- Email: [email protected]
Association behavior of β-casein
Journal of Colloid and Interface Science 258 (2003) 33–39 www.elsevier.com/locate/jcis
Association behavior of β-casein J.E. O’Connell,a,b V.Ya. Grinberg,c and C.G. de Kruif a,d,∗ a NIZO Food Research, 6710 BA Ede, The Netherlands b Food Chemistry Department, University College, Cork, Ireland c Institute of Biochemical Physics, Russian Academy of Sciences, Vavilov St. 28, 117813 Moscow GSP-1, Russia d Van’f Hoff Laboratory, Debye Institute, University of Utrecht, Utrecht, The Netherlands
Received 19 November 2001; accepted 11 October 2002
Abstract The association behavior of β-casein, a protein with a distinct amphipathic character, was studied. β-Casein exhibits markedly temperaturedependent association behavior; at low temperatures (< 10–15 ◦ C), monomers predominate, but as the temperature is increased, monomers associate, via hydrophobic bonding, into micelles. β-Casein micelles have a hydrodynamic radius of ∼12 nm, a radius of gyration of ∼8.3 nm, and an interaction radius of ∼15 nm. These data are fully consistent with a pervious fluffy particle. The association behavior of β-casein is also strongly affected by concentration and solvent quality. At low concentrations β-casein exhibits a critical micelle concentration (CMC) of approximately 0.05%, w/v, at 40 ◦ C. In the presence of 6 M urea the temperature dependence of β-casein’s association behavior is eliminated, leaving monomers predominantly. Temperature-dependent transformations in micelle morphology can be explained by changes in solvent quality, i.e., the temperature–protein hydrophobicity and temperature–voluminosity profiles of β-casein. The results obtained are consistent with the shell model as developed by Kegeles, in which a distribution of micelle sizes is formed. They contrast with the traditional description of the micellization of β-casein by a two-state model or by the closed-association model, i.e., monomers ⇔ micelles. 2003 Elsevier Science (USA). All rights reserved.
1. Introduction
αs2 -casein exhibit consecutive association equilibria, i.e., (αs2 -casein)i + αs2 -casein ↔ (αs2 -casein)i+1 ,
The caseins, which exist in bovine milk as large macromolecular (108 Da, ∼200-nm diameter) associates in colloidal dispersion, known as casein micelles. The four casein fractions differ markedly in many respects, including charge (αs1 - > αs2 - > β- > κ-casein), hydrophobicity (β- > κ> αs1 - > αs2 -casein), and calcium sensitivity (αs2 - > αs1 > β- > κ-casein) [1]. The secondary structure of the caseins also differ considerably. Protein charge is uniformly distributed in αs1 - and αs2 -casein, while β- and κ-casein have well defined polar and apolar regions and thus their secondary structure has a distinctly amphipathic nature. At the natural pH of milk (pH 6.6) the hydrophilic N-terminal of β-casein (amino acids 1 to 21) is highly charged (net charge of −11.5), while the hydrophobic C-terminal has little if any net charge. Consequently, the association behavior of the caseins and the factors affecting it differ considerably. αs1 - and
which are markedly affected by ionic strength and less so by temperature. Traditionally, the micellization of κ- and βcasein is described as exhibiting monomer–micelle association equilibria of the “all-or-nothing”, or closed-association type [2]; i.e., n monomers ↔ micelle, which is strongly affected by temperature and less so by ionic strength. The micellization number, n, is supposedly fixed, n ∼ 20–50. However, Mikheeva et al. [3] showed that the micellization of β-casein can be described quite adequately using the so-called shell model of Kegeles [4,5], in which a distribution of micelle sizes is formed centered around n/2, where n is the maximum (due to steric constraints) micellization number. Thus: β-cas0 + β-cas0 ↔ β-cas1 ,
* Corresponding author.
E-mail address: [email protected] (C.G. de Kruif).
β-cas0 + β-cas1 ↔ β-cas3 , ···
0021-9797/03/$ – see front matter 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0021-9797(02)00066-8
34
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
β-cas0 + β-casi ↔ β-casi+1 , ··· β-cas0 + β-casn−1 ↔ β-casn . Our results, described here, confirm that the shell model describes the micellization of β-casein. For instance, the shell model accounts naturally for the observed polydispersity in micelle size. The fact that β-casein is a major constituent of casein micelles and is also commonly used as a foaming and emulsifying agent means that its association behavior is of importance in the food industry. Centrifugal sedimentation, calorimetry, size-exclusion chromatography, electron microscopy, and light scattering techniques have been used to study the effects of environmental conditions (i.e., ionic strength, temperature, and concentration) on the micellization of β-casein and also on the characteristics of micelles (i.e., shape, size distribution, associative interactions [6–17]). However, the relationship between micelle morphology, temperature, and electrostatic and hydrophobic interactions has yet to be fully elucidated. This communication focuses on the characterization of β-casein micelles. The significance of electrostatic and hydrophobic interactions in micelles was studied using lightscattering techniques.
2. Materials and methods 2.1. Preparation of β-casein High-purity β-casein was obtained from Eurial (Rennes, France). A stock solution (50 g l−1 in 25 mM sodium phosphate, pH 6.5; Merck, D64271, Darmstadt, Germany) was prepared and exhaustively dialyzed (100 ml against 2 × 2 l of 25 mM sodium phosphate, pH 6.5, for 72 h at 4 ◦ C) using dialysis tubing with a nominal molecular weight cutoff of 12 to 14 kDa (T99-T108, Medicell International Ltd., 239 Liverpool Rd., London, N1 ILX, UK). Sodium azide (4%, w/v; Sigma Chemical Co., St. Louis, MO 63103, USA) was added to a final concentration of 0.02%, w/v. 2.2. Dynamic light scattering (DLS) DLS was performed using a Spectra Physics 275-mW Ar laser (λ 514.5 nm). Measuring times varied with the intensity of scattered light but were carried out long enough to collect at least 107 photons. Scattered light was detected, at 90◦ , by a photomultiplier, which was interfaced to an ALV 5000 digital correlator. Using the ALV 5000 correlator an autocorrelation function, G(2) (Q, τ ), was determined [18,19], G(2) (Q, τ ) = I (Q, t)I (Q, t + τ ) , (1) where I (Q, t) is the intensity at a wave vector, Q, and time, t, and τ is the interval time of the correlator. As τ approaches
Fig. 1. (A) Calculated intensity of autocorrelation function for homogeneous particles with a diameter of 22 nm and a size polydispersity β = 0.4 in a lognormal distribution, corresponding to about 17% standard deviation in a Gaussian distribution. The line through the points is a fit with a sum of two exponentials with about equal amplitudes. (B) Amplitude, A1 , of β-casein micelles (•) and A2 (◦) from the fit of experimental DLS data to a sum of two exponentials. (C) Ratio of DLS amplitudes A1 /A2 of 0.5%, w/v, β-casein in 25 mM sodium phosphate, pH 6.5
infinity G(2) (Q, τ ) equals I 2 and we define G(2)(τ ) = G(2) (Q, τ ) G(2)(Q, τ∞ ) .
(2)
From G(2) (τ ) the diffusion coefficient was calculated, which in turn can be used to determine the hydrodynamic radius using the Stokes–Einstein equation, RH = kT /6πηD,
(3)
where k is the Boltzmann constant, T is the temperature (Kelvin), η is the solvent viscosity, and D is the diffusion coefficient. DLS experiments were performed as outlined above and the data obtained were fitted to a sum of two exponentials, resulting in small residuals. Numerical simulation of a polydisperse scattering sample showed that the correlation function can be represented quite well by the sum of two exponentials with amplitudes A1 and A2 . We present such a calculation in Fig. 1A. The two-exponential fit of the simulated data is excellent and superimposes exactly the calculated correlation function. Residuals were smaller than 0.001 for all points. Fitting the data to a cumulant resulted in residuals that were, on average five times larger, especially at short times.
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
The fast mode, in fitting the experimental data, gives the hydrodynamic radius of β-casein micelles. It should be emphasized that attributing the “slow” mode to a second type of particles (as would have occurred in software provided with experimental setups) would have led to nonmeaningful results. The “slow” mode gives an apparent particle size of 80–100 nm. Such particles were, however, not “visible” in static scattering techniques. Therefore, we attribute the “slow” mode to polydispersity. As temperature increases β-casein progressively forms more monodisperse micelles, so that the amplitude for the slow mode decreases. The apparent mean size of the micelles remains remarkably constant, which is consistent with the shell model. The sum of the amplitudes, A1 and A2 , equals a constant under all experimental conditions. Theoretically this value should be 1 but in our case it was 0.62, but this depended on the choice of the pinholes used. It must be mentioned that when the data are fitted to the cumulant the result depends on the choice of τ and the time window of the correlation function, which is limited in early DLS experiments [13]. The double exponential fit appeared to give consistent data in terms of particle diffusion coefficient. Based on the above it is concluded that the micellization of β-casein as affected by temperature, urea, and concentration can be followed.
35
2.4. Surface hydrophobicity The surface hydrophobicity of β-casein was determined as outlined by O’Connell and Fox [21]. To 1.9 ml of βcasein solution (5 g l−1 in 25 mM sodium phosphate, pH 6.5) 0.1 ml of 2.5 mM 8-anilino-1-naphthalene sulfonic acid (ANS; Sigma Chemical Co.) in 25 mM sodium phosphate, pH 6.5, was added and vortexed and the fluorescence was measured at excitation and emission wavelengths of 390 and 480 nm, respectively. A standard curve of β-casein concentration vs fluorescence was constructed to ensure that sufficient ANS was present. The hydrophobicity of β-casein at different temperatures is expressed relative to β-casein at 25 ◦ C, which was attributed a value of 1. 2.5. Viscosity The viscosity of β-casein solutions was measured within the temperature range 0 to 40 ◦ C. Viscosity measurements were taken within the concentration range 0.5 to 12.5%, w/v, at a shear rate of 0.1498 s−1 using a low-shear viscometer (Contraves Low Shear-30, Contraves AG, Zurich, Switzerland).
3. Results and discussion 2.3. Small angle neutron scattering (SANS) SANS experiments were performed at 40 ◦ C on β-casein solutions, 5 to 20 g l−1 . The D-22 spectrometer supplied with cold thermal neutrons emitted from the core of the high-flux nuclear reactor at the Institute Max Von Laue–Paul Langevin in Grenoble, France, was used. The sample path length was 2 mm and the mean wavelength of the emitted neutrons was 1.0 nm with a width at half height of 9%. In SANS experiments the normalized scattering intensity I (Q), which depends on the wave vector Q, is measured. The wave vector is defined as 4π sin(θ/2)/λ, where θ is the angle at which the scattered intensity is detected and λ is the neutron wavelength. For homodisperse solutions I (Q) is related to the structure factor S(Q) and the scattering particle form factor P (Q) by [20] I (Q) = KcMP (Q)S(Q),
(4)
where c is the particle mass concentration, M is the molar mass, and K is the material constant, which depends on the optical constant and wavelength. From the I (Q) and assuming that the structure factor at 5 g/l equals 1 the structure factor at a given concentration was calculated by applying the equation S(Q)2 = I (Q)2 c1 /I (Q)1 c2 ,
(5)
which follows from Eq. (4) and where the subscript “1” refers to the sample at 5 g l−1 , the “2” to the other samples, and c equals the protein concentration. All curves were normalized at high Q, where S(Q) approaches unity.
3.1. Association behavior of β-casein: CMC, influence of temperature and urea Because of the distinct amphipathic character of βcasein, i.e., the presence of a highly hydrophobic C-region, a reduction in solvent quality (with increasing temperature) causes β-casein to associate into soap-like micelles with a hydrophobic core and a soft pervious exterior. When the available data on the association behavior of β-casein are considered a model emerges in which β-casein undergoes a temperature-induced concentration-dependent micellization process, with hydrophobic bonding being the principal associative force, while electrostatic and steric repulsive forces are responsible for ensuring micellar stability, i.e., that the micelles do not aggregate [6–14]. Micellization of β-casein is invariably described as a bimodal or two-state system, n A ↔ An , with the micelle and monomer in an equilibrium which is governed by the mass-action effect. There are strong indications that this closed-association model is not valid. In the current study the shell model of Kegeles [4,5] is applied to explain the association behavior of β-casein. As in the closed-association model, the shell model also assumes a two-state system, i.e., monomers and micelles. However, the shell model accounts for a distribution in micellar size, which the closed-association model does not. As outlined above, β-casein exhibits distinct amphiphilic properties and forms soap-like micelles. Hydrophobic bonding is the principal interactive force responsible for the
36
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
Fig. 2. (A) DLS amplitude A1 of β-casein micelles (•) and amplitude A2 (◦) and (B) amplitude ratio A1 /A2 of 0.5%, w/v, β-casein in 25 mM sodium phosphate, pH 6.5, at 40 ◦ C under the influence of increasing amounts of urea.
endothermic self-assembly of β-casein into micelles, and thus micellization is markedly temperature-dependent. In Figs. 1B and C the result of fitting the DLS data to a sum of two exponentials is plotted. The amplitude A1 is a measure of the number of micelles, while A2 is a measure of the polydispersity. It is shown that the amplitude of the βcasein micelles (A1 ) increases and that A2 decreases with increasing temperature. We attributed the slow mode to “polydispersity,” which is basically equivalent to the suggestion made by Leclerc and Calmettes [16,17], who attributed negative values of the second virial coefficient to relatively longrange correlation fluctuations. “These fluctuations might be regarded as appearing or disappearing aggregates whose structure is looser as full grown micelles.” Indeed, such a phenomenon would show up in DLS as a slow mode. In addition, with increasing temperature the amplitude of the slow mode decreased, but the scattering intensity decreased. This observation would exclude the presence of “physical particles,” because in that case intensity would decrease as well. It is noteworthy that A1 and A2 are relative amplitudes whose sum is an apparatus constant ≈ 0.62. Our interpretation of Fig. 1 is that the number of micelles increases and that polydispersity decreases as temperature is increased. The addition of urea to β-casein at 40 ◦ C monomerizes native β-casein, which confirms that hydrophobic bonding is the principal interactive force responsible for the micellization process (Figs. 2A and B). A critical micelle concentration (CMC) is also consistent with the shell model of Kegeles [4,5]. Within the limit of high cooperativity the CMC behaves in a very similar way to the two-state micelle model. In Fig. 3A we plot the
Fig. 3. (A) Critical micelle concentration at 40 ◦ C. DLS amplitude A1 of β-casein micelles (•) and A2 (◦) at 40 ◦ C, as a function of β-casein concentration. (B) Critical micelle temperature. Scattering amplitude A1 of 0.125 (•), 0.5 (◦), 1.25 (), or 2.5 ()%, w/v, β-casein solutions in sodium phosphate as a function of temperature.
amplitude A1 as a function of the β-casein concentration. In the preceding experiments the association of 0.5% βcasein was studied. Increasing the concentration (0.125 to 2.5%, w/v) decreased the association temperature of βcasein (Fig. 3A). The association behavior of very dilute βcasein solutions was also studied and, as shown in Fig. 3B, β-casein exhibits a critical micelle concentration (CMC) at approximately 0.05%, w/v, at 40 ◦ C. This CMC is a value intermediate to the CMCs of 0.03, 0.05 and 0.07%, w/v, reported by Schmidt and Payens [8], Tai and Kegeles [7], and Takase et al. [22], respectively. 3.2. Size and morphology of β-casein micelles The size and morphology of β-casein micelles at 40 ◦ C, where cohesive interactions strongly promote micellization, should be confined to within close parameters. It would be assumed that β-casein micelles are restricted to a finite size, considering that there must approach a point where charged polar heads of the N-terminal region of the molecules prevent the addition of more monomers and that the surface diameter is limited to not more than twice the chain length. Within the shell model the maximum number of monomers is set as an adjustable parameter. The calorimetric measurements of Mikheeva et al. [3] reported a molecular mass at 35 ◦ C (1%) equivalent to about 25 monomers. We found the same value from light-scattering experiments. Several studies reported an increase in either the hydrodynamic radius [11,13] or the radius of gyration [14] of β-casein micelles as a function of temperature. The results of the current study are similar to those of these studies, but to the authors’ knowl-
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
37
Fig. 5. Structure factor of 1.0 (•), 1.5 (◦), or 2.0 () %, w/v, β-casein solutions as measured with SANS.
Fig. 4. Micellar size at 40 ◦ C. Influence of temperature (A), urea concentration (B), and β-casein concentration (C) on the size of β-casein micelles in (A and C) 5% β-casein in 25 mM sodium phosphate, pH 6.5.
edge this is the first temperature–size profile over the entire temperature range from an essentially monomeric to an essentially micellar system where both the hydrodynamic radius and the radius of gyration are reported. With increasing temperature there is a concomitant increase in micelle size from 0 to 20 ◦ C and thereafter it remains constant (Fig. 4A). On the addition of urea (Fig. 4B) at 40 ◦ C the apparent size decreases, although some particles appear to remain in even in the presence of 6 M urea. This may be due to residual dust. We think so, because total scattering intensity decreases strongly and also because long-range correlations are less likely, as urea is a “good solvent.” The effect of dilution on the size of micelles at 40 ◦ C is far more pronounced (Fig. 4C). The upswing at low concentrations may be due to residual dust particles. The apparent dramatic increase in micelle size on dilution (but at the same time strongly decreasing scattering intensity) is therefore attributed to dust. The dissociation of native micelles and subsequent formation of lamellar micelles on dilution is possible, but unlikely. Changes in micelle morphology with temperature have also been reported. Kajiwara et al. [14] proposed that the shape of β-casein micelles changes from an oblate ellipsoid to a more spherical form with temperature. We investigated this possibility by measuring the radius of gyration of βcasein as a function of temperature; it was shown to increase with temperature (Fig. 4A). The data obtained is consistent with that of Kajiwara et al. [14], who used the ρ value (ρ = RG /RH ) to predict the shape and polydispersity of βcasein micelles. Our results would indicate that the micelles behave like hard spheres (calculated ρ (0.769) corresponds
to a value for√hard spheres [14]). For homogeneous particles the radius is 5/3 RG . Furthermore, we applied depolarized dynamic light scattering in a vertical polarized laser beam and a horizontally oriented grating for the detection at 90◦ , but could not detect any depolarized scattering. DLS results indicated no deviation from sphericity. This is consistent with the interpretation of the SANS data, where the structure factor was calculated at several concentrations and plotted as a function of Q, the scattering wave vector (Fig. 5). As shown, the Q-value at which the first peak occurs is largely independent of concentration, which suggests that β-casein micelles behave effectively as hard spheres at higher temperatures. Furthermore, the first peak occurs at Q = 0.2 nm, which translates into an interaction radius of 15 nm, which is larger than the hydrodynamic radius but in full agreement with a model of a micellar particle with a sparse brush on the outside. It is worth mentioning that even with the SANS data, which contain considerable noise, it is possible to generate a structure factor, which is by itself a strong indication of the concentration-independent size of the micelles. In a further attempt to elucidate the relationship between micellar dimensions and temperature, the molecular mass of β-casein, as affected by temperature, was determined. For this we used our DLS goniometer. This instrument is not ideal for that purpose because of the small and varying scattering volumes as a function of angle. We therefore attribute only a relative significance to the results. From 15 to 35 ◦ C there is a linear increase in molecular mass from 6.38 × 105 to 14.30×105 g mol−1 ; there is little further change at higher temperatures. At 35 ◦ C the average number of monomers per micelle is similar to results reported earlier [10–14] when temperature is accounted for. Our result contains considerable uncertainty due to inaccuracies in concentration, refractive index increment, extrapolation to Q = 0, etc.; usually such data overestimate M by a factor of up to 2. Allowing for this, the various results seem to be consistent with the shell model. When Figs. 1 and 4 are considered in conjunction with the increase in molecular mass with size it would appear that the number and diameter of micelles is largely determined within the temperature range 10 to 25 ◦ C, while
38
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
tion process. It is noteworthy that the temperature where the voluminosity begins to decrease coincides with the inflexion point in the temperature–self-association profile (Fig. 1). The viscosity profile of β-casein, as given in Fig. 7, is consistent with a picture where the viscosity of the dispersion is made up of a continuous phase containing various amounts of monomers. At the same time the increasing number of micelles (with a temperature-dependent specific volume) initially increases viscosity, but due to the decreasing specific volume the viscosity goes down. Using the quantitative micellization data of Mikheeva et al. [3] it is possible to model the micellization process along this line. Fig. 6. Surface hydrophobicity measured from the fluorescence of ANS of 0.5%, w/v, β-casein in 25 mM sodium phosphate, pH 6.5.
4. Conclusions The micellization of β-casein seems to follow a sequential process rather than a two-state or bimodal model as usually is assumed in the literature. The so-called shell model of Kegeles [5] accounts for all the observations made. Our data indicate an almost constant hydrodynamic diameter (as determined by DLS) of the β-casein micelles of 12 nm. β-Casein micelles have a scattering radius of gyration of 8.3 nm and an interaction radius of 15 nm. This is fully consistent with a pervious fluffy particle depicted by Leclerc and Calmettes [16,17] and by De Kruif and May [24] for κcasein, which shows behavior very similar to that of β-casein micelles.
References Fig. 7. Low shear viscosity of β-casein in 25 mM sodium phosphate, pH 6.5, within the concentration range of 0 to 12.5%, w/v.
any subsequent increase in molecular mass, i.e., from 25 to 35 ◦ C, has little effect on micelle diameter. The additional incorporation of monomers (at temperatures > 25 ◦ C) may be compensated for by a concurrent compacting of the micelle as solvent quality decreases with temperature, which is consistent with the calculations of Leclerc and Calmettes [16,17]. The significance of solvent quality in relation to micellization and micelle morphology is illustrated by the effect of temperature on the hydrophobicity and viscosity of β-casein solutions (Figs. 6 and 7). The number of surface or accessible hydrophobic sites in β-casein progressively increases from 2.5 to ∼25 ◦ C and then decreases at higher temperatures as the monomers present in the micelle pack together as solvent quality decreases (Fig. 6). This observation is in close (though not identical) agreement with the results of Creamer and Wheelock [23], who reported a maximum in the temperature–surface hydrophobicity profile of β-casein at ∼32 ◦ C. The temperature–viscosity profile of β-casein (Fig. 7) is interesting, as it shows a clear maximum at temperatures close to the midpoint temperature of the micelliza-
[1] H. Swaisgood, in: P.F. Fox (Ed.), Advanced Dairy Chemistry 1. Proteins, Elsevier Applied Science, London, 1992. [2] D.F. Evans, H. Wennerstrõm, The Colloidal Domain, Wiley–VCH, New York, 1999. [3] L.M. Mikheeva, N.V. Grinberg, V.Ya. Grinberg, A.R. Khokhlov, C.G. de Kruif, Langmuir (2003), in press. [4] G. Kegeles, J. Phys. Chem. 83 (1979) 1728–1732. [5] G. Kegeles, Ind. J. Biochem. Biophys. 29 (1991) 97–102. [6] G.P. Berry, L.K. Creamer, Biochemistry 14 (1975) 3542–3545. [7] M.S. Tai, G. Kegeles, Biophys. Chem. 20 (1984) 81–87. [8] D.G. Schmidt, T.A.J. Payens, J. Colloid Interface Sci. 39 (1972) 655– 662. [9] A.L. Andrews, D. Atkinson, M.T.A. Evans, E.G. Finer, J.P. Green, M.C. Phillips, R.N. Robertson, Biopolymers 18 (1979) 1105–1121. [10] S. Arima, R. Niki, K. Takase, J. Dairy Res. 46 (1979) 281–282. [11] W. Buchheim, D.G. Schmidt, J. Dairy Res. 46 (1979) 277–280. [12] M.T.A. Evans, M.C. Phillips, M.N. Jones, Biopolymers 18 (1979) 1123–1140. [13] A. Thurn, W. Burchard, R. Niki, Colloid Polym. Sci. 265 (1987) 653– 666. [14] K. Kajiwara, R. Niki, H. Urakana, Y. Hirayi, N. Donkai, M. Nagura, Biochim. Biophys. Acta 955 (1988) 128–134. [15] E. Leclerc, P. Calmettes, Physica B 234 (1997) 207–209. [16] E. Leclerc, P. Calmettes, Phys. Rev. Lett. 78 (1998) 150–153. [17] E. Leclerc, P. Calmettes, Physica B 241 (1998) 1141–1143. [18] J. Lyklema, Fundamentals of Colloid and Interface Science, Academic Press, London, 1991.
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
[19] J.K.G. Dhont, An Introduction to Dynamics of Colloids, Elsevier Science, Amsterdam, 1996. [20] M. Kerker, Scattering of Light and Other Electromagnetic Radiation, Academic Press, New York, 1969. [21] J.E. O’Connell, P.F. Fox, Int. Dairy J. 9 (1999) 839–847.
39
[22] K. Takase, R. Niki, S. Arima, Biochim. Biophys. Acta 622 (1980) 1–8. [23] L.K. Creamer, J.V. Wheelock, in: XIX International Dairy Congress, 1E, 1974, p. 257. [24] C.G. De Kruif, R.P. May, Eur. J. Biochem. 200 (1991) 431–436.
Association behavior of β-casein J.E. O’Connell,a,b V.Ya. Grinberg,c and C.G. de Kruif a,d,∗ a NIZO Food Research, 6710 BA Ede, The Netherlands b Food Chemistry Department, University College, Cork, Ireland c Institute of Biochemical Physics, Russian Academy of Sciences, Vavilov St. 28, 117813 Moscow GSP-1, Russia d Van’f Hoff Laboratory, Debye Institute, University of Utrecht, Utrecht, The Netherlands
Received 19 November 2001; accepted 11 October 2002
Abstract The association behavior of β-casein, a protein with a distinct amphipathic character, was studied. β-Casein exhibits markedly temperaturedependent association behavior; at low temperatures (< 10–15 ◦ C), monomers predominate, but as the temperature is increased, monomers associate, via hydrophobic bonding, into micelles. β-Casein micelles have a hydrodynamic radius of ∼12 nm, a radius of gyration of ∼8.3 nm, and an interaction radius of ∼15 nm. These data are fully consistent with a pervious fluffy particle. The association behavior of β-casein is also strongly affected by concentration and solvent quality. At low concentrations β-casein exhibits a critical micelle concentration (CMC) of approximately 0.05%, w/v, at 40 ◦ C. In the presence of 6 M urea the temperature dependence of β-casein’s association behavior is eliminated, leaving monomers predominantly. Temperature-dependent transformations in micelle morphology can be explained by changes in solvent quality, i.e., the temperature–protein hydrophobicity and temperature–voluminosity profiles of β-casein. The results obtained are consistent with the shell model as developed by Kegeles, in which a distribution of micelle sizes is formed. They contrast with the traditional description of the micellization of β-casein by a two-state model or by the closed-association model, i.e., monomers ⇔ micelles. 2003 Elsevier Science (USA). All rights reserved.
1. Introduction
αs2 -casein exhibit consecutive association equilibria, i.e., (αs2 -casein)i + αs2 -casein ↔ (αs2 -casein)i+1 ,
The caseins, which exist in bovine milk as large macromolecular (108 Da, ∼200-nm diameter) associates in colloidal dispersion, known as casein micelles. The four casein fractions differ markedly in many respects, including charge (αs1 - > αs2 - > β- > κ-casein), hydrophobicity (β- > κ> αs1 - > αs2 -casein), and calcium sensitivity (αs2 - > αs1 > β- > κ-casein) [1]. The secondary structure of the caseins also differ considerably. Protein charge is uniformly distributed in αs1 - and αs2 -casein, while β- and κ-casein have well defined polar and apolar regions and thus their secondary structure has a distinctly amphipathic nature. At the natural pH of milk (pH 6.6) the hydrophilic N-terminal of β-casein (amino acids 1 to 21) is highly charged (net charge of −11.5), while the hydrophobic C-terminal has little if any net charge. Consequently, the association behavior of the caseins and the factors affecting it differ considerably. αs1 - and
which are markedly affected by ionic strength and less so by temperature. Traditionally, the micellization of κ- and βcasein is described as exhibiting monomer–micelle association equilibria of the “all-or-nothing”, or closed-association type [2]; i.e., n monomers ↔ micelle, which is strongly affected by temperature and less so by ionic strength. The micellization number, n, is supposedly fixed, n ∼ 20–50. However, Mikheeva et al. [3] showed that the micellization of β-casein can be described quite adequately using the so-called shell model of Kegeles [4,5], in which a distribution of micelle sizes is formed centered around n/2, where n is the maximum (due to steric constraints) micellization number. Thus: β-cas0 + β-cas0 ↔ β-cas1 ,
* Corresponding author.
E-mail address: [email protected] (C.G. de Kruif).
β-cas0 + β-cas1 ↔ β-cas3 , ···
0021-9797/03/$ – see front matter 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0021-9797(02)00066-8
34
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
β-cas0 + β-casi ↔ β-casi+1 , ··· β-cas0 + β-casn−1 ↔ β-casn . Our results, described here, confirm that the shell model describes the micellization of β-casein. For instance, the shell model accounts naturally for the observed polydispersity in micelle size. The fact that β-casein is a major constituent of casein micelles and is also commonly used as a foaming and emulsifying agent means that its association behavior is of importance in the food industry. Centrifugal sedimentation, calorimetry, size-exclusion chromatography, electron microscopy, and light scattering techniques have been used to study the effects of environmental conditions (i.e., ionic strength, temperature, and concentration) on the micellization of β-casein and also on the characteristics of micelles (i.e., shape, size distribution, associative interactions [6–17]). However, the relationship between micelle morphology, temperature, and electrostatic and hydrophobic interactions has yet to be fully elucidated. This communication focuses on the characterization of β-casein micelles. The significance of electrostatic and hydrophobic interactions in micelles was studied using lightscattering techniques.
2. Materials and methods 2.1. Preparation of β-casein High-purity β-casein was obtained from Eurial (Rennes, France). A stock solution (50 g l−1 in 25 mM sodium phosphate, pH 6.5; Merck, D64271, Darmstadt, Germany) was prepared and exhaustively dialyzed (100 ml against 2 × 2 l of 25 mM sodium phosphate, pH 6.5, for 72 h at 4 ◦ C) using dialysis tubing with a nominal molecular weight cutoff of 12 to 14 kDa (T99-T108, Medicell International Ltd., 239 Liverpool Rd., London, N1 ILX, UK). Sodium azide (4%, w/v; Sigma Chemical Co., St. Louis, MO 63103, USA) was added to a final concentration of 0.02%, w/v. 2.2. Dynamic light scattering (DLS) DLS was performed using a Spectra Physics 275-mW Ar laser (λ 514.5 nm). Measuring times varied with the intensity of scattered light but were carried out long enough to collect at least 107 photons. Scattered light was detected, at 90◦ , by a photomultiplier, which was interfaced to an ALV 5000 digital correlator. Using the ALV 5000 correlator an autocorrelation function, G(2) (Q, τ ), was determined [18,19], G(2) (Q, τ ) = I (Q, t)I (Q, t + τ ) , (1) where I (Q, t) is the intensity at a wave vector, Q, and time, t, and τ is the interval time of the correlator. As τ approaches
Fig. 1. (A) Calculated intensity of autocorrelation function for homogeneous particles with a diameter of 22 nm and a size polydispersity β = 0.4 in a lognormal distribution, corresponding to about 17% standard deviation in a Gaussian distribution. The line through the points is a fit with a sum of two exponentials with about equal amplitudes. (B) Amplitude, A1 , of β-casein micelles (•) and A2 (◦) from the fit of experimental DLS data to a sum of two exponentials. (C) Ratio of DLS amplitudes A1 /A2 of 0.5%, w/v, β-casein in 25 mM sodium phosphate, pH 6.5
infinity G(2) (Q, τ ) equals I 2 and we define G(2)(τ ) = G(2) (Q, τ ) G(2)(Q, τ∞ ) .
(2)
From G(2) (τ ) the diffusion coefficient was calculated, which in turn can be used to determine the hydrodynamic radius using the Stokes–Einstein equation, RH = kT /6πηD,
(3)
where k is the Boltzmann constant, T is the temperature (Kelvin), η is the solvent viscosity, and D is the diffusion coefficient. DLS experiments were performed as outlined above and the data obtained were fitted to a sum of two exponentials, resulting in small residuals. Numerical simulation of a polydisperse scattering sample showed that the correlation function can be represented quite well by the sum of two exponentials with amplitudes A1 and A2 . We present such a calculation in Fig. 1A. The two-exponential fit of the simulated data is excellent and superimposes exactly the calculated correlation function. Residuals were smaller than 0.001 for all points. Fitting the data to a cumulant resulted in residuals that were, on average five times larger, especially at short times.
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
The fast mode, in fitting the experimental data, gives the hydrodynamic radius of β-casein micelles. It should be emphasized that attributing the “slow” mode to a second type of particles (as would have occurred in software provided with experimental setups) would have led to nonmeaningful results. The “slow” mode gives an apparent particle size of 80–100 nm. Such particles were, however, not “visible” in static scattering techniques. Therefore, we attribute the “slow” mode to polydispersity. As temperature increases β-casein progressively forms more monodisperse micelles, so that the amplitude for the slow mode decreases. The apparent mean size of the micelles remains remarkably constant, which is consistent with the shell model. The sum of the amplitudes, A1 and A2 , equals a constant under all experimental conditions. Theoretically this value should be 1 but in our case it was 0.62, but this depended on the choice of the pinholes used. It must be mentioned that when the data are fitted to the cumulant the result depends on the choice of τ and the time window of the correlation function, which is limited in early DLS experiments [13]. The double exponential fit appeared to give consistent data in terms of particle diffusion coefficient. Based on the above it is concluded that the micellization of β-casein as affected by temperature, urea, and concentration can be followed.
35
2.4. Surface hydrophobicity The surface hydrophobicity of β-casein was determined as outlined by O’Connell and Fox [21]. To 1.9 ml of βcasein solution (5 g l−1 in 25 mM sodium phosphate, pH 6.5) 0.1 ml of 2.5 mM 8-anilino-1-naphthalene sulfonic acid (ANS; Sigma Chemical Co.) in 25 mM sodium phosphate, pH 6.5, was added and vortexed and the fluorescence was measured at excitation and emission wavelengths of 390 and 480 nm, respectively. A standard curve of β-casein concentration vs fluorescence was constructed to ensure that sufficient ANS was present. The hydrophobicity of β-casein at different temperatures is expressed relative to β-casein at 25 ◦ C, which was attributed a value of 1. 2.5. Viscosity The viscosity of β-casein solutions was measured within the temperature range 0 to 40 ◦ C. Viscosity measurements were taken within the concentration range 0.5 to 12.5%, w/v, at a shear rate of 0.1498 s−1 using a low-shear viscometer (Contraves Low Shear-30, Contraves AG, Zurich, Switzerland).
3. Results and discussion 2.3. Small angle neutron scattering (SANS) SANS experiments were performed at 40 ◦ C on β-casein solutions, 5 to 20 g l−1 . The D-22 spectrometer supplied with cold thermal neutrons emitted from the core of the high-flux nuclear reactor at the Institute Max Von Laue–Paul Langevin in Grenoble, France, was used. The sample path length was 2 mm and the mean wavelength of the emitted neutrons was 1.0 nm with a width at half height of 9%. In SANS experiments the normalized scattering intensity I (Q), which depends on the wave vector Q, is measured. The wave vector is defined as 4π sin(θ/2)/λ, where θ is the angle at which the scattered intensity is detected and λ is the neutron wavelength. For homodisperse solutions I (Q) is related to the structure factor S(Q) and the scattering particle form factor P (Q) by [20] I (Q) = KcMP (Q)S(Q),
(4)
where c is the particle mass concentration, M is the molar mass, and K is the material constant, which depends on the optical constant and wavelength. From the I (Q) and assuming that the structure factor at 5 g/l equals 1 the structure factor at a given concentration was calculated by applying the equation S(Q)2 = I (Q)2 c1 /I (Q)1 c2 ,
(5)
which follows from Eq. (4) and where the subscript “1” refers to the sample at 5 g l−1 , the “2” to the other samples, and c equals the protein concentration. All curves were normalized at high Q, where S(Q) approaches unity.
3.1. Association behavior of β-casein: CMC, influence of temperature and urea Because of the distinct amphipathic character of βcasein, i.e., the presence of a highly hydrophobic C-region, a reduction in solvent quality (with increasing temperature) causes β-casein to associate into soap-like micelles with a hydrophobic core and a soft pervious exterior. When the available data on the association behavior of β-casein are considered a model emerges in which β-casein undergoes a temperature-induced concentration-dependent micellization process, with hydrophobic bonding being the principal associative force, while electrostatic and steric repulsive forces are responsible for ensuring micellar stability, i.e., that the micelles do not aggregate [6–14]. Micellization of β-casein is invariably described as a bimodal or two-state system, n A ↔ An , with the micelle and monomer in an equilibrium which is governed by the mass-action effect. There are strong indications that this closed-association model is not valid. In the current study the shell model of Kegeles [4,5] is applied to explain the association behavior of β-casein. As in the closed-association model, the shell model also assumes a two-state system, i.e., monomers and micelles. However, the shell model accounts for a distribution in micellar size, which the closed-association model does not. As outlined above, β-casein exhibits distinct amphiphilic properties and forms soap-like micelles. Hydrophobic bonding is the principal interactive force responsible for the
36
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
Fig. 2. (A) DLS amplitude A1 of β-casein micelles (•) and amplitude A2 (◦) and (B) amplitude ratio A1 /A2 of 0.5%, w/v, β-casein in 25 mM sodium phosphate, pH 6.5, at 40 ◦ C under the influence of increasing amounts of urea.
endothermic self-assembly of β-casein into micelles, and thus micellization is markedly temperature-dependent. In Figs. 1B and C the result of fitting the DLS data to a sum of two exponentials is plotted. The amplitude A1 is a measure of the number of micelles, while A2 is a measure of the polydispersity. It is shown that the amplitude of the βcasein micelles (A1 ) increases and that A2 decreases with increasing temperature. We attributed the slow mode to “polydispersity,” which is basically equivalent to the suggestion made by Leclerc and Calmettes [16,17], who attributed negative values of the second virial coefficient to relatively longrange correlation fluctuations. “These fluctuations might be regarded as appearing or disappearing aggregates whose structure is looser as full grown micelles.” Indeed, such a phenomenon would show up in DLS as a slow mode. In addition, with increasing temperature the amplitude of the slow mode decreased, but the scattering intensity decreased. This observation would exclude the presence of “physical particles,” because in that case intensity would decrease as well. It is noteworthy that A1 and A2 are relative amplitudes whose sum is an apparatus constant ≈ 0.62. Our interpretation of Fig. 1 is that the number of micelles increases and that polydispersity decreases as temperature is increased. The addition of urea to β-casein at 40 ◦ C monomerizes native β-casein, which confirms that hydrophobic bonding is the principal interactive force responsible for the micellization process (Figs. 2A and B). A critical micelle concentration (CMC) is also consistent with the shell model of Kegeles [4,5]. Within the limit of high cooperativity the CMC behaves in a very similar way to the two-state micelle model. In Fig. 3A we plot the
Fig. 3. (A) Critical micelle concentration at 40 ◦ C. DLS amplitude A1 of β-casein micelles (•) and A2 (◦) at 40 ◦ C, as a function of β-casein concentration. (B) Critical micelle temperature. Scattering amplitude A1 of 0.125 (•), 0.5 (◦), 1.25 (), or 2.5 ()%, w/v, β-casein solutions in sodium phosphate as a function of temperature.
amplitude A1 as a function of the β-casein concentration. In the preceding experiments the association of 0.5% βcasein was studied. Increasing the concentration (0.125 to 2.5%, w/v) decreased the association temperature of βcasein (Fig. 3A). The association behavior of very dilute βcasein solutions was also studied and, as shown in Fig. 3B, β-casein exhibits a critical micelle concentration (CMC) at approximately 0.05%, w/v, at 40 ◦ C. This CMC is a value intermediate to the CMCs of 0.03, 0.05 and 0.07%, w/v, reported by Schmidt and Payens [8], Tai and Kegeles [7], and Takase et al. [22], respectively. 3.2. Size and morphology of β-casein micelles The size and morphology of β-casein micelles at 40 ◦ C, where cohesive interactions strongly promote micellization, should be confined to within close parameters. It would be assumed that β-casein micelles are restricted to a finite size, considering that there must approach a point where charged polar heads of the N-terminal region of the molecules prevent the addition of more monomers and that the surface diameter is limited to not more than twice the chain length. Within the shell model the maximum number of monomers is set as an adjustable parameter. The calorimetric measurements of Mikheeva et al. [3] reported a molecular mass at 35 ◦ C (1%) equivalent to about 25 monomers. We found the same value from light-scattering experiments. Several studies reported an increase in either the hydrodynamic radius [11,13] or the radius of gyration [14] of β-casein micelles as a function of temperature. The results of the current study are similar to those of these studies, but to the authors’ knowl-
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
37
Fig. 5. Structure factor of 1.0 (•), 1.5 (◦), or 2.0 () %, w/v, β-casein solutions as measured with SANS.
Fig. 4. Micellar size at 40 ◦ C. Influence of temperature (A), urea concentration (B), and β-casein concentration (C) on the size of β-casein micelles in (A and C) 5% β-casein in 25 mM sodium phosphate, pH 6.5.
edge this is the first temperature–size profile over the entire temperature range from an essentially monomeric to an essentially micellar system where both the hydrodynamic radius and the radius of gyration are reported. With increasing temperature there is a concomitant increase in micelle size from 0 to 20 ◦ C and thereafter it remains constant (Fig. 4A). On the addition of urea (Fig. 4B) at 40 ◦ C the apparent size decreases, although some particles appear to remain in even in the presence of 6 M urea. This may be due to residual dust. We think so, because total scattering intensity decreases strongly and also because long-range correlations are less likely, as urea is a “good solvent.” The effect of dilution on the size of micelles at 40 ◦ C is far more pronounced (Fig. 4C). The upswing at low concentrations may be due to residual dust particles. The apparent dramatic increase in micelle size on dilution (but at the same time strongly decreasing scattering intensity) is therefore attributed to dust. The dissociation of native micelles and subsequent formation of lamellar micelles on dilution is possible, but unlikely. Changes in micelle morphology with temperature have also been reported. Kajiwara et al. [14] proposed that the shape of β-casein micelles changes from an oblate ellipsoid to a more spherical form with temperature. We investigated this possibility by measuring the radius of gyration of βcasein as a function of temperature; it was shown to increase with temperature (Fig. 4A). The data obtained is consistent with that of Kajiwara et al. [14], who used the ρ value (ρ = RG /RH ) to predict the shape and polydispersity of βcasein micelles. Our results would indicate that the micelles behave like hard spheres (calculated ρ (0.769) corresponds
to a value for√hard spheres [14]). For homogeneous particles the radius is 5/3 RG . Furthermore, we applied depolarized dynamic light scattering in a vertical polarized laser beam and a horizontally oriented grating for the detection at 90◦ , but could not detect any depolarized scattering. DLS results indicated no deviation from sphericity. This is consistent with the interpretation of the SANS data, where the structure factor was calculated at several concentrations and plotted as a function of Q, the scattering wave vector (Fig. 5). As shown, the Q-value at which the first peak occurs is largely independent of concentration, which suggests that β-casein micelles behave effectively as hard spheres at higher temperatures. Furthermore, the first peak occurs at Q = 0.2 nm, which translates into an interaction radius of 15 nm, which is larger than the hydrodynamic radius but in full agreement with a model of a micellar particle with a sparse brush on the outside. It is worth mentioning that even with the SANS data, which contain considerable noise, it is possible to generate a structure factor, which is by itself a strong indication of the concentration-independent size of the micelles. In a further attempt to elucidate the relationship between micellar dimensions and temperature, the molecular mass of β-casein, as affected by temperature, was determined. For this we used our DLS goniometer. This instrument is not ideal for that purpose because of the small and varying scattering volumes as a function of angle. We therefore attribute only a relative significance to the results. From 15 to 35 ◦ C there is a linear increase in molecular mass from 6.38 × 105 to 14.30×105 g mol−1 ; there is little further change at higher temperatures. At 35 ◦ C the average number of monomers per micelle is similar to results reported earlier [10–14] when temperature is accounted for. Our result contains considerable uncertainty due to inaccuracies in concentration, refractive index increment, extrapolation to Q = 0, etc.; usually such data overestimate M by a factor of up to 2. Allowing for this, the various results seem to be consistent with the shell model. When Figs. 1 and 4 are considered in conjunction with the increase in molecular mass with size it would appear that the number and diameter of micelles is largely determined within the temperature range 10 to 25 ◦ C, while
38
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
tion process. It is noteworthy that the temperature where the voluminosity begins to decrease coincides with the inflexion point in the temperature–self-association profile (Fig. 1). The viscosity profile of β-casein, as given in Fig. 7, is consistent with a picture where the viscosity of the dispersion is made up of a continuous phase containing various amounts of monomers. At the same time the increasing number of micelles (with a temperature-dependent specific volume) initially increases viscosity, but due to the decreasing specific volume the viscosity goes down. Using the quantitative micellization data of Mikheeva et al. [3] it is possible to model the micellization process along this line. Fig. 6. Surface hydrophobicity measured from the fluorescence of ANS of 0.5%, w/v, β-casein in 25 mM sodium phosphate, pH 6.5.
4. Conclusions The micellization of β-casein seems to follow a sequential process rather than a two-state or bimodal model as usually is assumed in the literature. The so-called shell model of Kegeles [5] accounts for all the observations made. Our data indicate an almost constant hydrodynamic diameter (as determined by DLS) of the β-casein micelles of 12 nm. β-Casein micelles have a scattering radius of gyration of 8.3 nm and an interaction radius of 15 nm. This is fully consistent with a pervious fluffy particle depicted by Leclerc and Calmettes [16,17] and by De Kruif and May [24] for κcasein, which shows behavior very similar to that of β-casein micelles.
References Fig. 7. Low shear viscosity of β-casein in 25 mM sodium phosphate, pH 6.5, within the concentration range of 0 to 12.5%, w/v.
any subsequent increase in molecular mass, i.e., from 25 to 35 ◦ C, has little effect on micelle diameter. The additional incorporation of monomers (at temperatures > 25 ◦ C) may be compensated for by a concurrent compacting of the micelle as solvent quality decreases with temperature, which is consistent with the calculations of Leclerc and Calmettes [16,17]. The significance of solvent quality in relation to micellization and micelle morphology is illustrated by the effect of temperature on the hydrophobicity and viscosity of β-casein solutions (Figs. 6 and 7). The number of surface or accessible hydrophobic sites in β-casein progressively increases from 2.5 to ∼25 ◦ C and then decreases at higher temperatures as the monomers present in the micelle pack together as solvent quality decreases (Fig. 6). This observation is in close (though not identical) agreement with the results of Creamer and Wheelock [23], who reported a maximum in the temperature–surface hydrophobicity profile of β-casein at ∼32 ◦ C. The temperature–viscosity profile of β-casein (Fig. 7) is interesting, as it shows a clear maximum at temperatures close to the midpoint temperature of the micelliza-
[1] H. Swaisgood, in: P.F. Fox (Ed.), Advanced Dairy Chemistry 1. Proteins, Elsevier Applied Science, London, 1992. [2] D.F. Evans, H. Wennerstrõm, The Colloidal Domain, Wiley–VCH, New York, 1999. [3] L.M. Mikheeva, N.V. Grinberg, V.Ya. Grinberg, A.R. Khokhlov, C.G. de Kruif, Langmuir (2003), in press. [4] G. Kegeles, J. Phys. Chem. 83 (1979) 1728–1732. [5] G. Kegeles, Ind. J. Biochem. Biophys. 29 (1991) 97–102. [6] G.P. Berry, L.K. Creamer, Biochemistry 14 (1975) 3542–3545. [7] M.S. Tai, G. Kegeles, Biophys. Chem. 20 (1984) 81–87. [8] D.G. Schmidt, T.A.J. Payens, J. Colloid Interface Sci. 39 (1972) 655– 662. [9] A.L. Andrews, D. Atkinson, M.T.A. Evans, E.G. Finer, J.P. Green, M.C. Phillips, R.N. Robertson, Biopolymers 18 (1979) 1105–1121. [10] S. Arima, R. Niki, K. Takase, J. Dairy Res. 46 (1979) 281–282. [11] W. Buchheim, D.G. Schmidt, J. Dairy Res. 46 (1979) 277–280. [12] M.T.A. Evans, M.C. Phillips, M.N. Jones, Biopolymers 18 (1979) 1123–1140. [13] A. Thurn, W. Burchard, R. Niki, Colloid Polym. Sci. 265 (1987) 653– 666. [14] K. Kajiwara, R. Niki, H. Urakana, Y. Hirayi, N. Donkai, M. Nagura, Biochim. Biophys. Acta 955 (1988) 128–134. [15] E. Leclerc, P. Calmettes, Physica B 234 (1997) 207–209. [16] E. Leclerc, P. Calmettes, Phys. Rev. Lett. 78 (1998) 150–153. [17] E. Leclerc, P. Calmettes, Physica B 241 (1998) 1141–1143. [18] J. Lyklema, Fundamentals of Colloid and Interface Science, Academic Press, London, 1991.
J.E. O’Connell et al. / Journal of Colloid and Interface Science 258 (2003) 33–39
[19] J.K.G. Dhont, An Introduction to Dynamics of Colloids, Elsevier Science, Amsterdam, 1996. [20] M. Kerker, Scattering of Light and Other Electromagnetic Radiation, Academic Press, New York, 1969. [21] J.E. O’Connell, P.F. Fox, Int. Dairy J. 9 (1999) 839–847.
39
[22] K. Takase, R. Niki, S. Arima, Biochim. Biophys. Acta 622 (1980) 1–8. [23] L.K. Creamer, J.V. Wheelock, in: XIX International Dairy Congress, 1E, 1974, p. 257. [24] C.G. De Kruif, R.P. May, Eur. J. Biochem. 200 (1991) 431–436.