The ferricytochrome-c (cyt-c) shows a complex unfolding
pathway characterized by a series of stable partially folded states. When titrated with HCl at low ionic strength, two transitions are
detected. At pH 2, cyt-c assumes the U1
unfolded state, whereas the successive addition of Cl
ion
from either HCl or NaCl induces the recompaction to a molten globule
conformation (A1 and
A2 states, respectively). A second unfolded
state (U2) is also observed at pH 12. Recent
data evidence different features for the local structure of the
heme in the different states. To derive relationships between local and
overall conformations, we analyzed the structural characteristics of
the different states by synchrotron small angle x-ray scattering. The
results show that in the acidic-unfolded U1
form the protein assumes a worm-like conformation, whereas in the
alkaline-unfolded U2 state, the cyt-c is
globular. Moreover, the molten globule states induced by adding HCl or
NaCl to U1 appear structurally different: in
the A1 state cyt-c is dimeric and less
compact, whereas in the A2 form the protein
reverts to a globular-like conformation. According to the local heme
structure, a molecular model for the different forms is derived.
 |
INTRODUCTION |
The present work deals with a study of the
structural properties of partially folded states of ferricytochrome-c
(cyt-c) in acidic and alkaline solutions using small angle x-ray
scattering (SAXS) by synchrotron radiation sources.
Determination of the structural characteristics of partially folded
intermediate state in proteins is crucial for understanding the
mechanism of folding and the principle of structure stabilization. Among partially denatured states, the molten globule has been devoted
of a special interest. The molten globule is a compact denatured
protein form, with a significantly native-like secondary structure but
a largely flexible and disordered tertiary structure. On the basis of
structural and kinetic studies, the molten globule has been proposed as
the major intermediate in globular protein folding (Goto et al.,
1990a
,b; Christensen and Pain, 1991).
Horse cyt-c is a well characterized globular protein both in the
crystalline and in solution states and it represents a very useful
model for protein folding studies (Scott and Mauk, 1996). In
particular, cyt-c has been shown to exist in three stable states in
acidic pH region, corresponding to the native (N), unfolded (U1) and compact intermediate or
molten globule (A) forms (Goto et al., 1990a; Ohgushi and
Wada, 1983; Fink et al., 1994), and in one unfolded state stable at
alkaline pH (the U2 state). In recent
years, the U2 form has been the object
of extensive studies, especially in connection with the role of
conformational changes in electron transfer (Wilson and Greenwood,
1996; Döpner et al., 1998; Rossel at al., 1998; Weber at al.,
1987).
The structural properties of the cyt-c unfolded states have been mainly
analyzed in terms of gyration radii and of changes in protein
compactness also using time-resolution techniques (Kataoka et al.,
1993; Banci et al., 1999; Pollack et al., 1999). Moreover, the
definition of the conformational characteristics of the molten globule
form is still an open problem. In fact, acid-unfolding of cyt-c occurs
substantially under conditions of low salt at pH 2 (U1 state); the subsequent addition of
Cl ion from either HCl or NaCl induces the
recompaction to a molten globule state, indicating that the chloride
anion should play a key role in acid-salt induced refolding. The state
induced by addition of more acid is defined as
A1 state, whereas the compact intermediate state obtained by addition of NaCl at pH 2 is the A2 state (Boffi et al., 2001). The
unfolding/refolding pathways of cyt-c is then sketched below:
Optical and circular dichroism spectroscopy showed that the two
intermediate states A1 and
A2 have similar secondary structure (Goto et al., 1990a,b). In a recent work, x-ray absorption spectroscopy (XAS) at Fe K-edge has been used to characterize the local environment of the heme iron of cyt-c in acidic and basic conditions (Boffi et al.,
2001): the comparison of XAS protein spectra with experimental model
systems showed that local site structures of the two intermediate states induced by adding HCl (A1) or
NaCl (A2) to the HCl denatured state
(U1) are different. These results
indicate that the formation of a native-like secondary structure
induced both by acid and salt is not related to the coordination at the
heme pocket level. This prompt us to reexamine this question by using a
more direct probe than the optical spectroscopy to characterize the
partially folded states of cyt-c.
SAXS is ideally suitable to this purpose (Trewhella, 1997; Pérez
et al., 2001 and reference therein). Note that while XAS is a
short-range probe sensitive to variations in the local structure, SAXS
is a powerful tool for analyzing the size and the shape of protein
molecules, thus information on the relationship between local and
overall conformation can be obtained by the coordinated use of both
techniques. In this work, SAXS has been then used to characterize the
size and the shape of cyt-c in all the stable states N,
U, and A. Moreover optical absorption
measurements have been performed to monitor the pH and salt induced
transitions between them.
| |
MATERIALS AND METHODS |
Cytochrome-c
Cyt-c is a well-know monomeric protein with a molecular mass of
12,000 Da, as calculated from the amino acids composition. The protein
dry volume and averaged electron density, also derived from the amino
acid composition using data reported by Jacrot and Zaccai (Jacrot,
1976; Jacrot and Zaccai, 1981), are 14,500 Å3
and 0.438 e Å3, respectively. The theoretical
radius of gyration, Rg,th, calculated as indicated below from the atomic coordinates, is 12.6 Å.
Horse heart cyt-c (type III) has been obtained from Sigma (St. Louis,
MO). Samples for both SAXS and optical absorption measurements have
been prepared dissolving the dry powder in bidistilled water. To remove
all the ions eventually contained in the commercial powder, the
solution has been subsequently dialyzed overnight against bidistilled
water. After dialysis, the samples have been diluted to the appropriate
concentration of 0.8 mM (corresponding to c = 10 g
l1). Final samples have been prepared from the
same native water solution by adding negligible volumes of high
concentrated acid-basic-salt solutions to reach the required molar
ratios; consequently, the added volumes did not change appreciably the
protein concentration. pH measurements and titrations have been made
using a Crison micro-pH 2000 pH-meter.
The experimental procedure has been the following: 1) a pH titration
has been carried out on the protein from pH 7 to 12 using NaOH; 2) a pH
titration has been carried out on the protein from pH 7 to 0.5 using
HCl in absence of salt; 3) a salt titration with NaCl has been carried
out over 0 to 500 mM range at the pH of maximal unfolding (40 mM HCl at
pH ~ 2).
Optical absorption measurements
All absorption spectroscopy measurements have been performed
using the JASCO V-750 instrument. The concentration of native cyt-c has
been determined on suitable dilute protein samples using extinction
coefficient 
of 1.06 105
M1 cm1 at 410 nm and
25°C. To make a comparison with folding studies based on structural
probes, the conformational transitions of cyt-c have been studied at
695 nm ( = 810 M1
cm1) and at 620 nm ( = 1100 M1 cm1), which allow us
to monitor changes of the heme absorption spectrum at a concentration
that is a hundred times higher than those usually used for optical spectroscopy.
SAXS experiments
SAXS experiments were performed using two different
synchrotron beamlines, namely the SAS beamline at the LNLS
(National Synchrotron Light Laboratory, Campinas, Brazil) and the ID2
beamline at ESRF (European Synchrotron Radiation Facility, Grenoble,
France). At LNLS, the scattering vector Q ranges
(Q is the scattering vector modulus defined as
Q = 4
sin
/
, being 2 the scattering angle, and the x-ray wavelength) from 0.024 to 0.22 Å1. At ESRF, the Q-range was 0.022 to 0.30 Å1. Cyt-c samples were measured at
room temperature in 1.5-mm glass capillaries at ESRF (1 mm at LNLS). To
avoid radiation damage, the exposure time was 0.1 s/frame at ESRF (200 s/frame at LNLS), and a lead shutter was used to protect the sample
from excess radiation within periods where no data were recorded. The
final scattering pattern was then averaged from at least 30 frames at ESRF (four frames at LNLS). The experimental intensities (radially averaged at ESRF) were corrected for background, buffer contributions, detector inhomogeneities, and sample transmission. The scattering cross-sections have been converted to absolute units
(cm1) by calibration with lupolen.
Data analysis
Basic equations
The small angle scattering of x-ray (or neutrons) is a powerful
tool that yields information on the overall shape and size of
biological macromolecules in solution. Its application is thus particularly useful in studying systems where large structural or
conformational changes take place, like processes of protein folding/unfolding (Trewhella, 1997; Kataoka et al., 1993, 1995; Pollack
et al., 1999; Pérez et al., 2001 and reference therein). In the
frame of the so-called two phase model, monodisperse proteins in a
diluted solution (in the order of 105 M) act as
randomly oriented scattering particles of homogeneous electron density
, dispersed in a solvent with different homogeneous electron density
s (Guinier and Fournet, 1955). In this
condition, the excess x-ray scattering intensity reduces to
|
(1)
|
in which c is the molar protein concentration,
= 
s is the contrast,
and V is the protein volume. P(Q) is
the averaged squared form factor, which contains information on the
size and shape of the protein molecule.
SM(Q) is the structure
factor based on the interference between waves scattered by different
proteins. Its effect can be neglected
(SM(Q) 
1) for diluted
solutions. An isotropic Fourier transform connects the form factor
P(Q) to the distance distribution function
p(r), the probability of finding pair of small
volume elements at a distance r within the entire volume of
protein,
|
(2)
|
The behavior of I(Q) at small Q
is approximated by the Guinier law (Guinier and Fournet, 1955)
|
(3)
|
in which Rg is the gyration
radius, defined for a homogeneous scattering particle by
|
(4)
|
and
|
(5)
|
is the scattering intensity at zero angle. For globular
proteins, the Guinier approximation is strictly valid only for
QRg 
1.3 (Feigin and Svergun, 1987).
It should be observed that Rg and
I(0) are respectively related to the average dimension of the protein and to the concentration of the scattering particles. In
the case of aggregation, for monodisperse scattering particles constituted by n monomers of volume (V), the Eq. 5 transforms to
|
(6)
|
Under the assumption of monodispersity, from the Guinier
analysis it is then possible to determine the aggregation state of the protein.
In the paper, we will made an extensive use of Kratky plots (i.e.,
Q2I(Q) versus
Q plot) to show the scattering data. The Kratky plot is
indeed a useful tool in SAXS analysis for the characterization of
globular protein and for the detection of intermediate folded states
(Kataoka et al., 1993, 1995; Semisotnov et al., 1996). For a globular
particle, the Kratky plot shows a typical peak, whose position mainly
depends on its gyration radius Rg. On
the other hand, when an unfolding process takes place, the peak usually vanishes and the curve tends to show a plateau when the protein assumes
a completely unfolded random-coil conformation.
Concerning the p(r) function, which contains
direct information on the protein structure, it should be observed that
it can be calculated from the experimental I(Q)
through the Eq. 2 only in a few cases, due to the limitation of the
accessible Q range. Therefore, to reconstruct the shape of
the scattering particles, different procedures should be adopted,
usually based on the comparison of the form factor of refined model
shapes to the experimentally observed scattering intensity. The
different method used in this work are summarized below.
Monte Carlo form factor for globular protein
When the crystallographic structure of a protein is known, the
distance distribution functions p(r) of the
protein can be calculated using Monte Carlo methods (Hansen, 1990;
Henderson, 1996; Ashton et al., 1997; Svergun, 1997; Mariani et al.,
2000). According to Mariani and coworkers (2000), the scattering
particle can be described by the s(r) function,
which gives the probability that the point r 
(r, 
r) (in which
r indicates the polar angles
r and 
r) lies within
the particle. For a compact particle (Stuhrmann et al., 1977), the
s(r) function can be written in terms of a unique
two-dimensional angular shape function 
(r), as
![s(<B><UP>r</UP></B>)=<FENCE><AR><R><C><UP>1</UP></C><C><UP>r≤ℱ</UP>(<UP>&ohgr;<SUB>r</SUB></UP>)</C></R><R><C><UP>exp</UP>{<UP>−</UP>[r−ℱ(&ohgr;<SUB><UP>r</UP></SUB>)]<SUP>2</SUP>/2&sfgr;<SUP>2</SUP>}</C><C><UP>r>ℱ</UP>(<UP>&ohgr;<SUB>r</SUB></UP>)</C></R></AR></FENCE>](/content/vol81/issue6/fulltext/3522/img008.gif)
|
(7)
|
in which 
is the width of the Gaussian that accounts for the
effect of the chain mobility on the protein surface or/and to the
presence of a hydration shell with density (and then electron density)
different from the density of the bulk water (Svergun et al., 1998).
The function (r) is evaluated from the envelope surface of the van der Waals spheres placed in the crystallographic coordinates. The p(r) histogram is then
calculated taking into account the distances between all pairs of
M points randomly generated and evaluated according to Eq. 7. The Monte Carlo form factor is then calculated by the Eq. 2.
For the native cyt-c, we used the high-resolution nuclear magnetic
resonance and restrained simulated annealing structure reported by
Banci and co-workers (1999) and deposited in the protein data bank
(http://www.rcsb.org/pdb/) with the entry code 1GIW. The
p(r) functions obtained using M = 5000 and = 0 (dry protein) and = 1.3 Å (hydrated
protein, see below) are reported in Fig. 4 (see below). The
corresponding radii of gyration, calculated using Eq. 4, were 12.6 and
13.8 Å, respectively, whereas the protein volumes, determined by the
ratio between the accepted and the total number of Monte Carlo moves
were 13,900 and 18,600 Å3, respectively.
Model form factors for unfolded protein
The low-resolution structure of unfolded or partially folded
states of a protein is quite a tough problem, which cannot be tackle
with a model-independent approach. In most cases, denatured proteins
are a collection of different conformers where the secondary structure
is approximately conserved. In the case of a completely unfolded chain,
a suitable model to be applied is the Debye one (Debye, 1947), which
is a flexible chain with a random walk-like conformation
(random-coil chain). In this model, the distribution of the scattering
centers in the real space follows a Gaussian statistics. The
corresponding form factor is given by
![P<SUB><UP>Debye</UP></SUB>(Q)=2[<UP>exp</UP>(<UP>−</UP>R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>Q<SUP>2</SUP>)+R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>Q<SUP>2</SUP>−1]/(R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>Q<SUP>2</SUP>)<SUP>2</SUP>](/content/vol81/issue6/fulltext/3522/img009.gif)
|
(8)
|
with R
= Lb/6.
b is the statistical segment (Kuhn) length, representing the
separation between two adjacent scattering centers, and L is
the contour length, a measure of the chain length. It should be
observed that to derive the Rg value, the
Guinier approximation (Eq. 3) should be used in the usual
Q-range for proteins in rather globular states only, whereas
for unfolded proteins the Debye equation should be preferred (for
example, see Pérez et al., 2001).
More physical models describing semiflexible or worm-like
polymers have been developed. A detailed analysis has been recently published by Pedersen and Schurtenberger (1996). They report a Monte
Carlo simulation study of the worm-like chain model of Kratky and Porod
(1949), both considering the presence or the absence of excluded volume
effects. Results are given in terms of approximated analytical
expressions, which have been parametrized to reproduce SAXS scattering
profiles. In particular, in the absence of excluded volume effects, the
worm-like chain form factor, regardless of the cross-section
contribution, can be determined by
![×<UP>exp</UP>[<UP>−</UP>((Qb)/q<SUB>1</SUB>)<SUP><UP>p<SUB>1</SUB></UP></SUP>]+<FENCE><FR><NU>1</NU><DE>LbQ<SUP>2</SUP></DE></FR>+<FR><NU>&pgr;</NU><DE>LQ</DE></FR></FENCE>](/content/vol81/issue6/fulltext/3522/img013.gif)
|
(9)
|
when L/b > 2, and by
![P<SUB>2</SUB>(Q)={2[<UP>exp</UP>(<UP>−</UP>⟨R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>⟩<SUB>0</SUB>Q<SUP>2</SUP>)+⟨R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>⟩<SUB>0</SUB>Q<SUP>2</SUP>−1]/(⟨R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>⟩<SUB>0</SUB>Q<SUP>2</SUP>)<SUP>2</SUP>}×<UP>exp</UP>[<UP>−</UP>((Qb)/q<SUB>2</SUB>)<SUP><UP>p<SUB>2</SUB></UP></SUP>]+<FENCE><FR><NU>a<SUB>1</SUB></NU><DE>LbQ<SUP>2</SUP></DE></FR>+<FR><NU>&pgr;</NU><DE>LQ</DE></FR></FENCE>](/content/vol81/issue6/fulltext/3522/img015.gif)
|
(10)
|
with
![⟨R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>⟩<SUB>0</SUB>=<FR><NU>Lb</NU><DE>6</DE></FR> <FENCE>1−<FR><NU>3</NU><DE>2n<SUB><UP>b</UP></SUB></DE></FR>+<FR><NU>3</NU><DE>2n<SUP><UP>2</UP></SUP><SUB><UP>b</UP></SUB></DE></FR>−<FR><NU>3</NU><DE>4n<SUP><UP>3</UP></SUP><SUB><UP>b</UP></SUB></DE></FR> [1−<UP>exp</UP>(<UP>−</UP>2n<SUB><UP>b</UP></SUB>)]</FENCE>](/content/vol81/issue6/fulltext/3522/img017.gif)
|
(11)
|
and nb = L/b when
L/b 2. The optimum parameter values are
q1 = 5.53, p1 = 5.33, a1 = 0.0625, p2 = 3.95, a2 = 11.7, and q2 = a2/L, as shown in Pedersen
and Schurtenberger (1996). In the case of finite section of the chains,
the form factors described by Eqs. 8 (random-coil chain) and 9 and 10 (worm-like chain) should be multiplied for the form factor of the
cross-section Ssc(Q). The
simplest approximation is to assume a local cylindrical shape, which
reads Ssc(Q) = [2J1(RQ)/(RQ)]2,
being J1(RQ) the first
order Bessel function and R the cross-section radius.
Protein shape reconstruction from the multipole expansion
method
In the case of monodisperse diluted protein, a well-established
method for particle shape reconstruction is based on the expansion in
series of spherical harmonics of the shape function,
|
(12)
|
Here K denotes the maximal rank of the spherical
harmonics
Yk,m(r)
(Svergun and Stuhrmann, 1991). Recently this method was improved,
introducing the group theory and the maximal entropy to deduce in a
more efficient way the shape of the scattering particles (Spinozzi et
al., 1998). The idea was to introduce the particle symmetry by
considering its point group 
, and then to explore the fitting
parameter space using a sequence of decreasing symmetries. The
resulting symmetryzed shape function is
|
(13)
|
in which the symmetry spherical harmonics
Y
(r) are
compatible with the group and
{ak,m} is a set of parameters that
can be determined to fit the experimental scattering curve. The finally
recovered shape is the best compromise between a simple and hence
symmetrical shape, i.e., a shape described by few parameters, and a
good agreement with the experimental data.
Data fitting
In all cases, the analysis of the experimental curves
I(Q) of NQ
points has been performed by minimizing the reduced chi squared:
|
(14)
|
in which P(Q) is the fitting form factor,
is a scaling factor corresponding to the fitted scattering
intensity at Q = 0 (see Eq. 15),
i is the experimental uncertainty of the
scattering curve at the point Qi, and
NP is the number of fitting parameters.
Estimate of the protein tertiary structure from the shape function:
the filling approach
If a relevant part of the secondary structure of a protein is
known, it can be useful to have a method that allows an estimate of the
tertiary structure compatible with the reconstructed two-dimensional shape function (r). Here we propose a simple
algorithm to accomplish this task. The
NS secondary structure elements,
namely -helix or -sheet domains, are separated by random coil
sequences, which can be thought of as hinge elements. Due to the low
resolution of the SAXS technique, it is useless to describe in detail
all conformational changes of the random sequences, but it can be a
reasonable simplification to leave fixed their structure except a free
rotation around the C
atom in the middle of
the sequence. In this way, the protein conformational degrees of
freedom are simply reduced to rotations (typically described by three
Euler angles) around those hinges. A tertiary structure will be defined by the NS 1 sets of three Euler
angles, {
(,
)}, which give the orientation of the
k-th domain with respect the (k 1)-th.
Moreover, the space configuration of the whole protein will be
determined by the translation vector, RP,
of its mass center and by the Euler angles, P,
describing its overall orientation. In the case of protein aggregates
(e.g., a dimer), we need of more translational and orientational
coordinates for describing the mutual position between single monomers.
That tertiary structure where the largest number of atoms are contained
in the volume enclosed by the SAS-reconstructed shape will be
the best estimate compatible with the experimental data. The "best
filling" conformation can thus be calculated by minimizing the root
mean square (rms)
|
(15)
|
in which, V is the protein volume,
NA the number of atoms,
s(r) the position function (Eq. 7),
V
the van der Waals volume of
the i-th atom, and ri its
position, which is a function of the configurational variables
RP, P,
{}. Here
s(ri) plays the role of a filter
function, which discards the i-th atom if it is not inside
the shape. The configurational variables are optimized by using a Monte
Carlo trial and error procedure.
It is worth notice that the filling method is here used only to test if
the reconstructed shape fits with the mean dimensions of the cyt-c
structure elements. Any attempt to derive the real tertiary structure
from SAXS data by using this approach falls outside the aim of this
work and will be the object of a specific paper.
| |
RESULTS AND DISCUSSIONS |
Optical absorption measurements
Acidification of a salt-free solution of cyt-c using HCl leads to
an increase of the absorbance peak at 620 nm, characteristic of
high-spin complexes. Optical spectra of cyt-c show the presence of two
subsequent transitions with increasing the HCl concentration (Fig.
1). The first transition is ascribed to
N
U1 and the second to
U1A1
state. The former is a cooperative transition, which occurs at [HCl] = 14 mM for 0.8 mM concentration of protein. The latter can be obtained
by further addition of HCl. The presence of these two transitions, when
the protein is titrated in absence of salt, is a well-documented
behavior of the cit-c (Ohgushi and Wada, 1983; Weber et al., 1987; Goto
et al., 1990a; Fink et al., 1994). As previously shown (Boffi et al.,
2001), the two transitions are well resolved at [cyt-c] = 0.8 mM,
whereas they partially overlap at higher concentrations. This fixes a
limit to the protein concentration that can be used in SAXS
measurements to characterize the U1
state. In fact, at high protein concentration, upon decreasing the pH,
the protein will go directly into A1
state from the N state.
View larger version (20K):
[in this window]
[in a new window]
|
FIGURE 1
Changes in extinction coefficients at 695 nm () and
at 620 nm (µ,  ) as a function of the concentration of NaOH (),
HCl (µ), and NaCl at pH 2(). The size of the symbol is
representative of the error. Lines are guide for eyes.
|
|
The addition of NaCl to the protein in the
U1 state also leads to a collapse into
a partially refolded intermediate state (A2) (Goto et al., 1990a,b). The
principal responsible for the refolding is thought to be the binding of
chloride anions. In fact, they minimize the intramolecular charge
repulsion that initially caused the protein unfolding. However, as
shown in Fig. 1, the at 620 nm of the NaCl induced state
A2 is larger than that of HCl induced
one, indicating that some differences between
A1 and
A2 conformations should occur. This
may reflect the difference in heme iron local structure indicated by
XAS measurements (Boffi et al., 2001).
In the alkaline-unfolded U2 state, the
absorption band of cyt-c at 695 nm is absent; the complete disappearing
of this band indicates that protein looses the methionine-80 as sixth
ligand to the Fe heme. The dependence on NaOH concentration of the
protein extinction coefficient at 695 nm is also shown in Fig. 1. In
this case the cyt-c unfolds in a single cooperative transition around [NaOH] = 5 mM, assuming the U2
unfolded state at [NaOH] = 30 mM.
SAXS measurements
SAXS experiments have been performed on cyt-c solutions at low
ionic strength at different pH (pH 7 in the native conformation N; pH 2 in the fully acid-unfolded state
U1; pH 12 in the fully alkaline-unfolded state U2) and
further decreasing the pH from 2 to 0.5 using HCl (to obtain the
A1 state) or increasing the ionic
strength at constant acidic pH from 0 to 500 mM using NaCl (to obtain
the A2 state). Examples of the
experimental SAXS profiles obtained in the different investigated
conditions are reported in the Fig. 2 in
the form of Kratky plots.
View larger version (22K):
[in this window]
[in a new window]
|
FIGURE 2
SAXS profiles obtained from cyt-c at concentration 0.8 mM in native (N), acid-denatured
(U1), acid-induced recompacted
(A1), salt-induced recompacted
(A2), and alkaline-denatured
(U2) states. All curves were measured at
ESRF, beamline ID02, except the native sample, which was measured at
the SAXS beamline of LNLS. The curves are in the form of Kratky plots
and are normalized for the I(0) values obtained by
applying the Guinier law (Table 1) and scaled by a factor 2 for
clarity. The solid curves correspond to the fits obtained by using the
Monte Carlo method (N, A1,
U2), the worm-like model
(U1), and the multipole expansion method
(A2), as described in the text.
|
|
The radii of gyration and the intensities scattered at
Q = 0 have been obtained by using both Guinier and
Debye laws (Eqs. 3 and 8). However, only the Guinier results are
reported for the following reasons: 1) the Kratky plots show the
presence of a more or less pronunced peak and 2) the conclusions
reached in this work (see next sections) give an indication that a
rather high degree of globularity is preserved in the investigated
cyt-c states. The fitting results are then shown in Fig.
3 and Table 1. It should be first observed that the
Rg do not show a significant trend
with protein concentration, indicating that interaction effects can be
disregarded, i.e., the obtained parameters can be equated to the actual
structural characteristics of the cyt-c in the different stable states.
Moreover, it can be noticed that the radii of gyration of the alkaline
unfolded state (U2) and of the molten
globule intermediate induced adding NaCl at pH 2 (A2 form) are rather similar to that
of the native protein (N). By contrast, larger
Rgs are observed both for the acidic
unfolded state (U1) and for the
acid-recompacted state (A1). These
experimental data can be compared with the radius of gyration of the
cyt-c determined from the crystallographic coordinates (Banci et al., 1999), Rg,th = 12.6 Å. Whereas the
small variations could be related to changes in the hydration
properties of the protein surface or to small conformational
adjustments induced by solvent effects, the large radius of gyration
observed at very low pH (U1 and
A1) strongly suggests the occurrence
of large conformational changes or the formation of protein aggregates.
From this point of view, it should be observed that a radius of
gyration of 32.4 (±1.6) Å has been reported by Kataoka and co-workers
(1993) for the cyt-c in 4 M guanidine-HCl, which is thought to fully
unfold the protein.
View larger version (13K):
[in this window]
[in a new window]
|
FIGURE 3
Guinier plots of SAXS data obtained from cyt-c at
concentration 0.8 mM in the native (N), acid-denatured
(U1), acid-induced recompacted
(A1), salt-induced recompacted
(A2), and alkaline-denatured
(U2) states. The curves are normalized for
the I(0) values obtained by applying the Guinier law
(Table 1) and scaled by a factor 5 for clarity. The solid lines
correspond to the fits obtained by using the Guinier law.
|
|
View this table:
[in this window]
[in a new window]
|
TABLE 1
Radii of gyration and forward scattering intensities
calculated by applying the Guinier approximation (Eq. 3) to SAXS data
from cyt-c in the native (N), acid-denatured
(U1), acid-induced recompacted
(A1), salt-induced recompacted
(A2), and alkaline-denatured
(U2) states
|
|
According to the Eq. 6, information on the protein aggregation
properties can be directly obtained from the scattered intensity at
zero angle. Assuming that the solution is monodisperse and using for
the cyt-c the molecular volume of 18,600 Å3 (see
above) and the electron density contrast calculated in the different
experimental conditions from the solution chemical composition (namely,
taking into account the presence of salts and counter-ions), the
aggregation numbers n in Table 1 have been determined. In particular, the presence of dimer is strongly suggested for the A1 state. However, molecular
information can be hardly derived from these parameters. Further
structural information has been then obtained, applying the different
approaches above described (see Materials and Methods) to analyze the
SAXS curves. The results are reported in the following paragraphs in
which the different cyt-c states are separately analyzed.
Native N state
The SAXS results obtained at pH 7 have been analyzed by fitting
the experimental curve with the cyt-c crystallographic structure (Banci
et al., 1999). However, because SAXS data indicate the presence of
scattering particles with a radius of gyration larger than that
estimated from the atomic coordinates, the presence of a border shell
around the protein, attributed to the mobility of the chains on the
protein surface or/and to the presence of a hydration shell, has been
taken into account (Svergun et al., 1998).
The calculated scattering intensity (in the form of Kratky plot) and
the best fit parameters are reported in Fig. 2 and Table 2, respectively. From the figure, the
quality of the fit can be directly appreciated: it is interesting to
note that the width of the Gaussian that accounts for the particle
border, , is 1.3 Å, in agreement with data obtained for other
globular proteins (Svergun et al., 1998; Baldini et al., 1999). The
corresponding scattering particle volume, as determined using the Monte
Carlo procedure described by Mariani and co-workers (2000), is 18,600 Å3, in perfect agreement with the particle
volume that can be determined from the calculated scattering curve from
the Porod invariant V = 22/
P(Q)Q2dQ (Guinier and Fournet, 1955).
View this table:
[in this window]
[in a new window]
|
TABLE 2
Fitting results of SAXS curves from cyt-c in the native
(N), acid-denatured (U1),
acid-induced recompacted (A1), salt-induced
recompacted (A2), and alkaline-denatured
(U2) states
|
|
For the sake of completeness, the distance distribution function
p(r) and the three-dimensional structure of the
N cyt-c are also reported in Figs.
4 and
5, respectively. The unimodal
shape of the p(r) function confirms the globular
and compact nature of the protein in the native conformation. It should
be observed that the largest dimension of the protein is ~40 Å.
View larger version (17K):
[in this window]
[in a new window]
|
FIGURE 4
Distance distribution functions obtained from the
crystallographic coordinates of cyt-c (dry) and by the
analysis of SAXS data obtained in the native (N),
acid-denatured (U1), acid-induced
recompacted (A1), salt-induced recompacted
(A2), and alkaline-denatured
(U2) states. For the dry, N,
A2, and U2 forms,
the p(r) curves have been calculated by
the Monte Carlo method with a variance of the Gaussian border shell
of 0, 1.3, 1.5, and 2.3 Å, respectively; for the
U1 and A1 states,
the p(r) curves have been obtained by
Fourier transforms (see Eq. 2) of the corresponding fitting curves
reported in Fig. 3.
|
|
View larger version (43K):
[in this window]
[in a new window]
|
FIGURE 5
Three-dimensional representations of the structure of
cyt-c in the native (N), acid-denatured
(U1), acid-induced recompacted
(A1), salt-induced recompacted
(A2), and alkaline-denatured
(U2) states, as resulting from the analysis
of the corresponding SAXS data. (Top) View of the cyt-c,
PDB code 1GIW (Banci et al. 1999), using RasMol 2.6 package
(www.umass.edu/microbio/rasmol) with the "strands" option. The
sequence of the 104 amino acids is represented in different colors
corresponding to four regions of secondary structure: 1 to 16 (magenta) contains the large N-terminal -helix; 17 to
57 (blue) includes the subsequent short -helix; 58 to
84 (black) contains the two other short -helices; 85 to 104 (cyan) includes the large C-terminal -helix.
The heme group and the two bound amino acids His-18 and Met-80 are
represented using the "spacefill" option and with colors red,
green, and orange, respectively. (N,
A2, U2)
"Spacefill" representation of the 1GIW structure used to analyze
SAXS data with the Monte Carlo method. The yellow border, corresponding
to an indicative variance = 3 Å of the Gaussian that accounts
for the expanded conformation detected in the different cases, is
shown. (U1) Sketch of a possible
conformation of the worm-like model constituted by two cylindrical
subunits (yellow). To best fill the entire volume, the
1GIW tertiary structure has been modified rotating the four regions of
secondary structure around the three separating amino acids (see text).
(A1) Shape function,
(r) (yellow), obtained by the
multipole expansion method. According to the C2 symmetry of
a dimer, two 1GIW structures with modified tertiary structures have
been used to best fill the reconstructed shape. The symmetry center
position and the opening angles have been chosen by the best filling
procedure described in the text.
|
|
Acid-denatured U1 state
At the neutral pH, cyt-c is a globular protein having two
strong-field ligands to the heme iron, an imidazole nitrogen of His-18
and a sulfur of Met-80, coordinated in axial positions of the heme
plane. When titrated with HCl in absence of salt (Fig. 1), the optical
spectra of the protein show a first transition at pH near 3, resulting
in the unfolded state U1, where the
heme axial ligand is probably lost. Indeed, in agreement with the
undefined nature of the sixth ligand (Scott and Mauk, 1996), the XANES
spectrum presents characteristics of a met-myoglobin at pH 4 (Boffi et al., 2001).
The present SAXS measurements confirm significant differences between
the native N and the acid-denatured
U1 states. If the radii of gyration
are very different (see Table 1), the clear peak detected in the Kratky
plot appears larger than that observed in the native state and less
steep at higher Q (compare in Fig. 2). The presence of a
peak in the Kratky plot is a clear indication that the protein state is
sufficiently globular, because the curve for an expanded unfolded
conformation is expected to show a plateau and then to increase
monotonically (Kataoka et al., 1993, 1995; Semisotnov et al., 1996).
However, this is strictly valid only for an infinitely thin, random
coil polymer, and the Kratky plot of a random coil polymer with a
finite section could be undistinguishable from that of a protein in a
globular state. To derive the conformational state of the
acid-denatured form of cyt-c, the SAXS curve has been then tentatively
fitted considering different models. As illustrated in Fig.
6, poor fitting results (in particular at high Q values) were obtained using Eq. 8, also considering a
finite section of the random-coil chain (Fig. 6 c) or
assuming the presence in solution of a mixture of cyt-c in the native
conformation and in a fully unfolded state, as described by the Debye
model (Fig. 6 d). Bad results were also obtained by
analyzing the SAXS experimental curve with the multipole expansion
method, which is based on the presence of compact and globular-like
particles (Spinozzi et al., 1998). On the other hand, good fit to the
data was obtained using the scattering model for semiflexible worm-like
polymers (see Eqs. 9 and 10) (Figs. 2 and 6 a). The
parameters that describe the chain are the radius of the finite
cylindrical cross-section, R, the contour length,
L, and the statistical segment length, b, which
is a measure of the flexibility of the chain. The best-fit parameters
are reported in Table 2, whereas the corresponding p(r) function is shown in Fig. 4.
View larger version (21K):
[in this window]
[in a new window]
|
FIGURE 6
Comparison between different fitting models applied to
the SAXS intensities from cyt-c in the acid-denatured
(U1) and acid-induced recompacted
(A1) forms. U1
state: the curves correspond to the fits obtained using the worm-like
model with a finite section (a), the Gaussian chain
model with a finite section (b), the two cylinders
flexibly jointed end-to-end model (c), a mixture between
the native cyt-c form factor, and a Gaussian chain form factor
(d). A1 state (scaled by a
factor 102 for clarity): fits obtained using the multipole
expansion method (a') and a mixture between the native
cyt-c form factor and the worm-like model form factor
(b').
|
|
Therefore, to derive the structural model, five relevant points need to
be resumed: 1) the aggregation number n (Table 1), derived
assuming a monodisperse solution, indicates that the protein is
monomeric; 2) the Kratky plot is uncompatible with an expanded fully
unfolded conformation of the protein (random-coil conformation), but it
is also proving the absence of globular-like particles; 3) the
worm-like chain model fits well the experimental data; 4) the
L/b ratio (which corresponds to the number of
chain element in the whole chain) is around 2 (see Table 2). It should
be noticed that in polypeptides, L/b can be
regarded as the number of structural subunits or segments of the
polymer that appears to behave randomly, and b is the
effective length of these random segments (Van Holde et al., 1998). The
fitted b value is 55 Å (see Table 2); 5) the distance
distribution function (see Fig. 4) shows a bimodal nature and indicates
that the largest dimension of the scattering particle is ~90 Å.
Therefore, we suggest a two-domain structure for the acid-denatured
U1 state of cyt-c; the two subunits,
which should be characterized by a some degree of order, are 55 Å long. As a confirmation, simulations based on a very simple bimodal
structural model made by two compact subunits flexibly jointed
end-to-end and with randomly distributed bond angles have been
performed. A very good fit to the scattering data was obtained using
cylindrical subunits of 55 Å height and 11.5 Å radius, in good
agreement with the worm-like model parameters (Fig. 6 b).
The scattering curve obtained from the two-cylinder model is reported
in Fig. 2, whereas the particle model is sketched in Fig. 5.
Because of the detected partial compactness of the subunits, it
could be suggested that during unfolding two protein domains, which
retain some degree of the secondary structure, move apart, rotating
around regions of unfolded polypeptide chain. This process is
consistent with the loss and/or the change of the heme axial ligand
detected by XAS (Boffi et al., 2001). Because the two larger N- and
C-terminal -helices of cyt-c are stable in acidic conditions (Jeng
et al., 1990), and because their dimensions compare with the height of
the cylindrical subunits, each protein domain may be essentially
constituted by one of the -helices. The filling procedure gave rise
to the picture sketched in the Fig. 5: the first cylinder is filled by
the amino acids forming the larger N-terminal -helix, the successive
short -helix, and the heme group bound to the His-18; the other
cylinder is in turn filled by the amino acids forming the C-terminal
-helix and the pair of remaining short -helices (where the Met-80
is located). Nevertheless, it should be observed that the geometrical
volume of the bimodal structure is two times larger that the nominal
protein volume, suggesting that the amount of secondary structure is
much lower than that indicated in the model picture and that the
exposure of unfolded polypeptide regions is accomplished through
favorable water interactions. Therefore, experimental data are
consistent with the indication that acid denaturation of proteins
results in states that are less unfolded than those obtained with high concentrations of urea or guanidinium chloride (Tanford, 1970; Fink,
1995).
Acid-induced recompacted A1 state
Starting from the U1 state, the
addition of NaCl or HCl realizes the collapse into compacted
intermediate states, indicating that the chloride anion should play a
key role in these acid- or salt-induced transition to molten globule
conformations. Acid-salt-induced transitions are a common phenomenon to
several quite different proteins. Therefore, the general idea is that
in the acidic denaturation, intramolecular charge repulsion is the
driving force for unfolding. The shielding of intramolecular
charge-charge repulsion forces in the
U1 state by anions brings to the
formation of a molten globule state (Kataoka et al., 1993; Fink, 1995).
However, in the present case, the recompacted states obtained after
addition of NaCl and HCl show very different radii of gyrations and
forward scattering intensities (see Table 1), suggesting a different structure or at least a different aggregation behavior, with the formation of dimeric particles in the case of the acid-induced recompacted state (A1). Noticeable is
that in both cases the Kratky plots (Fig. 2) indicate that the degree
of compactness and globularity of the scattering particles is quite high.
The structural analysis of the acid-recompacted
A1 state has been performed both
fitting the experimental curve with different protein models and
reconstructing the shape of the scattering particles using the
multipole approach. No satisfactory fits to the data have been obtained
neither considering the worm-like model nor using globular and dumbbell
dimeric models based on the crystallographic coordinates of cyt-c (also
taking into account the hydration shell). Moreover, any attempt to fit
the scattering curve using a combination of scattering functions
related to these protein models has been unsatisfactory (Fig. 6
b'). Structural information has been then directly derived
from the particle shape function reconstructed using the multipole
approach. The best fit curve is shown in Figs. 2 and 6 a'.
The particle shape and the corresponding p(r)
function, which are reported in Figs. 5 and 4, respectively, are
consistent with a dimeric compact structure. Considering that the
volume of the reconstructed particle is more than two times larger that the volume of the native cyt-c (see Table 2), the formation of dimers
appears clearly confirmed. From XAS measurements, it was derived that
in the A1 state the pentacoordinate
heme with an imidazole as fifth ligand represents the best model to fit
the local environment of the heme iron. Moreover, recent kinetic
studies of the refolding of cyt-c (Pollack et al., 1999) have
demonstrated that the molten globule-like intermediate can be trapped
because of the presence of non-native His coordination (His-26 or
His-33). The axial ligand, although non-native, increases the stability of compact conformation. These results indicate that the native ligands
are not necessarily required for the formation of compact states.
However, the presence of non-native bonds can lead to the exposure of
charged or hydrophobic surfaces that may determine association
processes increasing ionic strength. The comparison of the particle
shape and dimensions with data relative to the U1 state suggests a model in which
aggregate formation occurs by the specific association of partially
folded domains of cyt-c, which associate preferentially in an
intermolecular fashion to form dimers, as opposed to intramolecular
association leading to the native conformation. One of the possible
dimer conformations, obtained by the Monte Carlo filling procedure
under the assumption that the cyt-c have secondary but no tertiary
structure, is represented in Fig. 5. The quality of the shape-function
filling is a good support to classify the
A1 state as partially folded
intermediate, the general characteristics of which include substantial
secondary structure, little tertiary structure, substantial
compactness, and propensity to aggregation (Fink, 1995).
Salt-induced recompacted A2 state
From the unfolded state U1,
recompaction of cyt-c can be also induced by adding salt. The resulting
state A2 has a 5% larger radius of
gyration than that of the native state (see Table 1), a difference
similar to that already observed for other molten globul protein states
(Kataoka et al., 1995). Moreover, the Kratky plot reported in Fig. 2
shows that the degree of compactness and packing density is very high,
clearly indicating that the A2 specie is as compact as the native state. Fitting procedure based on the
crystallographic coordinates of the cyt-c showed that SAXS data can be
very well reproduced by only adjusting the thickness of the hydration
shell (see Fig. 2 and Table 2). According to the XAS experimental
evidence that in this condition the heme local structure is very
similar to that of the native state (Boffi et al., 2001), and to the CD
analysis that indicate at pH 2 and in the presence of chlorine anions a
significant presence of native-like secondary structure (Goto et al.,
1990a and Fink et al., 1994), we suggest that the recompaction process,
through favorable intramolecular interactions, realizes a globular and
compact conformation. Noticeable is that the hydration shell obtained
from the fitting procedure has a of 2.3 Å, larger than that
observed in the native state ( = 1.3 Å). This difference
confirms that the conformation of the salt-recompacted state is more
expanded than that of the native state, in agreement with the
hypothesis that molten globules have a fluctuating tertiary structure.
It is interesting to observe that these characteristics fit the
original definition by Ptitsyn for the molten globule state
(Ptitsyn, 1987; Fink, 1995).
The whole results indicate that both the heme local and cyt-c global
structures of the two acid- and salt-induced molten globule A1 and
A2 states are completely different.
Specific or nonspecific interactions both contribute to the
stabilization the compact intermediate states. Because the anions
concentration is the same, full protonation of protein residues or
variations in the properties of the electrostatic shielding due to the
nature of the added co-ions (Na+ compared by
H+) appear to critically affect the final structure.
Alkaline-denatured U2 state
The alkaline-denatured state of cyt-c,
U2, and the native N state
have similar Rg values. Moreover, the
peak observed in the Kratky plot reported in Fig. 2 indicates a
globular structure. Accordingly, a good fit to the data has been
obtained, calculating the scattering intensity from the
crystallographic structure of cyt-c only adjusting the thickness of the
border shell (see Table 2): it is interesting to note that the width of
the border shell results comparable with that obtained form the native
data. Finally, the Fig. 4 shows the p(r)
function, which confirms the globular and compact nature of the
denatured state. All these results suggest that the cyt-c in the
alkaline-denatured state has a compact conformation, although the small
changes observed in both radius of gyration (~2%) and (~15%)
could indicate that the mobility of the chains on the protein surface
is increased with respect to the native structure.
Such results are in agreement with the small increase observed in the
hydrodynamic radius by dielectric measurements (Scott and Mauk, 1996;
Rossel et al., 1998) and the XAS evidence that the alkaline unfolded
U2 state is characterized by a
six-coordinate, low-spin heme iron, which maintains the native
imidazole ligand of His-18, whereas the sulfur atom of Met-80 is
replaced by another strong-field ligand (Boffi et al., 2001).
Therefore, the effect caused by basic pH consists in a partially
unfolded process, which is mainly given by the displacement of the
methionine S-iron linkage and by the occurrence of a stable structure
packed as tightly as in the native state.
| |
CONCLUSIONS |
In solution, SAXS is provided to be a powerful technique for the
characterization of the unfolded states of cyt-c. In particular, the
full analysis of the scattering curves gives the possibility to
characterize compactness, globularity, and aggregation properties of
cyt-c in very different experimental conditions.
The first important implication is that the molten globule states
A1 (which forms at pH 2 from the acid
unfolded state for addition of more acid) and
A2 (which forms at the same pH after addition of salt) show fairly compact structures. In good agreement with previous CD measurements (Goto et al., 1990a; Fink et al., 1990),
these structures are compatible with the presence of large amounts of
secondary structure. However, in agreement with the changes in heme
binding detected by XAS measurements (Boffi et al., 2001), these two
states are characterized by significant differences in globularity and
aggregation properties. If a fluctuating tertiary structure
could be responsible for the expanded conformation observed in the
salt-recompacted A2 state,
intermolecular specific interactions, maybe involving hydrophobic or
charged surfaces but sensitive to the effect of added co-ions, could
stabilize the dimeric and substantially compact structure of the
acid-induced A1 state.
Concerning the structural property of the unfolded states, the
alkaline-denatured U2 form results
surprisingly globular, whereas we suggest for the acid-unfolded
U1 form a two-domain expanded structure. Therefore, both states appear characterized by some degree
of structure, in agreement with the indication that acid-denatured proteins posses a residual secondary structure (Tanford, 1970; Fink,
1995).
Much remains to be investigated, in particular for what concerns the
fine balance of forces that determine the degree of folding and the
stability of the different intermediates and the experimental evidence
that the recompacted states are similar or identical to intermediates
formed during the actual folding process. However, a final point should
be stressed: the combined analysis of the local heme environment
(obtained by XAS) and the global structure derived by a full analysis
of the SAXS curves, allows us to extract the model structure for cyt-c
in very different unfolding intermediates. This can be a good general
way to derive systematic structural information on folding and
unfolding processes for a number of proteins.
Address reprint requests to Dr. Francesco Spinozzi, Istituto di Scienze
Fisiche, Facoltà di Medicina, Università di Ancona, Via
Ranieri 65, I-60131 Ancona, Italy. Tel.: 39-071-2204608; Fax:
39-071-2204605; E-mail: f.spinozzi{at}alisf1.unian.it.
TOP
ABSTRACT
INTRODUCTION
![P<SUB>1</SUB>(Q)={2[<UP>exp</UP>(<UP>−</UP>R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>Q<SUP>2</SUP>)+R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>Q<SUP>2</SUP>−1]/(R<SUP><UP>2</UP></SUP><SUB><UP>g</UP></SUB>Q<SUP>2</SUP>)<SUP>2</SUP>](/content/vol81/issue6/fulltext/3522/img011.gif)
![×(1−<UP>exp</UP>[<UP>−</UP>((Qb)/q<SUB>1</SUB>)<SUP><UP>p<SUB>1</SUB></UP></SUP>])](/content/vol81/issue6/fulltext/3522/img014.gif)
![×(1−<UP>exp</UP>[<UP>−</UP>((Qb)/q<SUB>2</SUB>)<SUP><UP>p<SUB>2</SUB></UP></SUP>])](/content/vol81/issue6/fulltext/3522/img016.gif)
