloHiPIP to INVs. To
produce matHoloHiPIP, preHoloHiPIP reconstituted in vitro was
digested for 60 min on ice with 100 lgÆmL
)1
proteinase K. The reac-
tion was stopped by addition of phenylmethanesulfonyl fluoride
(10 m
M
,10 min,0 °C) in dimethyl sulfoxide. It was confirmed that this
phenylmethanesulfonyl fluoride treatment resulted in complete inac-
tivation of proteinase K. For assays with the unprocessed substrate,
preHoloHiPIP was incubated in parallel on ice without protease, and
treated thereafter with phenylmethanesulfonyl fluoride. Note that the
absence of the N-terminal signal peptide in the matHoloHiPIP results
in targeting deficiency even in the presence of ATP. (B)
55
Fe tracing
data obtained using standard targeting assays with either wild-type
preHoloHiPIP (RR) or its RRfiKK signal peptide mutant (KK)
derivative. (C)
55
Fe tracing data obtained using standard targeting
assays with preHoloHiPIP and INVs prepared from wild-type E. coli
(wt ¼ MC4100), a DtatC mutant (B1LK0) and a DtatABCDE mutant
(DADE). Incubation times (min) and ATP addition are also indicated.
All values are given as a percentage of ATP-dependent targeting of

preHoloHiPIP observed using wild-type E. coli INVs, and error bars
indicate the standard deviation observed under the assay conditions
used.
1216 T. Bru
¨
ser et al.(Eur. J. Biochem. 270) Ó FEBS 2003
model Sec substrate MalE. In protease-protection assays as
described in Materials and methods, we observed an
inhibition of 37 ± 5% of MalE targeting, whereas mem-
brane targeting of preHoloHiPIP was inhibited by
12.5 ± 7.5% under the same conditions.
Characterization of membrane-targeted preHoloHiPIP
To determine whether or not the targeted preHoloHiPIP
retained the [4Fe)4S] cluster, EPR spectroscopy was used.
After ATP-dependent preHoloHiPIP targeting, membrane
fractions exhibited EPR signals that were not detectable
when ATP was omitted from the assay (Fig. 6, upper two
traces). The difference of both spectra gave a typical
HiPIP spectrum (Fig. 6, lower trace). Moreover, mem-
brane-associated preHoloHiPIP could be degraded by
thermolysin or proteinase K treatment (Fig. 7A, lane 1).
These data indicate that, during ATP-dependent mem-
brane targeting, preHoloHiPIP retained its iron-sulfur
cofactor, and that the cofactor-binding domain was
exposed on the INV membrane surface, and not trans-
located across the membrane. Apparently, this protease
sensitivity of targeted preHoloHiPIP was induced by INV
binding (Fig. 7A, lanes 1/5), as nontargeted preHolo-
HiPIP could be digested only to mature size by various
proteases (Fig. 7A,B, lanes 5 and 6). This suggested to us

that membrane targeting alters the conformation of
preHoloHiPIP, thereby increasing its protease sensitivity.
Interestingly, when membrane-associated preHoloHiPIP
was treated with protease, a small peptide of similar size
to that of the signal peptide of HiPIP remained protected,
and could be detected with polyclonal antibodies raised
against the precursor of HiPIP (Fig. 7A, lane 1). The
detection of the protease-protected HiPIP fragment in
conjunction with the salt-wash-resistant association of the
precursor with the membranes suggests a membrane-
insertion process. As the C-terminus of preHoloHiPIP
binds the cofactor that is retained during membrane
targeting (Fig. 6), this membrane-insertion process
appears to be mediated by the N-terminus of this protein.
Thus, association of the preHoloHiPIP with the mem-
brane apparently reflects two distinct processes: (a) the
insertion of the N-terminus into the membrane, and (b) a
change in the preHoloHiPIP conformation.
Change in preHoloHiPIP conformation requires ATP
and not insertion of the signal peptide
When older preparations (over 1 week) of in vitro
reconstituted preHoloHiPIP were used instead of fresh
preparations, we found that ATP dependence of target-
ing disappeared (Fig. 7B, lane 4). These ÔagedÕ prepara-
tions exhibited significantly altered optical spectroscopic
properties such that the [4Fe)4S] cluster absorbance
significantly decreased and the absorption maximum in
the UV area was shifted from 283 nm to 275 nm,
indicating that up to 40% of the preparation shifted to a
modified conformation, lacking its cofactor (called pre-

MalHiPIP; Fig. 7C). Nonetheless, even when this pre-
MalHiPIP was targeted to the membrane, a peptide of
the size of the signal peptide became protease-protected,
as in the case of preHoloHiPIP (Fig. 7B, lanes 1 and 2).
These observations suggest that the membrane insertion
per se does not require ATP, but rather ATP is needed for
structural conversion of preHoloHiPIP into a Ôless tightlyÕ
folded, protease-digestible form capable of membrane
insertion. Attempts to purify this ATPase by affinity with
preHoloHiPIP covalently attached to an ÔAminolinkÕ matrix
resulted in isolation of a protein identified as DnaK by
N-terminal amino-acid sequencing. However, preliminary
data obtained using DnaK-deficient mutants suggest that
DnaK is not essential for the translocation of HiPIP in vivo,
and it is also not the ATPase responsible for the above
targeting process in vitro.
Fig. 6. Membrane-associated HiPIP precursor contains its high-poten-
tial iron-sulfur cofactor. Membranes from several independent stand-
ard targeting reactions with or without ATP were pooled and analyzed
by EPR spectroscopy (see Materials and methods). Total INVs
(130 lg) from five standard assays were used to obtain each spectrum.
The spectrum obtained with samples that contained ATP showed that
preHoloHiPIP was targeted to INVs and the [(+ATP) – (–ATP)]
difference spectrum revealed a typical EPR signature that is charac-
teristic of HoloHiPIP. The EPR spectroscopy conditions were as
described in Fig. 3, and spectra were averaged from 10 scans.
Ó FEBS 2003 Membrane targeting of HiPIP (Eur. J. Biochem. 270) 1217
Evidence for membrane targeting of HiPIP
in vivo
The in vitro data described above raised the possibility that

preHiPIP could also be targeted to the membrane in vivo.
However, membranes of wild-type E. coli expressing hip
from its own promoter did not contain readily detectable
preHiPIP (Fig. 2). As membrane-targeted preHoloHiPIP is
highly protease sensitive, we considered that preHiPIP
might be rapidly degraded in membranes. Thus we tested
membrane targeting of preHiPIP-KK in an E. coli strain
that expressed hip from the stronger T7 promoter (Fig. 8).
In such a strain, preHiPIP-KK was readily detected in
membranes that had been washed twice (once in low-salt
and once in high-salt buffer) to ensure high purity.
Moreover, degradation products of HiPIP were also
detected, consistent with the high protease sensitivity of
membrane-targeted HiPIP precursor. Thus, these results
suggest that membrane targeting of HiPIP also occurs
in vivo.
Discussion
Recent studies suggest that Tat substrates need to fold in the
cytoplasm before translocation to their final destinations.
For example, the Tat substrate glucose–fructose oxido-
reductase from Zymomonas mobilis requires correct folding
and cofactor binding for efficient translocation [35]. In
addition, it has been demonstrated that translocation of
c-type cytochromes via the Tat system requires cytoplasmic
attachment of its cofactor, which induces folding [18]. It is
also known that HiPIP can fold and assemble its cofactor in
the cytoplasm [36]. Therefore, we chose HiPIP from
A. vinosum, as a small and well-characterized [4Fe)4S]
cluster-containing protein with a signal sequence exhibiting
the known characteristics of typical Tat substrates (Fig. 1).

Using the E. coli system as the best-characterized bacterial
Tat system, we then established the Tat dependence of
HiPIP translocation in vivo to pave the way for experiments
in vitro.
Fig. 8. Detection of the preHiPIP in membranes of E. coli expressing
the RRfiKK signal peptide mutant derivative of HiPIP. Western-blot
analysis of membrane fractions from E. coli strains BL21 DE3 car-
rying plasmids pEXH15 (KK) or pEXH5 (RR). Membranes were
prepared from crude extract after low-speed centrifugation (30 min,
23 000 g,4°C), ultracentrifugation (143 000 g,2h,4°C), a first wash
in low-salt buffer (20 m
M
Tris/HCl, pH 8.0, followed by ultracentri-
fugation), and a second wash and sonication in high-salt buffer
(200 m
M
NaCl, 50 m
M
MgSO
4
,5m
M
Tris/HCl, pH 8.0, 250 m
M
sucrose, followed by ultracentrifugation) and resuspension in 5 m
M
MgSO
4
/5 m
M

Tris/HCl (pH 8.0)/250 m
M
sucrose for Western-blot
analysis. Each lane corresponds to material obtained from  100 lL
E. coli culture.
Fig. 7. Only correctly folded preHoloHiPIP requires ATP for mem-
brane targeting. In (A) the ATP dependence of membrane insertion of
preHoloHiPIP is shown by immunoblot analysis. Lanes 1–4 show the
analysis of filtered and washed INVs after preHoloHiPIP targeting.
The presence or absence of ATP in the assay mixture is indicated. The
material analyzed in lanes 1 and 2 was further subjected to protease
treatment [200 lgÆmL
)1
thermolysin (TL) for 40 min on ice]. In par-
allel assays, the effect of thermolysin treatment on soluble pre-
HoloHiPIP in STM buffer was tested (lanes 5 and 6). In (B) the same
analysis was carried out with preMalHiPIP. Lanes 1–4 show the
analysis of filtered and washed INVs after preMalHiPIP targeting.
The presence or absence of ATP in the assay mixture is indicated. The
material analyzed in lanes 1 and 2 was further subjected to protease
treatment [200 lgÆmL
)1
proteinase K (PK) for 40 min on ice]. In
parallel assays, the effect of proteinase K treatment on soluble
preMalHiPIP in STM buffer was tested (lanes 5 and 6). (C) Com-
parison of UV-visible spectra of preHoloHiPIP and preMalHiPIP
( 15 l
M
) in STM buffer. Abbreviations: pre, precursor; mat, mature
protein; ppf, protease-protected fragment.

1218 T. Bru
¨
ser et al.(Eur. J. Biochem. 270) Ó FEBS 2003
HiPIP is a
bona fide
Tat substrate in
E. coli
As expected, in vivo translocation of HiPIP required the
tatABC gene products as well as the twin arginine motif in
its signal peptide (Fig. 2). When the twin arginines in the
signal peptide are replaced by lysines, the translocation is
completely blocked, indicating that HiPIP is a substrate of
the Tat system [9]. Some accumulating precursor is
degraded to mature size, probably as the result of
cytoplasmic folding before degradation (compare with
Fig. 7A, lanes 5,6). There are known cases in which a Tat
substrate signal sequence from one bacterial species is not
accepted by the Tat system of another species. One such
example is the glucose–fructose oxidoreductase from
Z. mobilis, which is translocated by the E. coli Tat system
only when its signal peptide is substituted by a signal
peptide from the E. coli Tat substrate TorA [37]. Our results
indicate that A. vinosum HiPIP behaves as an efficient Tat
substrate in E. coli. Thus, the E. coli Tat system is not
restricted to proteins with endogenous Tat signal sequences,
and no general incompatibility between a Tat system and a
heterologous substrate is apparent in this case. The use of
heterologous but natural Tat substrates has the advantage
that results are more likely to be related to general
properties of the Tat system, as substrate-specific targeting

factors can be excluded. In particular, HiPIP is an excellent
tool for studies on the translocation of folded proteins,
because its structure is known, it is small, it has only one
cofactor, it is monomeric, and it is soluble. Its functionality
as an E. coli Tat substrate was the basis for the following
studies.
Folded HiPIP can be targeted to the
E. coli
membranes
We found that the preHoloHiPIP obtained by in vitro
refolding, starting with inclusion bodies and reconstituting
the [4Fe)4S] cluster, could be efficiently targeted in vitro to
inverted membrane vesicles in the presence of ATP and an
ATP-regeneration system (Fig. 4). The membrane-targeted
preHoloHiPIP could not be washed from the vesicles with
high-salt buffers. Furthermore, a 4-kDa fragment of
preHoloHiPIP became protease-protected on targeting
(Fig. 7A,B), suggesting that the targeting of preHoloHiPIP
is a membrane-insertion process. We believe that the
membrane-inserted peptide corresponds to the N-terminus
of preHoloHiPIP for the following reasons: (a) the
N-terminal signal sequence is required for membrane
insertion (Fig. 5A) because matHoloHiPIP cannot be
targeted to INVs; (b) the size of the protease-protected
fragment is that of the signal sequence; (c) the C-terminus of
HiPIP binds the cofactor and has a globular folded
structure, and thus is not available for membrane insertion.
The exact topology of the membrane-inserted N-terminus
of preHoloHiPIP remains to be determined (Fig. 9).
Membrane targeting of preHoloHiPIP appears to be a

highly efficient process that requires ATP. However, this
ATP dependence of membrane targeting vanishes when
preHoloHiPIP loses its cofactor on prolonged storage, i.e.
ÔagingÕ, that results in malfolding or partial unfolding
(preMalHiPIP, Fig. 7B). Moreover, from studies on HiPIP
structure flexibility and folding, it is known that only the
N-terminal half of mature HiPIP can be unfolded without
the loss of cofactor [38]. These facts suggest that ATP is not
required for the membrane-insertion process per se,butit
may rather serve to convert the protein structure, probably
its N-terminus, into an insertion-compatible conformation
(Fig.9).Theincreaseinproteasesensitivityonmembrane
insertion indicates that the mature part of HiPIP remains on
the cytoplasmic side of the membrane, and that its
conformation is affected by membrane insertion. Searching
for the ATPase responsible for membrane targeting of
preHoloHiPIP, we purified the ATP-dependent chaperone
DnaK from the membrane fraction by its affinity for
preHoloHiPIP and release by ATP (see Materials and
Fig. 9. Model for membrane targeting of
HiPIP precursor. It is proposed that mem-
brane targeting of native preHoloHiPIP
requires an ATP-dependent step as indicated
on the left, whereas ÔagedÕ and presumably
partially unfolded preMalHiPIP can be inser-
ted into the membrane without any ATP
requirement as shown on the right. The latter
ATP-independent step may involve additional
specific protein(s) of currently unknown
nature (not shown). Two alternatives for

membrane topology of the N-terminus of
membrane-inserted preHoloHiPIP (with
N-terminal outside or inside shown as con-
tinuous or dotted lines, respectively) are also
indicated. The TatABC-independent mem-
brane insertion observed in vitro is suggested
to precede the TatABC-dependent transloca-
tion observed in vivo, which is indicated with
discontinuous arrows and a question mark.
The fate of unfolded protein is unknown. See
the text for more details.
Ó FEBS 2003 Membrane targeting of HiPIP (Eur. J. Biochem. 270) 1219
methods). However, preliminary experiments using DnaK
–
mutant strains indicate that both in vivo translocation and
in vitro membrane targeting still occur in the absence of this
protein. Thus, although DnaK is a cytoplasmic protein, up
to 25% of which may be membrane-associated [39],
apparently it is not the ATPase observed during membrane
targeting of preHoloHiPIP. Why DnaK recognizes pre-
HoloHiPIP so efficiently in an ATP-dependent manner is at
present unclear, but, considering that this chaperone can
also bind specifically to other Tat signal peptides [40], its role
could be to ensure complete folding of Tat substrates before
membrane insertion.
How could membrane insertion take place?
Proteins may insert into the membrane in either a sponta-
neous or a catalyzed mode. Our observation that a
membrane potential can slow down the membrane-insertion
process (Fig. 4B) suggests that a positively charged segment

of preHoloHiPIP may be transferred across the membrane
during this process. If such a charge translocation across a
hydrophobic membrane takes place, then the process is likely
to require a protein factor for catalysis. Currently known
insertases, SecA and YidC, are thought to accept unfolded
substrates [41], and their involvement in Tat substrate
targeting has previously been ruled out in some cases
[42,43]. It is unlikely that the ATPase SecA accounts for
the membrane-insertion process described here, as no ATP
dependence is observed in the case of preMalHiPIP (see
Fig. 7B). Further evidence for the independence of the
targeting reaction from SecA was obtained by including
3m
M
sodium azide in the assay, a well-known SecA
inhibitor. At this concentration of azide, we observed only
about 10–15% inhibition of the targeting process, whereas
the targeting of the model Sec substrate MalE was about
37% inhibited. The higher azide sensitivity of Sec-dependent
translocation and the ATP independence of preMalHiPIP
targeting argue against the involvement of SecA in the
targeting of preHoloHiPIP, suggesting that the ATP
dependence may be due to another factor, which is required
in the case of the correctly folded substrate only. However,
SecA inhibition by azide is very leaky in vitro and therefore
we do not rule out at this stage that SecA may be involved in
the ATP-dependent targeting step of folded HiPIP. Thus, the
molecular nature of the protein factor(s) required for
preHoloHiPIP membrane insertion remains to be deter-
mined.

Does membrane targeting occur before
the translocation?
The fact that the twin arginine motif and the known Tat-
system components are not required for membrane inser-
tion of preHoloHiPIP indicates that the insertion process
seen here is not confined to Tat substrates. If Tat substrates
such as preHoloHiPIP are among the natural substrates of
this membrane-targeting pathway, then Tat-substrate
recognition and translocation by specific Tat-system compo-
nents must occur after membrane insertion, and what we
observe in vitro is the accumulation of the targeting
intermediate in the membrane. This hypothesis has to be
taken into consideration, because (a) HiPIP is a Tat
substrate in vivo (Fig. 2), (b) membrane targeting is adapted
to the folded state of the protein with an ATP-consuming
step (Fig. 7), (c) Tat-independent membrane targeting of
Tat substrates in vivo has been documented previously
[8,44], and (d) translocation after membrane targeting has
been described for the thylakoidal system [45]. As no in vitro
translocation of folded and cofactor-containing natural Tat
substrates could be demonstrated so far, we do not exclude
the alternative that the observed membrane targeting may
not be coupled to translocation. In this case, the data would
suggest that there exists a pathway for the biogenesis of
membrane proteins, which are allowed to fold before
membrane targeting.
In summary, we found that preHoloHiPIP can be inserted
into the cytoplasmic membrane and that only correctly
folded preHoloHiPIP requires ATP for this process. To our
knowledge, this is the first description of membrane targeting

of a [4Fe)4S] cluster-containing folded protein. This mem-
brane-targeting pathway may be of importance for the
biogenesis of membrane-bound redox proteins or for the
targeting of folded Tat substrates. The molecular basis of this
process is currently under investigation.
Acknowledgements
We are grateful to Tracy Palmer for providing us with various Tat
deletion mutants and to Ute Lindenstrauß for excellent technical
assistance. We are indebted to Jan R. Andreesen, Donna M. Gordon,
Bo Hou, Ralf Bernd Klo
¨
sgen, Debkumar Pain, Mecky Pohlschro
¨
der,
Philip Rea and Carsten Sanders for many valuable discussions and
help. This work was supported by grants DOE 91ER20052 and NIH
GM38237 to F. D. and by grant BMBF-LPD 9901/8-14 from the
German Academy of Natural Scientists Leopoldina to T. B.
References
1. Pugsley, A.P. (1993) The complete general secretory pathway in
Gram-negative bacteria. Microbiol. Rev. 57, 50–108.
2. Berks, B.C., Sargent, F. & Palmer, T. (2000) The Tat protein
export pathway. Mol. Microbiol. 35, 260–274.
3. Mu
¨
ller, M., Koch, H G., Beck, K. & Scha
¨
fer, U. (2001) Protein
traffic in bacteria: multiple routes from the ribosome to and across
the membrane. Prog. Nucleic Acid Res. 66, 107–157.

4. Robinson, C. & Bolhuis, A. (2001) Protein targeting by the twin
arginine translocation pathway. Nat. Rev. Mol. Cell. Biol. 2,
350–356.
5. Wexler,M.,Sargent,F.,Jack,R.L.,Stanley,N.R.,Bogsch,E.G.,
Robinson, C., Berks, B.C. & Palmer, T. (2000) TatD is a cyto-
plasmic protein with DNase activity. J. Biol. Chem. 275, 16717–
16722.
6. Sargent, F., Bogsch, E.G., Stanley, N.R., Wexler, M., Robinson,
C., Berks, B.C. & Palmer, T. (1998) Overlapping functions of
components of a bacterial Sec-independent protein export path-
way. J. Biol. Chem. 274, 36073–36082.
7. Sargent, F., Gohlke, U., de Leeuw, E., Stanley, N., Palmer, T.,
Saibil, H.R. & Berks, B.C. (2001) Purified components of the
Escherichia coli Tat protein transport system form a double-
layered ring structure. Eur. J. Biochem. 268, 3361–3367.
8. Bolhuis, A., Mathers, J.E., Thomas, J.D., Barrett, C.M.L. &
Robinson, C. (2001) TatB and TatC form a functional and
structural unit of the twin-arginine translocase from Escherichia
coli. J. Biol. Chem. 276, 20213–20219.
9. Stanley, N.R., Palmer, T. & Berks, B.C. (2000) The twin arginine
consensus motif of tat signal peptides is involved in Sec-independent
1220 T. Bru
¨
ser et al.(Eur. J. Biochem. 270) Ó FEBS 2003
protein targeting in Escherichia coli. J. Biol. Chem. 275, 11591–
11596.
10. Hinsley, A.P., Stanley, N.R., Palmer, T. & Berks, B.C. (2001) A
naturally occurring bacterial Tat signal peptide lacking one of the
ÔinvariantÕ arginine residues of the consensus targeting motif.
FEBS Lett. 497, 45–49.

11. Cristo
´
bal, S., de Gier, J W., Nielsen, H. & von Heijne, G. (1999)
Compeition between Sec- and Tat-dependent protein transloca-
tion in Escherichia coli. EMBO J. 18, 2982–2990.
12. Bogsch, E., Brink, S. & Robinson, C. (1997) Pathway specificity
for a DpH-dependent precursor thylacoid lumen protein is
governed by a ÔSec-avoidanceÕ motif in the transfer peptide and a
ÔSec-incompatibleÕ mature protein. EMBO J. 16, 3851–3859.
13. Bru
¨
ser,T.,Deutzmann,R.&Dahl,C.(1998)Evidenceagainstthe
double-arginine motif as the only determinant for protein trans-
location by a novel Sec-independent pathway in Escherichia coli.
FEMS Microbiol. Lett. 164, 329–336.
14. Santini, C L., Bernadac, A., Zhang, A., Ize, B., Blanco, C. & Wu,
L F. (2001) Translocation of jellyfish green fluorescent protein via
the Tat system of Escherichia coli and change of its periplasmic
localization in response to osmotic up-shock. J. Biol. Chem. 276,
8159–8164.
15. Thomas, J.D., Daniel, R.A., Errington, J. & Robinson, C. (2001)
Export of active green fluorescent protein to the periplasm by the
twin-arginine translocase (Tat) pathway in Escherichia coli. Mol.
Microbiol. 39, 47–53.
16. Rodrigue, A., Chanal, A., Beck, K., Mu
¨
ller, M. & Wu, L F.
(1999) Co-translocation of a periplasmic enzyme complex by a
hitchhiker mechanism through the bacterial Tat pathway. J. Biol.
Chem. 274, 13223–13228.

17. Mu
¨
hlenhoff, U. & Lill, R. (2000) Biogenesis of iron-sulfur proteins
in eukaryotes: a novel task of mitochondria that is inherited from
bacteria. Biochim. Biophys. Acta. 1459, 370–382.
18. Sanders, C., Wethkamp, N. & Lill, H. (2001) Transport of cyto-
chrome c derivatives by the bacterial Tat protein translocation
system. Mol. Microbiol. 41, 241–246.
19. Yahr, T.L. & Wickner, W.T. (2001) Functional reconstitution of
bacterial Tat translocation in vitro. EMBO J. 20, 1–8.
20. Alami, M., Trescher, D., Wu, L.F. & Mu
¨
ller, M. (2002) Separate
analysis of twin-arginine translocation (Tat)-specific membrane
binding and translocation in Escherichia coli. J. Biol. Chem. 277,
20499–20503.
21. Carter, C.W., Kraut, J., Freer, S.T., Xuong, N H., Alden, R.A. &
Bartsch, R.G. (1974) Two-Angstrom crystal structure of oxidized
Chromatium high potential iron protein. J. Biol. Chem. 249, 4212–
4225.
22. Banci, L., Bertini, I., Dikiy, A., Kastrau, D.H.W., Luchinat, C. &
Sompornpisut, P. (1995) The three-dimensional solution structure
of the reduced high-potential iron-sulphur protein from Chroma-
tium vinosum through NMR. Biochemistry 34, 206–219.
23. Bru
¨
ser, T., Tru
¨
per, H.G. & Dahl, C. (1997) Cloning and sequen-
cing of the gene encoding the high potential iron-sulphur protein

(HiPIP) from the purple sulphur bacterium Chromatium vinosum.
Biochim. Biophys. Acta 1352, 18–22.
24. Natarajan, K. & Cowan, J.A. (1997) Identification of a key
intermediate of relevance to iron-sulphur cluster biosynthesis.
mechanism of cluster assembly and implications for protein fold-
ing. J. Am. Chem. Soc. 119, 4082–4083.
25. Bogsch, E.G., Sargent, F., Stanley, N.R., Berks, B.C., Robinson,
C. & Palmer, T. (1998) An essential component of a novel bac-
terial protein export system with homologues in plastids and
mitochondria. J. Biol. Chem. 273, 18003–18006.
26. Bolhuis, A., Bogsch, E.G. & Robinson, C. (2000) Subunit inter-
actions in the twin-arginine translocase complex of Escherichia
coli. FEBS Lett. 472, 88–92.
27. Landers, J.W. & Zak, B. (1958) Determination of serum copper
andironinasinglesmallsample.Am.J.Clin.Pathol.29, 590–592.
28. Mu
¨
ller, M. & Blobel, G. (1984) In vitro translocation of bacterial
proteins across the plasma membrane of Escherichia coli. Proc.
NatlAcad.Sci.USA81, 7421–7425.
29. Watanabe, M. & Blobel, G. (1989) Binding of a soluble factor of
Escherichia coli to preproteins does not require ATP and appears
to be the first step in protein export. Proc. Natl Acad. Sci. USA 86,
2248–2252.
30. Laemmli, U.K. (1970) Cleavage of structural proteins during the
assembly of the head of bacteriophage T4. Nature (London) 227,
680–685.
31. Lowry, O.H., Rosenbrough, N.J., Farr, A.L. & Randall, R.J.
(1951) Protein measurement with the Folin phenol reagent. J. Biol.
Chem. 193, 265–275.

32. Matsudaira, P. (1987) Sequence from picomole quantities of
proteins electroblotted onto polyvinylidene difluoride membranes.
J. Biol. Chem. 262, 10035–10038.
33. Bartsch (1971) High potential iron proteins: bacterial. Methods
Enzymol. 23, 644–649.
34. Dunham, W.R., Hagen, W.R., Fee, J.A., Sands, R.H., Dunbar,
J.B. & Humblet, C. (1991) An investigation of Chromatium vino-
sum high-potential iron-sulphur protein by EPR and Mossbauer
spectroscopy: evidence for a freezing-induced dimerization in
NaCl solutions. Biochim. Biophys. Acta 1079, 253–262.
35. Halbig, D., Wiegert, T., Blaudeck, N., Freudl, R. & Sprenger,
G.A. (1999) The efficient export of NADP-containing glucose-
fructose oxidoreductase to the periplasm of Zymomonas mobilis
depends both on an intact twin-arginine motif in the signal peptide
and on the generation of a structural export signal induced by
cofactor binding. Eur. J. Biochem. 263, 543–551.
36. Agarwal,A.,Tan,J.,Eren,M.,Tevelev,A.,Lui,S.M.&Cowan,
J.A. (1993) Synthesis, cloning and expression of a synthetic gene
for high potential iron protein from Chromatium vinosum. Bio-
chem. Biophys. Res. Commun. 197, 1357–1362.
37. Blaudeck,N.,Sprenger,G.A.,Freudl,R.&Wiegert,T.(2001)
Specificity of signal peptide recognition in tat-dependent bacterial
protein translocation. J. Bacteriol. 183, 604–610.
38. Bertini, I., Cowan, J.A., Luchinat, C., Natarajan, K. & Piccoli, M.
(1997) characterization of a partially unfolded high potential iron
protein. Biochemistry 36, 9332–9339.
39. Bukau, B., Reilly, P., McCarty, J. & Walker, G.C. (1993)
Immunogold localization of the DnaK heat shock protein in
Escherichia coli cells. J. Gen. Microbiol. 139, 95–99.
40. Oresnik, I.J., Ladner, C.L. & Turner, R.J. (2001) Identification

of a twin-arginine leader-binding protein. Mol. Microbiol. 40,
323–331.
41. de Gier, J W. & Luirink, J. (2001) Biogenesis of inner membrane
proteins in Escherichia coli. Mol. Microbiol. 40, 314–322.
42. Santini, C L., Ize, B., Chanal, A., Mu
¨
ller, M., Giordano, G. &
Wu, L F. (1998) A novel sec-independent periplasmic
protein translocation pathway in Escherichia coli. EMBO J. 17,
101–112.
43. Moore, M., Harrison, S., Peterson, E.C. & Henry, R. (2000)
Chloroplast Oxa1p homolog Albino3 is required for post-trans-
lational integration of the light harvesting chlorophyll-
binding protein into thylakoid membranes. J. Biol. Chem. 275,
1529–1532.
44. Sargent, F., Stanley, N.R., Berks, B.C. & Palmer, T. (1999) Sec-
independent protein translocation in Escherichia coli:adistinct
and pivotal role for the TatB protein. J. Biol. Chem. 274, 36073–
36082.
45. Musser, S.M. & Theg, S.M. (2000) Characterization of the early
steps of OE17 precursor transport by the thylakoid delta pH/Tat
machinery. Eur. J. Biochem. 267, 2588–2598.
Ó FEBS 2003 Membrane targeting of HiPIP (Eur. J. Biochem. 270) 1221