You are on page 1of 10

278 Biomacromolecules 2001, 2, 278-287

β-Casein Adsorption at the Hydrophobized Silicon


Oxide-Aqueous Solution Interface and the Effect of Added
Electrolyte
Tommy Nylander,*,† Fredrik Tiberg,‡ Tsueu-Ju Su,§ Jian R. Lu,§ and Robert K. Thomas|
Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund University, Box 124,
S-221 00 Lund, Sweden, Institute for Surface Chemistry, Box 5607, S-114 86 Stockholm, Sweden,
Department of Chemistry, University of Surrey, Guildford GU2 5XH, U.K., and Physical and Theoretical
Chemistry Laboratory, Parks Road, Oxford OX1 3PJ, U.K.
Received October 24, 2000

The effect of the presence of NaCl, CaCl2, or MgCl2 at the same ionic strength on the structure of β-casein
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

layers adsorbed on hydrophobic surfaces has been investigated by neutron reflectivity measurements. The
Downloaded via OAK RIDGE NATL LABORATORY on October 2, 2019 at 19:03:53 (UTC).

data were fitted to a four-layer model. The volume fraction versus distance profiles have a similar shape
whether β-casein is adsorbed from NaCl, CaCl2, and MgCl2 of the same ionic strength or whether the protein
concentration is lowered 10 times. In particular at larger distances from the surface, the volume fraction
values are low and similar. However, close to the hydrophobic surface the volume fraction of protein decreases
in the order CaCl2 > MgCl2 > NaCl. We have also used a specific proteolytic enzyme, endoproteinase
Asp-N, which cleaves off the hydrophilic part of β-casein, as a tool to reveal the interfacial structure of the
protein. For all the different types of added electrolytes, endoproteinase Asp N only affects the outermost
β-casein layer. Subsequent addition of β-casein in all cases led to large increases in amounts adsorbed and
in the thickness of the outer layers.

Introduction There are strong similarities between β-casein and am-


The milk proteins and casein in particular have, due to phiphilic (di-)block copolymers in terms of their ability to
their excellent properties as stabilizers of emulsions, foams, associate in solution and in their interfacial behavior.
and dispersions, come to be used in a range of applications β-Casein, for instance, forms micellar-like aggregates (cmc
both in foods and in other applications, such as glue, paint, ≈ 0.5 mg/mL) in aqueous solutions.21,22 In comparison with
and putty.1-3 One of the major constituents of milk casein block copolymers the structure of interfacial layers of
is β-casein.4 Therefore, quite a number of studies have been proteins is hard to model due to the variety of monomers,
devoted to the behavior of β-casein at the air-aqueous and amino acid residues, present in the proteins. However,
oil-aqueous interfaces and the interfaces between hydro- proteins have the advantage that they are monodisperse.
phobic solid surfaces and an aqueous phase.5-20 These studies Thus, we know that β-casein contains 209 amino acid
have been conducted under various conditions, using a range residues in a particular order giving it a molecular mass of
of different experimental techniques and various types of 24 020 Da,4,23 although different genetic variants, where a
hydrophobic surfaces. However, they all seem to be consis- few of the amino acid residues are different, exist. Dickinson
tent in terms of the adsorbed amount, of the order of 3 mg/ et al. have modeled the structure of the adsorbed layer of a
m2, and in the structure of the adsorbed layer. One reason model polymer, resembling β-casein, on a hydrophobic
could be that the protein has such a strong amphiphilic surface,13,14,17 by using the self-consistent field theories of
character, with the amino acid sequence divided into one Scheutjens and Fleer.24 Their results qualitatively confirm
hydrophilic and one hydrophobic domain. Thus, β-casein the structure based on experimental studies. However, the
forms quite well structured monolayers on hydrophobic model calculations showed no evidence of two distinct layers
surfaces, in which the hydrophobic protein segments are with markedly different volume fractions of protein segments.
attached to the surface, forming a densely packed inner layer. Instead, the volume fraction of protein segments decreases
The highly charged N-terminal portion of this casein extends exponentially with the distance from the surface. They also
into the aqueous solution to form a brushlike structure. The found significant effects of ionic strength and pH on the
consequence is that the surface becomes more hydrophilic, segment density profiles. Similar effects have been demon-
and this is manifested by a change in the wetting properties.20 strated in a number of experimental studies.3,9,11,15,18 This is
* To whom correspondence should be addressed. E-mail:
due to the large number of charged and chargeable residues
Tommy.Nylander@fkem1.lu.se. in the N-terminal 50 residues of the protein. In fact the net
† Lund University.
‡ Institute for Surface Chemistry.
negative charge of this part is at least -16 at pH 7.4 These
§ University of Surrey. charge residues include five phosphorylated serines, which
| Physical and Theoretical Chemistry Laboratory. gives the protein a specific affinity to divalent ions such as
10.1021/bm0056308 CCC: $20.00 © 2001 American Chemical Society
Published on Web 02/15/2001
Casein Adsorption Biomacromolecules, Vol. 2, No. 1, 2001 279

calcium and magnesium.25,26 Our earlier study indicated that Rutherford Appleton Laboratory, ISIS, Didcot, U.K. A
the surface excess increased in the order NaCl < MgCl2 < detailed description of the experimental procedure and data
CaCl2 at constant ionic strength, although no significant evaluation employed in this study is given elsewhere.29-32
change in the layer thickness was observed.15 However, other The sample cell consisted of a Teflon trough clamped against
studies have indicated a decrease in layer thickness on a silicon block of dimensions 12.5 × 5 × 2.5 cm3.28 The
calcium addition.11,18 To investigate further the effect of collimated beam enters the end of the silicon block at a fixed
added electrolyte on the structure of β-casein layers on angle, is then reflected at a glancing angle from the solid-
hydrophobic surfaces, the present neutron reflectivity study liquid interface, and exits from the opposite end of the silicon
was undertaken. This has previously been demonstrated to block. The neutron reflectivity, that is the ratio of the
be a powerful technique for studying the structure of β-casein reflected and incoming beam intensities, is determined as a
layers at hydrophobic surfaces12 as well as the air-water function of momentum transfer (wave vector), κ, where κ
interface.8,11,13 We have also made use of a specific pro- ) 4π sin θ/λ (θ is the incidence angle and λ is the
teolytic enzyme, endoproteinase Asp-N,27 as a tool to reveal wavelength). Neutron wavelengths from 1 to 6 Å were used
the interfacial structure of β-casein.10,15,20 This protein has in these experiments. Each reflectivity profile was measured
four potential cleavage sites for the enzyme, where two are at three different glancing angles of 0.35°, 0.8°, and 1.8°,
located in the hydrophilic part (residues 43 and 47) and two and the results were combined. The beam intensity was
in the hydrophobic part (residues 129-184). calibrated using the totally reflected beam below the critical
angle with D2O in the cell. A flat background determined
Materials and Methods by extrapolation to high values of momentum transfer, κ,
The β-casein (genetic variant A1, Mw ) 24 000 g/mol) was further subtracted. The reflectivity profiles were always
was extracted from bovine milk and purified according to essentially flat for κ > 0.2 Å-1, although the limiting signal
the procedure described by Nylander and Wahlgren,10 where at this point was dependent on the H2O/D2O ratio. The
also details of the analyses of the sample are given. The background for the D2O runs was typically 2 × 10-6 and
sample was free from contamination of other proteins as the background for H2O was 3.5 × 10-6, given in terms of
determined by FPLC and in addition the protein was applied reflectivity.
on a Chelex-100 (200-400 mesh, BioRad) column for The structural parameters of the d-OTS layer on the surface
complete removal of cations before being lyophilized. of the silicon block were fully characterized before any
Endoproteinase Asp-N was purchased from Boehringen adsorption of the protein. To remove entrapped air on the
Mannheim Biochemica (catalog no. 1054589, lot 14184025). hydrophobic surface, the cell was initially filled with ethanol,
The water used was passed through an Elgastat ultrapure which was swept from the cell with excess of water. The
water system (UHQ). Deuterated water (99.9%, deuterated) characterization was then done in different isotopic composi-
was obtained from Fluorochem. All other chemicals used tions of water: pure D2O, water CMSiO2 (contrast matching
were of analytical grade. β-Casein (0.1 or 0.01 mg/mL) or for SiO2 with H2O/D2O weight ratio of 0.401/0.599), water
endoproteinase Asp-N (0.04 µg/mL) was dissolved in 50 mL CMSi (contrast matching for Si with H2O/D2O weight ratio
of 0.02 M imidazol-HCl buffers (pH 7.0) containing the of 0.595/0.405), and finally pure H2O. The cell was then
appropriate amount of sodium chloride, calcium chloride, filled with protein solution (about 40 mL), generally prepared
and magnesium chloride as described in the Results. Fresh from D2O, and the reflectivity profile was recorded versus
solutions were always prepared immediately before the time until no changes with time could be observed. It took
adsorption measurement was started. about 15 min after the protein was added until the reflectivity
The procedure for polishing the large face (111) of the measurement was started, and each measuring cycle took
silicon block and hydrophobizing with deuterated octa- about 40 min (depending on the neutron flux). At least two
decyltrichlorosilane (d-OTS, C18H37SiCl3) has been described measuring cycles were performed, and the reflectivity profiles
earlier.28 We will describe only the basic steps. The silicon were compared. If they were similar, the protein solution
block was initially soaked for 10 min in nitric acid followed was removed and the cell was filled with pure buffer prepared
by etching for 10 min in a pH-buffered mixture of NH4F from D2O (or H2O for the experiment with CaCl2). This
(40%)/HF (10%) at volume ratio of 7:1. The substrate was solution was then removed from the cell, and the cell was
then immersed in a mixture of 25% NH4OH (pro Analysi, filled with a fresh aliquot of buffer. The reflectivity profile
Merck), 30% H2O2 (pro analysi, Merck), and H2O (1:1:5, was then recorded as for the protein solution. The buffer
by volume) at 70 °C for 10 min, followed by thorough rinsing
was then changed to one prepared from H2O, and the
by water and finally CH2Cl2. The self-assembled layer of
reflectivity profile was recorded. This experimental protocol
d-OTS on the silica surface was formed by immersing the
was followed for the addition of enzyme and the second
block in a freshly prepared solution of d-OTS in CH2Cl2 at
addition of β-casein.
a concentration of 9.4 × 10-4 M for 1 h. The block was
then rinsed in CH2Cl2, ethanol, and water. Before use and The reflectivity profile, R(κ), is determined by the variation
between each adsorption experiment, the surface was cleaned of the scattering length density, F, along the normal to the
in 5% Decon 90 solution and the whole block was then surface, z, by
thoroughly rinsed in UHQ water.
The neutron reflection measurements were made on the 16π2
R(κ) ) |F̂(κ)|2 (1)
“white beam” time-of-flight reflectometer CRISP at the κ2
280 Biomacromolecules, Vol. 2, No. 1, 2001 Nylander et al.

Table 1. Constants Used for the Fitting of the Reflectivity Profiles Table 2. Characterization of the Hydrophobic Interface (d-OTS
(Taken from Reference 12) (95% D)/SiO2)a
density volume τ(3
material (g/cm3) (Å3) b (10-4 Å) F (10-6 Å-2) layer (Å)b FD2O FH2O φOTS ( 0.1b φw ( 0.1b φair ( 0.1b
H2O 0.9975 30 -0.168 -0.56 oxide 18 4.00 2.62 0.2
D2O 1.105 30 1.905 6.35 OTS (inner) 17 6.35 6.35 1.0 0.0 0.0
water CMSi 1.038 30 0.621 2.07 OTS (outer) 20 5.40 1.41 0.3 0.6 0.2
water CMSiO2 1.059 30 1.023 3.41 a τ is the layer thickness (Å), F
D2O and FH2O are the scattering length
Si 2.32 20 0.415 2.07 densities in D2O and H2O, respectively, and φOTS, φw, and φair are the
SiO2 2.16 47 1.585 3.41 volume fraction of OTS, water, and air, respectively b The errors given
-C18D37 0.7768 542 36.65 6.76 reflect the maximum uncertainty in the fitting and any possible coupling
β-casein 1.365 29594 532.6 1.80 of the fitting parameters.

where F(κ) is the one-dimensional Fourier transform of F- be 18 Å, in agreement with these earlier studies. As discussed
(z), that is by Fragneto et al.,12,28 the d-OTS layer could be divided into
two layers, where the layer next to the silicon oxide surface
F̂(κ) ) ∫-∞+∞ e-iκz F(z) dz is crystalline. Consequently, this 17 Å thick layer consists
almost entirely of d-OTS with only 0.01 volume fraction
In turn the scattering length density depends on the chemical entrapped air. This outer layer, however, is defective and
composition of the sample as (cf. ref 30) contains a significant amount of water and air. The thickness
of this layer was found to be about 20 Å, and the volume
F) ∑nibi (2) fraction of d-OTS is about 0.27. This defective layer is
thicker than the value, 11 Å, reported earlier by Fragneto et
where ni is the number density of element i and bi is its al.28 However, they reported a higher surface coverage of
scattering amplitude (scattering length). After assumption of 5.9 µmol/m2, compared with a d-OTS coverage of 1.7 µmol/
a structural model for the adsorbed layer, the reflectivity m2 calculated from the values given in Tables 1 and 2. The
profile is calculated. The calculated reflectivity is then outer d-OTS layer in the present study can therefore be
compared with the measured data, and the structural param- considered to be more defective than that obtained in the
eters are varied until the fit is optimized. The structural earlier studies. It is worth commenting at this stage that the
parameters used in the fitting are the number of layers, division of the OTS into two layers is somewhat artificial.
thickness (τ), and the corresponding scattering length density A more gradual distribution would probably also fit the data,
(F) for each layer. The surface excess, Γ, for each layer can but the composition at the two limits, the total amount on
be deduced directly from the derived scattering length density the surface, and the overall thickness, would be little changed
and thickness of the layer using from the simpler two-layer division. Here, and in the
Fτ discussion of the β-casein results, we have opted for using
Γ) (3)

the minimum number of distinct layers required to fit the
Na( mibi + nwbw)
data. While this may not be entirely realistic, it is the standard
where ∑mibi denotes the total scattering length of the protein practice unless there are other reasons for describing the
molecule, Na is Avogadro’s number, nw is the number of system in terms of smooth distributions (for example to fit
water molecules associated with each protein molecule, and a theoretical model). The errors quoted in Table 2 reflect
bw is the scattering length of water. The total scattering length the maximum uncertainty in the fitting and any possible
of the protein depends on its chemical composition, where coupling of the fitting parameters. In the fitting we have also
each number mi of component i has a scattering length bi. allowed for the presence of air. We do normally see some
The volume fraction of protein, φp, can for each layer be air present at the hydrophobic surface unless it has been
obtained from specially treated with ethanol and water. It is displaced when
adsorption of other material occurs. We note that the surfaces
F ) φpFp + (1 - φp)Fw (4) were very hydrophobic and had a water contact angle of
>100°. They also gave very reproducible scattering curves.
where Fp and Fw are the scattering length densities of protein Since the outer part of the OTS layer was defective,
and water, respectively. The parameters used for the calcula- β-casein was likely to penetrate into the outer part of the
tions are given in Table 1. OTS. Thus when it came to fitting the reflectivity data for
adsorbed protein, we used the parameters already given for
Results the oxide and lower OTS layer but allowed the scattering
length density of the outer layer to vary according to how
The characteristics of the silicon sample block with the much protein was in it. It was immediately clear when fitting
OTS layer are given in Table 2. the data for the protein that the volume fraction declines away
As reported previously,12,28 the only model that could fit from the surface. This allows the choice of using a functional
the reflectivity profiles recorded for all four of the different form for the decay or of dividing the protein into uniform
water contrasts for the OTS layer on its own was a three- layers each containing less protein than the last. The
layer model. The lower oxide layer thickness was found to resolution of the experiment cannot distinguish these two
Casein Adsorption Biomacromolecules, Vol. 2, No. 1, 2001 281

Table 3. Adsorption/Desorption of β-Casein on/from Hydrophobized Silicaa


adsorption rinse
0.05 M 0.05 M 0.017 M 0.017 M 0.05 M 0.05 M 0.017 M 0.017 M
NaCl NaClb CaCl2c MgCl2 NaCl NaClb CaCl2 MgCl2
τ0 ( 3d 20 20 20 20 20 20 20 20
τ1 ( 3d 8 8 6 8 8 8 6 8
τ2 ( 5d 25 15 25 25 25 15 25 25
τ3 ( 10d 25 25 30 30 25 30 25
F0 4.7 4.7 2.44-3.02 5.1 4.7 4.7 3.9 5.1
F1 4.2 4.4 1.47 3.8 4.2 4.4 3.4 3.8
F2 5.8 5.8 0.03 5.5 6.0 5.9 5.5 5.7
F3 6.1 6.1 -0.39 6.1 6.2 6.1 6.2
φ0 ( 0.08d 0.36-0.49 0.36-0.49 0.44-0.72 0.24-0.37 0.36-0.49 0.36-0.49 0.44-0.72 0.24-0.37
φ1 ( 0.1d 0.6 0.6 0.9 0.8 0.6 0.6 0.9 0.8
φ2 ( 0.05d 0.16 0.16 0.25 0.25 0.1 0.13 0.25 0.19
φ3 ( 0.03d 0.07 0.07 0.07 0.07 0.04 0.07 0.04
Γ0 ( 0.2d 1.0-1.3 1.0-1.3 1.2-2.0 0.7-1.0 1.0-1.3 1.0-1.3 1.2-2.0 0.7-1.0
Γ1 ( 0.1d 0.7 0.6 0.7 0.8 0.7 0.6 0.7 0.8
Γ2 ( 0.15d 0.6 0.4 0.9 0.9 0.3 0.3 0.9 0.6
Γ3 ( 0.1d 0.2 0.2 0.3 0.3 0.1 0.3 0.2
τT 58 48 61 63 33 48 61 58
ΓT/mg/m2 2.5-2.8 2.2-2.5 3.1-3.9 2.7-3.0 2.0-2.3 2.0-2.3 3.1-3.9 2.3-2.6
a The properties of adsorbed β-casein layers were obtained from a four-layer model fit to data obtained from neutron reflectivity measurements (in 0.02

M imidazole-HCl buffered D2O, pH ) 7, with added electrolyte) after exposure to 0.1 mg/mL β-casein solution unless stated otherwise. τi is the layer
thickness (Å), Fi is the scattering length density, φi is the volume fraction of protein, and Γi is the surface excess (mg/m2) of layer i, where i ) 0 is the OTS
layer next to the silica surface. Subscript T stands for the sum over all layers. b Containing 0.01 mg/mL β-casein. c In H2O. dThe errors given reflect the
maximum uncertainty in the fitting and any possible coupling of the fitting parameters.

cases and would not be able to distinguish at all clearly the


functional form of the decay. We therefore opted to use a
description in terms of the minimum number of distinct layers
required to fit the data. The experiment is sensitive to the
overall thickness of the layer until the volume composition
drops below about 10%; i.e., the uncertainty in each layer
thickness increases as there is less protein in it. It is also
sensitive to the total amount adsorbed and to the more
detailed distribution within the OTS layer. On the basis of
this we have estimated the errors as shown in the tables,
and these estimates include contributions that could arise
from coupling of the fitting parameters. Our data were
analyzed by implementing a four-layer model, where the Figure 1. Neutron reflectivity versus momentum transfer curves in
0.02 M imidazole-HCl buffered D2O solution, pH 7: the curve for the
lower layer is the outer, defective d-OTS-layer. Typical clean d-OTS block (×); the layer adsorbed from a 0.1 mg/mL β-casein
scattering curves for the pure d-OTS layer at various stages solution containing 0.0167 M CaCl2 after changing solution to 0.02
of the β-casein adsorption study are given in Figure 1. M imidazole-HCl buffered D2O solution (+); the layer after treatment
The results from the fit to the experimental scattering with endoproteinase Asp-N (0.04 µg/mL) and subsequent rinsing in
buffered D2O solution (4); the layer formed after second addition of
curves are given in Tables 3-5. For the sake of clarity, the a 0.1 mg/mL β-casein solution and subsequent rinsing in buffered
evolution of thickness and the surface excess in the four D2O solution (O) are shown together with the corresponding fits, which
different layers during the different steps in the adsorption are inserted as solid lines.
experiments are illustrated by bar charts in Figure 2-5. The
thickness of the outer d-OTS layer could in all cases be fitted 4b), while the amounts in the presence of NaCl and MgCl2
to 20 Å. However, during the course of adsorption the density (Figures 2b and 5b) are smaller but similar to each other.
of this layer changed, which indicates penetration of BCN. However the total thickness in the presence of the three
The Adsorption of BCN. The adsorption data for BCN electrolytes is quite similar (Figures 4a, 2a, and 5a, respec-
in the various electrolytes could all be fitted to the four- tively). To see the effect of the protein concentration, one
layer model (Table 3). additional experiment was performed in the presence of 0.05
In every case the adsorption of BCN on the OTS surface M NaCl, with a 10-fold lower protein concentration of 0.01
reaches a plateau after the first measuring cycle, which is mg/mL. This led to a decrease in the adsorbed amount
completed about 55 min after adding the protein. Here, we (Figure 3b) as well as in the total thickness of the protein
note that the time resolution of neutron reflectivity measure- layer (Figure 3a). The adsorbed amount in the first two layers
ments is rather limited and the plateau might have been is basically the same at this lower protein concentration
reached after a significantly shorter time. The largest total (Figure 3b) as in the presence of NaCl and MgCl2 (Figures
amount adsorbed is found in the presence of CaCl2 (Figure 2b and 5b), while it is slightly higher in the presence of CaCl2
282 Biomacromolecules, Vol. 2, No. 1, 2001 Nylander et al.

Table 4. Exposure to Endoproteinase Asp-N and Subsequent Rinsinga


endoproteinase rinse
0.05 M 0.05 M 0.017 M 0.017 M 0.05 M 0.05 M 0.017 M 0.017 M
NaCl NaClb CaCl2 MgCl2 NaCl NaClb CaCl2 MgCl2
τ0 ( 3c 20 20 20 20 20 20 20 20
τ1 ( 3c 8 8 6 8 8 8 6 8
τ2 ( 5c 20 15 25 25 20 15 25 25
τ3 ( 10c
F0 4.7 4.7 3.9 4.3 4.7 4.7 3.9 4.3
F1 4.2 4.4 4.2 4.6 4.2 4.4 4.2 4.6
F2 6.2 6.1 6.0 6.1 6.2 6.1 6.0 6.1
F3
φ0 ( 0.08c 0.36-0.49 0.36-0.49 0.44-0.72 0.47-0.6 0.36-0.49 0.36-0.49 0.44-0.72 0.47-0.6
φ1 ( 0.1c 0.6 0.6 0.6 0.5 0.6 0.6 0.6 0.5
φ2 ( 0.02c 0.04 0.07 0.1 0.07 0.04 0.07 0.1 0.07
φ3
Γ0 ( 0.2c 1.0-1.3 1.0-1.3 1.2-2.0 1.3-1.6 1.0-1.3 1.0-1.3 1.2-2.0 1.3-1.6
Γ1 ( 0.1c 0.7 0.6 0.5 0.6 0.7 0.6 0.5 0.6
Γ2 ( 0.07c 0.14 0.14 0.34 0.24 0.14 0.14 0.34 0.24
Γ3
τT 28 23 31 33 28 23 31 33
ΓT/(mg/m2) 1.8-2.1 1.7-2.0 2.0-2.8 2.1-2.4 1.8-2.1 1.7-2.0 2.0-2.8 2.1-2.4
a The properties of adsorbed β-casein layers after addition of an endoproteinase Asp-N solution, followed by rinsing with buffered D O. The results
2
were obtained from a four-layer model fit to data obtained from neutron reflectivity measurements in 0.02 M imidazole-HCl buffered D2O, pH ) 7. The
β-casein layers were adsorbed from buffers containing different electrolytes and a protein concentration of 0.1 mg/mL β-casein unless stated otherwise.
τi is the layer thickness (Å), Fi is the scattering length density, φi is the volume fraction of protein, and Γi is the surface excess (mg/m2) of layer i, where
i ) 0 is the OTS layer next to the silica surface. Subscript T stands for the sum over all layers. b Containing 0.01 mg/mL β-casein. cThe errors given
reflect the maximum uncertainty in the fitting and any possible coupling of the fitting parameters.

Table 5. Second Addition of β-Casein and Subsequent Rinsinga


second β-casein addition rinse
0.05 M 0.05 M 0.017 M 0.017 M 0.05 M 0.05 M 0.017 M 0.017 M
NaCl NaClb CaCl2 MgCl2 NaCl NaClb CaCl2 MgCl2
τ 0 ( 3c 20 20 20 20 20 20 20 20
τ1 ( 3c 8 8 6 8 8 8 6 8
τ2 ( 5c 30 23 25 34 30 23 25 34
τ3 ( 10c 35 30 35 37 25 25 35 30
F0 4.2 4.1 3.8 4.3 4.2 4.1 3.8 4.3
F1 3.2 3.8 3.2 3.1 3.2 3.8 3.2 3.1
F2 5.0 5.3 5.2 4.8 5.7 5.7 5.2 4.8
F3 5.9 6.1 6.1 5.6 6.2 6.2 6.1 5.8
φ0 ( 0.08c 0.5-0.63 0.53-0.66 0.47-0.71 0.47-0.6 0.5-0.63 0.53-0.66 0.47-0.71 0.47-0.6
φ1 ( 0.1c 0.9 0.8 0.9 1.0 0.9 0.8 0.9 1.0
φ2 ( 0.05c 0.39 0.31 0.34 0.45 0.19 0.19 0.34 0.45
φ3 ( 0.03c 0.13 0.07 0.07 0.22 0.04 0.04 0.07 0.16
Γ0 ( 0.2c 1.4-1.7 1.4-1.8 1.3-1.9 1.3-1.6 1.4-1.7 1.4-1.8 1.3-1.9 1.3-1.6
Γ1 ( 0.1c 1.0 0.8 0.8 1.0 1.0 0.8 0.8 1.0
Γ2 ( 0.15c 1.6 1.0 1.2 2.1 0.8 0.6 1.2 2.1
Γ3 ( 0.1c 0.6 0.3 0.3 1.1 0.1 0.1 0.3 0.7
τT 73 61 66 79 63 56 66 72
ΓT/(mg/m2) 4.6-4.9 3.5-3.9 3.6-4.2 5.5-5.8 3.3-3.6 2.9-3.3 3.6-4.2 5.1-5.4
a The properties of endoproteinase Asp-N treated β-casein layers after (second) addition of 0.1 mg/mL β-casein solution, followed by rinsing with

buffered D2O. The results were obtained from a four-layer model fit to data obtained from neutron reflectivity measurements in 0.02 M imidazole-HCl
buffered D2O, pH ) 7. The β-casein layers were adsorbed from buffers containing different electrolytes and a protein concentration of 0.1 mg/mL β-casein
unless stated otherwise. τi is the layer thickness (Å), Fi is the scattering length density, φi is the volume fraction of protein, and Γi is the surface excess
(mg/m2) of layer i, where i ) 0 is the OTS layer next to the silica surface. Subscript T stands for the sum over all layers. b Containing 0.01 mg/mL
β-casein. c The errors given reflect the maximum uncertainty in the fitting and any possible coupling of the fitting parameters.

(Figure 4b). Here, we also note that the build-up of β-casein at the surface. The thicknesses of the three inner layers are
layer in the presence of CaCl2 was time dependent unlike in basically the same for the different electrolytes, while the
the other conditions studied. It took about 2 h until the lower protein concentration leads to a substantial decrease
constant reflectivity profile was obtained. The slower equili- in thickness of the outermost layer. However, the amounts
bration time is related to the fact that a more dense layer is adsorbed are different and below we will discuss this further
formed, which set demands on interfacial ordering and thus in terms of protein volume fraction versus distance. The
requires more extensive rearrangements of casein molecules adsorbed amount and the thickness of the two outer layers
Casein Adsorption Biomacromolecules, Vol. 2, No. 1, 2001 283

Figure 2. Properties of the adsorbed layer obtained from fitting Figure 3. Properties of the adsorbed layer obtained from fitting
neutron reflectivity data to a four-layer model, where layer no. 0 is neutron reflectivity data to a four-layer model, where layer no. 0 is
closest to the surface (mixed d-OTS-β-casein layer) (solid black) and closest to the surface (mixed d-OTS-β-casein layer) (solid black) and
layer 1 (hatched lines with narrow spacing), layer 2 (hatched lines layer 1 (hatched lines with narrow spacing), layer 2 (hatched lines
with broad spacing), and layer 3 (light gray) are indicated in the bar with broad spacing), and layer 3 (light gray) are indicated in the bar
charts. The extension of the bars indicates the (a) thickness of the charts. The extension of the bars indicates the (a) thickness of the
different layers, τ, and (b) the amount of β-casein adsorbed, Γ. different layers, τ, and (b) the amount of β-casein adsorbed, Γ. The
β-Casein was adsorbed from a solution containing 0.1 mg/mL protein same experimental procedure was used as the one used to obtain
in 0.02 M imidazole-HCl and 0.05 M NaCl in D2O at pH 7 (BC), the data in Figure 2, except that the β-casein concentration was 0.01
followed by rinse with 0.02 M imidazole-HCl in D2O at pH 7 (BCr), mg/mL protein.
addition of endoproteinase Asp-N (0.04 µg/mL 0.02 M imidazole-HCl
in D2O at pH 7) (E), and rinse with 0.02 M imidazole-HCl in D2O at
pH 7 (Er). A second addition of β-casein (0.1 mg/mL in 0.02 M in amount by a third of the total mass and by half the layer
imidazole-HCl and in D2O at pH 7) was then made (2BC), followed thickness is observed for β-casein adsorbed from a solution
by rinse with 0.02 M imidazole-HCl in D2O at pH 7 (2BCr).
containing CaCl2 (Figure 4). The reduction in the total
amount is only about 10% for the other conditions, although
are similar in the presence of both divalent electrolytes
the reduction in total thickness was larger. It should be
(Figures 4 and 5), while it is substantially less in the presence
emphasized that the sample cell was rinsed with the same
of NaCl (Figure 2). A further decrease is observed if the
buffer, that is, buffer without added electrolyte, before the
concentration of BCN is lowered (Figure 3). Thus, it appears
endoproteinase Asp-N dissolved in 0.02 M imidazole buffer
as if the largest effect of changing electrolyte and protein
was added. This was done to make sure that enzyme activity
concentration is in the outer parts of the adsorbed layer.
was not affected by the added electrolyte ensuring that the
However, the picture is different in terms of protein volume
enzymatic reaction was carried out in equal conditions.
fraction (see Discussion). Rinsing with pure buffer without
Rinsing with enzyme free buffer without added electrolyte
added electrolyte does not have any effect on the layer
had no effect on the enzyme-treated layer.
adsorbed from the protein solution containing CaCl2. For all
other conditions a slight decrease of the adsorbed amount is The Effect of a Second Addition of β-Casein. To assess
observed in the outer layer only. In all cases very small further the effect of the enzyme treatment, and hence the
changes are observed in the thickness of the two outer layers. structure of the adsorbed layer, a second addition of β-casein
There is however one exception and that is for the presence to the enzyme-treated layer was made. The protein was added
of NaCl, where the outermost layer seems to disappear. from buffer solution without added electrolyte. In every case
However, it should be emphasized that there is very little the second addition led to large increases in amounts
protein in the outermost layer and the accuracy in judging adsorbed and in the thickness of the adsorbed layers. In fact,
the amount in such a layer is low. the layer became significantly thicker and contained signifi-
The Effect of Adding Endoproteinase AspN. For all the cantly larger amounts of β-casein than the layer initially
added electrolytes, endoproteinase Asp N only affects the formed on the d-OTS, hydrophobic surface (Table 5).
outermost β-casein layer (Table 4). In fact the data could The largest increase was observed in the presence of the
now be fitted to a three-layer model. The largest reduction divalent electrolyte, MgCl2 (Figure 5). Here the second
284 Biomacromolecules, Vol. 2, No. 1, 2001 Nylander et al.

Figure 4. Properties of the adsorbed layer obtained from fitting Figure 5. Properties of the adsorbed layer obtained from fitting
neutron reflectivity data to a four-layer model, where layer no. 0 is neutron reflectivity data to a four-layer model, where layer no. 0 is
closest to the surface (mixed d-OTS-β-casein layer) (solid black) and closest to the surface (mixed d-OTS-β-casein layer) (solid black) and
layer 1 (hatched lines with narrow spacing), layer 2 (hatched lines layer 1 (hatched lines with narrow spacing), layer 2 (hatched lines
with broad spacing), and layer 3 (light gray) are indicated in the bar with broad spacing), and layer 3 (light gray) are indicated in the bar
charts. The extension of the bars indicate the (a) thickness of the charts. The extension of the bars indicate the (a) thickness of the
different layers, τ, and (b) the amount of β-casein adsorbed, Γ. The different layers, τ, and (b) the amount of β-casein adsorbed, Γ. The
same experimental procedure was used as the one used to obtain same experimental procedure was used as the one used to obtain
the data in Figure 2, except that 0.1 mg/mL β-casein in 0.02 M the data in Figure 2, except that 0.1 mg/mL β-casein in 0.02 M
imidazole-HCl and 0.0167 M CaCl2 in H2O at pH 7 were used in the imidazole-HCl and 0.0167 M MgCl2 in D2O at pH 7 were used in the
first step (BC). first step (BC).

addition almost doubled the adsorbed amount, and a small


portion of this, about 8%, could be removed by rinsing with of β-casein on a d-OTS surface.12 Although our data are
buffer. Adsorption from a solution containing NaCl of the qualitatively almost identical to theirs, we have used a four-
same ionic strength also gives quite large amounts adsorbed layer model in this study. The reason is that we wanted to
upon the second addition of β-casein (Figure 2), but now make a fair comparison with our neutron reflective study of
about 28% of the mass was removed when rinsing with pure β-casein adsorption on a hydrophilic surface.32 In that study
buffer. When the protein concentration is lowered 10-fold, it turned out that the quality of fit increased significantly
the adsorption at the second addition is also lower (Figure when increasing the number of layers from two to three. Due
3) and about 15% of the amount could be desorbed. The to the presence of an imperfect outer layer of d-OTS, an
lowest increase in adsorbed amount on second addition of extra layer was added in the model to account for penetration
β-casein is observed for the enzyme-treated β-casein layer, of β-casein into this layer. The volume fraction of β-casein
adsorbed in the presence CaCl2. No desorption upon subse- segments versus distance from the crystalline inner d-OTS
quent rinsing with protein-free buffer could be observed. layer is plotted for different experimental conditions in Figure
Here we note that the largest reduction in adsorbed amount 6.
upon endoproteinase Asp-N addition was observed on the The data are taken from Table 3 and the figure plots the
layer adsorbed from CaCl2 solution. In all cases, apart from volume fraction at the center of each layer at the corre-
adsorption to initially adsorbed layer formed in the presence sponding distance. The volume fraction of β-casein in the
of NaCl, the second addition only causes changes in the three outer d-OTS layer has been calculated as the volume fraction
outermost layers. In the case of NaCl the volume fraction remaining after subtraction of that of the OTS. In any case,
of the β-casein in the innermost layer, φ0, also increases from the volume fraction of β-casein in this layer is substantially
about 0.6 to 0.9 or 0.8 at the lower β-casein concentration. smaller than that in the innermost pure β-casein layer, and
this causes the maximum in Figure 6. This is expected, as
the presence of d-OTS islands is likely to make packing of
Discussion
β-casein at the surface less efficient. The most obvious
In an earlier neutron reflectivity study Fragneto et al. used difference between the results for β-casein adsorption on the
a two-layer model to evaluate their data on the adsorption hydrophobic surface compared with the corresponding data
Casein Adsorption Biomacromolecules, Vol. 2, No. 1, 2001 285

and whether it gives just an average thickness, as for instance,


ellipsometry. Although segment density profiles can, in
principle, be calculated from neutron reflectivity, the models
usually employed consist of just a small number of layers.
The resolution is not sufficient to distinguish stepwise and
smooth distributions.
Leermakers et al.13,14 used self-consistent field modeling
to estimate the segment density profiles of adsorbed layers
of β-casein on hydrophobic surfaces. The total segment
density profile, φ(z), where z is the distance from the
hydrophobic surface, was found to fall off in a featureless
fashion from very high values (≈0.95) close to the interface
Figure 6. Volume fraction of adsorbed layers of β-casein obtained (small z) to values approaching the protein bulk concentration
from neutron reflectivity measurements plotted versus the distance for z > 50 Å. A very dilute tail region (φ(z) < 0.01),
from the surface. The data are given in Table 3, and the distance
corresponding to the N-terminal hydrophilic sequence of ≈40
values are taken as half the thickness of the layer plus the total
thickness of inner layers. Neutron reflectivity data from β-casein amino acid residues, was found to extend out from z ) 30-
adsorption on a hydrophilic surface is also inserted (0).32 The β-casein 70 Å to z ≈ 200 Å, depending on pH, ionic strength, and
concentration was 0.1 mg/mL 0.02 M imidazole-HCl at pH 7, unless protein bulk concentration. A typical result for the model
stated otherwise. The data were recorded in the presence of added
electrolyte: 0.05 M NaCl (2); 0.05 M NaCl and 0.01 mg/mL β-casein
used to represent β-casein on a hydrophobic surface at pH
(4); 0.0167 M CaCl2 (b); 0.00167 M MgCl2 ([). For comparison, the 7, taken from Atkinson et al.,13 is inserted in Figure 6 as a
total segment density profile versus distance for β-casein look alike solid line, where their data have been shifted 20 Å to allow
at pH 7 on a hydrophobic surface calculated by the self-consistent- for the d-OTS layer. This is to account for the 20 Å thick
field theory by Atkinson et al.13 is introduced as a solid line. It should
be noted that for fair comparison their data have been shifted 20 Å. incomplete layer of d-OTS. Their calculated curve agrees
This is to account for the 20 Å thick incomplete layer of d-OTS. quite well with our experimental data suggesting that the
real β-casein layer conforms well to the model protein and
for the hydrophilic surface is the very different distribution that the self-consistent field modeling satisfactorily accounts
of the volume fraction versus distance. Close to the surface, for the protein segment density profile.
the density of β-casein segments seems to be less or as dense The volume fraction versus distance profiles have the same
on the hydrophilic as on the hydrophobic surface. Although shape whether β-casein is adsorbed in the presence of NaCl,
there are a small number of data points, making comparison CaCl2, and MgCl2 of the same ionic strength or whether the
uncertain, the volume fraction on the hydrophilic surface protein concentration is lowered 10 times. In particular at
seems to decrease more or less linearly with distance whereas larger distances from the surface, the volume fraction values
a more drastic decrease at about 40 Å from the inner d-OTS are low and similar. This is also reflected in the fact that
layer is observed for the hydrophobic surface. The addition endoproteinase Asp-N reduces the outer part of the adsorbed
of endoproteinase Asp-N leads to a relatively smaller layer in all cases. The low volume fraction of protein
reduction in the amount and layer thickness of β-casein on segments in the outer layer makes it possible for the enzyme
the hydrophobic (present study) than on the hydrophilic to penetrate and cleave at the N-terminal hydrophilic part
surface.32 Similar findings were found in our previous of β-casein, residues 43 and/or 47. However, we note that
ellipsometry study.15 This indicates the difference structure the profile recorded at the lower β-casein concentration, 0.01
of the adsorbed layer, which also is indicated by the much mg/mL, is less extended. If we ignore the innermost mixed
slower adsorption kinetics on the hydrophilic surface.15,32 The β-casein and d-OTS layer, it is obvious that the largest
total thickness of the adsorbed layer also seems to be larger difference between results from the experiments carried out
on the hydrophilic surface than on a hydrophobic surface under different conditions is in the region close to the
(Figure 6), which was also observed in our earlier study hydrophobic surface. The volume fraction in this layer
where the forces between two β-casein covered surfaces were decreases in the order CaCl2 > MgCl2 > NaCl at constant
determined by using the interferometric surface force ap- ionic strength. This confirms the findings in our previous
paratus (SFA).16 However, in contrast to the SFA and neutron ellipsometry study, where the adsorbed amounts were found
reflectivity measurements, in the ellipsometry study we did to decrease in this order, while the layer thickness remained
not observe any difference in β-casein thickness between constant. It should be borne in mind that the binding of both
the hydrophobic and hydrophilic surfaces.15 Also the thick- calcium and magnesium ions to the five serine phosphate
ness values determined by ellipsometry were somewhat lower residues on β-casein is similar and of high affinity, as
than those observed in the present study and in previous determined by multinuclear magnetic resonance spectros-
neutron scattering and reflectivity studies.6,11,12 Even larger copy.26 Thus the large molar electrolyte-β-casein ratio of
thicknesses were observed in the SFA studies, ≈110-125 about 800:1, used in the present study, suggests that all these
Å,16,18 light scattering studies of adsorbed β-casein layers binding sites on the protein are saturated.
on emulsion droplets and particles, ≈100-120 Å,5-7,9 and In contrast to our study, Brooksbank et al.,9 Atkinson et
thickness of emulsion films as determined by microinter- al. 11 and Velev et al. 18 found that the β-casein layer
ferometry, ≈110 Å.18 These discrepancies have to do with thickness decreases in the presence of calcium. However in
how well each technique responds to low segment densities all of these studies different experimental conditions were
286 Biomacromolecules, Vol. 2, No. 1, 2001 Nylander et al.

used compared to those employed in the present study, which negative charge on the surfaces and hence favor formation
might have resulted in different packing on the hydrophobic of the second layer. The relative large reduction by endopro-
surface. In the study of Brooksbank et al., the calcium ions teinase Asp-N of the β-casein layer adsorbed in the presence
were added after adsorbing β-casein to the “hydrophobic” of calcium might lead to relative larger loss of bound calcium
latex particles. Atkinson et al. studied the adsorption at the ions from the adsorbed layer. Hence, we did not observe
air/water interface for which they reported a more compact the same effect of second addition of β-casein to an
layer, that is, thinner and denser, compared with the results endoproteinase Asp-N treated protein layer formed in the
in our study. They reported a significant decrease of the presence of calcium compared to the one formed in the
amount of β-casein adsorbed when adding 2.0 mM calcium presence of magnesium. The large adsorption during the
to a solution containing 0.05 mg/mL β-casein. However, it addition of β-casein after cleaving portions of the N-terminal
should be borne in mind that they studied the adsorption at part shows that this part is essential for β-casein to act as a
a liquid interface. This means that the lateral mobility is protective colloid. This confirms the findings from our earlier
considerably higher than at the solid-liquid interface. study, where we demonstrated the same behavior by addition
Although, the formation elastic protein networks at the air/ of β-lactoglobulin to an endoproteinase Asp-N treated
water has been reported, these are much weaker for a β-casein surface.10
β-casein than a globular protein like β-lactoglobulin.33 Thus,
by proper orientation of the serine phosphate moieties the
number of intermolecular calcium bridges can be increased. Acknowledgment. The financial support from the Swed-
Such a reorientation can, if it means that β-casein protrudes ish Research Council for Engineering Sciences (TFR) is
less into the subphase, lead to a lateral expansion of the layer, gratefully acknowledged.
which in turn means an increase in area per molecule. In
contrast to the findings of Brooksbank et al.9 and Atkinson References and Notes
et al.,11 Velev et al.18 found that the adsorbed amount of (1) Walstra, P.; Jenness, R. Dairy Chemistry and Physics; Wiley-
β-casein at the oil-water interface did not seem to be Interscience: New York, 1984.
affected by the presence of calcium. They measured their (2) Friberg, S.; Larsson, K. Food emulsions, 3 ed.; Marcel Dekker, Inc.:
New York, 1997.
thickness with different surface force techniques, for which (3) Dickinson, E. Int. Dairy J. 1999, 9, 305- -312.
the value of the thickness may be sensitive to the compress- (4) Eigel, W. N.; Butler, J. E.; Ernstrom, C. A.; Farrell, H. M., Jr.;
ibility of the layer. They found a decrease of thickness with Harwalkar, V. R.; Whitney, R. M. J. Dairy Sci. 1984, 67, 1599.
increasing calcium ion concentration until 12 mM, after (5) Dalgleish, D. G.; Leaver, J. J. Colloid Interface Sci. 1991, 141, 288-
294.
which no further decrease of the thickness was observed. (6) Mackie, A. R.; Mingins, J.; North, A. N. J. Chem. Soc., Faraday
Interestingly, above this concentration they observed a strong Trans. 1991, 87, 3043-3049.
adhesive force from which they concluded that cross-binding (7) Leaver, J.; Dagleish, D. G. J. Colloid Interface Sci. 1992, 149, 49-
55.
of the film surfaces occurs. It should also be noted that they (8) Dickinson, E.; Horne, D. S.; Phipps, J. S.; Richardson, R. M.
used a slightly lower pH of 6.5 in their study. One should Langmuir 1993, 9, 242-248.
also be aware that calcium can cause precipitation (formation (9) Brooksbank, D. V.; Davidson, C. M.; Horne, D. S.; Leaver, J. J.
of calcium bridges) even in imidazole buffer under conditions Chem. Soc., Faraday Trans. 1993, 89, 3419-3425.
(10) Nylander, T.; Wahlgren, N. M. J. Colloid Interface Sci. 1994, 162,
where magnesium does not.26,34,35 This could also explain 151-162.
why calcium gives a higher volume fraction in the first layer (11) Atkinson, P. J.; Dickinson, E.; Horne, D. S.; Richardson, R. M. J.
than magnesium. In addition we note that, although the time Chem. Soc., Faraday Trans. 1995, 91, 2847-2854.
(12) Fragneto, G.; Thomas, R. K.; Rennie, A. R.; Penfold, J. Science 1995,
resolution of the neutron reflection measurements is poor, 267, 657-660.
the adsorption in the presence of calcium seemed to be (13) Atkinson, P. J.; Dickinson, E.; Horne, D. S.; Leermakers, F. A. M.;
slightly time dependent, which might indicate some kind of Richardson, R. M. Ber. Bunsen-Ges. Phys. Chem. 1996, 100, 994-
rearrangement, possibly cross-linking. It is noteworthy that 998.
(14) Leermakers, F. A. M.; Atkinson, P. J.; Dickinson, E.; Horne, D. S.
the largest reduction in adsorbed amount upon endoproteinase J. Colloid Interface Sci. 1996, 178, 681-693.
Asp-N addition was observed on the layer adsorbed from (15) Kull, T.; Nylander, T.; Tiberg, F.; Wahlgren, M. Langmuir 1997,
CaCl2 solution. Cross-linking of the released N-terminal 13, 5141-5147.
(16) Nylander, T.; Wahlgren, N. M. Langmuir 1997, 13, 6219-6225.
peptide, which, in turn might precipitate in the bulk solution, (17) Dickinson, E.; Pinfield, V. J.; Horne, D. S.; Leermakers, F. A. M. J.
could facilitate this proteolytic reaction. The second addition Chem. Soc., Faraday Trans. 1997, 93, 1785-1790.
of β-casein to the enzyme-treated β-casein layer seems to (18) Velev, O. D.; Campbell, B. E.; Borwankar, R. P. Langmuir 1998,
14, 4122-4130.
restore the β-casein layer because no desorption upon
(19) Murray, B. S.; Dickinson, E.; McCarney, J. M.; Nelson, P. V.;
subsequent rinsing with protein-free buffer could be ob- Whittle, M. Langmuir 1998, 14, 3466-3469.
served. Such behavior suggests a strong interaction with the (20) Nylander, T.; Tiberg, F. Colloids Surf., B 1999, 15, 253-261.
remaining layer or with small uncovered regions on the (21) Schmidt, D. G.; Payens, T. A. J. J. Colloid Interface Sci. 1972, 39,
655-662.
hydrophobic surface. The very large increase observed on (22) Leclerc, E.; Calmettes, P. Phys. ReV. Lett. 1997, 78, 150-153.
the second addition of β-casein to an enzyme treated layer (23) Carles, C.; Huet, J. V.; Ribadeau-Dumas, B. FEBS Lett. 1988, 229,
adsorbed from a solution containing MgCl2 implies that 265-272.
(24) Fleer, G. J.; Cohen Stuart, M. A.; Scheutjens, J. M. H. M.; Cosgrove,
electrostatic forces somehow control the formation of the
T.; Vincent, B. Polymers at interfaces; Chapman and Hall: London,
second layer. It might be that some magnesium ions remain 1993.
at the surface, which might reduce or even reverse the (25) Dickson, I. R.; Perkins, D. J. Biochem. J. 1971, 124, 235-240.
Casein Adsorption Biomacromolecules, Vol. 2, No. 1, 2001 287

(26) Wahlgren, N. M.; Dejmek, P.; Drakenberg, T. J. Dairy Res. 1993, (31) Lu, J. R.; Thomas, R. K. The application of neutron and X-ray
60, 65-78. specular reflection to proteins at interfaces; Baszkin, A., Norde, W.,
(27) Drapeau, G. R. J. Biol. Chem. 1980, 255, 839-840. Ed.; Marcel Dekker: New York, 2000; pp 609-650.
(28) Fragneto, G.; Lu, J. R.; McDermott, D. C.; Thomas, R. K. Langmuir (32) Tiberg, F.; Nylander, T.; Su, T. J.; Lu, J. R.; Thomas, R. K.
1996, 12, 477-486. Manuscript in preparation.
(29) Penfold, J.; Richardson, R. M.; Zarbakhsh, A.; Webster, J. R. P.; (33) Mackie, A. R.; Gunning, A. P.; Wilde, P. J.; Morris, V. J. J. Colloid
Bucknall, D. G.; Rennie, A. R.; Jones, R. A. L.; Cosgrove, T.;
Interface Sci. 1999, 210, 157-166.
Thomas, R. K.; Higgins, J. S.; Fletcher, P. D. I.; Dickinson, E.; Roser,
(34) Parker, T. G.; Dalgleish, D. G. J. Dairy Res. 1981, 48, 71-.
S. J.; McLure, I. A.; Hillman, A. R.; Richards, R. W.; Staples, E. J.;
Burgess, A. N.; Simister, E. A.; White, J. W. J. Chem. Soc., Faraday (35) Wahlgren, N. M. A Nuclear Magnetic Resonance Approach to the
Trans. 1997, 93, 3899-3917. Milk System; University of Lund: Lund, Sweden, 1992.
(30) Lu, J. R.; Thomas, R. K. J. Chem. Soc., Faraday Trans. 1998, 94,
995-1018. BM0056308

You might also like