Heme binding to the second, lower-affinity site of the
global iron regulator Irr from Rhizobium leguminosarum
promotes oligomerization
Gaye F. White
1,2
, Chloe Singleton
1
, Jonathan D. Todd
2
, Myles R. Cheesman
1
, Andrew
W. B. Johnston
2
and Nick E. Le Brun
1
1 School of Chemistry, Centre for Molecular and Structural Biochemistry, University of East Anglia, Norwich Research Park, Norwich, UK
2 School of Biological Sciences, University of East Anglia, Norwich Research Park, Norwich, UK
Introduction
Iron is essential for almost all forms of life, fulfilling
functions ranging from respiration to DNA synthesis.
The metal occurs in cells as a protein cofactor in dif-
ferent forms: as the bare metal or, more commonly, in
heme and iron-sulfur clusters. Despite its abundance in
the biosphere, iron has poor bioavailability owing to
its propensity, in the presence of oxygen and water, to
form insoluble ferric oxy–hydroxide complexes [1],
such that cells of all types employ complex mecha-
nisms to recruit it from the environment. In addition,
Keywords
a-proteobacteria; fur; heme; iron;
transcriptional regulation
Correspondence
N. E. Le Brun, School of Chemistry,
University of East Anglia, Norwich NR4 7TJ,
UK
Fax: +44 1603 592003
Tel: +44 1603 592699
E-mail: n.le-brun@uea.ac.uk
(Received 3 February 2011, revised 17
March 2011, accepted 1 April 2011)
doi:10.1111/j.1742-4658.2011.08117.x
The iron responsive regulator Irr is found in a wide range of a-proteobac-
teria, where it regulates many genes in response to the essential but toxic
metal iron. Unlike Fur, the transcriptional regulator that is used for iron
homeostasis by almost all other bacterial lineages, Irr does not sense Fe
2+
directly, but, rather, interacts with a physiologically important form
of iron, namely heme. Recent studies of Irr from the N
2
-fixing symbiont
Rhizobium leguminosarum (Irr
Rl
) showed that it binds heme with submi-
cromolar affinity at a His-Xxx-His (HxH) motif. This caused the protein to
dissociate from its cognate DNA regulatory iron control element box
sequences, thus allowing expression of its target genes under iron-replete
conditions. In the present study, we report new insights into the mecha-
nisms and consequences of heme binding to Irr. In addition to the HxH
motif, Irr binds heme at a second, lower-affinity site. Spectroscopic studies
of wild-type Irr and His variants show that His46 and probably His66 are
involved in coordinating heme in a low-spin state at this second site. By
contrast to the well-studied Irr from Bradyrhizobium japonicum, neither
heme site of Irr
Rl
stabilizes ferrous heme. Furthermore, we show that
heme-free Irr
Rl
exists as a mixture of dimeric and larger, likely hexameric,
forms and that heme binding promotes Irr
Rl
oligomerization. Bioanalytical
studies of Irr
Rl
variants showed that this property is not dependent on the
HxH motif but is associated with heme binding at the second site.
Structured digital abstract
lIrr binds to irr by molecular sieving (View Interaction 1,2)
lIrr binds to irr by cosedimentation in solution (View interaction)
Abbreviations
EPR, electron paramagnetic resonance; HRM, heme regulatory motif; ICE, iron control element.
FEBS Journal 278 (2011) 2011–2021 ª2011 The Authors Journal compilation ª2011 FEBS 2011
the very properties that make iron essential also make
it toxic in oversupply, such that the amount and form
of iron in the cell must be carefully regulated [1,2].
In bacteria, several different iron regulatory systems
are now known. The best studied are the functionally
and structurally similar, but evolutionarily unrelated
global regulators, Fur and DtxR. Under iron-sufficient
conditions, both of these bind Fe
2+
, causing a confor-
mational change that increases the affinity of the pro-
tein for operator sequences located 5¢of the regulated
genes, resulting (usually) in transcriptional repression
[3,4]. When iron is scarce, Fe
2+
dissociates, releasing
the protein from DNA, thereby switching on genes
involved in, for example, iron recruitment and uptake.
The subphylum a-proteobacteria contains important
pathogens of animals (e.g. Brucella,Rickettsia) and
plants (Agrobacterium), symbionts (the N
2
-fixing rhizo-
bia), many of the most abundant bacteria in the
oceans (the Roseobacters and the SAR11 clade, includ-
ing Pelagibacter) and some laboratory model organ-
isms (e.g. Paracoccus,Rhodobacter), as well as being
the source of mitochondria. The relatively few studies
on iron-responsive gene regulation in members of this
important group have shown that these bacteria are
very different from those that use the Fur regulator.
Although many a-proteobacteria contain a Fur homo-
logue [5], in those cases where this gene product was
studied directly, it was shown to have a minor regula-
tory role, repressing a few genes in response to manga-
nese (and not iron) availability; hence, it was renamed
‘Mur’ for mangenese uptake regulator [6,7].
Instead, many a-proteobacteria contain another
transcriptional repressor called Irr (iron responsive reg-
ulator) which, although a member of the Fur super-
family, has important features that distinguish it from
Fur sensu stricto (see below). Irr functions to repress a
wide range of genes under low iron availability by
binding to cis-acting regulatory sequences known
as iron control element (ICE) boxes that are 5¢of the
target genes. It was first discovered in Bradyrhizobi-
um japonicum [8–12] and was also studied in Rhizo-
bium leguminosarum, where it represses a wide range of
genes in cells grown under iron-depleted conditions
[13,14], and in Brucella abortus, where it regulates
siderophore biosynthesis [15]. In addition to Irr, a-pro-
teobacteria that are closely related to Rhizobium,Agro-
bacterium,Brucella and Bartonella contain another
wide-ranging iron-responsive regulator called RirA
[16–19]. RirA belongs to the Rrf2 family of regulators,
which includes IscR and NsrR [20,21] and, similar to
them, is a predicted FeS cluster-binding protein. RirA
represses many genes under iron-replete conditions,
recognizing the cis-acting iron regulatory sequences
(RirA-boxes) that precede genes similar to those that
are commonly regulated by Fur in other organisms
[5,16–19]. Thus, RirA and Irr are ‘opposing’ regula-
tors, which repress different portfolios of target genes
in cells grown in sufficient (RirA) or deficient (Irr) lev-
els of iron. Because these two regulators respond to
the availability of iron in the form of iron-sulfur clus-
ters and heme, respectively, and exhibit regulatory
‘cross-talk’ in response to iron availability [14], they
may represent a combined regulatory system that
senses the physiologically relevant status of iron and
not just the concentration of the metal per se, as is the
case in those organisms that use Fur or DtxR [13].
Few Irr proteins have been studied in detail so far.
The first, from B. japonicum (Irr
Bj
), exhibits a highly
unusual regulatory mechanism. Under iron-replete
conditions, Irr
Bj
interacts with heme at two known
sites: an N-terminal heme regulatory motif (HRM)
and an internal, histidine-rich (HxH) motif [12], with
the heme being normally delivered to Irr
Bj
by ferroch-
elatase [22]. The heme-Irr
Bj
complex is extremely
unstable and is rapidly degraded. The mechanism by
which this remarkable response occurs is not known
but may involve heme- and oxygen-mediated oxidative
damage that acts as signal(s) for protease-mediated
degradation [23]. The end result is that, under iron-
replete conditions, Irr
Bj
is unavailable to act as a
repressor.
Another member of the rhizobia, R. leguminosarum,
which forms nodules on peas, clovers and beans, has
an Irr (Irr
Rl
) that is 57% identical to Irr
Bj
but lacks
the N-terminal heme regulatory motif. We recently
reported that Irr
Rl
functions by a very different mecha-
nism that does not involve degradation in response to
elevated iron. Instead of being degraded, the interac-
tion between Irr
Rl
and heme causes an allosteric
change in the protein, which prevents it from binding
to its cognate ICE box DNA sequences [24]. In vitro
studies showed that specific binding to DNA was abol-
ished by the direct addition of heme, and spectroscopic
studies of heme binding to wild-type Irr
Rl
and mutant
forms of the protein revealed two low-spin ferric
heme-binding sites. Substitution of the HxH motif
abolished one of these sites, with a concomitant loss of
DNA binding in vitro and regulatory function in vivo,
demonstrating a key role of this motif [24]. The prop-
erties and location of the second, lower-affinity heme-
binding site of Irr
Rl
have not been fully investigated,
although electron paramagnetic resonance (EPR) and
magnetic CD spectroscopic studies showed that the
heme bound at the second heme site is coordinated by
two His residues [24]. In the present study, we used
spectroscopic and bioanalytical methods to gain
Heme binding to Irr promotes oligomerization G. F. White et al.
2012 FEBS Journal 278 (2011) 2011–2021 ª2011 The Authors Journal compilation ª2011 FEBS
further insight into heme binding by R. leguminosarum
Irr, with a focus on the second heme-binding site.
These studies highlight important new features of Irr
that have relevance to Irr proteins from a wide range
of a-proteobacteria.
Results
Identification of the second heme-binding site of
Irr
Rl
: importance of His46 and His66
It was previously shown that the second heme-binding
site of Irr
Rl
is associated with low-spin S= ½ heme
with g-values of 2.95, 2.26 and 1.55, and a near-infra-
red magnetic CD band centred at approximately
1580 nm [24], demonstrating that the second heme iron
has bis-His ligation [25]. However, these previous stud-
ies did not establish which His residues are involved.
In total, Irr
Rl
has seven His residues (Fig. 1A), three
of which are at the HxH motif (His93, His94 and
His95). One or more of the remaining His residues
(His39, His46, His66 and His128) must therefore
supply the ligands for the second heme site. A set of
variants, in which each of these His residues was
substituted with alanine, had been previously generated
and their absolute heme spectra measured [24].
Although the spectra were similar in form to the wild-
type Irr
Rl
spectrum, indicating no change in the ratio
of high- to low-spin bound heme, the amounts of heme
bound differed according to which histidine had been
substituted. Thus, variants H39A and H128A were
essentially identical to the wild-type Irr
Rl
protein,
whereas H46A and H66A exhibited spectra with signif-
icantly lower heme intensity [24]. Consistent with this,
hemochromogen analyses revealed that, although
H39A and H128A Irr
Rl
proteins bound only margin-
ally less heme than wild-type (1.4 compared to
1.5 heme per protein), H66A (1.1 heme per pro-
tein) and H46A (0.9 heme per protein) exhibited sig-
nificantly lower heme binding (not shown).
EPR spectra of each of the single His variants were
recorded. Previous studies showed that the heme that
binds at the HxH motif is EPR silent, very likely result-
ing from the magnetic coupling of HxH hemes within a
BA
DC
Fig. 1. Spectroscopic studies of Irr
Rl
His variants. (A) Amino acid residue sequence of Irr
Rl
, with His residues highlighted. (B) EPR spectra of
wild-type (black), H39A (green), H46A (blue), H66A (purple) and H128A (orange) Irr variants. (C) UV-visible absorbance and (D) EPR spectra
of H93 ⁄94 ⁄95A (red), His39 ⁄93 ⁄94 ⁄95Ala (olive green), His39 ⁄46 ⁄93 ⁄94 ⁄95Ala (dark cyan), His39 ⁄46 ⁄66 ⁄93 ⁄94 ⁄95Ala (magenta) and His-
free (grey) Irr variants. Note that UV-visible and EPR spectra of wild-type and H93 ⁄94 ⁄95A Irr were reported previously [24] and are included
here for reference. Excess heme was added to each protein and non- or weakly-bound protein was removed by gel filtration and the protein
concentrated as necessary. Proteins (10 lMfor absorbance, 100 lMfor EPR) were in 50 mMTris-HCl, 50 mMKCl (pH 8). EPR measurement
conditions were: temperature, 10 K; microwave power, 2 mW; modulation amplitude, 10 G.
G. F. White et al. Heme binding to Irr promotes oligomerization
FEBS Journal 278 (2011) 2011–2021 ª2011 The Authors Journal compilation ª2011 FEBS 2013
proposed Irr dimer [24]. Therefore, the low-spin heme
signals observed in the EPR spectrum are the result of a
heme bound at the second heme site. We noted that the
low-spin heme signal intensity was decreased in both
the H46A and H66A variants, although it was unaf-
fected by substituting His39 or His128 (Fig. 1B), sug-
gesting that His46 and His66 may be involved in
binding heme at the lower-affinity site. This raised the
question of why EPR-active low-spin heme binding was
not completely abolished in one or more of these vari-
ants. One explanation for this is that a minority of the
heme bound at the HxH site may not be magnetically
coupled, most likely as a result of incomplete occupa-
tion, which would result in a small component of the
HxH-bound heme being EPR-active. Alternatively, loss
of the second heme site ligands could interfere with cou-
pling of heme at the HxH site. In either case, inactiva-
tion of both heme-binding sites would be required to
abolish all low-spin heme binding.
To test this model, a further set of site-directed vari-
ants was generated. Beginning with the H93 ⁄94 ⁄95A
variant (in which the three His residues at the HxH
motif are substituted for Ala residues and previously
referred to as the HHH variant) [24], each of the
remaining His residues was substituted stepwise,
sequentially generating H39 ⁄93 ⁄94 ⁄95A, H39 ⁄46 ⁄93 ⁄
94 ⁄95A and H39 ⁄46 ⁄66 ⁄93 ⁄94 ⁄95A, as well as an
entirely His-free Irr
Rl
protein, H39 ⁄46 ⁄66 ⁄93 ⁄94 ⁄95 ⁄
128A (His-free Irr
Rl
). The far-UV CD spectrum of the
fully His-free variant indicated that the protein was
folded with a secondary structure content similar to
the wild-type protein (Fig. S1). UV-visible absorbance
(Fig. 1C) and EPR analysis (Fig. 1D) of this His-free
variant after the addition of heme revealed only weak,
nonspecific (adventitious) binding (Fig. S2). Having
established the baselines for the fully His-free form of
Irr
Rl
, we used UV-visible absorbance to examine heme
binding to the Irr
Rl
variants that retained at least one
histidine residue (Fig. 1C). The data obtained showed
that, as additional His residues were substituted, the
form of the spectrum changed, with that of H39 ⁄93 ⁄
94 ⁄95A being similar to the H93 ⁄94 ⁄95A variant,
whereas that of H39 ⁄46 ⁄66 ⁄93 ⁄94 ⁄95A more closely
resembled the His-free variant. This reflects a decrease
in the proportion of heme binding in the low-spin con-
figuration and is consistent with a progression towards
only low-affinity adventitious, high-spin heme binding
in the absence of His residues (Fig. S2). EPR spectra
of the multi-His variants were recorded (Fig. 1D).
These showed that the ability to bind low-spin heme
was significantly diminished in the H39 ⁄93 ⁄94 ⁄95A
variant, and was lost entirely in H39 ⁄46 ⁄93 ⁄94 ⁄95A
Irr
Rl
.
Irr
Rl
stabilizes heme in the ferric form
Previous studies of heme binding to Irr from B. japoni-
cum indicated that the protein has both ferric and fer-
rous heme-binding sites [12]. To investigate whether
Irr
Rl
has a specific ferrous heme-binding site, Irr
Rl
was
titrated with heme at pH 7 using reduced hemin under
anerobic conditions in the presence of a two-fold
excess of sodium dithionite. As shown by UV-visible
absorbance spectra, heme was bound in a reduced
state, almost all of which was in the low-spin form,
with the Soret band at 426 nm and resolved a⁄bbands
at 528 and 559 nm [24] (Fig. 2A). Measurements of
A
426
(converted to fractional saturation) fitted well to
a single site model, giving a K
d
of 1 ± 0.2 ·10
)6
m
(Fig. 2A, inset). This indicated that ferrous heme has a
significantly lower affinity than ferric heme
(K
d
=1·10
)7
m) [24] (Fig. S3), and that ferrous
heme apparently binds at only one of the two heme-
binding sites of Irr
Rl
. Upon removing dithionite under
anerobic conditions, the UV-visible absorbance spec-
trum revealed that heme was bound only in the oxi-
dized state (not shown), consistent with a higher
affinity binding of ferric heme.
Heme binding to proteins is commonly pH-depen-
dent [26–28] and thus heme binding in both the oxi-
dized and reduced states was investigated at higher
pH. In the oxidized state, although the affinity was
found by absorbance and fluorescence titrations to
decrease somewhat with increasing pH (Doc. S1 and
Fig. S3A, B), the form of heme binding (as judged
from difference and absolute UV-visible spectra) was
unaltered (Doc. S1 and Fig. S3C). By contrast, at
pH 8, the reduced heme UV-visible difference spectrum
was quite different from that at pH 7, indicating that
only ferric heme was bound to the protein (Fig. 2B).
Even at a five-fold excess of dithionite, a mixture of
reduced and oxidized heme was observed, which again
was all converted to oxidized heme upon the anerobic
removal of excess dithionite (data not shown).
Although there is no indication that heme undergoes
redox cycling in Irr
Rl
, we attempted to determine the
reduction potential of Irr
Rl
-bound heme. Spectropoten-
tiometric titration experiments at pH 7 using dithionite
in the presence of mediators were unsuccessful because
a significant reduction of heme was not observed in
the accessible potential range (data not shown), consis-
tent with Irr
Rl
-bound heme having a very low reduc-
tion potential (i.e. similar to, or lower than, that of the
HSO
3
)
⁄S
2
O
42)
couple of < )500 mV at pH 7) [29].
Thus, Irr
Rl
has a considerable preference for heme in
the ferric state, and strongly promotes the oxidation of
heme when it encounters this ligand in the ferrous
Heme binding to Irr promotes oligomerization G. F. White et al.
2014 FEBS Journal 278 (2011) 2011–2021 ª2011 The Authors Journal compilation ª2011 FEBS
form. Therefore, it is clear that Irr
Rl
does not contain
a ferrous heme-binding site.
Heme binding to Irr
Rl
promotes oligomerization
The pathogen B. abortus is a member of the Rhizobi-
ales that is more closely related to Rhizobium than it is
to Bradyrhizobium. Its Irr protein (Irr
Ba
) also more clo-
sely resembles Irr
Rl
(67% identical) than Irr
Bj
(56%
identical) and it lacks the N-terminal HRM. Gel filtra-
tion studies on purified Irr (Irr
Ba
) from this species
had shown that it is likely dimeric [30], as are other
members of the Fur superfamily [31,32]. To analyse
the association state of Irr
Rl
, analytical gel filtration
experiments were performed. For the ‘as isolated’,
heme-free protein, two distinct peaks were detected in
the chromatograph (Fig. 3A). Thus, the heme-free pro-
tein exists as an equilibrium mixture of two principal
association states, which must be in slow exchange,
such that they can be separated by gel filtration. Cali-
bration of the column suggested that the lower molec-
ular weight species corresponds to a dimeric form,
consistent with our previous proposal that heme bound
at the HxH motif of Irr
Rl
is EPR-silent as a result of
magnetic coupling between two closely located HxH-
bound hemes. The higher molecular weight species cor-
responded to a much larger, oligomeric form, although
this could not be precisely defined by gel filtration
alone. Interestingly, addition of heme to Irr
Rl
resulted
in significant changes in the elution profile: only one
major band was observed, corresponding to the higher
molecular weight species (Fig. 3A). Analysis of elution
data at different wavelengths confirmed that the
observed profile changes were the result of a change in
the distribution between higher and lower molecular
weight species and not the preferential binding of heme
to the larger species (Fig. S4). Heme binding, however,
did not cause major changes in the secondary structure
because far-UV CD spectra of Irr
Rl
before and after
the addition of heme were very similar (Fig. 3B).
To determine the mass of the oligomeric species
more precisely, analytical ultracentrifugation experi-
ments were run for heme-Irr
Rl
at five different concen-
trations, in the range 5–19 lm, at 15 000 and at
17 000 r.p.m. Figure 3C shows the data obtained for
one representative concentration of heme-Irr
Rl
(10 lm).
Fitting of the data, at both speeds and for all concen-
trations, to a single component model gave a mass of
96.6 ± 8 kDa for the heme-bound Irr
Rl
sample (resid-
uals for the fits are shown in the inset to Fig. 3C). The
data thus indicate that the higher molecular weight
species of Irr
Rl
corresponds to a hexamer.
To examine the role (if any) of the HxH motif in
the heme-dependent oligomerization, the association
state of the H93 ⁄94 ⁄95A mutant Irr
Rl
was also investi-
gated. The gel filtration profile (Fig. 3D) was very sim-
ilar to that of the wild-type protein, and the addition
of heme to the mutant polypeptide caused a similar
shift in equilibrium towards the larger (hexameric) spe-
cies. Therefore, the process of oligomerization is not
directly connected to heme binding at the HxH motif
and must be associated with heme binding at the
lower-affinity site. To test this proposal, Irr
Rl
variants
containing single substitutions of His residues consid-
ered to be involved in heme binding were examined by
analytical gel filtration. H46A and H66A Irr
Rl
proteins
AB
Fig. 2. Binding of ferrous heme by Irr
Rl
. (A) UV-visible difference absorbance spectra recorded upon titration of ‘as isolated‘ (heme-free) Irr
(17 lM)in50mMTris-HCl, 50 mMKCl (pH 7) with ferrous heme. The trend of absorption changes are indicated by arrows. The inset shows
a plot of fractional saturation (from DA
426
) as a function of reduced heme concentration. A fit of the data to a single site model is drawn.
Pathlength, 1 cm; temperature, 20 C. (B) UV-visible difference absorbance spectra as recorded in (A), except that the pH of the Irr (18 lM)
solution was 8. Note that the form of the difference titration was different from that observed for the oxidized heme titration (Fig. S3A),
although this is because hemin remained in the ferrous state in the reference cuvette, thereby imparting a different form on the difference
spectrum.
G. F. White et al. Heme binding to Irr promotes oligomerization
FEBS Journal 278 (2011) 2011–2021 ª2011 The Authors Journal compilation ª2011 FEBS 2015
