Bacterial IscU is a well folded and functional single domain protein
Salvatore Adinolfi
1
, Francesca Rizzo
1
, Laura Masino
1
, Margie Nair
1
, Stephen R. Martin
1
, Annalisa Pastore
1
and Piero A. Temussi
1,2
1
National Institute of Medical Research, London, UK;
2
University of Naples Federico II, Napoli, Italy
Iron–sulfur clusters are widely represented in most organ-
isms, but the mechanism of their formation is not fully
understood. Of the two main proteins involved in cluster
formation, NifS/IscS and NifU/IscU, only the former has
been well studied from a structural point of view. Here we
report an extensive structural characterization of Escherichia
coli IscU. We show by a variety of physico-chemical tech-
niques that E. coli IscU construct can be expressed to high
purity as a monomeric protein, characterized by an ab fold
with high a-helix content. The high melting temperature and
the reversibility of the thermal unfolding curve (as measured
by CD spectroscopy) hint at a well ordered stable fold. The

a transient [2Fe)2S] cluster. Escherichia coli IscU is
homologous to the amino terminal domain of NifU with
which it shares three conserved cysteines and the binding of
a transient [2Fe-2S] cluster [6]. We undertook a systematic
study of the Isc proteins of E. coli. Considering the central
role of IscU in Fe–S cluster biosynthesis, suggested,
inter alia, by the fact that it is one of the most conserved
sequence motifs in nature and the fact that its three
dimensional structure has not yet been published, we
decided to start the study of bacterial IscU.
A further motivation for studying bacterial IscU is its
possible connection with CyaY, the bacterial orthologue of
frataxin, a small protein expressed at abnormally low levels
in Friedreich’s ataxia patients [7]. Consistent evidence shows
that Friedreich’s ataxia arises from disregulation of mito-
chondrial iron homeostasis, with concomitant oxidative
damage leading to neuronal death [8–13]. Accumulating
evidence suggests that frataxin is involved in iron meta-
bolism [14–20]. A possible function of CyaY in the complex
chain of events involved in Isc formation might be to supply
iron ions to IscU, as suggested by a recent report by Yoon &
Cowan [21] on the interaction between the corresponding
human orthologues. A detailed structural characterization
of IscU and, most of all, the nature of its interaction with
iron ions may help to clarify this function.
Using complementary biophysical and biochemical tech-
niques, we report here a structural characterization of this
protein and demonstrate that E. coli IscU can be obtained
as a recombinant well-folded protein. We demonstrate that
our construct can function as a scaffold for a transient Fe–S

buffer (20 m
M
Tris/HCl, pH 8, 150 m
M
NaCl, 10 m
M
2-mercaptoethanol) and subsequently sonicated and centri-
fuged. The protein was purified by affinity chromatography
using Ni-nitrilotriacetic acid gel or a glutathione S-sepharose
column. Tobacco etch virus protease cleavage was then
obtained by incubating the protein overnight at 4 °C.
Further purification from glutathione S-transferase was
carried out by gel filtration chromatography on a G-75
column (Pharmacia).
15
N-labelled samples of IscU for
nuclear magnetic resonance studies were produced by
growing the bacteria in minimal medium using ammonium
sulfate as sole source of nitrogen.
The protein was desalted by dialysis against the final
buffer (either 20 m
M
Hepes/KOH pH 7.5 or 20 m
M
Tris/
HCl pH 7.5–8.0 with NaCl 50–150 m
M
and 20 m
M
2-mercaptoethanol) and concentrated with an Amicon

with UV absorption optics (Beckman Coulter Ltd, High
Wycombe, UK). The measurements were performed at
20 °C using speeds of 9000, 12 000 and 20 000 r.p.m. and
rotor An 60 Ti. Protein concentrations were in the range
10–40 l
M
. Data were recorded using different ionic strength
conditions (20 m
M
Tris/HCl at pH 8 with 50 m
M
or
150 m
M
NaCl and 10 m
M
2-mercaptoethanol). Each meas-
urement was repeated after 6 h to ensure that equilibrium
had been reached and that proteolysis was not occurring. In
all datasets, the absorbance of the depleted area at a final
speed of 40 000 r.p.m. provided an experimental value for
the baseline offset. The data were analysed with the
ORIGIN
XL
-
A
/
XL
1 package (Beckman) and fitted to the following
equation:

buffer at pH 7.5 using protein
concentrations of 7–35 l
M
from two independent protein
preparations. The spectra were recorded in fused silica
cuvettes of 1 mm path length. Ten scans were averaged and
the appropriate buffer baseline was subtracted. Spectral
decomposition for secondary structure predictions was
achieved by combining the CONTIN, SELCON and
CDSSTR methods [22].
Near UV spectra required a 2 mm path length cuvette
and a 490 l
M
protein concentration. Variations of the CD
signal were studied as a function of temperature over the
range 10 °Cto95°C using a heating rate of 1 °Cper
minute.
Fluorescence measurements were recorded on a SPEX
Fluoromax spectrometer (Glen Spectra Ltd, Middlesex,
UK) fitted with a thermostatically controlled jacketed cell
holder and interfaced with a Neslab RTE-111 water bath
(Thermo-Neslab, Portsmouth, UK). Fluorescence emission
spectra in the range 300–450 nm were recorded at 20 °C
with an excitation wavelength of 290 nm.
Nuclear magnetic resonance spectroscopy
NMR spectra at 25 °CwererecordedonVarianINOVA
spectrometers (Varian Ltd, Walton-on-Thames, UK) oper-
ating at 500, 600 and 800 MHz
1
H frequency. Typically,

M
samples in 20 m
M
Tris/HCl at pH 7.0, 50 m
M
NaCl and 10 m
M
dithio-
threitol. The spectra were acquired at 11.7 T (500 MHz
proton frequency) using standard pulse sequences [26] and
analyzed using
NMRPIPE
/
NMRDRAW
or
XEASY
.ForT
1
and T
2
measurements, peak intensities were determined
for 109 amide resonances as a function of the relaxation
delay and the data were then fitted by least-squares fitting
to a single exponential. Correlation times were calculated
from the T
1
/T
2
ratios according to the so-called model-
free approach [27]. Experimental

containing 76 l
M
IscU in the presence of different IscS
ratios, chosen to have formation kinetics compatible with
the spectroscopic measurements (the optimal ratio is
% 1 : 28 IscS/IscU according to Kispal et al.[4]).Afivefold
excess (relative to IscU concentration) of freshly prepared
ferric ammonium citrate and 4 m
M
2-mercaptoethanol were
added to the mixture. All experiments were performed in
20 m
M
Tris/HCl at pH 7.5 and 200 m
M
NaCl buffer using
a glove box under an argon atmosphere to obtain an
anaerobic environment. The Fe–S cluster formation was
followed with a UV-visible spectrophotometer Cary 50 Bio
(Varian Ltd, Walton-on-Thames, UK) recording spectra at
different times. A solution of 76 l
M
nonreconstituted IscU
wasusedasblank.
Results
E. coli
IscU is a stable well folded protein
IscU could be purified to a homogeneous construct. The
secondary structure of the protein and its thermal stability
were first characterized by circular dichroism spectroscopy.

1
Hand
15
N dimensions of the
1
H-
15
N HSQC spectrum of uniformly
15
N labeled sample
of E. coli IscU confirms that the protein is well behaved
without relevant disordered regions (Fig. 2B). The num-
ber of observed backbone amide resonances (111), is
consistent with the expected ones (123). Figure 2C shows
the excellent quality of the
1
H homonuclear NOESY of
IscU.
NMR relaxation measurements (T
1
,T
2
and hetero-
nuclear
15
N-[
1
H] NOE values) were recorded at 25 °Cand
600 MHz on an
15

M
protein in 10 m
M
buffer at pH 7.5.
Ó FEBS 2004 Characterization of E. coli IscU (Eur. J. Biochem. 271) 2095
for all residues excluding only two resonances, located
at 7.8 p.p.m. and 126.3 p.p.m. and at 8.0 p.p.m. and
124.8 p.p.m., which have negative values (Fig. 3C). Nega-
tive NOEs indicate highly flexible regions that, in globular
proteins, are usually observed for the amides at the N- and
C-termini. Positive values are typical of relatively rigid
regions [27]. We can therefore conclude that the IscU is
compact, without relevant differences of the local flexibility.
Finally, the environment of the unique tryptophan
(Trp76) of the IscU sequence was probed by fluorescence
measurements. The tryptophan fluorescence emission spec-
trum shows an emission band at 355 nm (data not
shown), suggesting that this residue is highly exposed to
the solvent.
E. coli
IscU is a monomeric protein
The sample was characterized for its aggregation state, using
three independent techniques. The molecular mass of native
E. coli IscU was first estimated by gel filtration. The elution
profile of E. coli IscU presents a single peak at a molecular
mass corresponding to that of the monomer (13.9 kDa)
Fig. 2. Typical ID and 2D spectra of IscU. (A) 1D NMR spectrum of
non labeled IscU sample (0.5 m
M
)in20m

15
N-labeled
sample of E. coli IscU at 0.3 m
M
concentration. The pulse sequence
used for the
15
N-[
1
H] heteronuclear NOE measurement is that pub-
lished by Farrow et al. [28]. In the absence of the sequential assign-
ment of the spectra, residue numbers are ordered according to their
resonances.
2096 S. Adinolfi et al.(Eur. J. Biochem. 271) Ó FEBS 2004
(Fig. 4A). However, an additional peak, which could be
consistent with a dimer, appeared if no reducing agent was
used in the buffer.
Because the gel filtration profile strongly depends on the
protein shape, this result was confirmed in two different
concentration ranges by analytical ultracentrifugation, a
technique generally considered as the most accurate way to
detect oligomerization, and by estimating the correlation
time of the protein in solution from the NMR relaxation
measurements. The apparent molecular mass of IscU as
obtained by ultracentrifugation methods for 10–40 l
M
samples is 14.7 kDa, a value corresponding, within experi-
mental error, to the monomer molecular mass (Fig. 4B). No
significant differences were observed using different ionic
strength conditions.

IscU in the
presence of 0.5–5.0 l
M
IscS, a 5-fold excess of ferric
ammonium citrate (based on the concentration of IscU
monomer), and 4 m
M
2-mercaptoethanol. The IscU/IscS
ratio was used to vary the rate of cluster formation. A
simple, yet efficacious, way to characterize an Iron–sulfur
cluster is through its UV-visible absorption spectrum [30].
Samples of ÔapoÕ IscU, i.e. prior to cluster assembly, do not
have a visible chromophore but become red on anaerobic
treatment with catalytic amounts of IscS in the presence of
excess
L
-cysteine and a stoichiometric amount of ferric
ammonium citrate. The spectrum should contain charac-
teristic bands centered at % 320, 410 and 456 nm and a
pronounced shoulder at 510 nm [31]. Similarly to what is
observed for the assembly of a [Fe
2
S
2
]
2+
cluster in
A. vinelandii IscU we recorded a UV-visible absorption
spectrum (Fig. 5) with characteristic bands at 320, 407 and
447 nm and a pronounced shoulder at 513 nm, character-

perturbation of selected cross peaks in the spectrum that
correspond to the protons affected by the binding (e.g.
[32]). When paramagnetic species are present in solution
but not bound, a general and unspecific broadening of
the resonances mediated by the solvent is observed.
Binding of a paramagnetic species leads instead to major
Fig. 4. Probing the aggregation state of IscU. (Top) Calibration curve
for apparent molecular mass determination of E. coli IscU in native
conditions by gel filtration chromatography. A HiLoad 10/30 Super-
dex 75 column equilibrated with Tris/HCl buffer (pH 7.5), 100 m
M
NaCl was used, with ovalbumin (A; 43 kDa), chymotrypsinogen A (B;
25 kDa) and ribonuclease A (C; 13.7 kDa) as molecular standards for
the mass calibration. (Bottom) Sedimentation equilibrium distribution
of IscU measured using analytical ultracentrifugation. The data were
recorded at 20 °C and 20 000 r.p.m. Protein concentration was 20 l
M
in 20 m
M
Tris/HCl a 50 m
M
at pH 8.0, 50 m
M
NaCl and 10 m
M
2-mercaptoethanol. Lower panel: Experimental absorbance at 280 nm
as a function of the radial position and data fitting to the equation
reported in Materials and methods. Upper panel: Distribution of dif-
ferences between experimental and calculated values. The apparent
molecular mass for this experiment is 14 000 Da.

our construct of E. coli IscU is well folded. The high
melting temperature and the reversibility of the thermal
unfolding curve (as measured by CD) hint at a well
ordered stable fold. This view is confirmed by the
excellent dispersion of cross peaks in the
1
H-
15
NNMR
correlation spectrum, by the quality of the homonuclear
NOESY spectrum (Fig. 2C and data not shown) and by
relaxation data. Altogether our data do not support a
flexible molten globule-like state for E. coli IscU.
Until now the only monomeric IscUs identified have been
the human protein [37] and that from Haemophilus influ-
enzae, whereas homologues from other organisms gener-
ally have been described as dimers, e.g. the IscU from
T. maritima was shown to form a homodimer [35]. Here we
present conclusive evidence that the E. coli orthologue also
behaves as a monomeric protein. Monomeric IscU is
Fig. 6. Probing for Fe(II) binding of IscU by NMR. (A)
1
H-
15
NHSQC
spectrum of uniformly
15
NlabeledsampleofE. coli IscU; (B) as for
sample A, after addition of Fe(II) in a Fe(II)/IscU ratio of 2 : 1;
(C) same as sample A with a ratio Fe(II)/IscU of 5 : 1 after partial

[30].
Another important result from our study is that E. coli
IscU does not interact directly with iron ions, independently
of Isc formation. The use of NMR is probably conclusive in
this respect because this technique can reveal even very
weak interactions. The generally accepted mechanism for
biological Iron–sulfur cluster assembly is based on the
hypothesis that persulfides catalytically formed on IscS
can be transferred to IscU for cluster assembly through
association of the two proteins [30]. The alternative
mechanism, proposed by Nuth et al. [34], based on initial
binding of iron by IscU, is not consistent with the reluctance
of IscU to accept iron ions indicated by our NMR data.
Accordingly, the mechanism of Fe–S cluster assembly based
on initial binding of iron followed by delivery of sulfur
equivalents, proposed for the T. maritima IscU [34], does
not seem applicable to E. coli IscU.
Within this frame, the role of CyaY in bacteria may well
be that of an iron chaperone which passes the iron to IscU,
as first suggested by Yoon & Cowan [21], and more recently
supported by independent line of experimental evidence
both based on in vivo and in organello studies [37,38]. This
process might however, require preformation of a (tran-
sient?) ternary complex with IscU/IscS and occur cooper-
atively only when a source of sulfur is also available.
Further studies to describe the molecular details of these
multiple interactions will be needed to understand this
complex phenomenon.
Acknowledgement
The project was funded by Seek A Miracle/MDA and the Friedreich’s

Arch. Neurol. 56, 1201–1208.
10. Wong, A., Yang, J., Cavadini, P., Gellera, C., Lonnerdal, B.,
Taroni, F. & Cortopassi, G. (1999) The Friedreich’s ataxia
mutation confers cellular sensitivity to oxidant stress which is
rescued by chelators of iron and calcium and inhibitors of apop-
tosis. Hum. Mol. Genet. 8, 425–430.
11. Delatycki, M.B., Camakaris, J., Brooks, H., Evans-Whipp, T.,
Thorburn, D.R., Williamson, R. & Forrest, S.M. (1999) Direct
evidence that mitochondrial iron accumulation occurs in Frie-
dreich ataxia. Ann. Neurol. 45, 673–675.
12. Wilson, R.B. & Roof, D.M. (1997) Respiratory deficiency due to
loss of mitochondrial DNA in yeast lacking the frataxin homo-
logue. Nat. Genet. 16, 352–357.
13. Lamarche, J.B., Cote, M. & Lemieux, B. (1980) The cardiomyo-
pathy of Friedreich’s ataxia morphological observations in 3 cases.
Can. J. Neurol. Sci. 7, 389–396.
14. Musco, G., Stier, G., Kolmerer, B., Adinolfi, S., Martin, S.,
Frenkiel, T., Gibson, T. & Pastore, A. (2000) A structural
understanding of genetic diseases: the solution structure of fra-
taxin, the protein responsible for Friedreich ataxia. Structure 8,
695–707.
15. Foury, F. & Cazzadini, O. (1997) Deletion of the yeast homologue
of the human gene associated with Friedreich’s ataxia elicits iron
accumulation in mitochondria. FEBS Lett. 411, 373–377.
16. Babcock, M., de Silva, D., Oaks, R., Davis-Kaplan, S., Jiraler-
spong, S., Montermini, L., Pandolfo, M. & Kaplan, J. (1997)
Regulation of mitochondrial iron accumulation by Yfh1p, a
putative homolog of frataxin. Science 276, 1709–1712.
17. Foury, F. (1999) Low iron concentration and aconitase deficiency
in a yeast frataxin homologue deficient strain. FEBS Lett. 6,

(1992) Pulse sequences for removal of the effects of cross
Ó FEBS 2004 Characterization of E. coli IscU (Eur. J. Biochem. 271) 2099
correlation between dipolar and chemical shift anisotropy
relaxation mechanisms on the measureament of heteronuclear T1
and T2 values in proteins. J. Mag. Res. 97, 359–375.
27. Neuhaus, D. & Williamson, M.P. (2000) The Nuclear Overhauser
Effect in Structural and Conformational Analysis. 2nd edition. John
Wiley & Sons, New York.
28. Farrow, N.A., Muhandiram, R., Singer, A.U., Pascal, S.M., Kay,
C.M., Gish, G., Shoelson, S.E., Pawson, T., Forman-Kay, J.D. &
Kay, L.E. (1994) Backbone dynamics of a free and a phospho-
peptide complexed Src homology 2 domain studied by
15
NNMR
relaxation. Biochemistry 33, 5984–6003.
29. Maciejewski, M.W., Liu, D., Prasad, R., Wilson, S.H. & Mullen,
G.P. (2000) Backbone dynamics and refined solution structure of
the N-terminal domain of DNA polymerase beta. Correlation
with DNA binding and dRP lyase activity. J. Mol. Biol. 296,
229–253.
30. Agar, J.N., Zheng, L., Cash, V.L., Dean, D.R. & Johnson, M.K.
(2000) Role of the IscU protein in Iron–sulfur cluster biosynthesis:
IscS-mediated assembly of a [Fe
2
-S
2
] cluster in IscU. J. Am. Chem.
Soc. 122, 2136–2137.
31. Dailey, H.A., Finnegan, M.G. & Johnson, M.K. (1994) Human
ferrochelatase is an Iron–sulfur protein. Biochemistry 33, 403–407.

EJB4112sm.htm
Fig. S1. Near UV thermal unfolding curve.
Fig. S2. 2D NOESY of IscU.
2100 S. Adinolfi et al.(Eur. J. Biochem. 271) Ó FEBS 2004