Báo cáo khoa học: Trigger factor interacts with the signal peptide of nascent Tat substrates but does not play a critical role in Tat-mediated export pptx

Trigger factor interacts with the signal peptide of nascent Tat
substrates but does not play a critical role in Tat-mediated export
Wouter S. P. Jong
1
, Corinne M. ten Hagen-Jongman
1
, Pierre Genevaux
2
, Josef Brunner
3
, Bauke Oudega
1
and Joen Luirink
1
1
Department of Molecular Microbiology, Institute of Molecular Cell Biology, Vrije Universiteit, Amsterdam, the Netherlands;
2
Department of Microbiology and Molecular Medicine, Centre Me
´
dical Universitaire, Geneva, Switzerland;
3
Institute of Biochemistry,
Eidgeno
¨
ssische Technische Hochschule Zu
¨
rich, Zu
¨
rich, Switzerland
Twin-arginine translocation (Tat)-mediated protein trans-
port across the bacterial cytoplasmic membrane occurs only

after synthesis and folding of the substrate protein that
contains a signal peptide with a characteristic t win-arginine
motif. This implies that p remature contact between the T at
signal peptide a nd the T at translocon in the membrane must
be prevented. We used site-specific photo-crosslinking to
demonstrate that t he signal peptide of nascent T at proteins
is in close proximity to the c haperone and peptidyl-prolyl
isomerase trigger f actor (TF). The contact with TF was
strictly dependent on the context of the translating ribosome,
started early in biogenesis when the nascent chain left the
ribosome near L 23, and p ersisted until the c hain reached i ts
full length. Despite this exclusive and prolonged contact,
depletion o r o verexpression of TF had little effect on the
kinetics and efficiency of the Tat export process.
Keywords: Escherichia coli; protein targeting; signal peptide;
trigger f actor; twin-arginine translocation.
In Escherichia coli, most proteins that reside in the
periplasmic space are synthesized as preproteins with a
cleavable N-terminal signal peptide that mediates targeting
to the inner membrane. Signal peptides classically have a
tri-partite structure with a positively charged N-region, a
hydrophobic core, and a polar C-region that contains the
signal peptidase cleavage site [1]. The majority of periplas-
mic proteins are targeted to the main protein-conducting
channel, the SecYEG complex, via the post-translational
SecB/SecA pathway (reviewed in [2]).
Recently, the cytosolic chaperone and folding catalyst
trigger factor (TF) was shown to have a significant impact
on the efficiency of Sec-mediated transport. Inactivation of
the gene e ncoding TF increased the rate of transport and

suppressed the requirement for the chaperone and targeting
factor SecB, whereas overproduction of TF impeded
transport [ 3]. TF is in part associated with the ribosomal
protein L23 that is located near t he major n ascent ch ain
exit site [4]. In vitro crosslinking studies showed that TF
can be crosslinked to a variety of n ascent polypeptides
when they emerge from the ribosome near L23 [5–9].
Interestingly, L23 also serves as a docking s ite f or the
bacterial signal recognition particle (SRP) that delivers
preproteins at the SecYEG complex in a cotranslational
targeting mechanism [9]. Whether or not TF c ontrols the
entry o f proteins i nto the SRP pathway is not fully clear
(reviewed in [10]).
The twin -arginine translocation ( Tat) pathway has been
identified as a second post-translational targeting/trans-
location pathway t hat operates i ndependently of the Sec
pathway (reviewed in [11]). In contrast to the Sec pathway,
the Tat pathway has the striking a bility to mediate the
export o f s ubstrates that have acquired a fully folded or even
oligomeric confo rmation in the c ytoplasm. T at substrates
possess a signal peptide of the ÔclassicalÕ tri-partite stru cture
but including a h ighly conserved ( S/T)RRxFLK consensus
motif between the N-region and the hydrophobic core [ 12].
This motif provides specificity for the Tat machinery
consisting of the i ntegral membrane proteins TatA/E, TatB
and TatC [11].
Little is known about the molecular mechanism of
targeting and export of Tat-dependent proteins. In partic-
ular, information on the generic and s pecific interactions of
the T at signal peptide and mature domain with targeting

factors, chaperones a nd folding catalysts is scarce. The
cytosolic DmsD protein was s hown to have affinity for
immobilized Tat signal peptides of both dimethylsulfoxide
reductase (DmsA) and trimethylamine N-oxide reductase
(TorA) [13] and for the TatB/TatC components [14],
suggesting a r ole for DmsD i n guiding substrates to the
Tat machinery. H owever, recent in vivo studies suggested
that DmsD is not required f or targeting but rather has a
chaperone-like function in the assembly of certain Tat
proteins [15].
Correspondence to J. Luirink, Department of Molecular Micro-
biology, Institute of Molecular Cell Biology, Vrije Universiteit,
De Boelelaan 1087, 1081 HV Amst erdam, the Netherlands.
Fax: +31 20 4446979, Tel.: +31 2 0 4447175,
E-mail:
Abbreviations: HA, hemagglutinin; Tat, twin-arginine translocation;
TF, trigger factor; OmpA, outer membrane protein A; TorA, tri-
methylamine N-oxide reductase; (Tmd)Phe-tRNA
sup
,
L
-[3-(trifluoro-
methyl)-3-diazirin-3H -yl]phenylalanine-tRNA
sup
; IMVs, inverted
inner membrane vesicles; Ffh, fifty-four homologue; SRP, signal
recognition particle.
(Received 2 8 July 2 004, revised 6 October 2004,
accepted 18 October 2004)
Eur. J. Biochem. 271, 4779–4787 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04442.x

Here, we have used an unbiased in vitro translation and
photo-crosslinking appro ach to probe the molecular inter-
actions of model Tat substrates during synthesis and p rior
to tar geting to t he Tat machinery. We f ound that the signal
peptide of Tat-dependent proteins is extensively a nd exclu-
sively crosslinked t o r ibosomal components and TF during
synthesis. Interestingly, TF was found crosslinked until late
in translation but only in the context of the translating
ribosome. However, in vivo experiments revealed only a
small effect of TF on the kinetics and efficiency of Tat-
mediated export.
Experimental procedures
Strains, plasmids and media
E. coli K-12 strains and plasmids used in this study are listed
in Table 1 . Strains were routinely grown in M9-medeu m
[16] containing 0.1% casaminoacids (Difco, Detroit, MI,
USA). Where appropriate, streptomycin ( 50 lgÆmL
)1
),
chloramphenicol ( 15 lgÆmL
)1
), kanamycin (30 lgÆmL
)1
),
spectinomycin (50 lgÆmL
)1
) a nd ampicillin (100 lgÆmL
)1
)
were added to the medium.

Reagents and sera
Restriction enzymes and t he Expand-Long template PCR
system were supplied by Roche M olecular Biochemicals
GmbH (Mannheim, Germany). T4-DNA ligase was from
Epicentre Technologies ( Madison, WI, USA). Megashort
T7 transcription kit was f rom A mbion (Austin, TX, USA).
[
35
S]Methionine and protein A–Sepharose were obtained
from Amersham Biosciences (Uppsala, Sweden). All other
chemicals were supplied by S igma-Aldrich (Steinheim,
Germany). Antisera against L 23 and L29, TF, SufI, and
OmpA were provided by R. Brimacombe (Max Planck
Institute for Molec ular G enetics, Berlin, Germany), W.
Wickner (Dartmouth Medical School, Hannover, NH,
USA), T . P almer ( University of East Anglia, Norwich,
UK), and J. W. de Gier (Stockholm University, Sweden),
respectively. The rabbit polyclonal antiserum against the
human influenza hemagglutinin (HA)-epitope was from
Sigma.
Plasmid construction
Plasmid pC4Meth-100TorA/P2 was constructed by PCR,
using pTorA/P2 [ 17] a s a template and the primers
RRTorA-SacI-fw (5¢-GCGCG
GAGCTCAAGAAGGA
AGAAAAATAATGAAC-3¢, SacI site underlined) and
TorA/Lep2-BamHI-rv (5¢-GCAT
GGATCCCGCGCGC
TTGATGTAATC-3¢, BamHI site underlined). The result-
ing PCR fragment was cloned into pC4Meth [5] u sing the

SacI/BamHI sites. Amber (TAG) codons were then incor-
porated at position 13 or 24 via nested PCR as described
[18], resulting in pC4Meth-100TorA/P2TAG13 a nd
pC4Meth-100TorA/P2TAG24. Plasmids pC4Meth-57SufI,
pC4Meth93-SufI and pC4Meth-SufIHA (encoding SufI
with a C-terminal HA-epitope, preceded by a Pro-Gly-Gly
spacer) w ere constructed b y PCR using pNR30 (gift
from T. Palmer) as a t emplate. The fo rward primer w as
SufI-EcoRI-fw (5¢-GCCG
GAATTCTAATATGTCACTC
AGTCGGCGTC-3¢, Eco RI site underlined). The reverse
primers were 51SufI-BamHI-rv (5¢-ACGC
GGATCCAG
TCATAAACAGCGGTTGC-3¢, Bam HI site un derlined),
87SufI-BamHI-rv (5 ¢-ACGC
GGATCCAACATCGTCGC
CCTTCCA-3¢, BamHI site underlined) and SufIHA-
XbaI+ClaI-rv (5¢-ACTG
ATCGATCTAGATTACGCAT
AGTCAGGAACATCGTATGGGTAGCCGCCTGGCG
GTACCGGATTGACCAAC- 3¢, ClaI site underlined,
XbaI site in italics, HA-epito pe sequence in boldface). The
resulting fragments were cloned into pC4Meth using the
EcoRI/BamHI or EcoRI/ClaI restriction sites where appro-
priate. The amber codon at position 8 was incorporated
via nested PCR, resulting in pC4Meth-57SufITAG8,
pC4Meth-93SufITAG8 and pC4Meth-SufIHATAG8. The
in vivo expression plasmid pBAD18-SufIHA was construc-
ted as follows. First, the EcoRI/XbaI f ragment from
pNR30, including the SufI coding region and the first

18 bp upstream of the ATG-start codon, was c loned into
pBAD18 [19]. The resulting plasmid pBAD18-SufI w as
then used as a template i n PCR using the primers SufI-
EcoRI-fw and SufIHA-XbaI+ClaI-rv (see above).
Finally, the AatII/XbaI f ragment of t he ob tained PCR
product was inserted into pBAD18-SufI. Nucleotide
sequences were confirmed by semi-automated DNA
sequencing.
In vitro
transcription, translation and crosslinking
Truncated mRNA was prepared as described previously
[20] from HindIII linearized pC4Meth-100TorA/P2,
pC4Meth-57SufI o r p C4Meth-93SufI derivative plasmids.
Full-length SufIHATAG8 mRNA was prepared from ClaI
linearized pC4Meth-SufIHATAG8. In vitro translation,
photo-crosslinking and sodium carbon ate e xtraction were
carried out as described [18,21]. Samples were analyzed
Table 1. Bacterial strains and plasmids use d in this study.
Strain/plasmid Relevant genotype Reference
MC4100 F’araD139D(argF-lac)U169
rpsL150 relAI flb5301
ptsF25 rbsR
[40]
MC4100Dtig MC4100Dtig::Cm
r
[33]
MC4100DdnaKdnaJ MC4100DdnaKdnaJ
::Kan
r
thr::Tn10

[33]
MC4100DtigDdnaKdnaJ MC4100Dtig::Cm
r
D
dnaKdnaJ::Kan
r
thr::Tn10
[33]
MC4100DtatA/E MC4100DtatADtatE [41]
MC4100DtatB MC4100DtatB [41]
MC4100DtatC MC4100DtatC::XSpec
r
[26]
HDB37 MC4100araD [3]
pC4Meth-100TorA/
P2TAG13
pC4Meth, 94torA/
P2TAG13
This study
pC4Meth-100TorA/
P2TAG24
pC4Meth, 94torA/
P2TAG24
This study
pC4Meth-57SufITAG8 pC4Meth, 51sufITAG8 This study
pC4Meth-93SufITAG8 pC4Meth, 87sufITAG8 This study
pC4Meth-SufIHATAG8 pC4Meth, sufIHATAG8 This study
pBAD18-SufIHA pBAD18, sufIHA This study
pJH42 pBAD18, tig [3]
4780 W. S. P. Jong et al.(Eur. J. Biochem. 271) Ó FEBS 2004

directly by SDS/PAGE or immunoprecipitated first using
3-fold the amount used for direct analysis.
Pulse-chase analysis
Strain MC4100 and its Dtig , DdnaKdnaJ, Dtig DdnaKdnaJ
and DtatC mutant derivatives, all harboring pBAD18-
SufIHA, were grown overnight in M9-medium contain ing
0.4% glucose, diluted to a n a ttenuance a t 6 60 nm (D
660
)of
0.05infreshmediumandgrowntoaD
660
of 0.35. Strains
HDB37 and MC4100DtatA/E, both harboring p JH42, were
grown overnight in M9-medium, diluted to a D
660
of 0.05 in
fresh M9-medium and grown to an D
660
of 0.3.
Upon reaching the a ppropriate D
660
, cells were washed
and r esuspended in M9-medium containing a cysteine- and
methionine-free amino acid mix. After recovery for 15 min
(pBAD18-SufIHA harboring strains) o r 90 min (pJH42
harboring strains) at the appropriate temperatures, cells were
pulse-labeled with 10 lCiÆmL
)1
[
35

S]methionine for 1 min
andchasedwith2m
M
cold methionine for the times
indicated. To stop the chase, cells were precipitated w ith
10% trichloroacetic acid at 4 °C. Sample s were analyzed
either directly or upon immunoprecipitation by SDS/PAGE.
Sample analysis
Radiolabeled proteins were visualized by phosphor imagin g
using a Molecular Dynamics PhosphorImager 473 and
quantified using the
IMAGEQUANT
software from Molecular
Dynamics/Amersham Biosciences.
Results
The TorA signal peptide is close to trigger factor
early in biogenesis
Tat preproteins fold in the c ytosol, prior to export by t he
Tat machinery in the inner membrane that specifically
recognizes the Tat signal peptide [11,22]. It has been
suggested that t his signal p eptide is sheltered during
synthesis a nd folding by generic or specific f actors in the
cytosol to prevent premature interactions with the Tat
translocon [23,24]. The molecular interactions of the signal
peptides of model Tat substrates early in biosynthesis were
studied using a n in vitro translation and crosslinking
approach. Nascent chains of TorA/P2, a strictly Tat-
dependent chimera comprising the signal peptide of TorA
fused to t he periplasmic P2 domain of leader p eptidase [17],
were generated from truncated mRNA to a length of 100

amino acid residues in a cell- and membrane-free E. coli
lysate without addition of any purified proteins. The nascent
chains were radiolabeled with [
35
S]methionine. To specific-
ally probe the m olecular environment of the TorA signal
peptide, TAG-stop codons were incorporated in 100TorA/
P2 either at position 13, two residues downstream of
the c onserved a rginine p air, or within the h ydrophobic
core at position 24 (Fig. 1A). The TAG-codons were
suppressed during in vitro synthesis by t he addition of
L
-[3-
(trifluoromethyl)-3-diazirin-3H-yl]phenylalanine-tRNA
sup
[(Tmd)Phe-tRNA
sup
] which carries a photo-reactive probe.
After translation, one half of each sample was irradiated
AB C
Fig. 1. Photo-crosslinking to the signal peptide of nascent TorA/P2. (A) S chematic representation of nascent 100TorA/P2. The T orA signal peptide
is ind icated as a solid line. Posi tions of the conserved twin-arginine motif ( RR) and the sto p codons (TAG) t hat are suppressed with (Tmd )Phe-
tRNA
sup
are indicated. (B) In vitro translation o f 100TorA/P2TAG13. After translation, samples were irradiated with UV-light to induce
crosslinking or kept in th e dark as indicated. UV-irradiated ribosome-nascent chain complexes were immunoprecipitated (IP) with TF a ntiserum as
indicated. Prior to crosslinking, one sample was s plit into equal aliquots and incubated w ith 2 m
M
puromycin (Puro), 2 m
M

puromycin and 0.5
M
potassium acetate (HS), o r mock-treated with incubation buffer at 37 °Cfor10min.(C)In vitro translation of 100TorA/P2TAG13 and 100TorA/
P2TAG24 as f or (B ), carried out in the p resence o r a bsence of IMVs. Samples with IMVs were extracted with sodium carbonate and t he re sulting
carbonate-pellet ( p) and -supernatant (s) fractions were analyzed. Crosslinked p roducts, nascent c hains (NC), peptidyl-tRNA (*) and molecular
mass markers ( at the left side of the pane ls in kDa) are indicated.
Ó FEBS 2004 Role of trigger factor in Tat-mediated export (Eur. J. Biochem. 271) 4781
with UV-light to induce crosslinking, whereas the other half
was kept in the dark to serve as a negative control.
The TAG-codons at both positions wer e efficiently
suppressed by (Tmd)Phe-tRNA
sup
(data not shown),
resulting in nascent T orA/P2 of the expected m olecular
mass carrying t he photo-reactive probe at the indicated
position. UV-irradiation o f 1 00TorA/P2TAG13 resulted in
two c rosslinked products of  70–80 kDa (Fig. 1B, l ane 2).
Both adducts represented crosslinks to the cytosolic
chaperone TF as shown by immunoprecipitation ( Fig. 1B,
lane 3). TF was also crosslinked to position 24 within the
hydrophobic core (Fig. 1C, l ane 8) but with a different ratio
of the 70 and 80 kDa adducts (Fig. 1C, compare lanes 2 and
8). The observation that crosslink ing of nascent c hains to
TF results in a double banded pattern has been made
previously [5,8] but is not yet understood.
To investigate w hether crosslinking of TF to 100TorA/P2
is dependent on the context of the ribosome, nascent
100TorA/P2TAG13 was released from the r ibosome with
puromycin or puromycin in a Ôhigh saltÕ buffer after
translation but prior to c rosslinking. Both t reatments

diminished crosslinking to TF (Fig. 1B, lane 5 a nd 6),
indicating that association w ith the ribosome is crucial for
the interaction with TF.
Molecular i nteractions of 100TorA/P2TAG13 a nd
100TorA/P2TAG24 were also investigated in the presence
of inverted inner membrane vesicles (IMVs) that were
isolated from an E. coli MC4100 wild-type strain. After
translation a nd UV-irradiation, samples were extracted with
sodium carbonate to separate the membrane integrated from
the peripheral membrane a nd soluble p roteins. Using
100TorA/P2TAG13, no obvio us changes in crosslinking
pattern a ppeared compared to the situation when mem-
branes were notpresent (Fig. 1C, lanes 1–6). TF continued to
be the major crosslinking partner and no crosslinking
products were detected in the integral membrane f raction
(Fig. 1C, lanes 4 and 6 ). Interestingly, u pon addition of
IMVs, pos ition 2 4 o f the TorA signal peptide was found to
specifically crosslink, in addition to TF, two low molecular
mass proteins of  7kDaand 17 kDa (Fig. 1C, lane 10).
These adducts were detected in the supernatant fraction a fter
carbonate extraction, indicating that the crosslinked p artners
are peripheral membrane proteins and not one of the known
Tat p roteins [25]. The adducts could not be immunoprecip-
itated with various antisera tested and remain to be identified.
Taken together, the twin-arginine motif and the hydro-
phobic core region of the TorA signal p eptide are adjacent
to TF early during biogenesis. Additional c ontacts with two
yet unknown peripheral m embrane proteins a ppeared
restricted to th e hydrophobic core region. Other cytosolic
factors with affinity for signal peptides w ere not detectably

crosslinked.
The SufI signal peptide is in close proximity to TF, L23
and L29 early in biogenesis
To investigate whether the observed contact o f TF with the
TorA signal peptide is generic for Tat substrates, SufI was
included i n our crosslinking stud ies. SufI belongs t o the
multicopper oxidase superfamily but does not contain
copper cofactors [26]. It has been used extensively as a
model Tat substrate [26–29]. SufI nascent chains were
generated (as described above for 100TorA/P2), carrying a
photo-reactive probe at positio n 8, t wo amino acids
downstream from the conserved arginine pair ( Fig. 2A).
Nascent chains of 5 7 a nd 93 amino acids were synthesized.
93SufITAG8 is comparable to 100TorA/P2TAG13 with
respect to the position of the photo-reactive probe relative to
the peptidyl-transferase center (compare Figs 2A and 1A).
57SufITAG8 w as analyzed to monitor t he earliest inter-
actions in nascent SufI (Fig. 2A).
A
B
Fig. 2. Photo-crosslinking to the signal p eptide of nascent SufI. (A)
Schematic represen tation o f nascent Su fI c onstructs. T he SufI signal
peptide is ind icated as a solid line. Positions of the conserved twin-
arginine motif (RR) and the stopcodons (TAG) that are suppressed
with (Tmd)Phe-tRNA
sup
are indicated. (B) In vitro tran slation of
57SufITAG8 an d 93SufITAG8. After translation, one hal f of each
sample was irradiated with UV-lig ht to induce crosslinking and one
half was kep t in t he dark. UV-irradiated ribosome-nascent ch ain

complexes were i mmunoprecipitated (IP) with antiserum aga inst TF,
L23 or L29 as indicated. Molecular mass markers (kDa) are indicated
at th e left side of t he panels.
4782 W. S. P. Jong et al.(Eur. J. Biochem. 271) Ó FEBS 2004
Upon irradiation o f 57SufITAG8 a nd 93SufITAG8 w ith
UV-light, adducts o f  65–75 kDa were detected (Fig. 2B,
lanes 2 and 7) that represented crosslinking to TF, as shown
by immunoprecipitation (Fig. 2B, lanes 5 and 10). Using
57SufITAG8, two adducts of lower molecular mass w ere
identified as the ribosomal subunits L23 (Fig. 2B, l ane 3)
and L29 ( Fig. 2 B, lane 4). L23 a nd L29 are l ocated near the
major r ibosomal exit site [30]. 93SufITAG8 was also found
to crosslin k to L23 an d L29 (Fig. 2B, lanes 8 and 9), but to a
much lower e xtent. In con trast, t he 93-mer crosslinked
much more strongly to TF than the 57-mer (Fig. 2B,
compare lanes 2 and 7) .
Together, t he data suggest that nascent SufI leaves the
ribosome via the major ribosomal exit site and that, upon
extension o f the nascent c hain, the SufI signal peptide moves
away from L23 and L29 and interacts w ith TF. Further-
more, the combined data obtained with nascent TorA/P2
and SufI constructs strongly suggest that interaction of TF
with the signal peptide early i n biosynthesis i s generic for
Tat substrates.
The SufI signal peptide is close to TF until late
in translation
In the crosslinking experiments described above we
showed th at TF interacts with t he signal peptide of
relatively short nascent Tat substrates in which the signal
peptide is barely exposed. T o investigate the influence of

nascent chain l ength on the ability of TF to crosslink to
the Tat signal peptide we made u se of a full-length
version of SufI that carries an immunogenic HA-epitope
at its C-terminus (SufIHA), and a photo-reactive probe
at position 8 in the signal peptide (SufIHATAG8).
Translation o f this construct resulted in a ladder of
distinct truncates (Fig. 3 , lane 2) that could be released
from the ribosome upon incubatio n with EDTA (data
not shown). Only a small yield of full-length product was
obtained, as sho wn by i mmunoprecipitation using a n
antiserum d irected against the C-terminal HA-epitope
(Fig. 3, lane 1). UV -crosslinking of this random array of
translation intermediates resulted in numerous distinct
adducts (Fig. 3, lane 4). TF antiserum immunoprecipi-
tated many of these adducts up to a mass o f  120 kDa
(Fig. 3, lane 5), corresponding to a complex of TF and
an approximately full-length version of SufIHATAG8.
This suggests that the signal peptide of SufI interacts
with TF until completion of translation. As observed
before (Fig. 2B, lanes 2 and 7), efficient crosslinking to
TF started from a nascent-chain length between 57 and
93 amino acid residues (Fig. 3, compare lane 5 with lanes
6and7).
A crosslinked c omplex at  170 kDa [Fig. 3, lane 4,
indicated with (*)] was not precipitated with the TF
antiserum. Strikingly, this band appeared more intense
upon incubation of the translation mixture with EDTA
and was immunoprecipitated with a n antiserum against
the HA-epitope (data not shown), indicative of cross-
linking to the released full-length form of SufIHA.

Notably, an identical  170 kDa crosslinked complex
was observed upon completion of synthesis and release
of SufIHATAG8 from the ribosome in a tr anscription/
translation system optimized for t he production of full-
length SufIHA (data not shown). T he crosslinked
partner(s) in this complex h ave not been identified yet.
In the presence of membrane vesicles derived from a
strain that overproduces all components of the Tat
translocase, released full-length S ufIHA was crosslinked
to TatB (data not shown) consistent with earlier data
obtained by Alami and coworkers [29] and confirming
that our in v itro system sustains faith ful targeting o f Tat
substrates.
TF is dispensable for the export of SufI
in vivo
The in vitro crosslinking experiments described above
suggested that TF s equesters the signal peptide of a Tat
substrate while it is being synthesized on the r ibosome. To
investigate whether this interaction is functionally relevant
for the export of Tat p roteins, we monitored the effect of the
intracellular TF level on the efficiency and kinetics of SufI
export in vivo.
Steady state analysis of e ndogenous SufI in w hole cell
samples o f a Dtig mutant strain did not show accumulation
of the precursor of SufI (pre-SufI) (Fig. 4A, lane 4),
Fig. 3. Photo-crosslinking of TF to the signal peptide of nascent SufI
chains of various lengths. In vitro translation o f f ull-length Su fIHA-
TAG8 from non-truncated mRNA in the presence of (Tmd)Phe-
tRNA
sup

. Translation products were immunoprecipitated (IP) with
antiserum against the HA-epitope as indicated. After translation, half
of the sample w as irradiated with UV-light to induce crosslinking and
half was kept in the dark. C rosslinked material w as im munoprec ipi-
tated with TF antiserum as indicated. An  170 kDa complex that
could be precipitated with a nti-HA, but not with anti-TF serum is
indicated with an asterisk. Samples were a nalyzed by SDS/PAGE. F or
comparison, cro sslinks o f TF to 57SufI and 93SufI, im munoprec ipi-
tated with antiserum against TF (corresponding with Fig. 2B, lanes 5
and 10), were run on the same gel (lanes 6 and 7). Molecular m ass
markers ( kDa) are i ndicated at the left s ide of the pa nels.
Ó FEBS 2004 Role of trigger factor in Tat-mediated export (Eur. J. Biochem. 271) 4783
suggesting SufI export i s not significantly affected. TF has
been shown to o verlap with the DnaK chaperone machinery
with respect t o f olding and substrate specificity [31,32].
Interestingly, DnaK has also been shown to interact with an
immobilized Tat s ignal peptide [13]. In light of this evidence,
we monitored the effect of deletion of DnaK and its
co-chaperone DnaJ on SufI export. Notably, a s mall but
reproducible accumulation of pre-SufI was observed i n a
DdnaKdnaJ double mutant (Fig. 4A, lane 5). This effect was
enhanced in a Dtig DdnaKdnaJ triple mutant (Fig. 4A, lane
3). The latter strain was recently constructed and is viable
only in a narrow temperature range [33]. The precursor o f
the SecB-dependent outer membrane protein A (OmpA)
was not detected in any of t he mutants t ested. This is
consistent with published data that suggest acceleration
rather than deceleration of t he export of SecB-dependent
secretory proteins in the absence of TF [3].
We next investigated the kinetics o f SufI e xport by pulse-

chase analysis in t he different genetic backgrounds
(Fig. 4B). In this a ssay we h ad to rely on SufI, provided
with a C-terminal HA-tag, expresse d from an exp ression
vector to obtain detectable SufI signals. I n a wild-type
strain, tagged pre-SufI was processed to its mature form
with the slow kinetics characteristic of T at proteins (Fig. 4 B,
lanes 1–5) [17,23,26]. Furthermore, pre-SufIHA w as not
processed in a tatC minus background during t he chase
period (Fig. 4B, lanes 21–25), confirming that the HA-tag
does not change targeting pathway specificity. The kinetics
of processing appeared not significantly affected in a Dtig,
DdnaKdnaJ or Dtig DdnaKdnaJ mutant as compared to the
wild-type M C4100 s train (Fig. 4B, lanes 6 –10, 11 –15 and
16–20).
Together, the results demonstrate a s mall, additive and
specific effect of dnaKdnaJ and tig deletion on SufI e xport.
The e ffect is only observed in steady state and may be the
result of a s mall subpopulation of pre-SufI that accumulates
in an export-incompetent conformation due to impaired
folding or premature targeting.
To investigate the effect of increased i ntracellular TF
levels on the efficiency of Tat-mediated export, steady state
signals and export kinetics of endogeneous SufI were
monitored upon overproduction of TF from an inducible
expression vector (pJH42) (Fig. 5 B). When TF expression
was induced for 9 0 min prior to labeling, no significant
effect on the kinetics of SufI e xport was apparent (Fig. 5B,
compare l anes 5– 8 and 1–4). Massive overexpression of TF
under these conditions was c onfirmed by trichloroacetic
acid precipitation of the labeled cells (Fig. 5B, lanes 5–8). As

a control, export o f SufI in a Tat-deficient MC4100DtatA/E
strain w as completely blocked in the presence of pJH42
(Fig. 5B, lane 10). Similarly, steady state analysis did not
A
B
Fig. 4. In vivo analysis of S ufI export in Dtig, DdnaKdnaJ and Dtig
DdnaKdnaJ mu tants. S teady state (A) and pulse-chase (B) analysis of
SufI e xport in strains M C4100, MC4100Dtig,MC4100DdnaKdnaJ,
MC4100DtigDdnaKdnaJ and MC4100DtatC at 30 °C. (A) Cells were
grown i n medium c ontaining glucose (0.2%) t o an D
660
of  0.6 and
analyzed by SDS/PAGE an d immun oblotting using a nti-SufI (top )
and a nti-OmpA se rum (bottom). (B) Cells, harboring pBAD18-
SufIHA, were grown in medium containing glucose (0.4%) to an D
660
of 0.35, radiolabeled with [
35
S]methionine for 1 min and chased for the
times indicated. Expression of SufIHA was induced by addit ion o f
L
-arabinose (0.2%) 5 min prior to labeling. Samples were immuno-
precipitated using a n tiserum against SufI.
A
B
Fig. 5. In vivo analysis of SufI upon overexpression of TF. Steady state
(A) and pulse-chase ( B) analysis of SufI export in strains HDB37 and
MC4100DtatA/E, both harboring TF-overexpressing pla smid pJH42,
at 37 °C. (A) Cells were grown to an D
660

of  0.4 an d induced f or TF
overexpression by the a ddition of
L
-arabinose (0.2%) as in dicated.
Samples were t aken 0.5 h and 4 h a fter i nduction as ind icated. Cells
were analyzed b y SDS/PAGE and im munoblotting using antiserum
against SufI (top) an d OmpA ( center), or Coomassie Blue staining
(bottom). (B) Cells were grown to an D
660
of  0.3, radiolabeled with
[
35
S]methionine for 1 min a nd chased for the tim es i ndic ated. W here
indicated, overexpression of TF was induced b y the addition of 0.2%
L
-arabinose 9 0 m in prior to labeling. Before analysis by SDS/PAGE,
samples were i mmunoprecipitated using antiserum against S ufI (top)
and OmpA (center), or pre cipitated with trichloroacetic acid (bottom).
4784 W. S. P. Jong et al.(Eur. J. Biochem. 271) Ó FEBS 2004
reveal any effect on pre-SufI processing after 4 h of TF
overexpression (Fig. 5 A). Here, overexpression of TF was
evident from t he Coomassie s taining of whole cell samples
used for the immunoblot analysis (Fig. 5A, lanes 2 and 4).
In marked contrast, overproduction of TF decelerated the
export of OmpA in a ccordance with published data [3]
resulting in a substantial accumulation of pre-OmpA after
4 h of TF overproduction (Fig. 5A, lane 4). A pparently, TF
overproduction has a differential effect on the export of
proteins that follow disparate targeting/translocation path-
ways.

In conclusion, the d ata suggest that TF, althou gh
interacting with Tat signal peptides, does not play a critical
role in the export of Tat-dependent proteins.
Discussion
Molecular interactions of the signal peptide of two model
Tat proteins, TorA and SufI, were investigated during
in vitro biosynthesis in a n effort to identify ta rgetin g factors
or escort proteins that p lay a role in t he Tat targeting
process. Surprisingly, the chaperone and folding catalyst TF
was t he only cytosolic factor that was extensively cross-
linked to the Tat signal peptides. The association with TF
persisted during synthesis of the entire protein at the
ribosome. Deletion or overexpression of TF did not
significantly influence the efficiency or kinetics of Tat-
mediated translocation.
TF has been suggested to play a regulatory role in
controlling t he entry of secretory proteins in distinct
targeting/translocation pathways [ 6,8]. Photo-crosslink ing
experiments revealed contacts of the signal peptide of
nascent OmpA (a SecB-dependent outer membrane protein)
with fifty-four homologue (Ffh, the protein component of
the E. coli SRP), SecA, S ecB and TF added to a semire-
constituted in vitro tra nslation system [8]. Ffh- a nd SecA-
crosslinking occurred when the signal peptide had just
emerged from the ribosome (up to 89 amino acid nascent
chain length) whereas T F was crosslinked to t he signal
peptide o f slightly longer nascent chains. SecB was only
crosslinked to the signal peptide upon release of nascent
OmpA from the r ibosome. In comparison, the molecular
landscape of Tat signal peptides in a similar experimental

set-up is less complex ( this study). TF is the only photo-
crosslinked cytosolic protein that is detected, probed from
two postitions in the T orA signal peptide, close to the twin-
arginine motif (position 13) and in the (moderately hydro-
phobic) core region (position 24) (Fig. 1B,C). Similar results
were obtained using a l ysine-specific chemical crosslinker
(data not shown). Using SufI, photo-crosslinking to TF was
demonstrated from the shortest nascent SufI w ith exposed
Tat signal p eptide (57 amino acid nascent chain length;
probe at position 8 ) ( Fig. 2B) up t o full-length, but
ribosome associated SufI (Fig. 3).
What is the role o f TF in Tat-mediated export? Does it
prevent the cotranslational engagement of Tat-dependent
proteins in other targeting/translocation pathways? We
have no evidence for this conjecture. First, in t he absence of
TF, we could not identi fy any other partners (e.g. Ffh,
SecA) for n ascent Tat proteins using the in vitro crosslinking
approach described above (data n ot shown). Possibly, the
relatively mild hydrophobicity of the Tat signal peptide
prohibits inte raction w ith F fh [5,34,35]. Furthermore, the
ÔSec-avoidanceÕ motif in the C-terminal re gion of the Tat
signal peptide [ 36] might p revent rerouting via the S ec
pathway e ven when TF is absent. Second, deletion of tig
does not affect th e export of SufI which proceeds in vivo
with the slow kinetics that are characteristic for the Tat
translocation process ( Fig. 4). In contrast, t he export o f
SecB substrate s is markedly accele rated in the absence of TF
probably by the disclosure of a more direct cotranslational
targeting pathway to the Sec translocon [3]. Perhaps, the
proofreading activity of the Sec t ranslocase [37] prevents the

use of this alternative targeting p athway by Tat proteins. On
the other hand, overproduction o f TF inhibits SecB-
mediated transport [3] whereas Tat-mediated transport
proceeds unaffected (Fig. 5). The latter observation is not
unexpected because TF only associates with Tat substrates
during s ynthesis prior to their folding in an export-
competent conformation [22].
Does TF prevent a premature interaction of Tat substrates
with the Tat translocase? It seems conceivable t hat interac-
tion of nascent Tat proteins with the Tat translocase
compromises the folding process that is a prerequisite for
export. On the o ther hand, when IMVs were added during
synthesis of n ascent Tat proteins, crosslinking to the T at
translocase w as not observed i rrespective o f t he presence of
TF (data not shown). Also, the lack of s ignificant effect in a
Dtig mutant strain (Fig. 4) argues against such a seemingly
important function for TF in T at export.
In light of this n egative evidenc e we are i nclined to
believe that TF interacts by default with nascent Tat
proteins due to its location near the n ascent chain exit site
(see below). At present it is unclear whether TF keeps the
signal peptide close to the exit site [8] forcing a looped
conformation of the nascent chain or whether TF moves
from the ribosome with the Tat signal peptide for which it
may have a relatively high affinity. As p roposed for other
substrates, TF may prevent aggregation of nascent Tat
substrates in polysomes, a function that can be t aken over
by the DnaK/DnaJ-chaperone machinery. DnaK/DnaJ
and TF possess an overlapping substrate specificity but
DnaK/DnaJ does n ot dock a t ribosomes and plays a more

prominent r ole in post-translational folding [31,32]. Strik-
ingly, in the absence of DnaK/DnaJ a s mall s ubpopulation
of pre-SufI accumulated, an effect that was augmented in
the a bsence of TF (Fig . 4A). The relatively small effect,
even in the Dtig DdnaKdnaJ triple mutant, may relate to
the capacity o f Tat substrates to fold rapidly. Also, other
chaperones such as SecB may protect nascent Tat
polypeptides from unwelcome interactions [38]. Photo-
crosslinking of the mature domain of Tat proteins will be
required to settle this point.
When emerging from the ribosome, the SufI signal
peptide also crosslinked to L 23 and L29 that are located
near the exit site of t he main ribosomal tunnel and
constitute the TF attachment site [4,30] (Fig. 2). Similarly,
SecB-dependent secretory p roteins such as OmpA [8], SRP-
dependent inner membrane proteins, such as FtsQ [9] and
cytosolic proteins, such as RpoB [38] were shown to
crosslink L23/L29 early during b iogenesis. This suggests
that, irrespective of their final l ocation, E. coli proteins
follow the same pathway through the ribosome and leave
the ribosome at a universal exit site near L23/L29.
Ó FEBS 2004 Role of trigger factor in Tat-mediated export (Eur. J. Biochem. 271) 4785
The p resence of I MVs during synthesis of nascent TorA/
P2 gave rise to two extra crosslinking products (Fig. 1C). T he
adducts ( 7and 17 kDa) appeared specific for the
hydrophobic core of the TorA signal peptide (position 24)
and were sensitive to carbonate extraction, indicating that
they represent peripheral membrane proteins a nd are not
related to any of the known Tat translocase s ubunits. This
raises the i ntriguing possibility that (a subpopulation of)

TorA associates with a distinct membrane bound mach inery
early during translation. It has been suggested before that
translocation through the Tat translocase is preceded by a
Tat-independent targeting and insertion process that was
speculated to function in the quality c ontrol of Tat substrates
[39]. Translation, folding a nd membrane inse rtion may be
coordinated at this location. In this context, it is of interest to
note that the efficiency of translocation of Tat substrates is
dramatically improve d whe n the membrane v esicles are
present during translation instead of being added after
translation [28], an observation that has been difficult to
reconcile with a strictly post-translational t argeting and
translocation m echanism. Work is in progress to elucidate
the identity o f the  7and 17 kDa crosslinked partners.
Acknowledgements
We are g rateful to E .N.G. H ouben a nd M.L. Urbanus for their help
during initial stages of the project and to T . Palmer and F. Sargent f or
providing reagents a nd suggestions. We also thank C. Georgopoulos in
whose lab part of the work was carriedout.N.Harms,G.J.Haanand
M. Mu
¨
ller are acknowledged for their comments on the manuscript an d
stimulating discussions. P .G. was suppo rted by Swiss National Science
Foundation Grant F N-31–65403.
References
1. von H eijne, G. (1990) The signal peptide. J. Membr. Biol. 115,
195–201.
2. Mori, H . & Ito, K. (2001) The Sec protein-translocation pathway.
Trends Microbiol. 9, 494–500.
3. Lee, H.C. & Bernstein, H.D. (2002) Trigger factor retards protein

export in Escherichia c oli. J. Biol. C hem. 277, 4 3527–43535.
4. Kramer,G.,Rauch,T.,Rist,W.,Vorderwu
¨
lbecke, S., Patzelt, H.,
Schulze-Speck ing, A., Ban, N., D euerling, E. & Bukau, B. ( 2002)
L23 protein functions as a chaperone docking site on the ribo-
some. Nature 41 9, 171–174.
5. Valent, Q.A., de Gier, J .W.L., von Heijne, G., K endall, D.A.,
ten Hagen-Jongman, C.M., Oudega, B. & Luirink, J. (1997)
Nascent membrane and presecretory proteins synthesized in
Escherichia c oli associate with signal recognition particle and
trigger f actor. Mol. Microbiol. 25, 53–64.
6. Beck, K., Wu, L.F., Brunner, J. & Mu
¨
ller, M. (2000) Dis-
crimination between SRP- and SecA/S ecB-depend ent substrates
involves selective recognition of n ascent chains by SRP and t rigger
factor. EMBO J. 19, 134–143.
7. Beha , D., Deitermann, S., Mu
¨
ller, M. & Koch, H.G. (2003)
Export of beta-lactamase is independent of the signal recognition
particle. J. Biol. Chem. 278, 22161–22167.
8. Eisner, G., Koch, H.G., Beck, K., Brunner, J. & Mu
¨
ller, M. (2003)
Ligand crowding at a nasc ent signal sequence. J. Cell Bio l. 163,
35–44.
9. Ullers, R.S., H ouben, E .N., Rain e, A., te n Hagen- Jongman,
C.M.,Ehrenberg,M.,Brunner,J.,Oudega,B.,Harms,N.&

Luirink, J. (2003) Interplay of signal r ecognition particle and
trigger factor at L23 near the nascent chain exit site on the
Escherichia coli ri b os om e. J. Cell Biol. 161, 679–684.
10. Luirink, J. & S inning, I. (2004) SRP-mediated protein targeting:
structure a nd function revisited. Biochim. Biophys. Acta. 1697,
17–35.
11. Berks, B.C., Palmer, T. & Sargent, F. (2003) The Tat protein
translocation pathway and i ts role in microbial physiology. Adv.
Microb. Physiol. 47, 1 87–254.
12. Berks, B.C. (1996) A common export pathway f or proteins
binding c omplex red ox cofactors. Mol. Microbiol. 22, 393–404.
13. Oresnik, I.J., Ladner, C.L. & Turner, R.J. (2001) Identification
of a twin-arginine leader-binding protein. Mol. Microbiol. 40,
323–331.
14. Papish, A.L., Ladne r, C.L. & Turner, R .J. (2003) T he twin-
arginine leade r-binding protein, DmsD, interacts with the TatB
and TatC subun its of the Escherichia coli twin-arginine translo-
case. J. Biol. Chem. 278 , 32501–32506.
15. Ray,N.,Oates,J.,Turner,R.J.&Robinson,C.(2003)DmsDis
required for the biogenesis of DMSO reductase in Esch erichia coli
but not for the interaction of the DmsA signal peptide with the Tat
apparatus. FEBS L ett. 534, 1 56–160.
16. Miller, J.H. (1992) A short course in bacterial genetics: A laboratory
manual and handbook for Escherichia coli and related bacteria. Cold
Spring Harbor Laboratory Press, Cold Spring Harbor, N Y.
17. Cristobal, S. , de G ier, J.W., Nielsen, H. & von Heijne, G. (1999)
Competition between Sec- and TAT-depen dent protein translo-
cation in Escherichia c oli. EMBO J. 18, 2982–2990.
18.Scotti,P.A.,Urbanus,M.L.,Brunner,J.,deGier,J.W.,von
Heijne,G.,vanderDoes,C.,Driessen,A.J.,Oudega,B.&

Luirink, J. (2000) YidC, the Escherichia coli homologue of
mitochondrial Oxa1p, is a compo nent of the Sec tran slocase.
EMBO J. 19 , 542–549.
19. Guzman, L.M., Belin, D., Carson, M.J. & Beckwith, J. (1995)
Tight regulation, modulation, and high-level expression by vectors
containing the arabinose PBAD promoter. J. Bacteriol. 177, 4121–
4130.
20. Scotti, P.A., Valent, Q.A., Mant ing, E.H., Urbanus, M.L., Dri-
essen, A.J., O udega, B. & Lu irink, J. (19 99) SecA is not required
for signal recognition particle-mediated targeting and initial
membrane insertion o f a nascent inner membrane protein. J. Biol.
Chem. 274, 29883–29888.
21. Urbanus, M.L., Scotti, P.A., Fro
¨
derberg, L., Sa
¨
a
¨
f, A., de Gier,
J.W., Brunner, J ., Samuelson, J.C., Dalb ey, R.E ., O udeg a, B . &
Luirink, J. (2001) Sec-dependen t membrane p ro tein insertion:
sequential interaction of nascent FtsQ with SecY and YidC.
EMBO Report 2, 524–529.
22. DeLisa, M.P., Tullman, D . & Georgiou, G . (2003) Folding q uality
control in the export o f protein s by the bacte rial twin -arginine
translocation pathway. Proc. N atl Acad. Sc i. USA 100, 6115 –
6120.
23. Santini, C.L., Ize, B., Chanal, A., Mu
¨
ller, M., Giordano, G. &

Wu, L.F. (1998) A novel s ec-independent periplasmic p rotein
translocation pathway in Escherichia c oli. EMBO J. 17 , 101–112.
24. Berks, B.C., Sargent, F. & Palmer, T. (2000) The Ta t protein
export pathway. Mol. Microbiol. 35, 260–274.
25. De Leeuw, E., P orcelli, I., Sargent, F., Palmer, T. & Berks, B.C.
(2001) Membrane interactions and self association of the TatA
and TatB components of the t win-arginin e translocation pathway.
FEBS Lett. 506 , 143–148.
26. Stanley, N.R., Palmer, T. & B erks, B.C. (2000) The t win arginine
consensus motif o f Tat s ignal p eptides is involved i n Sec-
independent protein targeting in Escherichia coli. J. Biol. Chem.
275, 1 1591–11596.
27. Yahr, T.L. & Wickner, W.T. (2001) Functional reco nstitutio n of
bacterial Tat translocation in vitro. EMBO J. 20, 2472–2479.
4786 W. S. P. Jong et al.(Eur. J. Biochem. 271) Ó FEBS 2004
28. Alami, M., Trescher, D. , Wu, L.F. & Mu
¨
ller, M. (2 002) Separate
analysis o f twin-arginine tran slocation (Tat)-specific memb rane
binding a n d translocation in Escherichia coli. J. Biol. Chem. 277,
20499–20503.
29. Alami, M., Luke, I., Deite rmann, S., Eisner, G., Koch, H.G.,
Brunner, J. & Mu
¨
ller, M. (2003) Differential in teractions betw een
a twin-arginine signal peptide and its tran slocase in Escherichia
coli. Mol. Cell. 12 , 937–946.
30. Nissen, P., H ansen, J., Ban, N., Moore, P.B. & Steitz, T.A. (2000)
The structural basis of ribosome activity in peptide bond synthesis.
Science 289, 920–930.

31. Frydman, J. (2001) Folding of newly translated proteins in vivo:
the role o f molecular chaperon es. Annu. Rev. Biochem. 70,
603–647.
32. Deuerling, E., Patzelt, H., Vorderwulbecke, S., Rauch, T.,
Kramer, G ., Schaffitzel, E., Mogk, A., Schulze-Specking, A.,
Langen, H. & Bukau, B. (2003) Trigger Factor and DnaK possess
overlapping substrate pools and binding specificities. Mol.
Microbiol. 47 , 1317–1328.
33. Genevaux, P., Keppel, F., Schwager, F., Lange ndijk-Genevaux,
P.S., Hartl, F.U. & Georgopoulos, C. (2004) In vivo analysis of the
overlapping f unctions of DnaK and trigger factor. EMBO Report
5, 195–200.
34. Lee, H.C. & B ernstein, H.D. (2001) The t argeting pathway of
Escherichia coli presecretory and integral membrane proteins is
specified by the hydrophobicity of the targeting signal. Proc. Natl
Acad. Sc i. USA 98 , 3471–3476.
35. Adams, H., Scotti, P.A., De Cock, H., Luirink, J. & Tommassen,
J. (2002) The presence of a helix breaker i n the hydrophob ic core
of signal sequences o f se cretory pro teins prevents recognition by
the signal-recognition particle in Escherichia coli. Eur. J. Biochem.
269, 5 564–5571.
36. Bogsch, E., Brink, S. & R obinson, C. (1997) Pathway specificity
for a delta pH-dependent precursor thylakoid lumen protein is
governed by a ÔSec-avoidanceÕ motif in t h e transfer peptide and a
ÔSec-in compatib leÕ mature protein. EMBO J. 16, 3851–3859.
37. Osborne, R.S. & Silhavy, T.J. (1993) PrlA suppres sor mutations
cluster in regions corresponding to three distinct topological
domains. EMBO J. 12, 3391–3398.
38. Ullers, R.S., Luirink, J., Harms, N., Schwager, F., Georgopoulos,
C. & Genevaux, P. (2004) SecB is a bona fide generalized cha-

perone in Escherichia c oli. Proc. N atl Acad. Sci. USA 101, 7583–
7588.
39. Bru
¨
ser, T. & Sanders, C. (200 3) An alternative m odel of t he twin
arginine tra nslocation system. Microbiol. Res. 158 , 7–17.
40. Casadaban, M.J. (1976) Transposition a nd fusion of the lac genes
to selected promoters in Escherichia coli using bacteriop hage
lambda and M u. J. Mol. B iol. 104, 5 41–555.
41. Sargent, F., Stanley, N.R., Berks, B.C. & Palmer, T. (1999) S ec-
independent protein transloc at ion in Escherichia coli.Adistinct
and pivotal role for the TatB protein. J. Biol . Chem. 274, 36073–
36082.
Ó FEBS 2004 Role of trigger factor in Tat-mediated export (Eur. J. Biochem. 271) 4787