Tài liệu Báo cáo khoa học: The sensor protein KdpD inserts into the Escherichia coli membrane independent of the Sec translocase and YidC doc

The sensor protein KdpD inserts into the
Escherichia coli
membrane
independent of the Sec translocase and YidC
Sandra J. Facey and Andreas Kuhn
Institute of Microbiology and Molecular Biology, University of Hohenheim, Stuttgart, Germany
KdpD is a sensor kinase protein in the inner membrane of
Escherichia coli containing four transmembrane regions.
The periplasmic loops connecting the transmembrane
regions are intriguingly short and protease mapping allowed
us to only follow the translocation of the second periplasmic
loop. The results show that neither the Sec translocase nor
the YidC protein are required for membrane insertion of the
second loop of KdpD. To study the translocation of the first
periplasmic loop a short HA epitope tag was genetically
introduced into this region. The results show that also the
first loop was translocated independently of YidC and the
Sec translocase. We conclude that KdpD resembles a new
class of membrane proteins that insert into the membrane
without enzymatic assistance by the known translocases.
When the second periplasmic loop was extended by an
epitope tag to 27 amino acid residues, the membrane inser-
tion of this loop of KdpD depended on SecE and YidC. To
test whether the two periplasmic regions are translocated
independently of each other, the KdpD protein was split
between helix 2 and 3 into two approximately equal-sized
fragments. Both constructed fragments, which contained
KdpD-N (residues 1–448 of KdpD) and the KdpD-C
(residues 444–894 of KdpD), readily inserted into the
membrane. Similar to the epitope-tagged KdpD protein,
only KdpD-C depended on the presence of the Sec translo-

case and YidC. This confirms that the four transmembrane
helices of KdpD are inserted pairwise, each translocation
event involving two transmembrane helices and a periplas-
mic loop.
Keywords: Escherichia coli;membraneprotein;protein
translocation; epitope tag.
The inner membrane protein KdpD of Escherichia coli is
involved in osmoregulation. It comprises of 894 amino acid
residues organized as two hydrophilic domains that are
separated by four closely spaced transmembrane regions [1].
KdpD is functionally related to other sensor kinases like
PhoR and EnvZ and shows a moderate sequence homology
in parts of the C-terminal domain with other sensor kinases.
In the membrane, the KdpD protein forms a homodimer,
which has been proposed to be required for the kinase
function [2]. The transmembrane regions are necessary for
signal perception because mutants in the transmembrane
regions have been found that are defective in the osmotic
response [3]. To understand how the transmembrane helices
or the periplasmic loops sense an osmotic signal a precise
knowledge of the topology and membrane insertion of these
hydrophobic regions is crucial. Intriguingly, the two peri-
plasmic loops separating the transmembrane regions com-
prise of only four and 10 amino acid residues, respectively.
Multi spanning membrane proteins contain several
hydrophobic regions linked by hydrophilic loops of various
lengths ranging from a few amino acids to several hundred
residues, e.g. in SecD [4]. Long periplasmic loops are
translocated by the ATP-driven Sec translocase, whereas
small loops may be translocated by a synergistic mechanism

without the Sec translocase as has been observed for the
double-spanning M13 procoat protein [5,6]. Based on
results from a functional approach [7], a Sec-independent
insertion has also been suggested for melibiose permease,
which has six short periplasmic loops. Gafvelin and von
Heijne [8] have shown, through studying a tandem
construction of leader peptidase that spans the membrane
four times, that short periplasmic loops of about 25 residues
were translocated independently of SecA, whereas long
loops of 250 residues required the SecA-driven translocase.
However, De Gier et al. [9] found by using the tightly
controlled SecE mutant strain, that the SecYE translocase
may be involved in the translocation of a 25 residue
periplasmic loop. The authors suggested that the hydro-
phobicity of the transmembrane region determines the
requirement of the Sec translocase.
Proteins that are destined to be translocated across or
inserted into the bacterial inner membrane are targeted to
the translocation sites by multiple mechanisms. In E. coli,
secretory proteins are targeted to the inner membrane by
means of the chaperone SecB, which directs the newly
synthesized protein to the SecA subunit of the translocase
complex of the Sec pathway, and whose membrane-
integrated components are SecY, E, and G [10]. In contrast,
polytopic membrane proteins are targeted to the membrane
by an essential ribonucleoprotein complex that is closely
related to the eukaryotic signal recognition particle (SRP).
E. coli contains Ffh (P48), which together with 4.5S RNA,
Correspondence to A. Kuhn, Institute of Microbiology and
Molecular Biology, University of Hohenheim, 70599 Stuttgart

Germany. Fax: + 49 711 4592238, Tel.: + 49 711 4592222,
E-mail:
Abbreviations: HA, haemagglutinin; SRP, signal recognition particle;
IPTG, isopropyl 1-thio-b-
D
-galactoside; CCCP, carbonyl cyanide
p-chlorophenylhydrazone; pmf, proton motive force.
(Received 11 December 2002, accepted 20 February 2003)
Eur. J. Biochem. 270, 1724–1734 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03531.x
represents the bacterial homologue of the SRP [10].
Membrane translocation is then catalysed by SecY and
SecE; SecA and SecG are not required for most membrane
proteins [11].
A new bacterial membrane protein insertion pathway was
recently discovered involving YidC, a protein homologous
to the mitochondrial Oxa-1p. YidC was found to be
required for the insertion of Sec-independent membrane
proteins and is also involved in the membrane integration of
Sec-dependent proteins, whereas exported proteins, such as
OmpA, were not affected (reviewed in [12]). In the absence
of YidC, Sec-independent proteins accumulated at the
cytoplasmic side of the membrane, whereas Sec-dependent
membrane proteins were jammed in the Sec translocase
[13,14].
To understand the translocation process of multispan-
ning membrane proteins, we have investigated the mechan-
ism of how the sensor kinase protein KdpD inserts into the
membrane. We found that KdpD inserts into the membrane
independently of the Sec translocase and YidC. However,
when the two small periplasmic regions of the protein were

extended by short epitopes, we found that the translocation
of the first periplasmic region was still independent of the
Sec translocase and YidC, but the second extended
periplasmic region required the Sec translocase and YidC.
Unexpectedly, the introduction of the epitope tag into the
second periplasmic region was the main cause for the
requirement for YidC.
Materials and methods
Plasmid constructions
K. Jung and K. Altendorf (Universita
¨
tOsnabru
¨
ck,
Germany) kindly provided the plasmids, pPV5 and pBD
carrying the kdpD gene in pKK233-3 and pBAD18,
respectively [15,16]. The strategy to generate the two
truncated halves of the protein was to cut KdpD in
approximately the middle between helix 2 and 3. By means
of site-directed mutagenesis, a stop codon (TAG) and an
NdeI restriction site was introduced between helix 2 and 3.
The constructed fragments containing KdpD-N (i.e. coding
the amino acid residues 1–448 of KdpD) and KdpD-C (i.e.
coding the amino acid residues 444–894 of KdpD) were
cloned into the expression vector pT7-7.
The epitope tags within the fragments were constructed
by first introducing a MunI restriction site between the first
and second and between the third and fourth helices by site-
directed mutagenesis. The epitope tags were introduced into
the opened MunI sites of the respective plasmids by ligating

two short complementary oligonucleotides with AATT
overhangs. These complementary oligonucleotides code
either for a haemagglutinin (HA)- or a T7-epitope tag with
a spacer of four amino acid residues. Each of the tagged
constructs was sequenced to confirm the correct in-frame
fusion of the epitope cassettes.
Strains, plasmids, and growth conditions
Cloning and mutagenesis experiments were performed with
E. coli XL1-Blue recA1 thi supE44 endA1 hsdR17 gyrA96
relA1 lac F¢ (proAB
+
lacI
q
lacZDM15 Tn10) (Stratagene).
The pT7-7 expression vector with the kdpD gene was
transferred into the E. coli BL21(DE3)pLysS strain which
expresses the T7 RNA polymerase under the inducible
lacUV5 promoter [17].
The SecE-depletion strain CM124 [18] was cultured in
M9 minimal medium supplemented with 0.4% glucose and
0.2%
L
-arabinose. To deplete cells for SecE, overnight
cultures were washed once with M9 medium and back-
diluted 1 : 20 into fresh M9 medium in the absence of
L
-arabinose. Depletion of SecE was checked by monitoring
the accumulation of the precursor to the outer membrane
protein A (proOmpA).
The YidC-depletion strain JS7131 [13] was cultured in

Luria–Bertani medium supplemented with 0.2% arabinose.
To deplete cells for YidC, overnight cultures were grown in
0.2% arabinose and then washed twice with LB to remove
cells of arabinose and back-diluted 1 : 50 into fresh Luria–
Bertani medium with 0.2% glucose. Depletion of YidC was
checked by immunoprecipitating the labelled cells with
antibodies to YidC.
Media preparation and bacterial manipulations were
performed according to standard methods [19]. Where
appropriate, ampicillin (100 lgÆmL
)1
, final concentration),
kanamycin (50 lgÆmL
)1
, final concentration) and chloram-
phenicol (25 lgÆmL
)1
, final concentration) were added to
the medium.
Wild-type KdpD, KdpD containing the HA- and T7-
epitope tags, KdpD-N containing the N-terminal fragment
with the HA-epitope and KdpD-C containing the
C-terminal fragment with the T7-epitope were expressed
by
L
-arabinose induction from the pBAD18 vector [20] in
strain MC1061 and by isopropyl thio-b-
D
-galactoside
(IPTG) induction from the vectors pT7-7, pMS119 [21]

and pDHB5700 [9] in strains BL21(DE3)pLysS, JS7131 and
CM124, respectively.
Antibodies
The T7-tag monoclonal antibody recognizing the 11 amino
acid T7 peptide (MASMTGGQQMG) was purchased from
Novagen. The anti-HA recognizes the HA peptide sequence
(YPYDVPDYA) derived from the human influenza HA
protein [22]. The anti-HA monoclonal antibody was
purchased from Boehringer. Polyclonal antibody against
KdpD was a gift from K. Jung and K. Altendorf
(Universita
¨
t Osnabru
¨
ck, Germany).
Protease mapping assay
For all experiments, cells were grown to midlogarithmic
phase. Cells harboring the plasmid-encoded proteins were
induced for 10 min either with IPTG (1 m
M
, final concen-
tration) or for 1 h with
L
-arabinose (0.2%, final concentra-
tion). Unless otherwise stated, cells were labelled with
[
35
S]methionine for 5 min and chased with excess
L
-methio-

nine for 5 min. For spheroplasting, cells were centrifuged at
12 000 g and resuspended in 500 lL of ice-cold spheroplast
buffer (40% w/v sucrose, 33 m
M
Tris/HCl, pH 8.0). Lyso-
zyme (5 lgÆmL
)1
, final concentration) and EDTA (1 m
M
,
final concentration) were added for 15 min. Aliquots of the
spheroplast suspension were incubated on ice for 1 h either in
the presence or absence of proteinase K (0.5 mgÆmL
)1
final
Ó FEBS 2003 KdpD membrane insertion (Eur. J. Biochem. 270) 1725
concentration). A lysis control was included by adding 2.5%
Triton X-100 and proteinase K for 1 h. After addition of
phenylmethanesulfonyl fluoride (0.33 mgÆmL
)1
, final con-
centration), samples were precipitated with trichloroacetic
acid (20%, final concentration), resuspended in 10 m
M
Tris/
2% SDS, pH 8.0 and immunoprecipitated with antibodies
against HA, T7, KdpD, OmpA (a periplasmic control), or
GroE (a cytoplasmic control, results not shown). Samples
were analysed by SDS/PAGE and phosphorimaging.
For the azide and carbonyl cyanide p-chlorophenyl-

hydrazone (CCCP) studies, the cells (0.5 mL cultures) were
pretreated by the addition of 10 lL of sodium azide
(100 m
M
) for 5 min or by the addition of 2.5 lLofCCCP
(10 m
M
) for 45 s, prior to labelling of the cells.
Results
Membrane insertion of the KdpD protein
The membrane insertion of the KdpD protein is difficult to
analyse because the translocated periplasmic regions are
comprised of only four and 10 amino acid residues,
respectively. We observed that proteinase K did not cleave
the protein in the first periplasmic loop, probably because
this loop is too short and does not extend far enough away
from the membrane surface to be accessible to the protease.
Cleavage in the second periplasmic loop occurred partially
and led to a protease protected fragment of 47 kDa that
was recognized by the KdpD antibody that detects the
C-terminal cytoplasmic domain. The generation of the
protease protected fragment allowed the investigation of
how the second (10 amino acid residues long) periplasmic
region of the wild-type KdpD is translocated.
First, the involvement of SecA was investigated using
sodium azide (Fig. 1A). Sodium azide has been shown to
inhibit SecA activity at 2 m
M
concentration [23]. To address
the role of SecA in KdpD membrane insertion, bacteria

weretreatedwith2m
M
sodium azide for 5 min prior to
[
35
S]methionine addition. After a pulse of 5 min, a fraction
of the radioactively labelled KdpD protein was accessible to
proteinase K added to the outside of the cells either in the
absence or presence of sodium azide (Fig. 1A, lower panel).
Translocation of the second periplasmic loop of KdpD was
followed by the generation of the C-terminal 47 kDa
proteolytic fragment. The results show that its formation
was not affected when the function of SecA was perturbed
by azide (compare lanes 2 and 5). Following lysis of the cells
with detergent, we confirmed that the smaller fragment was
readily digested (lanes 3 and 6). As expected, proOmpA was
rapidly converted to OmpA in the absence of azide (upper
panel, lane 1). In the presence of azide, the Sec-dependent
proOmpA accumulated in the cytoplasm of the cells and
was not digested by the protease (lanes 4 and 5).
To test the role of integral translocase components, the
involvement of SecE in KdpD membrane insertion was
investigated. This was performed by using the strain
CM124, in which SecE can be depleted efficiently. In this
strain, the secE gene expression is under the control of the
arabinose-inducible araBAD promoter [24]. In the presence
of the repressor glucose and absence of arabinose, SecE is
not expressed. CM124 cells were grown in the presence of
glucose or arabinose, respectively, and analysed for KdpD
membrane insertion. When SecE was depleted, KdpD was

still inserted because the proteolytic fragment was detectable
in equal amounts (Fig. 1B, lower panel; compare lanes 2
and 5). As a control, the translocation of proOmpA was
monitored (upper panel). As expected, proOmpA translo-
cation was blocked under SecE-depleted conditions and not
digested by the protease.
The dependence of KdpD insertion on the proton motive
force (pmf) was studied after treatment of the cells with
CCCP, a protonophore that dissipates the pmf [25]. The
pmf was collapsed by adding 50 l
M
CCCP, 45 s before
labelling the cells with [
35
S]methionine. CCCP reduced the
efficiency of the translocation of the second periplasmic
loop of KdpD as indicated by the reduced appearance of the
C-terminal fragment (Fig. 1C, lower panel; compare lanes 2
and 5). Immunoprecipitation with OmpA antiserum
showed the accumulation of the nontranslocated precursor
(proOmpA), which was not digested by proteinase K
(Fig. 1C, upper panel).
The role of YidC in the membrane insertion of KdpD
was examined in the depletion strain JS7131, where YidC
expression is under the control of an araBAD promoter and
operator [13]. YidC expression was induced with arabinose
and tightly repressed in the presence of glucose. To deplete
YidC, the cells were grown for 3 h with glucose and then
analysed for KdpD insertion (Fig. 1D, lower panel). Under
both conditions, KdpD inserted into the membrane as

judged by the appearance of the C-terminal fragment (lanes
2 and 5). As a control, the accumulation of M13 procoat
protein was analysed in a parallel culture (Fig. 1D, upper
panel). The results show that under YidC-depleted condi-
tions procoat accumulated and was not digested by the
protease. Taken together, these results suggest that the
second periplasmic loop of the wild-type KdpD protein is
inserted into the membrane in the absence of SecA, SecE
and YidC.
Short epitopes introduced into the periplasmic regions
allow the analysis of insertion events
To analyse the translocation of the two periplasmic regions
of KdpD in detail, short epitope tags were introduced into
these regions (Fig. 2). Oligonucleotide-directed insertion
was used to introduce a 15 residue HA-tag derived from the
human influenza haemagglutinin protein between helix 1
and 2 and a 17 residue T7-tag of the T7 major capsid protein
between helix 3 and 4. A specific monoclonal antibody (anti-
HA or anti-T7) was then used to monitor the location of the
epitope-tagged region with respect to the KdpD protein in
the membrane. The KdpD protein with the epitope tags was
readily digested by proteinase K in both periplasmic regions
(Fig. 3A). The periplasmic location of the epitope-tagged
regions is consistent with the proposed membrane topology
of KdpD [1] and shows that now both regions are well
exposed away from the membrane surface and easily
accessible by the protease.
To address the role of SecA in the membrane assembly of
KdpD containing the HA- and the T7-epitopes in the
respective loops, bacteria were treated with 2 m

M
sodium
azide for 5 min prior to [
35
S]methionine addition. Figure 3A
(middle and lower panel) shows that both periplasmic loops
of KdpD are translocated in the absence (lane 2) and in the
1726 S. J. Facey and A. Kuhn (Eur. J. Biochem. 270) Ó FEBS 2003
presence (lane 5) of sodium azide, under conditions in which
proOmpA translocation is reduced (Fig. 3A, upper panel).
This suggests that SecA is not necessary for membrane
insertion of KdpD.
The requirement of the Sec translocase was tested in the
CM124 strain where SecE is depleted when the cells are
grown in the absence of arabinose. When the cells expres-
sing KdpD with the HA- and the T7-epitopes were grown in
the presence of glucose to deplete SecE (Fig. 3B, middle and
lower panel), the membrane translocation of only the first
periplasmic loop of KdpD was efficient (middle panel,
compare lanes 2 and 5). The translocation of the second
periplasmic loop of KdpD was only about 70% efficient
indicating a dependence on SecE (Fig. 3B, lower panel). In
the same cells, proOmpA export was totally blocked by the
depletion of SecE (Fig. 3B, upper panel). This differs from
the results obtained with the wild-type KdpD protein, where
the translocation of the second periplasmic loop without the
epitope tag was not affected by SecE depletion (Fig. 1B).
To assess the effect of the pmf on the membrane insertion
of KdpD containing the epitope tags, the protonophor
CCCP (50 l

M
) was added 45 s prior to pulse-labelling of
the cells. Figure 4A (middle and lower panels) shows the
Fig. 1. The translocation of the second periplasmic loop of KdpD is
independent of SecA, SecE and YidC, but is sensitive to the membrane
potential. (A) Protease mapping of KdpD in the absence (–) and
presence (+) of sodium azide to block SecA function. E. coli strain
MC1061 expressing the wild-type KdpD was grown at 37 °Ctomid-
log phase, induced for 1 h with 0.2% arabinose and labelled with
[
35
S]methionine for 5 min. The cells were converted to spheroplasts
and incubated with (lanes 2 and 5) or without proteinase K (lanes 1
and4)atafinalconcentrationof0.5mgÆmL
)1
on ice for 1 h. A lysis
control was included by adding proteinase K (0.5 mgÆmL
)1
, final
concentration) and 2.5% Triton X-100 (lanes 3 and 6). All samples
were precipitated with 20% trichloroacetic acid, immunoprecipitated
with antiserum to OmpA (upper panel) and KdpD (lower panel) and
analysed by SDS/PAGE and visualized by phosphorimaging. The
positions of the molecular weight standards (SeeBlue
TM
Pre-Stained
Standard, from Invitrogen) are marked on the right. (B) Strain CM124
expressing KdpD was grown in M9 minimal medium containing
arabinose (lanes 1–3). For depletion of SecE (lanes 4–6), cells were
grown in the absence of arabinose for 8 h. The cells were then induced

with 1 m
M
IPTG for 10 min. Cells were pulse-labelled for 5 min and
chased with 500 lgÆmL
)1
cold
L
-methionine for 5 min and subse-
quently analysed as described as above. As a control, proOmpA
processing was monitored in parallel to verify SecE depletion. (C)
Protease mapping of KdpD in the absence (–) and presence (+) of the
protonophore CCCP to dissipate the pmf. CCCP was added 45 s prior
to labelling at a final concentration of 50 l
M
. E. coli MC1061 bearing
pBAD18 encoding wild-type KdpD was induced with arabinose for
1 h, labelled with [
35
S]methionine for 5 min and chased with
500 lgÆmL
)1
cold
L
-methionine for 5 min as described above. Clea-
vage of proOmpA was monitored as a control (upper panel). (D) To
test the requirement of YidC, the YidC depletion strain JS7131 was
induced with arabinose or tightly repressed in the presence of glucose.
E. coli strain JS7131 containing the cloned kdpD gene (pMS119kdpD)
was grown in LB with either 0.2% arabinose (YidC
+

) or 0.2% glucose
(YidC
–
) for 3 h. One millimolar IPTG was added for 10 min to induce
expression and the cells were pulse-labelled for 1 min, then converted
to spheroplasts by lysozyme treatment and osmotic shock. Translo-
cation of the YidC-dependent M13 coat protein was monitored in
parallel by proteinase K treatment of spheroplasts (upper panel).
Samples were immunoprecipitated with antiserum to M13 coat protein
(upper panel) and with antiserum to KdpD, respectively (lower panel).
Ó FEBS 2003 KdpD membrane insertion (Eur. J. Biochem. 270) 1727
membrane translocation of the periplasmic loops of KdpD
in the absence and in the presence of CCCP. These results
demonstrate that the pmf is required for efficient membrane
insertion of KdpD with the tags. This is in agreement with
the wild-type KdpD, which is also sensitive to the pmf for
efficient membrane assembly (Fig. 1C, lower panel).
We also investigated the involvement of YidC for the
translocation of KdpD with the two epitopes in the YidC-
depleted strain JS7131. Figure 4B (middle panel) shows that
in cells grown with glucose to deplete YidC, the first
periplasmic loop was normally translocated and did not
differ from the cells grown with arabinose (compare lanes 2
and 5). The translocation of the second periplasmic loop
(Fig. 4B, lower panel), however, was affected in the cells
with depleted YidC. This indicates that the two periplasmic
loops of KdpD with the epitope tags are translocated
differently. Whereas the first loop translocates in the
absence of SecA, SecYE and YidC, but depends on the
pmf, the translocation of the second loop is supported by

SecYE and YidC.
Membrane insertion of split osmosensor fragments
The kdpD gene encoding the HA- and the T7-epitopes was
split into 2 approximately equal-sized fragments between
helix 2 and 3. The constructed fragments containing KdpD-
N (i.e. coding the amino acid residues 1–448 of KdpD) and
KdpD-C (i.e. coding the amino acid residues 444–894 of
KdpD) were subcloned into pT7-7. The KdpD fragments
were stably expressed as truncated N- or C-terminal halves,
each with double-spanning membrane helices.
As described above, we used the protease accessibility
assay to analyse the insertion of the KdpD truncated halves
into the membrane. Both truncated halves, termed KdpD-N
and KdpD-C, were readily inserted into the inner mem-
brane and the epitopes were digested by the externally
added protease. Intriguingly, a stable dimeric form was
observed only for KdpD-N (Fig. 5A). The membrane
Fig. 3. The involvement of SecA (A) and SecE (B) in the translocation of
the individual membrane loops. (A) The kdpD gene containing the
epitope tags was expressed in strain MC1061 in the presence (lanes
1–3) or absence (lanes 4–6) of sodium azide. Cells were pulse-labelled
with [
35
S]methionine for 5 min and then converted to spheroplasts as
described in the legend to Fig. 1. The epitope-tagged KdpD protein
was immunoprecipitated with antiserum to HA (for the epitope in the
first periplasmic loop; middle panel) and to T7 major capsid protein
(for the epitope in the second periplasmic loop; lower panel), respect-
ively,andthenanalysedbySDS/PAGEandvisualizedbyphos-
phorimaging. OmpA accumulated in its precursor form (proOmpA)

in the azide treated cells (upper panel, lanes 4–5). (B) CM124
cells expressing the epitope-tagged KdpD were pulse-labelled with
[
35
S]methionine for 5 min and chased for 5 min either in the presence
of arabinose to induce expression of SecE (lanes 1–3) or in the absence
of arabinose to deplete SecE (lanes 4–6). Translocation of the Sec-
dependent protein OmpA was monitored in parallel after a 1-min
pulse-labelling (upper panel).
Fig. 2. Membrane topology of KdpD (A) and introduction of epitopes to
extend the short periplasmic regions of KdpD (B). (A) Oligonucleotide-
directed mutagenesis was used to integrate a HA-epitope derived from
the human influenza haemagglutinin protein into the first periplasmic
loop of KdpD and a T7-epitope of the T7 major capsid protein into the
second periplasmic loop of KdpD. (B) lists the amino acid sequences of
each of the two extra-membrane loops before and after the insertion of
the epitopes. Insertion of the epitopes (underlined) has the following
consequences for length (number of amino acid residues) and net
charge of the loops (without/with tag); Helix 1/2: (4/19) ()1/)3); Helix
3/4: (10/27) (0/0).
1728 S. J. Facey and A. Kuhn (Eur. J. Biochem. 270) Ó FEBS 2003
insertion of the N- and C-terminal halves was then studied
in CM124 cells where SecE was depleted (Fig. 5A,B). The
cells were induced, labelled with [
35
S]methionine for 5 min,
chased for 5 min, immediately converted to spheroplasts
and treated with proteinase K. The samples were immuno-
precipitated with antibodies to the respective tags (anti-HA
or anti-T7) and analysed by SDS/PAGE, and the bands

were visualized on a phosphorimager. The translocation of
KdpD-N was not affected by the depletion of SecE
(Fig. 5A), whereas KdpD-C was clearly affected by the
SecE depletion (Fig. 5B). In both experiments, the trans-
location and cleavage of proOmpA was efficiently blocked
when SecE was depleted (upper panels). In agreement with
the results obtained from studies with the four-spanning
KdpD protein containing the epitope tags (Fig. 3B), the first
periplasmic loop was translocated across the membrane in a
Sec-independent fashion, whereas the translocation of the
second periplasmic loop with the tag indicated a dependence
on SecE for efficient insertion.
Membrane potential is required for the insertion
of KdpD-N
To test whether the translocation of the periplasmic loops
requires the pmf, the location of the loops was analysed in
the presence of CCCP. As shown in Fig. 6A, CCCP
completely blocked translocation of KdpD-N. The protein
was not accessible to the externally added proteinase K,
indicating that it remains in the cytoplasm. Intriguingly, the
formation of the dimeric form was also blocked. In contrast,
the membrane insertion of KdpD-C was partially affected
by the addition of CCCP (Fig. 6B), and most of the protein
Fig. 4. The involvement of the electrochemical membrane potential (A)
and YidC (B) in the translocation of the individual membrane loops. (A)
Proteinase K mapping of the epitope-tagged KdpD protein in the
absence (–) and presence (+) of CCCP. E. coli MC1061 cells bearing
the pBAD18-plasmid coding for the epitope-tagged KdpD protein
were labelled with [
35

S]methionine for 5 min at 37 °C and chased with
500 lgÆmL
)1
L
-methionine for 5 min. Cells were then converted to
spheroplasts and analysed as described in Fig. 3. Dissipation of the
membrane potential was checked by monitoring the accumulation of
proOmpA. (B) Proteinase K mapping of the epitope-tagged KdpD
protein in the YidC depletion strain, JS7131. Cells were grown in the
presence of arabinose (YidC
+
) or in the presence of glucose (YidC
–
)
and pulse-labelled for 5 min. The cells were then converted to
spheroplasts and treated with or without proteinase K for 1 h, and
analysed as described in Fig. 3. OmpA processing was monitored in
parallel after a 1-min pulse-labelling (upper panel).
Fig. 5. Effects of SecE depletion on the translocation of the split KdpD
proteins. CM124 cells expressing KdpD-N (A) or KdpD-C (B) were
grown in M9 minimal medium either in the presence (SecE
+
)or
absence of arabinose (SecE
–
). Cells were pulse-labelled with
[
35
S]methionine for 5 min and chased for 5 min with 500 lgÆmL
)1

L
-methionine and analysed as outlined in the legend to Fig. 1. Samples
were immunoprecipitated with antiserum to HA (for KdpD-N) and to
T7 major capsid protein (for KdpD-C), respectively. OmpA processing
was monitored in parallel to check spheroplasting and SecE depletion
(upper panels). The extra band observed in the lower part of A is the
dimer of KdpD-N.
Ó FEBS 2003 KdpD membrane insertion (Eur. J. Biochem. 270) 1729
was accessible to proteinase K. This demonstrates that the
pmf is required for the insertion of KdpD-N, but has only a
slight effect on KdpD-C.
YidC is required for efficient insertion of KdpD-C,
but not for KdpD-N
We investigated the effect of YidC depletion on the
translocation of KdpD-N and KdpD-C in the strain
JS7131. When YidC was present (cells grown with arabi-
nose), both proteins were readily inserted into the membrane
and digested with proteinase K (Fig. 7A and B, lanes 1 and
2). In YidC-deficient cells (grown with glucose), KdpD-N
inserted normally into the membrane and was digested with
proteinase K (Fig. 7A, compare lanes 2 and 5). Likewise,
dimer formation was also not affected. Therefore, translo-
cation of KdpD-N is independent of YidC.
In contrast, the translocation of KdpD-C with the
T7-epitope was affected in the cells grown with glucose,
indicating a dependence on YidC for efficient insertion
(Fig. 7B). When YidC was not depleted (YidC
+
), KdpD-C
was efficiently inserted and digested with proteinase K

(lanes 1 and 2). Because the wild-type KdpD protein
without the tags was inserted independently of YidC
(Fig. 1D), the introduction of an epitope tag might affect
the membrane insertion.
To test this, the membrane insertion of KdpD-C with
(Fig. 8A) and without the epitope tag (Fig. 8B) was
followed with the KdpD antibody which recognizes the
C-terminal cytoplasmic domain of KdpD. Therefore, if the
periplasmic loop is cleaved by the protease, only a small
shift of the molecular mass of the protein is expected
because the antibody recognizes the remaining C-terminal
domain. In the presence of YidC, the shift of the molecular
mass of KdpD-C with the tag was complete when protei-
nase K was added externally (Fig. 8A, lane 2). When the
cells were depleted for YidC, the generation of the shift was
inhibited showing that the periplasmic loop was not
translocated (Fig. 8A, compare lanes 2 and 5). In contrast,
the untagged KdpD-C was only partially shifted (Fig. 8B).
This is because the short periplasmic region is not well
exposed at the cell surface, in agreement with the observa-
tions from the wild-type KdpD. When YidC was depleted,
membrane insertion of KdpD-C without the epitope tag
appeared almost as efficient as that of the YidC-containing
cells (Fig. 8B, compare lanes 2 and 5). Taken together, these
results suggest that the presence of the epitope tag is the
reason why KdpD-C requires the assistance of YidC.
Discussion
The present study was initiated to understand how multi-
spanning membrane proteins with short periplasmic loops
are inserted into the membrane bilayer. Most studies on

Fig. 6. The KdpD-N fragment (A) requires the electrochemical mem-
brane potential for membrane insertion, whereas the KdpD-C fragment
(B) is only slightly affected. MC1061 cells with plasmids expressing the
mutant proteins were analysed with (+) or without CCCP (–) as
described in the legend to Fig. 1. Cells bearing plasmids encoding these
proteins were pulse-labelled with [
35
S]methionine for 5 min and chased
for 5 min. OmpA accumulated in its precursor form (proOmpA) in
CCCP treated cells (upper panels, lanes 4–5).
Fig. 7. YidC is required for efficient membrane insertion of KdpD-C (B)
but not for KdpD-N (A). Plasmids encoding KdpD-N (A) or KdpD-C
(B) were transformed into E. coli JS7131. The cells were analysed in
pulse-labelling experiments under YidC-depleted or YidC-expressing
conditions as described in Fig. 1. After subjecting the cells to a pro-
tease accessibility assay, the proteins were immunoprecipitated with
antiserum to HA (A), to T7 major capsid protein (B) and analysed by
SDS/PAGE and phosphorimaging.
1730 S. J. Facey and A. Kuhn (Eur. J. Biochem. 270) Ó FEBS 2003
multispanning proteins made so far have focussed on the
translocation of large domains [26–28]. Short periplasmic
regions are difficult to analyse, since they hide as an
antigenic target and resist proteolytic assessment [29–31].
We used the four-spanning membrane protein KdpD as a
model system. It contains two periplasmic loops of four and
10 amino acid residues. The first periplasmic region of
KdpD proved resistant to proteinase K, whereas the second
periplasmic loop of the KdpD protein was partially
accessible to externally added protease and the digestion
resulted in a smaller C-terminal fragment. We found that

only about 50% of the protein was digested by the protease.
When the periplasmic region was extended by 17 amino acid
residues, more than 95% of the protein was accessible,
suggesting that the short periplasmic region in KdpD is
affected in its surface exposure, not in its membrane
translocation. The analysis of the membrane insertion of
the wild-type KdpD showed that the translocation of the
second periplasmic loop is independent of SecA, SecE, and
YidC, and is only affected by the loss of the membrane
potential (Fig. 1).
To analyse the translocation of the two periplasmic
regions of KdpD short epitopes were introduced into these
regions. Antibodies specific for each epitope were used for
immunoprecipitation showing that the translocation of
both periplasmic loops can be analysed individually. This
enabled the testing of whether the Sec translocase is
involved in the membrane insertion process. Using the
strain CM124, where the SecE content can be extensively
depleted [9], we observed that the first periplasmic loop of
KdpD was translocated normally across the membrane
(Fig. 3B). Because in the absence of SecE, SecY is rapidly
degraded [32], we conclude that the translocation of the first
loop is independent of SecYE. Likewise, the inactivation of
SecA by azide [23] did not affect the membrane insertion,
suggesting that wild-type KdpD is inserted Sec-independ-
ently. This is different to most other known membrane
proteins that require at least the integral components of the
Sec translocase for membrane insertion. Mannitol permease
and SecY require SecYE for insertion, but are independent
of SecA and SecG [33], whereas leader peptidase and YidC

require SecYEG and SecA [34–36]. The different require-
ments suggest that translocation components function as
modules responsible for specific tasks. For example, leader
peptidase has a large C-terminal domain in the periplasm
that requires SecA in addition to SecYEG [37]. Similarly,
large periplasmic loops extending 100 amino acid residues in
M13 procoat mutants, need SecA and SecYE for translo-
cation, whereas small loops do not stimulate the transloca-
tion ATPase of SecA [6,38]. The result obtained here that
KdpD is independent of SecA is therefore consistent with
previous findings.
The results obtained for the KdpD protein showed that
the use of short epitopes can provide valuable data for the
analysis of how specific regions of a membrane protein are
translocated across the membrane. The analysis of the
translocation requirements showed that the first periplasmic
loop of KdpD with the epitope tag was independent of the
Sec components, whereas the longer second periplasmic
loop of KdpD required SecE and YidC for efficient
translocation (Figs 3B and 4B). This indicates that the
multispanning membrane protein actually translocates in
pairs of transmembrane helices and that individual pairs
may have different insertion requirements, depending on the
connecting loops. Interestingly, the two translocation events
observed for KdpD with the epitope tags correspond to
Fig. 8. YidC is required for efficient membrane
insertion of KdpD-C with the epitope tag (A)
but not for KdpD-C without the epitope tag (B).
JS7131 cells bearing the pMS119 plasmids
encoding either KdpD-C with the T7 epitope

tag (A) or KdpD-C without the tag (B) were
depleted of YidC as described in the legend
of Fig. 1. After subjecting the cells to a
protease accessibility assay, the proteins were
immunoprecipitated with antiserum to KdpD
andanalysedbySDS/PAGEandphosphori-
maging. PK, proteinase K.
Ó FEBS 2003 KdpD membrane insertion (Eur. J. Biochem. 270) 1731
those of the split double-spanning proteins (Fig. 5). This
underlines that membrane proteins are inserted not in a
linear movement, but rather as individual domains. Experi-
ments with leader peptidase had shown earlier that the
N-terminal tail and the large C-terminal domain are
separately translocated [39]. The pairwise organization of
multispanning membrane proteins is also suggested from
single-molecule force spectroscopy where a molecular
tweezer was connected to the C-terminus of bacteriorho-
dopsin [40]. When the protein was pulled out of the
membrane, two transmembrane regions were preferentially
released together.
Unexpectedly, YidC is not important for the membrane
insertion of the KdpD wild-type protein (Fig. 1D). Other
Sec-independent proteins, such as Pf3 coat and M13
procoat strongly depend on YidC [14,41]. In contrast to
KdpD, the M13 procoat protein has a periplasmic region of
20 amino acid residues including five charged residues.
Interestingly, different mutants with alterations in the loop
region of procoat have shown that the number of the
charged residues determines the extent of YidC dependency.
A mutant that has no charged residue in the 20 amino acid

loop showed only a minor interference by YidC depletion
[14]. This might explain why KdpD is independent of Sec
and YidC as the periplasmic loops are much shorter and the
translocation of these periplasmic regions should require
less energy. An extension of the second loop of KdpD by 17
amino acid residues indeed resulted in the requirement of
the YidC protein, suggesting that YidC promotes the
translocation of larger periplasmic regions.
Interestingly, the two periplasmic loops of KdpD that
were extended with short epitope tags differed also for their
need of a membrane potential. Whereas KdpD-N is not
translocated in the absence of a potential, KdpD-C was only
marginally affected. Potential-dependent translocation of
negatively charged regions has been extensively studied with
the M13 procoat protein. The periplasmic loop of the
procoat protein has a net negative charge of )3. Procoat
mutants were studied where the charge of the periplasmic
loop has been changed [42]. Only the negatively charged
regions show potential dependence and the more negatively
charged residues present in the loop region of procoat the
higher is the potential dependency. The procoat mutant
with a net charge of )1 in the periplasmic loop was only
marginally affected. In agreement with this, the KdpD-N
protein with the HA-tag has three aspartic acyl residues in
the periplasmic loop, which might contribute to the strong
dependency on the membrane potential.
For the Sec-independent Pf3 coat protein it was shown
that a mutant with a longer hydrophobic region inserts
independent of YidC and of the electrochemical membrane
potential [43,44]. It was proposed that the hydrophobic effect

of the transmembrane region might drive the insertion step
and that this process can occur without any other protein.
Under limited hydrophobicity, the electrochemical mem-
brane potential and YidC become then essential factors.
These findings can be applied to the insertion of the KpdD
protein. If a protein can autonomously insert into the
membrane, the hydrophobic energy from the insertion of
the hydrophobic parts of the protein should compensate the
energy costs of the transfer of its hydrophilic part. Taking the
hydrophobicity scale [43] to calculate the free energy that
the transmembrane regions of KdpD can contribute to the
membrane insertion we get about DG
>
¼ )144 kJÆmol
)1
for
the first two helices. The transfer of the periplasmic
loop between these helices to translocate costs DG
>
¼
65 kJÆmol
)1
, which should allow an autonomous insertion.
However, when the HA-tag is included in the hydrophi-
lic region the energy cost increases to about DG
>
¼
200 kJÆmol
)1
. This would not allow membrane insertion

and might explain the strong dependence of KdpD-N on the
pmf. The membrane insertion of the helices 3 and 4 con-
tributes with only DG
>
¼ )63 kJÆmol
)1
. The second peri-
plasmic loop of the wild-type costs DG
>
¼ 105 kJÆmol
)1
,
and with the added T7-epitope DG
>
¼ 150 kJÆmol
)1
is required to pass the membrane. The hydrophobic contri-
bution cannot compensate the energy costs of the transfer of
the periplasmic loop. This may explain why YidC and Sec
play a role in the translocation of the C-terminal loop with
the T7-epitope tag.
Taken together, the data presented here show that KdpD
inserts unassisted from the Sec translocase and YidC into
the inner membrane of E. coli.Thisismostlikelybecause
KdpD has very short periplasmic regions that cost little
energy to translocate suggesting that the membrane inser-
tion occurs autonomously. The unassisted insertion path-
way may also be used by a large number of E. coli
membrane proteins with short periplasmic loops that have
not yet been analysed for membrane insertion. So far, the

unassisted membrane insertion pathway is known from
thylakoids [45,46], where a subset of membrane proteins
show independence of SRP, the Sec components and Alb3,
the plant homologue of YidC.
Acknowledgements
We would like to thank Drs K. Jung and K-H. Altendorf for
generously providing us with the initial plasmids (pPV5, pBD) and
KdpD antiserum and Drs H-G. Koch, M. Mu
¨
ller, R. Dalbey and
D. Kiefer for valuable discussions. This work was supported by the
Deutsche Forschungsgemeinschaft Sonderforschungsbereich 495.
References
1. Zimmann, P., Puppe, W. & Altendorf, K. (1995) Membrane
topology analysis of the sensor kinase KdpD of Escherichia coli.
J. Biol. Chem. 270, 28282–28288.
2. Heermann, R., Altendorf, K. & Jung, K. (1998) The turgor sensor
KdpD of Escherichia coli is a homodimer. Biochim. Biophys. Acta
1415, 114–124.
3. Puppe, W., Zimmann, P., Jung, K., Lucassen, M. & Altendorf, K.
(1996) Characterization of truncated forms of the KdpD protein,
thesensorkinaseoftheK
+
-translocating Kdp system of
Escherichia coli. J. Biol. Chem. 271, 25027–25034.
4. Pogliano, K.J. & Beckwith, J. (1994) Genetic and molecular
characterization of the Escherichia coli secD operon and its pro-
ducts. J. Bacteriol. 176, 804–814.
5. Cao, G., Chen, S., Whitley, P., von Heijne, G., Kuhn, A. &
Dalbey, R.E. (1994) Synergistic insertion of two hydrophobic

regions drives Sec-independent membrane protein assembly.
J. Biol. Chem. 269, 26898–26903.
6. Roos,T.,Kiefer,D.,Hugenschmidt,S.,Economou,A.&Kuhn,
A. (2001) Indecisive M13 procoat protein mutants bind to SecA
but do not activate the translocation ATPase. J. Biol. Chem. 276,
37909–37915.
1732 S. J. Facey and A. Kuhn (Eur. J. Biochem. 270) Ó FEBS 2003
7. Bassilana, M. & Gwizdek, C. (1996) In vivo membrane assembly of
the E. coli polytopic protein, melibiose permease, occurs via a Sec-
independent process which requires the proton motive force.
EMBO J. 15, 5202–5208.
8. Gafvelin, G. & von Heijne, G. (1994) Topological ÔfrustrationÕ in
multispanning E. coli inner membrane proteins. Cell 77, 401–412.
9. De Gier, J W.L., Scotti, P.A., Sa
¨
a
¨
f, A., Valent, Q.A., Kuhn, A.,
Luirink, J. & von Heijne, G. (1998) Differential use of the signal
recognition particle translocase targeting pathway for inner
membrane protein assembly in Escherichia coli. Proc.NatlAcad.
Sci. USA 95, 14646–14651.
10. Driessen, A.J., Manting, E.H. & van der Does, C. (2001) The
structural basis of protein targeting and translocation in bacteria.
Nature Struct. Biol. 8, 492–498.
11. Koch, H G. & Mu
¨
ller, M. (2000) Dissecting the translocase and
integrase functions of the Escherichia coli SecYEG translocon.
J. Cell Biol. 150, 689–694.

12. Dalbey, R.E. & Kuhn, A. (2000) Evolutionarily related insertion
pathways of bacterial, mitochondrial, and thylakoid membrane
proteins. Annu. Rev. Cell. Dev. Biol. 16, 51–87.
13. Samuelson, J.C., Chen, M., Jiang, F., Mo
¨
ller, I., Wiedmann, M.,
Kuhn, A., Phillips, G.J. & Dalbey, R.E. (2000) YidC mediates
membrane protein insertion in bacteria. Nature 406, 637–641.
14. Samuelson, J.C., Jiang, F., Yi, L., Chen, M., de Gier, J.W., Kuhn,
A. & Dalbey, R.E. (2001) Function of YidC for the insertion of
M13 procoat protein in E. coli: Translocation of mutants that
show differences in their membrane potential dependence and Sec-
requirement. J. Biol. Chem. 276, 34847–34852.
15. Walderhaug, M.O., Polarek, J.W., Voelkner, P., Daniel, J.M.,
Hesse,J.E.,Altendorf,K.&Epstein,W.(1992)KdpDandKdpE,
proteins that control expression of the kdpABC operon, are
members of the two-component sensor-effector class of regulators.
J. Bacteriol. 174, 2152–2159.
16. Jung, K. & Altendorf, K. (1998) Truncation of amino acids 12–
128 causes deregulation of the phosphatase activity of the sensor
kinase KdpD of Escherichia coli. J. Biol. Chem. 273, 17406–17410.
17. Studier, F.W., Rosenberg, A.H., Dunn, J.J. & Dubendorff, J.W.
(1990) Use of T7 RNA polymerase to direct expression of cloned
genes. Methods Enzymol. 185, 60–89.
18. Traxler, B. & Murphy, C. (1996) Insertion of the polytopic
membrane protein MalF is dependent on the bacterial secretion
machinery. J. Biol. Chem. 271, 12394–12400.
19. Maniatis, T., Fritsch, E.F. & Sambrook, J. (1982) Molecular
Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory,
Cold Spring Harbor, NY.

20. Guzman,L M.,Belin,D.,Carson,M.J.&Beckwith,J.(1995)
Tight regulation, modulation, and high-level expression by vectors
containing the arabinose P
BAD
promoter. J. Bacteriol. 177, 4121–
4130.
21. Balzer, D., Ziegelin, G., Pansegrau, W., Kruft, V. & Lanka, E.
(1992) KorB protein of promiscuous plasmid RP4 recognizes
inverted sequence repetitions in regions essential for conjugative
plasmid transfer. Nucleic Acids Res. 20, 1851–1858.
22. Wilson, I.A., Niman, H.L., Houghten, R.A., Cherenson, A.R.,
Connolly,M.L.&Lerner,R.A.(1984)Thestructureofananti-
genic determinant in a protein. Cell 37, 767–778.
23. Oliver, D.B., Cabelli, R.J., Dolan, K.M. & Jarosik, G.P. (1990)
Azide-resistant mutants of Escherichia coli alter the SecA protein,
an azide-sensitive component of the protein export machinery.
Proc.NatlAcad.Sci.USA87, 8227–8231.
24. Murphy, C.K. & Beckwith, J. (1994) Residues essential for the
function of SecE, a membrane component of the Escherichia coli
secretion apparatus, are located in a conserved cytoplasmic region.
Proc.NatlAcad.Sci.USA91, 2557–2561.
25. Daniels, C.J., Bole, D.G., Quay, S.C. & Oxender, D.L. (1981)
Role for membrane potential in the secretion of protein into the
periplasm of Escherichia coli. Proc. Natl Acad. Sci. USA 78, 5396–
5400.
26. McGovern, K. & Beckwith, J. (1991) Membrane insertion of the
Escherichia coli MalF protein in cells with impaired secretion
machinery. J. Biol. Chem. 266, 20870–20876.
27. MacFarlane, J. & Mu
¨

ller, M. (1995) The functional integration of
a polytopic membrane protein of Escherichia coli is dependent on
the bacterial signal-recognition particle. Eur. J. Biochem. 233,766–
771.
28. Whitley,P.,Zander,T.,Ehrmann,M.,Haardt,M.,Bremer,E.&
von Heijne, G. (1994) Sec-independent translocation of a 100-
residue periplasmic N-terminal tail in the E. coli inner membrane
protein ProW. EMBO J. 13, 4653–4661.
29. Herzlinger, D., Viitanen, P., Carrasco, N. & Kaback, H.R. (1984)
Monoclonal antibodies against the lac carrier protein from
Escherichia coli. 2. Binding studies with membrane vesicles and
proteoliposomes reconstituted with purified lac carrier protein.
Biochemistry 23, 3688–3693.
30. Nilsson, I. & von Heijne, G. (1993) Determination of the distance
between oligosaccharyltransferase active site and the endoplasmic
reticulum membrane. J. Biol. Chem. 268, 5798–5801.
31. Kiefer, D., Hu, X., Dalbey, R. & Kuhn, A. (1997) Negatively
charged amino acid residues play an active role in orienting the
Sec-independent Pf3 coat protein in the Escherichia coli inner
membrane. EMBO J. 16, 2197–2204.
32. Kihara, A., Akiyama, Y. & Ito, K. (1995) FtsH is required for
proteolytic elimination of uncomplexed forms of SecY, an essen-
tial protein translocase subunit. Proc. Natl Acad. Sci. USA 92,
4532–4536.
33.Beck,K.,Wu,L F.,Brunner,J.&Mu
¨
ller, M. (2000) Dis-
crimination between SRP- and SecA/SecB-dependent substrates
involves selective recognition of nascent chains by SRP and trigger
factor. EMBO J. 19, 134–143.

34. Wolfe, P.B., Rice, M. & Wickner, W. (1985) Effects of two sec
genes on protein assembly into the plasma membrane of Escheri-
chia coli. J. Biol. Chem. 260, 1836–1841.
35.Urbanus,M.L.,Froederberg,L.,Drew,D.,deGier,J W.,
Brunner, J., Oudega, B. & Luirink, J. (2002) Targeting, insertion
and localization of E. coli YidC. J. Biol. Chem. 277, 12718–12723.
36. Koch,H G.,Moser,M.,Schimz,K.L.&Mu
¨
ller, M. (2002) The
integration of YidC into the cytoplasmic membrane of Escherichia
coli requires the signal recognition particle, SecA and SecYEG.
J. Biol. Chem. 277, 5715–5718.
37. Andersson, H. & von Heijne, G. (1993) Sec dependent and sec
independent assembly of E. coli inner membrane proteins: the
topological rules depend on chain length. EMBO J. 12, 683–691.
38. Kuhn, A. (1988) Alterations in the extracellular domain of M13
procoat protein make its membrane insertion dependent on secA
and secY.Eur.J.Biochem.177, 267–271.
39. Lee, J I., Kuhn, A. & Dalbey, R.E. (1992) Distinct domains of
an oligotopic membrane protein are Sec-dependent and Sec-
independent for membrane insertion. J. Biol. Chem. 267, 938–943.
40. Oesterhelt,F.,Oesterhelt,D.,Pfeiffer,M.,Engel,A.,Gaub,H.E.
&Mu
¨
ller, D.J. (2000) Unfolding pathways of individual bacterio-
rhodopsins. Science 288, 143–146.
41. Chen, M., Samuelson, J.C., Jiang, F., Mu
¨
ller, M., Kuhn, A. &
Dalbey, R.E. (2002) Direct interaction of YidC with the Sec-in-

dependent Pf3 coat protein during its membrane protein insertion.
J. Biol. Chem. 277, 7670–7675.
42. Cao, G., Kuhn, A. & Dalbey, R.E. (1995) The translocation of
negatively charged residues across the membrane is driven by the
electrochemical potential: evidence for an electrophoresis-like
membrane transfer mechanism. EMBO J. 14, 866–875.
43. Kiefer, D. & Kuhn, A. (1999) Hydrophobic forces drive sponta-
neous membrane insertion of the bacteriophage Pf3 coat protein
without topological control. EMBO J. 18, 6299–6306.
Ó FEBS 2003 KdpD membrane insertion (Eur. J. Biochem. 270) 1733
44. Ridder, A.N.J.A., Kuhn, A., Killian, A. & de Kruijff, B. (2001)
Anionic lipids stimulate Sec-independent insertion of a membrane
protein lacking charged amino acid side chains. EMBO Report 2,
403–408.
45. Mant, A., Woolhead, C.A., Moore, M., Henry, R. & Robinson,
C. (2001) Insertion of PsaK into the thylakoid membrane in a
ÔhorseshoeÕ conformation occurs in the absence of signal recogni-
tion particle, nucleoside triphosphates, or functional Albino3.
J. Biol. Chem. 276, 36200–36206.
46. Woolhead, C.A., Thompson, S.J., Moore, M., Tissier, C.,
Mant, A., Rodger, A., Henry, R. & Robinson, C. (2001)
Distinct Albino3-dependent and – independent pathways for
thylakoid membrane protein insertion. J. Biol. Chem. 276, 40841–
40846.
1734 S. J. Facey and A. Kuhn (Eur. J. Biochem. 270) Ó FEBS 2003