Báo cáo Y học: Diffusion through channel derivatives of the Escherichia coli FhuA transport protein doc

Diffusion through channel derivatives of the
Escherichia coli
FhuA
transport protein
Michael Braun
1
, Helmut Killmann
1
, Elke Maier
2
, Roland Benz
2
and Volkmar Braun
1
1
Mikrobiologie/Membranphysiologie, Universita
¨
tTu
¨
bingen, Germany;
2
Lehrstuhl fu
¨
r Biotechnologie, Theodor-Boveri-Institut
(Biozentrum), Universita
¨
tWu
¨
rzburg, Germany
FhuA is a multifunctional protein in the outer membrane of
Escherichia coli that actively transports [Fe

3+
]ferrichrome,
the antibiotics albomycin and rifamycin CGP 4832, and
mediates sensitivity of cells to the unrelated phages T5, T1,
/80 and UC-1, and to colicin M and microcin J25. The
energy source of active transport is the proton motive force
of the cytoplasmic membrane that is required for all FhuA
functions except for infection by phage T5. The FhuA crystal
structure reveals 22 antiparallel transmembrane b-strands
that form a b-barrel which is closed by a globular N-terminal
domain. FhuA still displays active transport and sensitivity
to all ligands except microcin J25 when the globular domain
(residues 5–160) is excised and supports weakly unspecific
diffusion of substrates across the outer membrane. Here it is
shown that isolated FhuAD5–160 supported diffusion of
ions through artificial planar lipid bilayer membranes but
did not form stable channels. The double mutant FhuAD5–
160 D322–336 lacking in addition to the globular domain
most of the large surface loop 4 which partially constricts the
channel entrance, displayed an increased single-channel
conductance but formed no stable channels. It transported
in vivo [Fe
3+
]ferrichrome with 45% of the rate of wild-type
FhuA and did not increase sensitivity of cells to antibiotics.
In contrast, a second FhuA double mutant derivative which
in addition to the globular domain contained a deletion of
residues 335–355 comprising one-third of surface loop 4 and
half of the transmembrane b-strand 8 formed stable channels
in lipid bilayers with a large single-channel conductance of

2.5 nS in 1
M
KCl. Cells that synthesized FhuAD5–160
D335–355 showed an increased sensitivity to antibiotics and
supported diffusion of maltodextrins, SDS and ferrichrome
across the outer membrane. FhuAD5–160 D335–355 showed
no FhuA specific functions such as active transport of
[Fe
3+
]ferrichrome or sensitivity to the other FhuA ligands.
It is concluded that FhuAD5–160 D335–355 assumes a
conformation that is incompatible with any of the FhuA
functions.
Keywords: channel; Escherichia coli; FhuA transport
protein.
The outer membrane of Escherichia coli forms a permeabi-
lity barrier for hydrophilic substrates larger than 600 D [1].
Smaller substrates diffuse through water-filled pores formed
by the porins. Most of the siderophores secreted by bacteria
and fungi to solubilize Fe
3+
are larger than 600 Da [2]. In
addition, the concentration of the Fe
3+
siderophores is very
low so that diffusion does not provide sufficient iron for
growth, which is in the order of 10
5
ions per E. coli cell per
generation. Therefore, Fe

3+
siderophores are actively taken
up by an energy-requiring transport process. The proton
motive force of the cytoplasmic membrane drives their
transport across the outer membrane [3] and ATP energizes
uptake across the cytoplasmic membrane [4]. FhuA trans-
ports ferrichrome, the structurally related antibiotic albo-
mycin, the unrelated antibiotic rifamycin CGP 4832, and
serves as receptor for the phages T5, T1, /80 and UC-1, for
the toxic colicin M protein and the toxic microcin J25
peptide [4].
The crystal structure of FhuA reveals 22 antiparallel
transmembrane b-strands that form a b-barrel which is
closed by a globular domain, also called cork [5] or plug [6].
It is thought that energy input from the cytoplasmic
membrane opens the channel of the b-barrel in that the
globular domain somehow moves and ferrichrome dissoci-
ates from its binding site which is formed by 10 amino acid
residues [5]. The energy transfer from the cytoplasmic
membrane to FhuA in the outer membrane is mediated by
the TonB protein that is located in the periplasm and
anchored by the N-proximal end in the cytoplasmic
membrane [7]. Two additional proteins, ExbB and ExbD,
are associated with TonB and are required for TonB
activity. The subcellular localization of ExbD is similar to
TonB [8], whereas ExbB is anchored with three transmem-
brane segments in the cytoplasmic membrane with most of
the protein in the cytoplasm [9].
FhuA changes its conformation upon binding of ferri-
chrome. The crystal structure shows a small movement

(1 A
˚
) of the globular domain towards the bound ferri-
chrome but a large movement (17 A
˚
) of residues 19 and 22
of the globular domain that are exposed to the periplasm.
Binding of ferrichrome enhances interaction of FhuA with
TonB [10], which may be facilitated by the structural
transition of 17 A
˚
. Residues 1–18 are not seen in the crystal
and therefore are thought to be flexible. This segment
contains the TonB box (residues 7–11) which has been
Correspondence to V. Braun, Mikrobiologie/Membranphysiologie,
Universita
¨
tTu
¨
bingen, Auf der Morgenstelle 28, D-72076 Tu
¨
bingen,
Germany. Fax: + 49 7071 295843, Tel.: + 49 7071 2972096,
E-mail:
Abbreviations: LDAO, N,N-dimethyldodecylamine-N-oxide.
(Received 3 June 2002, revised 6 August 2002,
accepted 21 August 2002)
Eur. J. Biochem. 269, 4948–4959 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03195.x
named according to the finding that the amino acid
replacements I9P (isoleucine in position 9 replaced by

proline) and V11D abolished the TonB-dependent FhuA
activities [11] and reduced interaction of FhuA with TonB
[12]. The Q160L and Q160K replacements in TonB partially
restored the activities of the FhuA TonB box mutants.
These results supported the notion that FhuA interacts
through the TonB box with region 160 of TonB. As in
FhuA the TonB box is contained near the N terminus of all
active outer membrane transport proteins and the group B
colicins which also require TonB to kill cells. Further
support of the TonB box concept comes from the sponta-
neous disulfide formation between cysteine residues intro-
duced into the TonB box of the BtuB protein and cysteine
residues introduced into region 160 of TonB. BtuB is similar
to FhuA and actively transports vitamin B
12
across the
outer membrane [13]. Site-directed spin labelling and
electron paramagnetic resonance studies revealed that
binding of vitamin B
12
to BtuB alters the conformation
and the dynamics of the TonB box segment [14].
At the time when the crystal structure of FhuA was not
yet available we arrived at an early tentative transmembrane
model of FhuA that proposed a prominent loop at the cell
surface from residue 316 to residue 355 [15]. The model was
experimentally derived from the proteolytic cleavage of
peptides of up to 16 amino acids which had been inserted
into FhuA, and by computer-based prediction programs. In
support of this notion we showed that this region serves as

the binding site for the phages T1, T5 and /80, because
synthetic peptides covering this region inhibited infection by
the phages [16]. Under the assumption that the surface loop
might also control a putative channel of FhuA we excised
residues 322–355 which indeed converted FhuA into an
open channel that exhibited stable single-channel conduct-
ance in artificial lipid bilayer membranes [17]. Excision of
residues 322–336 and 335–355 resulted in no stable single-
channel conductance [18]. The crystal structure later
revealed that the largest surface loop (L4), indeed extends
from residues 318–339 and that residues 340–355 are located
above the outer membrane lipid bilayer and form half of the
b-sheet number 8. The cork domain revealed by the crystal
structure could not be predicted by the methods used.
To understand the role of the cork domain in channel
formation of FhuA we have constructed FhuAD5–160
based on the crystal structure under the assumption that
excision of the entire globular domain would convert FhuA
into an open channel similar to the channel formed by
FhuAD322–355 [17] and abolish all TonB-dependent FhuA
activities. However, FhuAD5–160 still displayed all TonB-
related activities between 40 and 100% of wild-type FhuA
activity, depending on the function tested, and the per-
meability of the outer membrane for substrates and
antibiotics increased only slightly [19]. These findings were
supported by a study using FhuAD5–160 derivatives of
Salmonella paratyphi BandSalmonella enterica serovar
Typhimurium which in addition showed that hybrid
proteins consisting of the b-barrel of one strain and the
globular domain of another strain were functional [20]. An

investigation with an E. coli deletion derivative of FepA, the
outer membrane transport protein for ferric enterobactin, in
which the cork domain was removed and comparison with
FhuAD5–160 supported the TonB dependent activities of
the b-barrels [21].
To gain further insights into the mode of action of
FhuAD5–160, and in particular to the transport activity of
the b-barrel, we determined in this report single-channel
conductance of isolated FhuAD5–160 incorporated into
artificial bilayer membranes. We wanted to relate the
in vitro activities of FhuAD5–160 with the in vivo activities.
Since FhuAD5–160 formed no stable channels in vitro we
deleted loop 4 which constricts half the channel entrance of
FhuA to about half the area of the total cross-section [6]. As
no stable channels were formed by FhuAD5–160 D322–336,
we combined deletion D5–160 with deletion D335–355
which as a single deletion did not display stable single-
channel conductance [18]. FhuAD5–160 D335–355 formed
large stable channels which were consistent with the in vivo
unspecific increase of the outer membrane per-
meability. FhuAD5–160 D335–355 did not actively trans-
port [Fe
3+
]ferrichrome, in contrast with FhuAD5–160
D322–336 which displayed TonB-dependent transport
activity.
MATERIALS AND METHODS
Bacterial strains, plasmids and growth conditions
The E. coli strains and plasmids used are listed in Table 1.
Cells were grown in TY medium [10 gÆL

)1
bactotryptone
(Difco Laboratories), 5 gÆL
)1
yeast extract, 5 gÆL
)1
NaCl]
or NB medium (8 gÆL
)1
nutrient broth, 5 gÆL
)1
NaCl,
pH 7) at 37 °C. To reduce the available iron of the NB
medium, 2,2¢-dipyridyl (0.2 m
M
) was added (NBD
medium). The antibiotic ampicillin (40 gÆmL
)1
)wasadded
when required.
Plasmids pHK234 and pHK237 were digested with MluI
and SalI and ligated into MluI/SalI-cleaved plasmid
pHK763 resulting in plasmids p7634 and p7637, respect-
ively. Plasmids p7634 and p7637 were digested with BspEI
and EcoRI and ligated into BspEI/EcoRI-cleaved plasmid
pBK7 resulting in plasmids pDM234 and pDM237,
respectively.
Plasmid pSKF405-04 was digested with MluIandBstEII
and ligated into MluI/BstEII-cleaved plasmid pHK763
resulting in plasmid p76405. To introduce a His

6
-tag 10 lg
of the primers His1 5¢-GATCATCACCATCACCATCAC-
3¢ and His2 5¢-GATCGTGATGGTGATGGTGAT-3¢
were mixed and incubated for 30 min at 83 °C. The mixture
was then allowed to cool down slowly to 25 °C during
30 min followed by an incubation for 5 min at room
temperature. The annealed His1 and His2 primers were
ligated into BglII-cleaved plasmid p76405 resulting in
plasmid pHK763H.
Plasmid pHK763H was digested with SalIandEcoRI
and ligated into SalI/EcoRI-cleaved plasmids pBK7, p7634,
pDM234, p7637, and pDM237 resulting in plasmids
pBK7H, p7634H, pDM234H, p7637H, and pDM237H,
respectively.
Recombinant DNA techniques
Isolation of plasmids, use of restriction enzymes, liga-
tion, agarose gel electrophoresis, and transformation
followed standard techniques [22]. All genetic construc-
tions were examined by DNA sequencing using the
dideoxy chain-termination method with fluorescence-
Ó FEBS 2002 FhuA transport protein of E. coli (Eur. J. Biochem. 269) 4949
labelled or unlabelled nucleotides (Auto Read Sequencing
Kit, Pharmacia Biotech) and the A.L.F. sequencer
(Pharmacia).
Protein analytical methods
To show the FhuA proteins in cells that were grown under
the same conditions as the phenotype assays were per-
formed, E. coli HK97 fhuA was transformed with fhuA
wild-type and fhuA mutant genes and grown overnight in

NB medium. The overnight cultures were used to inoculate
fresh NB medium and the cultures were grown at 37 °Ctoa
D
578
of 1.0 before the outer membrane fractions were
isolated by lysing cells with lysozyme–EDTA, followed by
solubilization of the cytoplasmic membrane in 0.2% Triton
X-100 and differential centrifugation. The proteins of the
undissolved outer membrane fraction were dissolved by
heating in sample buffer, separated by SDS/PAGE and
stained with Serva blue [23].
Phenotype assays
All phenotype assays were carried out with freshly trans-
formed E. coli K-12 strains HK97 aroB fhuA fhuE and
HK99 aroB fhuA tonB. These strains carry the same four
amino acid replacements and an amino acid deletion in fhuA
and contain the mutated FhuA protein in the outer
membrane [23]. The plasmid-encoded fhuA genes in the
transformants were transcribed from the fhuA promoter.
Sensitivity of cells to the FhuA ligands was tested by
spotting 10-fold (phages T1, T5, /80, and colicin M) or
3-fold (microcin J25, rifamycin CGP 4832, and albomycin)
diluted solutions (3 lL) on TY agar plates overlaid with
3 mL TY soft agar containing 10
8
cells of the strain to be
tested. The colicin M solution was a crude extract of a strain
carrying plasmid pTO4 cma cmi [24]. The microcin J25
solution was a supernatant of E. coli MC4100 carrying the
plasmid pTUC203 mcjABCD [25] after growth of the

transformants in brain heart infusion medium (37 gÆL
)1
;
Difco Laboratories) at 37 °C.
Growth inhibition by SDS and various antibiotics was
determined by placing filter paper discs supplemented with
10 lL of the agents in concentrations as indicated on TY
agar plates overlaid with 3 mL TY soft agar containing 10
8
cells of the strain to be tested. The tests were performed in
parallel with TY agar plates and TY soft agar both
containing ferrichrome at a final concentration of 1 l
M
.
Growth promotion by siderophores was tested by placing
filter paper discs containing 10 lL of a siderophore solution
of different concentrations on NBD agar plates overlaid
with 3 mL NB soft agar containing 10
8
cells of the strain to
be tested. After overnight incubation, the diameter
and the growth density around the filter paper discs were
determined.
Table 1. E. coli strains and plasmids used in this study.
Strain or plasmid Genotype or phenotype Reference or source
Strains
AB2847 aroB tsx malT thi [34]
HK99 AB2847 tonB fhuA [17]
CH1857 AB2847 DfhuACDB tonB [19]
HK97 F

–
araD139 lacU169 rpsL150 relA1 flbB5301 deoC1
ptsF25 rbsR aroB thi fhuE::kplac Mu53fhuA
[23]
BL21 (DE3) omp8 F
–
hsdS
B
B(r
B
–
m
B
–
) gal ompT dcm (DE3) DlamB
ompF::Tn5 DompA T7 polymerase under lacUV5 control
[35]
CH21 BL21 (DE3) omp8 fhuA [20]
KB419 lamB Krieger-Brauer
Plasmids
pHK763 pT7-6 fhuA wild-type [17]
pHK763H pT7-6 fhuA (P
405
PDH
6
DLA V
406
) This study
pAB pT7-6 fhuA with BamHI E159D [36]
pBK7 pT7-6 fhuA D5–160 E3D [19]

pBK7H pT7-6 fhuA D5–160 E3D (P
405
PDH
6
DLA V
406
) This study
pHK234 pBluescript SK + fhuAD322–336 (P
321
PDL K
337
) [18]
p7634 pT7-6 fhuA D322–336 (P
321
PDL K
337
) This study
p7634H pT7-6 fhuA D322–336 (P
321
PDL K
337
)(P
405
PDH
6
DLA V
406
) This study
pDM234 pT7-6 fhuA D5–160 D322–336 E3D (P
321

PDL K
337
) This study
pDM234H pT7-6 fhuA D5–160 D322–336 E3D (P
321
PDL K
337
)(P
405
PDH
6
DLA V
406
) This study
pHK237 pBluescript SK + fhuAD335–355 (P
334
DL S
356
) [18]
p7637 pT7-6 fhuA D335–355 (P
334
DL S
356
) This study
p7637H pT7-6 fhuA D335–355 (P
334
DL S
356
)(P
405

PDH
6
DLA V
406
) This study
pDM237 pT7-6 fhuA D5–160 D335–355 E3D (P
334
DL S
356
) This study
pDM237H pT7-6 fhuA D5–160 D335–355 E3D (P
334
DL S
356
)(P
405
PDH
6
DLA V
406
) This study
pSKF405-04 pBluescript SK + fhuA (P
405
PDLA V
406
) [15]
p76405 pT7-6 fhuA (P
405
PDLA V
406

) This study
pTO4 pBR322 cma cmi [24]
pTUC203 pACYC184 mcjABCD [25]
pT7-6 Amp
r
[37]
4950 M. Braun et al. (Eur. J. Biochem. 269) Ó FEBS 2002
Growth promotion by maltodextrins was tested with the
lamB strain KB419 (as a control) and with strain KB419
transformed with plasmids encoding various FhuA deriv-
atives. Overnight cultures were washed twice with M9
medium and adjusted to a D
578
of1.0inM9minimal
medium. The test was performed by placing filter paper
discs supplemented with 10 lL of a 40% solution of
maltodextrins (maltose to maltohexaose) on M9 minimal
agar plates that contained no other carbon source overlaid
with M9 minimal top agar containing 100 lL of the strain
to be tested. After incubation overnight at 37 °Cthe
diameter and the growth density around the filter discs were
determined.
Ferrichrome uptake assays
E. coli K-12 strains HK97 aroB fhuA fhuE,andHK99aroB
fhuA tonB, freshly transformed with the plasmids to be
tested were grown overnight on TY plates. Cells were
washed and suspended in transport medium (M9 salts [26],
0.4% glucose), and the cell density was adjusted to a D
578
of

0.5. Free iron ions were removed by adding 25 lL10m
M
nitrilotriacetate, pH 7.0 to 1 mL cells. After incubation for
5min at37°C, time-dependent ferrichrome uptake was
started by adding [
55
Fe
3+
]ferrichrome to a final concentra-
tion of 1 l
M
in the case of transport assays and 10 l
M
when
diffusion of ferrichrome into cells was tested. In the latter
case, E. coli HK99 aroB fhuA tonB wasusedasthetest
strain and a 150-fold surplus of nonradioactive ferrichrome
was added as a chase after 17 min to remove adsorbed
ferrichrome molecules. The concentration-dependent
uptake assays were started by adding [
55
Fe
3+
]ferrichrome
to final concentrations of 1, 3, 6, or 10 l
M
. Samples of
50 lL or 100 lL were withdrawn and added to 10 mL
0.1
M

LiCl. Cells were harvested on cellulose nitrate filters
(poresize0.45lm; Sartorius AG) and washed twice with
5mL0.1
M
LiCl. The filters were dried, and the radio-
activity was determined by liquid scintillation counting.
Ferrichrome binding assay
E. coli CH1857 aroB DfhuACDB tonB, freshly transformed
with the plasmids to be tested were grown overnight on TY
plates. Thirty lL of a 50% NaI solution (density 1.5 gÆcm
)3
)
wasoverlaidwith80lL silicone oil PN200 (density
1.03 gÆcm
)3
in a 400-lL microtest tube). The binding assay
was started by adding [
55
Fe
3+
]ferrichrome to a final
concentration of 1 l
M
to cells prepared as described in
ferrichrome uptake assays. At the times indicated, samples
of 50 or 100 lL were withdrawn and applied onto the
silicone oil layer in the microtest tube. The tubes were
centrifuged immediately for 90 s in a Beckman Microfuge
E. The cells passed the silicone oil layer according to their
density (1.2–1.3 gÆcm

)3
),andaccumulatedontopoftheNaI
layer. The residual radiolabelled ferrichrome in the binding
medium remained on top of the silicone layer. After the
centrifugation step, cells were stored in the test tube at
room temperature until the uptake assay was completed
(H. Killmann and G. Gestwa, unpublished data).
At the end of the assay, the microtest tubes were cut with
a scalpel in the middle of the silicone oil layer and the lower
part of the test tube containing the cells was placed upside
down in a fresh test tube and centrifuged for 10 min at
10 000 g. The empty tube was removed, and the mixture of
NaI, silicone oil and cells was suspended in 900 lLH
2
Oand
completely transferred to a 20-mL polyethylene vial. After
adding 10 mL liquid scintillation counting cocktail,
the radioactivity was determined by liquid scintillation
counting.
Purification of the FhuA proteins
The E. coli K-12 strain CH21 freshly transformed with the
plasmids encoding the FhuA derivatives carrying a His
6
-tag
at residue 405 were grown in 250 mL TY medium to a D
578
of 0.5 before T7 RNA polymerase synthesis was induced by
adding isopropyl-b-
D
-thiogalactopyranoside to a final con-

centration of 1 m
M
. After shaking the cultures for addi-
tional 3 h at 37 °C the cells were harvested by
centrifugation. After suspending in 15 mL 0.2
M
Tris/HCl
(pH 8.0) that contained 1 mg deoxyribonuclease and two
tablets of COMPLETE (Roche), cells were disrupted with a
French press. The disrupted cells were mixed with 15 mL
extraction buffer (50 m
M
Tris/HCl pH 8.0, 10 m
M
MgCl
2
,
2% Triton X-100) and incubated for 10 min at room
temperature. The outer membrane fractions were isolated
by centrifugation for 1 h with 30 000 g. To remove the
added Triton X-100 the outer membranes were washed four
times with 10 mL H
2
O. The outer membranes were
suspended in 12 mL solubilization buffer containing
50 m
M
Tris/HCl pH 8.0, 1 m
M
EDTA, 1% N,N-dimethyl-

dodecylamine-N-oxide (LDAO). To solubilize the FhuA
proteins samples of 1 mL were shaken overnight at 15 °C,
then centrifuged and 5 mL of the pooled supernatant
fractions dialysed for 4 h at room temperature in 500 vols
binding buffer containing 50 m
M
sodium phosphate,
300 m
M
NaCl, 10 m
M
imidazole, 0.1% LDAO (pH 8.0).
The protein solutions were concentrated to 1 mL by
ultrafiltration (Centricon YM30, Millipore) before loading
onto a Ni
2+
–NTA agarose column equilibrated with
binding buffer. Chromatography was performed as des-
cribed by the manufacturer (Qiagen) with the exception that
all buffers contained 0.1% LDAO. The elution buffer
containing the purified proteins was replaced by a buffer
containing 0.2
M
Tris/HCl pH 8.0, 0.2% LDAO using
PD10 columns (Pharmacia).
Black lipid bilayer membrane experiments
Membranes were formed from a 1% (w/v) solution of
diphytanoyl PtdCho (Avanti Polar Lipids) in n-decane in a
Teflon cell consisting of two aqueous compartments con-
nected by a circular hole with an area of approximately

0.4 mm
2
[27,28]. The aqueous salt solutions (analytical
grade; Merck) were used unbuffered and had a pH of
approximately 6. The temperature was kept at 20 °C
throughout the experiments. The single-channel measure-
ments were performed with a pair of Ag/AgCl electrodes
(withsaltbridges)switchedinserieswithavoltagesource
and a current amplifier (Keithley 427). The amplified signal
was monitored with a storage oscilloscope and recorded
with a strip chart recorder. Stock solutions containing the
FhuA deletion derivatives were added after the lipid
membrane turned optically black to reflected light.
For determination of zero-current membrane potentials
the membranes were formed in a 50-m
M
KCl solution and
Ó FEBS 2002 FhuA transport protein of E. coli (Eur. J. Biochem. 269) 4951
insertion of pores was observed until a conductance of at
least 0.1 nS was reached corresponding to the formation of
a sufficient number of channels. Then the instrumentation
was switched to the measurement of the zero-current
potentials and a KCl gradient was established by adding
3
M
KCl solution to one side of the membrane while
stirring. The zero-current membrane voltage reached its
stationary value about 2–5 min after addition of the
concentrated KCl-solution and was analysed using the
Goldman–Hodgkin–Katz equation [29].

RESULTS
FhuAD5–160 does not form stable channels
in lipid bilayers
The FhuA b-barrel lacking the central N-terminal globular
domain was incorporated into lipid bilayer membranes to
determine the increase in conductance. To purify the
protein, the fhuAD5–160 gene was cloned into plasmid
pBK7H (Table 1) downstream of the phage T7 gene 10
promoter and specifically transcribed by the T7 RNA
polymerase. FhuAD5–160 was labelled with a His
6
-tag at
residue 405 [30] which is exposed at the cell surface [5,6].
FhuAD5–160 was contained in the outer membrane fraction
in lower amounts than wild-type FhuA (Fig. 1A). It was
purified by affinity chromatography using an Ni
2+
–NTA
agarose column (Fig. 1B). Control purifications were per-
formed with strain CH21 transformed with the pT7-6
expression vector.
FhuAD5–160 was added to the aqueous phase on one or
both sides bathing a black lipid bilayer membrane formed
from diphytanoyl PtdCho/n-decane across a small circular
hole. FhuAD5–160 increased the membrane conductance as
demonstrated by the current recording in 1
M
KCl (Fig. 2).
The conductance increase did not occur in a step-wise
fashion as found with porins of Gram-negative bacteria

[27,31]. This means that FhuAD5–160 failed to form stable
channels. The current recordings revealed a high degree of
current noise. The most frequently observed conductance
step was about 0.5 nS in 1
M
KCl. No conductance increase
was recorded when purified wild-type FhuA was added to
the lipid bilayer membranes. This was also the case in the
control recordings with purified samples of the control
strain CH21 transformed with pT7-6. Control measure-
ments were carried out with a 100-fold concentrated protein
solution compared to the measurements of the different
FhuA deletion derivatives (data not shown).
FhuAD5–160 D322–336 increased the permeability
of lipid bilayer membranes
The loop L4 reduces the entrance of the surface cavity of
FhuA to about half its diameter [6]. To see whether loop 4
restricts the permeability of the open channel that was
formed by removal of the globular domain, we constructed
Fig. 1. (A) Stained proteins after SDS/PAGE
of outer membrane fractions of E. coli HK97
fhuA transformed with plasmids (listed in
Table 1) encoding the FhuA proteins indicated
in the figure. Arrows denote wild-type FhuA
and the various FhuA deletion derivatives.
The molecular masses of standard proteins in
kDa are indicated. (B) His-tagged proteins
obtained by chromatography on Ni–agarose
columns as they were used for the lipid bilayer
experiments. Ten-lLsamplesofa0.5

mgÆmL
)1
protein solution were applied
per lane.
Fig. 2. Single-channel recording of a diphytanoyl PtdCho/n-decane
membrane in the presence of the FhuAD5–160 mutant. The aqueous
phase contained 1
M
KCl (pH 6) and  50 ng FhuAD5–160ÆmL
)1
.
The applied membrane potential was 20 mV; T ¼ 20 °C. Note
that the current did not increase in a step-wise fashion but showed a
high current noise indicating rapid fluctuations of the channel-forming
unit.
4952 M. Braun et al. (Eur. J. Biochem. 269) Ó FEBS 2002
His
6
-tagged FhuAD5–160 D322–336. The purified deletion
derivative (Fig. 1) increased the conductance of lipid bilayer
membranes but did not show uniform single-channel
conductance or a step-wise increase (Fig. 3). Higher time
resolution of the current than shown in Fig. 2 revealed
frequent and rapid opening and closing of channels which
was also observed with FhuAD5–160 (Fig. 2) but had a
much higher amplitude. The conductance of FhuAD5–160
D322–336 was higher than that of FhuAD5–160 (Table 2)
which indicated that removal of half of the L4 loop
increased the conductance of the b-barrel.
FhuAD5–160 D335–355 forms stable channels

Because FhuAD5–160 did not form stable channels in the
reconstitution experiments we attempted to combine the
D5–160 excision with the 322–355 deletion which we have
previously shown converts FhuA into a stable channel [17].
We failed to observe transformants which expressed
FhuAD5–160 D322–355, presumably because the protein
was toxic to cells. We then constructed FhuAD5–160 D335–
355. As reported previously [18] FhuAD335–355 increased
the membrane conductance only slightly and did not form
stable channels in lipid bilayer membranes. SDS/PAGE of
isolated outer membrane fractions identified the FhuAD5–
160 D335–355 protein. The amounts were lower than those
of wild-type FhuA or FhuAD5–160 (Fig. 1A).
FhuAD5–160 D335–355 purified as His
6
-tagged deriv-
ative on nickel agarose (Fig. 1B) inserted readily into planar
lipid bilayers and produced discrete step-wise current
increase at a transmembrane potential of 20 mV (Fig. 4).
If each step corresponded to a single channel, the unitary
conductance was  2.5nSin1
M
KCl. The conductance
steps were fairly homogeneous as shown by the histogram
(Fig. 5). Only a small number of small steps were observed,
which probably represent smaller substates of the open
channel (see Table 2 for a summary of the results of the lipid
bilayer experiments). The conductance of the FhuAD5–160
D335–355 mutant was smaller than that of FhuAD322–355
(3 nS) determined previously under otherwise identical

conditions [17].
Single-channel analysis of the FhuA deletion mutants
Table 2 shows the average single-channel conductance G of
the FhuA mutant proteins as a function of the KCl
concentration in the aqueous phase. Measurements were
performed from 0.1 to 3
M
KCl and with 1
M
LiCl and 1
M
KAc. Only FhuAD5–160 D335–355 displayed a linear
relationship between single-channel conductance and KCl
Table 2. Single-channel properties of the various FhuA deletion mutants in different salt solutions. The membranes were formed of diphytanoyl
PtdCho dissolved in n-decane. The aqueous solutions were used unbuffered and had a pH of  6 unless otherwise indicated. The applied voltage
was 20 mV, and the temperature was 20 °C. The average single-channel conductance, G, was calculated from at least 80 single events. The selectivity
of the different mutants in KCl was derived from zero-current membrane potential measurements. ND, not determined.
Salt Concentration c [
M
]
FhuA mutants
FhuAD5–160 FhuAD5–160 D322–336 FhuAD5–160 D335–355
Single-channel conductance G [nS]
KCl 0.1 0.2 0.4 0.3
0.3 0.3 0.9 0.7
1.0 0.5 1.3 2.5
3.0 0.5 1.5 6.5
LiCl 1.0 0.5 0.5 1.3
KAc (pH 7) 1.0 0.8 1 1.5
KCl + 0.1% SDS 1.0 0.75 ND ND

Selectivity P
K
/P
Cl
11 2.5 3.3
Fig. 3. Single-channel recording of a diphytanoyl PtdCho/n-decane
membrane in the presence of FhuAD5–160 D322–336. The aqueous
phase contained 1
M
KCl (pH 6) and  50 ng FhuAD5–160 D322–
336ÆmL
)1
. The applied membrane potential was 20 mV; T ¼ 20 °C.
Note that the time resolution of the current recording is higher than
that of Figs 2 and 4.
Ó FEBS 2002 FhuA transport protein of E. coli (Eur. J. Biochem. 269) 4953
concentration, which is expected for wide water-filled chan-
nels similar to those formed by Gram-negative bacterial
porins [32,33]. The single-channel conductance showed a
minor restriction as it increased 21-fold while the KCl
concentration was increased 30-fold. FhuAD5–160 D335–
355 showed a higher conductance in KAc as compared with
LiCl which suggested some preference for cations over
anions. For the other mutant proteins no clear dependence
on the aqueous salt concentrations was recorded which may
be caused by the rapid transition of the FhuA deletion
channels between different conductance states as Figs 2 and
3 indicate. Furthermore, with the latter mutant proteins no
clear selectivity for ions was observed.
Selectivity of the FhuA mutant proteins

Zero-current membrane potential measurements were per-
formed to further determine the selectivity of the FhuA
mutant proteins. After the incorporation of 100–1000
channels into the membranes, the KCl concentration on
one side of the membranes was raised from 0.1 to 0.5
M
by
the addition of concentrated KCl. The more dilute side of
themembrane(0.1
M
) became positive which indicated
preferential movement of potassium ions through the
mutant channels. These data demonstrate selectivity for
cations and support the data obtained from the single-
channel experiments (Table 2). The zero-current membrane
potentials for KCl were on average between 14 and 32 mV
at a fivefold KCl gradient across the membranes. Analysis
of the potential using the Goldman–Hodgkin–Katz equa-
tion [29] suggested that anions also move through the
channel because the ratio of the permeabilities P
K+
divided
by P
Cl-
was between 2.5 and 11 (Table 2). The ion selectivity
of the mutant proteins with higher single-channel conduct-
ance was lower than that of FhuAD5–160 with a smaller
conductance.
Active transport of ferrichrome by FhuA deletion
derivatives

Previously, we have shown that FhuAD5–160 exhibits all
TonB-dependent FhuA activities [19] except uptake of
microcin J25 [20]. Therefore, we tested whether FhuAD5–
160 retained ferrichrome transport activities when addi-
tional deletions were introduced. E. coli HK97 with a
chromosomal fhuA mutation was transformed with the
plasmids carrying fhuA deletion genes. The amounts of the
FhuA mutant proteins present in cells used to determine
the properties of the mutants are shown in Fig. 1A. Cells
containing FhuAD5–160 transported [
55
Fe
3+
]ferrichrome
with a rate that amounted to 59% of wild-type FhuA. This
comparison has to take into account that the cells
contained less FhuAD5–160 than wild-type FhuA. There
is no linear increase of the ferrichrome transport rates with
increasing concentrations of FhuA above a certain FhuA
concentration. The ferrichrome transport rate of FhuAD5–
160 D322–336 amounted to 45% and that of FhuAD322–
336 to 75% of the rate of wild-type FhuA (Fig. 6A and
Table 3). In contrast, FhuAD5–160 D335–355 did not
transport ferrichrome (Fig. 6A and Table 3). Since
FhuAD335–355 was also transport inactive, removal of
the globular domain did not convert FhuAD335–355 into
an active transporter.
Fig. 4. Single-channel recording of a diphytanoyl PtdCho/n-decane
membrane in the presence of FhuAD5–160 D335–355. The aqueous
phase contained 1

M
KCl (pH 6) and  50 ng FhuAD5–160 D335–
355ÆmL
)1
. The applied membrane potential was 20 mV, T ¼ 20 °C.
Fig. 5. Histogram of the probability P(G)oftheoccurrenceofagiven
conductivity unit observed with membranes formed of diphytanoyl
PtdCho/n-decane in the presence of FhuAD5–160 D335–355 mutant.
P(G) is the probability that a given conductance increment G is ob-
served in the single-channel experiments. It was calculated by dividing
the number of fluctuations similar to those of Fig. 3 with a given
conductance increment by the total number of conductance fluctua-
tions. The aqueous phase contained 1
M
KCl (pH 6) and  50 ng
FhuAD5–160 D335–355ÆmL
)1
. The applied membrane potential was
20 mV; T ¼ 20 °C. The average single-channel conductance was
2.5 nS for 94 single-channel events (right-hand maximum).
4954 M. Braun et al. (Eur. J. Biochem. 269) Ó FEBS 2002
Increase of outer membrane permeability
by the FhuA deletion derivatives
As FhuAD5–160 D335–355 formed , in lipid bilayer mem-
branes, stable open diffusion channels for ions such as K
+
and Cl
–
, we examined whether FhuAD5–160 D335–355 also
formed open channels in the outer membrane of E. coli

cells. To study diffusion of ferrichrome into the periplasm,
we used the fhuA tonB double mutant HK99 that was
devoid of active transport across the outer membrane but
actively transported ferrichrome via the FhuBCD proteins
from the periplasm across the cytoplasmic membrane into
the cytoplasm. [
55
Fe
3+
]Ferrichrome remains in the cyto-
plasm when cells are washed on filters to remove excess
radioactivity, whereas [
55
Fe
3+
]ferrichrome that has entered
only the periplasm is washed away. As HK99 is devoid of
FhuA activity, [
55
Fe
3+
]ferrichrome can only pass the outer
membrane via the FhuA deletion derivatives synthesized
after transformation of HK99 with plasmids carrying the
fhuA deletion genes. To measure diffusion we increased the
concentration of [
55
Fe
3+
]ferrichrome from 1 l

M
used in
the transport assay to 10 l
M
. Of the FhuA deletion
derivatives tested, only FhuAD5–160 D335–355 supported
uptake of [
55
Fe
3+
]ferrichrome via diffusion across the outer
membrane and subsequent active transport across the
cytoplasmic membrane (Fig. 6B). Addition of excess non-
radioactive ferrichrome (1.5 m
M
) after 21 min of transport
released a small portion (10%) of the [
55
Fe
3+
]ferrichrome
which was probably bound to periplasmic FhuD and some
may have been unspecifically bound to cells and the filters.
Ninety per cent of [
55
Fe
3+
]ferrichrome has been taken up
from the periplasm into the cytoplasm from where it was no
longer released during the chase.

Wild-type FhuA and the FhuA deletion mutants other
than FhuAD5–160 D335–355 did not support diffusion of
[
55
Fe
3+
]ferrichrome into the periplasm of the HK99 trans-
formants (Fig. 6B). [
55
Fe
3+
]Ferrichrome bound to HK99
synthesizing plasmid-encoded wild-type FhuA remained
constant during the 21 min incubation period. [
55
Fe
3+
]
Ferrichrome was released by the chase with nonradioactive
ferrichrome which is considered to be the amount of
[
55
Fe
3+
]ferrichrome that is bound to FhuA. HK99
Table 3. Ferrichrome binding and transport rates of wild-type FhuA and various FhuA deletion derivatives. The E. coli strains CH1857 fhuACDB tonB
and HK97 fhuA were transformed with the plasmids listed in Table 1 that encoded the FhuA proteins listed in the left panel.
FhuA proteins
Fc transport rates per min into HK97
a

(% wild-type)
Fc binding to CH1857 Iron ions/cell
b
(% wild-type)
FhuA wild-type 9960 (100%) 10198 (100%)
FhuAD5–160 5905 (59%) 3108 (30%)
FhuAD322–336 7456 (75%) 8972 (88%)
FhuAD5–160 D322–336 4492 (45%) 1171 (11%)
FhuAD335–355 0 2146 (21%)
FhuAD5–160 D335–355
c
––
a
[
55
Fe
3+
]Ferrichrome transport rates per minute were calculated from the linear region between 5 and 13 min of Fig. 6A. The rate was
related to the transport rate of wild-type FhuA (100%).
b
The mean values of [
55
Fe
3+
]ferrichrome (Fc) bound to the FhuA derivatives
minus the mean values after addition of 150 l
M
nonradioactive ferrichrome (chase) was taken as the fraction that is bound to FhuA. The
percentage is related to ferrichrome bound to wild-type FhuA (100%).
c

FhuAD5–160 D335–355 did not take up ferrichrome by active
transport but by TonB independent diffusion. The high concentration of iron ions measured during the binding assay (17597 iron ions per
cell) was due to the diffusion of ferrichrome into the periplasm and does not reflect the binding capacitiy of FhuAD5–160 D335–355 for
ferrichrome.
Fig. 6. (A) Time-dependent transport of [
55
Fe
3+
]ferrichrome (1 l
M
)
into E. coli HK97, (B) Time-dependent uptake of [
55
Fe
3+
]ferrichrome
(10 l
M
)intoEcoliHK99. After 21 min a 150-fold surplus of non-
radioactive ferrichrome was added as a chase. The E. coli strains
HK97 fhuA aroB and HK99 fhuA tonB aroB were transformed with
plasmids (listed in Table 1) encoding the FhuA deletion derivatives
indicated in the figure.
Ó FEBS 2002 FhuA transport protein of E. coli (Eur. J. Biochem. 269) 4955
FhuAD322–336 bound similar amounts of [
55
Fe
3+
]ferri-
chrome as HK99 wild-type FhuA. Cells of the other FhuA

deletion derivatives contained only very small amounts of
bound [
55
Fe
3+
]ferrichrome (Fig. 6B).
To support the conclusion of a diffusive entry of
[
55
Fe
3+
]ferrichrome across the outer membrane of HK99
FhuAD5–160 D335–355, we determined the concentration-
dependent uptake rate of [
55
Fe
3+
]ferrichrome and com-
pared it with the uptake into HK99 transformed with
plasmids that encoded wild-type FhuA and the other FhuA
deletion derivatives. Only HK99 FhuAD5–160 D335–355
showed a linear increase of uptake with increasing concen-
trations of [
55
Fe
3+
]ferrichrome from 1 to 6 l
M
(Fig. 7).
From 6 to 10 l

M
[
55
Fe
3+
]ferrichrome the uptake into HK99
FhuAD5–160 D335–355 was no longer linear, presumably
because transport across the cytoplasmic membrane became
rate limiting. The FhuA deletion mutants showed only a
comparatively small increase of [
55
Fe
3+
]ferrichrome associ-
ated with the cells which in the case of HK99 FhuAD322–
336 and FhuA wild-type reflected mostly binding to FhuA.
At higher concentrations [
55
Fe
3+
]ferrichrome diffused
somewhat through the other FhuA deletion derivatives into
the periplasm (Fig. 7).
Fig. 6 not only reveals transport of [
55
Fe
3+
]ferrichrome
but also binding of [
55

Fe
3+
]ferrichrome. For cells that
synthesized wild-type FhuA and FhuAD322–336 transport
into HK97 (Fig. 6A) correlated qualitatively with binding
to HK99 (Fig. 6B). However, although HK97 FhuAD5–
160andHK97FhuAD5–160 D322–336 transported
[
55
Fe
3+
]ferrichrome, binding of [
55
Fe
3+
]ferrichrome was
very low. Because during the transport assay cells on filters
were washed twice with 5 mL 0.1
M
LiCl, weakly bound
[
55
Fe
3+
]ferrichrome could have been released from cells and
washed through the filter. Therefore, binding was deter-
mined by a recently devised method in which cells are
centrifuged through silicone oil and collected above a layer
of a NaI solution. During this procedure cells remain viable
(H. Killmann and G. Gestwa, unpublished data). This

method was used for the determination of [
55
Fe
3+
]ferri-
chrome binding to FhuA and the FhuA deletion derivatives.
The used E. coli CH1857 is a tonB mutant deficient in
ferrichrome transport across the outer membrane and it
lacks the fhuABCD genes for transport across the cytoplas-
mic membrane. The only [
55
Fe
3+
]ferrichrome binding site
left is that of plasmid encoded FhuA and its derivatives. In
fact, cells that synthesized FhuAD5–160 bound [
55
Fe
3+
]
ferrichrome to 30% of the level of the wild-type FhuA.
Reduction of [
55
Fe
3+
]ferrichrome binding probably resulted
from the loss of four binding sites delivered by the cork
domain out of a total of 10 ferrichrome binding sites
in FhuA. This result demonstrated that centrifugation
through silicone oil was a much milder procedure than

washing of cells on filters, and indicates weak binding of
[
55
Fe
3+
]ferrichrome to FhuAD5–160. Cells synthesizing
FhuAD5–160 D322–336 bound 11% [
55
Fe
3+
]ferrichrome,
FhuAD322–336 88% and FhuAD335-335 showed 21%
binding compared to cells with wild-type FhuA (Table 3).
Binding to FhuAD5–160 D335–355 could not be determined
as ferrichrome diffused into the periplasm and remained
there during centrifugation through silicone oil (determined
value was 17 597 iron ions per cell).
Sensitivity of cells synthesizing FhuA deletion
derivatives to antibiotics
Another approach to identify water-filled protein channels
in the outer membrane is the determination of the sensitivity
of cells to antibiotics which are prevented from entering cells
by the permeability barrier of the outer membrane. Novo-
biocin, erythromycin, rifamycin and vancomycin are anti-
biotics which are too large to diffuse rapidly through the
pores formed by the porins (Table 4). Cells of HK97 wild-
type FhuA were resistant to the indicated antibiotics except
rifamycin (Table 4). All FhuA deletion derivatives increased
sensitivity to the antibiotics. The highest sensitivity to the
antibiotics was conferred by FhuAD335–355 and FhuAD5–

160 D335–355. Sensitivity mediated by FhuAD5–160 was
increased by the additional deletion 335–355, whereas the
sensitivity of FhuAD335–355 against SDS and Novobiocin
was increased only slightly when the cork domain had been
removed. Another indicator for outer membrane permeabi-
lity is sensitivity to SDS to which E. coli K-12 is resistant
as long as the outer membrane barrier is intact. Only
FhuAD335–355 and FhuAD5–160 D335–355 rendered cells
sensitive to SDS.
Sensitivity to the antibiotics was also determined in the
presence of 1 l
M
ferrichrome to test whether binding of
ferrichrome and the concomitant structural changes in
FhuA affected the permeability of the FhuA deletion
derivatives. We observed only slight effects if any (data not
shown) which might be caused by the low binding of
ferrichrome to the FhuA deletion derivatives. This conclu-
sion was supported by the decrease of the sensitivity of
Fig. 7. Concentration-dependent uptake of [
55
Fe
3+
]ferrichrome into
E. coli HK99 fhuA tonB ar oB expressing the plasmid-encoded FhuA
proteins indicated in the figure.
4956 M. Braun et al. (Eur. J. Biochem. 269) Ó FEBS 2002
HK97 FhuAD322–336 to rifamycin by ferrichrome which
binds to FhuAD322–336 (Table 3).
Growth of FhuA deletion mutants on maltodextrins

LamB is the maltoporin through which maltodextrins diffuse
across the outer membrane into the periplasm of E. coli [31].
If lamB is deleted maltodextrins larger than maltotriose
diffuse too slowly into the periplasm to support growth on
maltodextrins as the sole carbon source. We used E. coli
KB419 which is a lamB mutant with no polar effect on
transport genes required for maltose transport across the
cytoplasmic membrane. Among the FhuA mutants tested
E. coli KB419 FhuAD5–160 D335–355 formed the largest
growth zone on maltotetraose and could grow on malto-
pentaose (Table 4). In contrast, the growth zone of E. coli
KB419 FhuAD335–355 on maltotetraose was smaller and no
growth occurred on maltopentaose. The other FhuA dele-
tion derivatives did notsupport growth on maltotetraose and
maltopentaose except FhuAD5–160 (Table 4). Deletion
322–336 in FhuAD5–160 D322–336 did not increase but
somewhat reduced growth of KB419 on the maltodextrins,
as compared with strains expressing only FhuAD5–160.
FhuA activities of the FhuA deletion mutants
FhuAD5–160 confers to E. coli HK97 devoid of wild-type
FhuA sensitivity to the FhuA-specific phages T1, T5, /80,
to colicin M and albomycin to the same or similar degree as
wild-type FhuA. We examined sensitivities of HK97 that
synthesized the various FhuA deletion derivatives. HK97
FhuAD5–160 D335–355 was resistant to all FhuA ligands as
was HK97 FhuAD335–355. In contrast, HK97 FhuAD5–
160 D322–335 displayed sensitivity to T5, colicin M and
albomycin which, however, was lower than the sensitivity of
HK97 FhuAD322–335. The sensitivity to T5 was reduced
100-fold and the growth inhibition zones caused by colicin

M and albomycin were turbid indicating partial inhibition,
whereas those of HK97 FhuAD322–335 were clear (data
not shown).
DISCUSSION
The globular domain of FhuA tightly closes the channel
formed by the b-barrel and for this reason was designated
cork [5] or plug [6]. Although binding of ferrichrome causes
a large structural change in the crystal structure of FhuA it
does not open the channel. Excision of the entire globular
domain resulted in FhuAD5–160 that showed TonB-
dependent active ferrichrome transport. The b-barrel
domain alone functioned as an active transporter. With
higher concentrations of ferrichrome than are required for
transport, tonB mutant cells grew on ferrichrome as sole iron
source, indicating an open channel [19]. Apparently, ferri-
chrome diffused through FhuAD5–160 in contrast to wild-
type FhuA that failed to support growth of a tonB mutant.
FhuAD5–160 also somewhat increased sensitivity of cells to
antibiotics that are prevented from access to their target site
by the permeability barrier of the outer membrane [19].
Maltodextrins that pass through the outer membrane via the
LamB maltoporin entered the periplasm via FhuAD5–160 in
a lamB mutant. These results indicated that FhuAD5–160
increased unspecifically the permeability of the outer
Table 4. Growth inhibition of E. coli HK97 fhuA by SDS and antibtiotics and growth promotion of E. coli KB419 lamB by maltodextrins. Sensitivity of E. coli HK97 fhuA transformants and growth promotion of
E. coli KB419 lamB transformants expressing various plasmid-encoded FhuA deletion derivatives as indicated in the table. The molar masses and the absolute amounts of the inhibitors and the maltodextrins
added to the filter paper discs are indicated. The size of the zones of growth inhibition (in mm) of E. coli HK97 with SDS and the antibiotics is given with subtraction of the sensitivity of the control strain HK97
pT7-6 including the diameter of the filter paper disc (6 mm). The growth zones of E. coli KB419 with the maltodextrins is given in mm with subtraction of the filter paper disc (6 mm). No sensitivity or no growth
is indicated by a single line (–).
Growth inhibition of E. coli HK97 fhuA by Growth of E. coli KB419 lamB on

SDS
288 Da
750 lg
Novobiocin
634 Da
30 lg
Erythromycin
734 Da
15 lg
Rifamycin
823 Da
5 lg
Vancomycin
1485 Da
20 lg
Maltotriose
504 Da
4 mg
Maltotetraose
667 Da
4 mg
Maltopentaose
829 Da
4 mg
FhuA wild-type – – – 2 – 6 – –
FhuAD5–160 – 3 6 7 3 14 7 3
FhuAD322–336 – – – 4 – 6 – –
FhuAD5–160 D322–336 – 1 3 5 1 10 – –
FhuAD335–355 2 6 10 9 6 8 4 –
FhuAD5–160 D335–355 3 8 10 9 5 16 14 7

Ó FEBS 2002 FhuA transport protein of E. coli (Eur. J. Biochem. 269) 4957
membrane. However, cells synthesizing FhuAD5–160
remained resistant to SDS which suggested that the
FhuAD5–160 channel still restricted entry of molecules.
In this paper, a correlation between data obtained in vivo
with those obtained in vitro was attempted. Determination
of the single-channel conductance of isolated FhuAD5–160
in artificial lipid bilayer membranes revealed an increase in
conductance for KCl and other ions with rapidly changing
amplitudes and no permanently open stable channels. This
finding was unexpected as the crystal structure of FhuA
suggests that removal of the cork domain should result in an
open channel. The conductance showed rapid transitions
between different states probably through rapid opening
and closing of the FhuAD5–160 channel or, less likely, rapid
membrane entry and exit of the protein. The instability of
the channels formed by FhuAD5–160 may have several
causes. It contains surface loop 4 above the outer cavity of
FhuA through which ferrichrome gains access to the
ferrichrome binding site. Loop 4 constricts the channel
entrance to about half the area of the total cross-section [6].
Removal of the globular domain may increase the flexibility
of loop 4 and other surface exposed loops so that rapidly
moving loops may cause the frequent transition of open and
closed states of FhuAD5–160. In addition, among the more
than 60 residues that are exposed to the interior of the
b-barrel channel and fix the globular domain, some amino
acid side chains may become flexible when the globular
domain is removed, and depending on their orientation may
modulate the movement of ions through the channel.

Furthermore, the b-barrel of FhuAD5–160 may be less rigid
than is suggested by the crystallographic B-factor (52 A
˚
2
)of
complete FhuA [6]. Removal of the globular domain may
increase the flexibility of the b-strands resulting in b-barrels
with changing diameters of the elliptical cross shape.
Although FhuAD5–160 D322–336 lacked the cork
domain and most of the L4 loop, it formed no stable
channels in lipid bilayer membranes, and showed no or only
a marginal concentration-dependent increase in ferrichrome
diffusion across the outer membrane, but increased some-
what the outer membrane permeability for antibiotics and
maltodextrins. Obviously, removal of the L4 loop did not
subtantially increase diffusion through the b-barrel.
FhuAD322–336 exhibited 75% of the [
55
Fe
3+
]ferrichrome
transport activity of wild-type FhuA. The transport activity
was reduced further when the cork domain was deleted also,
probably because four ferrichrome binding sites were
removed with the cork domain which resulted in only
11% binding related to wild-type FhuA. FhuAD5–160
D322–336 still functioned as an active transporter and
displayed all the other FhuA related functions.
Deletion 335–355 was within the b-barrel and caused
inactivation of FhuAD335–355 and probably also of

FhuAD5–160 D335–355 as transporters and as receptors
for the phages and colicin M. Deletion 335–355 distorted
the b-barrel such that the FhuA deletion derivative was not
only unable to display TonB-dependent activities but also
the TonB-independent infection by phage T5. However, it
formed channels through which ferrichrome diffused into
the periplasm with a rate that increased linearly with the
ferrichrome concentration. Maltodextrins served as carbon
sources for cells synthesizing FhuAD5–160 D335–355 and
the deletion derivative was sensitive to SDS. It is likely that
these hydrophilic/amphipatic compounds diffused through
FhuAD5–160 D335–355 into the periplasm. We favour the
same interpretation for the increase in sensitivity of the
antibiotics although they are more hydrophobic and may
enter cells through diffusion through the outer membrane if
incorporation of FhuAD5–160 D335–355 disturbed outer
membrane integrity. The results obtained in vivo correlated
with the results obtained in vitro. The recordings of the
single-channel conductance of FhuAD5–160 D335–355
revealed stable single channels of rather uniform size with
high single-channel conductance. Each step probably cor-
responded to the incorporation of one channel-forming unit
into the lipid bilayer.
Our conclusion that the b-barrel of FhuA largely
determines the transport properties including the response
to TonB was confirmed by a similar study on FepA and its
comparison with FhuAD5–160 [21]. FepA is the outer
membrane transport protein for Fe
3+
enterobactin. The

crystal structure of FepA is similar to that of FhuA [34].
Excision of the globular domain (residues 17–150) resulted
in a protein that exhibited Fe
3+
enterobactin transport that
depended on TonB. However, the V
max
of the transport rate
(0.3% of wild-type FepA) was much lower than the V
max
of
FhuAD5–160 (32% of wild-type FhuA) [21]. Despite the
low transport rate of FepAD17–150 it supports the conclu-
sion that the b-barrel alone functions as an active trans-
porter. In wild-type FhuA the cork must be dislocated so
that the channel of the b-barrel opens which may require the
concerted action of TonB on the b-barrel and the cork.
ACKNOWLEDGEMENTS
We thank K.A. Brune for critical reading of the manuscript. This work
was supported by the Deutsche Forschungsgemeinschaft (BR330/20–1,
Forschergruppe Bakterielle Zellhu
¨
lle: Synthese, Funktion und Wirkort)
and the Fonds der Chemischen Industrie.
REFERENCES
1. Nikaido, H. (1992) Porins and specific channels of bacterial outer
membranes. Mol. Microbiol. 6, 435–442.
2. Drechsel, H. & Winkelmann, G. (1997) Iron chelation and side-
rophores in transition metals in microbial metabolism. In Tran-
sition Metals in Microbial Metabolism (Winkelmann, G. &

Carrano, C.J., eds), pp. 1–49. Harwood Academic Publishers,
Amsterdam.
3. Bradbeer, C. (1993) The proton motive force drives the outer
membrane transport of cobalamin in Escherichia coli. J. Bacteriol.
175, 3146–3150.
4. Braun, V., Hantke, K. & Ko
¨
ster, W. (1998) Bacterial iron trans-
port: mechanisms, genetics, and regulation. In Metal Ions in Bio-
logical Systems, 35. Iron Transport and Storage in Microorganisms.
(Sigel, A. & Sigel, H., eds), pp. 67–145. Marcel Dekker, New York.
5. Ferguson, A.D., Hofmann, E., Coulton, J.W., Diederichs, K. &
Welte, W. (1998) Siderophore-mediated iron transport: crystal
structure of FhuA with bound lipopolysaccharide. Science 282,
2215–2220.
6. Locher, K.P., Rees, B., Koebnik, R., Mitschler, A., Moulinier, L.,
Rosenbusch, J.P. & Moras, D. (1998) Transmembrane signaling
across the ligand-gated FhuA receptor: crystal structures of free
and ferrichrome-bound states reveal allosteric changes. Cell 95,
771–778.
7. Postle, K. (1993) TonB protein and energy transduction between
membranes. J. Bioenerg. Biomembr. 25, 591–601.
8. Kampfenkel, K. & Braun, V. (1992) Membrane topology of the
Escherichia coli ExbD protein. J. Bacteriol. 174, 5485–5487.
4958 M. Braun et al. (Eur. J. Biochem. 269) Ó FEBS 2002
9. Kampfenkel, K. & Braun, V. (1993) Topology of the ExbB protein
in the cytoplasmic membrane of Escherichia coli. J. Biol. Chem.
268, 6050–6057.
10. Moeck, G.S., Coulton, J.W. & Postle, K. (1997) Cell envelope
signaling in Escherichia coli. Ligand binding to the ferrichrome-

iron receptor FhuA promotes interaction with the energy-trans-
duction protein TonB. J. Biol. Chem. 272, 28391–28397.
11. Scho
¨
ffler, H. & Braun, V. (1989) Transport across the outer
membrane of Escherichia coli K12 via the FhuA receptor is
regulated by the TonB protein of the cytoplasmic membrane. Mol.
Gen. Genet. 217, 378–383.
12. Gu
¨
nter, K. & Braun, V. (1990) In vivo evidence for FhuA outer
membrane interaction with the TonB inner membrane protein of
Escherichia coli. FEBS Lett. 274, 85–88.
13. Cadieux, N. & Kadner, R.J. (1999) Site-directed disulfide bonding
reveals an interaction site between energy-coupling protein TonB
and BtuB, the outer membrane cobalamin transporter. Proc. Natl
Acad.Sci.USA96, 10673–10678.
14. Merianos,H.J.,Cadieux,N.,Lin,C.H.,Kadner,R.J.&Cafiso,
D.S. (2000) Substrate-induced exposure of an energy-coupling
motif of a membrane transporter. Nature Struct. Biol. 7, 205–209.
15. Koebnik, R. & Braun, V. (1993) Insertion derivatives containing
segments of up to 16 amino acids identify surface- and periplasm-
exposed regions of the FhuA outer membrane receptor of
Escherichia coli K-12. J. Bacteriol. 175, 826–839.
16. Killmann, H., Videnov, G., Jung, G., Schwarz, H. & Braun, V.
(1995) Identification of receptor binding sites by competitive
peptide mapping: phages T1, T5, and /80 and colicin M bind to
the gating loop of FhuA. J. Bacteriol. 177, 694–698.
17. Killmann, H., Benz, R. & Braun, V. (1993) Conversion of the
FhuA transport protein into a diffusion channel through the outer

membrane of Escherichia coli. EMBO J. 12, 3007–3016.
18. Killmann, H., Benz, R. & Braun, V. (1996) Properties of the FhuA
channel in the Escherichia coli outer membrane after deletion of
FhuA portions within and outside the predicted gating loop.
J. Bacteriol. 178, 6913–6920.
19. Braun, M., Killmann, H. & Braun, V. (1999) The b-barrel domain
of FhuAD5–160 is sufficient for TonB-dependent FhuA activities
of Escherichia coli. Mol. Microbiol. 33, 1037–1049.
20. Killmann, H., Braun, M., Herrmann, C. & Braun, V. (2001) FhuA
barrel-cork hybrids are active transporters and receptors. J. Bac-
teriol. 183, 3476–3487.
21. Scott,D.C.,Cao,Z.,Qi,Z.,Bauler,M.,Igo,J.D.,Newton,S.M.
& Klebba, P.E. (2001) Exchangeability of N termini in the ligand-
gated porins of Escherichia coli. J. Biol. Chem. 276, 13025–13033.
22. Sambrook, J.E.F. & Fritsch, E.F. & Maniatis, T. (1989) Molecular
Cloning: a Laboratory Manual, 2nd edn. Cold Spring Harbor
Laboratory Press, Cold Spring Harbor, New York.
23. Killmann, H. & Braun, V. (1992) An aspartate deletion mutation
defines a binding site of the multifunctional FhuA outer mem-
brane receptor of Escherichia coli K-12. J. Bacteriol. 174, 3479–
3486.
24. O
¨
lschla
¨
ger, T., Schramm, E. & Braun, V. (1984) Cloning and
expression of the activity and immunity genes of colicins B and M
on ColBM plasmids. Mol. Gen. Genet. 196, 482–487.
25. Solbiati, J.O., Ciaccio, M., Farias, R.N. & Salomon, R.A. (1996)
Genetic analysis of plasmid determinants for microcin J25 pro-

duction and immunity. J. Bacteriol. 178, 3661–3663.
26. Miller, J.H. (1972) Experiments in Molecular Genetics. Cold Spring
Harbor Laboratory Press, Cold Spring Harbor, New York.
27. Benz,R.,Janko,K.,Boos,W.&La
¨
uger, P. (1978) Formation of
large, ion-permeable membrane channels by the matrix protein
(porin) of Escherichia coli. Biochim. Biophys. Acta 511, 305–319.
28. Benz, R., Schmid, A. & Hancock, R.E.W. (1985) Ion selectivity of
gram-negative bacterial porins. J. Bacteriol. 162, 722–727.
29. Benz,R.,Janko,K.&La
¨
uger, P. (1979) Ionic selectivity of pores
formed by the matrix protein (porin) of Escherichia coli. Biochim.
Biophys. Acta 551, 238–247.
30. Ferguson, A.D., Breed, J., Diederichs, K., Welte, W. & Coulton,
J.W. (1998) An internal affinity-tag for purification and crystal-
lization of the siderophore receptor FhuA, integral outer mem-
brane protein from Escherichia coli K-12. Protein Sci. 7, 1636–
1638.
31. Benz, R. (1994) Solute uptake through bacterial outer membranes.
In Bacterial Cell Wall (Ghuysen, J.M. & Hakenbeck, R., eds),
pp. 397–423. Elsevier Science, Amsterdam.
32. Benz, R. (1988) Structure and function of porins from gram-
negative bacteria. Annu. Rev. Microbiol. 42, 359–393.
33.Weiss,M.S.,Kreusch,A.,Schiltz,E.,Nestel,U.,Welte,W.,
Weckesser, J. & Schulz, G.E. (1991) The structure of porin from
Rhodobacter capsulatus at 1.8 A
˚
resolution. FEBS Lett. 280,

379–382.
34. Hantke, K. & Braun, V. (1978) Functional interaction of the tonA/
tonB receptor system in Escherichia coli. J. Bacteriol. 135, 190–197.
35. Prilipov, A., Phale, P.S., Van Gelder, P., Rosenbusch, J.P. &
Koebnik, R. (1998) Coupling site-directed mutagenesis with high-
level expression: large scale production of mutant porins from
E. coli. FEMS Microbiol. Lett. 163, 65–72.
36. Killmann, H., Herrmann, C., Wolff, H. & Braun, V. (1998)
Identification of a new site for ferrichrome transport by compar-
ison of the FhuA proteins of Escherichia coli, Salmonella para-
typhi, Salmonella typhimurium, and Pantoea agglomerans.
J. Bacteriol. 180, 3845–3852.
37. Tabor, S. & Richardson, C.C. (1985) A bacteriophage T7 RNA
polymerase/promoter system for controlled exclusive expression
of specific genes. Proc.NatlAcad.Sci.USA82, 1074–1078.
Ó FEBS 2002 FhuA transport protein of E. coli (Eur. J. Biochem. 269) 4959