Casein Micelles Structure Configuration Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Up to date various models have been proposed to configure the structure of casein micelles. The first model, called "coat-core" model, was described by Waugh & Noble (1965). This involves a core consisting of α- and/or β-caseins enveloped by a coat of κ-casein. Thus, the micelle size depends on the proportion of caseins between the coat and the core. The "coat-core" model hypothesizes that the exterior (κ-casein) prevents the interior (α- and β-caseins) from precipitation; however, it can not explain the stabilizing role of colloidal calcium phosphate (CCP) (Waugh & Noble, 1965; Payens, 1966; Parry Jr & Carroll, 1969).

Submicelle model, which was first introduced several decades ago, has been accepted by many authors (Lin et al., 1971; Slattery & Evard, 1973; Schmidt, 1982; Walstra, 1990). According to the theory of "submicelle" model, the casein micelle is roughly spherical, composed of smaller proteinaceous subunits, called submicelles, which have variable composition. Submicelles have two major types: one including αs and β-caseins (the κ-casein poor submicelle), and the other consisting of αs and -caseins (κ-casein rich submicelle) (Walstra, 1999). Submicelle contains a hydrophobic core and is covered by a hydrophilic coat with κ-casein existing as a hairy layer (Walstra & Jenness, 1984). These subunits link together via CCP and thus integrate into a micelle, where the κ-casein poor submicelles are located in the interior of the micelle while those rich in κ-casein surround the surface, giving the whole micelle a κ-casein-rich surface layer (called "hairy" layer) (Schmidt, 1982). The hydrophilic layer plays an important role in micelle stability, such as preventing any further aggregation of submicelles or protecting micelles from flocculation by steric and electrostatic repulsion (Walstra, 1999). This model is actually consistent with the fact that the loss of the hairy layer leads to the lessening in micellar hydrodynamic diameter (Dalgleish et al., 2004).

Figure 2. 1 The schematic of the submicelle model of the casein micelle (Horne, 2006)

Another category of models involved the properties of the individual casein constituents, causing or directing the formation of the internal structure of the casein micelle (Wong, 1988). The first internal structure model accounts for the formation of a porous network of proteins due to the interactions between caseins (Rose & Colvin, 1966b; Ribadeau Dumas, 1968). Thus, the casein distribution in the "internal structure" model is more throughout the micelle than that in the "coat-core" model. The size of micelle is based on the component of caseins. This model debates the hypothesis that κ-casein resides on the micellar surface (Ladisch, 1987). Disagreement with the notion of submicelle model, Holt et al. (1996) enhanced an alternative internal structure that casein micelles were some how spherical, highly hydrated, and fairly open particle. While the submicelle model emphasizes the role of hydrophobic interactions on micelle size, the Holt model believes that the interactions between the caseins and calcium phosphate function holding the micelle together. Later on, De Kruif & Holt (2003) refines this model in which the main caseins are present through the micelle, forming a homogeneous web of caseins; and the subunits are calcium phosphate nanoclusters distributing in the centres from which the casein micelles grow. This model considers κ-casein unessential and therefore fails to explain how this molecule controls the micelle size or makes a location where it can control micelle stability (Walstra, 1999; Horne, 2006).

Figure 2. 2 The schematic of network formation in the Holt model (Horne, 2006)

Better surface images have recently recorded using a cold field-emission ultra high resolution scanning electron microscope (FESEM) by Dalgleish et al. (2004). They revealed casein micelle structure as a rough, tubular surface structure rather than spherical subunits. In another study, Pignon et al. (2004), who presented small-angle and ultra small-angle x-ray scattering (SAXS and USAXS) measurements of casein micelles over an exceptionally wide scattering vector range, showed that the larger structures consisted of globular objects and the smaller structures were branch-like structures where casein micelles were organized as an entangled network. In a similar approach, Marchin et al. (2007) concluded from SAXS/ USAXS studies that casein micelles comprised a complex network of protein chains, with smaller substructures being colloidal calcium phosphate nanoclusters rather than submicelles. Using differential centrifugation to prepare size-fractionated casein micelles and observing their structure by using cryo transmission electron microscopy (cryo-TEM), the authors indicated that the largest casein micelles were more polydispersed and spherical in shape while the smallest ones appeared as an elaborate structure and therefore their boundary was difficult to recognize. In an agreement with Pignon et al. (2004), they assumed that the structure of micelles with different sizes might be made up by different steps of building. Large micelles could be developed at the highest polymerization stage, whereas smaller particles could be formed at intermediate stages of building.

Figure 2. 3 Electron micrograph of an individual casein micelle, made using the technique of field-emission scanningelectron microscopy without metal coating (Dalgleish et al., 2004).

In general, the structure of casein micelles is still a matter of debate since many microscope technologies have been introduced. Due to the experimental limitations, previous methods used to investigate the micellar structure could not offer a complete understanding of casein micelle characterization. Take previous small-angle x-ray scattering (SAXS) technique as an example (Kumosinski et al., 1988), because of technical limitations regarding the radiation wavelength responding to the total particle size, only a small amount of the casein micelles in the whole fraction is studied and thus the result is given inadequate (Marchin et al., 2007). Similarly, scanning electron microscopy provides no information about the micelle surface structure since artifacts may add up during the sample preparation process, such as staining, fixation, exchanging water for ethanol, air drying, or metal coating (McMahon & McManus, 1998). Nowadays, the improvement of the visualization techniques has led electron microscopy a powerful tool to clarify the structure of colloidal caseins. Depending on the purpose of observation of casein micelle structure, some of the following methods can be used. The surface images can be observed by using scanning electron microscopy without metal coating (Dalgleish et al., 2004) while cross-sections of the internal structure can be seen by TEM of freeze-fractured cryoprotected casein micelle suspensions (Karlsson et al., 2007). Total images (surface and internal images) can be figured out by TEM of freeze-dried surface immobilized casein micelles without resin embedding and sectioning (McMahon & McManus, 1998; McMahon & Oommen, 2008) and by cryo-TEM of thin vitrified films of casein micelle suspensions (Marchin et al., 2007).


Casein micelles are found approximately 92% of protein and 8% of mainly inorganic, salts, calcium phosphate. The caseins, known as predominant components in micelles, are present four main types: αs1-, αs2-, β- and κ-caseins (Whitney, 1988; Swaisgood, 2003).

The composition of the individual caseins varies according to the micellar size (Park et al., 1999) whereas smaller micelles has a higher amount of κ-casein compared to larger micellars (Ekstrand & Larssonraznikiewicz, 1978; Ekstrand et al., 1980; Davies & Law, 1983; Donnelly et al., 1984; McNeill & Donnelly, 1987; Dalgleish et al., 1989; Pierre et al., 1999; O'Connell & Fox, 2000; Marchin et al., 2007). Up to date, there has been a number of opposite evidences regarding the relationship between individual casein components and the size of casein micelles. By using quantitative column chromatography on hydroxyapatite, McGann et al. (1980) and Donnelly et al. (1984) reported that the dercrease in micellar size was followed by the increment of β- and κ-casein contents, opposed to the proportion of α-casein. In contrast, Ekstrand & Larssonraznikiewicz (1978) showed that the κ-casein increased with decreasing the size of micelles from the large to the medium size but then dropped in the smallest fraction by using diethylaminoethyl (DEAE)-cellulose ion exchange chromatography. These results, however, are questionable because the method applied to determine the individual casein proportion does not completely resolve the whole casein (Swaisgood, 2003). Defense against the previous findings has been suggested by a number of researchers, Dalgleish et al. (1989), who employed fast protein liquid chromatography (FPLC) for quantification of proteins, supported that smaller micelle fractions had a higher amount of κ-casein and lower quantity of β-casein compare to larger micelle fractions. Interestingly, the proportion of α-casein was size independent. These results are in agreement with those by Marchin et al.(2007), who used reversed phase-HPLC (RP-HPLC) analysis of the casein fractions. Moreover, Dalgleish et al. (1989) emphasized that the content of calcium and inorganic phosphate were also largely independent of the micellar size.

In term of the influence of microfiltration with various pore size membranes on casein compostition, Tziboula et al. (1998) showed that although there was no significant difference in the total casein content between the retentates or the permeates obtained from 0.8 and 1.4 µm membranes, the concentration of caseins in the retentates was higher than that in the permeates whereas the proportion of whey protein remained the same at either permeate or retentate stream. The authors also indicated that the permeates had lower β-and αs2 casein contents and higher amounts of - and αs1 caseins than the retentates was logical with the changes of micelle size in both filtrates. Nevertheless, experimental analysis proved the differences were mirror, putting the selectivity of 0.8 and 1.4 µm membranes to separate native phosphocaseinate in different sizes on questions.

As technical methods concerned, electrophoresis and chromatography are the two major methods to determine the protein components in the casein micelle fractions.

Basically, the identification of milk proteins can be specified by gel electrophoresis. The electrophoresis in either starch (SGE) or polyacrylamide (PAGE) gels are widely used for casein studies since they both give similar analytical results (Wake and Baldwin, 1961; Peterson, 1963). However, PAGE is to some extent easier to use and thus has become a standard electrophoretic method for analysis of caseins (Fox, 2003). Currently, SDS-PAGE analyser can separate the four bovine caseins in the presence of a reducing agent and may provide several distinct bands equivalent with increasing mobility (αs1-, αs2-, β- and κ-casein) (Tremblay et al. 2003). However, αs1 and β-caseins behave atypically, leading to the higher Mrs values. β-casein, which has very high surface hydrophobicity, binds a disproportionately high quantity of SDS and, subsequently, has a higher electrophoretic mobility than αs1-casein, although it is a larger molecule (Creamer & Richardson 1984).

It is evident that because the quantification of protein-stained zones following gel electrophoresis is quite difficult, the high resolution of chromatographic methods appears to be better than that obtained with the former techniques (Swaisgood, 2003). Initially, (DEAE)-cellulose ion exchange chromatography (Ekstrand & Larssonraznikiewicz, 1978) was used for determination of the amount of different caseins a milk sample but appeared inefficient due to incomplete resolution. Whole casein could also be fractionated by quantitative column chromatography on hydroxyapatite (McGann et al., 1980; Donnelly et al., 1984); however, resolution does not seem good either. Recently, ion exchange FPLC has been proven a powerful method for the studies of the quantification of various caseins in the whole protein, with a good resolution for αs1-, αs2- and κ-caseins (Dalgleish et al., 1989). An alternative method to ion exchange, known as RP-HPLC, has been widely used for separation of individual caseins on C4, C8, and C18 reveresed-phase columns (Strange et al., 1992; Bobe et al., 1998; Veloso, Teixeira & Ferreira, 2002).

Physicochemical properties

Particle size and size distribution

The size and size distribution of casein micelles has been investigated by numerous literatures. Regarding the measurement of casein micelles, there are two major methods used to determine micelle size: electron microscopy and light scattering. Other methods ordinarily combine fractionating casein micelles by size, such as ultracentrifugation and chromatography, and measuring average sizes using sedimentation, diffusion and light scattering technique (Ladisch, 1987).

Electron microscopy is considered an excellent technique that allows direct examination of the particles. Table 2.1 summarizes the results of casein micelle size obtained by various methods. The table points out the variances in the values obtained by using microscopy technique (Karlsson et al., 2007; Donnelly et al., 1984; Rose & Colvin, 1966a; Carroll et al., 1968; Ekstrand & Larssonraznikiewicz, 1978; Ekstrand et al., 1980; McGann et al., 1980; Dalgleish et al., 1989; Holt et al., 1973; McNeill & Donnelly, 1987; Marchin et al., 2007). Most of researchers note the average particle diameter of 100 nm (Holt et al., 1978; Walstra, 1990; Marchin et al., 2007) but the range of micelle size is readily different. Some authors describe the wide range of size distribution, from 20 up to 600 nm (Schmidt, 1982; de Kruif, 1998) while the others give a decreasing distribution, which is from 50 to 170 nm (Pierre et al., 1999; Carroll et al., 1968; Holt et al., 1978; McGann et al., 1980; Donnelly et al., 1984). Recently, the new microscopic techniques has been introduced, making the experiments more reliable owing to the ability to detect very small particles of these methods. With the use of cryo-TEM, Marchin et al. (2007) reported that micelle size had a range of 50-300 nm, the same as previous study by de Kruif (de Kruif, 1998; Marchin et al., 2007).

Light scattering measurements, on the other hand, give larger values of the average diameter than those obtained by electron microscopy. By using dynamic light scattering method in combination with an ultracentrifugal procedure, the diameter of micelle fractions was recognized from 125 to 487 nm (Niki et al., 1994). Yet, in another study by Marchin et al. (Marchin et al., 2007) the result is relatively different, with narrower mean dimensions of 112 to 288 nm. It might be explained that although the authors use the same size measurement method, the different sample preparation procedures could result in various values. Also, it is recommended that the mean diameter obtained by electron microscopy indicates number-length mean whereas that received by dynamic light scattering is the average hydrodynamic diameter; therefore, different techniques give different means.

To sum up, although the results given by using microscopy allow to detect very small particles, sample preparation for electron microscopy is laborious and slow. Moreover, when once few particles are chosen as representative sampling to measure the mean diameter of particles, the results could be exaggerated. Dynamic light scattering method, though can give an error if there occurs a number of strange particles like fat globules in the samples, has been considered a well established method for size analysis due to its rapidity, repeatability for reliable results; and high resolution.

Zeta Potential

It is well known that the net charge on the surface of the micelle has contributed to the stability of the casein micelle. The surface charge could be figured out through the zeta potential. Thus, zeta potential measurements are used to assess the stability of colloidal systems. Since casein micelles carry a negative net charge, their zeta potential is negative. An increase or a decrease of zeta potential discussed below refers to its absolute value.

The zeta potential of native casein micelles in milk system at natural pH (pH~6.8) was between -20 mV to -17 mV, depending on the medium and the measurement method (Dalgleish, 1984; Wade et al., 1996; Michalski et al., 2002). Michalski et al. also put a note that the unfiltered skim milk had lower zeta potential than the filtered skim milk, which were -12.9 and -20.1 mV, respectively. The apparent zeta potential for milk with fat content of 1 g/kg was reported approximately -16 mV. If casein micelles were redispersed in SMUF (simulated milk ultrafitrate), this value of casein micelles was found more or less -18mV, close to that in milk (Anema & Klostermeyer, 1996). Therefore, it is said that resuspension of casein micelle in SMUF does not significant impact on the stability of micelles.

For the analysis of the colloidal stability, it was suggested that the dispersion of casein micelles in natural milk serum, milk ultracentrifugate should be avoided because the presence of some non micellar caseins or whey proteins might contribute to the unreliable or misleading results for the properties of micelles (Horne & Davidson, 1986). This explanation was also confirmed by Darling & Dickson (1979), who showed the low ζ-potential values of -10.19 mV and -15.5 mV, corresponding to the micelle dispersions of milk ultrafiltrate and milk centrifugate.

In the literature, the investigation of environmental conditions, such as pH, salt concentration and temperature, impacting on the zeta potential of casein micelles has been the focus of many workers. In general, the absolute ζ-potential tends to decrease by the addition of salts, ion Ca2+ (Darling & Dickson, 1979; Dalgleish, 1984; Holt & Dalgleish, 1986; Anema & Klostermeyer, 1996; Rabiller-Baudry et al., 2005), lessening pH (Darling & Dickson, 1979; Anema & Klostermeyer, 1996; Anema & Li, 2003; Anema et al., 2005) or decreasing temperature (Dalgleish, 1984; Darling & Dickson, 1979). These are beyond the scope of this review and thus will not be discussed further.

The characterization of the casein micelles with different sizes regarding the zeta potential has been reported restrictively. Darling & Dickson (1979) found that there were no significant difference between the electrophoretic mobility values of large casein micelles in artificial serum and those of small ones. This hypothesis was in conformity with the information obtained by O'Connell & Fox (O'Connell & Fox, 2000), who proposed that the zeta potential of casein micelles was independent of their size. However, the authors' conclusion contradicted the data, in which the zeta potential of the intermediate micelles (-17.4 mV) was higher than that of the large and small ones (the average value of approximately -15mV). This fluctuation might be due to the use of an inappropriate buffer, the ultracentrifugate, to disperse the casein micelles (Horne & Davidson, 1986). On the contrary, Park et al.(1999) concluded that the absolute zeta potential decreased as the size decreased, with -7.4 mV, -6.1 mV and -1.1mV for the large, medium and small casein micelles dispersed in artificial serum, respectively. In a similar approach, Hartanto (2007) showed the higher values, which were from -16.7 mV (for the largest micelles) to -10.7 mV(for the smallest micelles). However, the zeta potential of control milk was -11.2 mV (Hartanto, 2007), lower than that reported by Michalski et al. (2002). As the author mentioned, the possibility of this relatively low value was that because the small amount of fat globules still stayed in the sample, the zetasizer was unable to detect differences between both signals and proposed an average value; therefore, influenced the calculated zeta potentials.

The comparision of the absolute zeta potential resulted from the different literatures is something of a hard work owing to the differences in sample preparation procedures, environmental conditions and the variation of approach to calculate the zeta potential. It is recommended that since the zeta potential is calculated from the electrophoretic mobility using the Henry equation and only an empirical equivalent to the surface potentials, zeta potential is only a good approximation for complex systems (Michalski et al., 2002). However, zeta potential measurements are considered an effective tool for comparative studies and have been used by a numer of researchers.

The electrophoretic mobility of casein micelles can be determined by using several techniques, such as electro-osmosis (Payens, 1966), moving boundary electrophoresis (Darling & Dickson, 1979; Ekstrand & Larssonraznikiewicz, 1984), electroacoustics or electrokinetic sonic amplitude (Wade et al., 1996) and laser doppler electrophoresis (Dalgleish, 1984; Holt & Dalgleish, 1986; Anema & Klostermeyer, 1996; Park et al., 1999). The summary of the previous studies can be listed in table 2.2.


In the literature, the voluminosity (v) values of casein micelles were substantially fluctuated between different methods of determination (Thompson et al., 1969; Lin et al., 1971; Dewan et al., 1973; Schmidt et al., 1973; Richardson et al., 1974; Holt, Parker & Dalgleish, 1975; Bloomfield, 1976; Sood, Sidhu & Dewan, 1976; McMahon & Brown, 1984; Walstra, 1990). After applying some corrections, Walstra (1979) summarized that methods based on hydrodynamic radius gave a value of approximatively 3.9 ml/g dry casein whereas other methods (microscopy, sediment volume) about 2.2 ml/g dry casein. Methods depending on hydrodynamic radius (light scattering, diffusion coefficient and intrinsic viscosity) gave a much higher voluminosity than the others mentioned. The variance could be due to the fact that the micelles were not completely globular. Assuming that the casein micelle structure derived from submicelles, the hydrodynamic thickness of the hairy layer on the micelle surface would considerably increased the specific hydrodynamic volume of the whole casein micelle compared to that of the core particles of the typical micelle measured by microscopy or sediment hairy.

The voluminosity of the micelles markedly increased with decreasing tempature (Dewan et al., 1973; Walstra, 1990), the average voluminosity of about 4.0 at 25 °C and 3.0 at 50°C. The increase in voluminosity may be due to swelling of the micelle core and a limited disintegration of micelles into smaller ones. However, Walstra (1990) suggested that the data below 20oC were uncertain, showing the conversed trend because of the partial dissolution of β-casein at low temperature (Schmidt, 1982). The voluminosity of the casein micelles also decreased when pH was adjusted from 6.5 to 5.5 and increased in the presence of NaCl (Karlsson et al., 2005). The alteration could be explained by the loss of the -casein layer on the surface of micelle, the altered ionic strength and the change of the average distance between casein. Interestingly, when pH went down to 5.2, the voluminosity was reversed to increased, indicating that there was an acid-induced aggregation of the casein micelles. Regarding the ultrafiltration of skim milk to produce the casein micelle concentrates, Karlsson et al. (2005) reported that during concentration the water was removed and the distance between the casein micelles decreased, leading to the decrease in the voluminosity and altered interactions between them.

Concerning the voluminosity as a function of casein micelle size, Dewan & Bloomfield (1973) indicated that size-fractionated casein micelles by rate -zone ultracentrifugation were uniform spheres of a large, constant voluminosity. On the contrary, Satio & Yasuo (Satio & Yasuo, 1981) used three ultracentrifugal steps (10,000 - 47,000 - 107,000g at 0 - 1oC for 60 minutes) to produce three different fractions of casein micelles. The voluminosity of intermediate fraction was markly higher than that of the large and small micelles, with about 5.0, 4.18 and 4.32, respectively. Assuming that the water content was partially squeezed out of the micelles during centrifugation, the authors indicated the uncompressed state of medium casein micelles during centrifugation or the recovery of compressed fractions of intermediate micelles after removing centrifugal force. However, the results are unreliable due to the use of inappropriate buffer to redisperse micelles in solution. Debate against the above results, Sood, Sidhu & Dewan (1976) found the apparent voluminosity from relative viscosities varied from 5.0 ml/g for the smallest to 3.55 ml/g for the largest micelles. Based upon the assumption of a "hairy" micelle, the magnitude is equivalent to what would be expected that the larger micelles had lower voluminosities than the smaller ones. The similar conclusion was reported by Donnelly et al. (1984), who showed that the increase in -casein component, which was from 4.8% of total caseins in the large fraction of micelles to 11% in the small fraction, resulted in the increment in micelle voluminosity by about 40%.


Hydration, which plays a role in the stability of casein micelles, is calculated from the amount of water associated with the protein expressed as mass of water per unit mass of protein (Morris et al., 2000). According to Dewan & Bloomfield (1973), the hydration measured from voluminosity data estimated on the basis of dynamic light scattering was 3.7g water per gram protein, about twice as the hydration measured from the water content of casein micelle pellets. The variance was thought that because a certain amount of water was squeezed out of the pellets by the ultracentrifugal force, the hydration value obtained by pellet hydration method was less than that by dynamic light scattering. Using a modern analytical ultracentrifugation combined with capillary viscometry, Morris et al. (2000) reported a hydration of 3.4 g water/g protein which was in accordance with the value from dynamic light scattering.

Hydration decreased with increasing casein micelle size, from 2.8 down to 2.1 g water/ g protein corresponding to the smallest and the largest fractions (O'Connell & Fox, 2000). Nonetheless, the difference was not significant, opening the idea that centrifugation did not transfer the water held within the micelle to the supernatant (Carr et al., 2003). The hydration of casein micelle could also be declined by heating. But it seems that thermal treatment does not markedly impact on the micelle hydration, rather than the hydration reduces during heating as a consequence of the decrement in pH (O'Connell & Fox, 2003).