Journal of Dairy Science
Volume 92, Issue 4 , Pages 1338-1353, April 2009

Review of the chemistry of αS2-casein and the generation of a homologous molecular model to explain its properties1

  • H.M. Farrell Jr.

      Affiliations

    • Dairy Processing and Products Research Unit, USDA, Agricultural Research Service, Eastern Regional Research Center, 600 East Mermaid Lane, Wyndmoor, PA 19038
    • Corresponding Author InformationCorresponding author.
    • ,
  • E.L. Malin

      Affiliations

    • Dairy Processing and Products Research Unit, USDA, Agricultural Research Service, Eastern Regional Research Center, 600 East Mermaid Lane, Wyndmoor, PA 19038
    • ,
  • E.M. Brown

      Affiliations

    • Dairy Processing and Products Research Unit, USDA, Agricultural Research Service, Eastern Regional Research Center, 600 East Mermaid Lane, Wyndmoor, PA 19038
    • ,
  • A. Mora-Gutierrez

      Affiliations

    • Cooperative Agricultural Research Center, Prairie View A&M University, Prairie View, TX 77446

Received 12 September 2008; accepted 18 November 2008.

Article Outline

Abstract 

αS2-Casein (αS2-CN) comprises up to 10% of the casein fraction in bovine milk. The role of αS2-CN in casein micelles has not been studied in detail in part because of a lack of structural information on the molecule. Interest in the utilization of this molecule in dairy products and nutrition has been renewed by work in 3 areas: biological activity via potentially biologically active peptides, functionality in cheeses and products, and nutrition in terms of calcium uptake. To help clarify the behavior of αS2-CN in its structure-function relationships in milk and its possible applications in dairy products, this paper reviews the chemistry of the protein and presents a working 3-dimensional molecular model for this casein. The model was produced by threading the backbone sequence of the protein onto a homologous protein: chloride intracellular channel protein-4. Overall, the model is in good agreement with experimental data for the protein, although the amount of helix may be over-predicted. The model, however, offers a unique view of the highly positive C-terminal portion of the molecule as a surface-accessible area. This region may be the site for interactions with κ-carrageenan, phosphate, and other anions. In addition, most of the physiologically active peptides isolated from αS2-CN occur in this region. This structure should be viewed as a working model that can be changed as more precise experimental data are obtained.

Keywords: casein structure, casein molecular model, protein functionality, value-added milk product

 

Back to Article Outline

Introduction 

αS2-Casein (αS2-CN) comprises up to 10% of the casein fraction in bovine milk; it consists of 2 major and several minor components exhibiting varying levels of posttranslational phosphorylation (Swaisgood, 1992; Farrell et al., 2004) as well as minor degrees of intermolecular disulfide bonding (Rasmussen et al., 1992). Studies of the association behavior of αS2-CN (Snoeren et al., 1980) and of its secondary structure (Hoagland et al., 2001) are limited compared with other caseins. Also, the role of αS2-CN in casein micelles has not been studied in detail. Interest in the use of this molecule in dairy products and nutrition has been renewed by work in 3 areas: biological activity, functionality, and nutrition.

αS2-Casein peptides derived from its C-terminal section have been found, in in vitro assays, to have unique antibacterial properties (Zucht et al., 1995; Recio and Visser, 1999; McCann et al., 2005; Lopez-Exposito et al., 2006). Moreover, αS2-CN peptides with angiotensin-I enzyme inhibitor properties have been identified (Tauzin et al., 2002). In vitro experiments (Tanabe et al., 2006) suggested that αS2-CN peptides could serve to dampen allergenic responses. Thus, αS2-CN-enriched preparations or proteolytic digests may have use in value-added or health-promoting dairy products (Lopez-Exposito and Recio, 2008). In addition, Kizawa et al. (1996) have purified peptides from αS2-CN that inhibit calmodulin-activation of cyclic nucleotide phosphodiesterase, a major regulatory component of a variety of metabolic pathways. The peptides from αS2-CN apparently bind to the regulatory protein calmodulin interfering with its binding to the enzyme. The physiological implications here are unknown, but other peptides from the same area of the molecule may be involved in promoting intestinal health and well being.

The second area relates to work on the functionality of caprine milks. Caprine caseins, in contrast to bovine caseins, vary considerably in the types of casein present: some are poor in αS1-CN and some are richer in αS2-CN. It has been speculated that the differences in αS2-CN content account for the unique physicochemical characteristics of caprine caseins and explain a large part of differences in the technological behavior between caprine and bovine milks (Mora-Gutierrez et al., 1996, 1997; Mora-Gutierrez and Farrell, 2001). In this context, the proportions of αS-CN fractions are in large part responsible for the differences observed between the 2 milks in cheesemaking (Ambrosoli et al., 1988; Grosclaude, 1988; Remeuf et al., 1995). It has also been suggested that caprine caseins rich in αS2-CN offer enhanced sites for interactions with the common food additive κ-carrageenan (Farrell and Mora-Gutierrez, 2006).

Finally, from a nutritional point of view, a recent study compared native and calcium-fortified bovine and caprine cheeses in terms of calcium absorption and deposition in growing male rats (Mora-Gutierrez et al., 2007a). In the latter study, the caprine milk used was low in αS1-CN but high in αS2-CN. Significant differences in calcium absorption were found among the cheeses with the data yielding the pattern caprine calcium-fortified>caprine = bovine calcium-fortified>bovine>control (Mora-Gutierrez et al., 2007a). The enhanced calcium absorption resulted in the production of greater bone mass and correspondingly increased resistance to bone fracture in the rat femurs (Mora-Gutierrez et al., 2007a). In recent work it has been shown that calcium phosphopeptides (CPP) added to milks enhance calcium adsorption and deposition from added calcium carbonate in growing male rats (Mora-Gutierrez et al., 2007b). In particular, the CPP from caprine casein high in αS2-CN was the most effective.

To help clarify the role of αS2-CN in its structure-function relationships in milk and its possible applications in value-added dairy products, this paper reviews the chemistry of the protein and presents a working 3-dimensional (3D) model for the protein.

Back to Article Outline

Materials and Methods 

Sequence Alignments 

All sequence alignments and homology predictions were obtained using the T-coffee program of Notredame et al. (2000) found on the ExPASy Web site under tools-alignment (http://tcoffee.vital-it.ch/cgi-bin/Tcoffee/tcoffee_cgi/index.cgi?stage1=1&daction=TCOFFEE::Regular).

Molecular Modeling of Bovine αS2-CN 

A quick route to an approximate 3D model is to use another protein with a strong homology and a known 3D structure (Sawyer et al., 2002) as a guide. Such structures for the caseins do not appear to be available through conventional database searches. Results from the T-coffee program of Notredame et al. (2000) suggested that the chloride intracellular channel (CLIC) molecules, although not closely related to αS2-CN, had sufficient homology to proceed with the modeling work. The model was generated from the human CLIC-4 model (Q9Y696_C00001) found at http://swissmodel.expasy.org/repository/smr.php?job=2&sptr_ac=Q9Y696&model_name=Q9Y696_C00001. The Q9Y696 model was initially based on the x-ray template of a soluble fragment of human CLIC-1 (O00299; PDB file 1kOm). Initial alignments and set-up for the model were accomplished using the program Deep View/Swiss PDB Viewer 3.7 obtained from the ExPASy site. After initial manipulations by the author, the model was submitted as a project to SWISS-MODEL, an automated homology-modeling server; such modeling is computer based and produced without human intervention (Kopp and Schwede, 2004). Therefore, the model for αS2-CN provides some basis for caution in that it is based on a secondary homology and initially produced without human intervention. Interpretation of the model must be carefully considered in that it has the characteristics of an x-ray structure but lacks the certitude of direct crystallographic data.

After viewing the original model it was apparent that the putative site for casein-casein interactions predicted by Kumosinski et al. (1994) was partially blocked (this could also represent the membrane interaction site in the CLIC proteins). Several manipulations of the bond angles at the base of the proposed site were attempted and the easiest change to accentuate the movement of the interaction site finally proved to be a minor adjustment of the phi/psi angles of Ala 116 from -60/−60 to -60/0. The pseudo-charge representation of the Ala adjusted molecule used the electrostatic surface calculation of the Deep View program; the parameters for this representation were dielectric constants of the solvent and the protein set at 80 and 5, respectively, the red color represents a value of -1.8esu (electrostatic units), the white 0, and the blue is +1.8esu.

Energy minimization of the molecule was carried out in vacuo using the program GROMOS96 found within the Deep View program from the ExPASy site (Stocker and van Gunsteren, 2000). The default settings with 20 steps of steepest descent and with a nonbonded cutoff of 10Å were used with multiple rounds of minimization. The minimization was considered finished when the change between rounds was <10%. Peptide models from αS2-CN were excised from the final energy-minimized model, capped, and again energy minimized as described above.

Back to Article Outline

Results and Discussion 

General Chemical Properties 

αS2-Casein was the last of the bovine caseins to be sequenced; it also has the most unique primary structure of the caseins (Brignon et al., 1977). The primary structure of the protein has been established by chemical sequencing (Brignon et al., 1977) and confirmed by cDNA sequencing (Stewart et al., 1987) and genomic DNA sequencing (Groenen et al., 1993). As shown in Figure 1, αS2-CN consists of 207 amino acids with a molecular weight from sequence of 25,226 for the A variant with 11 phosphorylated serine residues (Farrell et al., 2004). There are 4 known genetic variants; the A variant predominates in western cattle (Farrell et al., 2004). Unlike the other caseins, which are rather uniformly phosphorylated, the number of phosphorylated residues in αS2-CN varies from 10 to 13 (Farrell et al., 2004). The protein contains no known covalently bound carbohydrate residues, but does have 2 cysteine residues. In contrast to κ-CN, which occurs in milk as a large disulfide linked aggregate (Farrell et al., 2003), αS2-CN appears to occur primarily as a monomer (Snoeren et al., 1980) with an internal disulfide; however, it can form dimers with the chains either parallel or antiparallel to each other (Rasmussen et al., 1992, 1994). In addition, minor amounts of αS2-CN-κ-CN-linked polymers have been isolated (Rasmussen and Petersen, 1991). A convenient large-scale method has been developed for purification of the protein, which may facilitate future research (Vreeman and von Reil, 1990).

Brignon et al. (1977) pointed out that αS2-CN has 2 very large segments of about 80 residues that display sequence homology with each other and that may have arisen by gene duplication. Segments of the αS2-CN sequences were reanalyzed using the T-coffee multiple sequence alignment tool (Notredame et al., 2000); the best homologous alignment was for residues 42–122 and124–207 with a significant score of 50%. Curiously, the last half of the second segment (residues 170–207), although homologous (T-coffee = 50) with the last half of the first segment 86–123, is more hydrophobic than the latter and has a higher positive charge (Table 1). Residues 170–207 represent the largest section of positively charged residues in any of the caseins.

Table 1. Linear profile of the αS2-CN molecule derived from its primary structure
Residues consideredNet charge1Charge frequency2Average hydrophobicity2
1–41−7.50.39860
42–80−14.50.51780
81–125+2.00.151,460
126–170−5.00.49861
171–207+6.50.241,674

1For the form of the molecule with 11 phosphorylated serine residues; serine phosphate = −2, histidine = +0.5.

2Calculated as described by Bigelow (1967).

The N-terminal section of the molecule (residues 1–35) not only has a strong homology with all of the other casein phosphopeptide regions but also displays homology with the first half of each of the 2 large segments of αS2-CN (residues 45–80 and 125–149, respectively) with a 3-way T-coffee score of 50%; all are also homologous to other casein phosphopeptides. This being the case, the overall molecule (Table 1) has 5 distinct areas composed of only 2 repeating sequences (a and b): area-a1, residues 1–41: a typical casein phosphopeptide region with high charge and low hydrophobicity; area-a2, residues 42–80: another typical phosphopeptide region; area-b1, residues 81–125: somewhat basic but with a high hydrophobicity; area-a3, residues 126–170: a phosphopeptide homolog with a high negative charge but low phosphate; and area-b2, residues 171–207: high hydrophobicity but with an extremely high positive charge (in summary: a-a-b-a-b).

It would appear that the sequences discussed above might give rise to the physical-chemical properties of the αS2-CN molecule. The number of anionic clusters is related to the calcium binding properties of αS2-CN. For example, αS2-CN is more sensitive to Ca2+ than αS1-CN (Toma and Nakai, 1973), with almost complete precipitation occurring in 2mM Ca2+ for αS2-CN at pH 7, whereas precipitation of αS1-CN requires 6mM Ca2+ (Aoki et al., 1985). Snoeren et al. (1980) concluded, from a combination of light scattering and viscosity studies, that the self-association of αS2-CN is isodesmic and produces spherical polymers of about 4nm at 0.6 M NaCl. The alternating charged areas appear to make this casein the least susceptible to aggregation as initially purified from milk. Snoeren and coworkers (1980) presented a model in which this positively charged tail participates in the self-association of the protein with one of the negative charged sections; they argued for a surface position for the positively charged cluster. As noted above, caprine caseins (in contrast to bovine caseins) vary considerably in the types of casein present: some are poor in αS1-CN and some are richer in αS2-CN (Farrell, 1999). Thus, it is speculated that caprine caseins rich in αS2-CN offer enhanced sites for interactions with negatively charged molecules such as κ-carrageenan (Farrell and Mora-Gutierrez, 2006); again this assumes that this positive cluster is on the surface of associated-whole casein. To test these theories in silico, a 3D model was developed and a preliminary report given (Farrell and Mora-Gutierrez, 2006).

A Homologous Molecular Model for αS2-Casein 

The concept that αS2-CN might be involved in the binding of a highly negatively charged species such as κ-carrageenan is supported by an examination of its linear amino acid sequence. As noted above, the molecule contains 3 segments that contain highly charged negative residues due primarily to phosphoseryl and glutamyl residues; these segments center on residues 10, 60, and 130. In contrast, the C-terminal 47 residues carry a net charge of +9.5. Because of the net hydrophobicity of the latter segment, it is uncertain whether this segment would be surface-accessible for interaction with the highly negatively charged κ-carrageenan. Hoagland et al. (2001) assessed possible structural arrangements that are in accord with currently available spectral data and one could argue from this analysis that the positively charged section might be surface-accessible. This latter argument is supported by the physical-chemical data of Snoeren et al. (1980). Although 3D molecular models have been developed for the other 3 major caseins (Kumosinski et al., 1994), no models have been published for αS2-CN. One reason for this is the segmental domain nature of the molecule as detailed above and as first noted by Swaisgood (1992). In contrast, the other caseins are more linear amphiphiles with only 2 domains, one hydrophilic and one hydrophobic.

The CLIC molecules form a novel class of recently discovered proteins (Ashley, 2003; Paul and Beitel, 2003). These proteins are often found free in the cytosol of cells, but have the ability to alter their conformation and insert themselves into membranes within the cell (hence the name intracellular); in model systems, they allow chloride movement through membranes once inserted. The CLIC proteins interact with a variety of intracellular membranes such as mitochondrial, nuclear, Golgi, and sarcoplasmic (Fernandez-Salas et al., 1999; Shanks et al., 2002; Board et al., 2004; Suh et al., 2004). It is not known if the CLIC proteins bind directly to the membranes or to receptors on the membranes, nor is it established which portions of the molecules are inserted into the membranes (Berryman et al., 2004). Within this family there are 5 members termed CLIC proteins 1 to 5 that share strong structural homologies with each other. Known defects in these proteins can cause several molecular disease states (Cromer et al., 2002). Recently it has been shown that the loss of EXC-4, a CLIC protein of Caenorhabditis elegans, prevents opening of excretory vesicles (Berry et al., 2003). The interaction of the CLIC proteins by insertion into model bilayers appeared reminiscent of the interactions of β-CN with model membranes as detailed by Casanova and Dickinson (1998). Comparisons of the amino acid sequences of human CLIC proteins with bovine β-CN using the T-coffee program of Notredame et al. (2000) showed only marginal homology. Proteins CLIC-3 and CLIC-4, however, showed good homology with αS2-CN as seen in Table 2. Although the degree of homology between the CLIC proteins and αS2-CN is only about 58%, there was sufficient similarity to proceed with the production of a working 3D model by threading the αS2-CN structure onto a model of CLIC-4. It must be noted that the CLIC-4 model was itself a working model produced from the crystal structure of a soluble monomeric fragment of human CLIC-1 found in the Swiss Model Repository of the ExPASy Web site. A crystal structure for CLIC-4 is now available (Li et al., 2006), and the ExPASy tool Magic-Fit is able to align the model used to generate the αS2-CN model with the new crystal structure of CLIC-4 with a root mean square (RMS) difference of only 0.86Å. Thus, little change in the predicted αS2-CN model is expected to occur in future studies. The homologous model produced for αS2-CN is shown as a ribboned backbone trace in Figure 2 (left).

Table 2. T-Coffee1 scores for comparison of bovine αS2- and β-CN with human chloride intracellular channel (CLIC) proteins
Proteinβ-caseinαS2-caseinCLIC proteins
CLIC-13940
CLIC-23045
CLIC-33557982
CLIC-43458
CLIC-53735
αS2-Casein330

1Notredame et al. (2000).

2All human CLIC proteins average a score of 98 in pairwise comparisons.

  • View full-size image.
  • Figure 2. 

    Left: Trace of the backbone structure of αS2-CN derived by homologous modeling. The amino acid sequence of αS2-CN was threaded onto the model of CLIC-4. Right: Manipulation of Ala 116 to produce a hydrophobic core accessible for protein-protein interactions. Coiled ribbons represent α-helix, broad ribbons represent β-sheet, and arrowheads indicate the direction of travel of the chain from N- to C-terminal.

In previous models for κ-CN, portions of a hydrophobic section of the molecule (residues 17 to 65) were used for docking sites for other caseins (Kumosinski et al., 1994). This section of κ-CN is also somewhat homologous with the central hydrophobic portion of αS2-CN (residues 77 to 119) and to segments of the CLIC proteins, and could represent a protein-protein or protein-lipid interaction site. The homologous model showed this segment in αS2-CN to be partially exposed for such interactions. The manipulation of only one residue (Ala 116) caused this hydrophobic section to further protrude into space to provide a good site for potential interactions with other caseins (Figure 2, right). This manipulation may also shed some light on the CLIC proteins, as the literature suggests the necessity of C- and N-terminal portions in membrane insertion (Paul and Beitel, 2003). Here, a central hydrophobic section is implicated for potential hydrophobic interactions. To understand the relationship of the model to the domains indicated above (Table 1), Figure 3A shows the model as a Cα trace in stereo. The 2 phosphopeptide regions 8–21 and 54–63 (areas a1 and a2) represent the negatively charged left wing with the former being toward the front and the latter to the back; area b1, the somewhat basic area with a high hydrophobicity (81–125), represents the lower body of the molecule; and area a3 (126–170) with another phosphopeptide-like region and a high negative charge represents the top of the central body. Finally, area b2 (171–207) with high hydrophobicity and high positive charge is represented on right wing.

  • View full-size image.
  • Figure 3. 

    A) Stereo view of the Cα trace of the homologous model for αS2-CN as given in Figure 2 (right panel); B) mono view of the energy-minimized model with backbone visible and ribboned for defined secondary structures (coiled ribbons represent α-helix, and broad ribbons represent β-sheet), and the free SH groups at residues 36 and 40 are indicated.

Because the FASTA format used by the ExPASy program does not support the amino acid phosphoserine (Pser), Glu was used to replace these residues to mimic the potential charge distribution on the molecule at neutral pH. Eleven Ser that are potentially phosphorylated (Figure 1) were replaced with Glu. The molecule was then energy minimized using the GROMOS program and, after adjusting side chains, the energy minimization was continued until the net total energy decreased by less than 10% of the total of the previous cycle. The energy distribution patterns of the molecule before and after minimization are given in Table 3. Using the ExPASy tool Magic-Fit, the RMS difference between the initial CLIC-based structure and the αS2-CN energy-minimized model was 0.16Å, indicating little difference between the 2 models. Figure 3B shows a ribboned version of the energy-minimized model, which is quite similar to Figure 2 (right) and shows Cys 36, Phe 39, and Cys 40 side chains.

Table 3. Energy profiles of the αS2-CN molecular models following energy minimization using the GROMOS program
EnergyEnergy, KJ/mol
Initial molecule before minimizationAfter minimization
11Glu-SH111Glu-S-S2
Bond35,681165180
Angle3,7381,1701,207
Torsion1,0241,0641,106
Improper658349353
Nonbonded21,321−6,261−6,272
Electrostatic−7,060−7,740−7,717
Total55,359−11,250−11,140

1Following conversion of Ser to Glu.

2Following internal disulfide bond formation.

When the αS2-CN working model was fitted with a surface and a pseudo-electrostatic charge calculated, a trefoil-like representation was produced as shown in color (Figure 4, right). Note at 6 o’clock the hydrophobic central core extends downward and on the upper surface, the separation of charges produces both a highly negatively charged domain (red) as well as a unique positive surface (blue) for the potential attachment of κ-carrageenan. With the exception of small portions of κ-CN and αS1-CN (Malin et al., 2005), this represents the most definitive representation of positive charge on the caseins. The model on the left represents a 90° clockwise rotation to emphasize the positive surface charge. Thus, this working model supports the physico-chemical data suggestive of a role for αS2-CN in the interaction of casein and κ-carrageenan (Farrell and Mora-Gutierrez, 2006).

  • View full-size image.
  • Figure 4. 

    Pseudo charge surface representation of the αS2-CN model. On the right the molecule is in the same orientation as in Figures 2 and 3. The negative charges (red) on the top left and center arise from the anionic sections of the molecule. The positive charges (blue) on the right arise from the C-terminal section of the molecule. The region extending from the center to 6 o’clock represents the central hydrophobic core; it is lightly colored and this section is free to react with other casein molecules as has been speculated for the homologous section of κ-CN (Kumosinski et al., 1994). The molecule on the left is rotated 90° clockwise displaying the C-terminal positively charged areas.

Overall Shape and Size of the αS2-CN Model 

The overall dimensions of the molecule (Figure 4) are a vertical distance of 61.4Å and a horizontal distance of 58.0Å. The Stokes radius calculated from the longest dimension (sweep out volume) is 30.7Å. This yields a hydrated volume of 121,200Å3. The latter value is smaller than the experimental value of 180,000 A3 (Farrell et al., 2006b) and considerably smaller than the value of 240,000 from preliminary de novo modeling considerations (Hoagland et al., 2001; Hoagland and Farrell, 2002). Farrell et al. (2006b) argued, using the calculations and nomenclature of Uversky, that αS2-CN, like αS1-CN, would be a natively unfolded coil. The current model at 120,000Å3 makes the protein more compact but still natively unfolded, and it would be classed as a premolten globule. Note that the αS2-CN model may be more compact because it was threaded onto the CLIC-4 structure. Indeed, the αS2-CN model and the CLIC-4 model have an RMS difference of only 0.16Å, indicating that the CLIC-4 protein is also a premolten globule based upon its volume. In contrast, a compact globular protein with a sequence length of 205 residues would have a volume of 37,000Å3 (Farrell et al., 2006b).

An interesting conjecture may be made at this point concerning the highly positive C-terminal section pictured in Figure 4. This segment along with the N-terminal region of αS1-CN is one of the few areas in any of the caseins that could present a positive surface charge (Malin et al., 2005). If, indeed, the C-terminal segment is surface oriented at physiological ionic strengths, then it may readily react with either the phosphopeptide regions of other caseins or with inorganic phosphate. Experimentally, it has been shown that the presence of calcium enhances phosphate binding to casein (Visser et al., 1979) and that modification of amino groups decreases calcium phosphate binding to αS1-CN (Zhang and Aoki, 1995). The presence, surface accessibility, and stability of this highly positive region could anchor the developing colloidal calcium phosphate during micelle accumulation in the Golgi apparatus (Farrell et al., 2006a).

Correlation of Proteolytic Studies of αS2-CN with the Model 

Trypsin cleavage of αS2-CN has been studied by Tauzin et al. (2002), 2003). In the latter paper they studied the time of release of tryptic peptides as a marker of surface accessibility of sections of the αS2-CN molecule. They concluded that the C-terminal sites in the positively charged region (153–207) were released with t1/2 values (where t1/2 is the time required to release one half of the susceptible peptide bonds) of 22 to 30min. Next, the N-terminal sites in the 1–90 region were more slowly released (32–35min), whereas those sites in the center of the molecule (92–150) were the slowest to be hydrolyzed. In contrast, tryptic cleavage of purified κ-CN releases peptides from its hydrophilic C-terminal region with t1/2 values of 20 to 40s (Farrell et al. 1999). The ExPASy Accessibility tool measures the surface accessibility of all residues of a protein model on a color-coded scale (red, orange, yellow, green, and blue), with the estimates ranging from high (red) to low (blue). None of the potential trypsin sites of the αS2-CN model were colored red, indicating that the potential trypsin cleavage sites are not highly surface accessible. A scale of 1 to 5 was assigned for the colored tryptic hydrolysis sites. Trypsin sites with an accessibility value of 1 would be expected to be rapidly released and would have t1/2 values of several minutes, so the model is in agreement with the data indicating few if any rapidly released peptides. The C-terminal sites yielded an average numerical value of 3.5, indicating moderate to good accessibility in line with the 22- to 30-min t1/2 values for these peptides. The central sites (92–150) had an average accessibility of 4.5, indicating poor accessibility, again in line with the experimental data. Finally, the N-terminal sites had an average accessibility of 4.0 in accord with the intermediate hydrolysis times found by Tauzin et al. (2003). Overall, the αS2-CN model is in relatively good agreement with the average experimental data; however, the individual regions do contain 2 to 3 outlying points.

Trypsin cleavage of whole caseins over a protracted time yields readily isolated peptide fractions, including the so-called CPP fraction. The CPP fractions can be added to calcium-fortified foods and beverages to improve bone health in individuals with low absorption values (Heaney et al., 1994). It has been presumed that all of the CPP fractions are similar in terms of calcium absorption regardless of the casein fraction from which they are isolated. However, a recent study showed that CPP isolated from goat casein rich in αS2-CN is superior in the delivery of added calcium to developing rats (Mora-Gutierrez et al., 2007b). It is tempting to argue that the helical regions found in the model for 2 of 3 CPP fractions are responsible for this effect. This is, however, not likely for the phosphopeptide regions 8–13 and 54–63, as no experimental evidence supports α-helix in phosphoserine-rich segments (Hoagland et al., 2001; Huq et al., 2003). A more thorough examination of the CPP and their structures is needed before any firm conclusions can be reached regarding this potentially important result.

Plasmin is the major indigenous basic milk protease. This enzyme is responsible for the proteolytic fragments of β-CN known to occur in milk (Eigel et al., 1984). The action of plasmin on purified αS2-CN has been studied by LeBars and Gripon (1989) and by Visser et al. (1989). Both groups showed that αS2-CN is attacked by the enzyme, and similar peptides were released in both studies. Visser et al. (1989) demonstrated by SDS electrophoresis that a substantial amount of the αS2-CN had been broken down to lower molecular weight segment(s) after 1h of exposure to plasmin at pH 8.4. However, only 8 of about 30 potential cleavage sites were found and the isolated fragments were primarily from the positively charged C-terminal half of the protein. These fragments were similar to the ones detailed above for trypsin by Tauzin and coworkers (2003). Here, cleavages were also found yielding f(1–21) and f(1–24) from the N-terminal section. The latter cleavages yield classical CPP fractions. As is the case with the trypsin cleavage sites, the plasmin cleavage sites reside primarily in the more surface-accessible portions of the αS2-CN model.

Chymosin cleavage of purified αS2-CN was studied by McSweeney et al. (1994). There are 3 primary predicted sites for the cleavage of the protein by this enzyme (88–89, 163–164, and 174–175). All predicted cleavages were found after a period of 20h. Using the accessibility tool described above, the average value for these residues was 3.4, predicting a moderate time of release for the peptides of 22 to 30min. It should be noted that the first susceptible site (88–89) is in the helix-turn-helix motif reminiscent of a membrane insertion protein and therefore may not be available in the self-associated polymers of αS2-CN or in the casein micelles themselves. Molecular dynamics studies of the interaction of chymosin with model substrates show clearly that β-sheet is required for cleavage (Plowman and Creamer, 1995). Here, the model predicts primarily α-helix for 88–89 and 163–164. At first glance, this appears to be contradictory to the αS2-CN model, but the 20-h time of hydrolysis for αS2-CN is quite long (McSweeney et al., 1994) relative to the time of hydrolysis of κ-CN and αS1-CN. It is possible that some slow αS2-CN conformation changes could occur and lead to altered association patterns and hydrolysis. The first of the established sites for chymosin hydrolysis occurs in a region (77–118) homologous to κ-CN (42–84), and it is this section of κ-CN that has been shown to undergo time and temperature conversions to β-sheet with accompanying formation of amyloid bodies (Farrell et al., 2003). Amyloid bodies containing residues 81–125 of αS2-CN have been isolated from bovine mammary gland (Niewold et al., 1999), and Thorn et al. (2008) have shown that a time-dependent conversion of α-helix to β-sheet occurs prior to the formation of amyloid bodies in vitro. Thus, it is possible that the native molecule does contain α-helix, but with time it converts to β-sheet for cleavage by chymosin. Alternatively the αS2-CN model may be incorrect and this area could represent β-sheet. The major segment that is in question occurs in the hydrophobic segment of the molecule, which may represent a major protein-protein interaction site (residues 86–96 with a turn at 97–100 and helix from 101 to 114). For κ-casein, this center has been predicted to be a major interaction site as β-sheet, whereas hydrophobic helices perform the same function in membrane proteins. As noted above, residues 97–100 are turns and, regardless of the surrounding motifs, this would result in a double-sided section with a strong hydrophobic surface structure for protein-protein interactions. Interestingly, the cleavage site 174–175 in the model is in the positive C-terminal region with a β-sheet conformation and would be predicted to be cleaved normally by chymosin in these experiments.

Diaz et al. (1996) and Gagnaire and Leonil (1998) studied the release of tryptic peptides from micellar casein. The latter researchers found that for αS2-CN, only10 of 30 potential sites were cleaved by the enzyme. Short peptides from Lys residues 21 and 24 were cleaved but not readily released from the micelles; in contrast, a limited number of peptides from the positive C-terminal were released. Moreover, Diaz et al. (1996) found that peptides from β-CN and αS1-CN were released quickly, but those of αS2-CN were released more slowly and in lesser amounts. The slow release of these peptides suggested that, of the calcium-sensitive caseins, αS2-CN is the least accessible in the micelles. Most of the short peptides identified were from the C-terminal portion of the molecule. This region, although relatively hydrophobic, carries a high net positive charge. The αS2-CN model as noted above does not show these cleavage sites to be exceptionally accessible; apparently they are less so in the micelles. However, the electrostatic calculation of Figure 4 shows that this region is dominated by the charges from these Lys and Arg residues. This too raises the question of whether or not these residues have a specific function in the micelle; that is, phosphate binding as noted above. Binding to calcium phosphate by the Lys and Arg residues in the positively charged C-terminal of αS2-CN would reduce their accessibility to trypsin in the micelle.

The time-dependent cleavage of caseins in Cheddar cheese was studied by Basch et al. (1989) who found that the cleavage of αS2-CN was slowest of all caseins with a half-life of 20 wk. For comparison, the chymosin cleavage of αS1-CN is complete in less than 1 d, and residual plasmin/trypsin cleavages of β-CN occur in 3 wk. Thus, the chymosin and plasmin/trypsin cleavage sites of αS2-CN are not readily accessible within the cheese matrix.

Secondary Structure Predictions and the αS2-CN Model 

Several groups have published secondary structure predictions for αS2-CN and these have been summarized by Tauzin et al. (2003) who also presented the results of a consensus algorithm for the secondary structure of the protein. In past de novo modeling studies, the method of Garnier and coworkers (1978) was used to help generate casein working models such as those presented by Kumosinski et al. (1994). Therefore, for comparison with the model, its predictions for αS2-CN will be presented along with the predictions of Hoagland et al. (2001) who attempted to correlate the predictions with spectroscopic evidence as well as the predictions of Tauzin and coworkers (2003). Because these predictions as well as the de novo models of Kumosinski and coworkers (1994) and the homologous model presented herein are meant to serve as guides to further research, the secondary structure results for the model will be compared with the predictions only in terms of overall percentages, and only gross deviations from the predictions will be discussed in detail. In addition, the polyproline II (PPII) predictions of Adzhubei and Sternberg (1993) are included because the importance of these structures in caseins has only recently been recognized (Farrell et al., 2001; Barron et al., 2002). The results from analysis of the phi/psi angles of the model are compared with both experimental data and the above predictions in Table 4.

Table 4. Summary of the overall percentage of secondary structure of the αS2-CN model and comparison with secondary structure prediction algorithms and with experimental data
StructureModelTauzin et al. (2003)Hoagland et al. (2001)Garnier et al. (1978)1Polyproline II2Experimental3
α-Helix46453354 24 to 32
β-Sheet962615 27 to 37
All turns12 19 24 to 31
Polyproline II7 15
NC α- or β-419
Unspecified7491313 9 to 22

1Values calculated for αS2-CN using the method of Garnier et al. (1978).

2Adzhubei and Sternberg (1993).

3Hoagland et al. (2001).

4Noncontinuous α-helix or β-sheet as described in the text.

For the αS2-CN model the Ramachandran plot is extremely simple with about 60% of the residues in the α-helical region, 2 residues in the lower left “forbidden zone,” a few in the turns region, and the remainder in the upper left quadrant or β-sheet/extended region. However, the actual continuous β-sheet found in the model is quite low compared with most of the predicted values and the experimental data given in Table 4. One reason for this is that PPII would fall in this region and be included as sheet in the Fourier transform infrared data of Hoagland et al. (2001); using the predictions of Adzhubei and Sternberg (1993), 15% PPII was predicted for αS2-CN. However, analysis of the αS2-CN model showed that only 7%, residues 123–128, 174–177, and 200–202, are in the PPII configuration (continuous left-handed helix). In addition, there are considerable segments of the molecule that have 3 to 4 continuous sheet phi/psi angles that are interrupted by 1 or 2 α-helical angles resulting in loops or turn-like structures. In Table 2 these are termed noncontinuous α-helix and β-sheet and account for 19% of the αS2-CN model.

With regard to α-helix, the overall predictions in the 50% area are closer to the αS2-CN model and sequence predictions are quite good for residues 81–207 of the model. Although the total percentage helix predicted is similar to the αS2-CN model for residues 1–89, there is a lack of any agreement between the actual residues predicted to be in α-helical structures and those found in the model. The percentage α-helix may be over-reported in the model because the template used (CLIC-4) has a high degree of α-helix (42%), which in turn may guide the threading experiment and influence the αS2-CN model. Indeed, the ExPASy tool Magic-Fit is able to align the parent and homolog molecules with an RMS difference of only 0.16Å, but the experimental data for α-helix do not support the high degree of helix found in the model. Because of the consensus in the C-terminal regions between the predictions and the model, the N-terminal areas of α-helix in the model must be viewed with suspicion. This is particularly true for the helices found in the model for the phosphopeptide regions 8–13 and 54–63, as no experimental evidence supports α-helix in phosphoserine-rich segments. Hoagland et al. (2001) calculated that residues 1–7 could exist in a helical conformation because of paired ions in the sequence. However, although Huq et al. (2003) have shown the potential in residues 1–41 of αS2-CN, actual helix was observed only in the presence of added calcium. In fact, riboflavin binding protein contains a phosphopeptide region homologous to those of the major caseins; in the crystal structure of riboflavin binding protein (Monaco, 1997) the phosphopeptide region is missing because of extremely high temperature factors that produce too much motion and uncertainty. Thus, as noted above, predicted helical segments for the phosphopeptides in all of the caseins must be viewed with suspicion.

Finally, turn regions account for 12% of the αS2-CN model; there are 6 regions where the polypeptide chain reverses direction near the surface of the molecule and twists toward the more interior regions. These regions are 46–49, 65–68, 97–100, 148–151, 171–174, and 191–194. Interestingly, the predictions of Garnier et al. (1978), although close in percentage, are in accord only with the last 2 of the 6. The experimental predictions of turns found in Table 4 are greater; however, the databases used had no sequences present similar to the noncontinuous regions noted above.

Sulfhydryl-Disulfide Interactions 

αS2-Casein has 2 Cys residues: 36 and 40. Experimental evidence (Rasmussen et al., 1992, 1994) has disclosed that in whole casein, >85% of the protein exists with an intramolecular disulfide between these 2 residues. The remainder of the αS2-CN appears limited to dimers connected in a parallel or antiparallel fashion (Rasmussen et al., 1992, 1994). Unlike κ-CN where multiple polymers exist between its 2 Cys residues (Farrell et al., 2003), the number of αS2-CN polymers is small in both number and size. In the αS2-CN model the 2 Cys residues can be found in the upper left corner of the molecule (Figure 3B) in a surface-oriented position where they are readily surface accessible and where the 2 side chains can be rotated outward to facilitate docking of a second monomer. Docking in a parallel manner (36–36 and 40–40) allows for both hydrophobic tails to be orientated in the same direction, possibly sharing a common hydrophobic environment. Docking in an antiparallel fashion (36–40 and 40–36) causes the hydrophobic tails to be oriented away from each other. These latter interactions should occur between 2 molecules not sharing a common hydrophobic center, and therefore arise from surface interactions between 2 protein aggregates (submicelles). Such interactions must be limited, because, as noted above, the majority of the protein (85%) is monomeric.

With only a minor degree of side-chain manipulation, a disulfide bond was put in place between Cys 36 and Cys 40. This short distance forms a very tight loop, energy minimization readily removed any constraints, and the disulfide model had a total energy equivalent to the original –SH model (Table 3). Using the ExPASy Magic-Fit tool the RMS difference between the 2 models was 0.03Å, indicating little divergence between the 2 models. The tight loop introduced by this disulfide is shown in Figure 5. Therefore, the model is in accord with the experimental data regarding the sulfhydryl-disulfide interactions of αS2-CN.

  • View full-size image.
  • Figure 5. 

    Tight loop introduced by the internal disulfide bond of αS2-CN between Cys 36 and Cys 40. The disulfide bond lies between a stretch of helix and one of sheet. The Phe 39 side chain is shown as a point of reference.

Predicted Structures of the Biologically Active Peptides of αS2-CN 

Calmodulin is a rather ubiquitous, low molecular weight protein with multiple cellular functions ranging from regulating cell division to activation of membrane enzyme cascades; it was purified from bovine mammary gland by Thompson et al. (1987). It is thought that calmodulin acts by binding to positively charged areas on its target proteins; for example, calmodulin stimulates the activity of membrane-bound cyclic nucleotide phosphodiesterase. The work of Kizawa et al. (1996) clearly demonstrates that the peptide fragments of αS2-CN, which are homologous to calmodulin binding enzyme domains, bind to calmodulin and prevent its activation of the phosphodiesterase. These fractions represent f(164–179) and f(183–206/207); f(183–207) is the most reactive with a Ki of 2.6 μM for inhibition of calmodulin activation of the enzyme. This is in the range required to exert required physiological activity. In the αS2-CN model this latter segment (Figure 6A) is primarily composed of α-helix (residues 183–190, 194–199, and 204–207). The 3 sections of helix are separated by 2 nontraditional turns or tight loops. Note the high preponderance of hydrophobic residues in this section. The overall structure can be characterized as a flat open hairpin (Figure 6A). This is precisely the structure predicted to be necessary for calmodulin binding by Kizawa et al. (1996).

  • View full-size image.
  • Figure 6. 

    Bioactive peptides from the C-terminal of αS2-CN. A) Molecular model for the peptide fragment of αS2-CN (Kizawa et al., 1996) homologous to calmodulin-binding protein domains f (183–207). In the model the latter segment is primarily composed of α-helix (residues 183–190, 194–199, and 204–207). The 3 sections of helix are separated by 2 nontraditional turns or tight loops. The overall structure can be characterized as an open hairpin; B) molecular model for αS2-CN f (165–203) called casocidin I by Zucht et al. (1995). In this case the latter 2 sections of helix are again separated by a nontraditional turn or tight loop, but the first 2 are separated by a section of the molecule (172–180) that is linear in structure and is primarily in a polyproline II configuration (173–178). This makes the peptide quite planar with the overall appearance of a Greek letter Ω; C) molecular model for the major angiotensin I converting enzyme (ACE) inhibitor in tryptic digests of αS2-CN f (174–181) (Tauzin et al., 2002). Residues 173–180 are in the polyproline II conformation.

The first antibacterial peptide isolated from αS2-CN was called Casocidin I by Zucht et al. (1995). This peptide inhibited the growth of both Escherichia coli and Staphylococcus carnosus in a dose-dependent manner at concentrations above 300ng. The active peptide was identified as f(165–203). In the αS2-CN model this segment is also primarily composed of α-helix (residues 165–171, 181–190, and 194–199). In Figure 6B it can be seen that the latter 2 sections of helix are separated by a nontraditional turn or tight loop, but the first 2 are separated by a section of the molecule (172–180) that is a linear structure and is primarily in a PPII configuration (173–178). This makes the peptide quite planar with the overall appearance of a Greek capital letter Ω as seen in Figure 6B. Note again the high preponderance of hydrophobic residues. Recio and Visser (1999) found 2 antibacterial peptides in peptic digests of αS2-CN representing sequences similar to those found by Zucht and coworkers (1995); namely, f(164–179) and f(183–207). The latter fraction had consistently higher activity. Finally, McCann et al. (2005) identified 5 antibacterial peptides from chymosin digests of αS2-CN; all were from the C-terminal end with f(164–207) and f(172–207) being most active against Listeria innocua. The activities of the latter peptides were equivalent to the bacteriocin nisin. Note that the structures would be nearly identical to that of Figure 6B. Lopez-Esposito et al. (2006) studied ovine αS2-CN peptides with antibacterial action against E. coli. The highest potency was found for f(165–181), which again would be predicted to be structurally similar to Figure 6B.

Tauzin et al. (2002) discovered inhibitors of the angiotensin-I-converting enzyme in tryptic digests of αS2-CN. The fragments were identified as f(174–181) and f(174–179). These segments are contained within the N-terminal of the Casocidin I structure noted above. Here, the motif is predicted to be only the extended planar structure discussed above. Residues 173–180 are in the PPII conformation (Figure 6C). It is interesting that these segments contain the PPII conformation. It has been proffered that this conformation may be highly involved in protein-protein interactions regulating several metabolic pathways and protein-mediated intracellular cytoskeleton changes (Kay et al., 2000). Once more, the molecule is flat with a high level of hydrophobic residues.

Intestinal absorption can be studied in vitro by the use of Caco-2 cell cultures. These cells, derived from a colon carcinoma line, form monolayers in culture and mimic intestinal brush borders (Tanabe et al., 2006). One potential route of food allergy is the permeation of the intestinal cell wall by intact proteins. Nondenatured ovalbumin can permeate Caco-2 cells in culture. Tanabe and coworkers (2006) discovered that a small peptide from αS2-CN f(102–111) can prevent ovalbumin permeation of Caco-2 cells in vitro. Although not predicted to be helical by the programs cited above (Table 4), this segment of the αS2-CN model is a tightly wound helix (Figure 7A, bottom). It contains the correctly aligned H-bonds for an α-helix as calculated by the ExPASy tool, and is highly hydrophobic. In Figures 3 and 4 it represents the right side of the hydrophobic foot at 6 o’clock, which has been hypothesized to be related to protein-protein interactions between αS2-CN and other caseins. Its hydrophobic residues (perhaps stabilized by the internal H-bonds) may bind to pores in the cell membranes and prevent permeation by ovalbumin. Alternatively, once excised from the molecule it could unfold to β-sheet as seen in Figure 7A (top); this segment, as discussed above, is a portion of the αS2-CN found in amyloid bodies in mammary gland f(81–112) by Niewold et al. (1999), and amyloid bodies are generally β-sheet-based structures.

  • View full-size image.
  • Figure 7. 

    Peptides from αS2-CN potentially active in the intestine. A) Molecular model for the potential antiallergenic peptide from αS2-CN f (102–111) as discovered by Tanabe et al. (2006). Note the tight helical conformation and the presence of Trp (bottom) and the segment as a potential β-sheet (top); B) molecular model for αS2-CN f (34–59) as excised from the total model (right) and compared with the crystal structure (left) for defensin 6 (Szyk et al., 2006).

Overall intestinal health and well-being is the result of interactions between the cells of the intestinal mucosa and the bacterial flora that surrounds them. In the normal course of events proteins called defensins are the first line of surveillance; they are secreted by the Paneth cells of the intestinal mucosa and bind to and monitor helpful and potentially harmful bacteria. These defensins have low molecular weight and have known crystal structures (Szyk et al., 2006). Defensin 6 has an overall homology to αS2-CN of 68% using the T-coffee program; this is for the pre-pro-defensin. When the latter is cleaved by proteases to the active form of defensin 6, the homology score remains the same, but only the first 21 residues remain in alignment with the αS2-CN residues f(34–59), the remainder form a gapped alignment with αS2-CN f(164–174). Nevertheless, the molecular model for the first section of αS2-CN f(34–59) is strikingly similar to the total crystal structure of defensin 6 (Figure 7B). Thus, αS2-CN may be predicted to yield peptides with defensin-like activity and aid the immune system in fighting bacteria.

Back to Article Outline

Conclusions 

A 3D molecular model for αS2-CN has been produced by threading the backbone sequence of the protein onto a homologous protein, CLIC-4. In this model Glu was used in place of Ser to mimic phosphoserine in the overall size and charge distribution on the protein. Overall, the model is in good agreement with experimental data for the protein; however, the amount of helix may be over-predicted. The model does, however, offer a unique view of the highly positive C-terminal portion of the molecule as a surface accessible area. This region may be the site for interactions with κ-carrageenan, phosphate, and other anions. In addition, many of the physiologically active and antibiotic peptides isolated from αS2-CN occur in this region; others come from the functional hydrophobic foot as well as the center of the molecule. Based on molecular similarities, αS2-CN is predicted to yield peptides with defensin-like activity, which may aid the immune system in fighting bacteria. This structure should be viewed as a working model with the ability to be changed as more precise experimental data are obtained. As such, the model completes the series of working models for the caseins (Kumosinski et al., 1994).

Back to Article Outline

References 

  1. Adzhubei AA, Sternberg MJE. Left-handed polyproline II helices commonly occur in globular proteins. J. Mol. Biol. 1993;229:472–493
  2. Ambrosoli R, DiStaiso L, Mazzocco P. Content of αS1-casein and coagulation properties of goat milk. J. Dairy Sci. 1988;71:24–28
  3. Aoki T, Toyooka K, Kako Y. Role of phosphate groups in the calcium sensitivity of αS2-casein. J. Dairy Sci. 1985;68:1624–1629
  4. Ashley RH. Challenging accepted ion channel biology: p64 and the CLIC family of putative intracellular anion channel proteins. Mol. Membr. Biol. 2003;20:1–11
  5. Barron LD, Blanch EW, Hecht A. Unfolded proteins studied by Raman optical activity. Adv. Protein Chem. 2002;62:51–90
  6. Basch JJ, Farrell HM, Walsh RA, Konstance RP, Kumosinski TF. Development of a quantitative model for enzyme-catalyzed, time dependent changes in protein composition of cheddar cheese during storage. J. Dairy Sci. 1989;72:591–603
  7. Berry KL, Bulow HE, Hall DH, Hobert O. A C. elegans CLIC-like protein required for intracellular tube formation and maintenance. Science. 2003;302:2134–2137
  8. Berryman M, Bruno J, Price J, Edwards JC. CLIC-5A functions as a chloride channel in vitro and associates with the cortical actin cytoskeleton in vivo and in vitro. J. Biol. Chem. 2004;279:34794–34801
  9. Bigelow CC. On the average hydrophobicity of proteins and the relation between it and protein structure. J. Theor. Biol. 1967;16:187–211
  10. Board PG, Coggan M, Watson S, Gage PW, Dulhunty AF. CLIC-2 modulates cardiac ryanodine receptor Ca2+ release channels. Int. J. Biochem. Cell Biol. 2004;36:1599–1612
  11. Brignon G, Ribadeau Dumas B, Mercier J-C, Pelissier J-P, Das BC. The complete amino acid sequence of bovine αS2-casein. FEBS Lett. 1977;76:274–278
  12. Casanova H, Dickinson EJ. Rheology and flocculation of oil-in-water emulsions made with mixtures of αS1-casein and β-casein. J. Colloid Interface Sci. 1998;207:82–89
  13. Cromer BA, Morton CJ, Board PG, Parker MW. From glutathione transferase to pore in a CLIC. Eur. Biophys. J. 2002;31:356–364
  14. Diaz O, Gouldsworthy AM, Leaver J. Identification of peptides released from casein micelles by limited trypsinolysis. J. Agric. Food Chem. 1996;44:2517–2522
  15. Eigel WN, Butler JE, Ernstrom CA, Farrell HM, Harwalkar VR, Jenness R, et al. Nomenclature of proteins of cow's milk: Fifth revision. J. Dairy Sci. 1984;67:1599–1631
  16. Farrell HM. Milk composition and synthesis. In:  Knobil E,  Neill JD editor. Encyclopedia of Reproduction. San Diego, CA: Academic Press Publishers; 1999;p. 256–263
  17. Farrell HM, Cooke PH, Wickham ED, Piotrowski EG, Hoagland PD. Environmental influences on bovine κ-casein: Reduction and conversion to fibrillar (amyloid) structures. J. Protein Chem. 2003;22:259–273
  18. Farrell HM, Jimenez-Flores R, Bleck GT, Brown EM, Butler JE, Creamer LK, et al. Nomenclature of the proteins of cows’ milk—Sixth revision. J. Dairy Sci. 2004;87:1641–1674
  19. Farrell HM, Malin EL, Brown EM, Qi PX. Casein micelle structure: What can be learned from structural biology?. Curr. Opin. Colloid Interface Sci. 2006;11:135–147
  20. Farrell HM, Mora-Gutierrez A. κ-Carrageenan interaction with bovine and caprine caseins as shown by sedimentation and NMR spectroscopic techniques. Implication of surface charge by a homologous three dimensional model for αS2-casein. In:  Fishman ML,  Qi PX,  Wicker L editor. Advances in Biopolymers: Molecules, Clusters, Networks, and Interactions. Washington, DC: American Chemical Society; 2006;p. 93–114
  21. Farrell HM, Qi PX, Uversky VN. New views of protein structure: Applications to the caseins: Protein structure and functionality. In:  Fishman ML,  Qi PX,  Wicker L editor. Advances in Biopolymers: Molecules, Clusters, Networks, and Interactions. Washington, DC: American Chemical Society; 2006;p. 52–70
  22. Farrell HM, Wickham ED, Dower HJ, Piotrowski EG, Hoagland PD, Cooke PH, et al. Characterization of the particles of purified κ-casein: Trypsin as a probe of surface-accessible residues. J. Protein Chem. 1999;18:637–652
  23. Farrell HM, Wickham ED, Unruh JJ, Qi PX, Hoagland PD. Secondary structural studies of bovine caseins: Temperature dependence of β-casein structure as analyzed by circular dichroism and FTIR spectroscopy and correlation with micellization. Food Hydrocolloids. 2001;15:341–354
  24. Fernandez-Salas E, Sagar M, Cheng C, Yuspa SH, Weinberg WC. p53 and tumor necrosis factor alpha regulate the expression of a mitochondrial chloride channel protein. J. Biol. Chem. 1999;274:36488–36497
  25. Gagnaire V, Leonil J. Preferential sites of tryptic cleavage on the major caseins within the micelle. Lait. 1998;78:471–489
  26. Garnier J, Osguthorpe DJ, Robinson B. Analysis of the accuracy and implications of simple methods for predicting the secondary structure of globular proteins. J. Mol. Biol. 1978;120:97–120
  27. Groenen MAM, Dijkhof REM, Verstege AJM, van der Poel JJ. The complete sequence of the gene encoding bovine αS2-casein. Gene. 1993;123:187–193
  28. Grosclaude F. Le polymorphosine genetique des principales 1actoproteines bovines. Relationsavec la quantite’, la composition et les aptitudes fromageres du lait. Le polymorphosine genetique des principales 1actoproteines bovines. Relationsavec la quantite ∼ la composition et les aptitudes fromageres du lait. INRA Prod. Anim. 1988;1:5–17
  29. Heaney RP, Saito Y, Orimo H. Effect of casein phosphopeptide on absorbability of co-ingested calcium in normal postmenopausal women. J. Bone Miner. Metab. 1994;12:77–81
  30. Hoagland PD, Farrell HM. Molecular modeling of bovine αS2-casein. J. Dairy Sci. 2002;85(Suppl. 1):341. (Abstr.)
  31. Hoagland PD, Unruh JJ, Wickham ED, Farrell HM. Secondary structure of bovine αS2-casein: Theoretical and experimental approaches. J. Dairy Sci. 2001;84:1944–1949
  32. Huq NL, Cross KJ, Reynolds EC. Nascent helix in the multiphosphorylated peptide αS2-casein (2–20). J. Pept. Sci. 2003;9:386–392
  33. Kay BK, Williamson MP, Sudol M. The importance of being proline: Interactions of proline-rich motifs in signaling proteins with their cognate domains. FASEB J. 2000;14:231–241
  34. Kizawa K, Naganuma K, Murakami U. Interactions of amphiphilic peptides derived from αS2-casein with calmodulin. J. Dairy Sci. 1996;79:1728–1733
  35. Kopp J, Schwede T. The SWISS-MODEL Repository of annotated three-dimensional protein structure homology models. Nucleic Acids Res. 2004;32:D230–D234
  36. Kumosinski TF, King G, Farrell HM. An energy minimized three dimensional working model for casein submicelles. J. Protein Chem. 1994;13:681–700
  37. Le Bars D, Gripon J-C. Specificity of plasmin towards bovine αS2-casein. J. Dairy Res. 1989;56:817–821
  38. Li YF, Li DF, Zeng ZH, Wang DC. Trimeric structure of the wild soluble chloride intracellular ion channel CLIC4 observed in crystals. Biochem. Biophys. Res. Commun. 2006;343:1272–1278
  39. Lopez-Exposito I, Gomez-Ruiz JA, Amigo L, Recio I. Identification of antibacterial peptides from ovine αS2-casein. Int. Dairy J. 2006;16:1072–1080
  40. Lopez-Exposito I, Recio I. Protective effect of milk peptides: Antibacterial and antitumor properties. Adv. Exp. Med. Biol. 2008;606:271–293
  41. Malin EL, Brown EM, Wickham ED, Farrell HM. Contributions of terminal peptides to the associative behavior of αS1-casein. J. Dairy Sci. 2005;88:2318–2328
  42. McCann KB, Shiell BJ, Michalski WP, Lee A, Wan J, Roginski H, et al. Isolation and characterization of antibacterial peptides from the f(164–207) region of bovine αS2-casein. Int. Dairy J. 2005;15:133–143
  43. McSweeney PLH, Olson NF, Fox PF, Healy A. Proteolysis of bovine αS2-casein by chymosin. Z. Lebensm. Unters. Forsch. 1994;119:429–432
  44. Monaco HL. Crystal structure of chicken riboflavin binding protein. EMBO J. 1997;16:1475–1483
  45. Mora-Gutierrez A, Farrell HM. Hydration of native and cold solubilized rennet-treated caprine caseins as determined by oxygen 17 nuclear magnetic resonance. J. Dairy Sci. 2001;84(E Suppl.):E93–E99
  46. Mora-Gutierrez A, Farrell HM, Attaie R, McWhinney VJ, Wang C. Effects of bovine and caprine Monterey Jack cheeses fortified with milk calcium on bone mineralization in rats. Int. Dairy J. 2007;17:255–267
  47. Mora-Gutierrez A, Farrell HM, Attaie R, McWhinney VJ, Wang C. Influence of bovine and caprine casein phosphopeptides differing in αS1-casein content in determining the absorption of calcium from bovine and caprine calcium-fortified milks in rats. J. Dairy Res. 2007;74:356–366
  48. Mora-Gutierrez A, Kumosinski TF, Farrell HM. Comparison of hydration behavior of bovine and caprine caseins as determined by oxygen-17 nuclear magnetic resonance: pH/pD dependence. J. Agric. Food Chem. 1996;44:796–803
  49. Mora-Gutierrez A, Kumosinski TF, Farrell HM. Oxygen-17 nuclear magnetic resonance studies of bovine and caprine casein hydration and activity in deuterated sugar solutions. J. Agric. Food Chem. 1997;45:4545–4553
  50. Niewold TA, Murphy CL, Hulskamp CAM, Tooten PCJ, Gruys E. Casein related amyloid, characterization of a new and unique amyloid protein isolated from bovine corpora amylacea. Amyloid. Int. J. Exp. Clin. Invest. 1999;6:244–249
  51. Notredame C, Higgens D, Heringa J. T-Coffee: A novel method for fast and accurate multiple sequence alignment. J. Mol. Biol. 2000;302:205–217
  52. Paul SM, Beitel GJ. Developmental biology. Tubulogenesis CLICs into place. Science. 2003;302:2077–2078
  53. Plowman JE, Creamer LK. Restrained molecular dynamics study of the interaction between bovine κ-casein peptide 98–111 and bovine chymosin and porcine pepsin. J. Dairy Res. 1995;62:451–467
  54. Rasmussen L, Hojrup P, Petersen TE. Localization of two interchain disulfide bridges in dimers of bovine αS2-casein. Parallel and antiparallel alignments of the polypeptide chains. Eur. J. Biochem. 1992;203:381–386
  55. Rasmussen L, Hojrup P, Petersen TE. Disulphide arrangement in bovine caseins: Localization of intra chain disulphide bridges in monomers of κ- and αS2-casein from bovine milk. J. Dairy Res. 1994;61:485–493
  56. Rasmussen L, Petersen TE. Purification of disulphide-linked αS2- and κ- casein from bovine milk. J. Dairy Res. 1991;58:187–193
  57. Recio I, Visser S. Identification of two distinct antibacterial domains within the sequence of bovine αS2-casein. Biochim. Biophys. Acta. 1999;1428:314–326
  58. Remeuf F, Verdalet-Guzman I, Lenoir J. Technological adjustment of goat milk containing low synthesis rate αS1-casein variants. Int. Dairy J. 1995;5:381–392
  59. Sawyer L, Barlow PN, Boland MJ, Creamer LK, Denton H, Edwards PJB, et al. Milk protein structure—What can it tell the dairy industry?. Int Dairy J. 2002;12:299–310
  60. Shanks RA, Larocca MC, Berryman M, Edwards JC, Urshdinani T, Navarre J, et al. AKAP450 at the Golgi apparatus. J. Biol. Chem. 2002;277:40973–40980
  61. Snoeren TH, van Markwijk B, van Montfort R. Some physical chemical properties of bovine αS2-casein. Biochim. Biophys. Acta. 1980;622:268–276
  62. Stewart AF, Bonsing J, Beattie CJ, Shah F, Willis FIM, Mackinlay AG. Complete nucleotide sequences of bovine αS2- and β-casein cDNAs: Comparisons with related sequences in other species. Mol. Biol. Evol. 1987;4:231–241
  63. Stocker U, van Gunsteren WF. Molecular dynamics simulation of hen egg white lysozyme: A test of the GROMOS96 force field against nuclear magnetic resonance data. Proteins. 2000;40:145–153
  64. Suh KS, Mutoh M, Nagashima K, Fernandez-Salas E, Edwards LE, Hayes DD, et al. The organellar chloride channel protein CLIC4/mtCLIC translocates to the nucleus in response to cellular stress and accelerates apoptosis. J. Biol. Chem. 2004;279:4632–4641
  65. Swaisgood HE. Chemistry of the caseins. In:  Fox PF editors. Advanced Dairy Chemistry. I: Proteins. New York, NY: Elsevier Applied Science; 1992;p. 63–110
  66. Szyk A, Wu Z, Tucker K, Yang D, Lu W, Lubkowski J. Crystal structures of human defensins HNP4 HD5 and HD6. Protein Sci. 2006;15:2749–2760
  67. Tanabe S, Isobe N, Miyauchi E, Kobayashi S, Suzuki M, Oda M. Identification of a peptide in enzymatic hydrolysate of cheese that inhibits ovalbumin penetration in Caco-2 cells. J. Agric. Food Chem. 2006;54:6904–6908
  68. Tauzin J, Miclo L, Gaillard J-L. Angiotensin-I-converting enzyme inhibitory peptides from tryptic hydrolysate of bovine αS2-casein. FEBS Lett. 2002;531:369–374
  69. Tauzin J, Miclo L, Roth S, Molle D, Gaillard J-L. Tryptic hydrolysis of bovine αS2-casein: Identification and release kinetics of peptides. Int. Dairy J. 2003;13:15–27
  70. Thompson MP, Piazza GJ, Brower DP, Bingham EW, Farrell HM. Isolation and characterization of bovine mammary calmodulin. J. Dairy Sci. 1987;70:1551–1556
  71. Thorn DC, Ecroyd H, Sunde M, Poon S, Carver JA. Amyloid fibril formation by bovine milk αS2-casein occurs under physiological conditions yet is prevented by its natural counterpart αS1-casein. Biochemistry. 2008;47:3926–3936
  72. Toma SJ, Nakai S. Calcium sensitivity and molecular weight of αS2-casein. J. Dairy Sci. 1973;56:1559–1562
  73. Visser J, Schaier RW, van Gorkom M. The role of calcium phosphate and citrate ions in the stabilization of casein micelles. J. Dairy Res. 1979;46:333–335
  74. Visser S, Slangen KJ, Alting AC, Vreeman HJ. Specificity of bovine plasmin in its action on bovine αS2-casein. Milchwissenschaft. 1989;44:335–339
  75. Vreeman HJ, von Reil JAM. The large-scale isolation of αS2-casein from bovine casein. Neth. Milk Dairy J. 1990;44:43–48
  76. Zhang ZP, Aoki T. Effect of modification of amino groups on crosslinking of casein by micellar calcium phosphate. J. Dairy Sci. 1995;78:36–43
  77. Zucht HD, Raida M, Adermann K, Magert HJ, Forssmann WG. Casocidin-I: A casein-αS2 derived peptide exhibits antibacterial activity. FEBS Lett. 1995;372:185–188

PII: S0022-0302(09)70445-X

doi:10.3168/jds.2008-1711

Journal of Dairy Science
Volume 92, Issue 4 , Pages 1338-1353, April 2009

Article Tools

* Image Usage & Resolution