Conformational changes of b-lactoglobulin in sodium bis(2-ethylhexyl)
sulfosuccinate reverse micelles
A fluorescence and CD study
Suzana M. Andrade, Teresa I. Carvalho, M. Isabel Viseu and Sı
´lvia M. B. Costa
Centro de Quı´mica Estrutural, Complexo 1, Instituto Superior Te
´cnico, Lisboa, Portugal
The effect of b-lactoglobulin encapsulation in sodium bis
(2-ethylhexyl) sulfosuccinate reverse micelles on the envi-
ronment of protein and on Trp was analysed at different
water contents (x
0
). CD data underlined the distortion of the
b-sheet and a less constrained tertiary structure as the x
0
increased, in agreement with a concomitant red shift and
a decrease in the signal intensity obtained in steady-state
fluorescence measurements. Fluorescence lifetimes, evalu-
ated by biexponential analysis, were s
1
¼1.28 ns and
s
2
¼3.36 ns in neutral water. In reverse micelles, decay-
associated spectra indicated the occurrence of important
environmental changes associated with x
0
. Bimolecular
fluorescence quenching by CCl
4
and acrylamide was
employed to analyse alterations in the accessibility of the two
Trp residues in b-lactoglobulin, induced by changes in x
0
.
The average bimolecular quenching constant <kCCl4
q>was
found not to depend on x
0
, confirming the insolubility of this
quencher in the aqueous interface, while <kacrylamide
q>
increases with x
0
. The drastic decrease with x
0
of k
q
, asso-
ciated with the longest lifetime, kCCl4
q2 , comparatively to the
increase of kacrylamide
q2 , emphasizes the location of b-lacto-
globulin in the aqueous interfacial region especially at
x
0
‡10. The fact that kacrylamide
q2 (x
0
¼30) kacrylamide
q2
(water) also confirms the important conformational changes
of encapsulated b-lactoglobulin.
Keywords:b-lactoglobulin; conformation; quenching;
reverse micelles.
Many biological phenomena occur at interfaces rather than
in homogeneous solution, and protein–surfactant inter-
actions play a key role in the reactions involving membrane
proteins [1,2]. The role of reverse micelles (RM) has been
pointed out as a convenient membrane-mimetic medium
for the study of interactions with bioactive peptides [3]. In
particular, RM formed using the anionic surfactant, sodium
bis(2-ethylhexyl) sulfosuccinate (AOT), have been widely
reported for extractive separation and purification of
proteins [4,5].
Briefly, RM can be described as water nanodroplets
dispersed in water-immiscible apolar solvents, stabilized by
a monolayer of surfactant with its nonpolar tails protruding
into the oil and the polar headgroups in direct contact with
the central water core [6]. The droplet size can be altered
with a concomitant change on the properties of the water
inside the RM. As water is added, the radius of the water
pool (range 1.5–10 nm) increases as a function of the
water : surfactant ratio (x
0
). RM are protein-sized and,
consequently, proteins and other biopolymers can be
accommodated in different microenvironments according
to their physico-chemical nature and the properties of the
interfacial layer. The presence of proteins results in struc-
tural changes in both the biomolecules and the micellar
aggregates.
Milk proteins are widely valued within the food industry
for their emulsifying and emulsion-stabilizing properties.
These proteins become rapidly adsorbed at the oil/water
interface generated during emulsification [7]. b-Lactoglo-
bulin (bLG), is a globular, acid-stable protein of 162
residues, which constitutes approximately two-thirds of the
whey fraction of ruminant milk. The structural similarity
of bLG to retinol-binding protein has been noted, and
crystallography confirmed the typical lipocalin topology,
containing a b-barrel or calyx composed of eight antiparallel
b-strands, b
A
to b
H
[8]. bLG exists as a dimer in solutions of
physiological pH, but exhibits complex association equili-
bria, shifting between monomer, dimer, tetramer, octamer,
and monomer again, upon lowering the solution pH from
8.5 to 2.0 [9].
It is of special interest that bLG has a marked high
a-helical propensity [10,11] and an afibtransition was
detected, by time-resolved CD spectroscopy, during its
folding process [12]. Thus, bLGmayserveasamodelforthis
conformational change associated with the prion diseases or
with Alzheimer’s disease [13]. In spite of the vast number of
studies, involving bLG, which have been carried out over the
past 60 years, the biological function of this protein is still
unclear. Its inclusion in the lipocalin family led to the
suggestion of a transport role. In fact, bLG exhibits affinity
for a variety of hydrophobic ligands, such as retinol, fatty
acids, etc. [14,15]. The fact that bLG increases lipase activity
Correspondence to S. M. Andrade, Centro de Quı
´mica Estrutural,
Complexo 1, Instituto Superior Te
´cnico, 1049–001 Lisboa Codex,
Portugal. Fax: + 351 21 8464455, Tel.: + 351 21 8419389,
E-mail: sandrade@popsrv.ist.utl.pt
Abbreviations: AOT, sodium bis(2-ethylhexyl) sulfosuccinate; bLG,
b-lactoglobulin; DAS, decay associated spectra; GdnHCl, guanidine
hydrochloride; NAT, N-acetyltryptophan; NATA, N-acetyltrypto-
phanamide; RM, reverse micelles; x
0
, water : surfactant ratio.
(Received 8 October 2003, revised 4 December 2003,
accepted 22 December 2003)
Eur. J. Biochem. 271, 734–744 (2004) FEBS 2004 doi:10.1111/j.1432-1033.2004.03977.x
and contributes to the removal of free fatty acid suggested
that bLG could facilitate the digestion of milk fat [16].
The main goal of the present investigation was to obtain
information on the conformation of bLG when it is
encapsulated in AOT RM at a wide range of water-pool
sizes. The study of the interaction between bLG and AOT
RM was carried out using CD, and steady-state and time-
resolved fluorescence techniques. In spite of the widespread
use of the intrinsic fluorescence of Trp as a probe of
microenvironmental changes, only a few publications have
reported on the photophysics of proteins in RM [17–19].
bLG has two Trp residues that are differently exposed to the
water solvent (Scheme 1): Trp19, facing into the base of
the hydrophobic pocket, is essentially inaccessible to the
solvent, whereas Trp61, at the end of strand b
C
,isrelatively
exposed [8]. The guanidino group of Arg124 lies only 3–4 A
˚
from the indole ring of Trp19, and Trp61 is close to a
disulfide bridge. As both groups can be efficient quenchers
of Trp fluorescence, some discrepancy has been found in the
literature as to which residue the bLG fluorescence can be
attributed [8,20]. Quenching studies involving acrylamide
and CCl
4
provided evidence of a different accessibility of
these quenchers to Trp residues, which depended on the
quencher location in the RM and on x
0
.
Materials and methods
Sample preparation
Bovine bLG (AB mixture), chromatographically purified
and lyophilized to ‡90% purity (Sigma; catalogue no.
L-3908), N-acetyltryptophanamide (NATA) (Sigma; cata-
logue no. A-6501) and AOT of 99% purity (Sigma; catalogue
no. D-4422), were used without further purification. Acryl-
amide (99% purity, electrophoresis grade) (Aldrich; cata-
logue no. 14,866–0) and guanidine hydrochloride (GdnHCl;
99% purity) (Aldrich; catalogue no. 177253–100G) were
both used as received. All solvents were of spectroscopic
grade.
A stock solution of 0.1
M
AOT/iso-octane was prepared
and checked for fluorescence emission, which was negligible
at the experimental conditions used. RM solutions were
then prepared by the direct addition of bidistilled water to
the surfactant/hydrocarbon mixture. The protein was added
by the injection method and freshly prepared prior to use.
All the volume injected was considered as water and used
to calculate x
0
(x
0
¼[H
2
O]/[AOT]). A transparent solution
was always obtained after shaking for a few seconds. The
amount of water in dry micelle solution (x
o
¼0.15) was
determined by the Karl-Fischer method. bLG concentra-
tions were determined spectrophotometrically, using the
molar extinction coefficient e
280nm
¼17 600
M
)1
Æcm
)1
[21]
for the bLG protein monomer. The final concentration of
bLG was calculated relative to the total volume of the RM
solution and was kept small to ensure that (a) the
absorbance (A) was never > 0.1 and (b) multiple occupancy
would be statistically unlikely (assuming a Poisson distri-
bution). All measurements were made at 24 ± 1 C.
Absorption and CD spectroscopy
A Jasco V-560 spectrophotometer, together with a 10 mm
quartz cuvette, was used in UV-Vis absorption measure-
ments. CD spectra were obtained using a Jasco J-720
spectropolarimeter (Hachioji City, Tokyo). The protein
spectra were measured using 10 mm (for near-UV) and
2 mm (for far-UV) quartz cells. The solutions containing
5l
M
(far-UV) or 14 l
M
(near-UV) bLG were scanned at
20 nmÆmin
)1
, with a 0.2 nm step resolution, a 1 nm band-
width and a sensitivity of 10 millidegrees (mdeg). An average
of 5–10 scans was recorded and corrected by subtracting the
baseline spectrum of unfilled RM of the same composition.
The CD signal (in mdeg) was converted to molar ellipticity [h]
(deg cm
2
Ædmol
)1
), defined as [h]¼h
obs
(10cl)
)1
,whereh
obs
(mdeg) is the experimental ellipticity, c (molÆdm
)3
)isthe
monomeric protein concentration, and l (cm) is the cell path
length. The secondary structure content was evaluated by
using the
SELCON
3 program [22] from the
DICROPROT
2000
package (release 1.0.4), available free from the Internet
(http://dicroprot-pbil.ibcp.fr).
Steady-state and time-resolved fluorescence
spectroscopy
Fluorescence measurements were recorded using a Perkin-
Elmer LS 50B spectrofluorimeter, with excitation at
295 nm. The instrumental response at each wavelength
was corrected by means of a curve obtained using appro-
priate fluorescence standards together with the standard
provided with the instrument. The quantum yields of
NATA and bLG were determined relative to that of Trp
alone, at pH 7.0 and in aerated aqueous solution (/¼0.13)
[19], with appropriate corrections for the refractive index of
the solvent in AOT solutions. Steady-state fluorescence data
of bLG obtained at different water concentrations were
fitted to the following equation:
F¼FoþFwK1
d½H2On
1þK1
d½H2OnEqn ð1Þ
where Fis the fluorescence intensity (corrected for absorp-
tion at the excitation wavelength) and F
o
and F
w
are,
respectively, the fluorescence intensities in the absence and
Scheme 1. Ribbon diagram of a single unit of bovine b-lactoglobulin
(bLG) drawn using
SWISS PDBVIEWER
, version 3.7, with PDB file 1BEB.
The locations of Trp19 and Trp61 are indicated.
FEBS 2004 Conformational changes of bLG in reverse micelles (Eur. J. Biochem. 271) 735
presence of water; K
d
is the dissociation constant for the
interaction of water with the protein; and nis the Hill
coefficient which accounts for the system heterogeneity, as
described previously [23].
Fluorescence decay profiles were obtained using the time-
correlated single-photon counting method [24] with a
Photon Technology International (PTI) instrument. Exci-
tation of Trp at 295 nm was made with the use of a lamp
filled with H
2
, and sample emission measurements were
performed until a maximum of 10
4
counts was obtained.
NATA lifetime (s
F
¼2.85 ± 0.05 ns) was used as a
standard to check the apparatus response on a daily basis.
Data analysis was performed by a deconvolution method
using a nonlinear least-squares fit programme, based on the
Marquardt algorithm. The goodness of fit was evaluated by
statistical parameters (reduced v
2
and Durbin–Watson) and
graphical methods (autocorrelation function and weighted
residuals).
The decay associated spectra (DAS) of Trp fluorescence
in bLG were obtained using the following equation [25]:
FiðkÞ¼FSSðkÞaisi
R
iaisi
¼FSS ðkÞfiðkÞEqn ð2Þ
where s
i
are the fluorescence lifetimes and a
i
(k)arethe
normalized pre-exponential factors of the exponential
functions used for the global fit analysis. For each k
em
,
the steady-state intensity F
SS
(k) is the weighted sum of the
intensities F
i
(k) associated with each decay component.
The decay profiles were obtained at 10 nm intervals in the
wavelength range of the steady-state spectra (310–400 nm).
For fluorescence quenching experiments, a 3
M
stock
solution of acrylamide was used and the protein fluores-
cence (F) was monitored at 340 nm. The following correc-
tion factor:
fc¼ODtotal
ODbLG
110ODbLG
110ODtotal
was applied to Fto account for the fact that acrylamide
absorbs at the excitation wavelength (e
295 nm
¼0.27 ±
0.03
M
)1
Æcm
)1
)[19].TheCCl
4
extinction coefficient at
295 nm was not measurable and so inner filter effects were
negligible. Quenching data were analysed using the Stern–
Volmer equation [26]:
F0=F¼ð1þKsv½QÞeV½Q¼ðs0=sÞeV½QEqn ð3Þ
where F
0
and Fare the fluorescence intensities in the absence
and in the presence of the quencher Q, respectively; K
sv
and
Vare related, respectively, to the fluorescence extinction rate
constant for the dynamic (K
sv
¼k
q
s
0
,wheres
0
is the
fluorescence lifetime in the absence of the quencher) and
static processes. In the case of different ground state sites,
individual components of static quenching would contribute
to the quenching given by the following equation:
F0
F¼X
n
i¼1
fi
ð1þKSVi ½QÞeVi½Q
"#
1
Eqn ð4Þ
where K
SVi
and V
i
are, respectively, the dynamic and static
quenching constants for each fluorescent component i,and
f
i
is its corresponding fractional contribution to the total
fluorescence [26].
The errors of the calculated parameters were accessed
using the propagation theory, and the distribution F of
Snedcor was used to confirm, with 99% confidence, the
relationship among the variables [27].
Results
CD spectra of bLG in AOT RM
The effect of the amount of water (x
0
) inside AOT RM on
bLG far-UV CD spectra was followed at a pH
ext
of 6.5
(pH of the aqueous solution containing the protein)
(Fig. 1A). The band with a minimum at 216 nm, charac-
teristic of bLG in water [28], gradually broadened and
deepened as x
0
increased, so that the minimum shifted to
lower wavelengths. This suggests some change in the bLG
native structure. The spectra showed increased noise below
Fig. 1. CD spectra. (A) CD spectra of b-lactoglobulin (bLG) in
sodium bis(2-ethylhexyl) sulfosuccinate (AOT) RM and in pure water.
Far-UV CD spectra in: 1, water; 2, AOT, x
0
¼3; 3, AOT, x
0
¼5;
4, AOT, x
0
¼15; 5, AOT, x
0
¼30; and 6, GdnHCl (6
M
). (B) CD
spectra of bLG in AOT (1–4) and SDS (5 and 6) aqueous solutions at
different surfactant concentrations. Inset: % of a-helix obtained with
SELCON
3 for SDS (d)andAOT(h) aqueous solutions.
736 S. M. Andrade et al. (Eur. J. Biochem. 271)FEBS 2004
210 nm, which made it difficult to detect a reliable CD
signal below 200 nm. AOT itself is chiral and although a
background subtraction was performed the spectra had
significant noise, which could not be well discounted. On the
other hand, the presence of the protein may cause changes
on the AOT chirality and also on the RM size, which might
contribute to incorrect background compensation as the
average scattering intensity of RM is highly dependent on
the micelle size. Therefore, a quantitative appreciation of the
spectral changes was not possible. Curiously, the data
obtained at a lower x
0
are more similar to those of the
native aqueous structure than the ones obtained at a higher
x
0
. A study carried out in hexane, at different hydration
levels, showed that proteins retain their native conformation
for lower concentrations of water (10%) than at higher
water concentrations [29]. Structural changes occur owing
to the collapse of water clusters, at the surface of the protein,
into larger clusters; this provides a medium for ion diffusion
and ion pair formation, which leads to the movement of the
charged groups of the protein in order to keep themselves
neutral.
As a means of testing the role of the surfactant on the
protein conformations, bLG was dissolved in aqueous
AOT. The far-UV CD spectra obtained at different
concentrations of AOT (Fig. 1B), show (a) the appearance
of a minimum at around 208 nm and a shoulder at around
222 nm increasing with AOT concentration and (b) less
noisy spectra, which allows for quantitative analysis down
to 190 nm. Both the values of ellipticity obtained at 222 nm,
which can be converted into a-helix content [30], as well as
the results obtained by applying the
SELCON
3program,
indicate the same trend of increasing a-helix content as
AOT concentration increases. Above 6 m
M
,aplateauseems
to be reached (Fig. 1B, inset). Curiously, this value is in the
range of the critical vesicle concentration (5–8 m
M
) deter-
mined for aqueous AOT in the presence of different
concentrations of poly(ethylene glycol) [31].
Aqueous solutions of an analogous anionic surfactant,
SDS, were also tested and the analysis of far-UV CD data
using
SELCON
3 (Fig. 1B, inset) confirms that the observed
spectral changes (Fig. 1B) are the result of an increase in
bLG a-helical content, when the SDS concentration
increases, similar to that obtained in AOT/water.
Changes in the secondary structure are accompanied by
tremendous alterations in near-UV CD signals (Fig. 2). The
CD spectrum of bLG in its native conformation presents
two peaks, at 286 and 293 nm, arising from the vibrational
fine structure of Trp residues [28]. These peaks are absent in
RM, even at high x
0
, thus suggesting that Trp residues in
such altered conformation are in a much less specific (more
symmetrical) environment and have a higher mobility than
in the native bLG. However, even at a x
0
of 3.0, the CD
signal between 260 and 300 nm is greater than that of bLG
in 6
M
GdnHCl, implying the existence of some ordered
tertiary structure, even at this low hydration level. A plot of
the ellipticities at 293 nm vs. x
0
(inset of Fig. 2 or Fig. 4)
shows a nearly sigmoid behaviour, which could be indicative
of a two-state transition. The mid-transition (x
0,mid
)of
around 6–7 corresponds to the level where AOT headgroups
are fully hydrated and water molecules start to be free for
the protein hydration. Increasing x
0
also leads to a decrease
in [h]
270
(Fig. 2, inset), with the appearance of a broad band
in the 265–280 nm region, which is not detectable in
aqueous solution or in ethanol/water mixtures.
Fluorescence of bLG in AOT RM
Steady-state fluorescence. The fluorescence spectra
obtained depend strongly on the amount of solubilized
water, Fig. 3. There is a concomitant red shift, and a
significant decrease in the fluorescence quantum yield, as
x
0
increases. Comparatively to free aqueous solution
(k
max
¼338 nm), the spectra at x
0
<10 are blue-shifted
(up to 5 nm), suggesting that Trp residues are less exposed
at lower x
0
. This may be associated with a decrease in the
local dielectric constant and consequent lowering of the
average polarity of the Trp environment and/or with con-
formational changes of the protein that are accompanied by
Fig. 2. CD spectra. Near-UV CD spectra in: 1, water; 2, sodium bis
(2-ethylhexyl) sulfosuccinate (AOT), x
0
¼5; 3, AOT, x
0
¼30; and
4, GdnHCl (6
M
). Insets: molar ellipticities at 270 and 293 nm as a
function of x
0
andinwater.
Fig. 3. Fluorescence spectra. Fluorescence spectra of b-lactoglobulin
(bLG) (k
exc
¼295 nm) in sodium bis(2-ethylhexyl) sulfosuccinate
(AOT) reverse micelles (RM) at x
0
¼5 (1) and 30 (2); in pure water
(3); in 6
M
GdnHCl(4)andatatemperature(T)of75C(5).Inset:
Wavelengths of maximum emission (j) and fluorescence quantum
yields (s) as a function of x
0
andinwater.
FEBS 2004 Conformational changes of bLG in reverse micelles (Eur. J. Biochem. 271) 737
dislocation of local quenchers. The inset of Fig. 3 shows a
decrease in the intrinsic Trp fluorescence as x
0
increases.
The line through the data points was fitted to Eqn (1).
The value of K
d
¼0.16 ± 0.05
M
,withaQ
max
¼[(F
0
–
Fw)/F
0
]·100 ¼41 ± 5%, implies that both Trp residues
in bLG are probably not effectively quenched. The free
energy of this interaction (DG¼–RT ·lnK
d
,molar
standard state) at 25 Cis)4.6±0.1kJÆmol
)1
(equivalent
to the energy of one conventional hydrogen bond).
Data from both CD and steady-state fluorescence
spectroscopies in RM were converted to a normalized scale
between 0 and 1, to compare their variation and the mid-
point transition (Fig. 4). The ensemble of data provides
evidence of a common x
0,mid
between 6 and 7.
Time-resolved fluorescence. Fluorescence lifetime analysis
fitted well to a biexponential model throughout all studied
x
0
, similarly to water. Both lifetimes decreased upon
increasing the water content, although never reaching the
values obtained in free aqueous solution (Fig. 5), followed by
changes in the population associated with each lifetime
component. The weight of the shorter lifetime, which is the
major component in water (f
1
¼0.86), is reduced in RM,
becoming the major component only at x
0
‡10 and reaching
f
1
¼0.61 at x
0
¼30. As for the long component, taking into
account the CD results we may invoke the existence of
conformational changes affecting the Trp environment, in
such a way that quenching groups (e.g. disulfide bridges) may
no longer be effective and thus contribute to a longer lifetime
ofTrpinAOTRMthaninwater.
Decay associated spectra. More detailed information
about the individual environments of Trp residues in the
protein was obtained from DAS (Fig. 6, Table 1), which
were constructed across the emission spectrum (see Mate-
rials and methods). In free water (Fig. 6C), almost the entire
fluorescence intensity (80%) was caused by the DAS
of the short-lifetime component emitting at 338 nm
(s
1
¼1.28 ns), linked to the more hydrophobic region (less
polar and/or less accessible to water). In AOT RM, DAS
were obtained at x
0
¼5(Fig.6A)andx
0
¼30 (Fig. 6B),
providing evidence of quite different features. At x
0
¼5,
there was a larger contribution of the long-lifetime compo-
nent (s
2
¼4.07 ns) which emits more in the red
(k¼340 nm) and with the highest fractional intensity
(f
2
¼0.53). The short component (s
1
¼1.61 ns) was
similar to that in free water but contributed less to the
overall fluorescence and was blue shifted (k¼330 nm).
This implies that upon encapsulation, some changes
occurred in the vicinity of the Trp residues. At x
0
¼30,
the short-lifetime component (s
1
¼1.43 ns) became the
Fig. 5. Fluorescence lifetimes and fraction of the short-lived component.
Fluorescence lifetimes, s
1
(j)ands
2
(m),andfractionoftheshort-
lived component, f
1
(s), of b-lactoglobulin (bLG) in sodium bis(2-
ethylhexyl) sulfosuccinate (AOT) reverse micelles (RM) as a function
of x
0
andinpurewater.
Fig. 4. Comparison between ellipticities at 270 and 293 nm, and fluor-
escence quantum yields (k
exc
¼295 nm) for b-lactoglobulin (bLG) in
sodium bis(2-ethylhexyl) sulfosuccinate (AOT) reverse micelles (RM) as
afunctionofx
0
; and ellipticity at 222 nm for bLG in AOT/water. These
parameters were normalized to a scale between 0 and 1.
Fig. 6. Decay associated spectra (DAS). DAS for b-lactoglobulin (bLG) fluorescence in sodium bis(2-ethylhexyl) sulfosuccinate (AOT) reverse
micelles (RM) and in water, calculated using Eqn (2). The two spectra correspond to two lifetime components, s
1
(s)ands
2
(j). The dotted lines
were obtained by fitting to a Gaussian function (see Table 1 for details). (A) AOT, x
0
¼5; (B) AOT, x
0
¼30; (C) water.
738 S. M. Andrade et al. (Eur. J. Biochem. 271)FEBS 2004