Báo cáo khoa học: High thermal and chemical stability of Thermus thermophilus seven-iron ferredoxin Linear clusters form at high pH on polypeptide unfolding doc

High thermal and chemical stability of
Thermus thermophilus
seven-iron ferredoxin
Linear clusters form at high pH on polypeptide unfolding
Susanne Griffin
1
, Catherine L. Higgins
1
, Tewfik Soulimane
2
and Pernilla Wittung-Stafshede
1
1
Department of Chemistry, Tulane University, New Orleans, LA, USA;
2
Paul Scherrer Institute, Structural Biology,
Villigen, Switzerland
To probe the stability of the seven-iron ferredoxin from
Thermus thermophilus (FdTt), we investigated its chemical
and thermal denaturation processes in solution. As predicted
from the crystal structure, FdTt is extremely resistant
to perturbation. The guanidine hydrochloride-induced
unfolding transition shows a midpoint at 6.5
M
(pH 7,
20 °C), and the thermal midpoint is above boiling, at 114 °C.
The stability of FdTt is much lower at acidic pH, suggesting
that electrostatic interactions are important for the high
stability at higher pH. On FdTt unfolding at alkaline pH,
new absorption bands at 520 nm and 610 nm appear tran-
siently, resulting from rearrangement of the cubic clusters

into linear three-iron species. A range of iron–sulfur proteins
has been found to accommodate these novel clusters in vitro,
although no biological function has yet been assigned.
Keywords: ferredoxin; linear iron–sulfur cluster; protein
unfolding; thermostability; Thermus thermophilus.
Many proteins require the binding of cofactors to perform
their biological activity. It has been demonstrated in vitro
that many proteins retain interactions with their cofactors
after polypeptide unfolding [1–6]. Therefore, it is possible
that cofactors bind to their corresponding polypeptides
before or during folding in vivo. Cofactors most often
stabilize the native states of the proteins with which they
interact [1–6]. However, the manner in which cofactors
affect polypeptide folding and unfolding pathways remains
poorly understood. Iron–sulfur ([Fe-S]) clusters represent
one of nature’s simplest, functionally versatile, and perhaps
most ancient cofactors [7]. The [Fe-S] clusters, which have
2, 3 or 4 irons, are usually attached to their protein partners
by four cysteine thiol ligands [7–9]. Proteins that contain
one or more [Fe-S] clusters represent a large class of
structurally and functionally diverse proteins that are
essential players in the life-sustaining processes of respir-
ation, nitrogen fixation, and photosynthesis. In these
proteins, the [Fe-S] clusters participate as agents of electron
transfer, substrate activation, catalysis, and environmental
sensing [7,10]. Most [Fe-S] proteins have low reduction
potential and are known as ferredoxins. Given the struc-
tural simplicity of [Fe-S] clusters and the participation of
ferredoxins in so many metabolic processes, it is somewhat
surprising that the pathways for biological formation of

[Fe-S] clusters and their incorporation into proteins are
only now beginning to emerge [7].
The origin of protein thermostability is still an unsolved
problem, and its understanding presents a great intellectual
challenge to scientists, not to mention its potentially
enormous biotechnological impact. Proteins from thermo-
philic organisms offer a unique opportunity to study the
determinants of thermostability [11,12]. Although these
proteins are often very similar in sequence and structure to
their mesophilic homologues (this is true also for mesophilic
and thermophilic ferredoxins), they are much more resistant
to thermal denaturation and inactivation. In the case of
thermostable ferredoxins, it is not clear if subtle features
around the [Fe-S] cluster site contribute to the additional
stability or if higher stability is a result of polypeptide
properties only [13]. Earlier efforts to determine the origin of
thermostability in monomeric proteins (most often without
cofactors) have led to several hypotheses, such as stabiliza-
tion by an increased number of ionic interactions, an
increased extent of hydrophobic-surface burial, an increased
number of prolines, and smaller surface loops [12].
Although evidence for these and other modes of stabiliza-
tion can be found in specific examples, none applies to all or
even most thermostable proteins. If there are general rules
for how thermophilic proteins attain their stability, then it is
clear that they do not lie exclusively in individual inter-
actions but may be based on properties of the whole
molecule [14].
In this investigation, we focus on the seven-iron (one
[4Fe-4S]

2+/1+
and one [3Fe-4S]
+1/
° cluster) ferredoxin
from Thermus thermophilus (hereafter called FdTt). T. ther-
mophilus is a Gram-negative aerobic bacterium found
in hot springs, thermal vents, and thermal spas. It grows
at temperatures of 50–82 °C, with optimum growth at
65–72 °C [15]. Its seven-iron ferredoxin, FdTt, is a small
single-chain, single-domain protein with 78 residues. The
Correspondence to P. Wittung-Stafshede, Department of Chemistry,
Tulane University, 6823 St Charles Avenue, New Orleans,
LA 70118, USA. Fax: + 1 504 865 5596, Tel.: + 1 504 862 8943,
E-mail:
Abbreviations: FdAv, ferredoxin from Azotobacter vinelandii;FdTt,
ferredoxin from Thermus thermophilus; GdnHCl, guanidine hydro-
chloride; T
m
, midpoint temperature of thermal unfolding transition.
(Received 2 July 2003, revised 23 September 2003,
accepted 7 October 2003)
Eur. J. Biochem. 270, 4736–4743 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03873.x
crystal structure of FdTt was recently solved at 1.64 A
˚
resolution [16], but there has been no in vitro characteriza-
tion of the protein. FdTt has a (bab)
2
core structure
enveloping the two [Fe-S] clusters with an additional
C-terminal a-helix further away from the clusters (Fig. 1).

According to the crystal structure of FdTt, the improved
polar and hydrophobic interactions lead to extensive cross-
linking of the secondary-structure elements (compared with
mesophilic ferredoxins), which is believed to result in high
stability. It appears that most of the stabilizing features of
FdTt are found in the vicinity of the [3Fe-4S] cluster, which
is the cluster that has functional importance [16]. We here
report a detailed biophysical characterization of the stability
of FdTt to thermal and chemical perturbation in solution
in vitro. Interestingly, the cubic clusters in FdTt rearrange
into new, linear three-iron species on polypeptide unfolding
at high pH.
Materials and methods
Materials
The chemical denaturant guanidine hydrochloride
(GdnHCl) was of highest purity. All chemicals were
purchased from Sigma. The FdTt protein was purified as
previously described [16].
Chemically induced unfolding
GdnHCl was used to promote FdTt unfolding at different
pH values (2.5, 4, 7, 10 and 10.5) at 20 °C. An FdTt
concentration of 20 l
M
was used unless specified otherwise.
Buffers of 25 m
M
concentration were used unless otherwise
specified. Samples were incubated (at 20 °C) for various
lengths of time from 5 min to 120 h before spectroscopic
measurements were taken. Unfolding was monitored by

far-UV CD (200–300 nm) on a JASCO-810 instrument
(1 mm cell), by visible absorption (250–700 nm) on Cary-50
and Cary-100 spectrophotometers (1 cm cell), and by
tryptophan emission (300–450 nm, excitation at 280 nm)
on a Cary Eclipse instrument. For each set of conditions, the
transition midpoint ([GdnHCl]
1/2
) was obtained by direct
inspection of the data. Different buffers were used for
experiments at different pH values: glycine/HCl was used
for pH 2.5, citric acid/sodium phosphate was used for pH 4,
phosphate was used for pH 7, and KCl/NaOH was used for
pH 10 and 10.5.
EPR experiments were performed with liquid nitrogen at
110 K on a Bruker EMX instrument (A. Tsai, University of
Texas Medical School, Houston, TX, USA) using 10 mW
microwave power and 9.3 GHz microwave frequency. The
EPR samples were 100 l
M
FdTt in buffer, pH 10.5 (folded
protein) and 100 l
M
FdTt incubated for 15 min in 6.5
M
GdnHCl, pH 10.5, 20 °C (unfolded protein with linear
[Fe-S] cluster).
Thermally induced unfolding
Thermally induced unfolding of FdTt was monitored by
visible absorption, fluorescence, and far-UV CD methods.
To probe the reaction by cluster integrity, the absorption at

408 nm was monitored (for FdTt samples with different pH
and GdnHCl conditions) as a function of temperature. The
signal from 20 l
M
FdTt was recorded on increasing the
temperature at rates of 0.5, 0.25, or 0.125 °C per minute
(one data point collected per second; from 25 °Cto95°C).
At the end of each experiment, the temperature was
decreased to 25 °C, and a full absorption spectrum was
taken to check for refolding. The midpoint of transition
(T
m
) at each pH and GdnHCl concentration was deter-
mined by direct inspection of the absorption vs. temperature
data. T
m
values obtained for FdTt in the presence of
different GdnHCl concentrations (but the same pH and
scan rate) were used to linearly extrapolate to 0
M
GdnHCl,
to obtain the T
m
for FdTt without denaturant at each pH
(and each scan rate). Next the extrapolated T
m
values for
Fig. 1. Ribbon diagram of FdTt. Protein data
bank file 1H98. The iron–sulfur clusters
(iron space-filled red, sulfur space-filled

yellow) and secondary-structure elements
(a-helices in red, b-strands in gold, random
coil in white) are highlighted.
Ó FEBS 2003 Stability of T. thermophilus seven-iron ferredoxin (Eur. J. Biochem. 270) 4737
0
M
GdnHCl obtained at the three different scan rates were
plotted as a function of 1/(scan rate) to yield the T
m
for
FdTt at that pH at infinite scan rate, i.e. where 1/(scan
rate) ¼ 0.
The T
m
values obtained by monitoring visible absorption
were compared with those derived from far-UV CD
(monitored at 220 nm) and tryptophan emission experi-
ments (emission at 357 nm; excitation at 280 nm). In the
case of the thermal far-UV CD-monitored experiments, we
used an OLIS spectropolarimeter with a digitally controlled
water bath (Julabo, Allentown, PA, USA); the approximate
scan rate was 0.5 °CÆmin
)1
.
Results
Chemically induced FdTt unfolding
Folded FdTt has characteristic visible absorption at 408 nm
resulting from the intact [Fe-S] clusters which disappears as
the protein unfolds (Fig. 2A). FdTt has one tryptophan
at position 64, and tyrosines at positions 33, 55 and 67 in

the primary structure [16]. Folded FdTt shows very little
tryptophan fluorescence because of energy transfer to the
iron–sulfur clusters. As the protein unfolds, the tryptophan
emission (at  350 nm) increases dramatically because of
cluster and tryptophan separation and presumably cluster
decomposition (Fig. 2B). Folded FdTt has positive CD
absorption  230 nm, arising from the tryptophan and
tyrosine contribution, and a negative CD feature at 220 nm,
characteristic of the presence of secondary structure. Both
CD bands lose intensity as the protein unfolds (Fig. 2C),
and the CD spectrum on unfolding resembles that of a
random-coil polypeptide. Taken together, these spectro-
scopic techniques can be used to probe the unfolding
reaction of FdTt via cluster integrity (visible absorption),
cluster–tryptophan distance (emission), and secondary
structure (far-UV CD) (Fig. 2).
To probe the unfolding mechanism for FdTt, visible
absorption, fluorescence, and far-UV CD probes were
monitored as a function of time after the protein had been
mixed with a high concentration of the chemical denaturant
GdnHCl (7.9
M
GdnHCl final concentration, 20 °C).
Despite the high denaturant concentration, more than 6 h
were required for the signals to reach their endpoints at pH
7(t
1/2
¼ 50 ± 10 min). At pH 2.5, the kinetics for the same
reaction were faster (t
1/2

¼ 10 ± 5 min). Identical (within
error) kinetic traces were observed regardless of far-
UV CD; 408-nm absorption or fluorescence signals were
used as the detection method (absorption and CD changes
shown in Fig. 3) for each condition. This observation
suggests that FdTt unfolding is a single process (at pH 2.5
and 7), in which polypeptide unfolding and cluster degra-
dation occur simultaneously. At pH 10, however, the
kinetic process monitored by visible absorption did not
match that probed by far-UV CD (discussed below).
Next, GdnHCl titrations at 20 °C were performed at
various pH values, and FdTt unfolding was probed at
different incubation times (from 2 to 48 h) using the three
spectroscopic methods. The GdnHCl-induced unfolding
process is irreversible at all pH values, probably because the
clusters decompose on unfolding (in accord with complete
visible-absorption disappearance, Fig. 2A). It is also
possible that cysteine oxidation takes place in the unfolded
state, hampering refolding. Irreversible unfolding has been
reported for other ferredoxins [17–20]. Because of the
irreversibility of the reaction, no thermodynamic data, such
Fig. 2. Visible absorption (A), tryptophan emission (B), and far-UV CD
(C) of native FdTt (solid line, 0
M
GdnHCl,pH 7,20°C) and denatured
FdTt (dotted line, 7.9
M
GdnHCl 48 h incubation, pH 7, 20 °C).
4738 S. Griffin et al.(Eur. J. Biochem. 270) Ó FEBS 2003
as DG

U
(H
2
O), could be obtained for FdTt; instead, we
report unfolding-transition midpoints as a function of pH
and incubation time at 20 °C (summarized in Table 1). The
transition midpoints shift to lower GdnHCl concentration
as the incubation time is increased as expected for an
irreversible reaction (Table 1).
We observe single unfolding transitions with all spectro-
scopic probes under all conditions except at high pH (see
below). The different probing methods, reporting on
different properties of FdTt, gave similar transition mid-
points for FdTt samples incubated for the same amount of
time and at the same pH condition (Table 1), supporting the
suggestion that FdTt unfolding is a two-state process with
cluster degradation occurring simultaneously with polypep-
tide unfolding. For example, the midpoint of the unfolding
transition appeared at  6.5
M
GdnHCl after 48 h of
incubation at pH 7 (20 °C) as monitored by all three
probes. At pH values lower than 7, the apparent stability of
FdTt decreased significantly (Fig. 4); for example, at
pH 2.5, the midpoint of transition was 1.5
M
GdnHCl
(48 h incubation, 20 °C).
Transition midpoints for FdTt samples with and without
500 m

M
NaCl were identical (20 °C, pH 7), indicating that
protein stability is not affected by the presence of NaCl
(data not shown). The effect of protein concentration was
investigated in a separate experiment by comparing unfold-
ing transitions for 10 l
M
and 80 l
M
FdTt at pH 2.5. The
transition midpoints, monitored by visible absorption, was
not significantly different for the two different protein
concentrations. In this experiment (pH 2.5, 20 °C), the
GdnHCl-induced midpoints were 1.7 and 1.6
M
GdnHCl
after 2 h of incubation and 1.1 and 1.0
M
GdnHCl after
24 h of incubation for 80 and 10 l
M
FdTt, respectively.
Thermally induced FdTt unfolding
In buffer at pH 7, FdTt does not unfold below 100 °C.
Therefore, thermally induced unfolding experiments were
conducted in the presence of different concentrations of
GdnHCl (not high enough to unfold FdTt at 20 °C). Like
GdnHCl-induced unfolding at 20 °C, thermal unfolding of
FdTt occurred in a single transition which was irreversible.
In most thermal experiments, visible absorption was used

as detection probe, although some experiments were also
probed by far-UV CD and fluorescence. At each condition,
identical T
m
values were observed with the different
detection probes. For each scan rate studied (0.5, 0.25
and 0.13 °CÆmin
)1
), thermal midpoints (T
m
) obtained at
Fig. 3. FdTt unfolding kinetics (20 °C), measured by visible absorption
at 408 nm, on addition of 7.9
M
GdnHCl at pH 7 (solid line) and pH 2.5
(dashed line). Inset: far-UV CD detection of the same process (solid
line,pH7;dashedline,pH2.5).
Table 1. GdnHCl-induced unfolding midpoints (M) for FdTt at four pH
values probed by visible absorption (Abs), tryptophan emission (FL), and
far-UV CD (CD) after different incubation times (20 °C). The values
have an error of ± 0.2
M
. *, Intermediate with new absorption bands
forms on unfolding; therefore, unfolding midpoints are not reliable by
this technique.
pH Time (h)
Unfolding midpoints (M)
Abs FL CD
2.5 2 2.2 2.3 2.2
8 2.0 2.2 2.2

24 1.6 2.0 1.7
48 1.5 1.5 1.5
4 2 4.7 4.7 4.5
8 4.2 4.4 3.9
24 3.8 4.0 3.8
48 3.5 3.7 3.5
7 2 No unfolding No unfolding No unfolding
8 7.1 7.3 7.4
24 6.7 7.0 6.8
48 6.4 6.5 6.6
10 2 * 6.2 6.2
8 * 5.6 5.8
24 * 5.6 5.6
48 * 5.5 5.4
Fig. 4. GdnHCl concentrations at which the midpoint of unfolding
transitions occur as a function of pH. Incubation times 2 h (d), 8 h (j),
24 h (r), and 48 h (m).
Ó FEBS 2003 Stability of T. thermophilus seven-iron ferredoxin (Eur. J. Biochem. 270) 4739
different concentrations of GdnHCl were extrapolated to
give a T
m
value for each scan rate at 0
M
GdnHCl. Next,
these T
m
values at 0
M
GdnHCl for different scan rates were
extrapolated to (1/scan rate) ¼ 0, to give a T

m
correspond-
ing to infinite scan rate conditions. The method of
extrapolation to infinite scan rate has been used before to
eliminate irreversible time-dependent steps in protein-
unfolding reactions [21]. The thermal midpoints for FdTt
unfolding (at infinite scan rate) are 69 °C (pH 2.5), 91 °C
(pH 4), 114 °C (pH 7), and 90 °C (pH 10.5) (Fig. 5). Thus,
optimum stability of FdTt to heat also occurs around
neutral pH, and the thermal stability is, like the resistance to
chemical denaturation, dramatically reduced at lower pH.
Formation of linear clusters at high pH
On GdnHCl-induced unfolding of FdTt at high pH
(20 °C), new visible absorption bands at 520 nm
and 610 nm appeared transiently before complete
disappearance of the visible absorption occurred (Fig. 6A).
The new peaks formed with rate constants that depended
on the concentration of GdnHCl (pH 10, 20 °C). In 7.9
M
GdnHCl, the new bands formed rapidly and became more
intense (maximum intensity reached within 25 min) than at
6
M
GdnHCl where the formation was slower (maximum
intensity reached within 80 min). Identical absorption
bands at 520 nm and 610 nm have been observed transi-
ently in some other ferredoxins and in beef aconitase under
various conditions that perturb the protein structure
[19,20,22]. In those cases, it was concluded from EPR
and Mo

¨
ssbauer studies and comparison with small model
compounds [23] that the new absorption features resulted
from linear three-iron clusters bound to the unfolded,
or partially unfolded, polypeptide (Fig. 6A, inset). The
formation of the new absorption bands for FdTt correlated
with the disappearance of the far-UV CD signal, implying
Fig. 5. (A) T
m
vs. concentration of GdnHCl at pH 7 for three different
scan rates [(j)0.5°CÆmin
-1
;(d)0.25°CÆmin
-1
;(m)0.13°CÆmin
-1
)]
and (B) T
m
(at infinite scan rate) as a function of pH.
Fig. 6. Formation of linear species at high pH. (A)Visible absorption of
FdTt on addition of 7.9
M
GdnHCl (pH 10.5) after 1 min (solid line),
15 min (dotted line), and 24 h (dotted-dashed line) incubation (inset:
schematic drawing of a linear three-iron cluster). (B) EPR spectra
(110 K, 10 mW, 9.3 GHz) of folded FdTt in pH 10.5 (thin line) and
FdTt at pH 10.5 incubated for 15 min in 6.5
M
GdnHCl (thick line;

corresponding to dotted line in A).
4740 S. Griffin et al.(Eur. J. Biochem. 270) Ó FEBS 2003
that polypeptide unfolding triggered linear-cluster forma-
tion. Support for the idea that the new absorption bands
also correspond to linear three-iron clusters in FdTt comes
from EPR experiments (Fig. 6B). At pH 10.5, FdTt
exhibits a typical [3Fe)4S] cluster resonance at g ¼ 2.02
and a minor contribution at g ¼ 4.3 from adventitious
iron in solution. On incubation in 6.5
M
GdnHCl for
15 min (pH 10.5, 20 °C; to reach maximum intensity at
610 nm), the signal from the [3Fe)4S] center decreases
with the concomitant 15-fold increase in the g ¼ 4.3
resonance and a small peak at g ¼ 9.5 (features charac-
teristic of an S ¼ 5/2 system in a rhombic environment).
The S ¼ 5/2 signal is compatible with the presence of a
linear three-iron cluster and very similar to that reported
for purple aconitase, another seven-iron ferredoxin, and
model compounds [19,22,23]. The linear [Fe–S] cluster
remained in the unfolded protein for several hours (5–10 h;
20 °C, pH 10.5, 6.5–7.0
M
GdnHCl) before complete loss
of visible absorption occurred.
Discussion
Understanding and probing protein stability at high
temperatures and extreme conditions is relevant for a
variety of biochemical and biotechnological applications.
Intriguingly, the stability of a protein can be increased

by the optimization of a few interactions without large
structural modifications. To further understand the
mechanisms of increased stability in bacterial ferredoxins,
we characterized chemical and thermal denaturation of
the seven-iron T. thermophilus ferredoxin (FdTt) in
solution. This protein was recently crystallized [16], but
in vitro solution studies have been lacking. The data
presented here thus constitute the groundwork for future
experiments in which strategic FdTt mutants, designed
on the basis of the crystal structure, can be directly
compared with the biophysical behavior of the wild-type
form.
Our solution study shows that wild-type FdTt is
extremely stable to heat and chemical denaturants. An
explanation for this behavior, in terms of the sum of many
minor effects, is found on analyzing the crystal structure.
FdTt shares 64% sequence identity with the mesophilic
ferredoxin from Azotobacter vinelandii (FdAv), which is
a protein with the same overall structure as FdTt but
significantly less stable in solution [16]. Like other
mesophilic ferredoxins, FdAv has a stretch of 29 residues
at the C-terminus, which is absent from FdTt. This stretch
of residues protects the [3Fe)4S] cluster in FdAv from
solvent. Hence, the [3Fe)4S] cluster is more accessible to
solvent in FdTt, and shielding of this cluster from solvent
cannot be vital for FdTt function. Because FdTt has a
shorter C-terminus than FdAv, it has less accessible solvent
surfacearea,whichmayaidinresistancetoperturbation
[11,12]. On comparing the crystal structures, polar residues
at the surface of FdTt replace topologically equivalent

negatively charged residues in FdAv. Moreover, an a-helix
in FdTt, replacing a 3
10
-helix in FdAv, is stabilized with
alanine residues [16], and the (bab)
2
core of FdTt is
stabilized by additional hydrogen bonds between side
chains and the main chain, as compared with FdAv. Also,
FdTt has more glycine residues than FdAv, which may
minimize conformational strain in the folded state [16].
FdAv uses a cluster of glutamic and aspartic acid residues
to electrostatically interact with its physiological electron-
transfer partner. In FdTt, this region of the protein’s
surface is less charged. This difference in electrostatics is
thought to increase FdTt stability by reducing unfavorable
repulsions between negatively charged residues. The cor-
responding reduced electrostatic attraction between FdTt
and its electron-transfer partners may be compensated for
by faster protein diffusion rates at the higher temperatures
[16].
In addition to our biophysical study of FdTt presented
here, five other thermostable seven-iron ferredoxins have
been studied in vitro with respect to thermostability: from
the thermostable bacteria Bacillus schlegelii [24] and
Bacillus acidocaldarius [25] and the thermostable archaea
Acidianus ambivalens [13,20] and Sulfolobus sp. strain 7
[26]. Three of these proteins (and FdTt) are stable above
the boiling point of water at pH 7. Ferredoxin from
B. schlegelii begins to unfold at 90 °C [24], ferredoxin

from B. acidocaldarius is completely denatured at 88 °C
[25,27], ferredoxin from A. ambivalens (FdA; species with
zinc ion) has a thermal midpoint of 122 °C(pH7)[13],
and another ferredoxin from A. ambivalens (FdB; species
without zinc) has a thermal midpoint of 108 °C (pH 6.5)
[18]. The seven-iron ferredoxin from Sulfolobus sp. strain 7
has a thermal midpoint of 109 °C [28]. FdTt, with its
114 °C thermal midpoint at pH 7, is thus one of the most
stable seven-iron ferredoxins (in fact, among proteins in
general) investigated to date. Our GdnHCl-induced
unfolding data at 20 °C for FdTt can only be compared
withsimilarworkontheA. ambivalens ferredoxin (FdA;
species with zinc). In the case of that protein, the
GdnHCl-induced unfolding midpoints appear at 7.1
M
(pH 7, 20 °C), 2.3
M
(pH 2.5, 20 °C), and 6.3
M
(pH 10,
20 °C) [13]. Thus both proteins have similar extreme
resistance to chemical denaturants at pH 7 and higher.
The dramatic reduction in apparent stability at low pH for
both FdTt and A. ambivalens ferredoxin (using both
chemical and thermal perturbation) implies that electro-
static interactions contribute significantly to the proteins’
integrity at the higher pH values. This occurs because, at
low pH values, salt bridges are easily broken due to
protonation of aspartic and glutamic acid residues (which
have pK

a
values around 4).
On GdnHCl-induced polypeptide unfolding at high pH,
we find that the clusters in FdTt transiently rearrange into
intermediate species before complete cluster degradation
occurs. Clusters with the same spectroscopic features as
we found in FdTt at high pH have been shown in other
studies to be linear [3Fe)4S] clusters still bound to the
polypeptides (Fig. 6A, inset) [19]. Our spectroscopic data
(absorption and EPR) support the suggestion that in
FdTt also the cubic clusters rearrange into linear species
on protein unfolding under alkaline conditions. The
presence of a linear [3Fe)4S] cluster was first observed
in the protein bovine heart aconitase, at high pH where
the protein structure was perturbed [22]. Recently, the
same linear cluster was discovered on in vitro unfolding of
seven-iron ferredoxins from A. ambivalens and S. acidoc-
aldarius [13,19,20]. The observation of a linear cluster in
FdTt and other seven-iron ferredoxins, and the recent
Ó FEBS 2003 Stability of T. thermophilus seven-iron ferredoxin (Eur. J. Biochem. 270) 4741
observations of this cluster in [2Fe)2S] ferredoxins from
Aquifex aeolicus ([17] and unpublished data), suggest a
more general relevance of this type of linear cluster in
nature. In Table 2, we summarize the known systems and
solution conditions in which linear three-iron clusters have
been observed in vitro. No biological function for linear
[3Fe)4S] clusters is yet known, although reconstituted,
recombinant human cytosolic iron regulatory protein 1
has been found to contain such a cluster under physio-
logical conditions [29]. We speculate that [Fe–S] cluster

rearrangements induced by protein-conformational chan-
ges may be used for regulatory purposes in vivo.The
linear cluster may be a storage or transport form for iron
and sulfide in the cells ready for use in resynthesis of cubic
(functional) clusters.
Summary
Examination of the crystal structure of the seven-iron
ferredoxin from T. thermophilus has suggested that it
represents the minimal functional unit of this type of
protein[16].Inagreement,wefindFdTttobeaverystable
protein in solution in vitro: temperatures above boiling or
high denaturant concentrations and long incubation times
are necessary to perturb it. From our work in solution at
different pH values, it is clear that electrostatic interactions
play a significant role in governing the high stability of
FdTt. As unfolding is very slow (hours), even in high
concentrations of denaturant, there appears to be a kinetic
barrier to FdTt unfolding. Slow unfolding kinetics may be a
general mechanism governing high stability of thermostable
proteins. On polypeptide unfolding at high pH, linear
three-iron clusters form in FdTt. Recent discoveries of the
transient appearance of such linear clusters in many
different iron–sulfur proteins imply that they may be of
biological relevance.
Acknowledgements
This work was supported by the National Institute of Health
(GM5966301A2) (P.W S.), the Louisiana Board of Regents
(C.L.H.), and the Newcomb College Fellows Program (Tulane
University, New Orleans, Louisiana) (S.G.). We thank Ah-lim Tsai
(University of Texas Medical School, Houston) and John S. Olson

(Rice University, Houston) for help with EPR experiments.
References
1. Bertini, I., Cowan, J.A., Luchinat, C., Natarajan, K. &
Piccioli, M. (1997) Characterization of a partially unfolded high
potential iron protein. Biochemistry 36, 9332–9339.
2. Goedken, E.R., Keck, J.L., Berger, J.M. & Marqusee, S.
(2000) Divalent metal cofactor binding in the kinetic folding
trajectory of Escherichia coli ribonuclease HI. Protein Sci. 9,
1914–1921.
3. Leckner, J., Bonander, N., Wittung-Stafshede, P., Malm-
strom, B.G. & Karlsson, B.G. (1997) The effect of the metal ion
on the folding energetics of azurin: a comparison of the native,
zinc and apoprotein. Biochim. Biophys. Acta 1342, 19–27.
4. Pozdnyakova, I., Guidry, J. & Wittung-Stafshede, P. (2000)
Copper triggered beta-hairpin formation. Initiation site for azurin
folding? J. Am. Chem. Soc. 122, 6337–6338.
5. Wittung-Stafshede, P., Hill, M.G., Gomez, E., Di Bilio, A.,
Karlsson, G., Leckner, J., Winkler, J.R., Gray, H.B. &
Malmstrom, B.G. (1998) Reduction potentials of blue and purple
copper proteins in their unfolded states: a closer look at
rack-induced coordination. J. Biol. Inorg. Chem. 3, 367–370.
6. Robinson, C.R., Liu, Y., Thomson, J.A., Sturtevant, J.M. &
Sligar, S.G. (1997) Energetics of heme binding to native and de-
natured states of cytochrome b
562
. Biochemistry 36, 16141–16146.
7. Frazzon, J. & Dean, R.D. (2001) Feedback regulation of iron-
sulfur cluster biosynthesis. Proc. Natl Acad. Sci. USA 98,
14751–14753.
8. Mortenson, L.E., Valentine, R.C. & Carnahan, J.E. (1962) An

electron transport factor from Clostridium pasteurianum. Biochem.
Biophys. Res. Commun. 7, 448–452.
9. Tagawa, K. & Arnon, D.I. (1962) Ferredoxins as electron carriers
in photosynthesis and in the biological production and con-
sumption of hydrogen gas. Nature (London) 195, 537–543.
10. Meyer, J. (2001) Ferredoxins of the third kind. FEBS Lett. 509,
1–5.
11. Jaenicke, R. & Bohm, G. (1998) The stability of proteins in
extreme environments. Curr. Opin. Struct. Biol. 8, 738–748.
12. Vogt, G. & Argos, P. (1997) Protein thermal stability: hydrogen
bonds or internal packing? Fold. Des. 2, S40–S46.
13. Moczygemba, C., Guidry, J., Jones, K., Gomes, C., Teixeira, M.
& Wittung-Stafshede, P. (2001) High stability of a ferredoxin from
the hyperthermophilic archaeon Acidianus ambivalens: involve-
ment of electrostatic interactions and cofactors. Protein Sci. 10,
1539–1548.
14. Szilagyi, A. & Zavodszky, P. (2000) Structural differences between
mesophilic, moderately thermophilic and extremely thermophilic
protein subunits: results of a comprehensive survey. Struct. Fold.
Des. 8, 493–504.
Table 2. Proteins found to accommodate linear three-iron clusters in vitro and the corresponding solution conditions.
Protein Origin Normal cluster(s) Conditions for linear cluster
Ferredoxin[this work] T. thermophilus [3Fe)4S]/[4Fe)4S] High pH and [GdnHCl]
Ferredoxin [19] S acidocaldarius [3Fe)4S]/[4Fe)4S] High pH and [GdnHCl]
Ferredoxin [20] A. ambivalens [3Fe)4S]/[4Fe)4S] High pH and [GdnHCl]
Aconitase [22] Bovine heart [4Fe)4S] High pH or [urea]
Dihydroxy acid dehydratase [30] E. coli [4Fe)4S] Exposure to oxygen
Ferredoxin [31] R. marinus [3Fe)4S] High [GdnHCl]
Iron regulatory protein 1 [29] Human cytosol [3Fe)4S] Physiological
Fd1, Fd4 and Fd5 [17]

a
A. aeolicus [2Fe)2S] High pH and [GdnHCl]
Ferredoxin
a
Spinach [2Fe)2S] High pH and [GdnHCl]
Ferredoxin
a
C. pasteurianum [2Fe)2S] High pH and [GdnHCl]
a
Unpublished data.
4742 S. Griffin et al.(Eur. J. Biochem. 270) Ó FEBS 2003
15. Tabata, K., Kosuge, T., Nakahara, T. & Hoshino, T. (1993)
Physical map of the extremely thermophilic bacterium Thermus
thermophilus HB27 chromosome. FEBS Lett. 331, 81–85.
16. Macedo-Ribeiro, S., Martins, B., Pereira, P.J.B., Buse, G., Huber,
R. & Soulimane, T. (2001) New insights into the thermostability of
bacterial ferredoxins: high-resolution crystal structure of the
seven-iron ferredoxin from Thermus thermophilus. J. Biol. Inorg.
Chem. 6, 663–674.
17. Higgins, C., Meyer, J. & Wittung-Stafshede, P. (2002) Exceptional
stability of a [2Fe-2S] ferredoxin from hyperthermophilic bacter-
ium Aquifex aeolicus. Biochim. Biophys. Acta 1599, 82–89.
18. Janssen, S., Trinca
˜
o, J., Teixeira, M., Scha
¨
fer, G. & Anemu
¨
ller, S.
(2001) Ferredoxins from the archaeon Acidanus ambivalens:

overexpression and characterization of the non-zinc-containing
ferredoxin FdB. Biol. Chem. 382, 1501–1507.
19. Jones, K., Gomes, C., Huber, H., Teixeira, T. & Wittung-
Stafshede, P. (2002) Formation of a linear [3Fe-4S] cluster in a
seven-iron ferredoxin triggered by polypeptide unfolding. J. Biol.
Inorg. Chem. 7, 357–362.
20. Wittung-Stafshede, P., Gomes, C. & Teixeira, M. (2000) Stability
and folding of the ferredoxin from the hyperthermophilic
archaeon Acidianus ambivalens. J. Inorg. Biochem. 78, 35–41.
21. La-Rosa, C., Milardo, D., Grasso, D., Guzzi, R. & Sprotelli, L.
(1995) Thermodynamics of the thermal unfolding of azurin.
J. Phys. Chem. 99, 14864–14870.
22. Kennedy,M.C.,Kent,T.,Emptage,M.,Merkle,H.,Beinert,H.&
Mu
¨
nck, E. (1984) Evidence for the formation of a linear [3Fe-4S]
cluster in partially unfolded aconitase. J. Biol. Chem. 259, 14463–
14471.
23. Hagen, K., Watson, A. & Holm, R.H. (1983) Synthetic routes
to Fe
2
S
2
,Fe
3
S
4
,Fe
4
S

4
,andFe
6
S
9
clusters from the common
precursor [Fe (SC
2
H
5
)
4
]
2–
: structures and properties of
[Fe
3
S
4
(SR)
4
]
3–
and [Fe
6
S
9
(SC
2
H

5
)
2
]
4–
, examples of the newest
types of Fe-S-SR clusters. J. Am. Chem. Soc. 105, 3905–3915.
24. Aono, S., Bentrop, D., Bertini, I., Donaire, A., Luchinat, C.,
Niikura, Y. & Rosato, A. (1998) Solution structure of oxidized
Fe
7
S
8
ferredoxin from thermophilic bacterium Bacillus schlegelii
by H
1
NMR spectroscopy. Biochemistry 37, 9812–9826.
25. D’Auria, S., Rossi, M., Herman, P. & Lakowicz, J. (2000) Pyru-
vate kinase from the Thermophilic eubacterium Bacillus acid-
ocaldarius as probe to monitor the sodium concentrations in the
blood. Biophys. Chem. 84, 167–176.
26. Fujii, T., Hata, Y., Oozeki, M., Moriyama, H., Wakagi, T.,
Tanaka,N.&Oshima,T.(1997)Thecrystalstructureofzinc-
containing ferredoxin from the thermoacidophilic archaeon Sul-
folobus sp. strain 7. Biochemistry 36, 1505–1513.
27. Schlatter, D., Waldvogel, S., Zu
¨
lli, F., Suter, F. & Portmann. &
Zuber, H. (1985) Purification, amino-acid sequence and some
properties of the ferredoxin isolated from Bacillus acidocaldarius.

Biol. Chem. Hoppe-Seyler 366, 223–231.
28. Kojoh, K., Matsuzawa, H. & Wakagi, T. (1999) Zinc and an
N-terminal extra stretch of the ferredoxin from a thermo-
acidophilic archaeon stabilize the molecule at high temperature.
Eur. J. Biochem. 264, 85–91.
29. Gailer, J., George, G., Pickering, I., Prince, R., Kohlhepp, P.,
Zhang, D., Walker, A. & Winzerling, J. (2001) Human cytosolic
iron regulatory protein 1 contains a linear iron-sulfur cluster.
J. Am. Chem. Soc. 123, 10121–10122.
30. Flint, D., Emptage, M., Finnegan, M., Fu, W. & Johnson, M.
(1993) The role and properties of the iron-sulfur cluster in
Escherichia coli dihydroxy-acid dehydratase. J. Biol. Chem. 268,
14732–14742.
31. Pereira, M., Jones, K., Campos, M., Melo, A., Saraiva, L., Louro,
R., Wittung-Stafshede, P. & Teixeira, M. (2002) A ferredoxin from
the thermohalophilic bacterium Rhodothermus marinus. Biochim.
Biophys. Acta 1601,1–8.
Ó FEBS 2003 Stability of T. thermophilus seven-iron ferredoxin (Eur. J. Biochem. 270) 4743