exchangeability of n-termini in the ligand-gated porins of
TRANSCRIPT
Exchangeability of N-Termini in the Ligand-Gated Porins of
Escherichia coli
Daniel C. Scott, Zhenghua Cao, Zengbiao Qi, Matthew Bauler, John D. Igo,
Salete M. C. Newton & P. E. Klebba*
Department of Chemistry & Biochemistry
University of Oklahoma
Norman, OK 73019
Running title: Exchange of LGP N-termini
* Corresponding author: [email protected]
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
2
Summary
The ferric siderophore transporters of the Gram-negative bacterial outer membrane manifest an unique
architecture: their N-termini fold into a globular domain that lodges within, and physically obstructs, a
transmembrane porin $-barrel formed by their C-termini. We exchanged and deleted the N-termini of
two such siderophore receptors, FepA and FhuA, which recognize and transport ferric enterobactin and
ferrichrome, respectively. The resultant chimeric proteins and empty $-barrels avidly bound appropriate
ligands, including iron complexes, protein toxins, and viruses. Thus, the ability to recognize and
discriminate these molecules fully originates in the transmembrane $-barrel domain. Both the hybrid
and the deletion proteins also transported the ferric siderophore that they bound. The FepA constructs
showed less transport activity than wild type receptor protein, but the FhuA constructs functioned with
turnover numbers that were equivalent to wild type. The mutant proteins displayed the full range of
transport functionalities, in spite of their aberrant or missing N-termini, confirming (Braun et al., Mol.
Microbiol. 33:1037-1049, 1999) that the globular domain within the pore is dispensable to the
siderophore internalization reaction, and when present, acts without specificity during solute uptake.
These and other data suggest a transport process in which siderophore receptors undergo multiple
conformational states that ultimately expel the N-terminus from the channel concomitant with solute
internalization.
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
3
Introduction
Gram-negative bacteria recognize and transport ferric siderophores (1,2) through proteins in their outer
membrane (OMa). They secrete these chelators in the iron-deficient environments that they encounter in
the wild and in the host. Because of its involvement in cellular processes, iron is essential to survival, and
bacteria possess efficient systems to obtain it (3-6). However, their competitors in the microbial world
parasitize these iron uptake pathways. Bacteriocins and phages bind to siderophore receptors as an initial
step in the penetration of the cell wall (7-9). Other microbes are not the only antagonists who target the
iron portals of bacteria: synthetic antibiotics that couple antibacterial agents to organic iron complexes
also enter through siderophore receptors (10-13). Finally, iron acquisition is crucial to the pathogenesis of
Gram-negative bacteria, including Salmonella, Neisseria, Yersinia, Vibrio and Haemophilus (14-20). In
response, higher organisms defend themselves by the sequestration of iron in proteins like ferritin,
transferrin, lactoferrin and conalbumin (21,22). The competition for iron in vivo is so fierce that one
pathogenic species, Borrelia, evolved completely novel biochemical systems that do not require the metal
(23).
The OM proteins that initiate the uptake of iron, antibiotics, colicins and phage function as ligand-
gated porins {LGP; (24-26)}. They contain a transmembrane channel formed by amphiphilic $-strands
that project to the cell surface as large loops (27-29). In this sense the proteins are fundamentally porins
in nature (30-32). However, LGP differ from other porins because they bind the molecules they transport
with high affinity, and because they contain a globular N-terminus that resides within their transmembrane
channel (Fig. 1). When ligands adsorb to siderophore receptors, they trigger unknown events that induce
their transport into the cell (33-35). In this sense the receptors are ligand-gated. Finally, LGP transport
a Abbreviations: OM, outer membrane; LGP, ligand-gated porin; FeEnt, ferric enterobactin; Fc, ferrichrome; MAb, monoclonal antibody; SDS, sodium dodecyl sulfate; LDS, lithium dodecyl sulfate; MOPS, morphilino propane sulfonic acid; PCR,
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
4
requires proton motive force (36-38) and the participation of another cell envelope protein, TonB (39-41),
that presumably promotes the opening of the closed channels so ligands may enter. The crystal
structures of the ferric enterobactin receptor, FepA, (27) and the ferrichrome receptor, FhuA, (28,29)
defined the architecture of siderophore transporters, but did not answer several pressing questions about
their biochemical mechanisms.
Their ligand recognition and transport processes, for example, are only vaguely defined. The OM
receptors of Escherichia coli discriminate numerous ferric siderophore complexes, that in spite of their
similar chelation geometry (hexacoordinate) and size (600 - 1000 Da), are distinct in composition and
charge (1,2,42). Each ferric siderophore enters through a different OM iron transporter. Ferric
enterobactin (FeEnt), the native E. coli siderophore, penetrates through FepA (43), while ferrichrome
(Fc), a fungal product that E. coli also utilizes, passes through FhuA (44). Does LGP specificity derive
from residues in their surface loops, or from amino acids in the N-domain, at the entrance to the
membrane channel (Fig. 1)? Experiments on a FhuA mutant devoid of its N-terminal domain (45), were
relevant to this issue, as well as that of ligand transport. In those studies the N-domain of FhuA was not
necessary for Fc, colicin and bacteriophage uptake. The experiments we report herein duplicated and
confirmed those findings on FhuA, and characterized comparable constructions from FepA, that gave
almost identical results. Furthermore, when we genetically exchanged the N-termini of FepA and FhuA,
both hybrid proteins still bound and transported their appropriate ligands.
Experimental Procedures
Bacterial Strains, Plasmids and media. Bacteria harboring plasmids (Table 1) carrying the genes of
interest were cultured in LB broth or MOPS minimal media (46). It was impractical to create single-copy
polymerase chain reaction; TBS, tris-buffered saline; SulfoEGS, ethylene glycobis(sulfosuccimidylsuccinate).
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
5
chromosomal derivatives of the numerous constructs we generated. Instead, we individually transferred
all of the genes, including wild type FepA and FhuA, to the low-copy plasmid pHSG575, which exists in
E. coli at a level of 2-3 copies/cell (47).
Genetic engineering. We used PCR to join the FepA N-terminus (residues 1-152) to the FhuA C-
terminus (residues 160-723), and the FhuA N-terminus (residues 1-155) to the FepA C-terminus (residues
149-724). We first cloned the N and C-terminal domains of FepA and FhuA, and then ligated them
together. For instance, for FepNFhu$, we PCR-amplified FepN, incorporating PstI and BamHI sites in
the forward and reverse primers, respectively, and inserted the product into the low copy plasmid
pHSG575 (47). We then PCR amplified Fhu$, flanked by BamHI and SacI restriction sites, and joined it
to FepN in pHSG575. Additionally, we employed oligonucleotide-directed mutagenesis (Quik Change,
Strategize Corp., San Diego) to delete residues 17-150 of FepA (creating Fep$), residues 3-150 of FepA
(Fep$2), and residues 5-160 of FhuA (Fhu$). Except when noted, fepA, fhuA, and their derivatives were
all analyzed on pHSG575, which was used for binding, nutrition and transport experiments.
Binding and transport. The adsorption of 59FeEnt and 59FeFc was measured with metabolically
inactive KDF541 (46) expressing FepA or FhuA, respectively. 59FeEnt and 59FeFc were prepared at a
specific activity of approximately 200 cpm/pM and chromatographically purified. Binding manipulations
were performed at 0o C. A mid-log bacterial culture was chilled on ice for 1 hour and an aliquot
(containing approximately 5 x 107 cells), was pipetted into a 50 ml test tube and incubated on ice. A 25
ml volume of ice cold MOPS minimal media, containing varying concentrations of 59Fe-siderophore, was
poured into the tube to achieve rapid and thorough mixing. After 1 minute the binding reactions, which
were performed in triplicate, were filtered through 0.45 : nitrocellulose, the filters were were washed
with 10 ml of 0.9% LiCl and counted in a Packard Cobra gamma counter. The initial adsorption reaction
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
6
reached equilibrium within 5 s at physiological temperatures, and within 1 minute on ice. The FepA-
deficient strain KDF541 was simultaneously tested as a negative control, and any non-specific adsorption
of 59FeEnt by this strain was subtracted from the experimental samples. Whenever necessary because of
low cpm bound, the assay samples were counted for an extended period of time (up to 30 minutes) to
decrease standard error. Binding data were analyzed using the Bound vs Total equation of Grafit 4.013
(Erithacus Ltd., Middlesex, UK).
Ferric siderophore uptake was qualitatively evaluated by nutrition tests (44), and the transport of
59FeEnt and 59FeFc was quantitatively measured in live bacteria (46). For ferric siderophore acquisition,
{and general and specific porin-mediated transport as well (48)} the outer membrane transport stage
constitutes the rate-limiting step of the overall uptake process (4,6,24,26,46,49-51). The periplasmic
binding proteins (51) and the components of the inner membrane permease complexes are either
constitutively expressed, or de-repressed by iron starvation. Therefore, our transport data describes the
outer membrane components of the process, FepA, FhuA, and their mutant derivatives.
Transport manipulations were performed at 37o C. A volume of mid-log bacterial culture (50 - 100
:l containing approximately 5 x 107 cells) was pipetted into a 50 ml test tube and incubated in a 37o C
water bath. Without delay, 25 ml of pre-warmed MOPS minimal media, containing glucose (0.2%),
appropriate nutritional supplements and varying concentrations of 59FeEnt, was poured into the tube to
achieve rapid and thorough mixing. The transport reactions were quenched by the addition of a 1,000-fold
excess of non-radioactive FeEnt, immediately filtered through 0.45 : nitrocellulose, the filters were
washed with 10 ml of 0.9% LiCl and counted in a Packard Cobra gamma counter. Kinetic parameters
were determined from the initial rates of FeEnt uptake, which were calculated at each substrate
concentration from two independent measurements made in triplicate at 5 s and 15 s: cpm bound to the
cells at 5s were subtracted from the cpm associated with the cells at 15s (10 s uptakes). For some mutants
with low uptake rates (FepNFhuβ, FhuNFepβ, Fepβ) we extended the uptake period to 60 minutes (6). In
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
7
all transport experiments the FepA-deficient strain KDF541 was simultaneously tested as a negative
control, and any non-specific adsorption of 59FeEnt by this strain was subtracted from the experimental
samples. Transport results were analyzed according to the Michaelis-Menten equation, using Grafit 4.013.
Bacteriocin and bacteriophage susceptibility. Serial dilutions of colicins B, D or M, and bacteriophage
T5 and N80 were prepared in LB broth in microtiter plates, and 5 ul volumes of the dilutions were
transferred to LB plates seeded with the strain of interest, using a Clonemaster (Immusine Corp., San
Leandro, Ca). The titer of the phage and colicins was expressed as the reciprocal of the highest dilution
that cleared the bacterial lawn.
SDS-PAGE and western immunoblots. Proteins were separated on SDS-PAGE gels (52), and either
stained with coomassie blue or electrophoretically transferred to nitrocellulose paper. Western
immunoblots were incubated with anti-FepA MAbs 26 or 45 (52), developed with 125I protein A (49),
quantitated by image analysis with a Packard Instant Imager, and visualized by exposure of X-ray films.
The former antibody recognizes an epitope in the N-terminal globular domain of FepA {bounded by
residues 27and 37 (52)}; the latter binds in loop 4 of the C-terminal barrel domain, near residue 329
(49,52). For native gel electrophoresis, 50 ug of OM or 2 ug of purified protein was resuspended in LDS
sample buffer and sonicated in an ice water bath for five minutes. After a 1 min centrifugation in a
microfuge, the samples were electrophoresed overnight at 5 mA and 4 oC on LDS-PAGE gels (53).
Mutant protein expression and localization. The expression of FepA and its mutants was quantitated
by western immunoblots of whole cell lysates with anti-FepA monoclonal antibodies 26 and 45 and 125I-
protein A. Protein concentrations were determined by image analysis, relative to standards. The
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
8
localization of the chimeric proteins was determined by fractionation of inner and outer membranes on
sucrose gradients (54), followed by western immunoblots of the fractions.
Enzyme-linked immunosorbent assays (ELISA). Purified OM fragments or purified, denatured FepA,
were suspended in 10 mM ammonium acetate, 10 mM ammonium carbonate, pH 8.3, at 10 ug/ml and 0.5
ug/ml, respectively, dispensed into microtiter plates (Immulon), and incubated at 4 oC overnight. All
further incubation steps were performed at 25 oC. In the morning, excess or unabsorbed antigen was
removed by 3 washes with Tris-buffered saline containing 0.05% Tween 20, pH 7.4 (TBS-tween), and
the plates were blocked with 2% BSA in TBS for 30 min. The plates were washed 3 times with TBS-
tween, and serial dilutions of anti-FepA MAbs in 50 µl of blocking buffer were added and incubated for
1 hour. The blocking buffer was removed and 50 µl of a 1/100 dilution of goat-anti-mouse
immunoglobulin-alkaline phosphatase (Sigma) in blocking buffer was added to the plates. After
incubation for 1 hour at 25 oC, the plates were washed 3 times and developed with 50 ul of p-nitrophenyl
phosphate (1mg/ml; Sigma). After a 1 hour incubation at 25 oC, 50 µl of 2N NaOH was added to stop the
reaction and absorbance was measured at 405 nm with a microplate reader.
OM protein crosslinking. Bacteria grown in MOPS minimal media, or sucrose gradient-purified OM
fractions, were suspended at 109 cells/ml or 10 mg/ml, respectively, in 4 mM SulfoEGS, a water soluble
homobifunctional crosslinker (mol. wt. 661 D) that preferentially reacts with primary amines, for 2 hours
at 0 oC. The compound contains a 16Å spacer arm, that is cleaved by reaction with hydroxylamine. After
crosslinking, cells or OM proteins were solubilized in sample buffer, subjected to SDS-PAGE, and stained
with coomassie blue. When indicated, 5 uM ferric enterobactin was added to the cells prior to
crosslinking. Crosslinked bands were excised from the gels, cleaved with hydroxylamine, electroeluted
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
9
and re-electrophoresed: the identity of the crosslinked proteins was determined by sequence analysis of
their N-terminal 15 residues (Protein and Nucleic Acid Sequence Facility, Medical College of Wisconsin).
Results
Genetic exchange of LGP N-termini. To understand the basis of ligand specificity in FepA and FhuA
we switched their N-terminal domains and biochemically characterized the resulting hybrid proteins.
Although Braun et al. (45) reported the ability of an N-terminal deletion of FhuA to transport, they did
not consider the affinity of the mutant for its ligands, nor the kinetic parameters of the uptake process.
Our experiments duplicated their construction, created comparable deletion mutants of FepA, and created
the two chimeric proteins that switched the N-domains of FepA and FhuA. In the latter case, the protein
engineering conserved three structural features of FepA and FhuA: the globular N-domain, the $-barrel
of the C-domain, and the $-turn that joins them (Fig. 1). The resulting clones avoided the deletion or
introduction of amino acids in the junction sequence: FepNFhu$ contained the N-terminus and $-turn of
FepA connected to the $-barrel of FhuA; FhuNFep$ contained the N-terminus of FhuA linked to the $-
turn and $-barrel of FepA.
The hybrid proteins functioned with unexpected efficiency and selectivity, conferred by the $-
barrels and their loops. The N-terminal domains did not affect the discrimination of ferric siderophores.
FhuNFep$ bound FeEnt but not ferrichrome, whereas FepNFhu$ bound ferrichrome but not FeEnt. FepA
adsorbs FeEnt with sub-nanomolar affinity, that perservered in the FhuNFep$ chimera: the Kd of its
binding reaction with FeEnt was 0.2 nM (Fig 2, Table 2). The reverse hybrid (FepNFhu$) bound
ferrichrome with equivalent affinity to that of wild type FhuA: the Kd its binding reaction with Fc was
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
10
0.6 nM (Fig 2, Table 2). So, in spite of their heterologous N-termini, the chimeras maintained the
selectivity of their $-barrels, with the affinity of wild type receptors.
The constructs that deleted the N-domains of FepA and FhuA (Fep$ and Fhuβ) also bound ferric
siderophores with wild type affinities (Fig. 2, Table 2). Our Fhu$ construct was identical to that of Braun
et al (45). Regarding the β-barrel-only clones of FepA, we engineered two different derivatives. In
fepA∆17-151 (Fep$) we removed most of the the N-domain, but left amino acids 1-16, which include the
TonB-box region, fused to G151. Besides Fep$ we engineered fepA)3-151 (Fep$2), which eliminated
the TonB box. Except for a slight difference in expression (Fep$ was better), the two clones functioned
identically in binding, nutrition and transport assays; we only report data for Fep$. The empty $-barrels
of FhuA and FepA avidly bound their appropriate ferric siderophore, indicating that the surface loops
select the iron complexes that the bacteria encounter in their environments: the loops identify correct iron
chelates, and reject improper ones. Although they were irrelevant to ferric siderophore recognition, the
N-terminal domains of FepA and FhuA were needed for optimum binding. Bacteria expressing
constructions that exchanged or deleted them adsorbed ferric siderophores with much lower capacities
(Fig. 2, Table 2). This reduction did not result from poor expression (Table 2), improper localization or
instability (Fig. 4): rather, biochemical characterizations indicated that only a fraction (~10%) of the
mutant proteins bound ferric siderophores. Thus, a correct N-domain, which was not needed for
discrimination of the metal chelates, was essential for their maximal binding. Without their homologous
N-domains, mutants of FepA and FhuA adsorbed much lower amounts of ferric siderophores to the cells.
Interactions with colicins and bacteriophage. The $-barrels of the mutant proteins also dictated
interactions with bacteriocins and bacteriophages. The FepA $-barrel conferred susceptibility to colicins
B and D, but not M or the phages that use FhuA, again demonstrating the irrelevance of the N-domain to
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
11
ligand selection. Likewise, the FhuA barrel recognized T5 and N80, and the transposition of the FepA N-
terminus into it did not change this specificity (Table 2). So, the antagonism between iron chelates and
noxious agents for adsorption to FepA and FhuA derives from competition for structural elements in their
$-barrels and loops, in accord with the prior localization of residues in FepA that are needed for efficient
colicin reception. Yet, the exchange of N-termini reduced the efficiency of colicin B and D killing,
suggesting that these bacteriocins further interact with the globular domain during the implementation of
their toxicity. Unlike Braun et al. (45), we could not detect the ability of Fhuβ to confer susceptibility to
colicin M. This discrepancy may stem from the low activity of our colicin M preparation.
The N-domain in transport. The occlusion of the FepA and FhuA channels by their own N-termini
raises another question: what is the function and structural disposition of the N-domain during transport?
Passage of ferric siderophores requires a minimum diameter of about 15 Å, which does not exist in either
FepA or FhuA. Does the N-terminal domain exit the $-barrel during metal uptake, or change shape
within the pore to create a sufficient opening for ligand transit? The former mechanism faces an
energetic barrier, because the crystal structures of FepA and FhuA describe many residues in appropriate
positions for hydrogen bonds between the N-domains and barrel walls. The latter notion, that the
conformation of the N-terminus temporarily changes to create a pore, appears more plausible, but transient
channels of such magnitude have not been observed in membrane proteins. Furthermore, although
colicins generally comprise a slender, elongated shape, their entry into or passage through even an open
porin channel faces severe steric obstacles (55).
Uptake experiments with the mutant proteins, especially those containing the FhuA barrel, bore
relevance to these questions. In spite of its binding ability, the presence of the FepA N-terminus within it
made it unlikely for FepNFhu$ to transport Fc. Yet the chimera efficiently internalized the hydroxamate
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
12
siderophore. Fc stimulated the growth of bacteria expressing FepNFhu$ as much as strains producing
FhuA (Fig. 3), in a TonB-dependent manner. Quantitative determinations explained this result: despite a
slightly lower overall uptake affinity for Fc than FhuA (Km values of 2.5 nM and 0.6 nM, respectively;
Fig. 2, Table 2), the chimeric protein displayed a comparable turnover number to the wild type protein (k3
values of 6.5/min and 4/min, respectively; Table 2). Furthermore, the isolated FhuA $-barrel transported
Fc with similar affinity (Km = 3.6 nM), and at a faster rate (k3 = 8.3/min). Thus neither was an N-domain
required for Fc uptake, nor did the introduction of the FepA N-terminus into the FhuA barrel impair the
internalization reaction: the 160 amino acids of the globular domain were superfluous to Fc transport.
Mutant proteins containing the FepA $-barrel were less efficient, but they corroborated the dispensability
of the N-terminus to transport: both FhuNFep$ and Fep$ conferred TonB-dependent FeEnt uptake, albeit
slower than wild type FepA (k3 values of 0.9/min, 0.3/min and 6.4/min , respectively; Fig. 3, Table 2).
The low transport rates of FhuNFep$ and Fep$ explained their marginal abilities in nutrition tests (Fig.
3a). All of the plasmids failed to support growth in KDF571 {(24) tonB; Fig. 3}. We also constructed
and analyzed a second, similar chimera, FhuNFep$2, containing the N-terminus and turn of FhuA linked
to the barrel of FepA, whose biochemical properties were equivalent to those of FhuNFep$ (data not
shown). Accuracy was not compromised in the binding and transport measurements of strains with low
capacity (FepNFhuβ, FhuNFepβ, Fepβ and Fhuβ). The 59Fe binding and uptake assays were sufficiently
sensitive to quantitate even slight binding relative to the negative controls, which are absolute (46,49).
The close conformity of the data to non-linear fit calculations from the Bound vs Total and Michaelis-
Menten equations (Figs. 2 and 3, insets) substantiated the accuracy of the measurements. For bacteria
expressing FepNFhuβ, FhuNFepβ and Fepβ, the samples with the lowest capacities, the mean standard
error of affinity measurements (Kd and Km) was 9.5%. For the same strains the mean standard error of
capacity and Vmax measurements was 2.3%.
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
13
Conformation of the hybrid proteins. The tertiary structures of the mutant proteins were pertinent to the
understanding of their binding and transport abilities. We characterized the conformations of the
chimeric proteins by native electrophoresis, immunochemistry, and non-specific permeability. The $-
barrels of OM proteins are resistant to denaturation by SDS, and they have a compact native structure that
imparts enhanced mobility in PAGE (56,57). Denaturation of OM proteins by heating in ionic detergents
eliminates this globular structure, reverting the abnormal electrophoretic behavior. In their native states
FepA and FhuA exhibit the same compact shape and increased electrophoretic migration (52,58), but they
are less stable than general or specific porins (59), and therefore more sensitive to denaturation. For these
reasons, electrophoretic behavior was a revealing measure of the tertiary structure of the chimeric
proteins. When electrophoresed in LDS at 4oC, FhuNFep$ and FepNFhu$ displayed the same rapid
mobility as FepA and FhuA, respectively (Fig. 4), intimating that all four proteins possessed a similarly
compact globular structure.
As a further indication of the folding of FepNFhu$, we determined the accessibility of epitopes
within its (FepA) N-terminus to MAb binding. In ELISAs utilizing four different anti-FepA MAbs that
bind epitopes in the N-terminal 150 residues, FepNFhu$ was indistinguishable from wild type FepA (Fig.
5). The antibodies against the N-domain did not react with either protein unless the antigen was
denatured before adsorption to the plates, suggesting that the N-domain of the chimeric protein was
sequestered within Fhu$, in a comparable manner to the N-domain within the barrel of the wild type
protein. The unavailability of antibodies to the FhuA N-terminus (J. A. Coulton, personal
communication) precluded a similar analysis of FhuNFep$. However, we cytofluorimetrically assessed
its proper folding, with anti-FepA MAbs against surface epitopes in the loops of its (FepA) C-terminal
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
14
domain. Like FepNFhu$ in ELISA, FhuNFep$ was indistinguishable from wild type FepA in flow
cytometry (Table 2).
Finally, neither chimeric protein conferred increased permeability to large antibiotics, whereas
bacteria expressing either Fep$ or Fhu$ showed enhanced susceptibility to bacitracin (1500 D; Table 2),
as expected for OM proteins containing a large, open channel. In total, the three different approaches all
suggested that both hybrid proteins assembled with N-termini lodged within their heterologous β-barrels.
Crosslinking with SulfoEGS. In the presence of formaldehyde, FepA crosslinks to another cell
envelope protein, TonB {for review see (4)}. However, skepticism about formaldehyde as an indicator of
proximity led us to test other crosslinking agents with more specific chemical targets. The bifunctional,
cleavable crosslinking agent SulfoEGS, selectively reacts with primary amines, and is too large (661 kD)
to pass through the aqueous channels of the OM, restricting its chemical action to Lys residues on the
membrane surface. When applied to live bacteria or purified E. coli outer membranes, SulfoEGS
generated two prominent protein complexes that included FepA, of approximate molecular weights 100
(Band 1) and 120 kD (Band 2; Fig. 6). The identity of the crosslinked protein(s) in the 100 kD complex is
currently unknown. When cleaved, the 120kD product yielded, in addition to FepA, the major OM
proteins OmpF/C and OmpA, as determined by Edman degradations of their N-terminal 15 residues (Fig.
6). The ostensible discrepancy in the concentration of FepA, OmpF/C and OmpA in lane 9 of Fig. 6
occurs because the molecular mass of FepA (81 kD) is almost 3-fold higher than those of OmpF/C or
OmpA. Accordingly, in SDS-PAGE of a (dissociated) 1:1 complex of FepA:OmpA, for example, a nearly
3-fold lower intensity is expected for the OmpA band. Two conditions influenced the crosslinking
reaction between FepA and the major OM proteins. First, both in vivo and in vitro the binding of FeEnt
to FepA eliminated Band 2, and drastically reduced the level of Band 1. The removal of the N-domain, in
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
15
Fep$, produced analogous results, reducing the levels of the 100 kD and 120 kD products to barely
detectable levels (Fig. 6).
Discussion
Since its discovery (39,40), the TonB-dependence of ferric siderophore acquisition has remained
obscure and controversial. The crystal structures of FepA and FhuA did not explain this aspect of ligand
uptake, although a change in the structure of the N-terminal extremity of FhuA during Fc binding (28,29)
raised the possibility of conformational signaling between the receptor protein and TonB, or another
component of the transport system. The localization of the “TonB-box” of siderophore transporters within
the region of the N-terminus that changed conformation in response to ligand binding was superficially
consistent with this notion of signal transduction between the bound receptor protein and TonB. In this
light it was unexpected that a mutant FhuA protein devoid of the N-terminal globular domain effectively
transported Fc (45). Our experiments had several objectives relevant to this phenomenon: (i) to verify the
phenotype of the N-domain deletion of FhuA, (ii) to determine whether equivalent mutants of FepA
transport FeEnt, (iii) to consider the functional exchangeability of the FepA and FhuA N-termini, and (iv)
to provide thermodynamic and kinetic descriptions of the binding and transport reactions by such mutant
receptors.
Equilibrium binding studies unequivocally demonstrated that the specificity of LGP, and their
high affinity for ferric siderophores, derive from their surface loops. The abilities of both Fep$ and Fhu$
to correctly discriminate FeEnt and Fc, and to bind them with wild-type affinity, confirms this point.
These data also concur with the identification of aromatic residues in the exterior-most portion of the
loops around the entrance to the FepA vestibule, that function in FeEnt recognition (49). In the crystal
structure of FhuA, Fc lodged deep within the vestibule interior, in contact with residues at the top of the
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
16
globular domain, suggesting that the N-domain contributes to the overall efficacy and specificity of the
ligand binding process. However, the sub-nanomolar binding affinity of the two proteins that lack the N-
domain, Fhu$ (for Fc) and Fep$ (for FeEnt) indicate that in both receptor proteins the globular domain
does not significantly affect ligand recognition or binding. The correct adsorption of colicins and phages
by both chimeras reiterates that if the N-domain acts at all in the binding reaction, then it’s contributions
are minimal and secondary to those of the surface loops.
One of the mutant constructs (FhuNFepβ) had a lower expression level, and others conferred lower
binding capacities than wild type FepA. To compare the uptake rates of the strains it was therefore
necessary to determine the number of molecules of ligand transported per receptor protein per unit of time.
k3 describes this rate, in essence the turnover number of each of the transport proteins. It derives from the
simple Michaelis-Menten description of the siderophore uptake reaction:
k1 k3 FeXout + Receptor W FeX-Receptor 6 Receptor + FeXin . k2
k3 is the only measure available by which to compare, in vivo, populations of proteins with different
capacities and concentrations in the membrane. In our study k3 is exactly equivalent to the term kcat, the
recommended basis for rate comparisons of enzymes: k3 = Vmax/Cap ; kcat = Vmax/ET .
Besides its high affinity, Fhu$ transported Fc in a TonB-dependent manner, with a better turnover
number than wild type FhuA (k3 in Table 2). Although the comparable FepA derivatives had reduced
transport rates, their activities must be related to the following series of controls: fepA bacteria, tonB
bacteria, and fepA+, energy depleted bacteria. These control strains/conditions show absolute inability to
transport any of the relevant ligands {Table 2, Fig. 3 (24,46,49,50)}. Therefore, even the least active
construct, FhuNFepβ, showed the complete range of functionalities. These data lead to the same
conclusion as that reached by others (45): the N-termini of FhuA and FepA are not needed for the
transport of Fc and FeEnt. Besides their importance to the understanding of ferric siderophore transport
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
17
(see below), these data demonstrate that TonB does not function by an interaction with the N-terminal
150 amino acids of ferric siderophore receptors, in spite of indirect evidence from genetic suppression
(60-62), crosslinking (63-68) and affinity chromatography (69). Thus the concepts of transmembrane
signaling and/or functional interactions between LGP and TonB, through the TonB-box, are specious.
Instead, the results suggest the contrary, that subsequent to the binding of metal chelates, the surface
loops of LGP undergo TonB-dependent conformational changes during the transport of metal chelates.
This conclusion from genetic engineering concurs with biophysical studies of loop conformational
dynamics in vivo (25).
Also contrary to previous suppositions (27-29,70) the globular domain apparently does not create a
siderophore-specific transit pathway to the periplasm. The idea of a substrate-customized channel
conflicts with the efficient internalization of the neutral Fc iron complex by FepNFhu$, which contains an
N-domain adapted to a negatively charged, catecholate ferric siderophore. The heterologous N-domain of
FepNFhu$ did not impair Fc uptake: like Fhu$, the chimera transported at a faster rate than wild type
FhuA. The FeEnt transport activity of FhuNFepβ reiterated this point. The ability of the chimeras to
internalize the ferric siderophore that their $-barrel domains adsorbed undercuts the notion of a transient,
ligand-specific pathway to the periplasm. On the contrary, bound metal chelates enter through a non-
specific route, neither created nor regulated by the N-terminus.
If it does not act in solute recognition or uptake, then what is the function of the N-terminus?
Constructs without an N-domain, or with an aberrant one, adsorbed ferric siderophores to much lower
capacities than the wild type receptors, suggesting that in the native proteins the proper insertion of the N-
domain into the barrel facilitates binding. Such an effect may occur by protein-protein interactions
between the globular domain and the barrel, that optimize the conformation of surface loops. In the
populations of empty $-barrel proteins that we studied only a fraction (~10%) of the molecules bound
ferric siderophores. However, the percentage of receptor proteins that adopted an appropriate
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
18
conformation, perhaps by random conformational motion (the flexibility of the loops of FepA was
apparent in its X-ray structure determination), attained wild-type affinity. So, in spite of the fact that it
does not act in ligand discrimination, the globular domain enhances the binding capacity of ferric
siderophore receptors. In this sense though, the N-domains of FepA and FhuA were not interchangeable,
because like Fep$ and Fhu$ the chimeric receptors also showed reduced binding capacity. We addressed
the possibility that in FepNFhu$ and FhuNFep$ the globular domains do not enter the porin $-barrels, but
remain suspended in the periplasm. Electrophoretic and immunochemical characterizations argued
against this alternative view of their structure: both hybrid proteins were identical to wild-type FepA and
FhuA in these tests. It is therefore likely that the globular domains of the chimeras reside within their $-
barrels, but the absence of appropriate interactions between the heterologous N- and C-domains prevents
the attainment of wild type binding capacity.
In addition to its function in the optimization of ferric siderophore binding, presence of the
globular domain within the $-barrel may preserve the integrity of the cell envelope to noxious small
molecules. The bacterial OM creates a selective permeability barrier that permits uptake of nutrients and
vitamins, generally smaller in mass than 600 D, and excludes larger, toxic, hydrophobic molecules,
including the natural detergents of the animal gut (48). In this light the obstruction of the large LGP
channel, which when open renders the bacteria susceptible to bile salts in the gut, is an important
rationale for the arrangement of the binding-receptive state that we propose. This general trend of OM
protein design appears in all other structurally characterized porins, including the E. coli transporters
OmpF and LamB, which both contain a uniquely oriented loop (L3) that restricts permeability through
their channels (30-32).
The crystal structure of FhuA created controversy about potential conformational changes that may
occur in the surface loops of LGP during ferric siderophore binding and transport. Although ample
evidence from biophysical studies suggested structural dynamics within FepA (25,33,71,72) and FhuA
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
19
(34) the absence of different loop conformations in the crystals of ligand-bound and ligand-free FhuA
challenged this deduction. Nevertheless, it is most relevant to consider conformational change in the in
vivo environment, free from the constraints of crystallography. A crystal structure captures only one of
the many conformational states that a protein may adopt, as aptly stated by Gutfreund: "...from a picture of
a racehorse, you can’t tell how fast it can run." (73). The crosslinking experiments reported herein provide
further contraindications to the static picture surmised from crystallography. When live bacteria were
exposed to sulfoEGS, FepA crosslinked to the major OM proteins OmpF/C and OmpA. Binding of
FeEnt eliminated the crosslinking, as did the removal of the N-domain, in Fep$. The experiment
demonstrates two loop conformations in vivo: an unoccupied form in which the surface loops reach close
proximity to contiguous OM proteins, and a bound form with closed loops that do not crosslink to other
proteins. The failure of Fep$ to crosslink to OmpF/C and OmpA suggests the need for the N-domain to
enter the barrel to create the open state.
We’ve previously objected to the use of formaldehyde crosslinking in the study of proposed
physical interactions between TonB and FepA, because of lack of control data on the crosslinking of TonB
to non-TonB-dependent OM proteins, uncertainties about the nature and identities of chemical targets of
the formaldehyde-activated residues within the crosslinked proteins, and the absence of quantitation of the
very small amount of total crosslinked products (4). The crosslinking experiments we report attempted to
satisfy these same concerns, by employing a reagent with well defined targets, and methods {coomassie
blue-stained SDS-PAGE gels and 125I-proteinA immunoblots} that were amenable to the identification of
all detectable crosslinked products, and their precise quantitation. In our experiments up to 70% of the
total FepA in the OM became involved in cross links. Band 2 (Fig. 6) was abundant enough that we
excised it from gels and determined the N-terminal sequences of its components.
The several experimentally defined forms of LGP provide insight into the iron transport process.
First, structural differences exist between the unbound and bound receptor protein, implicating
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
20
conformational change as a main element of the ligand binding process. During ferric siderophore
acquisition LGP likely progress between multiple conformations; at least one of the transitions between
structural forms requires energy and/or the participation of TonB. According to this view, FepA initially
assumes an “open” state, with the N-domain inside the barrel, where it compels the unoccupied loops to
an extended, optimized configuration for binding. FeEnt adsorption to the loops stimulates their
coalescence, creating a “closed “ complex. Besides our crosslinking data, ESR spectroscopy (33) (72)
described these open and closed forms, and kinetic fluorescence analyzes (71) showed the movement of
FeEnt between two distinct sites during binding. Consistent with this view, site-directed mutagenesis
defined residues in an exterior binding site {Y272, F329; Y260 (49)} and an interior site {R316 (50) and
Y260 (49)}. After the ferric siderophore attains binding equilibrium, it sits poised above the globular N-
domain, itself situated within the channel. Crystallographic experiments presumably captured and
described such a closed complex of FhuA. The transport-competent, empty $-barrels of FhuA and FepA
constitute another noteworthy form of LGP. Their very existence and functionality suggests that the N-
domain exits the barrel as ligands traverse the OM. On this basis we propose that TonB directly or
indirectly promotes loop dynamics that force bound ligands into the underlying pore, but this does not
occur by its binding to the “TonB-box” or another site within the N-domain of siderophore receptors,
because neither Fhu$ nor Fep$ contain these regions. Rather, we suggest that in the native receptors the
TonB-dependent dislodgement of the bound solute, driven by an inward contraction of the surface loops,
concomitantly ejects the N-domain and releases iron into the now open channel to the periplasm. Such
proposed movement of the N-terminus in and out of the channel superficially reprises one of the first
postulates of membrane protein function, the “ball & chain” hypothesis (74). The possibility of hydrogen
bonds between the residues on the surfaces of the globular domain and the β-barrel walls suggests an
energetic barrier to this reaction mechanism, perhaps explaining the energy dependence of iron transport
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
21
across the OM. For FhuA and FepA such movement of the globular domain does not cause or regulate
ligand uptake, but rather occurs in response to ligand internalization. This model also rationalizes the
problematical uptake of siderophore antibiotics, which often double the mass of an iron complex without
preventing its recognition and transport. A large, non-specific uptake pathway, comparable to those that
exist in the open $-barrels of FepA and FhuA, is necessary to accommodate the passage of such Trojan
horse antibiotics through the OM.
Acknowledgements
The authors thank Ines Chen, Paul Cook, Charles Earhart, Giovanna Ferro-Luzzi Ames, and Emil
Gotschlich for reading the manuscript, T. Hashimoto-Gotoh for pHSG575, and Marjorie Montague for
technical assistance This work was supported by NSF grant MCB9709418 and NIH grant GM53836.
Figure Legends
Figure 1. Crystal structures of FepA (top left), FhuA (top right), and the chimeric constructions
FhuNFep$$ (bottom left) and FepNFhu$$ (bottom right). The N-terminal domains of FepA and FhuA
are red and orange, their $-barrels are green and blue, and the $-turns between these regions are cyan and
magenta, respectively. The junction regions are also shown in space filling representations, including
Arg and Lys (green), Ser and Thr (white), and Glu (cyan); Pro and Gly are grey.
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
22
Figure 2. FeEnt and Fc binding. The adsorption of 59FeEnt (open symbols) and 59FeFc (filled
symbols) was measured with metabolically inactive KDF541 expressing FepA or FhuA (F), FhuNFep$
(ªª), FepNFhu$ (««), Fep$ (G), and Fhu$ (�), carried on pHSG575.
Figure 3. FeEnt and Fc transport. (Top) Ferric siderophore uptake was qualitatively evaluated by
nutrition tests. All of the plasmids failed to support growth in the tonB strain KDF571, producing tests
identical to those of KDF541. (Bottom) The transport of 59FeEnt (open symbols) and 59FeFc (filled
symbols) was quantitated in live bacteria. Strains and their symbols are the same as in Fig. 2, except
(inset) uptake by KDF571 (tonB) harboring pFep$ (x) and pFhuNFep$ (*).
Figure 4. Expression, localization and non-denaturing electrophoresis of mutant proteins. Panel
A: Expression. Western immunoblots stained with anti-FepA MAbs 26 (top) and 45 (bottom) and
developed with 125I-protein A. Lanes 1-3: 1, 3 and 5 ug of purified FepA, respectively; Lanes 4-10 each
contain lysates of 2. 5 x 108 bacterial cells, except lane 9, which contained 108 cells. Lane 4: KDF541
(fepA, fhuA); 5:KDF541/pFepNFhu$; 6:KDF541/pFhuNFep$; 7:KDF541/pITS23 (fepA+); 8: BN1071
(fepA+, fhuA+); 9: KDF541/pITS449(fepA+); 10: KDF541/pFep$. Expression levels were quantitated by
image analysis, and tabulated in Table 2. Panel B: Localization. Inner and outer membranes from
bacteria expressing the chimeric proteins were fractionated on sucrose gradients, and 30 ug of each sample
was analyzed by western immunoblots with monoclonal antibodies 26 (top) and 45 (bottom). Lane 1:
KDF541 cell lysate; lanes 2-4: KDF541/pITS23 lysate, IM and OM , respectively; lanes 5-7:
KDF541/pFhuNFep$ lysate, IM and OM; lanes 8-10: KDF541/pFepNFhu$ lysate, IM and OM. Panel
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
23
C: Non-denaturing SDS-PAGE. The left panel is a coomassie blue-stained LDS-PAGE gel; the right
panel is a western immunoblot stained with a mixture of anti-FepA MAbs 26 and 45 and developed with
125I-protein A. Lane 1, molecular mass markers (Bio Rad, Hercules CA); lanes 2 and 3; purified,
denatured FepA and FhuA, respectively; lanes 4 and 5: purified, non-denatured FepA and FhuA,
respectively; lanes 6-8: OM fragments (30 ug) from KDF541 expressing FepA, FhuNFep$, and
FepNFhuβ.
Figure 5. Analysis of FepNFhu$$ by ELISA. Outer membranes from KDF541 carrying pITS23 (left) or
pFepNFhu$ (center), or purified, denatured FepA (right) were adsorbed to microtiter plates and tested
with anti-FepA MAbs 1 (�) , 12 (�), 27 (�), 41 (�), 33 (�), and 45 (�), or normal mouse serum (x).
The first four antibodies recognize different epitopes in the N-terminal globular domain, while the latter
two bind epitopes in surface loops of the barrel. Panels A and B plot mean values from 3 experiments;
Panel C shows the results of a single experiment.
Figure 6. Crosslinking of bacteria with Sulfo-EGS. A. Outer membranes from BN1071 (lanes 1-3)
or KDF541/pITS449 (lanes 4-6) were incubated with the cleavable crosslinker Sulfo-EGS, subjected to
SDS-PAGE, and stained with coomassie blue. Lanes 1 and 4: no crosslinking; lanes 2 and 5: crosslinked
by SulfoEGS. Lanes 3 and 6: crosslinked by SulfoEGS in the presence of 5 uM FeEnt. Lane 7:
molecular weight standards myosin (200 kD), $-galactosidase (116 kD), phosphorylase b (97 kD) bovine
serum albumin (66 kD), ovalbumin (45 kD), and carbonic anhydrase (31 kD). Lane 8: purified FepA.
Lane 9: band 2 in lane 2 was excised, cleaved with hydroxylamine, electroeluted and re-electrophoresed:
note the resulting proteins that migrate with apparent molecular masses of 34 kD and 36kD, whose
identity as OmpF/C and OmpA was verified by sequence analysis of their N-terminal 15 residues (data not
shown). B. Outer membranes from KDF41/pITS23 (Lanes 1-3), or cell lysates of KDF541 harboring
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
24
pITS23, (lanes 4-6) or pFep$ (Lanes 7-9) were either untreated (lanes 1,4, 7), or crosslinked with Sulfo-
EGS in the absence (lanes 2, 5, 8) or presence of 5 uM FeEnt (lanes 3, 6, 9), and cell lysates were
transferred to nitrocellulose and analyzed by immunoblot with anti-FepA MAb 45, developed with 125I-
proteinA.
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
25
Genotype Reference Strains BN1071 F- thi, entA, pro, trp, rpsL (75) KDF541 BN1071 recA, fepA, fhuA, cir (24) KDF571 KDF541 tonB (24) AN193 F- pro leu, trp, thi, fhuA, lac, rpsL, entA (75) Plasmids pHSG575 (47) PITS449 fepA+ on pUC18 (24) pITS23 fepA+ on pHSG575 ¶ this study pFep$ fepA∆17-150 “ pFep$2 fepA∆3-150 “ pFepNFhuβ fepA1-152::fhuA160-723 “ pITS11 fhuA+ on pHSG575 “ pFhu$ fhuA∆5-160 “ pFhuNFepβ fhuA 1-155::fepA149-724 “ pFhuNFepβ2 fhuA 1-160::fepA153-724 “
Table 1 Strains and Plasmids.
¶ All constructions were analyzed as derivatives of the low copy plasmid PHSG575.
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
26
Ferric Enterobactin Protein Ligands†
Binding¶ Transport§
MAbs‡
Antibiotics¢
Strain/plasmid Kd Cap Nutr¥ Km Vmax k3
ColB
ColD 33 45 N B E
Exp.^̂
KDF541 NB 0 0 NT 0 0 R R 0.15 0.13 14 0 0 0
“ /pFepA 0.4 64.0 18 0.40 411 6.4 105 4x104 7.8 14.1 14 0 8 1.2x105
“ /pFhuNFep$ 0.2 1.6 25* 1.7 1.6 0.9 103 4x101 7.6 8.3 14 0 8 1.5x104
“ /pFepA$ 0.6 4.3 20* 1.45 1.4 0.3 4x104 2x103 8.9 13.8 14 8 10 0.8x104
Ferrichrome ColM T5 NN80
AN193 NB 0 0 NT 0 0 R R R ND ND 14 0 8 0
“ /pFhuA 0.6 23 19 0.6 87 4 2x102 2x106 108 ND ND 14 0 8 ND
“ /pFepNFhu$ 0.6 1.6 19 2.5 9.8 6.5 R 105 107 0.2 0.18 14 0 8 9.4x104
“ /pFhu$ 1.0 3.0 21 3.6 25 8.3 2x102 105 4x107 ND ND 16 8 8 ND
Table 2. Biochemical properties of FepA and FepA mutants.
¶ Kd (nM) and Capacity (pMol bound /109 cells) were determined from the concentration dependence of
FeEnt binding, by analyzing the mean values from three independent experiments with Grafit 4
(Erithacus), using the “Bound vs Total” equation. For FeEnt, the mean standard errors for Kd and
Capacity were 16% and 4%, respectively; For Fc, the mean standard errors for Kd and Capacity were 15%
and 2% . NB: no binding.
¥ Nutrition tests express the diameters (mm) of growth halos. The results of a single experiment are
given, but the experiments were repeated several times without significant variation. An asterisk (*)
indicates that a very faint halo was observed (see Fig. 3).
§ Km (nM) and Vmax (pMol/min/109 cells) of uptake were determined from the concentration dependence
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
27
of FeEnt transport, by analyzing the mean values from three independent experiments with Grafit 4
(Erithacus), using the “Enzyme Kinetics” equation. For FeEnt, the mean standard errors for Km and Vmax
were 17% and 4%, respectively. For Fc, the mean standard errors for Km and Vmax were 27% and 6%.
The turnover number for each protein (k3; molecules transported/min) was calculated by dividing Vmax by
capacity. NT: no transport.
† Colicin and bacteriophage susceptibility was determined by measuring the killing of KDF541
expressing the wild type and mutant proteins by limiting dilutions of colicins B, D, and M, and phage T5
and N80. The results are expressed as the titer of the killing agents on the different bacterial strains.
‡ Anti-FepA MAbs 33 and 45, that recognize sites in loops 2 and 4, respectively, were used to measure
the exposure of surface loops on the bacteria expressing the proteins of interest, relative to the FepA-
deficient strain KDF541. The tabulated values are mean fluorescent intensity. The data are from a single
experiment, but the experiment was repeated several times with little variation.
¢ The tabulated values are the diameter (mm) of the zone of clearing created by a 6 mm filter paper disk,
imbedded with an antibiotic (N: Neomycin, 30 ug/ml; E: Erythromycin, 15 ug/ml; B: Bacitracin, 10
IU/ml), on a lawn of the bacteria.
^ The expression of FepA and its chimeras (copies/cell) was measured by SDS-PAGE and immunoblots
with anti-FepA MAbs and 125I-protein A (Fig. 4).
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
28
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
29
D.C. Scott et al., Fig. 2
[FeEnt] (nM)
0 2 4 6 8 10
pM
ol B
ou
nd
/109
Cel
ls
0
20
40
60
0 2 40
0.5
1
1.5
2
[Fc] (nM)
0 5 10 15 20
0
10
20
0 5 10 150
0.5
1
1.5
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
30
D.C. Scott et al., Fig. 3b
[Fc] (nM)0 10 20 30 40
0
25
50
75
100
[FeEnt]( nM)0 2 4 6 8 10
pM
ol/m
in/1
0E9
cells
0
100
200
300
400
0 2 4 6 8 100
0.5
1
1.5
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
31
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
32
Log Dilution1 2 3 4 5 6 7
A B C
Log Dilution
1 2 3 4 5 6 7
Ab
sorb
ance
0
0.5
1
1.5
2
2.5
Log Dilution1 2 3 4 5 6 7
D.C. Scott et al., Fig. 5
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
33
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
34
1. Neilands, J. B. (1982) Annu Rev Microbiol 36, 285-309
2. Neilands, J. B. (1995) J Biol Chem 270(45), 26723-6
3. Braun, V. (1995) FEMS Microbiol Rev 16(4), 295-307
4. Klebba, P. E., and Newton, S. M. (1998) Curr Opin Microbiol 1(2), 238-247
5. van der Helm, D. (1998) Met Ions Biol Syst 35, 355-401
6. Thulasiraman, P., Newton, S. M., Xu, J., Raymond, K. N., Mai, C., Hall, A., Montague, M. A., and
Klebba, P. E. (1998) J Bacteriol 180(24), 6689-96
7. Guterman, S. K. (1971) Biochem Biophys Res Commun 44(5), 1149-55
8. Di Masi, D. R., White, J. C., Schnaitman, C. A., and Bradbeer, C. (1973) J Bacteriol 115(2), 506-
13
9. Wayne, R., Frick, K., and Neilands, J. B. (1976) J Bacteriol 126(1), 7-12
10. McKee, J. A., Sharma, S. K., and Miller, M. J. (1991) Bioconjug Chem 2(4), 281-91
11. Dolence, E. K., Minnick, A. A., Lin, C. E., Miller, M. J., and Payne, S. M. (1991) J Med Chem
34(3), 968-78
12. Miller, M. J., McKee, J. A., Minnick, A. A., and Dolence, E. K. (1991) Biol Met 4(1), 62-9
13. Dolence, E. K., Minnick, A. A., and Miller, M. J. (1990) J Med Chem 33(2), 461-4
14. Fernandez-Beros, M. E., Gonzalez, C., McIntosh, M. A., and Cabello, F. C. (1989) Infect Immun
57(4), 1271-5
15. Furman, M., Fica, A., Saxena, M., Di Fabio, J. L., and Cabello, F. C. (1994) Infect Immun 62(9),
4091-4
16. Cornelissen, C. N., and Sparling, P. F. (1994) Methods Enzymol 235, 356-63
17. Zhu, W., Hunt, D. J., Richardson, A. R., and Stojiljkovic, I. (2000) J Bacteriol 182(2), 439-47
18. Bearden, S. W., and Perry, R. D. (1999) Mol Microbiol 32(2), 403-14
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
35
19. Occhino, D. A., Wyckoff, E. E., Henderson, D. P., Wrona, T. J., and Payne, S. M. (1998) Mol
Microbiol 29(6), 1493-507
20. Cope, L. D., Hrkal, Z., and Hansen, E. J. (2000) Infect Immun 68(7), 4092-101
21. Jurado, R. L. (1997) Clin Infect Dis 25(4), 888-95
22. Cho, S. S., Lucas, J. J., and Hyndman, A. G. (1999) Brain Res 816(1), 229-33
23. Posey, J. E., and Gherardini, F. C. (2000) Science 288(5471), 1651-3
24. Rutz, J. M., Liu, J., Lyons, J. A., Goranson, J., Armstrong, S. K., McIntosh, M. A., Feix, J. B., and
Klebba, P. E. (1992) Science 258(5081), 471-5
25. Jiang, X., Payne, M. A., Cao, Z., Foster, S. B., Feix, J. B., Newton, S. M., and Klebba, P. E. (1997)
Science 276(5316), 1261-4
26. Killmann, H., Benz, R., and Braun, V. (1993) Embo J 12(8), 3007-16
27. Buchanan, S. K., Smith, B. S., Venkatramani, L., Xia, D., Esser, L., Palnitkar, M., Chakraborty,
R., van der Helm, D., and Deisenhofer, J. (1999) Nat Struct Biol 6(1), 56-63
28. Locher, K. P., Rees, B., Koebnik, R., Mitschler, A., Moulinier, L., Rosenbusch, J. P., and Moras,
D. (1998) Cell 95(6), 771-8
29. Ferguson, A. D., Hofmann, E., Coulton, J. W., Diederichs, K., and Welte, W. (1998) Science
282(5397), 2215-20
30. Weiss, M. S., Wacker, T., Weckesser, J., Welte, W., and Schulz, G. E. (1990) FEBS Lett 267(2),
268-72
31. Cowan, S. W., Schirmer, T., Rummel, G., Steiert, M., Ghosh, R., Pauptit, R. A., Jansonius, J. N.,
and Rosenbusch, J. P. (1992) Nature 358(6389), 727-33
32. Schirmer, T., Keller, T. A., Wang, Y. F., and Rosenbusch, J. P. (1995) Science 267(5197), 512-4
33. Liu, J., Rutz, J. M., Klebba, P. E., and Feix, J. B. (1994) Biochemistry 33(45), 13274-83
34. Moeck, G. S., Tawa, P., Xiang, H., Ismail, A. A., Turnbull, J. L., and Coulton, J. W. (1996) Mol
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
36
Microbiol 22(3), 459-71
35. Letellier, L., Locher, K. P., Plancon, L., and Rosenbusch, J. P. (1997) J Biol Chem 272(3), 1448-
51
36. Bradbeer, C. (1993) J Bacteriol 175(10), 3146-50
37. Reynolds, P. R., Mottur, G. P., and Bradbeer, C. (1980) J Biol Chem 255(9), 4313-9
38. Pugsley, A. P., and Reeves, P. (1976) J Bacteriol 127(1), 218-28
39. Wang, C. C., and Newton, A. (1971) J Biol Chem 246(7), 2147-51
40. Guterman, S. K., and Dann, L. (1973) J Bacteriol 114(3), 1225-30
41. Postle, K., and Good, R. F. (1983) Proc Natl Acad Sci U S A 80(17), 5235-9
42. Raymond, K. M. (1994) Pure&Appl.Chem. 66(4), 773-781
43. McIntosh, M. A., and Earhart, C. F. (1976) Biochem Biophys Res Commun 70(1), 315-22
44. Wayne, R., and Neilands, J. B. (1975) J Bacteriol 121(2), 497-503
45. Braun, M., Killmann, H., and Braun, V. (1999) Mol Microbiol 33(5), 1037-49
46. Newton, S. M., Igo, J. D., Scott, D. C., and Klebba, P. E. (1999) Mol Microbiol 32(6), 1153-1165
47. Hashimoto-Gotoh, T., Franklin, F. C., Nordheim, A., and Timmis, K. N. (1981) Gene 16(1-3),
227-35
48. Nikaido, H., and Vaara, M. (1985) Microbiol Rev 49(1), 1-32
49. Cao, Z., Qi, Z., Sprencel, C., Newton, S. M., and Klebba, P. E. (2000) Mol Microbiol 37(6), 1306-
17
50. Newton, S. M., Allen, J. S., Cao, Z., Qi, Z., Jiang, X., Sprencel, C., Igo, J. D., Foster, S. B., Payne,
M. A., and Klebba, P. E. (1997) Proc Natl Acad Sci U S A 94(9), 4560-5
51. Sprencel, C., Cao, Z., Qi, Z., Scott, D. C., Montague, M. A., Ivanoff, N., Xu, J., Raymond, K. M.,
Newton, S. M., and Klebba, P. E. (2000) J Bacteriol 182(19), 5359-64
52. Murphy, C. K., Kalve, V. I., and Klebba, P. E. (1990) J Bacteriol 172(5), 2736-46
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
37
53. Liu, J., Rutz, J. M., Feix, J. B., and Klebba, P. E. (1993) Proc Natl Acad Sci U S A 90(22), 10653-7
54. Smit, J., Kamio, Y., and Nikaido, H. (1975) J Bacteriol 124(2), 942-58
55. Wiener, M., Freymann, D., Ghosh, P., and Stroud, R. M. (1997) Nature 385(6615), 461-4
56. Schnaitman, C. A. (1973) Arch Biochem Biophys 157(2), 541-52
57. Mizushima, S. (1974) Biochem Biophys Res Commun 61(4), 1221-6
58. Locher, K. P., and Rosenbusch, J. P. (1997) Eur J Biochem 247(3), 770-5
59. Klug, C. S., Su, W., Liu, J., Klebba, P. E., and Feix, J. B. (1995) Biochemistry 34(43), 14230-6
60. Schramm, E., Mende, J., Braun, V., and Kamp, R. M. (1987) J Bacteriol 169(7), 3350-7
61. Heller, K. J., Kadner, R. J., and Gunther, K. (1988) Gene 64(1), 147-53
62. Gudmundsdottir, A., Bell, P. E., Lundrigan, M. D., Bradbeer, C., and Kadner, R. J. (1989) J
Bacteriol 171(12), 6526-33
63. Skare, J. T., Ahmer, B. M., Seachord, C. L., Darveau, R. P., and Postle, K. (1993) J Biol Chem
268(22), 16302-8
64. Larsen, R. A., Wood, G. E., and Postle, K. (1993) Mol Microbiol 10(5), 943-53
65. Larsen, R. A., Foster-Hartnett, D., McIntosh, M. A., and Postle, K. (1997) J Bacteriol 179(10),
3213-21
66. Cadieux, N., and Kadner, R. J. (1999) Proc Natl Acad Sci U S A 96(19), 10673-8
67. Merianos, H. J., Cadieux, N., Lin, C. H., Kadner, R. J., and Cafiso, D. S. (2000) Nat Struct Biol
7(3), 205-9
68. Cadieux, N., Bradbeer, C., and Kadner, R. J. (2000) J Bacteriol 182(21), 5954-61
69. Moeck, G. S., Coulton, J. W., and Postle, K. (1997) J Biol Chem 272(45), 28391-7
70. Ferguson, A. D., Braun, V., Fiedler, H. P., Coulton, J. W., Diederichs, K., and Welte, W. (2000)
Protein Sci 9(5), 956-63
71. Payne, M. A., Igo, J. D., Cao, Z., Foster, S. B., Newton, S. M., and Klebba, P. E. (1997) J Biol
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
D.C. Scott et al.
38
Chem 272(35), 21950-5
72. Klug, C. S., Eaton, S. S., Eaton, G. R., and Feix, J. B. (1998) Biochemistry 37(25), 9016-23
73. Gutfreund, H., and Knowles, J. R. (1967) Essays in Biochemistry 3, 25-67
74. Armstrong, C. M., and Bezanilla, F. (1977) J Gen Physiol 70(5), 567-90
75. Klebba, P. E., McIntosh, M. A., and Neilands, J. B. (1982) J Bacteriol 149(3), 880-8
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from
Newton and Phillip E. KlebbaDaniel C. Scott, Zhenghua Cao, Zengbiao Qi, Matthew Bauler, John D. Igo, Salete M.C.
Exchangeability of N-termini in the ligand-gated porins of Escherichia coli
published online January 19, 2001J. Biol. Chem.
10.1074/jbc.M011282200Access the most updated version of this article at doi:
Alerts:
When a correction for this article is posted•
When this article is cited•
to choose from all of JBC's e-mail alertsClick here
by guest on April 12, 2018
http://ww
w.jbc.org/
Dow
nloaded from