Date post: | 20-Nov-2023 |
Category: |
Documents |
Upload: | independent |
View: | 0 times |
Download: | 0 times |
ORIGINAL PAPER
Ga3+ as a mechanistic probe in Fe3+ transport:characterization of Ga3+ interaction with FbpA
Katherine D. Weaver Æ Jared J. Heymann Æ Arnav Mehta Æ Petra L. Roulhac ÆDamon S. Anderson Æ Andrew J. Nowalk Æ Pratima Adhikari ÆTimothy A. Mietzner Æ Michael C. Fitzgerald Æ Alvin L. Crumbliss
Received: 5 February 2008 / Accepted: 4 April 2008 / Published online: 7 May 2008
� SBIC 2008
Abstract The obligate human pathogens Haemophilus
influenzae, Neisseria gonorrhoeae, and N. meningitidis uti-
lize a highly conserved, three-protein ATP-binding cassette
transporter (FbpABC) to shuttle free Fe3+ from the periplasm
and across the cytoplasmic membrane. The periplasmic
binding protein, ferric binding protein (FbpA), is capable of
transporting other trivalent cations, including Ga3+, which,
unlike Fe3+, is not redox-active. Because of a similar size and
charge as Fe3+, Ga3+ is widely used as a non-redox-active
Fe3+ substitute for studying metal complexation in proteins
and bacterial populations. The investigations reported here
elucidate the similarities and differences in FbpA seques-
tration of Ga3+ and Fe3+, focusing on metal selectivity and
the resulting transport function. The thermodynamic binding
constant for Ga3+ complexed with FbpA at pH 6.5, in 50 mM
4-morpholineethanesulfonic acid, 200 mM KCl, 5 mM
KH2PO4 was determined by UV-difference spectroscopy as
log K 0eff ¼ 13:7� 0:6: This represents a 105-fold weaker
binding relative to Fe3+ at identical conditions. The unfold-
ing/refolding behavior of Ga3+ and Fe3+ holo-FbpA were
also studied using a matrix-assisted laser desorption/ioni-
zation time-of-flight mass spectroscopy technique, stability
of unpurified proteins from rates of H/D exchange (SUP-
REX). This analysis indicates significant differences
between Fe3+ and Ga3+ sequestration with regard to protein
folding behavior. A series of kinetic experiments established
the lability of the Ga3+FbpA–PO4 assembly, and the simi-
larities/differences of stepwise loading of Fe3+ into apo- or
Ga3+-loaded FbpA. These biophysical characterization data
are used to interpret FbpA-mediated Ga3+ transport and
toxicity in cell culture studies.
Keywords Binding constant � Ferric binding protein �Iron transport � Gallium transport � Kinetics
Introduction
Iron is an essential nutrient and serves numerous functions
in vivo, in part owing to its facile, biologically accessible
redox chemistry. In aqueous media, iron commonly exists
in two stable oxidation states: Fe2+ (d6) and Fe3+ (d5). The
redox potential between these two states is controlled by a
number of factors, most importantly the first coordination
sphere, and couples iron to a range of metabolic activities
that must be carefully regulated to prevent cytotoxic
reactions [1].
The ability to scavenge metals, particularly iron, is critical
to bacterial infection of the human host. Iron is transported
through the periplasm of Gram-negative bacteria Hae-
mophilus influenzae, Neisseria gonorrhoeae, and
N. meningitidis by ferric binding protein (FbpA) [2, 3].
Electronic supplementary material The online version of thisarticle (doi:10.1007/s00775-008-0376-5) contains supplementarymaterial, which is available to authorized users.
K. D. Weaver � J. J. Heymann � A. Mehta �P. L. Roulhac � M. C. Fitzgerald � A. L. Crumbliss (&)
Department of Chemistry,
Duke University,
Durham, NC 27708-0346, USA
e-mail: [email protected]
D. S. Anderson
Molecular Cardiology Research Institute,
TUFTS-New England Medical Center,
Boston, MA 02111, USA
D. S. Anderson � A. J. Nowalk � P. Adhikari � T. A. Mietzner
Department of Molecular Genetics and Biochemistry,
University of Pittsburgh School of Medicine,
Pittsburgh, PA 15261, USA
123
J Biol Inorg Chem (2008) 13:887–898
DOI 10.1007/s00775-008-0376-5
FbpA is a 34-kDa protein with a single iron binding site that
utilizes four amino acid residues, including two tyrosines,
one histidine, and one glutamate residue. The inner coordi-
nation shell of iron is completed by water and a synergistic
anion ligand [4–6]. The process of high-affinity iron uptake
in bacteria was studied here using the free iron ATP-binding
cassette transporter (FbpABC). Orthologs of FbpABC are
found in bacterial species as diverse as N. gonorrhoeae,
H. influenzae, Pasteurella multocida, Mannheimia haemol-
ytica, Pseudomonas aeruginosa, Serratia marcescens, Bur-
kholderia cepacia, Yersinia enterocolitica, Actinobacillus
actinomycetemcomitans, Shewanella oneidensis, Synecho-
coccus elongatus, and Ehrlichia chaffeensis, among many
other bacteria [7].
Substrate selectivity exhibited by FbpA ensures that the
appropriate, biologically active metal ion will be trans-
ported, delivered, and inserted into the target proteins.
However, similar metals, when available in sufficient
environmental excess, may be able to replace Fe3+ as the
FbpA substrate. This metal mimicry results in metaboli-
cally inactive enzymes or unstable metalloproteins [8].
Factors controlling the partitioning of metals include ionic
size, oxidation state, and specific electron orbital charac-
teristics [8]. The spherically symmetrical, tripositive
cations Fe3+ and Ga3+ are both hard Lewis acids by virtue
of their high charge density. The similarity in ionic radius
between these trivalent metals (Fe3+ 0.65 A; Ga3+ 0.62 A)
permits gallium to substitute for iron in FbpA [9, 10].
Gallium has been utilized as a redox-inactive iron sub-
stitute owing to the commonly accepted premise that the
similarity between the charge and ionic radius of Ga3+ and
Fe3+ makes them indistinguishable by biological systems
[11]. Ga3+ substitution into metalloenzymes occurring at a
sufficient excess of Ga3+ over Fe3+ can result in a loss of
enzymatic function. These proteins are rendered inactive in
some cases owing to the inability to access the essential 3+/
2+ redox chemistry, resulting in cellular toxicity. Despite
the critical differences in redox activity between the two
metals, Ga3+ and Fe3+ appear in certain cases to utilize
common uptake and transport systems. The mechanism of
Ga3+ uptake is of interest owing to the applications of
gallium as an antitumor or imaging (through radiogallium67Ga) agent, along with its use to probe hypothesized Fe3+/
Fe2+ redox processes during transport. One such gallium-
based iron transport study demonstrated that gallium
enhances non-transferrin-mediated iron uptake by cells, but
rapidly binds to plasma transferrin (Tf), therefore inhibiting
the more prominent Tf-mediated iron acquisition [12].
Tf and FbpA belong to the same protein superfamily and
are responsible for iron transport in vertebrates and certain
pathogenic Gram-negative bacteria. FbpA and Tf are both
recognized by specific receptors, where their metal sub-
strate is transferred to the inner cellular machinery.
Gallium uptake by Tf follows a different mechanism from
that of iron, and interaction of GaTf with Tf receptor 1
(TfR) also differs from that of FeTf, implying that kinetics
may play a more prominent role than thermodynamics in
the TfR interactions of GaTf compared with FeTf [13].
Similarly, gallium-substituted ferric siderophore receptors
in bacteria show reduced rates of uptake [14]. The increase
in strength of Fe3+ binding to Tf over Ga3+ matches cor-
relations in metal ion acidity (hydrolysis constant pKa1 for
Fe3+ and Ga3+ is 2.68 and 3.09, respectively) [15, 16].
In biological studies probing metal sensitivity, Ga3+
exhibits significant toxicity toward siderophore-deficient
Escherichia coli expressing the complete FbpABC operon
[17]. This report demonstrates the ability of Ga3+ to use the
FbpABC transporter to enter the cytosol, imparting toxicity
through cellular iron deprivation [17]. Iron release from
FbpA and subsequent transcytoplasmic membrane trans-
port has been proposed as a redox-mediated process, as the
FbpA affinity for iron is decreased by approximately 12
orders of magnitude upon Fe3+/2+ reduction [18–20]. The
FbpABC-induced cytosolic presence of the redox-inactive
Ga3+ would appear to refute reductive-induced metal
release from FbpA. However, subtle differences in the
transport of Fe3+ and Ga3+ have been shown in other gal-
lium toxicity studies [21]. Gallium toxicity has been
eliminated through the biological selection of a gallium-
resistant mutation in the FbpABC transporter, which was
coupled with a restoration of wild-type levels of iron
transport activity [21]. Thus, the effects of gallium on
cellular iron metabolism are complex, requiring further
study to delineate these differences in metal transport
mechanisms.
In this study, the differences between Fe3+- and Ga3+-
loaded FbpA were investigated and a mechanism is pre-
sented for metal binding and selectivity. Through further
analysis of these properties, suggestions are offered
explaining the chemical origins of the biological differen-
tiation between these apparently in vivo ‘‘identical’’
cations.
Materials and methods
Ga(NO3)3 was purchased from Alfa Aesar. Chelex-100 was
purchased from Bio-Rad. DOWEX 50W-X8, 20–50-mesh
resin was obtained from J.T. Baker. Tryptic soy broth,
nutrient broth agar (NBA), KCl, KOH, NaOH, 4-mor-
pholineethanesulfonic acid (MES), trifluoroacetic acid,
methanol, and acetonitrile were purchased from Fisher.
2,20-Dipyridyl (dip), ampicillin, cetyltrimethylammonium
bromide (CTAB), FeCl3, Fe(ClO4)3, nitrilotriacetic acid
(NTA), tris(hydroxymethyl)aminomethane (Tris base),
sodium perchlorate, D2O (99.9 atom% D), deuterium
888 J Biol Inorg Chem (2008) 13:887–898
123
chloride (20 wt% in D2O, 99.5 atom% D), sodium deu-
teroxide (40 wt% in D2O, 99.9 atom% D), sinapinic acid,
aldolase (rabbit muscle) and soybean trypsin inhibitor,
were obtained from Sigma Aldrich. Sterile antimicrobial
discs were purchased from Difco (Detroit, MI, USA).
KH2PO4 was purchased from United States Biochemical
Corporation. Guanidinium chloride (GdmCl) was pur-
chased from Mallinckrodt. Deuterated guanidinium
hydrochloride was prepared by repeated dissolution and
lyophilization of fully protonated guanidine hydrochloride
in D2O until the calculated deuterium content was greater
than 99%.
Isolation and purification of FbpA
The technique of obtaining purified apo-FbpA cloned from
N. gonorrhoeae strain F62 and expressed in an E. coli
background was used as previously reported [18, 22].
Briefly, E. coli overexpressing FbpA was subjected to
extraction by cetyltrimethylammonium bromide in 0.1 M
Tris base at pH 8.0, followed by binding to a carboxy-
methyl Sepharose column. While bound to the column,
iron-loaded FbpA was converted to apo-FbpA by addition
of 10 vol of 1 mM citrate in 0.1 M Tris base at pH 8.0.
Apo-FbpA was eluted using a NaCl gradient in buffers
rendered iron-free by exposure to Chelex-100 prior to their
addition to the column. Fractions were collected in acid-
washed glassware and extensively dialyzed against Chelex-
100 treated 0.05 M MES with 0.2 M KCl and concentrated
using an Amicon centrifugal filter device (Millipore).
Stock solutions
FbpA concentrations were determined spectrophotometri-
cally using the molar extinction coefficients for the protein
at 278 nm for apo-FbpA (45,990 M-1 cm-1) and holo-
FbpA (51,100 M-1 cm-1), and at the synergistic anion-
and metal-sensitive ligand-to-metal charge transfer band at
481 nm for FeFbpA–PO4 (2,430 M-1 cm-1) [19]. Stock
5 mM acidic Ga(NO3)3 solutions were prepared from
strongly acidic solutions by slowly adjusting the pH to
approximately 2.1 with microliter quantities of 0.1 M
KOH. Stock 7.5 mM Ga(NTA)2 solutions were prepared
by adding 2 molar equiv of NTA to acidic Ga(NO3)3
solutions. The pH was adjusted with microliter quantities
of 0.1 M KOH to produce Ga(NTA)2 at pH * 5.5. Stock
5 mM Fe(NTA)2 solutions were prepared by adding
2 molar equiv of NTA to acidic FeCl3 solutions. The pH
was adjusted using microliter quantities of 0.1 M KOH to
produce Fe(NTA)2 at pH * 5.5. A filtered stock sodium
perchlorate solution was prepared and standardized by
passage through a DOWEX 50W-X8 strongly acidic cat-
ion-exchange column in the H+ form, followed by titration
against standardized NaOH to a phenolphthalein end point.
Kinetics buffer solutions of 0.05 M MES, 0.150 M NaClO4
were prepared and adjusted to pH 6.5 with a minimal
volume of NaOH monitored with an Orion model 230 pH
meter. A stock solution of 0.1 M Fe(NTA) was prepared by
slowly adding the appropriate volume of a previously
standardized Fe(ClO4)3 solution [23] to a vigorously stirred
solution of NTA in kinetics buffer, to obtain a final molar
ratio of Fe to NTA of 1:1.1. GaFbpA–PO4 solutions used
for kinetic analysis were prepared by the addition of
Ga(NTA) to apo-FbpA in kinetics buffer while monitoring
the addition using a UV-difference spectroscopic technique
(described later) until a stable absorbance was achieved.
All solutions were prepared in ultrapure 18 MX cm
equivalent water (Hydropure).
Gallium inhibition of FbpABC expressing E. coli
Bacteriologic assays were performed as previously
described [21] in an E. coli strain H-1443 background. N.
gonorrhoeae derived FbpA (nFbpA) is amenable to large-
scale purification and solubility when compared with H.
influenzae FbpA (hFbpA) and was used for the thermody-
namic and kinetic studies. However, for the cell-based
assays, the closely related genetic construct of the H. in-
fluenzae operon hitABC was used because toxicity is
substantially decreased in the E. coli strain H-1443 back-
ground [24, 25]. For all practical purposes, purified
preparations of hFbpA and nFbpA behave identically with
regard to metal binding. The rationale for using hFbpA is
that E. coli is more tolerant of the hFbpA operon.
Bacterial growth conditions were as follows. Nutrient
broth combined with bacto agar to prepare NBA and sterile
supplement discs were purchased from Difco (Detroit, MI,
USA). Bacteria were inoculated on NBA supplemented
with 100 lg mL-1 ampicillin (NBAamp100) to maintain
the plasmid. Cells were plated on NBAamp100 containing
100 lM of the iron chelator 2,20-dipyridyl (dip) in order to
limit the amount of iron in the medium. This bacterial
propagation medium is referred to as NBAamp100dip100. A
100-lL aliquot of a 1 9 103 colony-forming units per
milliliter solution was spread on a standard nutrient agar
plate. E. coli strain H-1443 was selected owing to its
inability to scavenge iron using phenolate siderophores
resulting from a defect in its aromatic biosynthesis pathway
[24]. However, this strain is permissive for growth on this
medium by competition for iron using low-affinity systems
[24]. NBAamp100dip100 was used to select for E. coli
strains expressing (1) no hFbpABC components,
H-1443(pBR322), (2) hFbpA only (Haemophilus iron
transport protein, HitA), H-1443(pAHIDC), and (3) a
functional hFbpABC operon, H-1443(pAHIO) [17, 24].
Sterile discs were saturated with 25 lL of a 1 or 10 lM
J Biol Inorg Chem (2008) 13:887–898 889
123
gallium citrate solution and placed on the agar medium
immediately after bacterial seeding. The plates were
incubated for 24–36 h at 37 �C prior to photography
(Fig. 1). In the case of iron competition with gallium on
cellular growth medium, a disc containing 10 lM gallium
citrate was placed adjacent to a disc containing 10 lM iron
citrate and incubated under identical conditions (Fig. 1).
Determination of Ga3+ stability constant
Conditional stability constants for Ga3+ binding by FbpA in
the presence of phosphate were determined using UV-dif-
ference spectrophotometric measurements. Competitive
sequestration of Gaaq3+ by apo-FbpA and NTA in the
presence of excess phosphate was monitored spectropho-
tometrically as illustrated in Eq. 1:
Ga(NTA)2 þ apo-FbpA�PO4� GaFbpA�PO4 þ 2 NTA.
ð1Þ
NTA was used as a competing chelator with known
affinity constants for Gaaq3+ to permit evaluation of free Gaaq
3+
concentration in equilibrium with GaFbpA–PO4 [26]. In
these studies apo-FbpA was pre-equilibrated with
synergistic anion, phosphate, through overnight dialysis
at 4 �C against three 200-mL volumes of spectro-
photometry buffer (pH 6.5, 50 mM MES, 200 mM KCl,
and 5 mM KH2PO4). The pH selected was based on the
reported acidic conditions of the periplasm [27]. Two
tandem (dual compartment) quartz spectrophotometer cells
(Wilmad LabGlass, type 55, single-compartment
pathlength, b = 0.4375 cm) were used for the
UV-difference experiments. The cells were arranged in a
Cary 100 Bio UV–vis double-beam spectrophotometer
(Varian) with the compartments in series, such that the
sample and reference beam passed through two
compartments of a single cell. In each cell, 1.25 mL of
apo-FbpA was added to one compartment identified as PS
or PR (protein in sample beam line and reference beam line,
respectively). The total apo-FbpA present was determined
previously using e278 nm = 45,990 M-1 cm-1 for the
measured absorbance at 278 nm [19]. Likewise, into the
buffer compartments of each cell (sample and reference),
BS and BR, 1.25 mL of spectrophotometry buffer was
added. A baseline spectrum was recorded prior to
beginning the metal additions. Aliquots of 7.5 mM 1:2
gallium to NTA (pH * 5.5) were added to PS and BR, and
an equal volume of ultrapure water was added to PR to
correct for dilution. Spectra were recorded at 25 �C from
650 to 240 nm after each addition. The solution was
allowed to equilibrate for 90–160 min after each addition,
to reach spectral stability. The De of GaFbpA–PO4 at
equilibrium was calculated from DA246 nm (Fig. 2),
pathlength b = 0.4375 cm, and the concentration of total
FbpA present (Eq. 2):
Fig. 1 Haemophilus influenzae free iron ATP-binding cassette
transporter (hFbpABC) mediated gallium toxicity disc diffusion
assay. Fresh transformants grown to mid log phase in NBAamp100
were seeded onto nutrient broth agar supplemented with
100 lg mL-1 ampicillin containing 100 lM of the iron chelator
2,20-dipyridyl. a–c Sterile discs (white dots) containing 1 lM (left) or
10 lM (right) gallium citrate were applied to the plates and allowed
to incubate overnight. Plates were digitally scanned for presentation.
a Escherichia coli strain H-1443(pBR322), vector only, no H.influenzae ferric binding protein (hFbpA): no periplasm to cytosol
transporter and no gallium toxicity realized; b E. coli strain H-
1443(pAHIDC), hFbpA: minimal zone of toxicity around 10 lM disc
related to gallium binding by ferric binding protein (FbpA) and taken
up by low-affinity iron transport mechanisms; c E. coli strain
H-1443(pAHIO), hFbpABC: large zone of toxicity observed around
10 lM disc with functional periplasmic binding protein and ABC
transporter; and d sterile disc containing 10 lM gallium citrate placed
2 cm from the disc containing 10 lM iron citrate, H-1443(pAHIO),
hFbpABC: the zone with no growth indicates gallium toxicity around
the disc with 10 lM gallium citrate which is rescued (outlined in red)
by 10 lM iron citrate
890 J Biol Inorg Chem (2008) 13:887–898
123
De ¼ DA
½FbpA�T � b: ð2Þ
The equilibrium concentration of Gaaq3+ and the effective
binding constant K 0eff
� �for the formation of GaFbpA–PO4
from Gaaq3+ and apo-FbpA–PO4 at pH 6.5 according to
Eq. 3 were derived by variation of previously described
methods as described in Table S1 [18, 19]:
Ga3þaq þ apo-FbpA�PO4�GaFbpA�PO4: ð3Þ
The value for K 0eff was calculated using known
equilibrium constants and mass-balance equations
involving the usual Ringbom coefficients for the Ga3+/
NTA system [26, 28]. As a test of this method, K 0eff for
FeFbpA–PO4 by using Fe(NTA) was found to be in good
agreement with values determined previously [18, 19]. The
details of the model and use of Ringbom coefficients are
described in the electronic supplementary material, along
with an internal calibration and check of the model using
data for Gaaq3+ sequestration by FbpA at pH 7.3 [9].
Stability of unpurified proteins from rates of H/D
exchange analysis
Stability of unpurified proteins from rates of H/D exchange
(SUPREX) data were collected according to established
protocols [29]. SUPREX analyses were initiated by the
tenfold dilution of 1-lL aliquots of the appropriate FbpA
UV-difference reaction mixtures into a series of deuterated
H/D exchange buffers. The deuterated H/D exchange buf-
fers contained 50 mM MES, 200 mM KCl (pD 6.5) and
concentrations of deuterated GdmCl that ranged from 0 to
6 M. The GdmCl concentration in each buffer was deter-
mined with a Bausch & Lomb refractometer as previously
described [30]. Measurements of pH were performed with
a Jenco 6072 pH meter equipped with a Futura calomel
electrode from Beckmann Instruments. To correct for iso-
tope effects, the measured pH of each D2O solution was
converted to pD by adding 0.4 to the measured pH value
[31].
After dilution of the FbpA complex into the series of H/
D exchange buffers, the resulting solutions were incubated
at 25 �C, and allowed to exchange for 1 h. After 1 h, a
1-lL aliquot of the H/D exchange reaction in each protein-
containing exchange buffer was quenched with the addition
of trifluoroacetic acid (final concentration 0.3% v/v), and
the protein samples in each exchange buffer were subjected
to a concentration and desalting step using C4 ZipTipsTM
(Millipore) prior to matrix-assisted laser desorption/ioni-
zation (MALDI) mass spectral analysis. The concentration
and desalting step was performed as previously described
[32].
MALDI mass spectra were acquired using a Bruker
Ultraflex II mass spectrometer. Spectra were collected in
the linear mode in conjunction with a smartbeam Nd:YAG
(355 nm) laser. Sinapinic acid was used as the matrix in all
experiments. Positive ion mass spectra were collected
manually using the following parameters: 25-kV acceler-
ation voltage, 23.5-kV grid voltage, 75-V guide wire
voltage, and 225-ns delay time. Each mass spectrum was
the sum of 32 laser shots. Raw MALDI spectra were
treated by either of two methods:
1. Data were processed with a Microsoft Excel macro that
performed the following operations: a 19-point floating
average smoothing of the data, a two-point mass
calibration using the protein ion signals of the internal
calibrants (aldolase, trypsin inhibitor), and a center-of-
mass determination for the protein’s [M + H]+ peak.
2. Data were calibrated using the software supplied by
the instrument manufacturer.
Ten replicate mass spectra were collected for the protein
samples taken from each H/D exchange in order to
determine an average molecular mass of the deuterated
r0 2 4 6 8 10 32
∆A
bsor
banc
e
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
-0.005
0.005
0.015
0.025
0.035
0.045
0.055
0.065
240 290 340 390
Wavelength (nm)
∆ A
bs
2
1
3
4
5
6
7
2
1
3
4
5
6
7
25
Fig. 2 UV-difference spectra for Ga(NTA)2 (pH * 5.5), where NTA
is nitrilotriacetic acid, additions to apo-FbpA in 50 mM 4-morphol-
ineethanesulfonic acid (MES), 200 mM KCl, 5 mM KH2PO4, at pH
6.5. A baseline was recorded against apo-FbpA prior to additions of
stock Ga3+ solution, 7.5 9 10-3 M Ga(NTA)2. FbpA concentration
7.2 9 10-6 M, V = 1.25 mL, r = [Ga]T/[FbpA]T: 1 0.0, 2 0.4, 3 0.8,
4 1.7, 5 3.3, 6 5.7, 7 32.0. Total and Gaaq3+ concentrations are defined
in Table S1. Inset: Titration curve obtained from difference spectra at
246 nm. Error bars signify the standard deviation of three indepen-
dent experimental determinations
J Biol Inorg Chem (2008) 13:887–898 891
123
protein at each denaturant concentration. Change in mass
values were determined by subtracting the molecular mass
of the fully protonated FbpA protein (i.e., 33,598 Da) from
the average molecular mass of the protein ascertained in
the MALDI analyses. These values were used to generate
FbpA SUPREX curves (i.e., plots of change in mass versus
GdmCl concentration). The plots were fit to a four-
parameter sigmoidal equation using Sigma PlotTM to
extract a CSUPREX1/2 value (the denaturant concentration at
the transition midpoint).
Stopped-flow spectrophotometric kinetic measurements
All kinetics experiments were performed in a buffered
solution of 50 mM MES, 150 mM NaClO4, 5 mM
Na2HPO4 at pH 6.5, with temperature regulated at
25 ± 0.1 �C, under pseudo-first-order conditions with
Fe(NTA) in at least tenfold excess of protein concentration
(40 lM FbpA). The reaction of GaFbpA–PO4 with iron
was studied using an Applied Photophysics stopped-flow
(SX.18MV) spectrophotometer, capable of monitoring in
absorbance and fluorescence modes. Reactions monitored
in rapid-scan absorbance mode utilized a photodiode-array
detector, recording from 375 to 725 nm. The ligand-to-
metal charge transfer band of FeFbpA–PO4 (kmax =
481 nm) was used as the visible absorbance spectral han-
dle, while FbpA binding site tyrosine fluorescence
quenching was studied using excitation at 280 nm. Both
single-wavelength techniques acquired data using a stan-
dard photomultiplier tube detector. Data were recorded for
1,000 s (approximately 17 min) for each run, using a log
time base collection method, providing proportionally
more data at early times, enabling the simultaneous study
of reactions with broadly varying half-lives. The kinetic
results represent an average of at least three independent
kinetic mixing experiments for each set of conditions. Data
were analyzed using Origin 7.5 software, fitting the fluo-
rescence data, single-wavelength data (481 nm), and rapid-
scan data (450, 465, 480, and 550 nm) to appropriate
degree exponential functions, as determined by best fit.
Results
Gallium inhibition of bacterial growth
In various forms, we have demonstrated recombinant
E. coli expressing the FbpABC operon to be remarkably
susceptible to gallium exposure. Figure 1 demonstrates this
sensitivity using a disc diffusion assay in which different
plasmid constructs expressing (1) no FbpA encoded
information (Fig. 1a), (2) FbpA only (Fig. 1b), and (3) the
FbpABC operon (Fig. 1c) were transformed into E. coli
strain H-1443. These isogenic constructs were grown on
the iron-limited media described in ‘‘Materials and meth-
ods.’’ The results show a notable concentration-dependent
concentric zone of inhibition around the gallium citrate
disc in only the FbpABC construct (Fig. 1c), whereas no
notable inhibition was exhibited in the construct with no
FbpA (Fig. 1a) or there was minimal inhibition around the
construct with FbpA only (Fig. 1b).
This experimental approach was extended to Fe3+/Ga3+
competition. As demonstrated in Fig. 1d, the concentric
zone of inhibition around the gallium citrate disc was
rescued by the diffusion of iron citrate (as indicated by the
bracketed areas). Taken together, these results are consis-
tent with our findings that Ga3+ can act as a competitor for
iron transport, but a substantially greater excess of gallium
is required to inhibit iron stimulation [21].
The delivery of iron from FbpA to the cytoplasmic
membrane permease and resultant transport across the
membrane into the cytosol may incorporate a mechanism
where reduction does or does not play a role. If the release
of iron is a redox-mediated event, then non-redox-active
iron analogs such as Ga3+ will not enter the cytoplasm via
the same mechanism. While the cellular toxicity results
were not unexpected, they are incongruent with the
hypothesis that a reductive process is central to the removal
of iron from FbpA to the FbpBC ABC permease complex.
Implicit in this biological paradox is the assumption of
similar Fe3+–FbpA and Ga3+–FbpA binding events. To
investigate this assumption, a rigorous evaluation of the
GaFbpA–PO4 interaction, similar to that previously
described for Fe3+ [18, 19], was initiated.
Conditional binding constant K 0eff
� �for GaFbpA–PO4
The thermodynamic stability of the GaFbpA–PO4 com-
plex was determined through the titration of a 1:2
gallium to NTA solution into apo-FbpA–PO4 at pH 6.50
in 50 mM MES, 200 mM KCl, 5 mM KH2PO4 as
illustrated in Eq. 1. The reaction was monitored using
UV-difference spectroscopy, recording the spectral
change of apo-FbpA–PO4 for a series of gallium addi-
tions. The UV-difference spectrum of apo-FbpA exhibits
two positive bands at approximately 246 and 295 nm
upon binding Ga3+. The increasing absorbance of these
bands is attributed to metal coordination to tyrosine with
concomitant proton loss, as reported for gallium-loaded
Tf, lactoferrin (Lf), and FbpA complexes [9, 33, 34].
Typical UV-difference spectra exhibiting binding of Ga3+
(using 1:2 gallium to NTA) to apo-FbpA–PO4 are shown
in Fig. 2.
As described in ‘‘Materials and methods’’ and the
electronic supplementary material, the conditional stability
constant for Ga3+ binding to apo-FbpA–PO4 at pH 6.5 was
892 J Biol Inorg Chem (2008) 13:887–898
123
calculated according to Eq. 3 to be log K 0eff ¼ 13:7� 0:6
in 50 mM MES, 200 mM KCl, 5 mM KH2PO4 (Table S1).
Although an equilibrium constant was not reported by Guo
et al. [9] for pH 7.3, it was implied that binding of Ga3+ to
apo-FbpA was weak relative to iron, especially in com-
parison with Ga–Tf or Ga–Lf binding [9, 33, 34]. Using
data reported by Guo et al. [9], we estimate that at pH 7.3
log K 0eff ¼ 15� 1: When comparing the log K value
determined in this work at pH 6.5 with that calculated from
the data reported by Guo et al. at pH 7.3, we find that the
difference is within a reasonable approximation, owing to a
predicted dependence on H+ concentration based on
similarities between FbpA and Tf [35].
SUPREX analysis
The SUPREX behavior of GaFbpA–PO4 was studied at pD
6.5 to simulate the estimated conditions in the periplasm.
Shown in Fig. 3 are the SUPREX data recorded for
GaFbpA–PO4 using a 1-h H/D exchange time. SUPREX
curves were also recorded for apo-FbpA and FeFbpA–PO4
samples using a 1-h H/D exchange time (Fig. 3). The Fe-
FbpA–PO4 SUPREX data shown in Fig. 3 are consistent
with previously described SUPREX behavior for FeFbpA–
PO4 [36]. SUPREX curves at different H/D exchange times
ranging up to 3 h (data not shown) were also recorded and
were quite similar to those shown in Fig. 3. Contrary to
previous observations for FeFbpA–PO4, the CSUPREX1/2 value
for GaFbpA–PO4 does not change with the H/D exchange
time [36].
Kinetics of GaFbpA–PO4 reacted with Fe(NTA)
Time-dependent spectral changes for the reaction of
Fe(NTA) with apo-FbpA (Eq. 4) and GaFbpA–PO4 (Eq. 5)
were monitored by stopped-flow absorbance spectropho-
tometry. The absorbance increase in the 450–550-nm
region of the spectrum was used to monitor iron binding to
FbpA (Fig. 4) for Eqs. 4 and 5 at 25 �C, pH 6.5, 50 mM
MES, 150 mM NaClO4, 5 mM Na2HPO4:
Fe(NTA)þ apo-FbpA�PO4! FeFbpA�PO4þNTA ð4ÞFe(NTA)þ GaFbpA�PO4 ! FeFbpA�PO4 þ Ga(NTA):
ð5Þ
Owing to the relatively weak GaFbpA–PO4 binding
under these defined conditions, Eq. 4 must be considered as
a parallel path with Eq. 5 to form FeFbpA–PO4. The
relative amounts of GaFbpA–PO4 and apo-FbpA in Eqs. 4
and 5 were controlled by varying the relative gallium
concentration. Rate constants obtained in absorbance and
in fluorescence mode as a function of reactant
concentration are listed in Table S2. For reactions studied
in absorbance mode, the change in absorbance versus time
data were fit to exponential functions of varying degree,
yielding four distinct rate constants (Table S2). In each
reaction, data from the first 200 ms were fit to a single-
exponential curve, while data from 0.2 to 1,000 s were fit
to a triple-exponential function. An exponential fit was
applied to the fluorescence data, plotting change in signal
as a function of time, for which a double-exponential
Fig. 3 Stability of unpurified proteins from rates of H/D exchange
(SUPREX) curve for apo-FbpA–PO4 equilibrated with Ga(NO3)3, or
apo-FbpA–PO4 equilibrated with Fe(NTA)2, or apo-FbpA–PO4 alone
in 50 mM MES, 200 mM KCl, 5 mM KH2PO4 pH 6.5. Dilute sample
protocol was accomplished using C4 ZipTips; exchange time 1 h;
exchange buffer 50 mM MES, 200 mM KCl, pD 6.5; 1.5 9 10-6 M
FbpA. Circles FeFbpA–PO4, squares GaFbpA–PO4, trianglesapo-FbpA
400 450 500 550 600 650 700-0.04
0.00
0.04
0.08
0.12
∆ A
bso
rban
ce
Wavelength (nm)
0 2 400 600 800 10000.00
0.04
0.08
0.12
∆A
bsor
b anc
e
Time (s)200
Fig. 4 Representative change in absorbance data for the addition of
Fe(NTA) to GaFbpA–PO4 from 0 to 1,000 s. Inset: Change in
absorbance at 481 nm as a function of time. Data shown are the
average of three trials and are baseline-corrected for Fe(NTA)
absorbance. The conditions were as follows: 2.0 mM Fe(N-
TA), 0.04 mM FbpA, 0.12 mM gallium, in 50 mM MES, 150 mM
NaClO4, 5 mM Na2HPO4, pH 6.5, 25 �C
J Biol Inorg Chem (2008) 13:887–898 893
123
function yielded the best fit for data over 0–1,000 s at these
defined conditions. These two rate constants are identical to
those observed for the second and third rate constants
obtained from the absorbance data (Table S2).
An examination of the kinetic data for Eq. 5 provides
mechanistic information for the metal exchange process;
i.e., the ease by which Fe3+ in the form of Fe(NTA) can
displace Ga3+ from GaFbpA–PO4. Overall, Eq. 5 occurs in
four kinetically distinguishable steps. The first step occurs
with an observed first-order rate constant (k1obs) indepen-
dent of gallium concentration, with a half-life of
approximately 35 ms. At the individual conditions of each
reaction, the range of gallium concentrations provides a
significant variation in apo-FbpA concentration when
considering the relatively weak binding of aqueous gallium
to FbpA (Eq. 1, log Kd = 6.1) calculated using the method
described in the electronic supplementary material. The
proposed mechanism of apo-FbpA iron loading occurs by
an initial rapid uptake of iron by apo-FbpA within the first
200 ms of the reaction (Eq. 6) [37]. For each set of reaction
conditions presented here, k1obs is consistent with this
explanation of iron binding. When further examining the
single-exponential fit of these data (Eq. 7), one finds that
the change in FeFbpA concentration is proportional to the
pre-exponential amplitude (a) of the function. Assuming
that rapid iron binding by FbpA in this step is dependent on
the effective concentration of apo-FbpA (Table S3), we
expect a linear trend between the change in apo-FbpA
concentration and amplitude (Fig. 5).
Fe(NTA)þ apo-FbpA�PO4 ! Fe(NTA)(FbpA�PO4Þ k1
ð6Þ
�D[apo-FbpA] ¼ D[Fe(NTA)(FbpA�PO4Þ� / DA
¼ a e�kt þ b: ð7Þ
It should be noted that the identical reaction monitored
by fluorescence did not exhibit the tyrosine fluorescence
quenching, which would be expected with this rapid first
step. However, this assignment of the rapid iron binding by
apo-FbpA is consistent with the relatively low available
concentration of apo-FbpA utilized in the fluorescence
experiments.
The full mechanistic interpretation of the kinetic data is
presented in the electronic supplementary material, and
parallels the previously described mechanism of iron
loading into apo-FbpA [37]. With this comparable mech-
anistic interpretation and the dependence on the relative
concentration of apo-FbpA, it is useful to consider the
overall reactions observed here occurring in two parallel
processes (Scheme 1).
The pathway on the right side of the scheme follows the
previously reported mechanism for Fe3+ loading by reac-
tion of Fe(NTA) with apo-FbpA (Eq. 4) [37], while the
pathway on left side follows steps attributed to the Ga3+-
loaded protein (Eq. 5). In developing the gallium-depen-
dent metal exchange mechanism, we find it is also useful to
consider that the apo-FbpA iron loading process is occur-
ring simultaneously. Because of the relative abundance of
initial apo-FbpA-PO4 and GaFbpA–PO4 species, it is rea-
sonable that the initial fast step of apo-FbpA is observable
(Eq. 4), but the subsequent steps are masked by the left
pathway involving GaFbpA–PO4 (Eq. 5), which results in
four kinetically distinguishable rate constants.
0 10 20 30 40
-0.004
-0.008
-0.012
-0.016
Am
plitu
de 1
(ab
sorb
ance
units
)
Calculated [apo-FbpA] (µM)
Fig. 5 Amplitude for exponential fit of k1 (Eqs. 6, 7) as a function of
calculated apo-FbpA concentration (Eq. 1; details in the electronic
supplementary material). Error bars indicate standard error at each
point. The conditions were as follows: 2.0 mM Fe(NTA), 0.04 mM
FbpA, 0.12–0.24 mM gallium, in 50 mM MES, 150 mM NaClO4,
5 mM Na2HPO4, pH 6.5, 25 �C
+ Fe(NTA)
GaFbpA-PO4
(NTA)FeFbpA-Ga(PO4)
FeFbpA-(NTA)Ga(PO4)
Ga3+ + FbpA-PO4
Fe(NTA)(FbpA-PO4)
(NTA)FeFbpA-(PO4)
FeFbpA-NTA(PO4)
FeFbpA-PO4
k2 Ga
k3 Ga
k4 Ga
k1 apo
k2 apo
k3 apo
k4 apo
(8 s)
(30 s)
(0.03 s)
(1 s)
(30 s)
(250 s) (5000 s)
STEP I
STEP II
STEP III
STEP IV
STEP II
STEP III
STEP IV
Fe(NTA) +
effK′ T
Scheme 1 Gallium competition in iron binding pathway of ferric
binding protein (FbpA). Values in parentheses represent the half-lives
of each kinetic step at 2 mM Fe(NTA), where NTA is nitrilotriacetic
acid, 5 mM PO43-, 40 lM FbpA. (Half-lives listed for the vertical
sequence on the right were taken from Gabricevic et al. [37])
894 J Biol Inorg Chem (2008) 13:887–898
123
Discussion
Bacteriologic implications
Gallium was used as a probe to compete for iron uptake
in a model system of FbpABC iron transport [17, 21].
The inherent rationale that underlies this approach is the
use of gallium as an atomic homolog of iron. A compa-
rable approach targeting the chemical similarities between
iron and gallium has been demonstrated for P. aeruginosa
[11]. In these studies, it is reported that low concentra-
tions of gallium inhibit P. aeruginosa growth and prevent
biofilm formation, while high levels of gallium kill
planktonic bacteria and decrease the bacterial load of
established biofilms. Further, a P. aeruginosa lung
infection model demonstrated corresponding effects in
mice, for which this susceptibility to gallium exposure
may be explained by the expression of a functional
FbpABC type transporter [7]. Both the model studies of
FbpABC iron uptake [7] and the study of P. aeruginosa
[11] rely on qualitative inhibition of in vitro and in vivo
bacterial growth. Lacking is a precise quantitative anal-
ysis of iron-versus-gallium binding and release in
periplasm to cytosol metal transport. To address this
deficiency in the literature, we determined the thermo-
dynamic binding constant for GaFbpA–PO4, with a direct
comparison to iron binding under comparable conditions.
The indication that FbpA selectively binds Fe3+ by sev-
eral orders of magnitude in stability over Ga3+, and that
Fe3+ is kinetically competent in displacing Ga3+ explains
the rescue in growth of gallium-stressed FbpABC trans-
formants (Fig. 1). The ability to sequester iron over
gallium imparts increased survival rates and reduces the
effects of gallium toxicity.
Conditional binding constant for GaFbpA–PO4
The reported binding affinity of Feaq3+ for apo-FbpA–PO4
log K 0eff ¼ 18:6� �
[19] suggests a 105-fold greater
selectivity for Fe3+ over Ga3+ log K 0eff ¼ 13:7� �
by FbpA at
pH 6.5. This difference is approximately 103 times greater
than that determined using similar relationships for Tf and
FeTfHCO3 log K 01 ¼ 21:70� �
and GaTfHCO3 log K 01�
¼19:71Þ [34]. This finding supports the previous assessment
that GaFbpA binding is weaker than either GaTf or
GaLf binding [9]. Previously reported conditional Fe3+
binding constants for FbpA in the presence of different
anions (sulfate \ oxalate * citrate \ NTA \ pyrophos-
phate \ arsenate \ phosphate) range from log K 0eff ¼ 16:2
(FeFbpA–SO4) to log K 0eff ¼ 18:6 (FeFbpA–PO4) at pH 6.5
[18–20]. It appears that Ga3+ binding in GaFbpA–PO4 at pH
6.5 is at least 2–3 orders of magnitude less stable than even
the weakest FeFbpA–X assembly.
SUPREX analysis
The CSUPREX1/2 values generated from SUPREX analysis of
a protein or protein–ligand complex have been shown to
provide qualitative information about the thermodynamic
stability of this complex [36]. Proteins and protein–
ligand complexes with high thermodynamic stability have
large CSUPREX1/2 values, whereas less thermodynamically
stable proteins and protein–ligand complexes have small
CSUPREX1/2 values. The SUPREX data collected for GaF-
bpA–PO4 and apo-FbpA do not have a well-defined
pretransition baseline; therefore, it is difficult to precisely
determine their CSUPREX1/2 values. However, it is clear that
SUPREX curves of these preparations are similar, and
that the CSUPREX1/2 values of the GaFbpA–PO4 and apo-
FbpA curves are shifted to a significantly lower dena-
turant concentration relative to the CSUPREX1/2 value of the
FeFbpA–PO4 curve (Fig. 3) at the same H/D exchange
time. This shift towards a lower denaturant concentration
indicates that the FeFbpA–PO4 complex is significantly
more stable than the GaFbpA–PO4 complex [36]. Fur-
thermore, the SUPREX data suggest that the
thermodynamics of Fe3+ binding to FbpA is more
favorable than that of Ga3+ binding to FbpA, which is
consistent with our equilibrium spectrophotometric titra-
tion analysis.
Previously, CSUPREX1/2 values have been shown to quantify
protein–ligand binding affinities [38]; however, this analysis
requires that the protein folding reaction under study be
highly cooperative so that the amide H/D exchange reac-
tion is dominated by a global exchange mechanism (i.e.,
amide H/D exchange occurs when the protein globally
unfolds). Under these conditions, the CSUPREX1/2 values of a
protein are predicted to move to lower denaturant concen-
trations when the H/D exchange time in the SUPREX
experiment is increased. Such behavior has previously been
noted for FeFbpA–PO4 [36]. Interestingly, such SUPREX
behavior was not observed with GaFbpA–PO4 or apo-FbpA–
PO4. Presumably the amide H/D exchange reactions for
these complexes in this study are dominated by more local
unfolding mechanisms. Thus, our SUPREX results for
GaFbpA–PO4 and apo-FbpA–PO4 suggest that these
assemblies are conformationally more flexible than
FeFbpA–PO4.
Collectively, these data demonstrate that GaFbpA–PO4
and FeFbpA–PO4 behave differently, and although Ga3+ is
used as a physiological analog, it is not a perfect probe for
Fe3+. Our data demonstrate that Ga3+ egress from GaF-
bpA–PO4 is more facile than Fe3+ egress from FeFbpA–
PO4. This conclusion is based on Kd values (KdGa [ Kd
Fe),
and SUPREX analysis indicating that GaFbpA–PO4
unfolding is more facile, which suggests an easier desta-
bilization/metal release on docking at the FbpB receptor.
J Biol Inorg Chem (2008) 13:887–898 895
123
Thus, when compared with Fe3+, Ga3+ exchange from
FbpA is more likely, which is further supported by our
kinetic data.
Kinetics of exchange from Ga/FeFbpA–PO4
Our kinetic data for Ga3+ exchange with Fe3+ in FbpA
indicate that Ga3+ loss from FbpA is more facile than Fe3+
loss. Harris and Messori [16] report that once the Tf has
adopted an open conformation, Ga3+ loss is more rapid
than Fe3+ loss. This is consistent with our SUPREX data
for FbpA, which demonstrate that GaFbpA–PO4 has a more
open conformation than FeFbpA–PO4. Without structural
studies of GaFbpA, it is difficult to ascertain if changes in
the orientations of certain residues play a role in metal
binding differences.
The reactions in Eqs. 4 and 5 occur simultaneously as
parallel paths in four kinetically distinguishable steps. The
first step occurs with a half-life of approximately 35 ms,
having an observed first order rate constant (k1obs) that is
independent of Ga3+ concentration. A decrease in rate
constant is correlated with increasing gallium concentra-
tion in step II (Scheme 1). By comparison with the
previously described iron loading of apo-FbpA, the
observed second step is a factor of 8 (at comparable con-
ditions; Scheme 1) slower in the presence of Ga3+, with the
observed rate constant leveling at approximately 0.06 s-1
(Fig. S1), compared with that for Fe3+ reaction with the
apo-FbpA, which displays saturation behavior and levels at
approximately 0.75 s-1 [37]. On the basis of this decrease
and the gallium concentration dependence, it is reasonable
to attribute this second step to the dissociation of Ga3+ from
the FbpA binding site, followed by the rapid binding of
Fe3+. Our results are consistent with the mechanism pro-
posed in Scheme 1 and attributable to a rearrangement of
the residues in the binding site in step III. The final step in
the metal exchange process is consistent with the final step
(step IV) of the apo-FbpA iron loading mechanism, pre-
viously assigned to an anion exchange process [37]. While
comparable to the final step of the apo-FbpA/Fe3+ loading
mechanism, the reaction observed here is somewhat faster.
This increased rate can be explained by the presence of
Ga3+ at the active site. While the Fe(NTA) + apo-FbpA
pathway relied strictly on anion exchange to produce Fe-
FbpA–PO4, the ability of Ga3+ to bind to NTA may
facilitate the removal of NTA in exchange for phosphate.
Additional factors in protein metal selectivity
The reason for the greater selectivity of FbpA for Fe3+ over
Ga3+ relative to Tf and Lf is not clear, but the structural
differences between these proteins may play a significant
role. As Tf (and Lf by inference) closes to bind iron and
CO32-, the open binding cleft reduces to a channel [39].
The iron binding center is buried deeply within the protein
at the back of the channel, with Fe3+ and CO32- both
protected from bulk solvent. This may contribute to the
high Tf–Fe3+ binding constant. The solvent channel for
FbpA is much shallower than that of Tf [4], which may
contribute to the relative lower Fe3+ binding constant in
FbpA. This shallow binding site in FbpA may also con-
tribute to the selection of Fe3+ over Ga3+.
The coordination of Fe3+ in FbpA, like that in Tf, adopts
an irregular octahedral structure. In general, the coordina-
tion chemistry of Ga3+ is similar to that of Fe3+. Despite the
commonality of coordination geometry, likely factors of
Fe3+ selectivity over Ga3+ in Tfs have been largely attrib-
uted to metal ion acidity or strength of hydroxide anion
binding [15]. In addition, dependencies on the concentra-
tions of synergistic and nonsynergistic anions alike affect
the structure and reactivity of metal binding sites [15, 40].
Preferential binding in other metal binding protein systems
reflects differences in ligand field stabilization energies and
other thermodynamic factors, including (1) metal ion
electroaffinities, (2) enthalpies of hydration, (3) underlying
folding landscape topographies, (4) solvent reorganization,
and (5) other forms of environmental coupling [41].
Studies by Barton [8] include size, charge, and electron
orbital characteristics as important factors controlling
metal selectivity. Crystal-field stabilization is not expected
to be of any significance for Ga3+ (filled 3d10) and Fe3+
(high-spin, half-filled 3d5) complexes and should not con-
tribute to selectivity between these two metals [14, 42]. On
the basis of the relatively shallow metal binding site of
FbpA, metal selectivity may be influenced by the enthalpy
of hydration. The greater hydration enthalpy, -DHhyd� , of
Ga3+ (4,640 kJ mol-1) over Fe3+ (4,530 kJ mol-1) may
play a role in selectivity [43]. The protein in essence serves
to dehydrate the metal cation as it strips away the waters of
hydration during chelation. Although dehydrating
Ga(OH2)63+ is more difficult by approximately 110 kJ
mol-1, this difference will be offset on binding to the
protein [41]. The calculated difference in free energy for
the binding affinity of GaFbpA–PO4 and FeFbpA–PO4 is
approximately 25 kJ mol-1.
The hardness of a Lewis acid depends on the metal
ion charge to radius (z/r) ratio with metals becoming
softer as this ratio becomes smaller. Ga3+ (z/r = 4.8) and
Fe3+ (z/r = 4.6) are both hard Lewis acids which prefer
oxygen and nitrogen donors (hard Lewis bases) like those
found in FbpA [10]. However, Fe3+ has a stronger
affinity for OH- as evidenced by OH- affinity constants
(log KOH of 11.32 for Fe3+ compared with 10.91 for
Ga3+) and this likely contributes to a stronger attraction
of Fe3+ for the deprotonated tyrosines at the binding site
[15, 16]. The difference in acidities together with the
896 J Biol Inorg Chem (2008) 13:887–898
123
difference in hydration enthalpies contributes to the
ability of FbpA to better stabilize Fe3+ over Ga3+, which
in turn enables the protein to adopt a fully folded, native
state in sequestering Fe3+.
Ga3+ also has a weaker affinity for monophosphorylated
peptides than does Fe3+ [44]. This is an important effect
when considering that one of the metal binding ligands is a
phosphate anion bound to the N-terminus (N-cap) of an
adjacent a-helix through the backbone amide of Ala141,
the hydroxyl of Ser139 (both from the C-terminal domain),
the Gln58 from the N-terminal domain of the protein, and
the iron in the case of FeFbpA–PO4 [5]. Weaker binding to
the coordinating phosphate would enable more efficient
dissociation. The phosphate anion in the open form of
FbpA (apo-FbpA) is linked to the C-terminal domain only.
Once the metal binds, phosphate shifts in position and
forms a bridge across the cleft connecting the C-terminal
and N-terminal domains of the protein. This change in
connectivity along with a 20� closure of the hinge region
assists in pulling FbpA into the closed form (holo-FbpA)
[4–6]. Thus, a difference in phosphate affinity would cause
changes in local conformational stability, which in turn
would contribute to differential H/D exchange behavior.
Local conformational fluctuations would affect electro-
static-potential-dependent variations and is anticipated to
be a factor in protein H/D exchange reactions [45]. If the
protein-bound coordinating phosphate has weaker interac-
tions with Ga3+, the exchange data should yield a
thermodynamic indication of an open-state conformation,
which is supported by our SUPREX data. GaFbpA–PO4
appears to be dominated by local unfolding as indicated in
the SUPREX data discussed earlier, and this may be
interpreted as a weaker interaction between phosphate and
Ga3+. The weaker interaction between Ga3+ and FbpA is
consistent with a metal-exchange-competent conformation
that retains resemblance to the native state, and therefore
receptor recognition, providing a platform for easy handoff
from FbpA to the FbpBC binding complex in the cytosolic
membrane without a need for reduction to a 2+ oxidation
state, as is hypothesized to be the case for Fe3+ [19, 20].
Conclusion
The SUPREX data show that FbpA folds differently in
the presence of Ga3+ than in the presence of Fe3+, and in
fact GaFbpA–PO4 behaves more like apo-FbpA than Fe-
FbpA–PO4 with regard to protein folding. This
differential behavior is consistent with our determination
of the conditional thermodynamic binding constants for
GaFbpA–PO4 and FeFbpA–PO4 which shows 105 lower
binding affinity for Ga3+. In addition to this significant
thermodynamic preference of FbpA for Fe3+ over Ga3+,
exchange kinetics indicate that Fe3+ can readily replace
Ga3+ in the metal binding pocket of FbpA. The Fe3+
metallation mechanism is similar for reaction with apo-
FbpA and GaFbpA–PO4. Facile displacement of Ga3+
from FbpA due to a lower affinity than for Fe3+ suggests
that FbpA is not saturated in the absence of excess Ga3+,
leaving appreciable apo-FbpA to react with entering Fe3+.
Furthermore, weaker affinity means Ga3+–protein side
chain bonds are more readily dissociated and Ga3+ is
replaced by Fe3+. Our biophysical results reported here
are consistent with the bioactivity results reported in
Fig. 1 where FbpABC clearly mediates the incorporation
of Ga3+ into the cytosol, similar to the native substrate,
Fe3+. Weaker binding and a different protein fold enable
Ga3+ delivery across FbpBC to the cytosol, while the
more tightly bound Fe3+ presumably must undergo
reduction to Fe2+ to be released through FbpBC into the
cytosol [19, 20]. Further, our biophysical results demon-
strate a mechanism for the recovery from Ga3+
cytotoxicity in the presence of excess Fe3+ as observed in
Fig. 1d. Weaker Ga3+ binding and kinetic lability enable
Fe3+ to readily displace Ga3+ from FbpA in the periplasm,
preventing its release to the cytosol where cytotoxic
effects may occur.
In summary, our results regarding FbpA protein folding,
thermodynamic metal binding affinity, and metal dis-
placement kinetics of Ga3+ by Fe3+, suggest that Ga3+ is not
an ideal redox-silent probe for investigating the mechanism
of FbpA-mediated iron delivery to the cytosol.
Acknowledgments A.L.C. thanks the NSF (CHE 0418006) for
financial support. J.J.H. received partial support from an NIH CBTE
training grant (T32GM8555).
References
1. Crichton RR (2001) Inorganic biochemistry of iron metabolism:
from molecular mechanism to clinical consequences, 2nd edn.
Wiley, New York
2. Cornelissen CN, Sparling PF (2004) Neisseria. In: Crosa JH (ed)
Iron transport in bacteria. ASM Press, Washington DC, pp 256–
272
3. Mietzner TA, Tencza SB, Adhikari P, Vaughan KG, Nowalk AJ
(1998) In: Vogt PK, Mahan MJ (eds) Current topics microbiology
and immunology, vol 225. Springer, Berlin, pp 113–135
4. Bruns CM, Nowalk AJ, Arvai AS, McTigue MA, Vaughan KG,
Mietzner TA, McRee DE (1997) Nat Struct Biol 4(11):919–924
5. McRee DE, Bruns CM, Williams PA, Mietzner TA, Nunn R
(1999) Structural basis of iron uptake in the pathogen Neisseriagonorrhoeae. RCSB Protein Data Bank
6. Bruns CM, Anderson DA, Vaughan KG, Williams PA, Nowalk
AJ, McRee DE, Mietzner TA (2001) Biochemistry 40:15631–
15637
7. Shouldice SR, McRee DE, Dougan DR, Tari LW, Schryvers AB
(2005) J Biol Chem 280:5820–5827
8. Barton LL (2005) Structural and functional relationships in
prokayotes. Springer, New York
J Biol Inorg Chem (2008) 13:887–898 897
123
9. Guo M, Harvey I, Yang W, Coghill L, Campopiano DJ, Parkin-
son JA, MacGillivray RTA, Harris WR, Sadler PJ (2003) J Biol
Chem 278(4):2490–2502
10. Shannon RD (1976) Acta Crystallogr Sect A 32:751–767
11. Kaneko Y, Thoendel M, Olakanmi O, Britigan BE, Singh PK
(2007) J Invest Med 117:877–888
12. Olakanmi O, Stokes JB, Britigan BE (2005) J Invest Med
53:143–153
13. Chikh Z, Ha-Duong N-T, Miquel G, El Hage Chahine JM (2007)
J Biol Inorg Chem 12:90–100
14. van der Helm D (1998) In: Sigel A, Sigel H (eds) Metal ions in
biological systems, vol 35. Marcel Dekker, New York, pp 355–
401
15. Sun H, Cox MC, Hongyan L, Sadler PJ (1997) In: Hill HAO,
Sadler PJ, Thomson AJ (eds) Metal sites in proteins and models:
iron centres, vol 88. Springer, Berlin, pp 71–102
16. Harris WR, Messori L (2002) Coord Chem Rev 228:237–262
17. Anderson DS, Adhikari P, Nowalk AJ, Chen CY, Mietzner TA
(2004) J Bacteriol 186(18):6220–6229
18. Taboy CH, Vaughan KG, Mietzner TA, Aisen P, Crumbliss AL
(2001) J Biol Chem 276(4):2719–2724
19. Dhungana S, Taboy CH, Anderson DA, Vaughan KG, Aisen P,
Mietzner TA, Crumbliss AL (2003) Proc Natl Acad Sci USA
100(7):3659–3664
20. Heymann JJ, Weaver KD, Mietzner TA, Crumbliss AL (2007) J
Am Chem Soc 129:9704–9712
21. Anderson DS, Adhikari P, Weaver KD, Crumbliss AL, Mietzner
TA (2007) J Bacteriol 189(14):5130–5141
22. Mietzner TA, Bolan G, Schoolnik GK, Morse SA (1987) J Exp
Med 165(4):1041–1057
23. Bastian R, Weberling R, Palilla F (1956) Anal Chem 28(4):459–
462
24. Adhikari P, Kirby SD, Nowalk AJ, Veraldi KL, Schryvers AB,
Mietzner TA (1995) J Biol Chem 270(42):25142–25149
25. Adhikari P, Berish SA, Nowalk AJ, Veraldi KL, Morse SA,
Mietzner TA (1996) J Bacteriol 178(7):2145–2149
26. Motekaitis RJ (2001) The National Institute of Standards and
Technology (NIST) Standard Reference Database 46, version 6.0
27. Ferguson SJ (1991) In: Mohan S, Dow C, Coles JA (eds) Pro-
karyotic structure and function: a new perspective. 47th
Symposium of the society for general microbiology. Cambridge
University Press, Cambridge
28. Ringbom A (1963) Complexation in analytical chemistry: a guide
for the critical selection of analytical methods based on com-
plexation reactions. Interscience, New York
29. Ghaemmaghami S, Fitzgerald MC, Oas TG (2000) Proc Natl
Acad Sci USA 97:8296–8301
30. Nozaki Y (1972) Methods Enzymol 26:43–50
31. Glasoe P, Long FA (1960) J Phys Chem 64:188–190
32. Powell KD, Fitzgerald MC (2001) Anal Chem 73:3300–3304
33. Harris WR, Pecoraro VL (1983) Biochemistry 22:292–299
34. Harris WR (1986) Biochemistry 25:803–808
35. Gelb MH, Harris DC (1980) Arch Biochem Biophys 200:93–98
36. Roulhac PL, Powell KD, Dhungana S, Weaver KD, Mietzner TA,
Crumbliss AL, Fitzgerald MC (2004) Biochemistry 43:15767–
15774
37. Gabricevic M, Anderson DA, Mietzner TA, Crumbliss AL (2004)
Kinetics and mechanism of iron(III) complexation by ferric
binding protein: the role of phosphate. Biochemistry 43:5811–
5819
38. Powell KD, Ghaemmaghami S, Wang MZ, Ma L, Oas TG,
Fitzgerald MC (2002) J Am Chem Soc 124:10256–10257
39. MacGillivray RTA, Moore SA, Chen J, Anderson BF, Baker H,
Luo Y, Bewley M, Smith CA, Murphy MEP, Wang Y, Mason
AB, Woodworth RC, Brayer GD, Baker EN (1998) Biochemistry
37(22):7919–7928
40. Harris DC, Aisen P (1989) In: Loehr TM (ed) Physical bioinor-
ganic chemistry. VCH Publishers, New York, pp 239–351
41. Lachenmann MJ, Ladbury JE, Dong J, Huang K, Carey P, Weiss
MA (2004) Biochemistry 43:13910–13925
42. Huheey JE, Keiter EA, Keiter RL (1993) Inorganic chemistry
principles of structure and reactivity, 4th edn. Harper Collins
College Publishers, New York
43. Smith DW (1977) J Chem Educ 54(9):540–542
44. Zhou W, Merrick BA, Khaledi MG, Tomer KB (2000) J Am Soc
Mass Spectrom 11:273–282
45. LeMaster DM, Anderson JS, Hernandez G (2006) Biochemistry
45:9956–9963
898 J Biol Inorg Chem (2008) 13:887–898
123