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: alvin.crumbliss@duke.edu

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