Here we report the preparation of geodin as a recombinant protein from Escherichia coli, its characterization through physico-chemical analyses, and a model of its 3D structure based on
Trang 1A bc-crystallin-type protein from a sponge
Concetta Giancola1, Elio Pizzo2, Antimo Di Maro3, Maria Vittoria Cubellis2and Giuseppe D’Alessio2
1 Department of Chemistry, University ‘Federico II’ of Naples, Italy
2 Department of Biological Chemistry, University ‘Federico II’ of Naples, Italy
3 Department of Life Sciences, Second University of Naples, Italy
Vertebrate crystallins are proteins that last an entire
life-time, tightly packed in the eye lens which they provide
with the appropriate refractive index essential for vision
There are three families of crystallins: the a-crystallins,
complex multimers made up of small proteins which
perform also chaperon-like functions; and the b- and
c-crystallins, which together constitute a superfamily of
homologous proteins including monomeric c-type and
oligomeric b-type proteins [1] bc-Crystallins have been
extensively studied as models of molecular evolution
[2–5] and for their structural features [6,7] The 3D
structures of many members of the superfamily have
been determined by X-ray crystallography [8–11], and
NMR [12,13] They indicate that the smallest structural
unit in the bc-crystallin superfamily is a b-stranded
Greek key motif, with two such motifs making up a
domain; two domains connected by a peptide linker
constitute a c-crystallin-type monomer, or a subunit of
a b-crystallin-type oligomer
A long evolutionary history can be traced for bc-crystallins, back to single-domain homologues from moulds [14], and bacteria [12], and a two-domain homologue from amphibians [15] Recently, a bc-crys-tallin-type gene from a sponge, Geodia cydonium, has been identified [16] and cloned [17] The finding that this is an intron-less gene [17] (as compared with ver-tebrate bc-crystallin genes, all of which are endowed with several introns) and the very early divergence of porifera ) the most primitive metazoans ) support the idea that the protein encoded by this gene, called geo-din, is the most ancient member of the metazoan bc-crystallin-type superfamily Thus, it appeared to be
of interest to inspect the structural features of this pro-tein, to verify the hypothesis that the same structural
Keywords
geodin; bc-crystallins; metazoans;
calorimetry; protein stability
Correspondence
G D’Alessio, Department of Biochemistry,
University of Naples Federico II, Via
Mezzocannone 16, 80134 Naples, Italy
Fax: +39 081 5521217
Tel: +39 081 2534731
E-mail: dalessio@unina.it
(Received 4 November 2004, revised 30
November 2004, accepted 20 December
2004)
doi:10.1111/j.1742-4658.2004.04536.x
Geodin is a protein encoded by a sponge gene homologous to genes from the bc-crystallins superfamily The interest for this crystallin-type protein stems from the phylogenesis of porifera, commonly called sponges, the earliest divergence event in the history of metazoans Here we report the preparation of geodin as a recombinant protein from Escherichia coli, its characterization through physico-chemical analyses, and a model of its 3D structure based on homology modelling Geodin is a monomeric protein of about 18 kDa, with an all-beta structure, as all other crystallins in the superfamily, but more prone to unfold in the presence of chemical denatu-rants, when compared with other homologues from the superfamily Its thermal unfolding, studied by far- and near-CD, and by calorimetry, is des-cribed by a two-state model Geodin appears to be structurally similar in many respects to the bacterial protein S crystallin, with which it also shares
a significant, albeit more modest stabilizing effect exerted by calcium ions These results suggest that the crystallin-type structural scaffold, employed
in the evolution of bacteria and moulds, was successfully recruited very early in the evolution of metazoa
Abbreviations
DSC, differential scanning microcalorimetry; GuHCl, guanidine hydrochloride; TFA, trifluoroacetic acid.
Trang 2scaffold was used in the evolution of bacteria and
moulds, and then very early recruited in the
metazo-ans Here we report the expression and purification of
recombinant geodin and its thermodynamic and
spect-roscopic characterization, and propose a putative
structure of the protein, as derived through homology
modelling
Results and Discussion
Preparation of recombinant geodin
As described in Experimental procedures, E coli cells
transformed with the cDNA encoding geodin, the
putative bc-crystallin-type protein from the sponge
G cydonium (UNIPROT accession number: O18426),
were lysed by sonication followed by centrifugation
By SDS⁄ PAGE, both the resulting supernatant and
pellet were found to contain a protein of 18 kDa,
expected for geodin on the basis of the amino acid
sequence encoded by the available gene sequence
(Fig 1)
Thus, both fractions were investigated: the soluble
fraction (S preparation) and the pelleted fraction, after
solubilization and renaturation (see Methods), called
IB preparation The S and IB preparations were
fract-ionated in parallel by size exclusion chromatography
followed by ion exchange chromatography The
chro-matography runs, described in the Experimental
proce-dures, are illustrated in Fig 2
When the proteins purified from the S or IB prep-arations, respectively, were analysed by SDS⁄ PAGE, they were both found to contain a single protein with the molecular size of geodin (Fig 1) Identical results were obtained when the two preparations
100 kDa
66 kDa
14 kDa
8 kDa
18 kDa
M 1 2 3 4
Fig 1 SDS ⁄ PAGE of protein preparations from the lysate of E coli
cells transformed with geodin encoding cDNA Lane 1, IB
prepar-ation from inclusion bodies; lane 2, S preparprepar-ation from the lysate
soluble fraction; lane 3, geodin purified from the IB preparation;
lane 4, geodin purified from the S preparation.
A
B
C
D
Fig 2 Gel filtration on Superdex G-75 of the proteins from the S preparation (A) and the IB preparation (B) from the lysate of E coli cells transformed with geodin encoding cDNA The inserts show the results of SDS ⁄ PAGE separations of individual fractions from each column as indicated by fraction numbers Fractions 48 through 56 were pooled and dialysed Each pool was then chroma-tographed on a cation exchange column (Resource-S) as illustrated
in C (for the S preparation) and D (for the IB preparation).
Trang 3were subjected to RP-HPLC on a C4 column as
des-cribed in Methods Also by this procedure, the
pro-tein, isolated from the S or IB preparation, was
found to be homogeneous and eluted in a single,
symmetrical peak from the HPLC column (data not
shown)
The two proteins eluted from the HPLC column,
analysed by MALDI-TOF MS, were found to have
the following molecular masses: 17 788, protein
puri-fied from S preparation; 17 785 protein puripuri-fied from
IB preparation These values compare very
satisfactor-ily with the geodin mass value calculated from its
amino acid sequence (17 781 kDa)
The amino acid sequence determination carried out
on the proteins purified from either the S or IB
prepa-rations gave the following N-terminal sequence
(one-letter code), identical for both protein preparations:
NH2–STAKVTLVTSGGSSQDFT–, which is exactly
the sequence deduced for the N terminus of the protein
encoded by the geod gene from G cydonium
These results indicated that: (a) the same protein
was contained in both the S and IB protein
prepara-tions, as derived from the soluble and insoluble
frac-tions, respectively, of the E coli lysate; (b) this protein
was geodin, the expression product of the geod gene
from the sponge G cydonium
Thus, a single form of geodin was expressed by E coli
under the conditions described above The finding that
the expressed recombinant geodin was distributed
between cytosol and inclusion bodies of transformed
E coli cells can be explained by considering that the
E coli synthetic machinery allowed the production of
free, soluble geodin up to a solubility limit, beyond
which the excess protein was sequestered in inclusion
bodies
Stability against urea and guanidinium chloride Figure 3 shows the CD spectra of geodin in far-UV and near-UV at T¼ 20 C and pH 5.0 and 7.0, respectively At both pH values, the far-UV CD spec-trum of the protein exhibits a strong positive band at
217 nm, which indicates a well defined b-conformation
in solution This suggests that geodin is an all-b pro-tein The different protonation state apparently does not affect the secondary structure, as no drastic chan-ges are observed in the far-UV signal (Fig 3) In the near-UV region the effects of protonation are signifi-cant in the protein tertiary structure with a higher exposure of chromophors at pH 5.0
As for vertebrate crystallins) such as mammalian
cB and bB2 crystallins) geodin tends to aggregate, especially at higher temperatures and concentrations However, because at relatively high concentrations geodin was more soluble at pH 5.0 than at pH 7.0, a
pH value closer to geodin pI value of 7.9, as calculated for from its amino acid sequence, we chose to study the thermodynamic properties of the protein at pH 5 Geodin stability against chemical denaturants was investigated by measuring the molar ellipticity at
217 nm, and the shift in fluorescence maximum wave-length, as a function of urea or guanidine (GuHCl) concentration As shown in Fig 4, all denaturation curves have monophasic, sigmoidal shapes with single midpoints of denaturation determined at 3.5 m and 1.2 m for urea- and GuHCl-induced denaturation, respectively When the denatured protein solutions were dialysed, their far-UV CD spectra were found to
be identical to that of the native protein This indicates that the unfolding of geodin as induced by chemical denaturants is reversible
Fig 3 Far-UV (A) and near-UV (B) CD
spec-tra of geodin at pH 5.0 (solid lines) and
pH 7.0 (dashed lines).
Trang 4Geodin appears to be very sensitive to chemical
denaturation, in fact more sensitive than a typical
crystallin-type protein of the vertebrate c family
(Table 1) In particular, with guanidinium⁄ HCl as a
denaturant, the midpoint of denaturation determined
for geodin is lower than those determined for other
monomeric crystallin-type proteins, but very close to
that determined for spherulin 3a The latter
crystal-lin-type protein is a homodimeric protein in which
two single-domain protomers associate into the
topo-logical equivalent of a monomeric crystallin Thus,
the finding of a similar dependence from chemical
denaturation between geodin and spherulin 3a might
be suggestive of weaker inter-domain interactions in
geodin, weaker than those at the interfaces of monomeric crystallins, but comparable with those
in a noncovalent two-domain homologue such as spherulin 3a
As for the denaturation of geodin in urea, as com-pared with other monomeric crystallin-type proteins, our findings (Table 1) suggest that the stability of geo-din in urea resembles that of protein S rather than that
of a mammalian crystallin such as cS-crystallin, a qualitative, indirect indication of a closer structural relationship between geodin and protein S Geodin also displays a single denaturation midpoint, detectable
at pH 5, but with no aggregation at any urea concen-tration (Fig 4, Table 1)
Fig 4 (A, B) Urea and (C, D) GuHCl-induced transitions of geodin at pH 5.0 The transi-tions were monitored by the shift in the wavelength corresponding to the maximum
of the fluorescence spectrum (A and C), and
by far-UV CD at 217 nm (B and D).
Trang 5Stability against temperature
When the thermal denaturation of geodin was
investi-gated by the dichroic absorption in far- and near-UV
(Fig 5), at 217 nm and 280 nm, respectively, the
curves showed cooperative transitions in agreement
with a two-state model For both, the midpoint
tem-perature was found to be 61C, which indicates a
sim-ultaneous collapse of secondary and tertiary structures
The calorimetric profile obtained by differential
scan-ning microcalorimetry (Fig 6), was not influenced by
protein concentration, although the endothermic peak
was followed by an exothermic process that coincides
with aggregation and precipitation, which lead to an
irreversible overall denaturation process
A description of an irreversible protein denaturation
is in the model proposed by Lumry and Eyring [18]:
N$ U $ F where N is the native protein, U is the reversibly
unfolded protein and F is the final, irreversibly
dena-tured protein Starting from the Lumry–Heyring
model, Sanchez-Ruiz [19] has shown that when the
transition is calorimetrically irreversible, but the
irre-versible step takes place with a significant rate at a
temperature even slightly above those corresponding to
the transition, the equilibrium thermodynamic analysis
is permissible
In this case the unfolding process:
N !K U
is described by the van’t Hoff equation:
@ln K
@T
P
¼DH 0
where K is the equilibrium constant and DH0 is the enthalpy that determines the variation of K with the absolute temperature The integrated form can be writ-ten as:
K¼ exp DH
0ðTÞ R
1
T
1
Tm
ð2Þ
At each temperature value, the enthalpy of this ther-modynamic system can be described as:
Table 1 Physico-chemical data for geodin and other two-domain
crystallin-type proteins Values of Tm, DH 0 and DG 0 , determined by
DSC, are from reference [39] for protein S, and reference [26] for
spherulin 3a and human-cS Data for denaturation with chemical
denaturants were obtained by CD spectroscopy for protein S [22],
spherulin 3a and human-cS [25].
c 1 ⁄ 2 GuCl
( M )
c 1⁄ 2 urea ( M )
T m (C)
DH 0 (kJÆmol)1)
DG 0 (293 K) (kJÆmol)1) Geodin 1.2 a 3.5 a 61.0 c 532 41
1.2b 3.6b 60.5d
+ Ca 2+ 65.0 c,d 570 48
Human-cS 2.6 8.0 75 750 84
Protein S 1.7 3.7 52 ⁄ 64 e
399 ⁄ 263 29 ⁄ 16 + Ca 2+ 1.9 4.8 64 ⁄ 65 e 454 ⁄ 332 39 ⁄ 25
Spherulin 3a 1.1 53.3 523 81
+ Ca 2+ 2.5 68.7 1020 137
a Data from near-UV CD spectroscopy b Data from fluorescence
spectroscopy. cData from near- and far-UV CD spectroscopy.
d Data from DSC e Values for first ⁄ second transition.
Fig 5 Temperature-induced unfolding transitions at pH 5.0 monit-ored by far-UV CD (A) and near-UV CD (B).
Trang 6HðTÞ ¼ fNHNþ fDHD¼ HNþ fDðHD HNÞ
where fN and fD are the fractions of molecules in the
native and denatured states, respectively HN and HD
are the corresponding enthalpies of native and
dena-tured states
Choosing the native state as reference state, the
fol-lowing equation for the excess enthalpy is obtained:
<DH0ðTÞ > ¼ HðTÞ HN ¼ fDDH0ðTmÞ
¼ ½K=ð1 þ KÞ DH0ðTmÞ ð4Þ which, derived with respect to the temperature and
based on Eqn (1), leads to:
<DC0ðTÞ > ¼½DH
0ðTmÞ2
RT2 ½K=ð1 þ KÞ2
þ DC0
PðTmÞ½K=ð1 þ KÞ ð5Þ This equation allows the simulation of a calorimetric
curve for a two-state transition [20,21] When the
experimental curve of geodin denaturation and the
DSC profile predicted by Eqn (5) were juxtaposed, a
satisfactory agreement was found (Fig 6) This
con-firms that the equilibrium thermodynamic treatment
can be applied to this case, and the Gibbs’ energy
value can be calculated
In Table 1 we compare the thermodynamic
parame-ters of geodin with those of two-domain crystallin-type
proteins for which values have been determined for
most physico-chemical parameters, such as monomeric
bacterial protein S and mammalian human-cS, and
spherulin, which is a two-domain noncovalent dimer
The latter two proteins, as well as geodin, display a single transition in the denaturation process Protein S instead undergoes a two-step denaturation, but the main transition is centred at 64.4C, closer to that of geodin (61C) when compared with the Tm values of
75C and 53 C determined for human-cS and spheru-lin, respectively Furthermore, the values forDG, as a parameter of thermodynamic stability, are of about
80 kJÆmol)1for human-cS and spherulin 3a, and of 41 and 45 kJÆmol)1for geodin and protein S, respectively The value of 45 kJÆmol)1for protein S is the value cal-culated for the overall unfolding process Thus, the conclusion can be proposed that, based on thermody-namic behaviour, geodin is a crystallin-type protein closer to protein S than to a mammalian (human-cS)
or mould (spherulin 3a) crystallin
Stability of geodin in the presence
of calcium ions Several two-domain members of the bc-crystallin superfamily are stabilized by calcium ions, such as pro-tein S [22], spherulin 3a [14,23,24], and c-crystallin [25] To investigate the effect of Ca2+on geodin stabil-ity, DSC and CD measurements were performed in parallel at pH 5.0 in ammonium acetate buffers con-taining either 1 mm CaCl2 or 1 mm Na2EDTA The near-UV spectrum, shown in Fig 7A, indicates that upon CaCl2 addition the positive band at 280 nm increases, revealing perturbation in the exposure of some aromatic residue(s)
Also the far-UV dichroic spectrum is affected, as shown in Fig 7B, with a less intense minimum at
217 nm and a shoulder at about 230 nm in the pres-ence of Ca2+ The latter findings, which suggest a reas-sessment in the secondary structural order of the protein upon calcium binding, are in contrast with the results obtained in the case of homologous protein S, for which calcium binding affects only the protein ter-tiary structure [26]
The CD melting profile determined for geodin at
217 nm in the presence of Ca2+ was found to be sig-moidal as that obtained in the absence of calcium ions, but the midpoint denaturation was increased to 65C (Fig 7C) A perfectly coincident value was obtained
by calorimetric measurements in the presence of cal-cium ions (Table 1) The higher melting temperature (increased by 4–5C) and higher denaturation enthalpy (Table 1) indicate that calcium ions provide
an additional stabilization to geodin In fact, the calcu-latedDG0at 293 K was found to be 48 kJÆmol)1, com-pared to 41 kJÆmol)1 calculated from measurements in the absence of Ca2+ions
Fig 6 Experimental (solid line) and simulated (dotted line)
calori-metric curves of geodin at pH 5.0.
Trang 7A structural model for geodin
We first searched for homologues of geodin in
Hom-strad (http://www-cryst.bioc.cam.ac.uk/data/align), a
curated database [27] of structure-based alignments for
homologous protein families, which makes use of
FuGUE (http://www-cryst.bioc.cam.ac.uk/fugue/
prfsearch.html [28] The first hit found by fugue (z
score¼ 20, average sequence length ¼ 174)
correspon-ded to the family of bc-crystallins The second hit (z
score¼ 13.5, sequence length ¼ 101) corresponded to
spherulin 3a, a homodimer in which each chain
con-tains a single crystallin-type domain [24] The same
two hits with statistically significant z scores were
obtained when the sequence of geodin was divided into
two halves and each half was used independently as a
query sequence Therefore, it is reasonable to assume
that geodin (sequence length¼ 163) has two tandemly
repeated domains, each with a crystallin-type fold We
submitted the geodin sequence to a threading server
(http://www.sbg.bio.ic.ac.k/~3dpssm), and obtained the
same fold
Geodin was then aligned with a structure-based
available alignment of bc-crystallins with
experiment-ally solved structures, stored in Homstrad and
decor-ated with joy [29] This makes visible structural
features, e.g buried residues (in upper case), exposed
residues (lower case) (Fig 8) The alignment of geodin
to this pre-existent alignment of crystallins was carried
out using fugue and was manually modified in the
N-terminal portion The improvement produced was
tested by comparative model validation using anolea [30,31] and prosaii [32]
Buried residues belonging to b-strands and facing the intradomain hydrophobic core of each domain, as well as other buried residues and positive-u glycine residues were found to be mostly conserved in geodin, which indicated the good quality of the alignment It should be noted that the sequence of protein S shows the insertion of a valine at position 50, where the sequences of the other homologues show a glycine The alternative to this insertion in the Homstrad align-ment would have required both an insertion plus a preceding gap In the alignment by fugue of geodin a glycine could align with the glycines of the other homologues, and only a gap was required
Recombinant geodin is monomeric, as are c-crystal-lins and protein S c-Crystalc-crystal-lins have a symmetric inter-domain interface made up by the second motifs
of both domains whereas protein S has a different interface, made up by the second motif of the first domain and the first motif of the second domain The inter-domain interface in c-crystallin (PDB code 4gcr)
is quite extended, made up of hydrophobic interactions between the triad M43⁄ F56 ⁄ I81 in the N-terminal domain and the other triad V132⁄ L145 ⁄ V170 in the C-terminal domain, and Q54–F145 and F56–Q143 hydrogen bonds These positions, marked by * in the alignment (shown in Fig 8), do not appear to be com-pletely conserved in geodin, as some of the hydro-phobic residues involved in c-crystallin are replaced by nonhydrophobic residues
Fig 7 The effects of Ca ions on geodin structure Near-UV (A) and far-UV (B) CD spectra of geodin at pH 5.0 in the presence of Na2EDTA (dotted line) or 1 m M CaCl 2 (solid line) In (C) the temperature unfolding transition of geodin in the presence of Ca2+(1 m M , solid line) is compared to that recorded in the absence of Ca 2+ (1 m M Na2EDTA, dotted line).
Trang 8The inter-domain interface in protein S (PDB code
1 prs) is instead less extended, made up of the
V65⁄ A67 ⁄ Y121 triad, and a N68–D107 hydrogen bond
[13] These positions, marked in the alignment by #,
are not conserved in geodin
Therefore, neither alignment could explain how
domains are assembled in geodin To infer the relative
orientation of domains in geodin, two independent
models were built with modeller As templates we
used, respectively, c-crystallins (PDB codes 4gcr,
1elp,1a45,1a5d) for a ‘mod-gamma’, and protein S
(PDB code 1 prs) for a ‘mod-prs’ anolea [30,31] and prosaii [32] were run on the models and, as a control,
on the template structures Even if the energy calcula-ted for c-crystallins is much lower than that calculacalcula-ted for protein S, anolea indicated that the energy of the model based on 1 prs is comparable with the energy of the model based on c-crystallins, while prosaii indica-ted that the model based on 1 prs is much better than that based on c-crystallins
Figure 9 shows that in mod-prs, the model built with protein S as a template, the inter-domain
Fig 8 Alignment of geodin amino acid sequence with the sequences of homologous bc-crystallins The sequences of bovine (PDB codes 1elp, 1a45, 4gcr, 2bb2) and murine (1a5d) crystallins, and of spore coat protein S from Myxococcus xanthus (1 prs) are decorated as follows: blue for b-strands; red for a-helices, brown for 3–10 helices Buried residues are in uppercase letters; residues with a positive u angle, in ital-ics; hydrogen bonds to main-chain amides in bold; hydrogen bonds to main-chain carbonyls are underlined Buried residues belonging to b-strands and facing the intra-domain hydrophobic core of each domain, other buried residues, and positive-u glycine residues conserved in geodin, are highlighted in green, blue and yellow, respectively The inter-domain interface residues in 4gcr are marked by asterisks The inter-domain interface residues in 1 prs are marked by #.
Trang 9interface is hydrophobic, as made up of W57, I58,
L101, P102 and P106 (shown as yellow as
ball-and-stick residues) Furthermore, correctly intercalated
resi-dues were found, both positively charged (K59 and
R107; blue ball-and-stick residues), and negatively
charged residues (D61 and D100; red ball-and-stick
residues, Fig 9) These charged residues, located at the
surface of the two domains, could also contribute with
hydrogen and⁄ or ionic bonds to the stability of geodin
inter-domain interface As these interactions were not
used as restraints with modeller, they further validate
the model built on protein S, and shown in Fig 9 On
the other hand, the model built using c-crystallins as
templates did not possess a hydrophobic interface or
any other interactions capable of stabilizing
mono-meric geodin
Since the key question at hand was to predict the
relative orientation of domains, we submitted the
mod-gamma, to a protein–protein interaction server,
http://www.biochem.ucl.ac.uk/bsm/PP/server [33] The
results, summarized in Table 2, indicated that the
interface in mod-prs is larger, more planar, although
less circular than that in mod-gamma The interacting
surfaces in mod-prs are more complementary, as
proved by a lower gap volume and gap volume index
(gap volume⁄ interface ASA), with a balanced number
of hydrophobic residues and, interestingly, a higher
number of interdomain hydrogen bonds, and salt
brid-ges In conclusion, the analysis confirms the results
reported above: although mod-prs and mod-gamma
represent alternative ways to assemble the two
domains of geodin, measurement of several descriptive
parameters of the interfaces and visual inspection both suggest that in geodin, as in protein S, the inter-domain interface is made up by the second motif in the first domain and the first motif of the second domain
As it has been found that geodin is stabilized by
Ca2+(see above), the 3D model of geodin with protein
S as a template was analysed to verify whether it could accommodate a Ca2+ binding site Protein S has two binding sites for Ca2+, one per domain They are formed by residues in the folded hairpin of the first Greek key motif and by residues in the loop connect-ing the penultimate and ultimate strands in the second Greek key motif of each domain The site with the highest affinity for Ca2+ in protein S is in the N-ter-minal domain and is defined by residues E10 and E71
It can be proposed that T10, S11 and E62, located in hydrophilic loops in the corresponding region of geo-din, could play the same role This tentative Ca2+ binding site was thus modelled (green ball-and-stick residues) in the geodin structure shown in Fig 9 Another presumable but weaker binding site could be identified in the C-terminal domain of geodin, defined
by Asn93 and Asn145
Spherulin 3A, a single-domain mould protein with crystallin fold, has two distinct Ca2+binding sites per domain, one site is located between the folded hairpin
of the first Greek key motif and the loop between the last two strands of the second Greek key motif [13] This site corresponds to the sites identified in protein S and modelled in geodin
Conclusions
The results of the physico-chemical studies reported above, and those from homology modelling, lead to
Fig 9 A model of geodin built using the structure of protein S
(1 prs) as a template Inter-domain interface residues are shown as
yellow ball-and-stick residues when they are hydrophobic; in red
when they are negatively charged; in blue when they are positively
charged Residues proposed to describe the binding site for Ca2+
are in green ball-and-stick notation The structure was drawn using
MOLSCRIPT [37] and rendered using RASTER 3 D [38].
Table 2 Domain–domain interface analysis of alternative models of geodin domain assembly.
Mod-prs domain-1
Mod-prs domain-2
Mod-gamma domain-1
Mod-gamma domain )2 Interface ASA 643.29 741.38 415.18 404.38
% Interface ASA 13.62 14.73 8.92 8.29 Planarity 2.14 2.13 1.02 1.28 Length ⁄ breadth 0.49 0.54 0.74 0.71
% Polar atoms
in interface
38.09 43.61 32.79 64.11
% Non–polar atoms in interface
61.90 56.30 67.20 35.80
Hydrogen bonds 3 3 1 1
Gap volume 1644.62 1644.62 2927.67 2927.67 Gap volume index 1.19 1.19 3.57 3.57
Trang 10the first structural description of geodin, and validate
its identification as a homologue of the bc-crystallin
superfamily Geodin is a monomeric, two-domain
protein, made up of b strands apparently folded in
Greek-key motifs, just as its mammalian, amphibian
and bacterial homologues Of particular interest is that
its source is a sponge, G cydonium, which makes
geo-din the most primitive metazoan bc-crystallin studied
so far
By both fluorescence and CD spectroscopic analyses,
geodin is found to be, with respect to other
bc-crystal-lins, more readily but reversibly unfolded by chemical
denaturants, with a single midpoint of denaturation,
both in urea and GuHCl These features are very
sim-ilar to those found for protein S [22] Consistently, our
conclusion from homology modelling strongly suggests
that the closest structural homologue to geodin is
bac-terial protein S
Geodin thermal denaturation, studied both by
near-and far-UV CD near-and calorimetry, shows that geodin
unfolds with a typical cooperative transition in
agree-ment with a two-state model, with a simultaneous
col-lapse of secondary and tertiary structures This is
difficult to explain, given the larger structural
differ-ences in primary structure and 3D architecture
between the two geodin domains, as they result from
the proposed sequence alignment and model,
com-pared with the more similar sequence and architecture
of the two domains in the other monomeric crystallins
studied so far It should be considered that, besides the
structural differences among the proteins under
com-parison, different pH values were adopted in the
experiments
The similarity in structural properties between geodin
and protein S extends to the stabilization exerted by
calcium ions on both crystallin-type proteins, although
the removal of calcium affects in geodin both
secon-dary and tertiary structure, whereas for protein S only
the tertiary structure is affected Furthermore, the
effect of calcium on geodin stability is less conspicuous,
with an increase inDG0 of denaturation of 7 kJÆmol)1,
compared to that reported for protein S (19 kJÆmol)1)
This outcome is due mainly to the higherDH0
contribu-tion for protein S denaturacontribu-tion in the presence of
Ca2+, in turn apparently due to strengthened
inter-domain interactions
As from the evolutionary point of view, it is of
interest that the crystallin-type structural organization,
already evolved in bacteria and moulds, was readily
recruited for porifera, the earliest metazoans
Further-more, it is interesting that in the model as derived for
geodin the inter-domain interfaces appear to be
stabil-ized by hydrophobic interactions, but also by a
number of polar interactions, such as H-bonds and salt linkages These contacts can be interpreted, as it has been proposed for vertebrate bc-crystallins [5], as remi-niscent of the polar, solvent exposed surfaces in the putative single-domain ancestor(s) that evolved to associate into two-domain geodin
Experimental procedures
Cloning and expression of geodin Cloning into the pET22b(+) vector (Novagen, Madison,
WI, USA) of the DNA segment encoding geodin, a puta-tive bc-crystallin-type protein from G cydonium, was carried out as described previously [17] The plasmid was used to transform E coli strain BL21 (DE3) (from AMS Biotechnology) For over-expression of geodin, the bacterial cultures were grown at 37C to D600¼ 1, then induced by addition of 0.1 m isopropyl-1-thio-d-galactopyranoside After overnight growth at room temperature, cells were pel-leted by centrifugation, and lysed by sonication An Ultra-sonic Ultra-sonicator (Heat System UltraUltra-sonic) was used at
20 kHz, with 30-s impulses, each followed by a 30-s rest period, for a total time of 15 min
A protein with the molecular size of geodin was detected
by SDS⁄ PAGE as a soluble protein in the lysate superna-tant (henceforth termed S preparation), but also in the insoluble material (Fig 1) Thus, the insoluble material, washed twice in 50 mm Tris⁄ HCl pH 8 containing 20 mm
Na2EDTA, and once in 100 mm Tris⁄ HCl pH 8 containing
1 mm Na2EDTA, was solubilized by denaturation with 6 m GuHCl in 25 mm sodium phosphate pH 7 (buffer P) Rena-turation followed, obtained through extensive dialysis against the same buffer This yielded another preparation containing a protein with the mobility of geodin, henceforth termed IB preparation, as presumably it consists of proteins sequestered in inclusion bodies
Both preparations S and IB were fractionated by gel fil-tration, carried out on a column of Superdex G-75 (Amer-sham Biosciences) equilibrated with buffer P containing 0.3 m sodium chloride Figure 2 illustrates that the fraction-ation of both S and IB preparfraction-ations yielded three main UV absorbing fractions Only in the middle peak from either S
or IB preparations SDS⁄ PAGE runs revealed the presence
of a protein band with the molecular size expected for geo-din The corresponding fractions (Fig 2A,B, insets) were pooled and dialysed against 50 mm ammonium acetate buf-fer pH 5
The Superdex G-75 fraction pools obtained from the S and IB preparations were each loaded onto a cation exchange Resource-S column (Amersham Biosciences) equilibrated in 50 mm ammonium acetate pH 5, and eluted with a 60-min linear gradient from 50 to 300 mm ammo-nium acetate As shown in Fig 2C most of the protein