Báo cáo Y học: A comparison of the urea-induced unfolding of apo¯avodoxin and ¯avodoxin from Desulfovibrio vulgaris potx

Eur. J. Biochem. 269, 212±223 (2002) Ó FEBS 2002

A comparison of the urea-induced unfolding of apo¯avodoxin
and ¯avodoxin from Desulfovibrio vulgaris
Â
Brian O Nuallain* and Stephen G. Mayhew
Department of Biochemistry, University College Dublin, Bel®eld, Dublin, Ireland

1 The kinetics and thermodynamics of the urea-induced
unfolding of ¯avodoxin and apo¯avodoxin from Desulfovibrio vulgaris were investigated by measuring changes in
¯avin and protein ¯uorescence. The reaction of urea with
¯avodoxin is up to 5000 times slower than the reaction with
the apoprotein (0.67 s)1 in 3 M urea in 25 mM sodium
phosphate at 25 °C), and it results in the dissociation of
FMN. The rate of unfolding of apo¯avodoxin depends on
the urea concentration, while the reaction with the holoprotein is independent of urea. The rates decrease in high salt
with the greater e€ect occurring with apoprotein. The ¯uorescence changes ®t two-state models for unfolding, but they
do not exclude the possibility of intermediates. Calculation
suggests that 21% and 30% of the amino-acid side chains
become exposed to solvent during unfolding of ¯avodoxin
and apo¯avodoxin, respectively. The equilibrium unfolding

curves move to greater concentrations of urea with increase
of ionic strength. This e€ect is larger with phosphate than
with chloride, and with apo¯avodoxin than with ¯avodoxin.
In low salt the conformational stability of the holoprotein is
greater than that of apo¯avodoxin, but in high salt the relative stabilities are reversed. It is calculated that two ions are
released during unfolding of the apoprotein. It is concluded
that the urea-dependent unfolding of ¯avodoxin from
D. vulgaris occurs because apoprotein in equilibrium with
FMN and holoprotein unfolds and shifts the equilibrium so

that ¯avodoxin dissociates. Small changes in ¯avin ¯uorescence occur at low concentrations of urea and these may
re¯ect binding of urea to the holoprotein.

Flavodoxins are small ¯avoproteins found in microorganisms and eukaryotic algae where they function as electron
carriers in oxidation±reduction reactions [1,2]. They consist
of a ®ve-stranded parallel b sheet of protein with a helices on
each side of the sheet, and a molecule of FMN bound
strongly but noncovalently between two loops on one side
of the molecule. The two methyl groups on the ¯avin are
exposed to solvent, the dimethylisoalloxazine moiety is
¯anked by hydrophobic residues, and the ribityl phosphate
side chain extends towards the centre of the protein (Fig. 1).
The ¯avodoxins are believed to function as 1-electron
carriers that operate between the semiquinone and hydroquinone forms of the ¯avin, and therefore the semiquinone
is probably the resting state of the folded protein in the cell.
The ¯avodoxins occur as short-chain proteins (137±148
residues) such as those from Desulfovibrio vulgaris, Desulfovibrio desulfuricans and Clostridium beijerinckii, and a
second group that has about 20 amino acids inserted in
one strand of the b sheet, and that includes proteins from
Azotobacter vinelandii and Anabaena PCC 7119. The
¯avodoxin fold is shared by a range of unrelated proteins
(nine superfamilies) with different functions [3].

The FMN of ¯avodoxins can be reversibly removed with
acid. The resulting apoproteins are stable, and it has been
proposed that they are useful models to investigate the
folding/unfolding reactions of a/b proteins [4±8]. However,
their ability to bind ¯avin is a property that has yet to be
explored in the context of protein folding, and it is likely that
they will also prove to be useful as models for folding of a/b

proteins that require a tightly bound organic cofactor for
activity. An early study showed that guanidine HCl dissociates ¯avodoxin from Clostridium pasteurianum into
apo¯avodoxin and FMN [9]. More recently, this denaturant
has been used to study unfolding of the apoproteins of
¯avodoxins from A. vinelandii [6±8] and D. desulfuricans
[10], and similar studies have been carried out with
apo¯avodoxin from Anabaena PCC 7119 but using urea
as the denaturant [4,5]. In the ®rst two cases, evidence was
obtained for a stable intermediate in the equilibrium
between the folded and unfolded states. In contrast, urea
causes apo¯avodoxin from Anabaena to unfold directly
without stabilizing an intermediate. The study with apo¯avodoxin from D. desulfuricans was the only one to
investigate the effect of bound FMN on the unfolding
equilibrium. It was concluded that FMN has no effect on
the stability of the protein, and that FMN remains tightly
bound to the unfolded protein [10].
The present paper compares the unfolding/folding reactions of ¯avodoxin from D. vulgaris by urea with the
corresponding reactions of its apoprotein. This ¯avodoxin
contains two residues of tryptophan and ®ve residues of
tyrosine. A tyrosine side chain (Y98) is almost coplanar with
the face of the dimethylisoalloxazine moiety that is closer to
solvent, while a tryptophan side chain (W60) is inclined at

Correspondence to S. G. Mayhew, Department of Biochemistry,
University College Dublin, Bel®eld, Dublin 4, Ireland.
Fax: + 353 1 2837211, Tel.: + 353 1 7061572,
E-mail:
*Present address: University of Tennessee Medical Centre,
1924 Alcoa Highway, Knoxville, TN 37920, USA.
(Received 10 July 2001, revised 18 October 2001, accepted 29 October

2001)

Keywords: apo¯avodoxin; urea unfolding; Desulfovibrio
vulgaris.


Ó FEBS 2002

Flavodoxin unfolding (Eur. J. Biochem. 269) 213

Fig. 1. Structure of ¯avodoxin from
D. vulgaris showing the relative positions of
the FMN, tryptophan side chains and tyrosine
side chains. The FMN is shown in yellow,
tryptophan side chains in magenta, and
tyrosine side chains in green. The ®gure was
produced with RASMOL.

an angle to the opposite face (Fig. 1). The ¯uorescence
emission by the side chains of the aromatic amino acids is
partly quenched when the apoprotein binds FMN, and the
intense greenish-yellow ¯uorescence of free ¯avin is about
99% quenched in the complex. Measurement of the
¯uorescence emissions due to these components can therefore be used to monitor changes in the ¯avin±protein
complex. Changes in the ¯uorescence of the apoprotein can
also provide information about the environment of the
aromatic amino acids. The kinetics of unfolding and
refolding in urea and the unfolding equilibrium have been
investigated. The experimental observations have been used
to determine the conformational stabilities of the holoprotein and apoprotein. A mechanism is proposed for

the unfolding/folding reactions of the two forms of the
protein in urea. This ¯avodoxin resembles ¯avodoxin from
D. desulfuricans both in amino-acid sequence (148 amino
acids) and in three-dimensional structure [11,12]. However,
we ®nd that the effects of denaturant on the ¯avodoxin and
apo¯avodoxin from D. vulgaris are surprisingly different
from those reported for the protein from D. desulfuricans
[10].

MATERIALS AND METHODS
Preparation and estimation of ¯avodoxin
and apo¯avodoxin
Flavodoxin from D. vulgaris was obtained as the recombinant protein that was puri®ed from extracts of E. coli [13].
Apo¯avodoxin was prepared by acid precipitation [14]. The
protein precipitate was dissolved in 25 mM sodium phosphate and 0.3 mM EDTA, pH 7.0 (buffer A), and dialysed
against the same buffer. The concentrations of holoprotein
and apoprotein were determined using absorption coef®-

cients at 458 nm (10 700 M)1ácm)1) and at 280 nm
(22 400 M)1ácm)1), respectively [13]. FMN (HPLC puri®ed;
a gift from A. F. Buckmann, Gesellschaft fur BiotechÈ
nologische Forschung Manscheroder, Weg, Germany)
was determined using an absorption coef®cient of
12 500 M)1ácm)1 at 445 nm [15]. Fluorescence titrations in
which apo¯avodoxin was added to a known concentration
of FMN [13] showed that the concentration of apoprotein
that bound FMN was at least 95% of the protein
determined from the UV absorbance and that therefore
the apoprotein was essentially fully active.
Reactions with urea

The effects of urea on ¯avodoxin and its apoprotein were
monitored by ¯uorimetry. The protein ¯uorescence of
¯avodoxin and its apoprotein was measured with excitation
at 280 nm, and emission at the wavelength that gave the
greatest difference between folded and unfolded protein
(380 nm and 315 nm for ¯avodoxin and apo¯avodoxin,
respectively; Fig. 2). The reaction of urea with ¯avodoxin
was also followed by monitoring changes in ¯avin ¯uorescence at 525 nm with excitation at 445 nm (see below).
Determination of unfolding rate constants
Studies on the rate of protein unfolding were carried out by
diluting a stock solution of ¯avodoxin or apo¯avodoxin in
buffer A at least 100-fold to obtain 1 lM protein in urea
(0.1±10 M). The relatively rapid unfolding of apo¯avodoxin
in urea and 25 mM sodium phosphate buffer pH 7 (buffer B)
was measured by using a stopped-¯ow spectro¯uorimeter
that consisted of a Rapid Kinetics Spectrometer Accessory
(Applied Photophysics Ltd; RX-1000) interfaced to the
optical system of a Baird Nova ¯uorimeter, a home-made


Â
214 B. O Nuallain and S. G. Mayhew (Eur. J. Biochem. 269)

was followed by a further very slow change in the protein
and ¯avin ¯uorescence from the holoprotein, and of the
protein ¯uorescence in experiments with the apoprotein, at
rates that were insigni®cant compared with the initial
reaction. The rates of the slow reactions were not affected by
the concentration of urea (0.6±5 M) or by changing the
phosphate buffer concentration in the range 25±250 mM,

indicating that these reactions are not associated with ureadependent unfolding. Therefore, the reactions were disregarded and the ¯uorescence at the end of the relatively rapid
phase of ¯uorescence change was taken as a measure of the
equilibrium between folded and unfolded protein.
The positions of the unfolded/folded equilibria of ¯avodoxin and apo¯avodoxin at different concentrations of urea
were also measured in refolding experiments. The protein
(50 lM) was unfolded in buffer B and 6 M urea as described
above. It was then diluted 50-fold into buffer B and urea
(0±6 M). Refolding of the diluted protein was monitored
from the changes in ¯avin and/or protein ¯uorescence.

5

Fluorescence (arbitrary units)

1

4

3
3

2

1

0
300

4

2

340

Ó FEBS 2002

380

420

Wavelength (nm)
Fig. 2. Fluorescence emission spectra of folded and unfolded ¯avodoxin
and apo¯avodoxin. (1) Folded apo¯avodoxin; (2) folded ¯avodoxin; (3)
unfolded ¯avodoxin; (4) unfolded apo¯avodoxin. The solutions contained at 25 °C: 1 lM protein; 25 mM sodium phosphate, pH 7.0; and
for (3) and (4) 6 M urea. The spectra (3) and (4) were recorded after all
¯uorescence changes were complete. Fluorescence excitation was at
280 nm.

signal ampli®er, an oscilloscope (Hameg Instruments 203-7)
and a digital storage adaptor (Thurlby±Thandor DSA524).
Progress curves for the unfolding of ¯avodoxin under all
conditions, and of apo¯avodoxin at high salt concentration,
were obtained by static ¯uorimetry. The reaction with
apo¯avodoxin was monitored continuously until the reaction was complete. In contrast, the much slower unfolding
reaction of ¯avodoxin was monitored at intervals; to
minimize photobleaching of FMN, the reaction mixtures
were stored in darkness between ¯uorescence measurements. Values for the ®rst-order rate constants for the
reactions were obtained by averaging for three measurements the slopes of plots of the logarithm of the change in
¯uorescence vs. time.
Equilibrium unfolding/refolding experiments

The unfolding/folding equilibrium of ¯avodoxin and apo¯avodoxin was determined by following the changes in
FMN and/or protein ¯uorescence. Readings were taken at
intervals until equilibrium had been reached. Assays were
carried out in triplicate, with each mixture containing in
1 mL at 25 °C: 0.25±23 lM protein; up to 7.1 M urea;
sodium phosphate buffer pH 7.0; and NaCl as indicated.
The mixtures were incubated for an appropriate period
(7±48 h for ¯avodoxin, depending on the phosphate concentration, and 20 min for incubations with apo¯avodoxin).
As is described in the Results, when the urea concentration
was suf®cient to cause only partial unfolding of ¯avodoxin
and apo¯avodoxin, a relatively rapid initial change occurred
in the ¯uorescence until equilibrium had been reached. This

Analysis of the equilibrium between folded
and urea-unfolded protein
The urea-unfolding curves for ¯avodoxin and its apoprotein
were analysed according to a two-state model that proposes
that the protein unfolds in a single step. The equilibrium can
be ®tted to an expression of the type [16,17]:
DGD ˆ DGW À mD

…1†

where, DGD is the change in free energy between the folded
and urea-unfolded states in the denaturant; D is the
concentration of denaturant; DGw is the change in free
energy between the folded and unfolded states in the
absence of the denaturant (the Ôconformational stabilityÕ of
the protein), and, m, is an empirically derived parameter, the
change in free energy between the folded and unfolded states

per molar concentration of denaturant. The parameter
m re¯ects the cooperativity of the two-state transition.
Cooperativity is used to describe the sensitivity of the
transition to denaturant concentration; it is not used to
mean the degree to which the transition approximates a
two-state transition. DGD can be derived directly from
experimental data (Eqn 2):
 
U
U
…SF À S†
…2†
and ˆ
DGD ˆ ÀRT ln
F
F
…S À SU †
where S is the observed ¯uorescence signal; SF and SU are
the ¯uorescence signals for the folded and unfolded protein,
respectively, and F and U are the proportions of the folded
and unfolded states; R is the gas constant; and T is the
temperature in K. The urea-unfolding curves for ¯avodoxin
and apo¯avodoxin were analysed using an equation derived
by Santoro & Bolen (Eqn 3) [18] that incorporates Eqns (1)
and (2).
S ˆ

SF ‡ SU eÀ…DGW ÀmD†aRT
1 ‡ eÀ…DGW ÀmD†aRT

…3†

Eqn (3) lacks parameters for the slopes of the baselines of SF
and SU, which are present in the equation derived by Santoro
& Bolen [18]. This is because SF and SU for ¯avodoxin and
apo¯avodoxin are essentially independent of the urea concentration (see below). The ®tting of an unfolding curve to


Ó FEBS 2002

Flavodoxin unfolding (Eur. J. Biochem. 269) 215

Eqn (3) gives empirically derived m and DGW values without
having to determine the value for DGD at each concentration
of urea as is required when calculating a value of DGW using
Eqn (1). In addition, the concentration of urea that gives halfunfolded protein (urea1/2) is obtained from the ®tted curve.
The urea-unfolding curves were also analysed by a
method [19] that is derived using experimentally measured
values for the change in solvation free energy when the side
chains of amino acids are transferred from water to
guanidine HCl and to urea (Eqn 4) [20±23].
DGD ˆ

DGW ‡ nDGsYm D
Kden ‡ D

…4†

where, n is the approximate number of amino-acid side
chains that become exposed on unfolding of the protein;

DGs,m is an empirically derived constant that represents an
average value for the free energy change for the solvation of
a buried amino-acid side chain that occurs on unfolding of
the protein when the concentration of denaturant is in®nite;
Kden is an empirical constant that represents the concentration of denaturant at which half DGs,m is achieved. The
values used for DGs,m (5.024 kJ mol)1) and Kden (25.25 M)
in urea were obtained from [19]. These values represent the
behaviour of an ÔaverageÕ protein, determined from the
solvent-excluded side chains of 55 proteins in the Protein
Data Bank, and using solvation energies of model compounds in guanidine HCl and in urea [20±23].
Assuming that salt ions bind preferentially to the folded
state, the number of salt ions (NaCl) that are released from
apo¯avodoxin when the protein unfolds can be obtained by
®tting the unfolding curves to Eqn (5) [24].
D…ln Kapp †
ˆ DL ˆ LU À LF
D…aL †

…5†

where Kapp is the unfolding/folding equilibrium constant, aL
is the mean activity of the salt, and LU and LF are the
number of salt ions bound by the unfolded and folded states
of the apoprotein, respectively.
Determination of kinetic and thermodynamic constants
for the binding of FMN to apo¯avodoxin
Rate and equilibrium constants were determined by
following the increase in ¯uorescence due to FMN release
when ¯avodoxin was diluted. Flavodoxin was diluted at
least 150-fold to obtain 1 lM protein in buffer B and a

concentration of urea (0±1 M) lower than that required to
unfold the holoprotein. Equilibrium was reached within
10 min. A value for the dissociation rate constant (ko€) was
determined from the progress curve. The dissociation of the
holoprotein can be described by Eqns (6) and (7).
koff

ÀÀ
FMN-apoprotein ÀÀ FMN ‡ apoprotein
AB
kon

Kd ˆ

koff
‰FMNŠ‰apoproteinŠ
x2
e
ˆ
ˆ
kon
a À xe
‰FMN À apoproteinŠ

…6†

…7†

where Kd, is the dissociation constant; [apoprotein] and
[FMN-apoprotein], are the concentrations of free apo¯avodoxin and the holoprotein, respectively; a, is the initial

concentration of the holoprotein and it represents 100%

relative ¯uorescence if complete dissociation occurs; and xe
is the concentration of FMN and apoprotein at equilibrium.
Solution of Eqn (7) between xi (initial), x and ti, t, yields the
integrated rate law for a ®rst-order, second-order equilibrium reaction (Eqn 8) [25].
koff t ˆ

xe
x…a À xe † ‡ xe a
ln
2a À xe
a…xe À x†

…8†

The values for xe, x, a, and t were obtained from each
progress curve. A plot of the right hand side of Eqn (8) vs.
time gives a straight line whose slope is ko€. Values for ko€
were determined from the average slope of the plots for
three measurements.
Values for the dissociation constant for the holoprotein
(Kd) were calculated from the end point of the progress
curves using Eqn (8). It was assumed that the concentrations
of apoprotein and free FMN in the equilibrium were the
same. As the experiments were carried out in a low
concentration of urea, it was necessary to correct for a
small proportion of unfolded apoprotein. This was calculated from the appropriate unfolding curves ®tted to Eqn
(3). Values for kon were then calculated by substituting the
values calculated for ko€ and Kd into Eqn (7).

Experimental data were ®tted to functions using the
computer program MAC CURVEFIT (version 1.3.5).

RESULTS
Unfolding/refolding reactions of apo¯avodoxin
Treatment with urea causes the ¯uorescence emission
maximum for apo¯avodoxin to decrease in intensity and
to shift from 336 nm to 351 nm (Fig. 2). The red shift is
consistent with the transfer of the aromatic residues to a
more polar environment [26] and it re¯ects unfolding of the
protein. The ¯uorescence changes occur rapidly, and at
concentrations of urea that are great enough to cause
complete unfolding, the progress curve follows a single
exponential (Fig. 3). This suggests that a two-state transition occurs in the conversion of the folded protein to the
unfolded state. The reactions at concentrations of urea that
are too low to cause complete unfolding were also found to
follow ®rst-order kinetics when they were analysed according to a model for reactions that approach an equilibrium
(data not shown) [25]. The rate constant calculated for the
reaction of apo¯avodoxin with 3 M urea in buffer B is
0.67 ‹ 0.032 s)1 (Table 1). The rate depends on the ionic
composition of the solution. These salt effects were not
examined in detail. However, it was observed that when the
phosphate concentration is increased from 25 mM to
250 mM, the rate constant decreases % 60-fold. Furthermore, when chloride ion was used to raise the ionic strength
to the same value as that of 250 mM phosphate, the decrease
in the rate constant for apo¯avodoxin was somewhat less,
indicating that the rate depends in addition on the nature of
the salt (Table 1). The rate constant increases exponentially
with increasing urea. Tanford [27] proposed that the rate
constant for unfolding of a protein in urea, ku, is related to

the rate constant in the absence of urea, kw, and to the urea
concentration by Eqn (9).
ln ku ˆ lnkw ‡ mu ‰ureaŠ

…9†


Â
216 B. O Nuallain and S. G. Mayhew (Eur. J. Biochem. 269)
0
ln ∆fluorescence

∆fluorescence (arbitrary units)

1

0.5

-1
-2
-3
-4
0

0

0

100

200

100

200 300
Time (s)

300

Time (s)
Fig. 3. The kinetics of unfolding of apo¯avodoxin by urea. The reactions contained at 25 °C: 1 lM apo¯avodoxin; sodium phosphate,
pH 7.0; and urea. The protein ¯uorescence was measured at 315 nm
with excitation at 280 nm. d, 25 mM phosphate and 3 M urea; h,
25 mM phosphate, 500 mM NaCl and 6 M urea; m, 250 mM phosphate
and 6 M urea. The inset shows the corresponding logarithmic plots.

where mu is proportional to the increase in exposure of the
protein to solvent on going from the folded to a transition
state. When this equation is used to analyse the unfolding of
apo¯avodoxin in 250 mM phosphate, the value determined
for kw (0.0066 s)1) is only about one third of ku determined
in strongly denaturing conditions (ku ˆ 0.021 ‹ 0.001 s)1
in 10 M urea; Fig. 4).
The refolding of urea-unfolded apo¯avodoxin occurs
rapidly. It was complete within 2 s of diluting the unfolded
protein 50-fold, but attempts to follow the reaction using
stopped-¯ow ¯uorescence measurements were unsuccessful
mainly because at this dilution, it was not possible to obtain
ef®cient mixing with the instrument available. The rate of
change of the protein ¯uorescence is decreased when FMN

is present in the diluting buffer so that the changes can be
followed in a conventional ¯uorimeter. Experiments in
which the ®nal 20±30% of the progress curve was monitored
suggest that the reaction in the presence of equimolar FMN
follows second-order kinetics (16.4 ‹ 2.1 ´ 105 M)1ás)1).
The rate constant was found to be similar to that observed

Ó FEBS 2002

with apo¯avodoxin that had not been through the unfolding
procedure (14.1 ‹ 3.1 ´ 105 M)1ás)1).
Plots of the extent of ¯uorescence change at equilibrium
vs. the concentration of urea also suggest that in the case of
this apo¯avodoxin a simple two-state transition occurs
between the folded and unfolded protein (Fig. 5). Dilution
of the completely unfolded apoprotein with urea of different
concentrations showed that the ¯uorescence at equilibrium
mirrored that observed during unfolding with urea, and that
the unfolding is reversible (Fig. 5). The conformational
stability in buffer B, determined by ®tting the ureaunfolding curve for 1 lM apo¯avodoxin, is 9.99 ‹ 0.4
kJ mol)1. A similar experiment was carried out using
50 mM Mops, pH 7, as the buffer to allow comparison with
published data for Anabaena apo¯avodoxin [4]. The value
obtained for the D. vulgaris protein in Mops (DGw ˆ
11.71 ‹ 1.43 kJámol)1) is similar to that in phosphate, but
smaller than the value for Anabaena apo¯avodoxin
(17.1 ‹ 0.5 kJámol)1). Analysis of the unfolding curve for
the D. vulgaris apoprotein in buffer B and using Eqn (4)
suggests that 39.4 ‹ 1.2 amino-acid side chains become
exposed to solvent when the protein unfolds.

An increase in the concentration of salt causes the
unfolding curve of apo¯avodoxin to shift to the right
(Fig. 5). The stability of the apoprotein increases about
threefold when the concentration of phosphate buffer is
increased from 25 mM to 250 mM (Table 2). The slope (m)
in the transition region of the unfolding curve shows only
small increases with increasing phosphate, the main effect
being a shift of the transition midpoint. Similar but slightly
smaller increases in the stability of apo¯avodoxin are
observed when the ionic strength is increased with NaCl
(Fig. 5; Table 2). By assuming that the increase in stability
of apo¯avodoxin with salt is due to the preferential binding
of salt to the folded protein, it can be calculated that
approximately two ions are released when the apoprotein
unfolds in urea (Fig. 5, inset). The values of m, DGw and the
concentration of urea to give half-unfolded apoprotein were
found to be independent of the protein concentration
(0.25±23 lM; Table 3).
Equilibrium unfolding curves for apo¯avodoxin were
also obtained using guanidine HCl as denaturant in buffer
B. The midpoint of the unfolding curve occurs at a lower
concentration of denaturant (1.0 M guanidine HCl vs. 1.35 M
with urea) and the slope is steeper (20.7 ‹ 1.4 kJámoláM)1).
The calculated conformational stability in guanidine HCl
(21.15 ‹ 1.45 kJámol)1) is about twice that calculated in
urea under the same conditions. It seems likely that the
denaturing effect of guanidine HCl is modulated by its

Table 1. E€ects of salt on the rates of urea-unfolding of ¯avodoxin and apo¯avodoxin. Values for the ®rst-order rate constants for the unfolding of the
protein (ku) in urea at pH 7.0 and 25 °C were determined as described in Figs 3 and 6. The errors are the standard deviations.

Bu€er
25 mM sodium phosphate
25 mM sodium phosphate
+ 500 mM NaCl
+ 250 mM NaPi

[urea]
(M)

104 ´ ku (s)1)
Flavodoxin

Apo¯avodoxin

3
6

±
1.42 ‹ 0.24

6729 ‹ 321
±

6
6
8

0.16 ‹ 0.01
±

0.21 ‹ 0.01

396 ‹ 15
115 ‹ 5.0
±


Ó FEBS 2002

Flavodoxin unfolding (Eur. J. Biochem. 269) 217

-3.6

u

ln k (s-1)

-4

-4.4

-4.8

0

4

8

12

Urea (M)
Fig. 4. The e€ect of urea concentration on the rate constant of unfolding
of apo¯avodoxin. The logarithm of the observed ®rst-order rate constant (ku) for the unfolding of apo¯avodoxin at 25 °C in 250 mM
sodium phosphate, pH 7.0, containing 5.2±10 M urea, is plotted
against the concentration of urea. The values for ku are the averages of
three kinetic traces. The error bars show the standard deviations.

contribution to the salt concentration and a resultant
increase in stabilization.
Unfolding/refolding reactions of ¯avodoxin
Flavodoxin completely unfolds by a single exponential in a
high concentration of urea, as was found to occur with
apo¯avodoxin. Identical kinetics are observed when the
reaction is followed by measuring the protein or the ¯avin
¯uorescence (Fig. 6). The ®nal ¯avin ¯uorescence is the
same as that of free FMN. Depending on the reaction
conditions the rate constant calculated for urea-unfolding of
the holoprotein is up to 5000 times smaller than that for the
apoprotein (Table 1). In further contrast to the reaction
with the apoprotein, the rate constant is independent of the
urea concentration [ku in 6 M urea ˆ 1.42 ‹ 0.02 ´
10)4 s)1 (3); ku in 10 M urea ˆ 1.48 ‹ 0.05 ´ 10)4 s)1
(3)], and similar to the rate constant determined for
dissociation of the holoprotein in the absence of urea
(ko€ ˆ 1.81 ´ 10)4 s)1). This suggests that the dissociation of

Fig. 5. E€ects of NaCl on the urea-unfolding/folding curve for apo¯avodoxin. Unfolding curves were determined at 25 °C for 1 lM
apo¯avodoxin in 25 mM sodium phosphate bu€er, pH 7, with 0 M (d),
0.1 M (r), 0.3 M (j) or 0.5 M (m) NaCl. Fluorescence excitation and

emission was measured at 280 nm and 315 nm, respectively. The
unfolding curves were ®tted using Eqn (3). A refolding curve was also
determined in the absence of NaCl (half ®lled square). The data points
are average values from three measurements. Inset: plots of the logarithm of the equilibrium constant (Kapp) for the unfolding of apo¯avodoxin calculated at 2.94 M (j) and 3.43 M (d) urea vs. the
logarithm of the mean activity of NaCl (ln a‹).

the holoprotein complex to apo¯avodoxin and FMN is the
rate-determining step during unfolding of this ¯avodoxin.
Salt inhibits the rate of unfolding of ¯avodoxin but the
inhibition is less than the corresponding salt inhibition
observed with the apoprotein (Table 1).
The equilibrium curves for urea-unfolding of the holoprotein suggest that the protein unfolds by a simple twostate transition, as was also concluded for the apoprotein.
The extent of the change of the ¯avin ¯uorescence at a given
concentration of urea is the same as the extent of change for
the protein ¯uorescence (Fig. 7).
It was observed that at concentrations of urea below that
required to establish the folded/unfolded equilibrium

Table 2. E€ects of salt on the energetic parameters for the conformational stabilities of ¯avodoxin and apo¯avodoxin. The parameters were
determined at 25 °C from urea-unfolding curves for 1 lM protein in the bu€er at pH 7.0 as indicated.
Bu€er
Flavodoxin
25 mM NaPi
+ 500 mM NaCl
250 mM NaPi
Apo¯avodoxin
25 mM NaPi
+ 100 mM NaCl
+ 300 mM NaCl

m
(kJámol)1áM)1)

Urea1/2
(M)

DGw
(kJámol)1)

)6.44 ‹ 0.4
)4.04 ‹ 0.5
)6.32 ‹ 0.4

2.66
5.01
4.71

17.59 ‹ 1.0
20.15 ‹ 2.6
29.62 ‹ 2.3

)7.50 ‹ 0.3
)6.97 ‹ 0.3
)7.93 ‹ 0.7

1.35
2.28
2.91

9.99 ‹ 0.4

15.73 ‹ 0.6
22.91 ‹ 2.0


Â
218 B. O Nuallain and S. G. Mayhew (Eur. J. Biochem. 269)

Ó FEBS 2002

Table 3. E€ects of protein concentration on the energetic parameters for the conformational stabilities of ¯avodoxin and apo¯avodoxin. The energetic
parameters were determined by ®tting unfolding curves as in Figs 5 and 7 using Eqn (3). Data were obtained at 25 °C in 25 mM sodium phosphate
and 0±7.1 M urea.
[protein]
7(M)

m
(kJámol)1áM)1)

urea1/2
(M)

DGw
(kJámol)1)

0.25
1.00
2.50
5.00
14.00
23.00

)
)
)
)
)
)

4.79
6.44
7.62
5.95
6.22
6.59

‹
‹
‹
‹
‹
‹

0.4
0.4
0.4
0.3
0.4
0.4

2.44

2.66
2.69
2.91
3.28
3.31

11.68
17.59
21.07
17.24
20.62
21.59

‹
‹
‹
‹
‹
‹

1.1
1.0
1.5
1.0
1.4
1.3

0.25
1.00
2.50

5.00
23.00

Protein

)
)
)
)
)

9.48
7.50
8.93
7.16
8.60

‹
‹
‹
‹
‹

0.6
0.3
0.4
0.4
0.5

1.04

1.35
1.42
1.40
1.40

9.81
9.99
12.59
10.05
12.18

‹
‹
‹
‹
‹

0.7
0.4
0.5
0.6
0.7

Flavodoxin

Apo¯avodoxin

(< 1 M urea in buffer B) the FMN ¯uorescence increases by
up to threefold (equivalent to % 3% of the FMN ¯uorescence of fully dissociated ¯avodoxin), and comparable
changes occur in the protein ¯uorescence. Several explanations for this effect were considered, including the possibility

that the holoprotein partly unfolds to an intermediate in
which FMN is still protein-bound, as proposed recently
for the guanidine HCl unfolding of ¯avodoxin from
D. desulfuricans [10]. However, kinetic analysis of the
increase with ¯avodoxin from D. vulgaris, and measurement of the ¯uorescence end point as a function of the urea

concentration, suggest that the effect results mainly from a
shift to the right of the holoprotein/apoprotein equilibrium
described by Eqn (6). The value determined for the
dissociation rate constant (ko€) is independent of the
concentration of urea (2.03 ‹ 0.53 ´ 10)4 s)1 for 0±1 M
urea; Table 4). The value for the holoprotein/apoprotein
dissociation constant (Kd) (corrected for the proportion of
apoprotein that is unfolded; see Materials and methods)
increases with increasing urea in the same range (Table 4).
As ko€ is unaffected by the urea concentration, this increase
in Kd must re¯ect a decrease in the value for the association

Fig. 6. Progress curves for the unfolding of ¯avodoxin by urea. The
experiments were carried out at 25 °C, pH 7.0, with 1 lM ¯avodoxin.
Protein (open symbols) and ¯avin (closed symbols) ¯uorescence was
measured at 380 nm and 525 nm with excitation at 280 nm and
445 nm, respectively. m,n, 25 mM phosphate and 6 M urea; j,h,
25mM phosphate, 500 mM NaCl and 6 M urea; d, 250 mM phosphate
and 8 M urea. The inset shows the corresponding logarithmic plots.

Fig. 7. E€ects of salt on the urea-unfolding curve of ¯avodoxin.
Unfolding curves were determined at 25 °C, pH 7, with 1 lM protein.
The curves were determined in 25 mM sodium phosphate without
NaCl (d,s) or with 0.5 M NaCl (m), or in 250 mM sodium phosphate

(j); they were ®tted using Eqn (3). The plots show changes with urea
concentration in the observed protein (d,m,j) and ¯avin ¯uorescence
(s). Each data point is an average from three samples.


Ó FEBS 2002

Flavodoxin unfolding (Eur. J. Biochem. 269) 219

Table 4. E€ect of urea on the rate constants and the equilibrium dissociation constant for the complex of FMN and apo¯avodoxin. Experimental data were obtained at 25 °C with a ®nal protein concentration
of 1 lM in 25 mM sodium phosphate pH 7.0 and urea as indicated. ko€
and Kd were determined experimentally; kon was calculated.
[urea]
(M)

10)5 ´ kon
(M)1ás)1)

104 ´ ko€
(s)1)

1010 ´ Kd
(M)

0.0
0.2
0.4
0.6
0.8
1.0

8.32
5.07
4.33
3.92
3.40
2.83

1.81
2.31
1.58
1.79
2.56
2.13

2.18
4.55
3.66
4.56
7.53
7.54

rate constant (kon; Table 4). It is concluded that the increase
in ¯avin ¯uorescence at low urea concentrations results
mainly from a direct effect of urea on the holoprotein that
weakens the FMN±protein interactions and shifts the
equilibrium in Eqn (6) to the right. In addition, apo¯avodoxin in equilibrium Eqn (6) starts to become unfolded at
these low concentrations of urea, and this contributes to
dissociation of ¯avodoxin by removing folded apoprotein
from the equilibrium.

When urea-unfolded ¯avodoxin is diluted to give a lower
concentration of urea the changes in FMN and protein
¯uorescence are reversed, indicating that the unfolding is
again reversible. Both ¯avin and protein ¯uorescence
quenching follow second order kinetics, and the rate
constant for the reaction (16.4 ‹ 1.9 M)1ás)1) is identical
to the value determined when unfolded apoprotein is
refolded in the presence of equimolar FMN, indicating that
the rate-determining step in the refolding pathway of the
holoprotein is the binding of FMN. However, the changes
are smaller than those observed during the unfolding
reaction, showing that refolding is incomplete. For example,
only % 80% of the holoprotein was found to refold in 0.12 M
urea as judged by the extent of ¯uorescence quenching; in
contrast, the addition of holoprotein to 0.12 M urea caused
less than 3% unfolding. The incomplete refolding of ureaunfolded ¯avodoxin is most likely due to inactivation of
apoprotein during the long period required to completely
unfold the holoprotein (7 h), in contrast to the much shorter
period used to unfold apoprotein (20 min). This conclusion
is supported by the observation that when unfolded
holoprotein is exposed to urea for a longer period than is
required to completely unfold the protein even less folded
protein is formed when the urea is subsequently diluted out
(6% recovery of folded holoprotein after 48 h unfolding). A
control experiment showed that when apoprotein is incubated in urea for up to 5 days before diluting out the
denaturant, the extent of refolding also progressively
decreases (data not shown). The urea-unfolding and refolding experiments were carried out in the absence of EDTA, a
chelating agent that is used to protect thiol groups from
heavy metal-catalysed oxidation. The inclusion of 2 mM
EDTA in the incubations did not improve the reversibility of

the reactions after long-term treatment with urea.
The conformational stability determined for the holoprotein of ¯avodoxin in buffer B (17.6 ‹ 1.0 kJámol)1,
Table 3) is almost twice that of the apoprotein. The greater

stability of the holoprotein results from an increase in the
transition midpoint which is approximately doubled. The ®t
of the urea-unfolding curve for ¯avodoxin with Eqn (4)
suggests that the number of amino-acid side chains that
become exposed when the holoprotein unfolds (30.6 ‹ 1.7)
is % 9 less than for the apoprotein.
In contrast to the apoprotein for which the transition
midpoint was found to be independent of the protein concentration, the transition midpoint for the holoprotein
increases gradually from % 2.4 M urea when the protein concentration is 0.25 lM to % 3.2 M urea at 23 lM protein
(Table 3). This increase may indicate that the protein associates to a polymeric form. However, there is no experimental evidence for such a phenomenon with this
¯avodoxin in the absence of urea. The values calculated
for the conformational stability and the slope (m) in the
transition region at each protein concentration do not
follow a de®nite pattern (Table 2), and therefore it is not
possible to draw a ®rm conclusion about the cause of this
increase in the transition midpoint.
When the concentration of salt is increased the unfolding
curve is shifted to the right (Fig. 7), as also occurs with the
apoprotein. However, the resultant increase in stability of
the holoprotein is smaller than the increase in stability that
occurs with the apoprotein. As a consequence, the stability
of the apoprotein in high salt is greater than the stability of
the holoprotein. Phosphate increases the slope in the
transition region of the unfolding curve for apo¯avodoxin
but it has no effect on the slope of the curve for ¯avodoxin.
Furthermore, the slope of the curve for the holoprotein is

smaller when NaCl rather than phosphate is used to
increase the ionic strength (Table 2, Fig. 7).

DISCUSSION
The experiments described in this paper show that
D. vulgaris apo¯avodoxin is reversibly unfolded by urea
in a reaction whose equilibrium midpoint depends on the
ionic composition of the solution. The change in protein
¯uorescence as a function of the concentration of urea is
consistent with a two-state model of unfolding, as are the
kinetics of the reaction. However, two observations suggest
that the apoprotein of D. vulgaris ¯avodoxin is not
completely unfolded in urea. First, the maximum ¯uorescence emission occurs at 351 nm, and is therefore blueshifted compared with the protein emission from unfolded
apo¯avodoxins from A. vinelandii [28] and Anabaena [4]
which occurs at 354±355 nm. The 3±4 nm difference
suggests that the side chains of the aromatic amino acids
in the urea-treated D. vulgaris protein may not all be as
exposed to solvent as those in the unfolded apo¯avodoxins
from the other two organisms [29]. Second, it is calculated
that only 30% of the amino-acid side chains become
exposed to solvent after urea treatment, a value that is
similar to the value that can be calculated for the ureaunfolding of apo¯avodoxin from Anabaena (using the data
of Fig. 6 in [4]).
The observation that the ®nal protein ¯uorescence of
urea-treated ¯avodoxin is the same as that of apoprotein
treated with the same concentration of urea suggests that
when the urea concentration is suf®ciently high both forms
of the protein unfold to the same extent. The observation
that the ¯uorescence due to FMN in urea-unfolded

Â
220 B. O Nuallain and S. G. Mayhew (Eur. J. Biochem. 269)

¯avodoxin is the same as that of protein-free FMN under
the same conditions indicates that the FMN is fully
dissociated from the protein. The kinetics of the two
unfolding reactions involve a single exponential, similar to
those observed with other small single-domain proteins [29],
and because the rate constants for the increase in FMN
¯uorescence and protein ¯uorescence are so similar, we
conclude that the rate-limiting step in unfolding of the
holoprotein is the slow dissociation of FMN from the
apoprotein. The only evidence so far that urea perturbs
the holoprotein at urea concentrations less than that
required in the transition region is the small increase in
protein and ¯avin ¯uorescence at urea concentrations up to
1 M. As was discussed earlier, the ¯uorescence increase
appears to be due to the release of FMN from the protein in
accordance with a shift of the equilibrium in Eqn (6) to the
right. This results partly from unfolding of apoprotein in the
holoprotein/apoprotein equilibrium at these low concentrations of urea, but presumably also from an interaction of
urea at the FMN binding site, possibly by hydrogenbonding to polar groups on the protein [31].
The conformational stabilities calculated for D. vulgaris
¯avodoxin and apo¯avodoxin in low salt are small by
comparison with other globular proteins for which values in
the range 21±60 kJámol)1 have been reported [32]. The value
for apo¯avodoxin is also low by comparison with the values
for apo¯avodoxins from Anabaena and A. vinelandii. The
low stability of the D. vulgaris protein might be because of

residual structure in the urea-unfolded protein, or because
of a stable intermediate not detected by ¯uorimetry that
could decrease the slope in the transition region of the
folding curve [17]. Stabilization by salts similar to that
observed with D. vulgaris apo¯avodoxin and ¯avodoxin has
been observed with Anabaena apo¯avodoxin [4] and with
the chemotactic protein CheY that has the ¯avodoxin-like
fold [33]. The increase in stabilization might originate from
effects such as those discussed above, including destabilization of a folding intermediate, as well as additional effects
such as preferential binding of salt ions to the folded protein
or an effect on the properties of the solvent. The use of a
single spectroscopic technique to monitor unfolding, as used
in the present study, cannot exclude the formation of stable
intermediates in the reaction, nor does it allow the
conclusion that the protein at the end point of the transition
is devoid of all secondary and tertiary structure. Measurements of the unfolding equilibrium by additional techniques
such as far UV circular dichroism might reveal different
unfolding equilibria, as recently observed in the guanidine HCl unfolding of apo¯avodoxins from A. vinelandii
[6±8] and D. desulfuricans [10]. It is known that other
proteins whose unfolding curves ®t the two-state model in
fact give intermediates in their unfolding/folding reactions
[34].
The larger values calculated for the conformational
stabilities of ¯avodoxin and its apoprotein in phosphate,
compared with NaCl, and the smaller rate of unfolding of
apoprotein in NaCl, indicates that the increased stabilization of the protein by salts cannot be explained simply by
a shielding of charged groups that might otherwise
destabilize the protein. The decrease in the m value for
¯avodoxin with NaCl is unusual but not unique because a
similar effect has been reported on the m value for an

equilibrium intermediate in the unfolding of apomyoglobin

Ó FEBS 2002

from Aplysia limacina [34]. In this case it was suggested
that the large decrease (% threefold) in the m value with
KCl is due to deviations from a proposed three-state
model.
It was calculated that the NaCl induced shift in the ureaunfolding curve for D. vulgaris apo¯avodoxin could be due
to preferential binding of two salt ions to the folded protein
(Fig. 5, inset). There is no direct evidence for the binding of
salts by this ¯avodoxin although it is known that the rate of
association of FMN with the apoprotein is inhibited by
dianionic phosphate, possibly due to competition between
FMN and phosphate for the binding site [35], and in the
apoprotein±ribo¯avin complex a phosphate or sulfate anion
occupies the site in the protein that is normally occupied by
the phosphate of FMN [36]. The interactions between ¯avin
and apo¯avodoxin also depend on the cation; the interaction is weaker in potassium salts than in sodium salts [13,37].
The multiple binding of ions to other a/b proteins and the
stabilization of folded states by such ions are well
established [34,38,39].
According to the Hofmeister series [40], phosphate
dianion and to a lesser extent chloride anion, disrupt the
structure of water, markedly increase its surface tension, and
decrease the solubility of nonpolar molecules (the so-called
salting-out effect). If the observed salt effects on ¯avodoxin
and apo¯avodoxin are due only to a change to the physical
properties of the solvent, large stabilizing effects should be
caused by phosphate relative to chloride. Phosphate should

then lead to larger shifts in the unfolding curves. It is
observed experimentally, however, that chloride and phosphate have similar effects, suggesting that the salt effects are
not due only to a change to the physical properties of the
solvent. It is concluded that the greater conformational
stabilities of the two forms of the protein at high concentrations of salt are probably due to a combination of factors
including the preferential binding of salt ions to the folded
protein, ionic strength effects, as well as to a change in the
physical properties of the solvent.
A detailed study of unfolding of the holoprotein of a
¯avodoxin has been reported for one other protein, namely
¯avodoxin from D. desulfuricans [10]. This protein was
unfolded with guanidine HCl in 3 mM phosphate pH 7. The
reactions differed in several ways from the unfolding
reactions of D. vulgaris ¯avodoxin described above. First,
the reactions with the D. desulfuricans holoprotein were
complete in less than 1 min rather than requiring hours to
reach completion. Second, the protein ¯uorescence increased
at low concentrations of denaturant without a change in the
¯avin ¯uorescence. Third, the FMN was found to be bound
tightly after the protein had been unfolded (calculated
Kd ˆ 0.2 nM [10]). The changes in protein ¯uorescence were
found to occur at smaller concentrations than changes in the
far UV circular dichroism of the protein, leading to the
conclusion that unfolding of this protein does not ®t a twostate model, but rather that it occurs through an intermediate partly unfolded state in which the FMN is still tightly
bound to the apoprotein.The conformational stabilities
determined for ¯avodoxin and apo¯avodoxin from
D. desulfuricans using guanidine HCl [10] are similar to
the stabilities determined for the corresponding proteins
from D. vulgaris but using urea in high salt. It seems likely
that the charge on guanidine HCl has a stabilizing effect

similar to that of a high concentration of phosphate or


Ó FEBS 2002

chloride ion. In support of this conclusion, the conformational stability of D. vulgaris apo¯avodoxin in guanidine HCl was found to be greater than the stability in urea.
The marked differences between the unfolding reactions
of ¯avodoxins from two species of Desulfovibrio are
somewhat surprising because the primary sequences and
overall crystal structures of the two proteins are very
similar [11,12]. Flavin binding by the apoprotein of
D. desulfuricans ¯avodoxin is reported to be stronger than
that of the apoprotein of D. vulgaris ¯avodoxin (Kd values
of 0.1 and 0.24 nM, respectively [35,41]). It should be noted
however, that the experimental data for the D. desulfuricans protein do not support such a small value for the Kd.
The value given in [41] was obtained from spectrophotometric measurements in which aliquots of apoprotein were
added to 43.8 lM FMN. It is clear that a Kd value as small
as 0.1 nM could not be measured by this method.
Recalculation of Kd from the experimental points that lie
off the straight lines of Fig. 4 in [41] indicates that its value
is 0.14 ‹ 0.1 lM. If the Kd value for the oxidized
D. desulfuricans protein is indeed % 700 times greater than
that of D. vulgaris ¯avodoxin, differences in the mechanisms of unfolding of the two proteins might be understandable. The conformations of the two loops of protein
that envelop the FMN are different in the two proteins
[11,12]. As a result, the carbonyl of glutamate 99 in
D. vulgaris ¯avodoxin points towards the solvent, while in
the D. desulfuricans protein it points towards the ¯avin
and is 0.29 nm from O(4) of the isoalloxazine structure. As
was noted by others [12], the orientation in the D. desulfuricans protein should lead to O-O repulsion and to a less
stable ¯avin±protein complex. It should be noted further

that neither the published Kd value of 0.1 nM for the
oxidized protein nor the re-estimated value of 0.14 lM
leads to the Kd values that have been published for the
semiquinone and hydroquinone forms of this ¯avodoxin
[40]; the values reported seem to greatly underestimate the
strengths of interaction between this apo¯avodoxin and
the two reduced forms of FMN.
Based on the observations described above, a scheme can
be devised for the unfolding/folding of D. vulgaris ¯avodoxin and apo¯avodoxin in urea (Fig. 8). It is proposed that
urea binds rapidly to the apoprotein to form a complex to
which FMN binds relatively weakly, possibly because
of competition between urea and ¯avin for the same
hydrogen-bonding groups on the protein and/or because of
a urea-induced change in the protein conformation. More
denaturant binds to apoprotein at greater urea concentrations (reaction I) leading to unfolded protein [(urea)x-apo*urea]. It is further proposed that when ¯avodoxin is treated
with urea, the denaturant reacts rapidly with both the
holoprotein (reactions E/F) and the apoprotein (reactions
C/D) so that a weaker holoprotein complex results (apourea and urea-apo-FMN appear in solution). The model
proposes that the subsequent unfolding/folding reactions of
the holoprotein can occur by two routes. One of these
involves unfolding of apoprotein (reaction I) and a consequent perturbation of the holoprotein/apoprotein equilibria
(reactions A/B and G/H). Note that the equilibria A/B and
G/H do not depend directly on the concentration of urea.
The other route involves further interaction of urea with the
holoprotein complex (urea-apo-FMN) and the direct
unfolding of this complex (reaction L).

Flavodoxin unfolding (Eur. J. Biochem. 269) 221

Fig. 8. Scheme for the unfolding/folding reactions of ¯avodoxin in urea.

apo, is folded apo¯avodoxin; urea-apo-FMN is a quasi-folded ¯avodoxin at low urea concentrations; apo-urea is quasi-folded apo¯avodoxin at low urea concentrations; and (urea)x-apo*-urea is unfolded
protein. The sum of (urea)x and urea is the concentration of urea
required to completely unfold the protein.

When the urea concentration is low, the small concentration of apoprotein (apo-urea) that is in equilibrium with
the holoprotein, unfolds rapidly to a new equilibrium that
includes completely unfolded protein [(urea)x-apo*-urea],
the two species of folded protein (apo-urea and urea-apoFMN), free FMN and urea. The FMN prevents the
apoprotein from unfolding completely and maintains a high
equilibrium concentration of apo-urea-FMN. As the direct
unfolding/folding reactions of the holoprotein complex
(reactions K/L) are very slow, the protein unfolding/folding
occurs mainly via the apoprotein routes through the apourea complex.
The scheme of Fig. 8 provides a working hypothesis for
the overall unfolding/folding reactions of D. vulgaris
apo¯avodoxin and ¯avodoxin, and it forms a basis for
further experimentation. It does not account for all of the
experimental observations on the system, in particular the
different effects of salt on the two forms of the protein that
cause the conformational stability of the apoprotein in
high salt to be greater than that of the holoprotein. It is possible that an intermediate occurs during folding/unfolding
of the holoprotein in high salt and that this decreases the
slope of the equilibrium curve, leading to an underestimate
of the conformational stability. The scheme proposes that
addition of free FMN should shift the unfolding equilibrium even further to the right. Such a shift might
be dif®cult to detect using ¯uorescence methods because
of high background emission from the added ¯avin.
However, it should be possible to test the scheme by using
an alternative method such as circular dichroism or nuclear
magnetic resonance spectroscopy, together with the use of

Â
222 B. O Nuallain and S. G. Mayhew (Eur. J. Biochem. 269)

modi®ed ¯avins that bind to the apo¯avodoxin, and
mutant apoproteins that modify the ¯avin-binding site
[42].

ACKNOWLEDGEMENTS
We are grateful for support by Enterprise Ireland and by the EU
through the Human Capital and Mobility Programme (CHRX-CT930166). We thank Dr A. F. Buckmann for a gift of puri®ed FMN.
È

REFERENCES
1. Mayhew, S.G. & Ludwig, M.L. (1975) Flavodoxins and ElectronTransferring Flavoproteins. In The Enzymes (Boyer, P.D., ed.),
Vol. 12B, 3rd edn, pp. 57±109. Academic Press, New York.
2. Mayhew, S.G. & Tollin, G. (1992) General properties of
¯avodoxins. In Chemistry and Biochemistry of Flavoenzymes
(Muller, F., ed.), Vol. 3, pp. 389±426. CRC Press, Boca Raton, FL.
È
3. Brenner, S.E., Chothia, C. & Hubbard, T.J.P. (1997) Population
statistics of protein structures: lessons from structural classi®cations. Curr. Opin. Struct. Biol. 7, 369±376.
Â
Â
4. Genzor, C.G., BeldarraõÂ n, A., Gomez-Moreno, C., LopezLacomba. J.L., Cortijo, M. & Sancho, J. (1996) Conformational
stability of apo¯avodoxin. Protein Sci. 5, 1376±1388.
Â
5. Maldonado, S., Jimenez, M.A., Langdon, G.M. & Sancho, J.
(1998) Cooperative stabilisation of a molten globule apo¯avodoxin fragment. Biochemistry 37, 10589±10596.

6. van Mierlo, C.P.M., van Dongen, W.M.A.M., Vergeldt, F., van
Berkel, W.J.H. & Steensma, E. (1998) The equilibrium unfolding
of Azotobacter vinelandii apo¯avodoxin II occurs via a relatively
stable folding intermediate. Protein Sci. 7, 2331±2344.
7. van Mierlo, C.P.M., van den Oever, J.M.P. & Steensma, E. (2000)
Apo¯avodoxin (un) folding followed at the residue level by NMR.
Protein Sci. 9, 145±157.
8. van Mierlo, C.P.M. & Steensma, E. (2000) Protein folding and
stability investigated by ¯uorescence, circular dichroism (CD), and
nuclear magnetic resonance (NMR) spectroscopy: the ¯avodoxin
story. J. Biotechnol. 79, 281±298.
9. Knight,E. Jr & Hardy, R.W.F. (1967) Flavodoxin. Chemical and
biological properties. J. Biol. Chem. 242, 1370±1374.
10. Apiyo, D., Guidry, J. & Wittung-Stafshede, P. (2000) No cofactor
e€ect on equilibrium unfolding of Desulfovibrio desulfuricans
¯avodoxin. Biochim. Biophys. Acta 1479, 214±224.
11. Watt, W., Tulinsky, A., Swenson, R.P. & Watenpaugh, K. (1991)
Comparison of the crystal structures of a ¯avodoxin in its three
oxidation states at cryogenic temperatures. J. Mol. Biol. 198,
195±208.
12. Romero, A., Caldiera, J., LeGall, J., Moura, I., Moura, J. &
Romao, M. (1996) Crystal structure of ¯avodoxin from Desulfovibrio desulfuricans ATCC 27774 in two oxidation states. Eur. J.
Biochem. 239, 190±196.
13. Curley, G.P., Carr, M.C., Mayhew, S.G. & Voordouw, G. (1991)
Redox and ¯avin-binding properties of recombinant ¯avodoxin
from Desulfovibrio vulgaris (Hildenborough). Eur. J. Biochem.
202, 1091±1100.
14. Wassink, J.M. & Mayhew, S.G. (1975) Fluorescence titration with
apo¯avodoxin: a sensitive assay for ribo¯avin 5¢-phosphate and
FAD in mixtures. Anal. Biochem. 68, 609±616.

15. Whitby, L.G. (1953) A new method for preparing ¯avin-adenine
dinucleotide. Biochem. J. 54, 437±442.
16. Greene, R.F.J. & Pace, C.N. (1974) Urea and guanidine hydrochloride denaturation of ribonuclease, lysozyme, a-chymotrypsin,
and b-lactoglobulin. J. Biol. Chem. 249, 5388±5393.
17. Pace, C.N. (1986) Determination and analysis of urea and guanidine hydrochloride denaturation curves. Methods Enzymol 131,
266±280.

Ó FEBS 2002
18. Santoro, M.M. & Bolen, D.W. (1988) Unfolding free-energy
changes determined by the linear extrapolation method. 1.
Unfolding of phenylmethanesulfonyl alpha-chymotrypsin using
di€erent denaturants. Biochemistry 27, 8063±8068.
19. Staniforth, R.A., Burston, S.G., Smith, C.J., Jackson, G.S.,
Badcoe, I.G., Atkinson, T., Holbrook, J.J. & Clarke, A.R. (1993)
The energetics and cooperativity of protein folding: a simple
experimental analysis based upon the solvation of internal residues. Biochemistry 32, 3842±3851.
20. Nozaki, Y. & Tanford, C. (1963) The solubility of amino acids and
related compounds in aqueous urea solutions. J. Biol. Chem. 238,
4074±4081.
21. Nozaki, Y. & Tanford, C. (1970) The solubility of amino acids,
diglycine, and triglycine in aqueous guanidine hydrochloride
solutions. J. Biol. Chem. 245, 1648±1652.
22. Pace, C.N. (1975) The stability of globular proteins. CRC Crit.
Rev. Biochem. 3, 1±43.
23. Wetlaufer, D.B., Malik, S.K., Stoller, L. & Con, R.L. (1964)
Non-polar group participation in the denaturation of proteins by
3
urea and guanidinium salts. Model compound studies. J. Am.
Chem. Soc. 86, 508±514.
24. Record, M.T., Anderson, C.F. & Lohman, T.M. (1978) Thermodynamic analysis of ion e€ects on the binding and conformational equilibria of proteins and nucleic acids: the roles of ion

association or release, screening, and ion e€ects on water activity.
Q. Rev. Biophys. 11, 103±178.
4 25. Frost, A.A. & Pearson, R.G. (1961) Kinetics and Mechanism, 2nd
edn. John Wiley, New York.
26. Burstein, E.A., Vedenkina, N.S. & Ivkova, M.N. (1973) Fluorescence and the location of tryptophan residues in protein molecules.
Photochem. Photobiol. 18, 263±279.
27. Tanford, C. (1968) Protein denaturation. Part A and B. Adv.
Protein Chem. 32, 121±282.
28. Steensma, E., Nijman, M.J.M., Bollen, Y.J.M., Adrie, D., Jager,
P., van den Berg, W.A.M., van Dongen, W.M.A.M. & van
Mierlo, C.P.M. (1998) Apparent local stability of the secondary
structure of Azotobacter vinelandii holo¯avodoxin II as probed by
hydrogen exchange: implications for redox potential regulation
and ¯avodoxin folding. Protein Sci. 7, 306±317.
Â
29. Colon, W. (1999) Analysis of protein structure by solution optical
spectroscopy. Methods Enzymol. 309, 605±632.
30. Jackson, S.E. & Fersht, A.R. (1991) Folding of chymotrypsin
inhibitor 2. Evidence for a two-state transition. Biochemistry 30,
10428±10435.
31. Makhatadze, G.I. & Privalov, P.L. (1992) Protein interactions
with urea and guanidinium chloride: a calorimetric study. J. Mol.
Biol. 226, 491±505.
32. Pace, C.N. (1990) Conformational stability of globular proteins.
Trends Biochem. Sci. 15, 14±17.
33. Filimanov, V.V., Rrieto, J., Martinez, J.C., Bruix, M., Mateo, P.L.
& Serrango, L. (1993) Thermodynamic analysis of the chemotactic
protein from Escherichia coli CheY. Biochemistry 32, 12906±
12921.
34. Barrick, D. & Baldwin, R.L. (1993) Three-state analysis of sperm

whale apomyoglobin folding. Biochemistry 32, 3790±3796.
35. Pueyo, J.J., Curley, G.P. & Mayhew, S.G. (1996) Kinetics and
thermodynamics of the binding of ribo¯avin, ribo¯avin 5¢-phosphate and ribo¯avin 3¢,5¢-bisphosphate by apo¯avodoxins.
Biochem. J. 313, 855±861.
36. Walsh, M.A., McCarthy, A., O'Farrell, P.A., McCardle, P.,
Cunningham, P.D., Mayhew, S.G. & Higgins, T.M. (1998)
X-ray crystal structure of the Desulfovibrio vulgaris (Hildenborough) apo¯avodoxin-ribo¯avin complex. Eur. J. Biochem. 258,
362±371.
37. O'Nuallain, B. (1998) An investigation of ligand interactions with
apo¯avodoxin from Desulfovibrio vulgaris and the cell adhesion


Ó FEBS 2002
molecule LFA-3. PhD Thesis, National University of Ireland,
Dublin.
38. Nelson, C.A., Hummel, J.P., Swenson, C.A. & Freidman, L.
(1962) Stabilisation of pancreatic ribonuclease against urea denaturation by anion binding. J. Biol. Chem. 237, 1575±1580.
39. Pace, C.N. & Grimsley, G.R. (1988) Ribonuclease T1 is stabilised
by cation and anion binding. Biochemistry 27, 3242±3246.
40. Collins, K.D. & Washabaugh, M.W. (1985) The Hofmeister
e€ect and the behaviour of water at interfaces. Q. Rev. Biophys.
18, 323±422.

Flavodoxin unfolding (Eur. J. Biochem. 269) 223
41. Caldeira, J., Palma, P.N., Regalla, M., Lampreia, J., Schafer, W.,
È
LeGall, J., Moura, I. & Moura, J.J.G. (1994) Primary sequence,
oxidation-reduction potentials and tertiary-structure prediction of
Desulfovibrio desulfuricans ATCC 27774 ¯avodoxin. Eur. J.
Biochem. 220, 987±995.

42. Mayhew, S.G., O'Connell, D.P., O'Farrell, P.A., Yalloway, G.N.
& Geoghegan, S.M. (1996) Regulation of the redox potentials of
¯avodoxins: modi®cation of the ¯avin binding site. Biochem. Soc.
Trans. 24, 122±127.