Báo cáo khoa học: Antioxidant Dps protein from the thermophilic cyanobacterium Thermosynechococcus elongatus An intrinsically stable cage-like structure endowed with enhanced stability potx

Antioxidant Dps protein from the thermophilic
cyanobacterium Thermosynechococcus elongatus
An intrinsically stable cage-like structure endowed with enhanced
stability
Stefano Franceschini*, Pierpaolo Ceci*, Flaminia Alaleona, Emilia Chiancone and Andrea Ilari
C.N.R. Institute of Molecular Biology and Pathology, University of Rome ‘La Sapienza’, Italy
The family of DNA-binding proteins from starved cells
(Dps) is part of a complex bacterial defence system that
protects DNA against oxidative damage [1–3]. Dps
proteins use hydrogen peroxide to oxidize intracellular
Fe(II) and thereby simultaneously remove the two
molecules that produce highly toxic hydroxyl radicals
via the Fenton reaction [4,5]. Subsequent sequestration
of the ferric ions thus formed in the protein cavity as
a micellar hydroxide core completes the detoxification
process. Thus, the structural features central to the
detoxifying activity of Dps are the characteristic cage-
like dodecameric assembly endowed with 23-tetrahedral
symmetry and the ferroxidase centre with its unique
intersubunit location at the twofold symmetry axes [2].
Keywords
Dps from Thermosynechococcus elongatus;
hydrogen peroxide; iron oxidation;
thermostability; X-ray structure
Correspondence
A. Ilari, Istituto di Biologia e Patologia
Molecolari CNR Dipartimento di Scienze
Biochimiche, Universita
`
di Roma ‘La
Sapienza’, P.le A. Moro, 5 00185 Rome,

Italy
Fax: +39 06 444 0062
Tel. +39 06 494 0543 ⁄ 499 10761
E-mail:
Database
The atomic coordinates and structure fac-
tors have been deposited in the Protein
Data Bank, Research Laboratory for Struc-
tural Bioinformatics, Rutgers University,
New Brunswick (,
PDB code 2C41)
*These authors contributed equally to this
work
(Received 21 July 2006, revised 4 September
2006, accepted 5 September 2006)
doi:10.1111/j.1742-4658.2006.05490.x
DNA-binding proteins from starved cells (Dps proteins) protect bacteria
primarily from oxidative damage. They are composed of 12 identical
subunits assembled with 23-symmetry to form a compact cage-like struc-
ture known to be stable at temperatures > 70 °C and over a wide pH
range. Thermosynechococcus elongatus Dps thermostability is increased
dramatically relative to mesophilic Dps proteins. Hydrophobic interac-
tions at the dimeric and trimeric interfaces called Dps-like are replaced
by salt bridges and hydrogen bonds, a common strategy in thermophiles.
Moreover, the buried surface area at the least-extended Dps-like inter-
face is significantly increased. A peculiarity of T. elongatus Dps is the
presence of a chloride ion coordinated with threefold symmetry-related
arginine residues lining the opening of the Dps-like pore toward the
internal cavity. T. elongatus Dps conserves the unusual intersubunit ferr-
oxidase centre that allows the Dps protein family to oxidize Fe(II) with

hydrogen peroxide, thereby inhibiting free radical production via Fenton
chemistry. This catalytic property is of special importance in T. elongatus
(which lacks the catalase gene) in the protection of DNA and photosys-
tems I and II from hydrogen peroxide-mediated oxidative damage.
Abbreviations
Dps, DNA-binding proteins from starved cells; Dps-Te, DNA-binding protein from starved cells of Thermosynechococcus elongatus.
FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS 4913
The cage-like structure of the Dps dodecamer is
expected to be resistant to dissociation into subunits
because closed symmetric systems, in which intersub-
unit interactions are maximized, tend to have lower
energies than asymmetric assemblies. Detailed studies,
performed as a function of pH, on Listeria innocua [6]
and Mycobacterium smegmatis Dps [7] confirm this con-
tention, but also highlight significant differences in the
tendency of the dodecamer to dissociate into subunits.
Thus, L. innocua Dps forms dimers only below pH 2.0
and monomers below pH 1.0, whereas dissociation of
M. smegmatis Dps into dimers is evident at pH 5.0
and proceeds to the monomer stage at pH 4.0. No spe-
cific information is available on the stability of the
dodecameric assembly as a function of temperature,
although this property is currently exploited during the
purification of Dps proteins. To investigate Dps ther-
mostability, a protein from the thermophilic cyano-
bacterium Thermosynechococcus elongatus (Dps-Te)
was chosen.
Thermosynechococcus elongatus inhabits Japanese
hot springs and grows optimally at around 55 °C [8].
It is a model system for studying the interplay of gen-

etic, biochemical and physiological phenomena in pho-
tosynthesis due to the availability of the complete
genome sequence [9], but it is also the source of highly
stable protein complexes that have been crystallized,
e.g. those of photosystems I and II [10,11]. The occur-
rence of oxygenic photosynthesis in T. elongatus adds
to the reaction of free Fe(II) with hydrogen peroxide
as an important source of reactive oxygen species.
Thus, superoxide radicals, hydrogen peroxide and
hydroxyl radicals are generated as a result of the pho-
tosynthetic transport of electrons from water to plasto-
quinone such that photosystems I and II are the main
targets of photodamage [12–17].
The T. elongatus genome contains two putative Dps-
encoding genes. The antioxidant activity provided by
the corresponding proteins is likely to have particular
importance in protecting DNA and photosystems I
and II against oxidative damage. In fact, the organism
does not appear suited to manage hydrogen peroxide
given the absence of a catalase gene coupled to the
presence of two superoxide dismutase genes [9]. Fol-
lowing expression in Escherichia coli, Dps-Te has been
characterized in terms of its X-ray crystal structure,
thermostability and antioxidant activity at various pH
values.
Analysis of the Dps-Te structure showed an
increased number of salt bridges at the subunit interfa-
ces with respect to mesophilic members of the family.
Such interactions, which are known to promote ther-
mostability in a number of proteins from thermophiles

and hyperthermophiles therefore play a crucial role in
conferring additional stability to an intrinsically stable
cage-like structure. In the structural comparison spe-
cial attention has been paid to the two types of pore
formed at the threefold interfaces. Pores at the
so-called ferritin-like interface are all of similar size and
are lined with negatively charged residues pointing to a
common function in the iron-uptake process; those at
the so-called Dps-like interface show marked variabil-
ity in their dimensions and chemical nature. Their
function may therefore differ in different organisms.
Results
Sequence analysis
Alignment of the Dps-Te sequence with the sequences
of six members of the Dps family was performed using
multalin [18] and is presented in Fig. 1A. The Dps-
Te sequence was compared with: (a) Dlp2 from Bacil-
lus anthracis (35% sequence identity) [19], used as
search model to solve the Dps-Te structure by molecu-
lar replacement; (b) L. innocua and M. smegmatis Dps
(sequence identity 30 and 24.2%, respectively) [20,21],
whose stability has been studied previously [6,7] and
E. coli Dps (22.7% sequence identity), the family pro-
totype; (c) Dps from the cyanobacterium Trichodes-
mium erythraeum (30% sequence identity) [22]; and (d)
Dps from the halophile Halobacterium salinarum (32%
sequence identity) [23]. Dps-Te contains all the distinc-
tive residues of the Dps family despite the low degree
of identity, namely the residues diagnostic of the inter-
subunit ferroxidase centre (His33, Asp60, His45 and

Glu64), the near-by Trp34 residue, and aspartates 125
and 130 lining the pore along the threefold symmetry
axes. Alignment also shows that Dps-Te lacks the long,
positively charged N-terminus involved in Dps–DNA
complex formation in E. coli Dps [3,24]. Further analy-
sis of the sequence using the Predict Protein server
() shows the presence of
three potential protein kinase C phosphorylation sites,
namely TLK (residues 6–8 and 14–16) and TVK
(residues 94–96). The first two are positioned on the
N-terminal tail and the third is on the BC loop, located
on the surface of the molecule in the assembled protein.
Monomer fold and dodecameric assembly
The Dps-Te monomer folds into the four-helix bundle
typical of Dps proteins and ferritins (Fig. 1B). The four
helices, A–D, are stabilized mainly by hydrophobic
interactions, an additional short a helix (BC) is in the
long loop connecting helices B and C. Superposition of
The thermostable T. elongatus Dps S. Franceschini et al.
4914 FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS
the Dps-Te monomer onto those of B. anthracis Dlp2,
L. innocua Dps and M. smegmatis Dps yields very
small RMSD values (0.893, 0.910 and 1.07 A
˚
, respect-
ively). A higher RMSD value pertains to superposition
of the Dps-Te monomer onto the E. coli Dps (1.54 A
˚
).
Figure 1B shows that the only significant differences

occur in the N- and C-terminal regions.
Twelve monomers assemble to form a hollow pro-
tein cage with 23-tetrahedral symmetry (external and
internal diameters 90 and 45 A
˚
, respectively). The
threefold symmetry-related subunits make two types of
interaction. One defines the so-called ‘ferritin-like’
interface because the interactions resemble those of
ferritin subunits along the threefold symmetry axes
[20], the other defines the interface specific to this pro-
tein family named ‘Dps-like’.
In Dps-Te, residues 2–7 (visible only in subunit C)
form a structured tail that protrudes from the dodeca-
meric assembly towards the solvent. These residues
have been refined without imposing non-crystallo-
graphic symmetry (NCS) restraints, indicating that
they assume different conformations. In the other
known Dps crystal structures the N-terminus is either
involved in interactions with the protein scaffold or is
not visible because of its flexibility. Thus, in L. innocua
Dps, the first six residues are not visible, whereas in
Dlp2 the first three are anchored to the C helix of the
same subunit via the Ser2 OH group, which is hydro-
gen bonded to the main chain oxygen of Val115
(Fig. 1B). In the E. coli Dps X-ray structure, the first
eight residues (containing two positively charged
lysines) are not visible, residues 9–15 are oriented
towards the solvent and residues 16–21 form a cove
that is bent toward the ‘ferritin-like’ interface. In

M. smegmatis Dps, the N-terminal tail formed by the
first 14 amino acids is likewise bent towards the
ferritin-like interfaces. Figure 1B also shows that
Dps-Te is characterized by a relatively long C-terminus
(residues 151–158) which is visible in all subunits and
forms a hook bent towards the Dps-like interface. The
longest C-terminal tail is found in M. smegmatis Dps
[7,21]. The few residues of this long tail that are visible
are likewise bent towards the Dps-like interface.
Trimeric ‘ferritin-like’ interface
The surface area buried at the ferritin-like interface is
quite extended (966 A
˚
2
per monomer) as it comprises
A
B
Fig. 1. Primary structure alignment (A) and monomer fold (B) of Dps proteins. (A) Proteins from T. elongatus (Dps-Te), B. anthracis (Dlp2),
L. innocua (Dps-Li), E. coli (Dps-Ec), M. smegmatis (Dps-Ms), H. salinarum Dps (DPS-HS) and T. erithraeum Dps (Dps-Er). The residues of
the ferroxidase centre are depicted in red, those lining the two types of pore are shown in green. The a helices are indicated by upper case
letters. (B) Structural overlay of the Ca trace of the Dps-Te monomer (Te, blue) with those of Dlp2 (Ba, red), Dps-Li (Li, green), Dps-Ec (Ec,
azure) and Dps-Ms (Ms, salmon). The N- and C-terminal regions are indicated by (N) and (C), respectively. Pictures were generated using
PYMOL (Delano Scientific LLC, San Carlos, LA; ).
S. Franceschini et al. The thermostable T. elongatus Dps
FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS 4915
the CD loop, the beginning of the D helix and the last
part of the B helix. It is stabilized by hydrophobic and
hydrophilic interactions and displays the features des-
cribed for E. coli and M. smegmatis Dps and Dlp2
[2,7,19]. The most buried hydrophobic side chains

belong to the highly conserved Trp144 residues Val136
(D helix), Ala117 (CD loop), Leu67 and Leu69 (B
helix). The hydrophilic residues stabilizing the interface
are the conserved Arg65 and Asp125 residues and
Asp70, Arg143 and Gln140, which are not conserved
in the other Dps proteins considered. In particular
(Fig. 2A, panels 1 and 2), Arg143 forms two strong
electrostatic bonds with Asp70 (distances: Asp O-d1–
Arg N-g1 ¼ 2.8 A
˚
, Asp O-d2–Arg N-g2 ¼ 2.8 A
˚
). As
in the other Dps proteins considered, the conserved
Arg65 residue contributes to stabilize the ferritin-like
interface (Table 1). In Dps-Te it is hydrogen bonded
to Gln140 (distance Gln140 O-e2–Arg65 N-g2 ¼
2.84 A
˚
) which also forms a weak electrostatic interac-
tion with Asp125, another conserved residue (distance
Asp125 O-d2–Arg65 N-e2 ¼ 4.8 A
˚
).
Trimeric ‘Dps-like’ interface
In all the mesophilic Dps proteins whose structure has
been solved to date the trimeric Dps-like interfaces are
the least extended ones and are stabilized mostly by
hydrophobic interactions. Thus, in L. innocua and
E. coli Dps hydrophilic interactions are absent and in

M. smegmatis Dps there is only a strong salt bridge
between Arg99 and Glu157 and two hydrogen bonds
(Table 1). At variance with these proteins, the Dps-like
interface of Dps-Te is stabilized by a large number of
hydrophilic interactions and is the most extended one
(1711 A
˚
2
per monomer), because it comprises the
C-terminal tail, the last part of the D helix and
the first B helix residues (Fig. 2B, panels 1 and 2).
The hydrophobic residues buried most deeply at the
interface are Val95, on the C helix and Trp144,
Phe145, Phe149 on the D helix. The residues engaged
in electrostatic interactions are Tyr37 and Gly38 on
the AB loop, Asp43, Arg42 and Glu50 on the first
part of the B helix, Lys96 on the last part of the
CD loop, Glu148 on the last part of the D helix,
and Gly153 and Asp154 on the C-terminus. Arg42
provides the strongest electrostatic interactions at this
interface because it forms a salt bridge via N-g1
with O-e2 of Glu50 (distance N-g1–O-e2 ¼ 2.8 A
˚
).
Moreover, three hydrogen bonds are formed between
Lys96 and Asp154, Glu148 and Gly38 and Tyr37
and Gly153 (Table 1). In the halophilic H. salinarum
Dps, the Dps–like interface is less extended (1008
A
˚

2
⁄ monomer) than in Dps-Te but is likewise stabil-
ized by a great number of hydrophilic interactions
(Table 1).
In the Dps protein from the N
2
-fixing cyanobacteri-
um T. erythraeum the interface lacks many of the resi-
dues involved in electrostatic interactions in Dps-Te,
i.e. Arg42 is Asn62 and Asp43 is Gln63.
Dimeric interface and ferroxidase centre
The dimeric interface is formed by helices A and B
and by the short BC helix placed at the centre of the
long loop connecting helices B and C (Fig. 2C). It
contains the two symmetry-related characteristic inter-
subunit ferroxidase centres. The absence of peaks with
values > 4 r in the F
obs
) F
calc
difference Fourier map
calculated before the introduction of water molecules
indicates that the ferroxidase centres are iron free.
However, two water molecules (A and B), placed at a
distance of  3A
˚
, are present at the iron-binding sites
(Fig. 3A). The A water molecule is coordinated by
N-e2 of His33 (distance N-e2–O ¼ 2.7 A
˚

) and by the
carboxylic oxygen atoms O-d1 and O-d2 of Asp60 (dis-
tances O-d1–O ¼ 2.53 A
˚
and O-d1–O ¼ 3.25 A
˚
).
Water molecule B is placed at 3.16 and 2.57 A
˚
,
respectively, from the Glu64 carboxylic oxygen O-e2
and O-e1 and at 2.86 A
˚
from N-e2 of His45.
The surface area buried upon dimerization (1180 A
˚
2
per monomer) is similar to those calculated for the
other members of the family [7]. In Dps-Te it is stabil-
ized mostly by hydrophilic interactions, whereas in the
mesophilic Dps proteins the dimeric interface is mainly
hydrophobic (Table 1, Fig. 2C, panel 2). In particular,
Lys30 is salt bridged to Asp60 (distance O-d2–N-f ¼
2.78 A
˚
), the Asp76 carboxylic oxygen forms a salt
bridge with the Lys31 nitrogen atom (distance O-d2–
N-f ¼ 2.83 A
˚
) and the O-d1 carboxylic oxygen of

Asp76 is hydrogen bonded to the Gly91 nitrogen
atom. Interestingly, also in the Dps protein from the
halophilic archaebacterium H. salinarum the dimeric
interface is stabilized mostly by hydrophilic interac-
tions and by two salt bridges between Arg8 and Glu56
(N-g2–O-e2 ¼ 2.8 A
˚
) of the twofold symmetry-related
subunits (Table 1).
The ‘ferritin-like’ and ‘Dps-like’ pores
Residues at the ferritin-like interface of Dps-Te form a
pore that connects the oligomer cavity to solvent and
is lined by negatively charged residues. In particular,
the opening on the protein surface ( 13.5 A
˚
diameter)
contains Glu118 and Glu122, respectively, whereas
that facing the protein cavity ( 7.5 A
˚
diameter) con-
tains the highly conserved Asp130 residues.
The thermostable T. elongatus Dps S. Franceschini et al.
4916 FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS
A
B
C
Fig. 2. Trimeric ferritin-like (A), trimeric Dps-like (B) and dimeric (C) interfaces of T. elongatus Dps. Panel 1, view along the interfaces; panel
2, blow-up indicating relevant interactions as detailed in the text. Pictures were generated using
PYMOL.
S. Franceschini et al. The thermostable T. elongatus Dps

FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS 4917
Comparison of the Dps-Te ‘ferritin-like’ pore with
those of other family members, such as L. innocua,
E. coli, H. salinarum and M. smegmatis Dps, shows
that its structural features are largely conserved. Thus,
the length is  10 A
˚
with the exception of H. salinarum
Dps in which it is  18 A
˚
, the diameter of outer open-
ing ranges between 9.0 and 13.5 A
˚
, and the opening on
the protein cavity between 7.0 and 8.0 A
˚
(Table 2).
These values pertain to distances between Ca atoms. In
particular, the negatively charged residues lining both
openings are conserved in all Dps proteins.
At the ‘Dps-like’ interface the subunits form another
pore with threefold symmetry. This pore shows great
variability among the Dps proteins considered
(Table 2). Thus, the length of the pore ranges from 7
to 21 A
˚
, whereas the size of the openings on the pro-
tein cavity and surface vary between 5 and 9.0 A
˚
,

respectively. Because these values refer to distances
between Ca atoms, in solution the pore is likely to
assume different conformations, e.g. ‘opened’ or
‘closed’, depending on the rotational conformations of
the residues lining the pore.
The nature of the residues along the Dps-like pores
is likewise variable. Interestingly, in the two extremo-
philic Dps proteins the opening on the protein surface
is lined with a hydrophobic residue (Val157 in Dps-Te
and Leu181 in H. salinarum Dps), whereas that on the
protein cavity contains the positive charges of the sym-
metry-related arginine residues (Arg42 in Dps-Te and
Arg62 in H. salinarum Dps). In Dps-Te, these Arg42
residues bind a chloride ion which occludes the pore
opening (distances: N-g1–Cl ¼ 3.2 A
˚
and N-g2–Cl ¼
3.4 A
˚
) (Fig. 3B).
Figure 4 shows that in the proteins analysed with
the exception of M. smegmatis Dps the Dps-like pores
have a constriction. In E. coli Dps, as described by
Grant et al. [2], the constriction is located near the
protein cavity and is lined by hydrophobic residues
(Ala61). In Dps-Te and L. innocua Dps the location is
similar, whereas in H. salinarum Dps the constriction
is in the middle of the pore. In the proteins, the con-
striction is lined by charged residues (Arg42, Asp43 in
Dps-Te and Arg61 Asp62 in H. salinarum Dps) or by

hydrophilic ones (Thr41 in L. innocua Dps).
Table 1. Electrostatic interactions stabilizing the interfaces in T. elongatus, H. salinarum, and E. coli Dps.
Dimeric interfaces (A
˚
) Ferritin-like interfaces (A
˚
) Dps-like interfaces (A
˚
)
T. elongatus Dps
Trp34 N-e1–O-d2 Asp60 2.7 Asp70 O-d1–N-g1 Arg143 2.8 Asp43 O-d1–N-e Arg42 2.4
Lys30 N-f –O-d2 Asp60 3.0 Asp70 O-d2–N-g2 Arg143 2.7 Asp43 O-d2 – H2O1206 2.7
Lys31 N-f –O-d2 Asp76 2.9 Arg65 N-d1–O-g1 Gln140 2.8 H2O1206–O Pro39 2.4
Gly91 N–O-d1 Asp76 2.8 Arg65 N-d2–O-g1 Gln140 2.9 Arg42 N-g1–O-e2 Glu50 2.8
Tyr37 O-g – O Gly153 2.7
Gly38 N–O Gln148 2.7
Lys96 N–O Asp154 3.0
Val95 N–O Asp154 2.4
H. salinarum Dps
Arg8 N-g2–O-e2 Glu56 2.8 Asp128 O-d1–N-g1 Arg21 2.9 Asp62 O-d2–N-g1 Arg61 2.8
Arg8 N–O Val112 2.6 Glu131 O-e1–N-g1 Arg21 2.6 His168 N-d1–O Val55 3.0
Ala9 O–N Val112 2.9 Arg134 N-g1–O Leu22 3.0 His168 O–N Gly57 2.8
Ala11 N–O-d1 Asp111 2.9 Arg134 N-g2–O Leu22 2.8 Asp178 O- d2–N Ile115 2.8
Tyr42 O-g
–O-g Tyr45 2.7 Glu138 O-e1–N-e2 His146 2.7 Thr174 O-c1–O-e1 Glu59 2.6
Tyr45 O-g –O-e2 Glu75 2.5 Glu160 O-e1–N-g1 Arg84 2.8 Leu175 N–O-e2 Glu59 2.8
His46 N-e2–N-e2 His46 2.7 Glu160 O-e1–N-g2 Arg84 2.6 Leu175 O–N-g1 Arg116 2.8
His52 N- e2–O- e1 Glu83 2.9 Glu167 O- e1–O-c Ser17 2.8
Trp53 N- e1–O-d1 Asp79 2.8 Asp172 O–N-g1 Arg8 3.0
Ala94 N–O- e2 Glu110 2.9 Asp172 O-d1–N-g2 Arg8 2.9

Ser95 N–O- e1 Glu110 2.9
Glu97 N–O- e1 Gln100 2.8
E. coli Dps
a
Lys48 N-f–O-d2 Asp78 2.8 Arg18 N-e –O-d1 Asp123 2.7
Arg70 N-g2–O-d2 Asp78 2.9 Arg83 N-g1–O-d1 Asp156 2.8
Arg133 N-g1–O-d1 Asp20 2.9
Lys134 N-f –O-d2 Asp20 3.0
a
Taken from Ceci et al. [3]
The thermostable T. elongatus Dps S. Franceschini et al.
4918 FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS
State of association as a function of pH
and temperature
The state of association was studied over the pH range
1.0–7.0 by HPLC-gel filtration. Representative elution
profiles presented in Fig. 5A show that the dodecamer-
ic architecture of Dps-Te is stable between pH 7.5 and
3.0. At pH 2.5, the chromatogram shows two addi-
tional small peaks, corresponding to a dimeric and a
high molecular mass species. However, the decrease in
the area of the peaks clearly points to marked precipi-
tation of the protein on the column. At pH 2.0, the
dimer peak disappears and the amount of precipitated
material increases. At pH 1.0, the elution pattern dis-
plays only one peak corresponding to a monomeric
species whose area is indicative of almost complete
precipitation of Dps-Te. It is worth noting that preci-
pitation does not take place in the case of L. innocua
and M. smegmatis Dps [6,7].

HPLC-gel filtration was supplemented by CD experi-
ments in the near-UV region (Fig. 5B). The CD spectra
at pH 7.0 showed two positive peaks at 289 and
282 nm due to
1
L
b
vibronic transitions of the Trp34
and Trp144 residues, and a negative peak at 297 nm
due to
1
L
a
vibronic transitions. In the other members
of the Dps family the tryptophan
1
L
b
vibronic trans-
itions produce negative peaks [6]. Spectra measured at
acid pH values show that the signal corresponding
to the Trp
1
L
a
and
1
L
b
vibronic transition decreases

dramatically at pH 2.0 and is lost completely at pH 1.0.
Protein stability as a function of temperature was
monitored in the far-UV CD region. The transition
from the native to the denatured state could not be
followed between pH 7.0 and 4.0 because of the high
stability of Dps-Te even at 100 °C. Thus, heat-
induced unfolding of Dps-Te was monitored at
pH 3.0 where the quaternary structure is conserved at
room temperature (Fig. 6). The denaturation process
of the mesophilic L. innocua and E. coli proteins was
followed under the same experimental conditions.
Whereas Dps-Te and L. innocua Dps undergo full
Fig. 3. Ferroxidase centre (A) and chloride-binding site (B) in
T. elongatus Dps. (B) The view is along the threefold axis with the
opening towards the protein cavity on the bottom. Water mole-
cules are depicted in red. Pictures were generated using
PYMOL.
Table 2. Characteristics of the ‘Dps-like’ pores in T. elongatus,
H. salinarum, E. coli, L. innocua and M. smegmatis Dps. The diam-
eter and length of the pores refer to the Ca-Ca distances between
the relevant symmetry-related residues.
Proteins
Residues
lining the pore
Diameter (A
˚
)
Length (A
˚
)surface cavity

T. elongatus Val157 (surface)
Leu40
Asp43 (cavity) 5.50 5.40  14
Arg42
H. salinarum Gln178
Leu181 (surface)
Val176 8.95 6.10  19
Glu59, Arg61
Asp62 (cavity)
E. coli Ala57 (surface)
Asn58 7.02 5.37  9.5
Ala61 (cavity)
L. innocua His37
Asn38 (surface) 6.80 6.26  13
Thr41 (cavity)
Glu44
M. smegmatis Pro45 (surface)
Asn46, Ile48 7.50 6.12  8.5
Gly49 (cavity)
S. Franceschini et al. The thermostable T. elongatus Dps
FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS 4919
denaturation, E. coli Dps still contains secondary
structure at 320 K, the apparent melting temperature,
an indication that protein aggregation takes place
before completion of the thermal melting process.
The T
m
values  353 for Dps-Te and  343 for
L. innocua Dps (calculated over a range of three
different experiments) can be taken as a measure of

thermostability, because the irreversibility of the
transitions depicted in Fig. 6 does not warrant the
calculation of thermodynamic parameters.
DNA-binding ability and DNA protection against
hydroxyl radicals
Binding of Dps-Te to DNA was analysed in vitro by
means of agarose gel electrophoresis experiments under
conditions where E. coli Dps is known to form large
complexes with DNA that do not enter the gel [7]. Dps–
DNA complexes were not detected when purified Dps-
Te (3 lm) was added to 20 nm supercoiled pET-11a
DNA in 30 mm Tris⁄ HCl containing 50 mm NaCl at
pH 6.5, 7.0 or 8.0 (Fig. 7A). Thus, in accordance with
the absence of the N-terminal extension used by E. coli
Dps in the interaction with DNA [3,24], Dps-Te is
unable to bind DNA. No complex formation was
observed when the protein concentration was increased
10-fold while keeping DNA constant (data not shown).
In order to establish whether T. elongatus Dps is
able to prevent hydroxyl radical-mediated DNA clea-
vage, an in vitro DNA damage assay was employed
[25]. The combined effect of 50 lm Fe(II) and
10 mm H
2
O
2
on the integrity of plasmid pET-11a
(5600 bp) was assessed in the presence and absence
of Dps-Te in 30 mm Tris ⁄ HCl, 50 mm NaCl, pH 7.5.
Under these conditions the hydroxyl radicals pro-

duced by the Fenton reaction degrade plasmid pET-
11a completely (Fig. 7B, lane 2). By inhibiting
hydroxyl radical formation the presence of Dps-Te
(Fig. 7B, lane 3) confers full protection to the DNA
plasmid.
AC
D
E
B
Fig. 4. Pores at the Dps-like interface of T. elongatus Dps (A), H. salinarum Dps (B), E. coli Dps (C), L. innocua Dps (D) and M. smegmatis Dps
(E). The residues forming the pores and their Van der Waal’s surfaces are indicated with a specific mention to the residues at the pore constric-
tions. (Left) View down the threefold axis. (Right) View from the opening on the protein surface. Pictures were generated using
PYMOL.
The thermostable T. elongatus Dps S. Franceschini et al.
4920 FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS
Iron incorporation kinetics
Dps-Te is able to oxidize and incorporate ferrous iron
in the presence of molecular oxygen at neutral pH at
25 and at 55 °C, the physiological temperature of the
bacterium (Fig. 8A). Progress curves measured after
the addition of 48 Fe(II) ⁄ dodecamer show that the
half-times of the iron-uptake reaction correspond to
600 and 200 s, at 25 and 55 °C, respectively. In the
absence of protein, the iron auto-oxidation process
leads to the precipitation of iron hydroxide at both
temperatures.
Dps-Te ferroxidation is more efficient with hydro-
gen peroxide as an oxidant, as described for other
AB
Fig. 7. DNA binding (A) and protection (B) by T. elongatus Dps. (A)

Lane 1, plasmid DNA; lane 2, plasmid DNA with Dps-Te. (B) Lane
1, plasmid DNA; lane 2, plasmid DNA with 50 m
M hydrogen per-
oxide, 50 l
M Fe(II); lane 3, plasmid DNA with 50 mM hydrogen per-
oxide, 50 l
M Fe(II) and 3 lM Dps-Te.
Fig. 5. Effect of pH on the state of association (A) and near-UV CD
spectra (B) of T. elongatus Dps-Te. At any given pH, protein solu-
tions at 1 mgÆmL
)1
were incubated at 25 °C for 24 h. (A) Elution
profiles upon HPLC-gel filtration after incubation at pH 1.0 (ÆÆÆ), 2.0
(—), 2.5 (ÆÆÆ), 3.0 (- - -), 7.0 (- Æ -). (B) Spectra recorded after incuba-
tion at pH 7.0 (- Æ -), 3.3 ( ), 2.0 (—), 1.0 (ÆÆÆ).
Fig. 6. Thermal denaturation of T. elongatus
Dps-Te, L. innocua Dps and E. coli Dps.
Spectra were recorded at 222 nm in 0.1 cm
quartz cuvettes; protein concentration
1mgÆmL
)1
; pH 3.3.
S. Franceschini et al. The thermostable T. elongatus Dps
FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS 4921
Dps proteins [4]. At pH 7.0 and 25 °C the half-time
of the reaction is 0.3 s upon addition of 100
Fe(II) ⁄ dodecamer (Fig. 8B), whereas at 55 °C the
half-time decreases to  0.035 s. Thus, oxidation of
Fe(II) by H
2

O
2
at room temperature is  2000-
fold faster than by molecular oxygen at room
temperature.
As expected, when using both oxidants an increase
in temperature from 25 to 55 °C results in an increase
in the initial rates of the reaction. The increase in rate
was approximately eightfold in the case of hydrogen
peroxide, and only threefold in the case of molecular
oxygen.
Discussion
The background to the present characterization of
T. elongatus Dps is provided by the recent, numerous
studies aimed at identifying the factors responsible for
the increased stability of proteins from thermophiles.
We were interested to establish which set of structural
devices is utilized by the Dps family to further stabilize
its characteristic shell-like assembly which is endowed
with an intrinsically high stability. Comparison of the
crystal structure of thermophilic Dps-Te with those of
mesophilic homologues indicates that the strategy
employed by T. elongatus is not only to increase the
number of intersubunit ion pairs and hydrogen bond-
ing interactions, a general strategy of thermophiles and
hyperthermophiles, but also to increase the amount of
buried surface of the least-extended Dps-like subunit
interface.
Prior to this study, the only reports that addressed
the stability of the Dps dodecamer regarded its ten-

dency to dissociate into subunits at acid pH and room
temperature. The protein systems revealed significant
differences in HPLC gel-filtration experiments. Thus,
the L. innocua dodecamer preserves its quaternary
structure at pH 2.0, whereas E. coli Dps starts dissoci-
ating at pH 2.5 and M. smegmatis Dps at pH 5.0
[6,7]. In these systems, dissociation gives rise to stable
dimers which in turn dissociate into stable monomers
when the pH is lowered further. Quite unexpectedly,
the thermophilic Dps-Te protein is less stable than the
L. innocua protein at room temperature. Thus, disrup-
tion of the Dps-Te assembly takes place at pH 2.5 as
shown by the disappearance of the dodecamer peak in
the HPLC patterns (Fig. 5A) and by the decrease in
rotational strength in the near-UV CD spectra
(Fig. 5B). Furthermore, the Dps-Te dimers and mono-
mers tend to aggregate and ⁄ or precipitate at variance
with those formed by L. innocua and M. smegmatis
Dps. The instability of the Dps-Te subunits in turn
implies that the subunit-dissociation process is irrevers-
ible, again at variance with that of L. innocua Dps [6].
The increased stability of Dps-Te relative to L. inno-
cua and E. coli Dps manifests itself at temperatures
>55 °C, the optimal growth temperature for the bac-
terium. At pH 7.0 and 80 °C, which corresponds to
the melting temperature of E. coli Dps, there is no
change in the secondary structure of Dps-Te (data not
shown). Thus, given the extremely high stability of
Dps-Te at neutral pH, thermal denaturation was stud-
ied at pH 3.0, a condition where the quaternary struc-

ture is conserved at room temperature. At this pH, the
melting temperature of Dps-Te is 10 or 30 °C higher
than those measured for the mesophilic L. innocua and
A

B
Fig. 8. Kinetics of iron oxidation ⁄ incorporation by T. elongatus Dps
using molecular oxygen (A) or hydrogen peroxide (B) as the oxidant.
Traces were measured at 310 nm wavelength, which monitors for-
mation of the ferric core because it corresponds to a d-d Fe(III) elec-
tronic transition at 25 and 55 °C. (A) Solutions of 17.5 l
M Fe(II) were
added to solutions of 0.25 l
M apoDps-Te [molar ratio 48 Fe(II) ⁄ Dps
dodecamer] in 50 m
M Mops, 150 mM NaCl buffer at pH 7.0. Tem-
peratures: 25 °C (—) and 55 °C (- - -). Fe(II) auto-oxidation: 25 °C(ÆÆÆÆÆ)
and 55 °C() Æ )). (B) Degassed solutions containing 1.0 l
M apoDps-
Te and 100 Fe(II) ⁄ Dps dodecamer in 50 m
M Mops, 150 mM NaCl
buffer at pH 7.0 were mixed with 50 l
M H
2
O
2
in the same buffer in
a stopped flow apparatus (Applied Photophysics).
The thermostable T. elongatus Dps S. Franceschini et al.
4922 FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS

E. coli proteins, respectively (Fig. 6). To elucidate the
factors responsible for such differences, the nature and
number of interactions at the different interfaces were
compared. In Dps-Te the trimeric and dimeric interfa-
ces contain significantly more hydrogen bonds and salt
bridges than in the mesophilic Dps proteins analysed
to date (Table 1).
Salt bridges, in particular, have been proposed to
play a crucial role in promoting protein thermostability
even though they appear to make little contribution to
protein stability at room temperature [26–28]. This
phenomenon is attributed to the fact that at room tem-
perature salt-bridge formation does not fully compen-
sate for the large enthalpic penalty due to desolvation
that accompanies association of two charged residues.
Conversely, the entropic contribution is favourable due
to the degrees of freedom gained by the previously
bound water molecules. Increasing the temperature,
the favourable entropic contribution increases more
than the enthalpic one, thus lowering the overall free
energy of salt-bridge formation [29,30]. Intriguingly,
the halophilic Dps from H. salinarum displays the
same structural features that stabilize the Dps–Te
interfaces, namely, a high number of intersubunit salt
bridges and an arginine cluster at the Dps-like trimeric
interface (Table 1) [23]. Whereas the former situation
applies only to few halophilic proteins, the latter struc-
tural feature, coupled to a negatively charged protein
surface, is often used to achieve stabilization at high
salt concentrations [31,32].

Analysis of the Dps-Te structure shows that the
dimeric interface is stabilized not only by the con-
served salt bridge between Lys30 and Asp60, but also
by the additional contribution of the salt bridge
between Asp76 and Lys31 and of the hydrogen bond
between Asp76 and Gly91 relative to the mesophilic
counterparts considered (Table 1). The Dps-like inter-
face likewise contains a large number of salt bridges
and hydrogen bonds that are conserved in the halophi-
lic H. salinarum Dps, but are absent in mesophilic Dps
proteins. In addition, it is unusually extended with a
significant increase in the amount of buried surface
area. The strongest electrostatic interaction at this
interface is provided by the salt bridge between the
nonconserved Arg42 residue and Glu50 (Fig. 2B).
Three hydrogen bonds furnish additional electrostatic
interactions, namely those formed between Lys96 and
Asp154, Glu148 and Gly38, and Tyr37 and Gly153
(Table 1).
The dimeric interface is of special importance
because it contains the unusual intersubunit bimetallic
ferroxidase centre characteristic of the Dps family. The
two iron-binding sites, A and B have different affinities
for the metal. In the available crystal structures, only
the A site is occupied by the metal. It is either fully
occupied as in L. innocua Dps, HP-NAP from
H. pylori and Dlp1 and Dlp2 from B. anthracis
[19,20,33], partially occupied as in A. tumefaciens Dps
[25], or is occupied by water molecules as in E. coli
Dps [2,34]. In Dps-Te the A and B sites contain two

water molecules replacing the iron atoms. The A water
molecule is coordinated by the two carboxylic oxygen
of Asp60 of one subunit and by the N-e2 atom of
His33 of the symmetry-related subunit, whereas water
molecule B is coordinated by O-e1 and O-e2 of Glu64
of one subunit and by N-e2 of His45 of the sym-
metry-related subunit. The absence of the metal in
Dps-Te points to a low affinity of the A site for iron.
According to Ilari et al. [20,34] this depends on the
nature and spatial arrangement of the residues sur-
rounding the ferroxidase site. In Dps-Te, the presence
of Lys30, which is engaged in a salt bridge with the
iron coordinating Asp60, likely decreases the affinity
of the site for iron. A similar situation applies to the
proteins of E. coli [34] and M. smegmatis [7], where
the A sites are empty and partially occupied, respect-
ively. In both proteins, a lysine residue in the vicinity
of the A site is engaged in electrostatic interactions
with an iron ligand (Lys48–Asp78 and Lys36–Asp66,
respectively).
The Dps-Te ferroxidase centre confers to the protein
the ability to protect DNA from hydroxyl radicals
even in the absence of DNA binding (Fig. 7B). Like
all members of the family Dps-Te uses hydrogen per-
oxide to oxidize Fe(II) and in this way simultaneously
eliminates the two molecules that give rise to hydroxyl
radicals via the Fenton reaction [35]. The significant
antioxidant properties of Dps-Te are also maintained
at 55 °C, the optimal growth temperature of T. elonga-
tus (Fig. 8A,B), a most important finding, because

reactive oxygen species have physiologically relevant
macromolecular targets other than DNA, like photo-
systems I and II.
Recently, the Dps protein from the N
2
-fixing marine
cyanobacterium T. erythraeum has been characterized.
It shows  30% sequence similarity to Dps-Te, binds
DNA, albeit with low affinity, but protects it from oxi-
dative damage. T. erythraeum carries out photosyn-
thesis in the presence of intense sunlight. Thus, the
Dps enzymatic activity could be particularly effective
in reducing photoactive damage facilitating survival of
the micro-organism.
Finally, the pores formed at the ferritin-like and
Dps-like interfaces deserve a comment for their func-
tional implications. They connect the protein cavity
with the outside and provide a passage for ions and
S. Franceschini et al. The thermostable T. elongatus Dps
FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS 4923
small molecules. The features of the ferritin-like pores
are largely conserved in all members of the family.
Thus, the openings on the protein surface (9–11 A
˚
)
and protein cavity ( 4.0 A
˚
) have similar dimensions
and are lined by negatively charged residues, e.g. in
Dps-Te, Glu118 and Glu120 and the highly conserved

Asp130, respectively. Canonical ferritins contain similar
negatively charged, hydrophilic pores at the threefold
symmetry axes that show negative values of the electro-
static potential. The electrostatic gradient thus formed
guides iron through the pores towards the ferroxidase
centre and the protein cavity [36,37]. Accordingly,
mutation of relevant aspartic and glutamic acid resi-
dues at the pore slows the rate of iron uptake [38,39].
Electrostatic calculations are lacking for the ‘ferritin-
like’ pore of Dps proteins. However, the similarity in
the disposition of the negatively charged residues is
expected to give rise to an electrostatic force directed
toward the protein cavity as in the case of ferritins.
Dps-like pores are highly variable in terms of dimen-
sions and the nature of the residues lining the openings
on the protein surface and protein cavity (Table 2).
The size variation is not surprising because these pores
are formed by flexible parts of the polypeptide chain,
like the C-terminus, the C-terminal part of the D helix
and the N-terminal part of the B helix. It may be
envisaged that these structural elements, although ‘fro-
zen’ in a single conformation in the crystal structure,
may move in solution and thereby enlarge the channel
openings. Intriguingly, in Dps-Te and H. salinarum
Dps, the two proteins from extremophiles examined,
the pores are in a ‘closed’ conformation (Table 2,
Fig. 4) and are lined with similar residues (arginines
on the cavity opening and hydrophobic residues on the
protein surface, namely valine in Dps-Te and leucine
in H. salinarum Dps).

A peculiarity of Dps-Te is the presence of strong
electronic density near the Arg42 residues that line the
pore. This has been tentatively identified as chloride
(Fig. 3B). It is tempting to speculate that the Dps-like
pores are involved in the iron reduction and exit
processes and hence may serve as an entrance for iron-
reducing agents. In turn, given their structural variabil-
ity, this common role would require the nature of the
endogenous reducing agents to differ widely in the
different organisms. If so, in Dps-Te the endogenous
reducing agent should be anionic in nature.
In conclusion, the thermostability of the intrinsically
stable dodecameric assemblage of Dps proteins is
enhanced in T. elongatus by means of two distinct
structural features that do not affect the stability at
room temperature, i.e. an increased density of inter-
subunit salt bridges, a common strategy used by
thermophiles and hyperthermophiles, and extension of
the buried surface area of the smallest subunit inter-
face. These distinctive structural features do not
increase the stability of Dps-Te at room temperature
relative to mesophilic Dps proteins over a wide pH
range. It is surprising that they be shared by the halo-
philic H. salinarum protein.
Experimental procedures
Construction of the Dps-Te gene
The dps gene was amplified by PCR from the genome of
T. elongatus BP1 (kindly provided by T. Kaneko, Kazusa
DNA Research Institute), using primers Dps-Te1 (5¢-CA
AAGGAGACT

CATATGAGTGCAACAACTAC-3¢) and
Dps-Te2 (5¢-CTACAA
AAGCTTAATCCGCAACTAACT
GAC-3¢). The NdeI and Hin dIII restriction sites are under-
lined. The amplified fragment (510 bp) was digested with
NdeI and HindIII, purified using the QIAquick PCR purifica-
tion kit (Qiagen, Valencia, CA) and cloned into the expres-
sion vector pET-22b (Novagen, Darmstadt, Germany)
digested with NdeI and HindIII. This plasmid was introduced
into E. coli BL21 (DE3) and sequenced by dideoxy sequen-
cing to confirm the presence of the correct gene.
Strains and media
E. coli strain BL21 (DE3) was grown at 37 °C on LB liquid
medium (10 gÆL
)1
tryptone, 5 gÆL
)1
yeast extract, 5 gÆL
)1
NaCl) or LB plates containing 50 lgÆmL
)1
ampicillin.
Expression and purification of Dps-Te
E. coli BL21 (DE3) cells harbouring the recombinant plas-
mid were grown at 37 °C in 1 L of ampicillin-containing
liquid LB medium to D
600
¼ 0.6. The dps gene was induced
by addition of 0.5 mm isopropyl thio-b-d-galactoside and
the culture was incubated further for 3–4 h.

Cells were harvested by centrifugation (15 000 g for
20 min, Rotor 34-14, Beckman Coulter Inc., Fullerton, CA)
and suspended in 10 mL of 50 mm Tris ⁄ HCl at pH 7.5,
containing 0.5 mm dithiothreitol, 1 mm EDTA, 500 mm
NaCl, and disrupted by sonication. The lysate was centri-
fuged at 15 000 g for 45 min and the supernatant was heated
to 75 °C for 10 min, cooled on ice, and then centrifuged
(Rotor JA-25.50, Beckman Coulter) to remove denatured
proteins. The recovered supernatant was precipitated using
two ammonium sulfate cuts at 30 and 60% saturation (w ⁄ v).
At 60% saturation Dps-Te precipitates and was recovered
after centrifugation (15 000 g for 45 min, Rotor JA-25.50,
Beckman Coulter). The Dps-Te-containing pellet was
resuspended and dialysed overnight against 20 mm Tris ⁄ HCl,
pH 7.5, and then loaded onto a DEAE cellulose column
The thermostable T. elongatus Dps S. Franceschini et al.
4924 FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS
(DE52) equilibrated with the same buffer. Dps-Te eluted with
200 mm NaCl; it was pooled and stored at )75 °C after
controlling the purity of the preparation using Coomassie
brilliant blue staining of SDS ⁄ 15% PAGE gels. The purified
protein did not contain iron.
Protein crystallization
Crystallization was achieved at 293 K by the hanging drop
vapour diffusion technique. A 2 lL volume of the protein
sample at 7 mgÆmL
)1
, equilibrated against 30 mm
Tris ⁄ HCl, pH 7.5 containing 0.2 m NaCl, was mixed with
an equal amount of the reservoir solution containing 0.1 m

sodium acetate at pH values between 4.0 and 5.0 and
PEG 4000 in a range between 10 and 14% w ⁄ v. Crystals
grew in 48 h to  0.4 · 0.3 · 0.2 mm
3
.
Data collection and processing
Data were collected as 0.5 oscillation frames using the
MAR CCD detector on the X-ray beamline at ELETTRA,
Basovizza (Trieste, Italy) at a wavelength of 1.2 A
˚
. Data
collection was performed at 100 K using 21% PEG 200 as
a cryoprotectant. Data analysis performed with denzo [40]
indicates that the crystals are centred orthorhombic (C
2221
)
with unit cell dimensions of a ¼ 122.97, b ¼ 122.88,
c ¼ 253.3 A
˚
. Data were scaled using scalepack [40] and
had R
merge
¼ 5.3% and v
2
¼ 0.65. The crystal contains 12
monomers per asymmetric unit, corresponding to the entire
molecule, with a V
M
¼ 2.08 A
˚

3
ÆDa
)1
and a solvent content
of  40.27%.
Structure solution and refinement
The structure was solved by molecular replacement using as
the search probe a polyalanine-truncated model built from
the B. anthracis Dlp2 tetramer (Protein Data Bank entry
1JIG). The rotational and translational searches, performed
with molrep [41] in the resolution range 10–3.0 A
˚
, produced
a clear solution. Refinement of the atomic coordinates and
displacement parameters were carried out applying the NCS
restraints to the 12 subunits of Dps-Te from residue 11 to
residue 154. Refinement was performed with refmac5 [42]
which used the maximum likelihood method. The refinement
statistics are presented in Table 3. Model building was per-
formed using the program package xtalview [43]. Water
molecules were added to the model manually. The final
model (a dodecamer) includes 1848 residues (154 residues
per monomer), 1500 water molecules, 4 chloride ions, 12 tri-
ethylene glycol molecules and 2 tetraethylene glycol mole-
cules. The final R
crys
at 1.8 A
˚
resolution was 16.9% with a
free R-value of 19.4%. The quality of the model was

assessed using procheck [44]. The most favoured regions of
the Ramachandran plot contain 96.4% nonglycine residues.
Atomic coordinates and the structure factors were deposited
in the Protein Data Bank (Accession number 2C41).
DNA-binding and protection assays
The DNA-binding ability of Dps-Te was assessed in gel-
shift assays using supercoiled pET-11a DNA (5600 bp,
20 nm) as a probe. DNA was purified using the Qiaprep
spin plasmid miniprep kit or Qiaquick Gel Extraction kit
(Qiagen, Chatsworth, CA), which ensure removal of impur-
ities and salts. DNA was incubated for 5 min at room tem-
perature alone or in the presence of the Dps protein (3 lm)
in 30 mm Tris ⁄ HCl, 50 mm NaCl at pH 6.5, 7.0 or 8.0. In
order to resolve the Dps–DNA complexes, electrophoresis
was carried out in 1% agarose gels in 0.04 m Tris-acetate
or BisTris-acetate buffers. Gels were stained with ethi-
dium bromide or Coomassie brilliant blue and imaged
using imagemaster vds (Amersham Biosciences, Uppsala,
Sweden).
DNA protection from oxidative damage was assessed
in vitro using 20 nm supercoiled pET-11a DNA. The assay
was carried out in 15 lLof30mm Tris ⁄ HCl, pH 7.5, con-
taining 100 mm NaCl. Plasmid DNA was allowed to inter-
act with Dps-Te (3 lm) for 5 min prior to the addition of
50 lm FeSO
4
. After 2 min, H
2
O
2

was added (10 mm final
concentration) and the mixture was incubated for 3 min at
room temperature to allow complete consumption of Fe(II).
Thereafter, 2% SDS was added to the reaction mixture and
incubated at 85 °C for 5 min. Plasmid DNA was resolved
Table 3. X-ray data analysis.
Data reduction
Cell parameters: a: 122.97 A
˚
a: 90.00°
b: 122.89 A
˚
b: 90.00°
c: 253.30 A
˚
c: 90.00°
Space group: C 2 2 2
1
v
2
: 0.649
R
merge
: 0.053
Collected reflections: 4242412
Unique reflections: 169902
Completeness of the dataset: 97.25%
Refinement
Resolution range: 30.00–1.81 A
˚

R
crys
: 0.169
R
free
: 0.194
Correlation factor: 0.912
Final model
<B> (calcd from the atomic model): 16.5 A
˚
2
r(B) 3.79 A
˚
2
Matthews coefficient: 2.08
Solvent content %: 40.27
Residues in core region of
Ramachandran plot (%):
96.4
Residues in additional allowed region: 3.6
Residues in generously allowed region (%) 0
S. Franceschini et al. The thermostable T. elongatus Dps
FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS 4925
by electrophoresis on 1% agarose gel in Tris acetate ⁄ EDTA
buffer. The gel was stained with ethidium bromide and
imaged using imagemaster vds.
Iron incorporation kinetics
Ferrous ammonium sulfate solutions were prepared freshly
in Thunberg tubes prior to the experiments and kept under
nitrogen gas.

Kinetic experiments of Fe(II) oxidation by O
2
were per-
formed on a Hewlett-Packard diode array spectrophotometer
(Palo Alto, CA) at 25 and 55 °Cin50mm Mops, 150 mm
NaCl, pH 7.0. Time-dependent absorbance traces were col-
lected at 310 nm, wavelength which monitors formation of
the ferric core since it corresponds to a d–d Fe(III) electronic
transition, and were analysed with origin 6.0 (Originlab
Corporation, Northampton, MA). Fe(II) was added to
0.25 lm apoDps in the same Mops buffer at 25 or 55 °C, at a
Fe(II) ⁄ dodecamer molar ratio of 48 : 1. During the course of
the reaction, protein solutions were maintained in air under
stirring. As a control, the rate of Fe(II) autoxidation was
measured in parallel.
The kinetics of Fe(II) oxidation by H
2
O
2
was followed in
a stopped flow apparatus (Applied Photophysics, Leather-
head, UK) over the temperature range 25–60 °C. Fe(II)
was added at a Fe(II) ⁄ dodecamer molar ratio of 100 : 1 to
degassed 1.0 lm apoDps in 50 mm Mops buffer at pH 7.0,
150 mm NaCl kept in a tonometer under nitrogen gas. This
solution was mixed with 50 lm H
2
O
2
in the same buffer.

The iron oxidation reaction was followed at 310 nm. The
traces were analysed with origin 6.0 (Microcal Inc.).
CD spectroscopy
CD measurements were carried out using Jasco J-715 CD
apparatus. The near-UV CD spectra were recorded at
20 °C in a 1 cm quartz cuvette on 0.5 mgÆmL
)1
protein
solutions after 16 h incubation at pH 7.0 (50 mm phosphate
buffer), pH 3.0 (50 mm glycine–HCl buffer), pH 2.0
(10 mm HCl) or pH 1.0 (100 mm HCl).
Thermal unfolding experiments were monitored in the
far-UV region in a stoppered 0.1 cm quartz cell. The
protein concentration was 1 mgÆmL
)1
at pH 3.0 (20 mm
glycine–HCl buffer).
HPLC and gel-filtration experiments
The state of association was analysed by size-exclusion
chromatography experiments on a Pharmacia HPLC system
at room temperature. Dps-Te samples (50 lL) were applied
to a TSK-GEL G3000SWxl 7.8 ⁄ 30 column (TOSOH Bio-
science, Tokyo, Japan) after 16 h incubation at pH 1.0, 2.0,
3.0 or 7.0 and were eluted with 0.1 m sodium phosphate
at pH 7.0 containing 0.1 m NaCl or 50 mm glycine–HCl
pH 3.0 at a flow rate of 0.8 mLÆmin
)1
. The incubation buf-
fers used were 0.1 m HCl pH 1.0, 10 mm HCl pH 2.0,
0.1 m glycine–HCl pH 3.0 and 0.1 m phosphate pH 7.0, all

in the presence of 0.1 m NaCl. Horse spleen ferritin
(450 kDa), E. coli Dps (221 kDa), sorcin (43 kDa) and
myoglobin (16.9 kDa) were run independently under the
same conditions to calibrate the column. All experiments
were performed in triplicate.
Acknowledgements
Stefano Franceschini is the recipient of a National
Research Council research fellowship. This work was
supported by grants from ‘FIRB 2003’ and
‘PRIN 2005’ (to EC). We thank Prof Alberto Boffi
for carrying out stopped-flow experiments, and Prof
Simonetta Stefanini and Dr Giuliano Bellapadrona for
their helpful discussions and valuable suggestions. We
also thank the beamline scientists of ELETTRA (Bas-
ovizza, Trieste, Italy) where the X-ray diffraction data
were collected.
References
1 Almiron M, Link AJ, Furlong D & Kolter R (1992) A
novel DNA-binding protein with regulatory and protec-
tive roles in starved Escherichia coli. Genes Dev 6, 2646–
2654.
2 Grant RA, Filman DJ, Finkel SE, Kolter R & Hogle
JM (1998) The crystal structure of Dps, a ferritin homo-
log that binds and protects DNA. Nat Struct Biol 5,
294–303.
3 Wolf SG, Frenkiel D, Arad T, Finkel SE, Kolter R &
Minski A (1999) DNA protection by stress-induced bio-
crystallization. Nature 400, 83–85.
4 Zhao G, Ceci P, Ilari A, Giangiacomo L, Laue TM,
Chiancone E & Chasteen ND (2002) Iron and hydrogen

peroxide detoxification properties of DNA-binding
protein from starved cells. A ferritin-like DNA-binding
protein of Escherichia coli. J Biol Chem 277, 27689–
27696.
5 Yamazaki I & Piette LH (1990) ESR spin-trapping stu-
dies on the reaction of Fe
2+
ions with H
2
O
2
-reactive
species in oxygen toxicity in biology. J Biol Chem 265,
13589–13594.
6 Chiaraluce R, Consalvi V, Cavallo S, Ilari A, Stefanini
S & Chiancone E (2000) The unusual dodecameric ferri-
tin from Listeria innocua dissociates below pH 2.0. Eur
J Biochem 267, 5733–5741.
7 Ceci P, Ilari A, Falvo E, Giangiacomo L & Chiancone
E (2005) Reassessment of protein stability, DNA bind-
ing and protection of Mycobacterium smegmatis Dps.
J Biol Chem 280, 34776–34785.
The thermostable T. elongatus Dps S. Franceschini et al.
4926 FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS
8 Yamaoka T, Satoh K & Katoh S (1978) Photosynthetic
activities of a thermophilic blue–green alga. Plant Cell
Physiol 19, 943–954.
9 Nakamura Y, Kaneko T, Sato S, Ikeuchi M, Katoh H,
Sasamoto S, Watanabe A, Iriguchi M, Kawashima K,
Kimura T et al. (2002) Complete genome structure of

the thermophilic cyanobacterium Thermosynechococcus
elongatus BP-1. DNA Res 9, 123–130.
10 Loll B, Kern J, Zouni A, Saenger W, Biesiadka J &
Irrgang KD (2005) The antenna system of photosystem II
from Thermosynechococcus elongatus at 3.2 A
˚
resolution.
Photosynth Res 86, 175–184.
11 Jordan P, Gromme P, Witt HT, Klukas O, Saenger W
& Krauss N (2001) Three-dimensional structure of cya-
nobacterial photosystem I at 2.5 A
˚
resolution. Nature
411, 909–917.
12 Lupinkova L & Komenda J (2004) Oxidative modifica-
tions of the photosystem II, D1 protein by reactive
oxygen species: from isolated protein to cyanobacterial
cells. Photochem Photobiol 79, 152–162.
13 He YY & Hader D (2002) Reactive oxygen species and
UV-B: effect on cyanobacteria. Photochem Photobiol Sci
1, 729–736.
14 He YY & Hader D (2002) Involvement of reactive oxy-
gen species in the UV-B damage to the cyanobacterium
Anabaena sp. J Photochem Photobiol B 66, 73–80.
15 Liu K, Sun J, Song YG, Liu B, Xu YK, Zhang SX,
Tian Q & Liu Y (2004) Superoxide, hydrogen peroxide
and hydroxyl radical in D1 ⁄ D2 ⁄ cytochrome b-559 phot-
osystem II reaction center complex. Photosynth Res 81,
41–47.
16 Sicora C, Mate Z & Vass I (2003) The interaction of

visible and UV-B light during photodamage and repair
of photosystem II. Photosynth Res 75, 127–137.
17 Allakhverdiev SI, Nishiyama Y, Takahashi S, Miyairi S,
Suzuki I & Murata N (2005) Systematic analysis of the
relation of electron transport and ATP synthesis to the
photodamage and repair of photosystem II in Synecho-
cystis. Plant Physiol 137, 263–273.
18 Corpet F (1988) Multiple sequence alignment with hier-
archical clustering. Nucleic Acids Res 16, 10881–10890.
19 Papinutto E, Dundon WG, Pitulis N, Battistutta R,
Montecucco C & Zanotti G (2002) Structure of two
iron-binding proteins from Bacillus anthracis. J Biol
Chem 277, 15093–15098.
20 Ilari A, Stefanini S, Chiancone E & Tsernoglou D
(2000) The dodecameric ferritin from Listeria innocua
contains a novel intersubunit iron-binding site. Nat
Struct Biol 7, 38–43.
21 Roy S, Gupta S, Das S, Sekar K, Chatterji D & Vijayan
M (2004) X-ray analysis of Mycobacterium smegmatis
Dps and a comparative study involving other Dps and
Dps-like molecules. J Mol Biol
339, 1103–1113.
22 Castruita M, Saito M, Shottel PC, Elmegreen LA,
Mineny S, Stiefel EI & Morel FM (2006) Overexpres-
sion and characterization of an iron storage and
DNA-binding Dps protein from Trichodesmium ery-
thraeum. Appl Environ Microbiol 72, 2918–2924.
23 Zeth K, Offermann S, Essen LO & Oesterhelt D (2004)
Iron-oxo clusters biomineralizing on protein surfaces:
structural analysis of Halobacterium salinarum DpsA in

its low- and high-iron states. Proc Natl Acad Sci USA
101, 13780–13785.
24 Ceci P, Cellai S, Falvo E, Rivetti C, Rossi GL & Chian-
cone E (2004) DNA condensation and self-aggregation
of Escherichia coli Dps are coupled phenomena related
to the properties of the N-terminus. Nucleic Acids Res
32, 5935–5944.
25 Ceci P, Ilari A, Falvo E & Chiancone E (2003) The
Dps protein of Agrobacterium tumefaciens does not
bind to DNA but protects it toward oxidative clea-
vage: X-ray crystal structure, iron binding, and hydro-
xyl-radical scavenging properties. J Biol Chem 278,
20319–20326.
26 Hennig M, Darimont B, Sterner R, Kirshner K & Jan-
sonius JN (1995) 2.0 A
˚
structure of indole-3-glycerol
phosphate synthase from the hyperthermophile Sulfolo-
bus solfataricus: possible determinants of protein stabi-
lity. Structure 3, 1295–1306.
27 Yip KS, Stillman TJ, Britton KL, Artymiuk PJ, Baker
PJ, Sedelnikova SE, Engel PC, Pasquo A, Chiaraluce R,
Consalvi V et al. (1995) The structure of Pyrococcus
furiosus glutamate dehydrogenase reveals a key role for
ion-pair networks in maintaining enzyme stability at
extreme temperatures. Structure 3, 1147–1158.
28 Wallon G, Kryger G, Lovett ST, Oshima T, Ringe D &
Petsko GA (1997) Crystal structures of Escherichia coli
and Salmonella typhimurium 3-isopropylmalate dehydro-
genase and comparison with their thermophilic counter-

part from Thermus thermophilus. J Mol Biol 266,
1016–1031.
29 Elcock HA (1998) The stability of salt bridges at high
temperatures: implications for hyperthermophilic pro-
teins. J Mol Biol 284, 489–502.
30 Szilagyi A & Zavodszky P (2000) Structural differences
between mesophilic, moderately thermophilic and extre-
mely thermophilic protein subunits: results of a compre-
hensive survey. Structure 8 , 493–504.
31 Tehei M & Zaccai G (2005) Adaptation to extreme
environments: macromolecular dynamics in complex
systems. Biochim Biophys Acta 1724, 404–410.
32 Irmia A, Ebel C, Madern D, Richard SB, Cosenza LW,
Zaccai G & Vellieux FM (2003) The oligomeric states of
Haloarcula marismortui malate dehydrogenase are modu-
lated by solvent components as shown by crystallographic
and biochemical studies. J Mol Biol 326, 859–873.
33 Zanotti G, Papinutto E, Dundon W, Battistutta R,
Seveso M, Del Giudice G, Rappuoli R & Montecucco
C (2002) Structure of the neutrophil-activating protein
from Helicobacter pylori. J Mol Biol 323, 125–130.
S. Franceschini et al. The thermostable T. elongatus Dps
FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS 4927
34 Ilari A, Ceci P, Ferrari D, Rossi GL & Chiancone E
(2002) Iron incorporation into Escherichia coli Dps gives
rise to a ferritin-like microcrystalline core. J Biol Chem
277, 37619–37623.
35 Yang X, Chiancone E, Stefanini S, Ilari A & Chasteen
N (2000) Iron oxidation and hydrolysis reactions of a
novel ferritin from Listeria innocua. Biochem J 349,

783–786.
36 Douglas T & Ripoll DR (1998) Calculated electrostatic
gradients in recombinant human H-chain ferritin.
Protein Sci 7, 1083–1091.
37 Takahashi T & Kuyucak S (2003) Functional properties
of threefold and fourfold channels in ferritin deduced
from electrostatic calculations. Biophys J 84, 2256–2263.
38 Treffry A, Bauminger ER, Hechel D, Hodson NW,
Nowik I, Yewdall SJ & Harrison PM (1993) Defining
the roles of the threefold channels in iron uptake, iron
oxidation and iron-core formation in ferritin: a study
aided by site-directed mutagenesis. Biochem J 296, 721–
728.
39 Levi S, Santambrogio P, Corsi B, Cozzi A & Arosio P
(1996) Evidence that residues exposed on the three-fold
channels have active roles in the mechanism of ferritin
iron incorporation. Biochem J 317, 467–473.
40 Ottwinowski Z & Minor W (1997) Processing of X-ray
diffraction data collected in oscillation mode. Methods
Enzymol 276, 307–326.
41 Vagin A & Teplyakov A (1997) MOLREP: an auto-
mated program for molecular replacement. J Appl Crys-
tallogr 30, 1022–1025.
42 Murshudov GN, Lebedev A, Vagin A, Wilson KS &
Dodson EJ (1999) Efficient anisotropic refinement of
macromolecular structures using FFT. Acta Crystallogr
Sect D Biol Crystallogr 55, 247–255.
43 McRee DE (1993) Practical Protein Crystallography.
Academic Press, Orlando, FL.
44 Laskowski RA, McArthur MW, Moss DS & Thornton

JM (1993) PROCHECK: a program to check the stereo-
chemical quality of protein structures. J Appl Crystal-
logr 26, 283–291.
The thermostable T. elongatus Dps S. Franceschini et al.
4928 FEBS Journal 273 (2006) 4913–4928 ª 2006 The Authors Journal compilation ª 2006 FEBS