- Email: [email protected]
Adsorption of milk proteins on to calcium phosphate particles
Journal of Colloid and Interface Science 394 (2013) 458–466
Contents lists available at SciVerse ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Adsorption of milk proteins on to calcium phosphate particles Lucile Tercinier a, Aiqian Ye a,⇑, Skelte Anema b, Anne Singh b, Harjinder Singh a a b
Riddet Institute, Massey University, Private Bag 11 222, Palmerston North 4442, New Zealand Fonterra Research Centre, Private Bag 11 029, Palmerston North 4442, New Zealand
a r t i c l e
i n f o
Article history: Received 21 September 2012 Accepted 27 November 2012 Available online 5 December 2012 Keywords: Milk proteins Caseins Whey proteins Calcium Phosphate Hydroxyapatite Adsorption Surface modification Suspension stability
a b s t r a c t The adsorption of caseins from sodium caseinate (SC) and whey proteins from whey protein isolate (WPI) on to particles of hydroxyapatite (HA) was studied. Confocal microscopy and zeta-potential measurements showed that both caseins and whey proteins bound to HA, resulting in an increase in the absolute value of the zeta-potential of the particles. This adsorption improved the suspension stability of the HA particles in water. For both protein sources, there was a preference in the protein adsorption: in the order b-casein > as-casein > j-casein for sodium-caseinate-coated particles; in the order b-lactoglobulin > a-lactalbumin for WPI-coated particles. The adsorption of caseins and whey proteins on to HA could be fitted using a simple Langmuir model, suggesting a single layer adsorption of caseins and whey proteins on to the HA surface. Possible mechanisms involved in the interaction between milk proteins and HA are discussed, in relation to the structure and the surface properties of both milk proteins and HA particles. Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction Milk proteins and insoluble calcium salts co-exist in many dairy products, such as calcium-fortified milks, beverages and yoghurts. A common insoluble calcium salt that is used in these products is tricalcium phosphate (TCP). Most commercial TCP is powdered hydroxyapatite (HA). HA is the main constituent of human bones and teeth and is the most thermodynamically stable of the calcium phosphate phases. As HA is barely soluble in water (solubility constant at 25 °C, Ksp = 4.7 1059 [1]), synthetic HA is one of the preferred calcium salts for the calcium fortification of food products. It is widely used in heat-treated calcium-fortified formulations such as UHT milks because it does not cause any heat instability, unlike soluble calcium salts [2,3]. Despite its industrial relevance, the interactions between HA particles and milk proteins have not been studied in detail, as HA is often considered to be inert in milk and is generally believed not to interact with other milk components. However, it is known that HA interacts strongly with proteins in a wide range of biological applications [4,5]. For example, HA can be used for the separation of proteins in high performance liquid
Abbreviations: ANOVA, analysis of variance; DIC, differential interference contrast; HA, hydroxyapatite; SC, sodium caseinate; SDS–PAGE, sodium dodecyl sulphate polyacrylamide gel electrophoresis; TCP, tricalcium phosphate; WPI, whey protein isolate. ⇑ Corresponding author. Fax: +64 6 350 5655. E-mail address: [email protected] (A. Ye). 0021-9797/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcis.2012.11.058
chromatography techniques [6]. Also, interactions between proteins and synthetic HA crystals have been widely studied for nano-ceramics and bone replacement, as the binding of proteins to bone substitutes can have a dramatic effect on the clinical success of a bone implant [7,8]. The surface of HA particles has been shown to play a critical role in their stability and behaviour in suspension. Therefore, a number of researchers have focused on the surface modification of these particles by electrostatic or steric stabilisation using the ability of HA to adsorb other components [4,5,9,10]. A wide range of proteins or peptides has been used to modify and stabilise the surface of HA and other insoluble calcium phosphates (these include bovine serum albumin, egg lysozyme, bovine serum fibrinogen [10] and lactoferrin [11]), as well as ions (Ca2+, PO3 [12], citrate [13] and 4 fluoride [14]) and other molecules (dodecyl alcohol [15] and silica [9]). This leads to the hypothesis that interactions may occur between milk proteins and HA particles when they are suspended in milk, resulting in the adsorption of the milk proteins on to the surface of the HA particles. To our knowledge, the adsorption of milk proteins on to foodgrade HA particles has not been reported. There are a few studies on the interactions between milk proteins and HA in relation to bone or dentistry topics [16–18]. van Kemenade and de Bruyn [16] looked at the effect of casein adsorption on the kinetics of HA precipitation whereas Ritzoulis et al. [17] used the binding property of caseins to HA to make a porous ceramic–protein composite material from sodium caseinate and HA. Devold et al. [18] also studied the in vitro
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
adsorption of milk proteins on tooth enamel and showed that, at neutral pH, caseins were preferentially bound to tooth enamel, forming protein pellicles. At acid pH, bovine serum albumin and lactoferrin were preferentially bound and the caseins did not adsorb, probably because the pH was close to the isoelectric point of casein. The adsorption of milk proteins on to the surface of HA is likely to be governed mainly by electrostatic interactions, but specific interactions between the phosphoserine residues of casein molecules and the calcium from HA may also be involved. The objective of this study was to examine the adsorption of the main proteins in milk, caseins and whey proteins, on to the surface of HA particles under a range of conditions. This will increase our understanding of the system formed when HA is suspended in milk, and could also lead to new strategies for stabilising insoluble calcium salts in a milk system. 2. Materials and methods 2.1. Materials Sodium caseinate (SC) (Alanate 180) and whey protein isolate (WPI) (Alacen 895) powders were obtained from Fonterra Cooperative Group Limited, New Zealand. The protein content was approximately 93% w/w in the SC powder and approximately 94% w/w in the WPI powder. Food-grade HA powder was purchased from Budenheim (TCP 53-83, Budenheim, Germany). Particle size analysis of the HA powder in water was carried out using a Malvern Mastersizer 2000 (Malvern Instruments Ltd., Malvern, Worcestershire, UK). A suspension of HA in water (1% w/w) was first prepared and introduced drop by drop into the wet dispersion unit of the Mastersizer until the desired obscuration was achieved. The particle size distribution was calculated based on a refractive index of 1.63 and an absorption factor of 0.001 and the median particle size d(0.5) was found to be 4.5 lm. The reported specific surface area of the powder, determined by the BET (Brunauer–Emmett–Teller) method, was 65 m2/g. The morphologies of the powders were observed by scanning electron microscopy. The particles were roughly spherical and each particle was made up of nano-sized crystals that were aggregated together, which explains the high surface area of the powder. 2.2. Preparation of HA/protein suspensions Two methods were used to prepare suspensions of HA in protein solutions with a range of different protein to HA ratios. – In the first method, stock solutions (6% w/w on a powder basis) of SC or WPI in Milli-Q water were prepared and stirred for at least 1 h and were left overnight at 4 °C to allow complete hydration. Protein solutions of different concentrations (0.1– 6% w/w) were prepared by diluting appropriate volumes of the stock solution of SC or WPI in Milli-Q water and stirring for at least 1 h. A constant amount of HA powder (10 g) was added to 90 g aliquots of the protein solutions of different concentrations. – In the second method, a constant amount of protein (100 lL of 1% w/w SC or WPI) was added to Eppendorf tubes containing HA suspensions of various concentrations (0.9 mL of HA suspensions of concentration between 0.1 and 3% w/w). A control was prepared by adding 100 lL of protein solution to 0.9 mL of water. The suspensions were stirred for 2 h at room temperature (approximately 20 °C). They were then centrifuged (3000g for 20 min) to separate the HA from the protein solution. The supernatants were carefully poured from the pellet, weighed and analysed
459
for residual protein content (Sections 2.3 and 2.4). The HA pellets were rinsed twice with Milli-Q water to remove the loosely bound proteins and were put aside for analysis. 2.3. Determination of surface protein concentration The supernatants of the suspensions prepared with a constant amount of HA and various protein concentrations were analysed for total protein using the Kjeldahl method. A factor of 6.38 was used to convert nitrogen content to protein content. The surface protein concentration (mg protein/m2 HA) was calculated from the surface area of the HA and the difference between the amount of protein used to prepare the suspension and that measured in the supernatant. The surface protein concentration was corrected to take into account the portion that was entrapped between particles in the centrifuged pellet but not adsorbed on to the particles (occluded solvent). The surface protein concentration was calculated using the following formula.
Surface protein concentration mi ½P i msup ½Psup ðmwetsed mdrysed Þ ½Psup ¼ 10 mdrysed ½SAHA where mi is the mass of the initial protein solution (g), msup is the mass of the supernatant (g), mwetsed is the mass of wet HA powder obtained after centrifugation (g), mdrysed is the mass of dry HA powder added to the initial protein solution (g), Pi is the measured protein concentration of the initial protein solution (g/100 g), Psup is the measured protein concentration of the supernatant (g/100 g) and SAHA is the surface area per gram of HA (65 m2/g). The portion of the non-adsorbed protein was shown to be 15% of the supernatant proteins at the worst (for high initial protein concentrations, when a large part of the initial proteins remained in the supernatants after adsorption). However, it is believed that this figure could be even lower than 15%, as the HA pellets were quite compact and the unbound or loosely bound proteins could have been squeezed out of the occluded solvent. 2.4. Determination of surface protein composition and preferential adsorption To identify which proteins bound to HA, supernatants prepared with a constant amount of protein solution added to different amounts of HA were analysed. The individual proteins in the supernatant were determined using sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE). A known amount of supernatant was mixed with SDS buffer (0.5 M Tris/HCl buffer, 2% w/w SDS and 0.01% w/w bromophenol blue, pH 6.8) and 5% w/w b-mercaptoethanol was added to the samples followed by heating at 95 °C for 5 min in a boiling water bath to reduce the proteins. A 10 lL aliquot of the solution was loaded on to SDS gels previously prepared on a Bio-Rad mini-gel slab electrophoresis unit (Bio-Rad Laboratories, Richmond, CA) and run at 200 V for 1 h. The protein bands were stained using a solution of Amido Black and the percentage composition of each sample was determined by scanning the bands for as- (as1- + as2-), b- and j-caseins (for samples prepared with SC) or b-lactoglobulin and a-lactalbumin (for samples prepared with WPI). The integrated intensities of the bands were determined using Molecular Dynamics ImageQuant TL software (version 7.0). The amount of protein remaining in the supernatants was expressed as a percentage of the proteins in the control (0.1% w/w SC or WPI without added HA). Each gel experiment was repeated at least three times. Variations were 4% for as-casein (as1- + as2-), 3% for b-casein, 6% for j-caseins and 3% for b-lactoglobulin and a-lactalbumin.
460
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
2.5. Zeta-potential measurements A Malvern Zetasizer Nano-ZS instrument and disposable DTS 1060C zeta-potential cells (Malvern Instruments Ltd., Malvern, Worcestershire, UK) were used for determining the zetapotential of the HA particles in suspension. The HA pellets, obtained after centrifugation, were suspended in Milli-Q water at a concentration of 0.05% w/w. The samples were transferred to the measuring cell and the cell was placed in the instrument. The temperature of the cell was maintained at 20 ± 0.2 °C. An applied voltage of 80 V was used in all experiments. The zetapotential was calculated from the measured electrophoretic mobility using the Smoluchowski model. An average of four measurements was used. 2.6. Confocal laser scanning microscopy A Leica (TCS 4D, Leica Lasertechnik, Heidelberg, Germany) confocal laser scanning microscope with a 100 oil immersion objective lens and an Ar/Kr laser with an excitation line of 488 nm was used to observe the HA particles and to determine whether or not the proteins had bound to the surface of the particles. Selected samples of the HA pellets obtained after centrifugation were suspended in Milli-Q water to a 0.05% w/w concentration. Approximately 3 mL of this suspension was placed in a test tube, Fast Green (fluorescent dye for protein) was added and the mixture was stirred. A sample was then placed on a microscope slide. The slide was covered with a coverslip and observed under the microscope. The images were acquired using two different channels of the microscope. The Fast Green fluorescence channel was used to visualise only the stained protein material. The differential interference contrast (DIC) channel was used to visualise the unstained calcium phosphate particles. 2.7. Turbidity HA pellets were resuspended in Milli-Q water to a 0.125% w/w concentration and turbidity/absorbance measurements were performed using a Jasco V-560 spectrophotometer (Japan Spectroscopic Co., Hachioji City, Japan) and 10 mm quartz glass cells. Absorbance values at a wavelength of 900 nm were recorded over 200 min, as they indicated the change in the turbidity of the samples and were related to the suspension stability of the particles.
buffers or other ions on the properties of both HA surface properties and milk proteins properties. Papers on protein adsorption to HA have reported mixing the samples for a few hours [7,12]. Therefore, the time of 2 h was chosen in this study as a reasonable time for equilibrium to occur. Very little changes in zeta-potential measurements and supernatant composition (protein concentration) were observed after 2 h of mixing, indicating that the equilibrium had been reached and the system was stable. When the particles were resuspended in water for zeta-potential and turbidity measurements, slow changes may have occurred because the equilibrium with the supernatant phase was disturbed. However, these changes were not expected to be significant in the time frame of the measurements, especially since the adsorption of proteins on HA has been reported to be mostly irreversible [19]. 3.1. Zeta-potential The zeta-potential indicates the surface charge property of the HA particles when suspended in water. The HA particles were suspended in protein solutions of different concentrations and were stirred for 2 h. The suspensions were then centrifuged, and the pellets were removed, rinsed three times with water and resuspended in water. The changes in zeta-potential of the suspended pellets as a function of the initial protein concentration of the solutions are shown in Fig. 1. HA was slightly negatively charged when suspended in water (about 11 mV). After mixing with SC, the zeta-potential of the HA progressively increased (became more negative) from 11 to 28 mV with SC addition up to about 2% protein, but did not change further at higher addition levels of SC. Similarly, for WPI, the zeta-potential progressively increased from 11 to 22 mV with the addition of up to 1.5% protein, with no further change at higher addition levels of WPI. As both caseins and whey proteins are negatively charged at neutral pH, the increase in zeta-potential (more negative value) suggested that proteins had adsorbed to the surface of the HA particles. As the zeta-potential value reached a maximum (plateau value) after the addition of 2% protein from SC and 1.5% protein from WPI, a saturation state of the HA surface with respect to protein may have been reached. The plateau value for pellets prepared with WPI (22 mV) was lower than that for pellets prepared with SC (28 mV). 3.2. Protein adsorption
2.8. Statistical analysis The adsorption data were fitted using two different models (Langmuir and Langmuir–Freundlich) to obtain quantified interpretable parameters that described the adsorption process. The model curves were fitted to the experimental data using the freeware computer program R (version 2.15.0) and a non-linear least-squares fitting algorithm to obtain the best-fit model parameters. 3. Results Solutions of SC and WPI were prepared and their pH values were recorded. The pH of the SC solutions was 6.7 and the pH of the WPI solutions was 6.8. The pH of a suspension of 10% (w/w) HA in water was 7.1. As all of the pH values were close to the natural pH of milk, water was used to prepare the protein solutions and no pH re-adjustment was made after mixing with HA particles. The purpose was to look at the adsorption behaviour of milk proteins in a simple media, without the additional effect of
3.2.1. Surface protein coverage The ratio of HA to protein was varied by adding different amounts of either SC or WPI to a constant amount of HA. The dispersions were stirred for 2 h and were centrifuged; the supernatants were then analysed for total protein content. The amount of protein bound to the HA surface was calculated by difference. Fig. 2 shows the amount of SC or WPI adsorbed to the surface of the HA particles as a function of the initial protein concentration. For HA particles mixed with SC, the surface protein concentration increased gradually with SC concentration up to about 2% (w/w, total protein) and then reached a maximum coverage value of 2.4 mg/m2. The surface protein concentration of the HA particles mixed with WPI was lower than that of the HA particles mixed with SC. The surface protein concentration increased gradually with WPI concentration up to 1.5% (w/w, total protein) to a maximum coverage value of 1.3 mg/m2. With respect to the initial protein concentration, the points at which the zeta-potential and the surface protein concentration plateau were coincident, as shown by comparing Figs. 1 and 2. The zeta-potential and surface protein coverage values increased up to the same threshold of 2%
461
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
Zeta-potential (mV)
(A)
(B)
-5
-5
-10
-10
-15
-15
-20
-20
-25
-25 -30
-30 0
1
2
3
4
5
0
6
1
2
3
4
5
6
Initial protein concentration (g/100 g)
Initial protein concentration (g/100 g)
Fig. 1. Effect of protein concentration on the zeta-potential of HA particles that were suspended in (A) SC or (B) WPI solutions of different initial concentrations, stirred for 2 h and centrifuged, and the pellets were then rinsed with water and resuspended in water.
-10
R2 = 0.79 R2 = 0.86
2.5
Zeta-potential (mV)
Surface protein concentration (mg/m2 )
3.0
2.0 1.5 1.0
-15
-20
-25
0.5
-30
0.0 0
1
2
3
4
5
6
Initial protein concentration (g/100 g) Fig. 2. Surface protein concentration (mg/m2) of HA particles as a function of protein concentration: ( ) SC-coated particles; ( ) WPI-coated particles.
protein for SC and 1.5% protein for WPI, showing a direct correlation between zeta-potential and surface protein coverage. This correlation was confirmed by the linear relationship between zeta-potential and surface protein coverage for both SC-coated particles and WPI-coated particles (Fig. 3).
3.2.2. Surface protein composition Various amounts of HA were added to a protein solution containing a constant amount of SC or WPI (0.1% w/w) to determine whether or not there was a preferential adsorption of the individual caseins or whey proteins to the HA particles. After mixing for 2 h, the suspensions were centrifuged and the supernatants were analysed for residual protein content by SDS–PAGE. Fig. 4 shows the SDS–PAGE gels obtained from these supernatant samples. For a constant amount of protein in solution, the amount of protein in the supernatant decreased as the ratio of HA to protein increased. The SDS–PAGE results showed that the ratios between the band intensities of the different caseins or whey proteins were different in the samples containing HA from that in the sample containing only protein (Fig. 4). For SC, the intensities of all the individual casein bands remaining in the supernatant decreased with increasing HA concentration. However, the intensity of the b-casein band decreased faster than the intensities of the as-casein
0.0
0.5
1.0
1.5
2.0
2.5
Surface protein coverage (mg/m 2 ) Fig. 3. Linear relationship between the surface protein coverage of the proteincoated particles and the zeta-potential of the corresponding particles suspended in water: (d) SC-coated particles; (s) WPI-coated particles; the dashed lines are linear regressions of the experimental points.
and j-casein bands, indicating that b-casein was preferentially adsorbed on to HA particles (Fig. 4A). b-Casein had almost entirely disappeared from the supernatant of the sample made with 1.25% w/w HA ( lane 7 on Fig. 4A) whereas as-casein (as1 + as2) still remained. All the available as-casein was bound to HA for HA concentrations greater than 1.5% w/w. A significant amount of j-casein remained in the supernatant of the sample made with 2% w/w HA, as shown by the j-casein band intensity ( lane 9 on Fig. 4A), indicating that j-casein was the least adsorbed among the caseins. Similarly, there was a preferential adsorption of b-lactoglobulin compared with a-lactalbumin for samples prepared with WPI, as the band intensities of b-lactoglobulin decreased faster than the band intensities of a-lactalbumin (Fig. 4B). All the available b-lactoglobulin was bound to the HA surface for HA concentrations greater than 3% w/w whereas some a-lactalbumin remained in the supernatants. Fig. 5 shows quantitative data for the remaining proteins present in the supernatants of HA samples prepared with SC. The concentrations of the individual caseins remaining in the supernatants are expressed as a function of the HA concentration. The b-, as- and j-casein concentrations (for the SC supernatants) are given as a percentage of their initial concentrations in the control (0.1% w/ w SC) without added HA.
462
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
(A)
of adsorbate (in this case, SC or WPI) adsorbed on to an adsorbent (in this case, HA) as a function of concentration once equilibrium has been reached (in this case, the supernatant protein concentration) at a given temperature. This model assumes that SC or WPI binds to a series of distinct empty sites on the surface of HA particles to form a complex. The surface is assumed to be energetically homogeneous, with the same adsorption energy for all adsorption sites. This model is applicable to reversible adsorption. It allows only a monolayer coverage and interactions between proteins adsorbed at the surface are not allowed. The Langmuir model is given by the following equation:
αs1-Casein αs2-Casein β-Casein κ-Casein
(B)
mabs K½P ¼ qm K½P þ 1 S
β-Lactoglobulin α-Lactalbumin
Fig. 4. SDS–PAGE gels of the supernatants of samples containing (A) 0.1% w/w SC and (B) 0.1% w/w WPI and different HA concentrations: lane 1, no added HA; lanes 2–9, 0.1, 0.25, 0.5, 0.75, 1, 1.25, 1.5 and 2% w/w added HA.
80
where mabs/S is the amount of protein bound to the surface expressed in mg/m2, [P] is the concentration of protein at equilibrium (g/100 g), qm is the maximum monolayer surface coverage and K (g/ 100 g) is the Langmuir equilibrium constant. Iafisco et al. [11] used both the Langmuir model and the Langmuir–Freundlich model to characterise the adsorption of lactoferrin on to HA nano-crystals. The Langmuir–Freundlich model takes into account some heterogeneity of the surface (thus all sites are not equal), as well as some possible interactions between the proteins once they are adsorbed at the surface. The Langmuir–Freundlich model is given by Eq. (2). KLF is the Langmuir–Freundlich affinity constant and n is a surface heterogeneity parameter. It still assumes a monolayer coverage but takes into account that the adsorption energy is not equal for all sites (in the case of a heterogeneous surface or if some interactions between proteins at the
(A) 3.0
40
2.5
0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
HA concentration (g/100 g) Fig. 5. Percentage of individual proteins remaining in the supernatants of samples prepared with a constant amount of protein (0.1% w/w SC) and different amounts of HA as a function of HA concentration: ( ) as-casein; (d) b-casein; ( ) j-casein.
The preference for binding was in the order b-casein > as-casein > j-casein (Fig. 5). For example, in the sample containing 1% w/ w HA and 0.1% w/w SC, it can be estimated that 19% of as-casein, 7% of b-casein and 51% of j-casein remained in the supernatant. Therefore, respectively, 81%, 93% and 49% of as-casein, b-casein and j-casein had bound to the surface of HA. In the sample containing 3% w/w HA, no remaining j-casein could be seen, suggesting that all the available j-casein was able to bind to HA when no other caseins were left in the protein solution. Similarly, the preference for binding was in the order b-lactoglobulin > a-lactalbumin for samples made with WPI (data not shown). Therefore, the SDS–PAGE results demonstrated that the ratio of protein to HA in the initial solution has an impact on the composition of the adsorbed protein layer. If the available HA surface is limited, there is a preference for binding, in the order b-lactoglobulin > a-lactalbumin for whey proteins and in the order b-casein > as-casein > j-casein for caseins.
2.0 1.5 1.0 0.5 0.0 0
1
2
3
4
5
6
1
2
3
4
5
6
(B) 2.0
Surface protein
20
Surface protein concentration (mg/m 2)
60
concentration (mg/m2 )
Supernatant residual casein content (%)
100
ð1Þ
1.5
1.0
0.5
0.0 0
Protein concentration at equilibrium (g/100 g)
3.2.3. Adsorption modelling The Langmuir model is the model that is most commonly used to describe adsorption processes [20,21]. It quantifies the amount
Fig. 6. Isotherms of milk proteins adsorbed on to HA and the different best-fit model curves: (A) SC and (B) WPI; (d) experimental data; (solid line) Langmuir model; (dotted line) Langmuir–Freundlich model.
463
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466 Table 1 Parameters for the adsorption of SC and WPI on to HA particles calculated according to the Langmuir and Langmuir–Freundlich models. Model
a
Affinity constant, K and KLF (g/100 g)
Maximum surface coverage, qm (mg/m2)
na
SC
Langmuir Langmuir–Freundlich
0.12 7.68
2.41 2.52
– 0.840
WPI
Langmuir Langmuir–Freundlich
0.04 23.05
1.24 1.24
– 0.998
n is a surface heterogeneity parameter.
surface occur for example). The closer the n value is to 1, the more homogeneous is the surface, or the less impact the surface has on the adsorption process.
mabs ðK LF ½PÞn ¼ qm S 1 þ ðK LF ½PÞn
ð2Þ
The curves for SC and WPI adsorption, calculated using the bestfit parameters, are displayed in Fig. 6A and B respectively. For the WPI adsorption data, there was no difference between the Langmuir and the Langmuir–Freundlich models, because the two curves completely overlapped. Table 1 gives the calculated parameters for each of the models. The affinity constant for the Langmuir model was larger for the SC data than for the WPI data (0.12 versus 0.04), which indicates a higher affinity for HA of caseins than of whey proteins. The maximum surface coverage values given by the models were very close to those observed experimentally in Fig. 2 (maximum surface coverage values of 2.4 and 1.3 mg/m2 for SC and WPI respectively). It is apparent from Eq. (2) that the Langmuir–Freundlich model becomes the same as the Langmuir model if n equals 1; n is a param-
(A)
eter that indicates the extent to which the heterogeneity of the surface affects the adsorption process. The surface heterogeneity has a larger impact on the adsorption process as the n value gets closer to zero. For the WPI data, n was calculated to be 0.998 and the curves overlapped, suggesting that the WPI adsorption process was not affected by the heterogeneity of the HA surface. However, for the SC data, n was calculated to be 0.840, which may indicate that the surface heterogeneity of HA affected the adsorption of the caseins.
3.3. Confocal microscopy The HA pellets prepared with 0.5, 1 and 3% w/w addition levels of SC or WPI solutions were resuspended in water and were observed by confocal microscopy using the Fast Green fluorescence channel and the DIC channel of the microscope. Fast Green dye specifically stains proteins. The control HA sample (a suspension of 0.05% w/w HA powder in water) showed that Fast Green could not stain HA, confirming that no protein was present on the HA particles. Fig. 7 shows the micrographs obtained for the resus-
(B)
10 μm
7.5 μm
(C)
(D)
5 μm
7. 5 μ m
Fig. 7. Confocal micrographs obtained for pellets made with 0.5% w/w SC solution or 0.5% w/w WPI solution, rinsed and resuspended in water (0.05% w/w) and stained with Fast Green: (A) SC, Fast Green channel; (B) SC, overlap of DIC and Fast Green channels; (C) WPI, Fast Green channel; (D) WPI, overlap of DIC and Fast Green channels; the arrows show the HA particles covered by proteins.
464
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
pended stained HA particles prepared with a 0.5% w/w SC solution (Fig. 7A and B) or a 0.5% w/w WPI solution (Fig. 7C and D). The samples were analysed both by using the Fast Green fluorescence channel only (Fig. 7A and C) and by overlapping the fluorescence images with the DIC images (Fig. 7B and D). Fig. 7A shows a typical image of SC-coated HA particles obtained on the Fast Green fluorescence channel. The HA particles were roughly spherical, with a black core surrounded by a green ring, i.e. the protein layer consisting of caseins. The micrographs of the WPI-coated HA particles appeared to be similar to those of the SC-coated particles, but the intensity of the green rings was lower, suggesting a lower amount of whey proteins on the surface of the HA particles and possibly a thinner layer (Fig. 7C). Fig. 7B and D were obtained by overlapping the Fast Green channel signal and the DIC channel signal. The presence of green patches on the white surface of the SC-coated HA particles (Fig. 7B) suggested that the surface was not covered homogeneously and that proteins might be bound preferentially to certain regions of the surface. The WPI-coated particles (Fig. 7D) showed a more uniform coverage of the proteins.
3.4. Suspension behaviour Fig. 8 shows photographs of HA pellets mixed with increasing concentrations of SC. The pellets were rinsed twice with water, were resuspended in water (10% w/w) and were left undisturbed for 24 h at room temperature. The number of particles remaining in suspension after 24 h increased with increasing SC concentration. A similar pattern was observed for WPI-coated particles (images not shown). This improvement in particle suspension was characterised by spectrophotometric measurements as a function of time. Fig. 9 shows the reduction in absorbance with time of 0.125% w/w suspensions of HA particles made with different concentrations of SC (Fig. 9A) or WPI (Fig. 9B). The reduction in absorbance was calculated using (At/A0) 100, where A0 is the initial absorbance and At is the absorbance at time t. Within 200 min, the absorbance of a suspension of HA powder (control with no protein bound) decreased to less than 5% of its original value. All the particles had sedimented to the bottom and the absorbance measured was close to the absorbance of the suspending medium (water). The absorbance of HA suspensions made with different SC or WPI solutions decreased less rapidly compared with the control, and this effect increased with increasing SC or WPI concentration. At low protein concentration (0.5% w/w), the improvement in suspension stability was better for WPI-coated
particles than for SC-coated particles. However, the suspension stability of the HA particles that were fully covered by proteins (HA suspensions made with 3% w/w SC and 3% w/w WPI) was better for particles covered with SC than for particles covered with WPI (45% of the original absorbance value for SC versus 35% for WPI).
4. Discussion The adsorption isotherms (Fig. 2) and the confocal micrographs (Fig. 7) clearly showed that milk proteins from SC and WPI adsorbed at the surface of HA particles, increasing their negative surface charge (Fig. 1). Devold et al. [18] have shown that, at neutral pH, casein molecules are able to bind to tooth enamel, mainly composed of HA. The adsorption on to HA particles of a large range of acidic and basic proteins, including many individual milk proteins, has also been studied by Reynolds and Wong [22]. They showed that as1-, b- and j-caseins, b-lactoglobulin and a-lactalbumin all bound to HA at pH 7, increasing the zeta-potential from 9.1 mV to more negative values between 13.1 mV (for b-lactoglobulin) and 24.1 mV (for as1-casein). The good correlation between the increase in zeta-potential and the increase in surface protein load (Fig. 3) confirmed that protein adsorption was responsible for the increase in surface charge. Small differences in the absolute values of the zeta-potential (at a plateau level) between SC-coated particles and WPI-coated particles may reflect differences in the charge densities of the protein molecules and/or differences in the conformations of the adsorbed proteins [23]. The general mechanism of proteins binding to HA is complicated. The surface of HA particles is heterogeneous and has multiple binding sites. Moreover, the surface morphology and the degree of crystallinity are very dependent on the process used for HA production [4]. Two different types of binding sites on the crystal surface of HA have been identified, and are usually called the C-site and the P-site [20,24]. After dispersion in an aqueous medium, the C-site is rich in calcium ions and positive charges and the P-site is rich in phosphate ions and negative charges. Therefore, HA can potentially bind to both positively and negatively charged regions of proteins. C-sites have been shown to interact with the carboxyl groups of proteins and P-sites have been shown to interact with the amino groups of proteins [6,25]. The initial interaction between these two pairs of groups is believed to be electrostatic in nature. However, it is known that the affinity of calcium ions in the C-site of HA for negatively charged groups in proteins (carboxyl) is higher than the affinity of the phosphate ions in the P-site for positively charged protein residues (amino groups) [20]. This is because coordination complexes are
Fig. 8. Suspension stability as a function of the SC concentration, after standing for 24 h. HA pellets were suspended in (O) water, (A) 0.5% w/w SC, (C) 1% w/w SC, (D) 1.5% w/ w SC and (F) 3% w/w SC, were rinsed with water and were resuspended in water.
465
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
formed between the C-sites and the carboxyl groups of the proteins, and those complexes are held together by much stronger interactions than the non-specific electrostatic interactions occurring between amino groups and phosphates. Therefore, the binding of both caseins and whey proteins to the surface of HA particles is likely to be driven by a mixed ion pairing between the C-sites of HA and the carboxyl groups of the milk proteins and, to a lesser extent, between the P-sites of HA and the amino groups of the milk proteins. HA chromatography is based on the same principle, and the mechanism of protein binding to HA columns has been explained and referred to as a mixed-mode ion-exchange mechanism [6,26]. In addition, casein molecules are phosphorylated proteins. as1-Casein, as2-casein and b-casein have a high content of phosphate groups (phosphoserine residues) that are not uniformly distributed in their sequences and are grouped in clusters [27]. Phosphoserine residues play a crucial role in the structure of casein micelles because they are able to bind cations, especially Ca2+ [28]. Because of their strong calcium-binding ability, they have also been shown to bind to the C-sites of HA [29]. Phosphoryl groups on proteins can interact with C-sites even more strongly than carboxyl groups [30]. Therefore, for SC, the possibility of a specific interaction between negative phosphoserine clusters in casein molecules and the C-site of HA has to be considered. This would explain the higher affinity of the caseins for HA particles compared with the whey proteins, as shown by the higher affinity constant of the Langmuir model and the higher maximum surface coverage (Table 1, Figs. 2 and 6). The involvement of the phosphoserine groups of the caseins in casein/HA interactions was also given as an explanation for the inhibitory action of as1- and b-casein on HA crystal growth by van Kemenade and de Bruyn [16]. They showed a strong complexation between the phosphoserine residues of as1- and b-casein and the growing calcium phosphate phase, leading to the formation of octacalcium phosphate at the expense of HA. The different adsorption levels of caseins and whey proteins may be also due to the different structures of the two proteins. Whey proteins have a small compact globular structure [31], probably form only a monolayer on the HA surface after adsorption, as seen by confocal microscopy (Fig. 7), and do not interact with each other once adsorbed. In contrast, caseins have disordered linear and flexible structures and they can aggregate to form selfassembled particles [32]. The charged groups of the caseins (mainly phosphoseryl and carboxyl groups) would therefore
interact with the HA surface, but the caseins would also associate with each other and form aggregates that would stay at the surface, resulting in a higher surface load and a thicker layer. Unfolded proteins often have lower affinity for HA because they have lower clustering of the carboxyl groups [29]. The preferential adsorption of b-casein to the HA surface (Figs. 4A and 5) may be related to the distribution of the phosphoseryl groups on the casein molecule. The N-terminus of bcasein contains all the phosphoseryl groups of the molecule and is highly negatively charged, whereas the rest of the molecule carries no net charge. Luo and Andrade [20] showed that the geometric distribution of charges on proteins plays an important role in their affinity for HA. The negative head carried by b-casein may therefore explain its preferential adsorption. The phosphoserine-rich regions of b-caseins might also have been more accessible than the phosphoserine residues of as1- and as2-caseins in our system, allowing more interactions between b-casein and the HA surface. j-Casein, because of its very low level of phosphoserine residues, was poorly adsorbed. The charge distribution over the surfaces of b-lactoglobulin and a-lactalbumin have been shown to be different, with a-lactalbumin carrying one main positive region and b-lactoglobulin carrying positive regions spread evenly along its surface [33,34]. The preferential adsorption of b-lactoglobulin over a-lactalbumin on to the HA surface might therefore be explained by a difference in the geometric distribution of the charges of the whey proteins. However, kinetics aspects could also explain the preferential adsorptions observed in this study, for both SC and WPI. For example, as there is more b-lactoglobulin than a-lactalbumin in WPI, b-lactoglobulin might have been able to reach the HA surface first. In this case, the adsorption of each protein from the WPI mixture would be determined primarily by the order and the rate of arrival at the interface, rather than by a difference in their molecular structures [35]. Kinetics aspects are also likely to partially explain the rate of adsorption of the individual caseins. Cooperative effects due to the lateral interactions between casein molecules once adsorbed at the surface could also be involved [20]. However, further investigation on the adsorption of the individual caseins and whey proteins on to HA would be necessary to fully understand the surface composition of HA particles covered by milk proteins. The modelling approach was useful to characterise the adsorption process and to confirm its characteristics. The simple Langmuir model considered that the HA surface was homogeneous, with the same adsorption energy for each site, and showed
(B)100
80
80
60
60
40
40
20
20
Variation of absorbance (%)
(A) 100
0
0 0
50
100
Time (min)
150
200
0
50
100
150
200
Time (min)
Fig. 9. Absorbance as a function of time of suspensions of HA particles (0.125% w/w) made with solutions of different (A) SC or (B) WPI concentrations: ( suspension in water; ( ) HA suspension made with 0.5% w/w SC or WPI; ( ) HA suspension made with 1% w/w SC or WPI; ( suspension made with 2% w/w SC or WPI; ( ) HA suspension made with 3% w/w SC or WPI.
) HA ) HA
466
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
a good fit for both the SC data and the WPI data. However, the surface of HA has been reported to be heterogeneous [20]. The difference reported in Table 1 between the heterogeneity parameter n in the Langmuir–Freundlich model for the WPI data (n = 1) and the SC data (n < 1) and the slightly better fit of the SC data with the Langmuir–Freundlich model (Fig. 6) may be explained by the difference in protein size and structure. As whey proteins have a compact and globular structure, they may not be affected by the surface heterogeneity because they can easily access the complete surface. However, as caseins are aggregated, surface heterogeneity may become important and some areas of the surface may become inaccessible for adsorption. Unlike the Langmuir model, the Langmuir–Freundlich model assumptions allowed interactions between adsorbed proteins at the HA surface. As interactions between caseins adsorbed at the surface are likely to occur when they associate and form aggregates, the Langmuir–Freundlich model was deemed to be a better choice for characterising the SC adsorption data. Nano- or micro-sized particles of HA are known to form spontaneous aggregates when suspended in aqueous media, mainly because of interparticle van der Waals interactions and hydrogen bonding [36]. As the zeta-potential in water of the particles used in this study was low (11 mV), they were not stable in water and some aggregation between particles occurred, leading to an increase in their average size and a fast sedimentation. The binding of proteins to the surface of the particles led to an increase in their surface charge (Fig. 3). As the surface charge became more negative, the repulsive forces between the particles increased, rendering them less prone to aggregation and sedimentation over time (Fig. 8). As the adsorption of caseins from SC led to a slightly more negative zeta-potential than the adsorption of whey proteins from WPI when the maximum surface concentration was reached (28 mV versus 22 mV), the improvement in suspension stability was slightly better for SC-coated particles (Fig. 9). Although this improvement can be referred to as electrostatic stabilisation, in the case of SC, steric stabilisation may also play a role. The charged groups of the caseins interacted with the surface of the particles, and the hydrophobic regions of the caseins may have interacted with each other, forming large aggregates on the particle surface. A multi-layer of caseins may have been formed, providing steric hindrance between the particles and reducing the rate of aggregation.
5. Conclusions This study showed that both caseins and whey proteins can adsorb to food-grade HA particles, changing their colloidal and suspension properties. The protein-coated particles were more negatively charged and more stable in suspension, because a slower sedimentation rate was observed over time when the particles were resuspended in water. Caseins had a higher affinity for HA and were able to adsorb more at the HA surface. The nature of the adsorbed layer was different between the two protein sources. The higher surface load observed for SC-coated particles and the brighter green intensity observed for the protein layers in the confocal pictures suggest that the aggregates of associated casein molecules formed a thick layer, whereas a low surface load and a less intense colour in the confocal pictures for the WPIcoated particles implies a thin single layer of protein on the HA surface. The adsorption mechanism was explained by specific interactions between the carboxyl and phosphoseryl groups of the proteins and the calcium sites on the HA surface and to a lesser
extent between the amino groups of the proteins and the phosphate sites on the HA surface. In conclusion, this work demonstrates the possibility of milk proteins binding to HA particles or other calcium phosphate particles. As the behaviour of HA was shown to change, this could be potentially of interest, for example, as a way to increase the stability of HA particles added to dairy beverages or in the development of solutions to control sediment formation and fouling on heat exchangers. Acknowledgments The authors gratefully acknowledge Dr. Barbara Kuhn-Sherlock and Quentin Béchet for their help with modelling and statistical analysis, Jianyu Chen from the Manawatu Microscopy and Imaging centre for her help with confocal microscopy, Claire Woodhall for proofreading the manuscript and Fonterra Research Centre for their financial support. References [1] H. McDowell, T.M. Gregory, W.E. Brown, J. Res. Nat. Bur. Stand. 81A (1977) 273. [2] E.D. Omoarukhe, N. On-Nom, A.S. Grandison, M.J. Lewis, Int. J. Dairy Technol. 63 (2010) 504. [3] M.J. Lewis, Int. J. Dairy Technol. 64 (2011) 1. [4] J. Norton, K.R. Malik, J.A. Darr, I. Rehman, Adv. Appl. Ceram. 105 (2006) 113. [5] V. Uskokovic´, D.P. Uskokovic´, J. Biomed. Mater. Res. B Appl. Biomater. 96 (2011) 152. [6] M.J. Gorbunoff, S.N. Timasheff, Anal. Biochem. 136 (1984) 440. [7] J.R. Sharpe, R.L. Sammons, P.M. Marquis, Biomaterials 18 (1997) 471. [8] M. Rouahi, E. Champion, O. Gallet, A. Jada, K. Anselme, Colloids Surf. B Biointerfaces 47 (2006) 10. [9] L. Borum, O.C. Wilson Jr., Biomaterials 24 (2003) 3681. [10] N. Brandes, P.B. Welzel, C. Werner, L.W. Kroh, J. Colloid Interface Sci. 299 (2006) 56. [11] M. Iafisco, M. Di Foggia, S. Bonora, M. Prat, N. Roveri, Dalton Trans. 40 (2011) 820. [12] X.D. Zhu, H.S. Fan, D.X. Li, Y.M. Xiao, X.D. Zhang, J. Biomed. Mater. Res. B Appl. Biomater. 82 (2007) 65. [13] M. Johnsson, C.F. Richardson, J.D. Sallis, G.H. Nancollas, Calcif. Tissue Int. 49 (1991) 134. [14] V. Sternitzke, R. Kaegi, J.-N. Audinot, E. Lewin, J.G. Hering, C.A. Johnson, Environ. Sci. Technol. 46 (2012) 802. [15] L. Borum-Nicholas, O.C. Wilson Jr., Biomaterials 24 (2003) 3671. [16] M.J.J.M. van Kemenade, P.L. de Bruyn, J. Colloid Interface Sci. 129 (1989) 1. [17] C. Ritzoulis, N. Scoutaris, K. Papademetriou, S. Stavroulias, C. Panayiotou, Food Hydrocolloids 19 (2005) 575. [18] T.G. Devold, M. Rykke, D. Isabey, E.S. SØrensen, B. Christensen, T. Langsrud, C. Svenning, B. Borch-Iohnsen, J. Karlsen, G.E. Vegarud, Int. Dairy J. 16 (2006) 1013. [19] M.J. Mura-Galelli, J.C. Voegel, S. Behr, E.F. Bres, P. Schaaf, Proc. Natl. Acad. Sci. USA 88 (1991) 5557. [20] Q. Luo, J.D. Andrade, J. Colloid Interface Sci. 200 (1998) 104. [21] M. Wahlgren, T. Arnebrant, Trends Biotechnol. 9 (1991) 201. [22] E.C. Reynolds, A. Wong, Infect. Immun. 39 (1983) 1285. [23] B.S. Chu, S. Ichikawa, S. Kanafusa, M. Nakajima, J. Sci. Food Agric. 88 (2008) 1764. [24] K. Kandori, T. Kuroda, S. Togashi, E. Katayama, J. Phys. Chem. B 115 (2011) 653. [25] M.J. Gorbunoff, S.N. Timasheff, Anal. Biochem. 136 (1984) 433. [26] A. Jungbauer, J. Chromatogr. A 1065 (2005) 3. [27] P.F. Fox, P.L.H. McSweeney, Advanced Dairy Chemistry, Kluwer Academic/ Plenum Publishers, New York, 2003. [28] D.S. Horne, in: A. Thompson, M. Boland, H. Singh (Eds.), Milk Proteins: From Expression to Food, Academic Press/Elsevier, San Diego, 2009, p. 133. [29] G. Bernardi, T. Kawasaki, Biochim. Biophys. Acta 160 (1968) 301. [30] T. Kawasaki, J. Chromatogr. A 544 (1991) 147. [31] E. Dickinson, Colloids Surf. B Biointerfaces 20 (2001) 197. [32] M. Srinivasan, H. Singh, P.A. Munro, Int. Dairy J. 9 (1999) 337. [33] R. de Vries, J. Chem. Phys. 120 (2004) 3475. [34] E. Casal, A. Montilla, F.J. Moreno, A. Olano, N. Corzo, J. Dairy Sci. 89 (2006) 1384. [35] E. Dickinson, Food Hydrocolloids 25 (2011) 1966. [36] Z. Sadeghian, J.G. Heinrich, F. Moztarzadeh, Ceram. Int. 32 (2006) 331.
Contents lists available at SciVerse ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Adsorption of milk proteins on to calcium phosphate particles Lucile Tercinier a, Aiqian Ye a,⇑, Skelte Anema b, Anne Singh b, Harjinder Singh a a b
Riddet Institute, Massey University, Private Bag 11 222, Palmerston North 4442, New Zealand Fonterra Research Centre, Private Bag 11 029, Palmerston North 4442, New Zealand
a r t i c l e
i n f o
Article history: Received 21 September 2012 Accepted 27 November 2012 Available online 5 December 2012 Keywords: Milk proteins Caseins Whey proteins Calcium Phosphate Hydroxyapatite Adsorption Surface modification Suspension stability
a b s t r a c t The adsorption of caseins from sodium caseinate (SC) and whey proteins from whey protein isolate (WPI) on to particles of hydroxyapatite (HA) was studied. Confocal microscopy and zeta-potential measurements showed that both caseins and whey proteins bound to HA, resulting in an increase in the absolute value of the zeta-potential of the particles. This adsorption improved the suspension stability of the HA particles in water. For both protein sources, there was a preference in the protein adsorption: in the order b-casein > as-casein > j-casein for sodium-caseinate-coated particles; in the order b-lactoglobulin > a-lactalbumin for WPI-coated particles. The adsorption of caseins and whey proteins on to HA could be fitted using a simple Langmuir model, suggesting a single layer adsorption of caseins and whey proteins on to the HA surface. Possible mechanisms involved in the interaction between milk proteins and HA are discussed, in relation to the structure and the surface properties of both milk proteins and HA particles. Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction Milk proteins and insoluble calcium salts co-exist in many dairy products, such as calcium-fortified milks, beverages and yoghurts. A common insoluble calcium salt that is used in these products is tricalcium phosphate (TCP). Most commercial TCP is powdered hydroxyapatite (HA). HA is the main constituent of human bones and teeth and is the most thermodynamically stable of the calcium phosphate phases. As HA is barely soluble in water (solubility constant at 25 °C, Ksp = 4.7 1059 [1]), synthetic HA is one of the preferred calcium salts for the calcium fortification of food products. It is widely used in heat-treated calcium-fortified formulations such as UHT milks because it does not cause any heat instability, unlike soluble calcium salts [2,3]. Despite its industrial relevance, the interactions between HA particles and milk proteins have not been studied in detail, as HA is often considered to be inert in milk and is generally believed not to interact with other milk components. However, it is known that HA interacts strongly with proteins in a wide range of biological applications [4,5]. For example, HA can be used for the separation of proteins in high performance liquid
Abbreviations: ANOVA, analysis of variance; DIC, differential interference contrast; HA, hydroxyapatite; SC, sodium caseinate; SDS–PAGE, sodium dodecyl sulphate polyacrylamide gel electrophoresis; TCP, tricalcium phosphate; WPI, whey protein isolate. ⇑ Corresponding author. Fax: +64 6 350 5655. E-mail address: [email protected] (A. Ye). 0021-9797/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcis.2012.11.058
chromatography techniques [6]. Also, interactions between proteins and synthetic HA crystals have been widely studied for nano-ceramics and bone replacement, as the binding of proteins to bone substitutes can have a dramatic effect on the clinical success of a bone implant [7,8]. The surface of HA particles has been shown to play a critical role in their stability and behaviour in suspension. Therefore, a number of researchers have focused on the surface modification of these particles by electrostatic or steric stabilisation using the ability of HA to adsorb other components [4,5,9,10]. A wide range of proteins or peptides has been used to modify and stabilise the surface of HA and other insoluble calcium phosphates (these include bovine serum albumin, egg lysozyme, bovine serum fibrinogen [10] and lactoferrin [11]), as well as ions (Ca2+, PO3 [12], citrate [13] and 4 fluoride [14]) and other molecules (dodecyl alcohol [15] and silica [9]). This leads to the hypothesis that interactions may occur between milk proteins and HA particles when they are suspended in milk, resulting in the adsorption of the milk proteins on to the surface of the HA particles. To our knowledge, the adsorption of milk proteins on to foodgrade HA particles has not been reported. There are a few studies on the interactions between milk proteins and HA in relation to bone or dentistry topics [16–18]. van Kemenade and de Bruyn [16] looked at the effect of casein adsorption on the kinetics of HA precipitation whereas Ritzoulis et al. [17] used the binding property of caseins to HA to make a porous ceramic–protein composite material from sodium caseinate and HA. Devold et al. [18] also studied the in vitro
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
adsorption of milk proteins on tooth enamel and showed that, at neutral pH, caseins were preferentially bound to tooth enamel, forming protein pellicles. At acid pH, bovine serum albumin and lactoferrin were preferentially bound and the caseins did not adsorb, probably because the pH was close to the isoelectric point of casein. The adsorption of milk proteins on to the surface of HA is likely to be governed mainly by electrostatic interactions, but specific interactions between the phosphoserine residues of casein molecules and the calcium from HA may also be involved. The objective of this study was to examine the adsorption of the main proteins in milk, caseins and whey proteins, on to the surface of HA particles under a range of conditions. This will increase our understanding of the system formed when HA is suspended in milk, and could also lead to new strategies for stabilising insoluble calcium salts in a milk system. 2. Materials and methods 2.1. Materials Sodium caseinate (SC) (Alanate 180) and whey protein isolate (WPI) (Alacen 895) powders were obtained from Fonterra Cooperative Group Limited, New Zealand. The protein content was approximately 93% w/w in the SC powder and approximately 94% w/w in the WPI powder. Food-grade HA powder was purchased from Budenheim (TCP 53-83, Budenheim, Germany). Particle size analysis of the HA powder in water was carried out using a Malvern Mastersizer 2000 (Malvern Instruments Ltd., Malvern, Worcestershire, UK). A suspension of HA in water (1% w/w) was first prepared and introduced drop by drop into the wet dispersion unit of the Mastersizer until the desired obscuration was achieved. The particle size distribution was calculated based on a refractive index of 1.63 and an absorption factor of 0.001 and the median particle size d(0.5) was found to be 4.5 lm. The reported specific surface area of the powder, determined by the BET (Brunauer–Emmett–Teller) method, was 65 m2/g. The morphologies of the powders were observed by scanning electron microscopy. The particles were roughly spherical and each particle was made up of nano-sized crystals that were aggregated together, which explains the high surface area of the powder. 2.2. Preparation of HA/protein suspensions Two methods were used to prepare suspensions of HA in protein solutions with a range of different protein to HA ratios. – In the first method, stock solutions (6% w/w on a powder basis) of SC or WPI in Milli-Q water were prepared and stirred for at least 1 h and were left overnight at 4 °C to allow complete hydration. Protein solutions of different concentrations (0.1– 6% w/w) were prepared by diluting appropriate volumes of the stock solution of SC or WPI in Milli-Q water and stirring for at least 1 h. A constant amount of HA powder (10 g) was added to 90 g aliquots of the protein solutions of different concentrations. – In the second method, a constant amount of protein (100 lL of 1% w/w SC or WPI) was added to Eppendorf tubes containing HA suspensions of various concentrations (0.9 mL of HA suspensions of concentration between 0.1 and 3% w/w). A control was prepared by adding 100 lL of protein solution to 0.9 mL of water. The suspensions were stirred for 2 h at room temperature (approximately 20 °C). They were then centrifuged (3000g for 20 min) to separate the HA from the protein solution. The supernatants were carefully poured from the pellet, weighed and analysed
459
for residual protein content (Sections 2.3 and 2.4). The HA pellets were rinsed twice with Milli-Q water to remove the loosely bound proteins and were put aside for analysis. 2.3. Determination of surface protein concentration The supernatants of the suspensions prepared with a constant amount of HA and various protein concentrations were analysed for total protein using the Kjeldahl method. A factor of 6.38 was used to convert nitrogen content to protein content. The surface protein concentration (mg protein/m2 HA) was calculated from the surface area of the HA and the difference between the amount of protein used to prepare the suspension and that measured in the supernatant. The surface protein concentration was corrected to take into account the portion that was entrapped between particles in the centrifuged pellet but not adsorbed on to the particles (occluded solvent). The surface protein concentration was calculated using the following formula.
Surface protein concentration mi ½P i msup ½Psup ðmwetsed mdrysed Þ ½Psup ¼ 10 mdrysed ½SAHA where mi is the mass of the initial protein solution (g), msup is the mass of the supernatant (g), mwetsed is the mass of wet HA powder obtained after centrifugation (g), mdrysed is the mass of dry HA powder added to the initial protein solution (g), Pi is the measured protein concentration of the initial protein solution (g/100 g), Psup is the measured protein concentration of the supernatant (g/100 g) and SAHA is the surface area per gram of HA (65 m2/g). The portion of the non-adsorbed protein was shown to be 15% of the supernatant proteins at the worst (for high initial protein concentrations, when a large part of the initial proteins remained in the supernatants after adsorption). However, it is believed that this figure could be even lower than 15%, as the HA pellets were quite compact and the unbound or loosely bound proteins could have been squeezed out of the occluded solvent. 2.4. Determination of surface protein composition and preferential adsorption To identify which proteins bound to HA, supernatants prepared with a constant amount of protein solution added to different amounts of HA were analysed. The individual proteins in the supernatant were determined using sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE). A known amount of supernatant was mixed with SDS buffer (0.5 M Tris/HCl buffer, 2% w/w SDS and 0.01% w/w bromophenol blue, pH 6.8) and 5% w/w b-mercaptoethanol was added to the samples followed by heating at 95 °C for 5 min in a boiling water bath to reduce the proteins. A 10 lL aliquot of the solution was loaded on to SDS gels previously prepared on a Bio-Rad mini-gel slab electrophoresis unit (Bio-Rad Laboratories, Richmond, CA) and run at 200 V for 1 h. The protein bands were stained using a solution of Amido Black and the percentage composition of each sample was determined by scanning the bands for as- (as1- + as2-), b- and j-caseins (for samples prepared with SC) or b-lactoglobulin and a-lactalbumin (for samples prepared with WPI). The integrated intensities of the bands were determined using Molecular Dynamics ImageQuant TL software (version 7.0). The amount of protein remaining in the supernatants was expressed as a percentage of the proteins in the control (0.1% w/w SC or WPI without added HA). Each gel experiment was repeated at least three times. Variations were 4% for as-casein (as1- + as2-), 3% for b-casein, 6% for j-caseins and 3% for b-lactoglobulin and a-lactalbumin.
460
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
2.5. Zeta-potential measurements A Malvern Zetasizer Nano-ZS instrument and disposable DTS 1060C zeta-potential cells (Malvern Instruments Ltd., Malvern, Worcestershire, UK) were used for determining the zetapotential of the HA particles in suspension. The HA pellets, obtained after centrifugation, were suspended in Milli-Q water at a concentration of 0.05% w/w. The samples were transferred to the measuring cell and the cell was placed in the instrument. The temperature of the cell was maintained at 20 ± 0.2 °C. An applied voltage of 80 V was used in all experiments. The zetapotential was calculated from the measured electrophoretic mobility using the Smoluchowski model. An average of four measurements was used. 2.6. Confocal laser scanning microscopy A Leica (TCS 4D, Leica Lasertechnik, Heidelberg, Germany) confocal laser scanning microscope with a 100 oil immersion objective lens and an Ar/Kr laser with an excitation line of 488 nm was used to observe the HA particles and to determine whether or not the proteins had bound to the surface of the particles. Selected samples of the HA pellets obtained after centrifugation were suspended in Milli-Q water to a 0.05% w/w concentration. Approximately 3 mL of this suspension was placed in a test tube, Fast Green (fluorescent dye for protein) was added and the mixture was stirred. A sample was then placed on a microscope slide. The slide was covered with a coverslip and observed under the microscope. The images were acquired using two different channels of the microscope. The Fast Green fluorescence channel was used to visualise only the stained protein material. The differential interference contrast (DIC) channel was used to visualise the unstained calcium phosphate particles. 2.7. Turbidity HA pellets were resuspended in Milli-Q water to a 0.125% w/w concentration and turbidity/absorbance measurements were performed using a Jasco V-560 spectrophotometer (Japan Spectroscopic Co., Hachioji City, Japan) and 10 mm quartz glass cells. Absorbance values at a wavelength of 900 nm were recorded over 200 min, as they indicated the change in the turbidity of the samples and were related to the suspension stability of the particles.
buffers or other ions on the properties of both HA surface properties and milk proteins properties. Papers on protein adsorption to HA have reported mixing the samples for a few hours [7,12]. Therefore, the time of 2 h was chosen in this study as a reasonable time for equilibrium to occur. Very little changes in zeta-potential measurements and supernatant composition (protein concentration) were observed after 2 h of mixing, indicating that the equilibrium had been reached and the system was stable. When the particles were resuspended in water for zeta-potential and turbidity measurements, slow changes may have occurred because the equilibrium with the supernatant phase was disturbed. However, these changes were not expected to be significant in the time frame of the measurements, especially since the adsorption of proteins on HA has been reported to be mostly irreversible [19]. 3.1. Zeta-potential The zeta-potential indicates the surface charge property of the HA particles when suspended in water. The HA particles were suspended in protein solutions of different concentrations and were stirred for 2 h. The suspensions were then centrifuged, and the pellets were removed, rinsed three times with water and resuspended in water. The changes in zeta-potential of the suspended pellets as a function of the initial protein concentration of the solutions are shown in Fig. 1. HA was slightly negatively charged when suspended in water (about 11 mV). After mixing with SC, the zeta-potential of the HA progressively increased (became more negative) from 11 to 28 mV with SC addition up to about 2% protein, but did not change further at higher addition levels of SC. Similarly, for WPI, the zeta-potential progressively increased from 11 to 22 mV with the addition of up to 1.5% protein, with no further change at higher addition levels of WPI. As both caseins and whey proteins are negatively charged at neutral pH, the increase in zeta-potential (more negative value) suggested that proteins had adsorbed to the surface of the HA particles. As the zeta-potential value reached a maximum (plateau value) after the addition of 2% protein from SC and 1.5% protein from WPI, a saturation state of the HA surface with respect to protein may have been reached. The plateau value for pellets prepared with WPI (22 mV) was lower than that for pellets prepared with SC (28 mV). 3.2. Protein adsorption
2.8. Statistical analysis The adsorption data were fitted using two different models (Langmuir and Langmuir–Freundlich) to obtain quantified interpretable parameters that described the adsorption process. The model curves were fitted to the experimental data using the freeware computer program R (version 2.15.0) and a non-linear least-squares fitting algorithm to obtain the best-fit model parameters. 3. Results Solutions of SC and WPI were prepared and their pH values were recorded. The pH of the SC solutions was 6.7 and the pH of the WPI solutions was 6.8. The pH of a suspension of 10% (w/w) HA in water was 7.1. As all of the pH values were close to the natural pH of milk, water was used to prepare the protein solutions and no pH re-adjustment was made after mixing with HA particles. The purpose was to look at the adsorption behaviour of milk proteins in a simple media, without the additional effect of
3.2.1. Surface protein coverage The ratio of HA to protein was varied by adding different amounts of either SC or WPI to a constant amount of HA. The dispersions were stirred for 2 h and were centrifuged; the supernatants were then analysed for total protein content. The amount of protein bound to the HA surface was calculated by difference. Fig. 2 shows the amount of SC or WPI adsorbed to the surface of the HA particles as a function of the initial protein concentration. For HA particles mixed with SC, the surface protein concentration increased gradually with SC concentration up to about 2% (w/w, total protein) and then reached a maximum coverage value of 2.4 mg/m2. The surface protein concentration of the HA particles mixed with WPI was lower than that of the HA particles mixed with SC. The surface protein concentration increased gradually with WPI concentration up to 1.5% (w/w, total protein) to a maximum coverage value of 1.3 mg/m2. With respect to the initial protein concentration, the points at which the zeta-potential and the surface protein concentration plateau were coincident, as shown by comparing Figs. 1 and 2. The zeta-potential and surface protein coverage values increased up to the same threshold of 2%
461
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
Zeta-potential (mV)
(A)
(B)
-5
-5
-10
-10
-15
-15
-20
-20
-25
-25 -30
-30 0
1
2
3
4
5
0
6
1
2
3
4
5
6
Initial protein concentration (g/100 g)
Initial protein concentration (g/100 g)
Fig. 1. Effect of protein concentration on the zeta-potential of HA particles that were suspended in (A) SC or (B) WPI solutions of different initial concentrations, stirred for 2 h and centrifuged, and the pellets were then rinsed with water and resuspended in water.
-10
R2 = 0.79 R2 = 0.86
2.5
Zeta-potential (mV)
Surface protein concentration (mg/m2 )
3.0
2.0 1.5 1.0
-15
-20
-25
0.5
-30
0.0 0
1
2
3
4
5
6
Initial protein concentration (g/100 g) Fig. 2. Surface protein concentration (mg/m2) of HA particles as a function of protein concentration: ( ) SC-coated particles; ( ) WPI-coated particles.
protein for SC and 1.5% protein for WPI, showing a direct correlation between zeta-potential and surface protein coverage. This correlation was confirmed by the linear relationship between zeta-potential and surface protein coverage for both SC-coated particles and WPI-coated particles (Fig. 3).
3.2.2. Surface protein composition Various amounts of HA were added to a protein solution containing a constant amount of SC or WPI (0.1% w/w) to determine whether or not there was a preferential adsorption of the individual caseins or whey proteins to the HA particles. After mixing for 2 h, the suspensions were centrifuged and the supernatants were analysed for residual protein content by SDS–PAGE. Fig. 4 shows the SDS–PAGE gels obtained from these supernatant samples. For a constant amount of protein in solution, the amount of protein in the supernatant decreased as the ratio of HA to protein increased. The SDS–PAGE results showed that the ratios between the band intensities of the different caseins or whey proteins were different in the samples containing HA from that in the sample containing only protein (Fig. 4). For SC, the intensities of all the individual casein bands remaining in the supernatant decreased with increasing HA concentration. However, the intensity of the b-casein band decreased faster than the intensities of the as-casein
0.0
0.5
1.0
1.5
2.0
2.5
Surface protein coverage (mg/m 2 ) Fig. 3. Linear relationship between the surface protein coverage of the proteincoated particles and the zeta-potential of the corresponding particles suspended in water: (d) SC-coated particles; (s) WPI-coated particles; the dashed lines are linear regressions of the experimental points.
and j-casein bands, indicating that b-casein was preferentially adsorbed on to HA particles (Fig. 4A). b-Casein had almost entirely disappeared from the supernatant of the sample made with 1.25% w/w HA ( lane 7 on Fig. 4A) whereas as-casein (as1 + as2) still remained. All the available as-casein was bound to HA for HA concentrations greater than 1.5% w/w. A significant amount of j-casein remained in the supernatant of the sample made with 2% w/w HA, as shown by the j-casein band intensity ( lane 9 on Fig. 4A), indicating that j-casein was the least adsorbed among the caseins. Similarly, there was a preferential adsorption of b-lactoglobulin compared with a-lactalbumin for samples prepared with WPI, as the band intensities of b-lactoglobulin decreased faster than the band intensities of a-lactalbumin (Fig. 4B). All the available b-lactoglobulin was bound to the HA surface for HA concentrations greater than 3% w/w whereas some a-lactalbumin remained in the supernatants. Fig. 5 shows quantitative data for the remaining proteins present in the supernatants of HA samples prepared with SC. The concentrations of the individual caseins remaining in the supernatants are expressed as a function of the HA concentration. The b-, as- and j-casein concentrations (for the SC supernatants) are given as a percentage of their initial concentrations in the control (0.1% w/ w SC) without added HA.
462
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
(A)
of adsorbate (in this case, SC or WPI) adsorbed on to an adsorbent (in this case, HA) as a function of concentration once equilibrium has been reached (in this case, the supernatant protein concentration) at a given temperature. This model assumes that SC or WPI binds to a series of distinct empty sites on the surface of HA particles to form a complex. The surface is assumed to be energetically homogeneous, with the same adsorption energy for all adsorption sites. This model is applicable to reversible adsorption. It allows only a monolayer coverage and interactions between proteins adsorbed at the surface are not allowed. The Langmuir model is given by the following equation:
αs1-Casein αs2-Casein β-Casein κ-Casein
(B)
mabs K½P ¼ qm K½P þ 1 S
β-Lactoglobulin α-Lactalbumin
Fig. 4. SDS–PAGE gels of the supernatants of samples containing (A) 0.1% w/w SC and (B) 0.1% w/w WPI and different HA concentrations: lane 1, no added HA; lanes 2–9, 0.1, 0.25, 0.5, 0.75, 1, 1.25, 1.5 and 2% w/w added HA.
80
where mabs/S is the amount of protein bound to the surface expressed in mg/m2, [P] is the concentration of protein at equilibrium (g/100 g), qm is the maximum monolayer surface coverage and K (g/ 100 g) is the Langmuir equilibrium constant. Iafisco et al. [11] used both the Langmuir model and the Langmuir–Freundlich model to characterise the adsorption of lactoferrin on to HA nano-crystals. The Langmuir–Freundlich model takes into account some heterogeneity of the surface (thus all sites are not equal), as well as some possible interactions between the proteins once they are adsorbed at the surface. The Langmuir–Freundlich model is given by Eq. (2). KLF is the Langmuir–Freundlich affinity constant and n is a surface heterogeneity parameter. It still assumes a monolayer coverage but takes into account that the adsorption energy is not equal for all sites (in the case of a heterogeneous surface or if some interactions between proteins at the
(A) 3.0
40
2.5
0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
HA concentration (g/100 g) Fig. 5. Percentage of individual proteins remaining in the supernatants of samples prepared with a constant amount of protein (0.1% w/w SC) and different amounts of HA as a function of HA concentration: ( ) as-casein; (d) b-casein; ( ) j-casein.
The preference for binding was in the order b-casein > as-casein > j-casein (Fig. 5). For example, in the sample containing 1% w/ w HA and 0.1% w/w SC, it can be estimated that 19% of as-casein, 7% of b-casein and 51% of j-casein remained in the supernatant. Therefore, respectively, 81%, 93% and 49% of as-casein, b-casein and j-casein had bound to the surface of HA. In the sample containing 3% w/w HA, no remaining j-casein could be seen, suggesting that all the available j-casein was able to bind to HA when no other caseins were left in the protein solution. Similarly, the preference for binding was in the order b-lactoglobulin > a-lactalbumin for samples made with WPI (data not shown). Therefore, the SDS–PAGE results demonstrated that the ratio of protein to HA in the initial solution has an impact on the composition of the adsorbed protein layer. If the available HA surface is limited, there is a preference for binding, in the order b-lactoglobulin > a-lactalbumin for whey proteins and in the order b-casein > as-casein > j-casein for caseins.
2.0 1.5 1.0 0.5 0.0 0
1
2
3
4
5
6
1
2
3
4
5
6
(B) 2.0
Surface protein
20
Surface protein concentration (mg/m 2)
60
concentration (mg/m2 )
Supernatant residual casein content (%)
100
ð1Þ
1.5
1.0
0.5
0.0 0
Protein concentration at equilibrium (g/100 g)
3.2.3. Adsorption modelling The Langmuir model is the model that is most commonly used to describe adsorption processes [20,21]. It quantifies the amount
Fig. 6. Isotherms of milk proteins adsorbed on to HA and the different best-fit model curves: (A) SC and (B) WPI; (d) experimental data; (solid line) Langmuir model; (dotted line) Langmuir–Freundlich model.
463
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466 Table 1 Parameters for the adsorption of SC and WPI on to HA particles calculated according to the Langmuir and Langmuir–Freundlich models. Model
a
Affinity constant, K and KLF (g/100 g)
Maximum surface coverage, qm (mg/m2)
na
SC
Langmuir Langmuir–Freundlich
0.12 7.68
2.41 2.52
– 0.840
WPI
Langmuir Langmuir–Freundlich
0.04 23.05
1.24 1.24
– 0.998
n is a surface heterogeneity parameter.
surface occur for example). The closer the n value is to 1, the more homogeneous is the surface, or the less impact the surface has on the adsorption process.
mabs ðK LF ½PÞn ¼ qm S 1 þ ðK LF ½PÞn
ð2Þ
The curves for SC and WPI adsorption, calculated using the bestfit parameters, are displayed in Fig. 6A and B respectively. For the WPI adsorption data, there was no difference between the Langmuir and the Langmuir–Freundlich models, because the two curves completely overlapped. Table 1 gives the calculated parameters for each of the models. The affinity constant for the Langmuir model was larger for the SC data than for the WPI data (0.12 versus 0.04), which indicates a higher affinity for HA of caseins than of whey proteins. The maximum surface coverage values given by the models were very close to those observed experimentally in Fig. 2 (maximum surface coverage values of 2.4 and 1.3 mg/m2 for SC and WPI respectively). It is apparent from Eq. (2) that the Langmuir–Freundlich model becomes the same as the Langmuir model if n equals 1; n is a param-
(A)
eter that indicates the extent to which the heterogeneity of the surface affects the adsorption process. The surface heterogeneity has a larger impact on the adsorption process as the n value gets closer to zero. For the WPI data, n was calculated to be 0.998 and the curves overlapped, suggesting that the WPI adsorption process was not affected by the heterogeneity of the HA surface. However, for the SC data, n was calculated to be 0.840, which may indicate that the surface heterogeneity of HA affected the adsorption of the caseins.
3.3. Confocal microscopy The HA pellets prepared with 0.5, 1 and 3% w/w addition levels of SC or WPI solutions were resuspended in water and were observed by confocal microscopy using the Fast Green fluorescence channel and the DIC channel of the microscope. Fast Green dye specifically stains proteins. The control HA sample (a suspension of 0.05% w/w HA powder in water) showed that Fast Green could not stain HA, confirming that no protein was present on the HA particles. Fig. 7 shows the micrographs obtained for the resus-
(B)
10 μm
7.5 μm
(C)
(D)
5 μm
7. 5 μ m
Fig. 7. Confocal micrographs obtained for pellets made with 0.5% w/w SC solution or 0.5% w/w WPI solution, rinsed and resuspended in water (0.05% w/w) and stained with Fast Green: (A) SC, Fast Green channel; (B) SC, overlap of DIC and Fast Green channels; (C) WPI, Fast Green channel; (D) WPI, overlap of DIC and Fast Green channels; the arrows show the HA particles covered by proteins.
464
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
pended stained HA particles prepared with a 0.5% w/w SC solution (Fig. 7A and B) or a 0.5% w/w WPI solution (Fig. 7C and D). The samples were analysed both by using the Fast Green fluorescence channel only (Fig. 7A and C) and by overlapping the fluorescence images with the DIC images (Fig. 7B and D). Fig. 7A shows a typical image of SC-coated HA particles obtained on the Fast Green fluorescence channel. The HA particles were roughly spherical, with a black core surrounded by a green ring, i.e. the protein layer consisting of caseins. The micrographs of the WPI-coated HA particles appeared to be similar to those of the SC-coated particles, but the intensity of the green rings was lower, suggesting a lower amount of whey proteins on the surface of the HA particles and possibly a thinner layer (Fig. 7C). Fig. 7B and D were obtained by overlapping the Fast Green channel signal and the DIC channel signal. The presence of green patches on the white surface of the SC-coated HA particles (Fig. 7B) suggested that the surface was not covered homogeneously and that proteins might be bound preferentially to certain regions of the surface. The WPI-coated particles (Fig. 7D) showed a more uniform coverage of the proteins.
3.4. Suspension behaviour Fig. 8 shows photographs of HA pellets mixed with increasing concentrations of SC. The pellets were rinsed twice with water, were resuspended in water (10% w/w) and were left undisturbed for 24 h at room temperature. The number of particles remaining in suspension after 24 h increased with increasing SC concentration. A similar pattern was observed for WPI-coated particles (images not shown). This improvement in particle suspension was characterised by spectrophotometric measurements as a function of time. Fig. 9 shows the reduction in absorbance with time of 0.125% w/w suspensions of HA particles made with different concentrations of SC (Fig. 9A) or WPI (Fig. 9B). The reduction in absorbance was calculated using (At/A0) 100, where A0 is the initial absorbance and At is the absorbance at time t. Within 200 min, the absorbance of a suspension of HA powder (control with no protein bound) decreased to less than 5% of its original value. All the particles had sedimented to the bottom and the absorbance measured was close to the absorbance of the suspending medium (water). The absorbance of HA suspensions made with different SC or WPI solutions decreased less rapidly compared with the control, and this effect increased with increasing SC or WPI concentration. At low protein concentration (0.5% w/w), the improvement in suspension stability was better for WPI-coated
particles than for SC-coated particles. However, the suspension stability of the HA particles that were fully covered by proteins (HA suspensions made with 3% w/w SC and 3% w/w WPI) was better for particles covered with SC than for particles covered with WPI (45% of the original absorbance value for SC versus 35% for WPI).
4. Discussion The adsorption isotherms (Fig. 2) and the confocal micrographs (Fig. 7) clearly showed that milk proteins from SC and WPI adsorbed at the surface of HA particles, increasing their negative surface charge (Fig. 1). Devold et al. [18] have shown that, at neutral pH, casein molecules are able to bind to tooth enamel, mainly composed of HA. The adsorption on to HA particles of a large range of acidic and basic proteins, including many individual milk proteins, has also been studied by Reynolds and Wong [22]. They showed that as1-, b- and j-caseins, b-lactoglobulin and a-lactalbumin all bound to HA at pH 7, increasing the zeta-potential from 9.1 mV to more negative values between 13.1 mV (for b-lactoglobulin) and 24.1 mV (for as1-casein). The good correlation between the increase in zeta-potential and the increase in surface protein load (Fig. 3) confirmed that protein adsorption was responsible for the increase in surface charge. Small differences in the absolute values of the zeta-potential (at a plateau level) between SC-coated particles and WPI-coated particles may reflect differences in the charge densities of the protein molecules and/or differences in the conformations of the adsorbed proteins [23]. The general mechanism of proteins binding to HA is complicated. The surface of HA particles is heterogeneous and has multiple binding sites. Moreover, the surface morphology and the degree of crystallinity are very dependent on the process used for HA production [4]. Two different types of binding sites on the crystal surface of HA have been identified, and are usually called the C-site and the P-site [20,24]. After dispersion in an aqueous medium, the C-site is rich in calcium ions and positive charges and the P-site is rich in phosphate ions and negative charges. Therefore, HA can potentially bind to both positively and negatively charged regions of proteins. C-sites have been shown to interact with the carboxyl groups of proteins and P-sites have been shown to interact with the amino groups of proteins [6,25]. The initial interaction between these two pairs of groups is believed to be electrostatic in nature. However, it is known that the affinity of calcium ions in the C-site of HA for negatively charged groups in proteins (carboxyl) is higher than the affinity of the phosphate ions in the P-site for positively charged protein residues (amino groups) [20]. This is because coordination complexes are
Fig. 8. Suspension stability as a function of the SC concentration, after standing for 24 h. HA pellets were suspended in (O) water, (A) 0.5% w/w SC, (C) 1% w/w SC, (D) 1.5% w/ w SC and (F) 3% w/w SC, were rinsed with water and were resuspended in water.
465
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
formed between the C-sites and the carboxyl groups of the proteins, and those complexes are held together by much stronger interactions than the non-specific electrostatic interactions occurring between amino groups and phosphates. Therefore, the binding of both caseins and whey proteins to the surface of HA particles is likely to be driven by a mixed ion pairing between the C-sites of HA and the carboxyl groups of the milk proteins and, to a lesser extent, between the P-sites of HA and the amino groups of the milk proteins. HA chromatography is based on the same principle, and the mechanism of protein binding to HA columns has been explained and referred to as a mixed-mode ion-exchange mechanism [6,26]. In addition, casein molecules are phosphorylated proteins. as1-Casein, as2-casein and b-casein have a high content of phosphate groups (phosphoserine residues) that are not uniformly distributed in their sequences and are grouped in clusters [27]. Phosphoserine residues play a crucial role in the structure of casein micelles because they are able to bind cations, especially Ca2+ [28]. Because of their strong calcium-binding ability, they have also been shown to bind to the C-sites of HA [29]. Phosphoryl groups on proteins can interact with C-sites even more strongly than carboxyl groups [30]. Therefore, for SC, the possibility of a specific interaction between negative phosphoserine clusters in casein molecules and the C-site of HA has to be considered. This would explain the higher affinity of the caseins for HA particles compared with the whey proteins, as shown by the higher affinity constant of the Langmuir model and the higher maximum surface coverage (Table 1, Figs. 2 and 6). The involvement of the phosphoserine groups of the caseins in casein/HA interactions was also given as an explanation for the inhibitory action of as1- and b-casein on HA crystal growth by van Kemenade and de Bruyn [16]. They showed a strong complexation between the phosphoserine residues of as1- and b-casein and the growing calcium phosphate phase, leading to the formation of octacalcium phosphate at the expense of HA. The different adsorption levels of caseins and whey proteins may be also due to the different structures of the two proteins. Whey proteins have a small compact globular structure [31], probably form only a monolayer on the HA surface after adsorption, as seen by confocal microscopy (Fig. 7), and do not interact with each other once adsorbed. In contrast, caseins have disordered linear and flexible structures and they can aggregate to form selfassembled particles [32]. The charged groups of the caseins (mainly phosphoseryl and carboxyl groups) would therefore
interact with the HA surface, but the caseins would also associate with each other and form aggregates that would stay at the surface, resulting in a higher surface load and a thicker layer. Unfolded proteins often have lower affinity for HA because they have lower clustering of the carboxyl groups [29]. The preferential adsorption of b-casein to the HA surface (Figs. 4A and 5) may be related to the distribution of the phosphoseryl groups on the casein molecule. The N-terminus of bcasein contains all the phosphoseryl groups of the molecule and is highly negatively charged, whereas the rest of the molecule carries no net charge. Luo and Andrade [20] showed that the geometric distribution of charges on proteins plays an important role in their affinity for HA. The negative head carried by b-casein may therefore explain its preferential adsorption. The phosphoserine-rich regions of b-caseins might also have been more accessible than the phosphoserine residues of as1- and as2-caseins in our system, allowing more interactions between b-casein and the HA surface. j-Casein, because of its very low level of phosphoserine residues, was poorly adsorbed. The charge distribution over the surfaces of b-lactoglobulin and a-lactalbumin have been shown to be different, with a-lactalbumin carrying one main positive region and b-lactoglobulin carrying positive regions spread evenly along its surface [33,34]. The preferential adsorption of b-lactoglobulin over a-lactalbumin on to the HA surface might therefore be explained by a difference in the geometric distribution of the charges of the whey proteins. However, kinetics aspects could also explain the preferential adsorptions observed in this study, for both SC and WPI. For example, as there is more b-lactoglobulin than a-lactalbumin in WPI, b-lactoglobulin might have been able to reach the HA surface first. In this case, the adsorption of each protein from the WPI mixture would be determined primarily by the order and the rate of arrival at the interface, rather than by a difference in their molecular structures [35]. Kinetics aspects are also likely to partially explain the rate of adsorption of the individual caseins. Cooperative effects due to the lateral interactions between casein molecules once adsorbed at the surface could also be involved [20]. However, further investigation on the adsorption of the individual caseins and whey proteins on to HA would be necessary to fully understand the surface composition of HA particles covered by milk proteins. The modelling approach was useful to characterise the adsorption process and to confirm its characteristics. The simple Langmuir model considered that the HA surface was homogeneous, with the same adsorption energy for each site, and showed
(B)100
80
80
60
60
40
40
20
20
Variation of absorbance (%)
(A) 100
0
0 0
50
100
Time (min)
150
200
0
50
100
150
200
Time (min)
Fig. 9. Absorbance as a function of time of suspensions of HA particles (0.125% w/w) made with solutions of different (A) SC or (B) WPI concentrations: ( suspension in water; ( ) HA suspension made with 0.5% w/w SC or WPI; ( ) HA suspension made with 1% w/w SC or WPI; ( suspension made with 2% w/w SC or WPI; ( ) HA suspension made with 3% w/w SC or WPI.
) HA ) HA
466
L. Tercinier et al. / Journal of Colloid and Interface Science 394 (2013) 458–466
a good fit for both the SC data and the WPI data. However, the surface of HA has been reported to be heterogeneous [20]. The difference reported in Table 1 between the heterogeneity parameter n in the Langmuir–Freundlich model for the WPI data (n = 1) and the SC data (n < 1) and the slightly better fit of the SC data with the Langmuir–Freundlich model (Fig. 6) may be explained by the difference in protein size and structure. As whey proteins have a compact and globular structure, they may not be affected by the surface heterogeneity because they can easily access the complete surface. However, as caseins are aggregated, surface heterogeneity may become important and some areas of the surface may become inaccessible for adsorption. Unlike the Langmuir model, the Langmuir–Freundlich model assumptions allowed interactions between adsorbed proteins at the HA surface. As interactions between caseins adsorbed at the surface are likely to occur when they associate and form aggregates, the Langmuir–Freundlich model was deemed to be a better choice for characterising the SC adsorption data. Nano- or micro-sized particles of HA are known to form spontaneous aggregates when suspended in aqueous media, mainly because of interparticle van der Waals interactions and hydrogen bonding [36]. As the zeta-potential in water of the particles used in this study was low (11 mV), they were not stable in water and some aggregation between particles occurred, leading to an increase in their average size and a fast sedimentation. The binding of proteins to the surface of the particles led to an increase in their surface charge (Fig. 3). As the surface charge became more negative, the repulsive forces between the particles increased, rendering them less prone to aggregation and sedimentation over time (Fig. 8). As the adsorption of caseins from SC led to a slightly more negative zeta-potential than the adsorption of whey proteins from WPI when the maximum surface concentration was reached (28 mV versus 22 mV), the improvement in suspension stability was slightly better for SC-coated particles (Fig. 9). Although this improvement can be referred to as electrostatic stabilisation, in the case of SC, steric stabilisation may also play a role. The charged groups of the caseins interacted with the surface of the particles, and the hydrophobic regions of the caseins may have interacted with each other, forming large aggregates on the particle surface. A multi-layer of caseins may have been formed, providing steric hindrance between the particles and reducing the rate of aggregation.
5. Conclusions This study showed that both caseins and whey proteins can adsorb to food-grade HA particles, changing their colloidal and suspension properties. The protein-coated particles were more negatively charged and more stable in suspension, because a slower sedimentation rate was observed over time when the particles were resuspended in water. Caseins had a higher affinity for HA and were able to adsorb more at the HA surface. The nature of the adsorbed layer was different between the two protein sources. The higher surface load observed for SC-coated particles and the brighter green intensity observed for the protein layers in the confocal pictures suggest that the aggregates of associated casein molecules formed a thick layer, whereas a low surface load and a less intense colour in the confocal pictures for the WPIcoated particles implies a thin single layer of protein on the HA surface. The adsorption mechanism was explained by specific interactions between the carboxyl and phosphoseryl groups of the proteins and the calcium sites on the HA surface and to a lesser
extent between the amino groups of the proteins and the phosphate sites on the HA surface. In conclusion, this work demonstrates the possibility of milk proteins binding to HA particles or other calcium phosphate particles. As the behaviour of HA was shown to change, this could be potentially of interest, for example, as a way to increase the stability of HA particles added to dairy beverages or in the development of solutions to control sediment formation and fouling on heat exchangers. Acknowledgments The authors gratefully acknowledge Dr. Barbara Kuhn-Sherlock and Quentin Béchet for their help with modelling and statistical analysis, Jianyu Chen from the Manawatu Microscopy and Imaging centre for her help with confocal microscopy, Claire Woodhall for proofreading the manuscript and Fonterra Research Centre for their financial support. References [1] H. McDowell, T.M. Gregory, W.E. Brown, J. Res. Nat. Bur. Stand. 81A (1977) 273. [2] E.D. Omoarukhe, N. On-Nom, A.S. Grandison, M.J. Lewis, Int. J. Dairy Technol. 63 (2010) 504. [3] M.J. Lewis, Int. J. Dairy Technol. 64 (2011) 1. [4] J. Norton, K.R. Malik, J.A. Darr, I. Rehman, Adv. Appl. Ceram. 105 (2006) 113. [5] V. Uskokovic´, D.P. Uskokovic´, J. Biomed. Mater. Res. B Appl. Biomater. 96 (2011) 152. [6] M.J. Gorbunoff, S.N. Timasheff, Anal. Biochem. 136 (1984) 440. [7] J.R. Sharpe, R.L. Sammons, P.M. Marquis, Biomaterials 18 (1997) 471. [8] M. Rouahi, E. Champion, O. Gallet, A. Jada, K. Anselme, Colloids Surf. B Biointerfaces 47 (2006) 10. [9] L. Borum, O.C. Wilson Jr., Biomaterials 24 (2003) 3681. [10] N. Brandes, P.B. Welzel, C. Werner, L.W. Kroh, J. Colloid Interface Sci. 299 (2006) 56. [11] M. Iafisco, M. Di Foggia, S. Bonora, M. Prat, N. Roveri, Dalton Trans. 40 (2011) 820. [12] X.D. Zhu, H.S. Fan, D.X. Li, Y.M. Xiao, X.D. Zhang, J. Biomed. Mater. Res. B Appl. Biomater. 82 (2007) 65. [13] M. Johnsson, C.F. Richardson, J.D. Sallis, G.H. Nancollas, Calcif. Tissue Int. 49 (1991) 134. [14] V. Sternitzke, R. Kaegi, J.-N. Audinot, E. Lewin, J.G. Hering, C.A. Johnson, Environ. Sci. Technol. 46 (2012) 802. [15] L. Borum-Nicholas, O.C. Wilson Jr., Biomaterials 24 (2003) 3671. [16] M.J.J.M. van Kemenade, P.L. de Bruyn, J. Colloid Interface Sci. 129 (1989) 1. [17] C. Ritzoulis, N. Scoutaris, K. Papademetriou, S. Stavroulias, C. Panayiotou, Food Hydrocolloids 19 (2005) 575. [18] T.G. Devold, M. Rykke, D. Isabey, E.S. SØrensen, B. Christensen, T. Langsrud, C. Svenning, B. Borch-Iohnsen, J. Karlsen, G.E. Vegarud, Int. Dairy J. 16 (2006) 1013. [19] M.J. Mura-Galelli, J.C. Voegel, S. Behr, E.F. Bres, P. Schaaf, Proc. Natl. Acad. Sci. USA 88 (1991) 5557. [20] Q. Luo, J.D. Andrade, J. Colloid Interface Sci. 200 (1998) 104. [21] M. Wahlgren, T. Arnebrant, Trends Biotechnol. 9 (1991) 201. [22] E.C. Reynolds, A. Wong, Infect. Immun. 39 (1983) 1285. [23] B.S. Chu, S. Ichikawa, S. Kanafusa, M. Nakajima, J. Sci. Food Agric. 88 (2008) 1764. [24] K. Kandori, T. Kuroda, S. Togashi, E. Katayama, J. Phys. Chem. B 115 (2011) 653. [25] M.J. Gorbunoff, S.N. Timasheff, Anal. Biochem. 136 (1984) 433. [26] A. Jungbauer, J. Chromatogr. A 1065 (2005) 3. [27] P.F. Fox, P.L.H. McSweeney, Advanced Dairy Chemistry, Kluwer Academic/ Plenum Publishers, New York, 2003. [28] D.S. Horne, in: A. Thompson, M. Boland, H. Singh (Eds.), Milk Proteins: From Expression to Food, Academic Press/Elsevier, San Diego, 2009, p. 133. [29] G. Bernardi, T. Kawasaki, Biochim. Biophys. Acta 160 (1968) 301. [30] T. Kawasaki, J. Chromatogr. A 544 (1991) 147. [31] E. Dickinson, Colloids Surf. B Biointerfaces 20 (2001) 197. [32] M. Srinivasan, H. Singh, P.A. Munro, Int. Dairy J. 9 (1999) 337. [33] R. de Vries, J. Chem. Phys. 120 (2004) 3475. [34] E. Casal, A. Montilla, F.J. Moreno, A. Olano, N. Corzo, J. Dairy Sci. 89 (2006) 1384. [35] E. Dickinson, Food Hydrocolloids 25 (2011) 1966. [36] Z. Sadeghian, J.G. Heinrich, F. Moztarzadeh, Ceram. Int. 32 (2006) 331.