Advanced Dairy Chemistry

3 Quantitation of Proteins in Milk and Milk Products 133 Shinmoto, H., Kobori, M., Tsushida, T. and Shinohara, K. tional properties as determined by Fourier Transform (1997). Competitive ELISA of bovine lactoferrin with Infrared Spectroscopy. Int. Dairy J. 8, 135–140. bispecific monoclonal antibodies. Biosci. Biotechnol. Surroca, Y., Haverkamp, J. and Heck, A.J.R. (2002). Biochem. 61, 1044–1046. Towards the understanding of molecular mechanisms in the early stages of heat-induced aggregation of Sjaunja, L.O. and Andersson, I. (1985). Laboratory exper- b-lactoglobulin AB. J. Chromatogr. A 970, 275–285. iments with a new IR milk analyzer, the Milko Scan. Swaisgood, H.E. (1992). Chemistry of the caseins, in, Acta Agric. Scand. 35, 345–352. Advanced Dairy Chemistry-1, Proteins, 2nd edn., P.F. Fox, ed., Elsevier Science Publishers, London, pp. Sjaunja, L.O. and Schaar, J. (1984). Determination of 63–110. casein in milk by infrared spectrophotometry. Swaisgood, H.E. (2005). The importance of disulfide Milchwissenschaft 39, 288–290. bridging. Biotechnol. Adv. 23, 7173. Tam, J.P. (1994). Immunization with peptide carrier com- Smith, E.B., Barbano, D.M., Lynch, J.M. and Fleming, plexes: traditional and multiple antigen peptide sys- J.R. (1993a). A quantitative linearity evaluation tems, in, Peptide Antigens. A Practical Approach, G.B. method for infrared milk analyzers. J. Assoc. Off. Wisdom, ed., Oxford University Press, pp. 83–114. Anal. Chem. Int. 76, 1300–1308. Thompson, A. K., Singh, H., Dalgleish, D. G. (2010). Use of surface plasmon resonance (SPR) to study the dis- Smith, E.B., Barbano, D.M., Lynch, J.M. and Fleming, sociation and polysaccharide binding of casein J.R. (1993b). Performance of homogenizers in infra- micelles and caseins. J. Agric. Food Chem. 58, red milk analyzers: a survey. J. Assoc. Off. Anal. Chem. 11962–11968. Int. 76, 1033–1041. Tijssen, P. (1988). Practice and theory of enzyme immu- noassays, in, Laboratory Techniques in Biochemistry Smith, E.B., Barbano, D.M., Lynch, J.M. and Fleming, and Molecular Biology, R.H. Burdon and P.H. Van J.R. (1995). Infrared analysis of milk: effect of homog- Knippenberg, eds., Elsevier, Amsterdam, pp. 1–7. enizer and optical filter selection on apparent homog- Tolkach, A., Steinle, S., and Kulozik, U. (2005). enization efficiency and repeatability. J. Assoc. Off Optimization of thermal pretreatment conditions for Anal. Chem. Int. 78, 1225–1233. the separation of native alpha-lactalbumin from whey protein concentrates by means of selective denatur- Smith, J.L., Hendrickson, W.A., Honzatko, R.B. and ation of beta-lactoglobulin. J. Food Sci. 70, Sheriff, S. (1986). Structural heterogeneity in protein E557–E566. crystals. Biochemistry 25, 5018–5027. Trieu Cuot, P. and Gripon, J.C. (1981). Electrofocusing and two dimensional electrophoresis of bovine caseins. Smith, R.D., Loo, J.A., Edmonds, C.G., Barinaga, C.J. J. Dairy Res. 48, 303–310. and Udseth, H.R. (1990). New developments in bio- Valence, F., Richoux, R., Thierry, A., Palva, A. and Lortal, chemical mass spectrometry: electrospray ionization. S. (1999). Autolysis of Lactobacillus helveticus and Anal. Biochem. 62, 882–899. Propionibacterium freudenreichii in Swiss cheeses: first evidence by using species specific lysis markers. Smolenski, G., Haines, S., Kwan, F.Y.S., Bond, J., Farr, J. Dairy Res. 65, 609–620. V., Davis, S.R., Stelwagen, K. and Wheeler, T.T. van de Voort, F.R., Sedman, J., Emo, G. and Ismael, A.A. (2007). Characterisation of host defence proteins in (1992). Assessment of Fourier transform infrared milk using a proteomic approach. J. Proteome Res. 6, analysis of milk. J. Assoc. Off. Anal. Chem. Int. 75, 207–215. 780–785. van Hekken, D.L. and Thompson, M.P. (1992). Application Spengler, B and Cotter, R. (1990). Ultraviolet laser des- of PhastSystem® to the resolution of bovine milk pro- orption/ionization mass spectrometry of proteins teins on urea polyacrylamide gel electrophoresis. above 100,000 daltons by pulsed ion extraction time of J. Dairy Sci. 75, 1204–1210. flight analysis. Anal. Chem. 62, 793–796. Vegarud, G.E., Molland, T.S., Brovold, M.J., Devoid, T.G., Alestrom, P., Steine, T., Rogne, S. and Langsrud, Spengler, B. (1997). Post source decay analysis in matrix T. (1989). Rapid separation of genetic variants of assisted laser desorption/ionization mass spec- caseins and whey proteins using urea modified gels trometry of biomolecules. J. Mass Spectrom. 32, and fast electrophoresis. Milchwissenschaft 44, 1019–1036. 689–691. Visschers, R.W. and de Jongh, H.H.J. (2005). Disulfide Stelwagen, K. and Lacy Hulbert, S.J. (1996). Effect of bond formation in food protein aggregation and gela- milking frequency on milk somatic cell count charac- tion. Biotechnol. Adv. 23, 75–80. teristics and mammary secretory cell damage in cows. Visser, S., Slangen, C.J. and Rollema, H.S. (1991). Am. J. Vet. Res. 57, 902–905. Phenotyping of bovine milk proteins by reversed phase Strange, E.D., van Hekken, D. and Thompson, M.P. (1991). Qualitative and quantitative determination of caseins with reverse phase and anion exchange HPLC. J. Food Sci. 56, 1415–1420. Strange, E.D., Malin, E.L., van Hekken, D.J.L. and Basch, J.J. (1992). Chromatographic and electrophoretic methods used for analysis of milk proteins. J. Chromatogr. 624, 81–102. Subirade, M., Loupil, F., Allain, A.F. and Paquin, P. (1998). Effect of dynamic high pressure on the second- ary structure of b-lactoglobulin and on its conforma-

134 D. Dupont et al. high performance liquid chromatography. Warme, P.K., Momany, F.A., Rumball, S.V., Tuttle, R.W. J. Chromatogr. 548, 361–370. and Scheraga, H.A. (1974). Computation of structures Visser, S., Slangen, C.J., Lagerwerf, F.M., van Dongen, of homologous proteins a-lactalbumin from lysozyme. W.D. and Haverkamp, J. (1995). Identification of a Biochemistry 13, 768–782. new genetic variant of bovine b-casein using reversed phase high performance liquid chromatography and Wehling, M.M. and Pierce, R.L. (1994). Comparison of mass spectrometric analysis. J. Chromatogr. A 711, sample handling and data treatment methods for deter- 141–150. mining moisture and fat in Cheddar cheese by near Visser, S., Slangen, K.J. and Rollema, H.S. (1986). High infrared spectroscopy. J. Agric. Food. Chem. 42, performance liquid chromatography of bovine caseins 2830–2835. with the application of various stationary phases. Milchwissenschaft 41, 559–562. Wishart, D.S., Sykes, B.D. and Richards, F.M. (1992). The Wake, R.G. and Baldwin, R.C. (1961). Analysis of casein chemical shift index: A fast and simple method for the fractions by zone electrophoresis in concentrated urea. assignment of protein secondary structure through Biochim. Biophys. Acta 47, 225–239. NMR spectroscopy. Biochemistry 31, 1647–1651. Ward, L.S. and Bastian, E.D. (1998). Isolation and identification of b-casein A1 4P and b-casein A2 4P in Wüthrich, K. (1989). Protein structure determination in commercial caseinates. J. Agric. Food Chem. 46, solution by nuclear magnetic resonance spectroscopy. 77–83. Science 243, 45–50. Walstra, P. (1999). Dairy Technology: Principles of Milk Properties and Processes. Marcel Dekker, New York. p. 727.

Chemistry of the Caseins 4 T. Huppertz 4.1 Introduction random coil or natively denatured proteins appears inaccurate as a definite degree secondary Milk protein constitutes an important part of the and tertiary structure has been identified for human diet. For the neonate, milk or infant the caseins. formula is the only type of food consumed; how- ever, whereas milk does not constitute a major The aim of this chapter is to provide the part of the diet after the neonatal stage of most current state-of-the-art with respect to the chem- other mammals, the human diet in many parts of istry of caseins. As with previous reviews on this the world continues to include high levels of dairy topic by Swaisgood (1982, 1992, 2003), casein products. The popularity of milk proteins in the composition and nomenclature, chemical compo- human diet is undoubtedly a result of the combi- sition and primary structure of the caseins, post- nation of their excellent nutritional value and translational modification, secondary structures high level of functionality. The relative ease of as well as physicochemical properties of caseins, isolation of proteins from milk has led not only to such as self-association and the interactions with the creation of a wide variety of functional and calcium, will be covered. For studies on the isola- nutritional milk protein ingredients, but also to tion of casein, the reader is referred to Swaisgood milk proteins being the best characterized of all (2003). Higher order structures of caseins and the food proteins. The primary structures of all milk structure and stability of the association colloids proteins have been determined and for all the in which caseins naturally exist, i.e. casein major whey proteins, three-dimensional struc- micelles, are outside the scope of this chapter and tures have been elucidated. Because of the fact are covered in Chaps. 5 and 6, respectively. The that attempts to crystallize caseins have thus far focus in this chapter will be on the caseins in remained unsuccessful, the full secondary and bovine milk; interspecies variability in casein tertiary structure of the caseins remains to be composition is covered in Chap. 13. elucidated. Although caseins have higher flexibi- lity than typical globular proteins, e.g. whey 4.2 Casein Composition proteins, the previous classification of caseins as and Nomenclature T. Huppertz (*) The American Dairy Science Association NIZO food research, P.O. Box 20, Committee on the Nomenclature, Classification 6710 BA, Ede, The Netherlands and Methodology of Milk Proteins originally e-mail: [email protected] defined the bovine caseins as those phosphop- roteins that precipitate from raw milk by P.L.H. McSweeney and P.F. Fox (eds.), Advanced Dairy Chemistry: Volume 1A: Proteins: Basic Aspects, 135 4th Edition, DOI 10.1007/978-1-4614-4714-6_1, © Springer Science+Business Media New York 2013

136 T. Huppertz Table 4.1 Current and former nomenclature of caseins phosphorylated, whereas glycosylation has been and major peptides derived therefrom. Reference proteins shown only for k-CN. As discussed in further are printed in italics detail later, as1-CN, as2-CN and b-CN are classified as the calcium-sensitive caseins, Current Former whereas k-CN is calcium insensitive. In addition to the aforementioned gene products, g-caseins as1- CN as1-CN A and l-caseins have been identified, which arise as1- CN A-8P as1-CN B from the hydrolysis of b-CN and as1-CN, respec- as1- CN B-8P as 0-CN tively, by the indigenous milk proteinase, plasmin. as1- CN B-9P as1-CN C Enzymatic hydrolysis of milk proteins by plasmin as1- CN C-8P as1-CN D is outside the scope of this chapter and is dealt as1- CN D-9P as1-CN E with in detail in Chap. 12. as1- CN E-8P as2- CN as6-CN A Current nomenclature of caseins and some as2- CN A-10P as4-CN A casein fractions, as well as former classifications as2- CN A-11P as3-CN A by which they were known, is shown in Table 4.1. as2- CN A-12P as2-CN A In such nomenclature, a Latin letter indicates as2- CN A-13P the generic variant of the proteins, whereas b-CN b-CN A1 differences in the degree of post-translational b-CN A1-5P b-CN A2 modification are indicated by an Arabic number, b-CN A2-5P b-CN A3 followed by the letter P to indicate that the post- b-CN A3-5P b-CN C translational variation arises from phosphoryla- b-CN D tion. For example, as1-CN B-8P refers to genetic b-CN C-4P b-CN E variant B of as1-CN containing eight phosphory- lated amino acid residues. For each of the caseins, b-CN D-4P k-CN A one of the variants outlined in Table 4.1 is consid- k-CN B ered to be the reference protein; these reference b-CN E-5P proteins are as1-CN B-8P, as2-CN A-11P, b-CN A2-5P and k-CN A-1P. k-CN k-CN A-1P 4.3 as1-Casein k-CN B-1P 4.3.1 Primary Structure of as1-Casein acidification to pH 4.6 at 20°C (Jenness et al., 1956). Subsequent reports by the committee The as1-CN family represents ~40% of total recommended that the caseins could be differen- casein in bovine milk. The reference protein for tiated according to their relative electrophoretic the as1-CN family is as1-CN B-8P, with ExPASy mobility in alkaline polyacrylamide or starch gels entry name and file number of CAS1_Bovin and containing urea, with or without b-mercaptoetha- P02662, respectively. The amino acid sequence nol (Whitney et al., 1976) or, more recently, of as1-CN B-8P, which predominates in the milk according to their primary amino acid sequences of Bos taurus and was first established by Mercier (Eigel et al., 1984; Farrell Jr et al., 2004). et al. (1971) and Grosclaude et al. (1973), is Accordingly, four gene products can be shown in Fig. 4.1. The protein consists of 199 identified: as1-casein (as1-CN), as2-casein (as2- amino acid residues, with 8 of the 16 Ser residues CN), b-casein (b-CN) and k-casein (k-CN). in the protein being phosphorylated, i.e. Ser45, Typical concentrations of as1-CN, as2-CN, b-CN Ser47, Ser64, Ser66, Ser67, Ser68, Ser75 and Ser115 and k-CN in bovine milk are 12–15, 3–4, 9–11 (Mercier et al., 1971). In as1-CN B-9P, previously and 2–4 g L-1, respectively, and the caseins denoted as0-CN, Ser41 is also phosphorylated account for ~75–80% of total milk protein. For all caseins, various genetic variants have been identified. In addition, all caseins show consider- able micro-heterogeneity, arising from post- translational modification; all caseins are

4 Chemistry of the Caseins 137 Fig. 4.1 Amino acid sequence of bovine as1-CN B-8P (Manson et al., 1977). De Kruif and Holt (2003) bicity all suggest a moderately hydrophobic protein. as1-CN B-8P contains 25 amino acid identified two centres of phosphorylation in as1- residues capable of carrying a positive charge and CN, i.e. f41–51, containing Ser41 (only in the 9P 40 capable of carrying a negative charge. A distri- variant), Ser45 and Ser47, and f61-70, containing bution of the charge over the polypeptide chain is residues Ser64, Ser66, Ser67 and Ser68. These shown in Fig. 4.2, which clearly highlights a centres of phosphorylation are crucial in the stabi- positively charged N-terminus and a high con- centration of negative charges, including the lization of the calcium phosphate nanoclusters in two clusters of phosphorylation, between resi- dues 30 and 80. A moderate and even distribu- the casein micelles (De Kruif and Holt, 2003). tion of positive and negative charges is found between residues 81 and 150, whereas the The amino acid composition and properties of remainder of the protein, with the exception of the 10 amino acid C-terminus, is largely as1-CN B-8P are shown in Table 4.2. Based on unchanged. Distribution of hydrophobicity, amino acid composition, the molecular mass of according to the scale of Tanford (1962), of as1- CN B is also shown in Fig. 4.2. In this scale, posi- the protein prior to post-translational modification tive values represent a hydrophobic character whereas negative values represent a hydrophilic is estimated at ~23.0 kDa, which increases to character. Some distinct patches of significant hydrophobicity can be observed, i.e. residues ~23.6 kDa as a result of the phosphorylation of 20–35 and 160–175. eight Ser residues. Based on the primary sequence, a pI of ~4.9 would be expected for as1-CN, but this decreases by ~0.5 pH units through the phos- phorylation of the eight Ser residues. Such values are in line with reported pI of as1-CN varying from 4.4 to 4.8 (Trieu-Cuot and Gripon, 1981; Eigel et al., 1984). The aliphatic index, grand average hydropathicity (GRAVY) and hydropho-

138 T. Huppertz Table 4.2 Amino acid composition and properties of as1-CN B-8P Amino acid as1- CN B-8P Total residues 199 Ala 9 Arg 6 Positively charged residues (Lys/Arg/His) 25 Asn 8 Negatively charged residues (Glu/Asp/SerP) 40 Asp 7 Aromatic residues (Tyr/Phe/Thr) 20 Cys 0 Gln 14 Molecular mass Glu 25 Based on primary structure 22,975 Da Gly 9 Including phosphorylation 23,599 Da His 5 Ile 11 pI Leu 17 Based on primary structure 4.91 Lys 14 Including phosphorylation 4.42 Met 5 Phe 8 Extinction coefficient at 280 nma 25900 M-1 cm-1 Pro 17 Ser 16 Absorbance at 1 g L-1 at 280 nma 1.127 Thr 5 Trp 2 Aliphatic indexa 75.43 Tyr 10 Val 11 Grand average of hydropathicity (GRAVY)a −0.704 HFave (kJ/residue)a 4.89 aValues are based on the primary structures of the protein and do not take into account post-translational modification of the structures Hydrophobicity 1,5 1,0 0,5 0,0 -0,5 -1,0 -1,5 1 0 Charge -1 -2 0 20 40 60 80 100 120 140 160 180 200 Fig. 4.2 Distribution of hydrophobicity (top) and charged 0.25 for the amino acids located 1, 2 or 3 positions from the centre of the window. Hydrophobicity was calculated based on residues (bottom) along the amino acid chain of as1-CN B-8P. the primary amino acid sequence in the absence of post-trans- Hydrophobicity was calculated using the scale of Tanford lational modification. Charged amino acid residues include Lys (+1), Arg (+1), His (+0.5), Glu (-1), Asp (-1), SerP (-2), the (1962) with values representing the average on a 7 amino acid N-terminus (+1) and the C-terminus (-1) window with the relative weight of each amino acid in the window being 1.0 for the centre amino acid and 0.75, 0.50 and

4 Chemistry of the Caseins 139 4.3.2 Genetic Variation of as1-Casein 4.3.3 Secondary Structure of as1- Casein In addition to the B-variant of as1-CN, a number of other genetic variants have been identified, an The secondary structure of as1-CN has been overview of which is shown in Table 4.3. In as1- studied using a number of different approaches. CN A, the amino acid residues 14–26 are missing While Fourier transform infrared (FTIR) spec- as a result of exon skipping (Grosclaude et al., troscopy studies by Byler and Susi (1986) found 1970); this variant has been found in Holstein no secondary structure in as1-CN, other studies Friesians, Red Holsteins and German Red cattle have reported varying degrees of secondary struc- (Ng-Kwai-Hang et al., 1984; Grosclaude, 1988; ture elements in as1-CN. The percentage of Erhardt, 1993). Variant as1-CN C predominates a-helix in as1-CN has been estimated as 5–15% in the milk of Bos indicus and Bos grunniens (Herskovits, 1966), 8–13% (Byler et al., 1988), (Eigel et al., 1984) and contains Gly instead of 20% (Creamer et al., 1981) or 13–15% (Malin Glu at position 192 (Grosclaude et al., 1969). et al., 2005). For b-sheet, values of 17–20% were In as1-CN D, which has been found in various reported (Byler et al., 1988; Creamer et al., breeds in France (Grosclaude, 1988) and Italy 1981), whereas Malin et al. (2005) reported (Mariani and Russo, 1975) as well as in Jerseys 34–46% extended b-sheet-like structures in as1- in the Netherlands (Corradini, 1969), the Ala CN. In addition, 29–35% b-turn structures have residue at position 53 is replaced by a phospho- been reported for as1-CN (Byler et al., 1988). In rylated Thr residue (Grosclaude et al., 1972). addition, the presence of polyproline II structures A replacement at position 59 of Gln by Lys and at in as1-CN is evident from Raman optical activity position 192 of Glu by Gly yields as1-CN E, spectra (Smyth et al., 2001). Higher order struc- which is found in Bos grunniens (Grosclaude tures of caseins are described in further detail et al., 1976), whereas as1-CN F contains Leu in Chap. 5. instead of SerP at position 66 and is found in German Black and White cattle (Erhardt, 1993). 4.3.4 Self Association of as1-Casein Finally, as1-CN G was discovered in Italian Brown cows (Mariani et al., 1995), but no amino Self-association of as1-CN is characterized by acid sequence has been reported for this variant progressive strongly pH- and ionic strength- to date, whereas as1-CN H arises from an eight dependent consecutive self-association to amino acid deletion at positions 51–58 (Mahe dimers, tetramers, hexamers, etc. (Ho and Waugh, et al., 1999). 1965; Payens and Schmidt, 1965, 1966; Schmidt and van Markwijk, 1968; Swaisgood and Table 4.3 Differences in the amino acid sequence of Timasheff, 1968; Schmidt, 1970a, b). At pH 6.6 genetic variants of as1-casein compared to as1-CN B-8P and ionic strength >0.003, the monomers exist in a rapidly equilibrating equilibrium with oligomers; Position increasing ionic strength results in increasing Variant 14–26 51–58 53 59 66 192 association constants and the appearance of larger A Deleted oligomers (Ho and Waugh, 1965; Schmidt and B Ala Gln SerP Glu van Markwijk, 1968; Schmidt, 1970b). The free C Gly energy for formation of the various oligomers is D ThrP comparable; hence, all species exist at apprecia- E Lys Gly ble concentrations, but they occur to different F Leu extents. At an ionic strength of 0.003, only mono- G mers are present, whereas at an ionic strength of H Deleted 0.01, a monomer–dimer equilibrium exists; at an

140 T. Huppertz ionic strength of 0.2, dimers and tetramers are association are driven by hydrogen bonding and favoured, while the formation of larger oligomers hydrophobic interactions in the absence of elec- becomes progressively less favourable (Ho and trostatic repulsion (Aoki et al., 1985). Waugh, 1965; Schmidt, 1970b). Likewise, as the pH is increased above 6.6, the electrostatic An extensive investigation into the calcium- repulsive free energy increases, resulting in binding and calcium-induced precipitation of smaller association constants yielding a lowered as1-CN by Dalgleish and Parker (1980) high- degree of association (Swaisgood and Timasheff, lighted that the binding of calcium by the protein 1968). The larger association constants, and decreases considerably with increasing ionic resulting much stronger association, of as1-CN C strength. In addition, the concentration of calcium compared to as1-CN B can be explained by the required to induce precipitation of as1-CN also change in electrostatic free energy (Schmidt, increases with increasing ionic strength, but not 1970a) due to its smaller net charge. However, proportionally to calcium binding, i.e. the degree as1-CN D behaves identically to as1-CN B of calcium binding which is required to induce (Schmidt, 1970a) although its net charge is precipitation of as1-CN decreases with increasing greater than that of as1-CN B. It should be noted ionic strength (Dalgleish and Parker, 1980). that the as1-CN B to as1-CN D substitution, at Calcium binding by as1-CN decreases when pH position 53 (Table 4.3) occurs in the polar domain, decreases below 7.0, but decreasing pH increases whereas the as1-CN B to as1-CN C substitution, the concentration of calcium required to induce at position 192 (Table 4.3), occurs in the hydro- precipitation of as1-CN (Dalgleish and Parker, phobic domain which is more likely to be in the 1980). Calcium-induced aggregation of as1-CN association contact surface. Enzymatic deimina- was described as a monomer–octamer equilib- tion of five of the six Arg residues of as1-CN rium, followed by Smoluchowski aggregation in reduces the susceptibility of the protein to self- which only the octamers participate (Dalgleish association (Azuma et al., 1991). et al., 1981). Dephosphorylation reduces the number of calcium-binding sites on the protein 4.3.5 Interactions of as1-Casein with and also reduces the stability of as1-CN to calcium- Calcium induced precipitation (Yamuachi et al., 1967; Bingham et al., 1972; Aoki et al., 1985). When considering the interactions of as1-CN, or Deimination of Arg residues in as1-CN enhances any of the other caseins, with calcium, or other calcium binding, as well as the stability of the cations, two aspects should be considered, i.e. the protein to calcium-induced precipitation (Azuma binding of calcium by the protein and the calcium- et al., 1991). induced precipitation of the protein by calcium. as1-CN is one of the calcium-sensitive caseins; Detailed analyses of the effects of calcium precipitation of as1-CN occurs in the range of binding on as1-CN have indicated several equi- 3–8 mM CaCl2 (Schmidt, 1969; Bingham et al., libria. The addition of up to 1 mM CaCl2 to as1- 1972; Toma and Nakai, 1973; Dalgleish and CN induces an exothermic process, possibly Parker, 1980; Aoki et al., 1985; Farrell Jr et al., hydrogen-bond formation (Holt et al., 1975), 1988) and occurs more readily for as1-CN B than binding of calcium only to phosphorylated Ser for as1-CN A (Farrell Jr et al., 1988). When CaCl2 residues (Ono et al., 1976) and the transfer of Tyr concentration exceeds ~0.1 mM, the solubility and Trp residues from an aqueous to an apolar of as1-CN increases again, due to the salting-in environment (Ono et al., 1976). As the concen- effect (Farrell Jr et al., 1988). Calcium-induced tration of CaCl2 is increased from 1 to 3 mM, the precipitates of as1-CN are readily solubilized in aforementioned exothermic phase is followed by 4 M urea, suggesting that no calcium-induced an increasingly endothermic reaction, possibly cross-linkage of proteins occurred and that the hydrophobic interactions (Holt et al., 1975); the driving forces behind the calcium-induced burying of the aromatic chromophores is abated (Ono et al., 1976), whereas calcium binding by both phosphorylated Ser and carboxylate- containing residues occurs (Ono et al., 1976);

4 Chemistry of the Caseins 141 turbidity increases slightly to a plateau level exhibits varying levels of phosphorylation (Holt et al., 1975) and increasing numbers of (Swaisgood, 1992; Farrell Jr et al., 2009) and bent-chain polymers are observed (Dosaka et al., intermolecular disulfide bonding (Rasmussen 1980). Finally, between 3 and 5 mM calcium et al., 1992, 1994). The reference protein for this chloride, the reaction becomes very endothermic family is as2-CN A-11P, a single-chain polypep- (Holt et al., 1975); binding of calcium, primarily tide with an internal disulfide bond with ExPASy to carboxylate-containing residues, continues entry name and file number CAS2_Bovin and (Ono et al., 1976); the turbidity increases dramati- P02663, respectively. The primary structure of cally (Holt et al., 1975) and precipitation even- as2-CN A-11P (Fig. 4.3), reported by Brignon tually occurs. These results suggest that binding et al. (1977), has been changed to Gln rather of Ca2+ to high-affinity phosphoseryl clusters in than Glu at position 87, as indicated by cDNA the polar domain alters its interaction with the sequencing (Stewart et al., 1987) and DNA hydrophobic domain, bringing about a conforma- sequencing (Groenen et al., 1993). In addition to tional change in that domain which allows some the aforementioned 11P variant of as2-CN A, 10P, association to occur. Further binding to carboxyl 12P and 13P forms of this protein have also been residues throughout the structure reduces the observed (Brignon et al., 1976). Three centres of electrostatic repulsion and, consequently, inter- phosphorylation have been identified, i.e. f8–16, action of the hydrophobic domains leads to the which contains the phosphorylated residues Ser8, formation of large aggregates. Ser9, Ser10 and Ser16; f56–63, which contains the phosphorylated residues Ser56, Ser57, Ser58 4.4 as2-Casein and Ser61; and f126–133, which contains the phosphorylated residues Ser129 and Ser131 (De 4.4.1 Primary Structure of as2-casein Kruif and Holt, 2003). The as2-CN family constitutes up to 10% of the The primary sequence of as2-CN A-11P, as total casein fraction in bovine milk and consists outlined in Fig. 4.3, contains two Cys residues, of two major and several minor components, and i.e. Cys36 and Cys40, which occur in intra- and intermolecular disulphide bonds. In as2-CN isolated from bovine milk, >85% of the protein is Fig. 4.3 Amino acid sequence of as2-CN A-11P

142 T. Huppertz Hydrophobicity 1,5 1,0 0,5 0,0 -0,5 -1,0 -1,5 1 Charge 0 -1 -2 0 20 40 60 80 100 120 140 160 180 200 Fig. 4.4 Distribution of hydrophobicity (top) and charged 2 or 3 positions from the centre of the window. residues (bottom) along the amino acid chain of as2-CN Hydrophobicity was calculated based on the primary A-11P. Hydrophobicity was calculated using the scale of amino acid sequence in the absence of post-translational Tanford (1962) with values representing the average on a modification. Charged amino acid residues include Lys 7 amino acid window with the relative weight of each (+1), Arg (+1), His (+0.5), Glu (-1), Asp (-1), SerP (-2), amino acid in the window being 1.0 for the centre amino the N-terminus (+1) and the C-terminus (-1) acid and 0.75, 0.50 and 0.25 for the amino acids located 1, in monomeric form containing the intramolecular Some properties of as2-CN are given in disulphide bond, with the remaining fraction of Table 4.4. The 207 amino acids yield a molecular as2-CN consisting of dimers, which can be mass of ~24.3 kDa, which further increases to oriented parallel or antiparallel (Rasmussen 25.2 kDa as a result of the phosphorylation of 11 et al., 1992, 1994). Brignon et al. (1977) pointed Ser residues. For the non-phosphorylated poly- out that two very large segments of as2-CN, of peptide chain of as2-CN, a pI of ~8.3 is predicted, ~80 residues, show very high sequence homology but the aforementioned phosphorylation of 11 with each other and may arise from gene duplica- Ser residues decreases pI considerably to ~4.9. tion. Sequence alignment by Farrell Jr et al. Because of the high level of charged residues, i.e. (2009) showed that the best homologous align- 33 residues able to carry a positive charge and 39 ment was for residues 42–122 and 124–207. capable of carrying a negative charge, as2-CN is According to Farrell Jr et al., (2009), the as2-CN generally regarded as the most hydrophilic of the molecule can be divided into five distinct regions. caseins. Residues 1–41 and 42–80 form typical casein phosphopeptide regions with high charge and 4.4.2 Genetic Polymorphism of as2- low hydrophobicity, whereas residues 81–125 Casein form a slightly positively charged region of high hydrophobicity and residues 126–170 form the The A variant of as2-CN is most frequently so-called phosphopeptide analogue, with high observed in Western breeds. The B variant was negative charge but low phosphate content; observed with low frequencies in Zebu cattle in finally, residues 171–207 have high positive South Africa, but a specific site of mutation for charge and high hydrophobicity (Farrell Jr et al., as2-CN B has not been identified to date. Variant 2009). Similar trends are available from the hydro- as2-CN C was observed in yaks in the Nepalese phobicity and charge distribution in Fig. 4.4.

4 Chemistry of the Caseins 143 Table 4.4 Amino acid composition and properties of as2-CN A-11P Amino acid as2-CN A-11P Ala 8 Total residues 207 Arg 6 Positively charged residues (Lys/Arg/His) 33 Asn 14 Negatively charged residues (Glu/Asp/SerP) 39 Asp 4 Aromatic residues (Tyr/Phe/Thr) 20 Cys 2 Gln 16 Molecular mass Glu 24 Based on primary structure 24,348 Da Gly 2 Including phosphorylation 25,206 Da His 3 Ile 11 pI Leu 13 Based on primary structure 8.34 Lys 24 Including phosphorylation 4.95 Met 4 Phe 6 Extinction coefficient at 280 nma 29,005 M-1 cm-1 Pro 10 Ser 17 Absorbance at 1 g L-1 at 280 nma 1.191 Thr 15 Trp 2 Aliphatic indexa 68.7 Tyr 12 Val 14 Grand average of hydropathicity (GRAVY)a −0.918 HFave (kJ/residue)a 4.64 aValues are based on the primary structures of the protein and do not take into account post-translational modification of the structures Table 4.5 Differences in the amino acid sequence of otide sequence that encodes amino acid residues genetic variants of as2-casein compared to as2-CN A-11P 51–59 (Bouniol et al., 1993). Position 4.4.3 Secondary Structure of as2-Casein Variant 33 47 51–59 130 Estimates of the secondary structure of as2-CN A Glu Ala Thr have been obtained using a variety of techniques and show considerable differences. Garnier et al. B (1978) suggested 54% a-helix, 15% b-sheet, 19% turns and 13% unspecified structure, whereas C Gly Thr Ile Hoagland et al. (2001) suggested 24–32% a-helix, 27–37% b-sheet, 24–31% turns and D Deleted 9–22% unspecified structure. Furthermore, Tauzin et al. (2003) suggested 45% a-helix, 6% valley and the Republic of Mongolia (Grosclaude b-sheet and 49% unspecified structure, whereas et al., 1976, 1982). As shown in Table 4.5, the C 15% polyproline II structure was suggested by variant differs from the A variant at positions 33, Adzhubei and Sernberg (1993). Most recently, 47 and 130, where Gly, Thr and Ile replace Glu, Farrell et al. (2009) suggested 46% a-helix, 9% Ala and Thr, respectively (Mahe and Grosclaude, b-sheet, 12% turns, 7% polyproline II, 19% non- 1982). Variant as2-CN D was observed in continuous a-helix or b-sheet and 7% unspecified Vosgienne and Montbeliarde breeds (Grosclaude et al., 1978) and in three Spanish breeds (Osta et al., 1995). The D variant differs from as2-CN A by the deletion of nine amino acid residues from positions 51–59 (Grosclaude et al., 1978), which is caused by the skipping of exon VIII, a 27-nucle-

144 T. Huppertz secondary structure. Higher order structures of and Nakai, 1973; Aoki et al., 1985). As for as1-CN caseins are described in further detail in Chap. 5. precipitates, calcium-induced precipitates of as2-CN are readily solubilized in 4 M urea, 4.4.4 Association Properties suggesting that no calcium-induced cross-linkage of as2-Casein of proteins occurs and that the driving forces behind the calcium-induced interaction are driven Given the aforementioned amphipathic and by hydrogen bonding and hydrophobic interac- highly charged structure of as2-CN, it is not tions in the absence of electrostatic repulsion surprising that its self-association properties (Aoki et al., 1985). This is further substantiated strongly depend on ionic strength (Snoeren et al., by the fact that dephosphorylation of as2-CN 1980). as2-CN associates less extensively than renders the protein insoluble at neutral pH, prob- as1-CN, but it does exhibit consecutive self- ably due to the low net charge on the protein at associations, the extent of which at 20°C reaches these conditions (Aoki et al., 1985; Table 4.4). a maximum at an ionic strength of 0.2–0.3, but decreases at higher ionic strength (Snoeren et al., 4.5 b-Casein 1980). This perhaps unexpected decrease in asso- ciation at higher ionic strengths may be due to 4.5.1 Primary Structure of b-Casein ionic suppression of electrostatic interactions between the N-terminal and the C-terminal The b-CN family constitutes up to 35% of the domains (Snoeren et al., 1980). Snoeren et al. casein of bovine milk. The reference protein for (1980) assumed that as2-CN particles under such this family, b-CN A2-5P, contains 209 residues conditions are spherical, which is indeed apparent and its ExPASy entry name and file number are from the electron micrographs reported by CASB_Bovin and P02666, respectively. The Thorn et al. (2008). However, when as2-CN is protein was chemically sequenced by Ribadeau- incubated at higher temperatures, e.g. 37 or 50°C, Dumas et al. (1972), sequenced from its cDNA ribbon-like fibrils with a diameter of ~12 nm by Jimenez-Flores et al. (1987) and Stewart et al. and length >1 mm, which occasionally form loop (1987) and from its gene by Bonsing et al. (1988). structures, are observed (Thorn et al., 2008). The The sequence for b-CN A2-5P is shown in formation of such fibrillar structures is optimal at Fig. 4.5. This sequence was corrected from the pH 6.5–6.7 and more extensive at higher temper- original sequences by Yan and Wold (1984) and ature. The presence of as1-CN inhibits fibril Carles et al. (1988) and differs from the original formation by as2-CN, whereas the presence of sequences at four positions: Glu for Gln at posi- b-CN has little effect on as2-CN fibril formation. tions 117, 175 and 195 and reversal of Pro137 Fibril formation is also reduced when the intra- and Leu138. The changes at residues 117 and and intermolecular disulphide bonds in as2-CN 175 were confirmed by both groups and by gene are disrupted by the reducing agent, dithiothreitol sequencing, whereas the reversal of residues (Thorn et al., 2008). 137 and 138 is not in agreement with cDNA- sequencing data (Jimenez-Flores et al., 1987), 4.4.5 Interactions of as2-Casein which is in accordance with the original data. with Calcium However, the Leu-Pro substitution is a one-base change, and mutations could occur and not be Of the caseins, as2-casein has the highest number observed by HPLC-mass spectroscopy (MS) of of phosphorylated residues and is also the peptides or by electrophoresis of the proteins. most sensitive to calcium-induced precipitation. Preference is, however, given to the two afore- Calcium-induced precipitation of as2-CN occurs mentioned independent protein-sequencing at calcium concentrations less than 2 mM (Toma reports. In a similar fashion, the change at posi- tion 195 is not in agreement with the cDNA

4 Chemistry of the Caseins 145 Fig. 4.5 Amino acid sequence of b-CN A2-5P results, but, in this case, three other lines of phosphorylated Ser residues, i.e. Ser15, Ser17, evidence support the occurrence of only Glu at Ser18, Ser19 and Ser35, of which the first four form residue 195, i.e. the two protein-sequencing a centre of phosphorylation (De Kruif and Holt, corrections noted previously, the invariance on 2003). The middle section of b-CN, i.e. residues electrophoresis of b-CN (f108–209) from the A1, 41–135, contains little charge and moderate A2 and A3 genetic variants (Groves, 1969); and hydrophobicity, whereas the C-terminal, section the purification from cheese of a bitter peptide 136–209, contains many of the apolar residues b-CN (f193–209), the sequence of which is iden- and is characterized by little charge and high tical to the chemically corrected sequences hydrophobicity. (Gouldsworthy et al., 1996). 4.5.2 Genetic Polymorphism Some features of b-CN A2-5P are shown in of b-Casein Table 4.6, whereas the distribution of charge and hydrophobicity over the molecule is shown in In addition to the aforementioned A2 variant of Fig. 4.6. This 209 amino acid protein has a b-CN, a number of other genetic variants have molecular mass which is increased from 23.6 kDa been observed. The amino acid substitutions for the primary structure to 24.0 kDa following giving rise to all variants of b-CN are given in phosphorylation of the aforementioned five Ser Table 4.7. In addition, Chung et al. (1995) residues. The pI of the non-phosphorylated amino identified variant A4 in native Korean cattle using acid is estimated at 5.1, which decreases to ~4.7 electrophoresis only; its substitutions compared as a result of phosphorylation, which is some- to the A2 reference protein are thus far unknown. what lower than experimental values of 4.8–5.0 The A1 variant of b-CN differs from the A2 variant observed by Trieu-Cuot and Gripon (1981). Some only by the substitution at position 67 of His for of the unique properties of b-CN are derived from Pro (Bonsing et al., 1988), whereas the A3 variant the fact that it is strongly amphipathic. The contains Gln instead of His at position 106 N-terminus of b-CN, residues 1–40, contains (Ribadeau-Dumas et al., 1970). In addition, b- essentially all the net charge of the molecule and CN B contains the aforementioned mutation for has a low hydrophobicity and contains only two Pro residues. This section also contains the five

146 T. Huppertz Table 4.6 Amino acid composition and properties of b-CN A2-5p Amino acid b-CN A2-5P Ala 5 Total residues 209 Arg 4 Positively charged residues (Lys/Arg/His) 20 Asn 5 Negatively charged residues (Glu/Asp/SerP) 28 Asp 4 Aromatic residues (Tyr/Phe/Thr) 14 Cys 0 Gln 20 Molecular mass Glu 19 Based on primary sequence 23,583 Da Gly 5 Including phosphorylation 23,973 Da His 5 Ile 10 pI Leu 22 Based on primary sequence 5.13 Lys 11 Including phosphorylation 4.65 Met 6 Phe 9 Extinction coefficient at 280 nma 11,460 M-1 cm-1 Pro 35 Ser 16 Absorbance at 1 g L-1 at 280 nma 0.486 Thr 9 Trp 1 Aliphatic indexa 88.5 Tyr 4 Val 19 Grand average of hydropathicity (GRAVY)a −0.355 HFave (kJ/residue)a 5.58 aValues are based on the primary structures of the protein and do not take into account post-translational modification of the structures Hydrophobicity 1,5 1,0 0,5 0,0 -0,5 -1,0 -1,5 1 Charge 0 -1 -2 0 20 40 60 80 100 120 140 160 180 200 Fig. 4.6 Distribution of hydrophobicity (top) and amino acids located 1, 2 or 3 positions from the centre charged residues (bottom) along the amino acid chain of the window. Hydrophobicity was calculated based of b-CN A2-5P. Hydrophobicity was calculated using on the primary amino acid sequence in the absence of the scale of Tanford (1962) with values representing post-translational modification. Charged amino acid the average on a 7 amino acid window with the relative residues include Lys (+1), Arg (+1), His (+0.5), Glu weight of each amino acid in the window being 1.0 for (-1), Asp (-1), SerP (-2), the N-terminus (+1) and the the centre amino acid and 0.75, 0.50 and 0.25 for the C-terminus (-1)

4 Chemistry of the Caseins 147 Table 4.7 Differences in the amino acid sequence of genetic variants of b-casein compared to b-CN A2-5P Position Variant 18 25 35 36 37 67 72 88 93 106 122 137/138 152 ? Gln A1 His Glu A2 SerP Arg SerP Glu Glu Pro Gln Leu Met His Ser Leu/Pro Pro A3 Gln B His Arg C Ser Lys His D Lys E Lys F His Leu G His Leu H1 Cys Ile H2 Glu Leu I Leu the A1 variant, as well as Arg for Ser at position (2002) and contains only the Leu for Met substi- 122 (Grosclaude et al., 1974a). Likewise, b-CN tution of the H2 variant at position 93. C is also a variant of b-CN A1, which is not phos- phorylated at Ser35 and contains Lys instead of 4.5.3 Secondary Structure of b-Casein Glu at position 37. b-CN D differs from b-CN A2 only at position 18, whereas it contains Lys Originally, b-CN was predicted to have little or instead of a phosphorylated Ser residue, whereas no secondary structure and, with the exception of b-CN E contains Lys instead of Glu at position 10% a-helix, was predicted to occur as a random 36 (Grosclaude et al., 1974b). Visser et al. (1995) coil (Herskovits, 1966; Noelken and Reibstein, identified b-CN F, which contains the A1 substi- 1968), which was further supported by the results tution in addition to Leu for Pro at residue 152. of Caessens et al. (1999). The presence of a-helix Dong and Ng-Kwai-Hang (1998) identified b- structure in b-CN was further shown by Creamer CN G-5P, which is similar to b-CN A1 and F but et al. (1981), Graham et al. (1984), Farrell Jr contains a Leu in place of Pro at either position et al. (2001) and Qi et al. (2004, 2005), with 137 or 138, depending on the sequence assigned, values ranging from 7 to 25%. However, 15–33% as the Pro-Leu reversal, as outlined above, is con- b-sheet structure was also reported to be present troversial. Han et al. (2000) showed that b-CN H1 in b-CN, as well as 20–30% turns (Creamer et al., represents two substitutions relative to the cor- 1981; Graham et al., 1984; Farrell Jr et al., 2001; rected reference b-CN A2, i.e. Arg to Cys at posi- Qi et al., 2004, 2005). Using optical rotary tion 25 and Leu to Ile at position 88. A genetic dispersion analysis, Garnier (1966) suggests that variant, discovered by Senocq et al. (2002), was polyproline II could be an important feature in termed b-CN H2, which differs from the A2 vari- b-casein structure. Subsequent studies have ant at two known positions, i.e. Leu instead of indeed confirmed the presence of 20–25% poly- Met at position 93 and Glu instead of Gln at posi- proline II structure in b-CN (Farrell Jr et al., tion 72; in addition, a substitution of Gln to Glu 2001; Syme et al., 2002; Qi et al., 2004). Higher occurs somewhere between residues 114 and 169 order structures of caseins are dealt with in detail but was not located (Senocq et al., 2002). Finally, in Chap. 5. the I variant of b-CN was described by Jann et al.

148 T. Huppertz 4.5.4 Association Properties by the number of monomers in the polymer, of b-Casein estimates of which have been shown to vary from 15 to 60 (Schmidt and Payens, 1972; The presence of distinct polar and hydrophobic Buchheim and Schmidt, 1979; Takase et al., domains in b-CN clearly manifests itself in the 1980; Thurn et al., 1987; Kajiwara et al., 1988; extremely temperature-dependent self-association Farrell Jr et al., 2001); the radius of gyration, behaviour of b-CN. At 0–4°C, primarily mono- with varying estimates of 7.3–13.5 nm (Andrews mers of b-CN are observed (Payens and Van et al., 1979; Thurn et al., 1987; Kajiwara et al., Markwijk, 1963), but even under these conditions, 1988); the Stokes radius of ~15 nm (Niki et al., polymeric structure is not entirely absent (Farrell 1977; Thurn et al., 1987) and the radius observed Jr et al., 2001). The hydrodynamic behaviour of by electron microscopy of 8–17 nm (Arima b-CN under these conditions approaches that of a et al., 1979; Buchheim and Schmidt, 1979). random coil, with the Stokes radius of 3.7 nm, Increasing ionic strength shifts the equilibrium determined by gel chromatography (Schmidt and towards the polymer micelle but affects the Payens, 1972), agreeing well with values obtained number of monomers in the micelle only slightly by sedimentation and viscosity, and is also consis- (Schmidt and Payens, 1972; Takase et al., 1980), tent with the 4–5-nm size of spherical particles whereas increasing the temperature shifts the observed by electron microscopy (Andrews et al., equilibrium position and increases the number of 1979). Small angle X-ray scattering indicates a monomers in the micelle (Takase et al., 1980). In radius of gyration of 4.6 nm (Schmidt and Payens, the theoretical ratio, radius of gyration/Stokes 1972, Andrews et al., 1979). radius is 0.775 for a hard sphere (Thurn et al., 1987; Kajiwara et al., 1988), while that observed As the temperature is increased above 4–5°C, for the b-CN polymer micelle is less than 0.6, b-CN undergoes a highly cooperative, reversible, suggesting the immobilization of water in a soft rapidly equilibrating discrete self-association, outer layer surrounding a more dense core yielding large polymers with a narrow size distri- (Kajiwara et al., 1988). bution (Payens and Van Markwijk, 1963; Payens and Heremans, 1969; Payens et al., 1969; Schmidt Removal of the C-terminal three hydrophobic and Payens, 1972; Niki et al., 1977; Andrews residues, Ile-Ile-Val, greatly reduces the associa- et al., 1979; Arima et al., 1979; Buchheim and tion (Thompson et al., 1967; Evans and Phillips, Schmidt, 1979; Evans and Phillips, 1979; Takase 1979), as does removal of the C-terminal 17 et al., 1980; Schmidt, 1982; Thurn et al., 1987; amino acids (Qi et al., 2005). Removal of these Kajiwara et al., 1988; Leclerc and Calmettes, 17 amino acids (Qi et al., 2005) or the 20 1997a, b, 1998; Farrell Jr et al., 2001; De Kruif C-terminal amino acids (Berry and Creamer, and Grinberg, 2002; O’Connell et al., 2003; Qi 1975) renders b-CN virtually incapable of binding et al., 2004, 2005; Gagnard et al., 2007). The the hydrophobic surface probe ANS. The impor- properties for this monomer–polymer equilib- tance of hydrophobic interactions in the micelli- rium can be treated using a shell model for the zation of b-CN is further exemplified by the polymer micelle with a continuous distribution of enhanced micellization when H2O is replaced by intermediates between the monomer and largest D2O (Evans and Phillips, 1979) or when ethanol polymer micelle (Tai and Kegeles, 1984; De Kuif is added (Mikheeva et al., 2003) and by the and Grinberg, 2002; O’Connell et al., 2003; reduced micellization of b-CN in the presence of Mikheeva et al., 2003). There appears to be a urea (Mikheeva et al., 2003). The importance of critical concentration above which micelles are charges on the N-terminus on the micellization of formed, ranging from less than 0.5 mg/mL to b-CN is strongly impaired by the absence of post- about 2 mg/mL (Schmidt and Payens, 1972; Niki translational phosphorylation but this loss of et al., 1977; Evans et al., 1979), which depends micellization is partially restored by duplication on the temperature, ionic strength and pH. The of the 6 N-terminal amino acids of b-CN in size of the polymer micelle has been characterized expression (Gagnard et al., 2007).

4 Chemistry of the Caseins 149 4.5.5 Interactions of b-Casein b-CN at the point of precipitation are also with Calcium and Other Cations observed (Parker and Dalgleish, 1981). Both dephosphorylation and glycation of b-casein have Compared to as1-CN and as2-CN, b-CN is less been shown to improve the stability of b-casein sensitive to calcium-induced precipitation. At to calcium-induced precipitation (Darewicz 37°C, b-CN precipitates in the range of 8–15 mM et al., 1999). Ca2+ at 37°C (Schmidt, 1969; Parker and Dalgleish, 1981; Farrell Jr et al., 1988). However, 4.6 k-Casein at 1°C, b-CN remains in solution at concentra- tions up to 400 mM CaCl2 (Farrell Jr et al., 1988). 4.6.1 Primary Structure of k-Casein Under physiological conditions, b-CN is capable of binding approximately seven calcium ions per Within the caseins, k-CN displays some rather molecule (Parker and Dalgleish, 1981; Baumy unique features. It is the smallest of the caseins, and Brule, 1988). Binding of calcium by b-CN is has a low level of phosphorylation, has a low increases with increasing temperature, whereas sensitivity to calcium and is the only one of the an increase in ionic strength reduces the binding caseins to occur in glycosylated form. The pri- of calcium by b-CN (Parker and Dalgleish, 1981; mary sequence of the 169 amino acid k-CN A Baumy and Brule, 1988). In addition, the bind- 1P, which is the parent protein of the k-CN ing of calcium by b-CN decreases with decreas- family and has the ExPASy entry name CASK_ ing pH (Baumy and Brule, 1988). The binding of Bovin and file accession number P02668, is other di- and trivalent cations has also been stud- shown in Fig. 4.7. Like for the other caseins, ied; binding of magnesium, zinc and manganese variable degrees of phosphorylation have also shows comparable dependence on pH and ionic been found for k-CN. The monophosphorylated strength to the binding of calcium, whereas the form of k-CN appears to be phosphorylated binding of iron and copper by b-CN is virtually exclusively at Ser149, whereas the diphosphory- independent of pH and ionic strength (Baumy lated form of k-CN is phosphorylated at Ser149 and Brule, 1988). The amount of calcium required and Ser121 (Mercier, 1981; Minkiewicz et al., to induce precipitation of b-CN decreases 1996; Talbo et al., 2001; Holland et al., 2006). strongly with increasing temperature, whereas For the triphosphorylated form of k-CN, decreases in the amount of calcium bound by Fig. 4.7 Primary amino acid sequence of k-CN A-1P

150 T. Huppertz Table 4.8 Amino acid composition and properties of k-CN A-1P Amino acid k-CN A-1P Ala 14 Total residues 169 Arg 5 Positively charged residues (Lys/Arg/His) 17 Asn 8 Negatively charged residues (Glu/Asp/SerP) 28 Asp 4 Aromatic residues (Tyr/Phe/Thr) 14 Cys 2 Gln 15 Molecular mass Glu 12 Based in primary sequence 18,974 Da Gly 2 Including phosphorylation 19,052 Da His 3 Ile 12 pI Leu 8 Based on primary sequence 5.93 Lys 9 Including phosphorylation 5.60 Met 2 Phe 4 Extinction coefficient at 280 nma 19035 M-1 cm-1 Pro 20 Ser 13 Absorbance at 1 g L-1 at 280 nma 1.003 Thr 15 Trp 1 Aliphatic indexa 73.3 Tyr 9 Val 11 Grand average of hydropathicitya −0.557 HFave (kJ/residue)a 5.12 aValues are based on the primary structures of the protein and do not take into account post- translational modification of the structures Holland et al. (2006) recently reported that the segment 1–116, but not in the C-terminal segment additional amino acid residue to be phosphory- 117–169. Hydrophobicity distributions high- lated is not a Ser residue, but Thr145. light, as for charges, an uneven distribution of hydrophobicity throughout k-CN. Segment 1–20 Some features of k-CN A-1P are shown in shows predominantly hydrophilic behaviour, Table 4.8, whereas the distribution of hydropho- whereas segment 21–110 contains some strongly bicity and charge over the protein chain are hydrophobic patches, which is in agreement with shown in Fig. 4.8. Based on the amino acid the absence of negatively charged and a low sequence, it can be deduced that of the 169 amino number of positively charged residues in this acids, 17 can be positively charged, whereas 28 segment. Segment 110–120 is strongly hydro- can be negatively charged and there are a further philic, whereas the remainder, i.e. segment 121– 14 aromatic residues. Both hydrophobicity and 169 shows some hydrophilic and hydrophobic charge are distributed unevenly throughout the areas. It should be noted that post-translational protein (Fig. 4.8). Negative charges are found phosphorylation and glycosylation occurring in only in the N-terminal fragment 1–20 and the this part of the protein will reduce hydropho- C-terminal fragment 115–169; the intermittent bicity considerably. fragment 21–114 is devoid of negatively charged residues. Additional negative charges arising Not taking into account post-translational from phosphorylation are also in the C-terminal modification, the molecular mass of k-CN A was segment 115–169, as would be negative charges reported as 19.0 kDa. Increases in mass arise arising from glycosylation, which, as discussed from post-translational phosphorylation and later, can occur on six Thr residues in this segment. glycosylation. Based on the amino acid sequence, Positive charges can be found in the N-terminal a pI for k-CN A of ~5.9 can be expected. However,

4 Chemistry of the Caseins 151 Fig. 4.8 Distribution of hydrophobicity (top) and acids located 1, 2 or 3 positions from the centre of the charged residues (bottom) along the amino acid chain of window. Hydrophobicity was calculated based on the k-CN A-1P. Hydrophobicity was calculated using the primary amino acid sequence in the absence of post- scale of Tanford (1962) with values representing the translational modification. Charged amino acid residues average on a 7 amino acid window with the relative include Lys (+1), Arg (+1), His (+0.5), Glu (-1), Asp (-1), weight of each amino acid in the window being 1.0 for the SerP (-2), the N-terminus (+1) and the C-terminus (-1) centre amino acid and 0.75, 0.50 and 0.25 for the amino experimental observations have shown consider- substitution at position 136 of Ile for Thr and at ably lower values for the isoelectric point of k-CN, position 148 of Ala for Asp (Mercier et al., 1973). as low as pH 3.5 (Holland et al., 2006), which is The C variant of k-CN differs from k-CN A by due to increased negative charges on the protein substitution of His for Arg at position 97 (Miranda arising from post-translational phosphorylation et al., 1993). The E variant of k-CN arises from a and glycosylation. For the non-glycosylated substitution at position 155, i.e. Gly for Ser monophosphorylated variants of k-CN A and B, (Miranda et al., 1993). k-CN F1 was discovered pI values of 5.56 and 5.81 were found by two- in both Zebu and Black and White hybrid cattle dimensional electrophoresis, with consistent and contains Val instead of Asp at position 148 reductions in pI apparent with increasing degree (Sulimova et al., 1992). k-CN F2 was reported to of phosphorylation and glycosylation (Holland be a variant of k-CN B, containing His instead of et al., 2004). Arg at position ten (Prinzenberg et al., 1996). Erhardt et al. (1996) reported the occurrence of 4.6.2 Genetic Variation of k-Casein k-CN G1 in alpine breeds, which, in addition to the substitutions occurring for k-CN B, also k-CN A predominates in Western breeds, with contains Cys instead of Arg at position 97. k-CN the exception of Jerseys (Thompson and Farrell G2 was shown to occur in the milk of Bos grun- Jr, 1974; Bech and Kristiansen, 1990; Ng-Kwai- niens and was shown to contain Ala instead of Hang and Grosclaude, 2003). In addition, a number Asp at position 148. In Pinzgauer cattle, of other variants of k-CN have also been Prinzenberg et al. (1999) identified k-CN H, identified (Table 4.9). The major other variant of which differed from k-CN A by an Ile for Thr k-CN is k-CN B, which differs from k-CN A by substitution at position 135. In another study, Prinzenberg et al. (1999) described k-CN I,

152 T. Huppertz Table 4.9 Differences in the amino acid sequence of 4.6.3 Glycosylation of k-Casein genetic variants of k-casein compared to k-CN A-1P Of the caseins, k-CN is the only one for which Position post-translational glycosylation has been shown to occur. Vreeman et al. (1986) observed that Variant 10 97 104 135 136 148 155 ~40% of k-CN is non-glycosylated, whereas the remainder can contain up to six glycans. A Arg Arg Ser Thr Thr Asp Ser Glycosylation sites in k-CN were found to be the Thr residues at positions 121, 131, 133, 142, 145 B Ile Ala and 165 (Pisano et al., 1994; Molle and Leonil, 1995; Minkiewicz et al., 1996). Holland et al. C His (2004, 2005, 2006) showed that the different glycoforms of k-CN can be separated readily by E Gly 2D electrophoresis on the basis of isoelectric point and molecular mass, yielding up to 16 F1 Val different spots for k-CN with isoelectric points down to ~3.5. Such separations have laid the F2 His Ile Ala basis for the recent elucidation of the glycosyla- tion pattern of k-CN. Using tandem MS sequenc- G1 Cys Ile Ala ing of chemically tagged peptides, it was observed that the mono-glycoform of k-CN was glycosy- G2 Ala lated exclusively at Thr131, the di-glycoform exclusively at Thr131 and Thr142 and the tri- H Ile glycoform at Thr131, Thr133 and Thr142 (Holland et al., 2005). The tetra-glycoform of k- I Ala CN B was shown to be glycosylated at Thr145, in addition to the three already-mentioned gly- J Ile Ala Arg cosylation sites, Thr131, Thr133 and Thr142 (Holland et al., 2006). The remaining two glyco- which differs from k-CN A by Ala for Ser substi- sylation sites of k-CN were not confirmed by tution at position 104. Finally, Mahe et al. (1999) Holland et al. (2006) but are most likely, as pro- described the occurrence of k-CN J, which seems posed by Pisano et al. (1994) and Minkewicz to have arisen from an Arg for Ser mutation at et al. (1996), to be Thr121 and Thr165. In general, position 155 in Bos taurus cattle on the Ivory k-CN B appears to be more heavily glycosylated Coast. As outlined previously, however, k-CN A than k-CN A, also displaying a more complex and B predominate strongly in Western breeds and variable glycosylation pattern (Coolbear of cattle. et al., 1996). From a technological perspective, the Phe105- A variety of glycans have been shown to be Met106 bond in k-CN is extremely important, as attached to k-CN, all of which have been shown it is the hydrolysis of this bond by chymosin, or to be attached to Thr residues. These glycans proteinases with comparable specificity, that consist of galactose (Gal), N-acetylglucosamine initiates the gelation of milk, which will ulti- (GalNAc) and N-acetyl neuraminic acid mately be processed into a cheese curd and a (NANA). The monosaccharide GalNac, the disa- ripened or unripened cheese. The N-terminal ccharide Galb(1–3)GalNac, the trisaccharides segment 1–105 arising from the chymosin- NANAca(2–3)Galb(1–3)GalNAc and Galb(1–3) induced hydrolysis of k-CN is called para-k-CN, whereas the C-terminal fragment 106–169 is called the caseinomacropeptide (CMP); when CMP is glycosylated, it is often referred to as glycomacropeptide (GMP). From Table 4.9, it is apparent that this sequence is conserved in all genetic variants of k-CN. However, for k-CN I, the adjoining Ser104 residue is replaced by the considerably more hydrophobic Ala residue. It is also worthwhile noticing that, as outlined further in later stages, all post-translational modifi- cations of k-CN occur in the CMP segment of the molecule.

4 Chemistry of the Caseins 153 [NANAca(2–6)]GalNac and the tetrasaccharide and dithiothreitol will significantly impact the NANAca(2–3)Galb(1–3)[NANAca(2–6)] oligomeric distributions of k-CN. As outlined in GalNac have been identified attached to k-CN. Table 4.9, k-CN G1 even contains a third Cys Saito and Itoh (1992) estimated the presence of residue, i.e. Cys97. The impact hereof on the 56.0% tetrasaccharide, 18.5% branched trisac- disulphide-bonding pattern has, however, not charide, 18.4% linear trisaccharide, 6.3% disac- been studied to date. charide and 0.8% monosaccharide. 4.6.4 Disulphide-Bonding Patterns 4.6.5 Secondary Structure of k-Casein of k-Casein The secondary structure of k-CN has been stud- The presence of the two Cys residues in k-CN, ied using a number of methods. NMR studies by i.e. Cys11 and Cys88, creates a complex disul- Rollema et al. (1988) suggest a high degree of phide-bonding pattern between k-CN molecules flexibility, particularly in the macropeptide part in bovine milk. Swaisgood et al. (1964) showed of k-CN. Some structure, however, has been that k-CN obtained without reduction was appar- detected for k-CN using spectroscopic methods ently randomly cross-linked by intermolecular such as FTIR and CD. Estimates suggest that k- disulphide bonds, to give oligomers, with the CN may contain 10–20% a-helix, 20–30% smallest detectable oligomer having a mass of b-structure and 15–25% turns (Byler and Susi, ~60 kDa, corresponding to a trimer. The exis- 1986; Griffin et al., 1986; Ono et al., 1987; tence of disulphide-cross-linked oligomers has Kumosinski et al., 1991, 1993; Sawyer and Holt, since been substantiated (Talbot and Waugh, 1993; Farrell Jr et al., 1996, 2003). The degree 1970; Farrell Jr et al., 1988; Groves et al., 1992), of estimated a-helical structure in k-CN with the further suggestion that, during biosyn- increases with increasing temperature (10– thesis, reduced monomers first interact with the 70°C), while the proportion of b-structure and calcium-sensitive caseins to form micelles, fol- turns decreases with temperature (Farrell et al., lowed by random cross-linking by oxidation 2003). In addition, analysis in the presence of (Pepper and Farrell Jr, 1982). In k-CN isolated alcohols also results in a higher degree of a-helix from bovine milk, only ~10% of total k-CN in k-CN. Several structural motifs have also appears to be in the monomeric form (Farrell Jr been suggested, including possible antiparallel et al., 1996). and parallel b-sheets or bab structure in the hydrophobic domain (Raap et al., 1983) and a Both disulphide-cross-linked oligomers and b-turn-b-strand-b-turn motif centred on the chy- reduced k-CN are capable of forming polymer mosin-sensitive Phe105-Met106 region (Creamer micelles and stabilizing calcium-sensitive et al., 1998). The latter motif appears to be con- caseins (Talbot and Waugh, 1970; Vreeman, served in k-CN from various species, as would 1979). In the monomeric form of k-CN, Cys11 be expected for specific sensitivity to aspartyl and Cys88 form an intramolecular disulphide proteinases (Holt and Sawyer, 1988). Using a bond. However, k-CN complexes arising to Raman optical activity study, Syme et al. (2002) octamers and larger have also been found in identified the presence of polyproline II helical bovine milk. These complexes contain an confirmation in k-CN. Some of the predicted apparently random distribution of disulphides, structure occurs in the polar macropeptide i.e. Cys11 to Cys88, Cys11 to Cys11, Cys88 to domain but the stability of ordered structure in a Cys11 and Cys88 to Cys88. Whether these region of such high net charge and apparent patterns remain after isolation of the k-CN from hydration would seem questionable and contra- milk is strongly dependent on the physicochemical dicts the great deals of flexibility; this part of the conditions of isolation. Particularly the presence molecule was found to exhibit in the NMR stud- of reducing agents such as b-mercaptoethanol ies by Rollema et al. (1988).

154 T. Huppertz 4.6.6 Association Behavior of k-Casein reduced k-CN micelles; such measurements led De Kruif and May (1991) to conclude that When isolated from milk, k-CN occurs in the reduced k-CN micelles are spherical and consist form of multimeric complexes. Analysis by of a dense core of ~6–7 nm, surrounded by a analytical ultracentrifugation suggests that the more open outer layer, protruding up to 14.7 nm weight average molecular weight of these com- from the centre of the core. The interactions plexes is ~1,180 kDa at 25°C and ~1,550 kDa at between micelles of reduced k-CN can be 37°C (Groves et al., 1998). Electron micros- described as that of the so-called hard spheres copy studies have shown a radius of 5.0–7.5 nm, (De Kruif and May, 1991). 9–10 nm (Parry and Carroll, 1969) or 8.9 nm (Farrell Jr et al., 1996). Similar values have When reduced and carboxymethylated k-CN been observed by gel permeation chromatogra- was incubated at 37°C, it was observed in addi- phy (9.4 nm; Pepper and Farrell Jr, 1982), tion to spherical particles, there was also a high dynamic light scattering (9.6 nm; Farrell Jr proportion of fibrillar structures present (Farrell et al., 1996) and small angle neutron scattering Jr et al., 2003). The formation of such fibrillar (SANS), for which values for values of a radius structures, with a diameter of 10–12 nm and of 7.4 nm (Thurn et al., 1987) and 8 nm (De lengths up to 600 nm, was subsequently shown to Kruif et al., 2002) have been reported. Micelle occur for native, reduced and carboxymethylated size, structure and interaction radius were found k-CN (Thorn et al., 2005; Ecroyd et al., 2008, to be independent of protein concentration (De 2010; Leonil et al., 2008). When native k-CN is Kruif et al., 2002). Both calcium and iron have used, it is the dissociated form that is involved been found to be present in isolated k-CN, and in fibril formation (Ecroyd et al., 2010). Fibril their chelation by EDTA has been reported to formation, which has been shown to result in an result in disruption of the k-CN particle, with increased proportion of b-sheet structure subsequent aggregation into particles with a (Ecroyd et al., 2008; Leonil et al., 2008), is more considerably broader size distribution (Farrell extensive at higher temperature (Thorn et al., Jr et al., 1996). 2005) and is more extensive for non-glycosylated k-CN than for its glycosylated counterpart Reduction of the disulphide bridges in afore- (Leonil et al., 2008). The presence of as-CNs or mentioned k-CN particles leads to amphipathic b-CN inhibits fibril formation (Thorn et al., monomers which can, like b-CN, associate into 2005; Leonil et al., 2008), whereas BSA does not micellar structures; unlike the micellization of b- inhibit fibril formation (Thorn et al., 2005). CN, micellization of reduced k-CN shows no Segment Tyr25-Lys86 of k-CN appears to be strong temperature dependence (Swaisgood incorporated into the protease-resistant core of et al., 1964; Vreeman et al., 1981). This suggests the fibrils (Ecroyd et al., 2008) whereas fragment that micellization of reduced k-CN is less domi- 106–169, i.e., the macropeptide, in either glyco- nated by hydrophobic interactions than micelli- sylated or non-glycosylated form, does not form zation of b-CN. For the monomer–polymer fibrils under comparable circumstances (Leonil micelle equilibrium of reduced k-CN, the critical et al., 2008). micelle concentration varies from 0.53 at an ionic strength of 0.1–0.24 mg/mL at an ionic strength 4.6.7 Interactions of k-Casein of 1.0 (Vreeman, 1979; Vreeman et al., 1977, with Calcium 1981). The degree of polymerization has been estimated at ~30 k-CN molecules per micelle, Compared to the other caseins, interactions of yielding a molecular mass of ~570–600 kDa calcium with k-CN have studied far less. This is (Vreeman, 1979; Vreeman et al., 1981, 1986) probably due to the fact that k-CN is, unlike as1- and an estimated diameter of 23 nm (Vreeman CN, as2-CN and b-CN, the so-called calcium et al., 1981). Such results are in agreement with insensitive, i.e., it is not precipitated in the pres- values derived from SANS measurements on

4 Chemistry of the Caseins 155 ence of excess calcium. Ono et al. (1980), studying Brignon, G., Ribadeau-Dumas, B. and Mercier, J.-C. the binding of calcium to k-CN, observed that (1976). Premiers elements de structure primaire des binding of calcium to phosphorylated Ser resi- caseines as2 bovines. FEBS Lett. 71, 111–116. dues reached a plateau at 1 mM CaCl2, whereas binding of calcium by carboxyl groups increased Brignon, G.B., Ribadeau-Dumas, B., Mercier, J.-C., linearly up to 3 mM CaCl2 and more slowly at Pelissier, J.-P. and Das, B.C. (1977). The complete higher concentrations. Spectra obtained from amino acid sequence of bovine aS2-casein. FEBS Lett. circular dichroism and UV analysis indicate that 76, 274–279. the binding of calcium to k-CN does not induce changes in the secondary structure of the protein Buchheim, W. and Schmidt, D.G. (1979). On the size of (Ono et al., 1980). Given the aforementioned monomers and polymers of b-casein. J. Dairy Res. 46, potential role of calcium in stabilizing k-CN 277–280. particles isolated from milk (Farrell et al., 1996), further study on the interactions of calcium with Byler, D.M. and Susi, H. (1986). Examination of the k-casein appears warranted. secondary structure of proteins by deconvoluted FTIR spectra. Biopolymers 25, 469–487. References Byler, D.M., Farrell, H.M., Jr. and Susi, H. (1988). Raman Adzhubei, A.A. and Sternberg, M.J.E. (1993). Left- spectroscopic study of casein structure. J. Dairy Sci. handed polyproline II helices commonly occur in 71, 2622–2629. globular proteins. J. Mol. Biol. 229, 472–493. Caessens, P.W.J.R., De Jongh, H.H.J., Noide, W. and Andrews, A.L., Atkinson, D., Evans, M.T.A., Finer, E.G., Gruppen, H. (1999). The absorption-induced secondary Green, J.P., Phillips, M.C. and Robertson, R.N. (1979). structure of b-casein and of distinct parts of its The conformation and aggregation of bovine b-casein sequence. Biochim. Biophys. Acta 1430, 73–83. A. I. Molecular aspects of thermal aggregation. Biopolymers 18, 1105–1121. Carles, C., Huet, J.-C. and Ribadeau-Dumas, B. (1988). A new strategy for primary structure determination of Aoki, T., Toyooka, K. and Kako, Y. (1985). Role of proteins: application to b-casein. FEBS Lett. 229, phosphate groups in the calcium sensitivity of as2- 265–272. casein. J. Dairy Sci. 68, 1624–1629. Chung, E.R., Han, S.K. and Rhim, T.J. (1995). Milk Arima, S., Niki, R. and Takase, K. (1979). Structure of protein polymorphisms as genetic markers in Korean b-casein. J. Dairy Res. 46, 281–282. native cattle. Asian Austral. J. Anim. Sci. 8, 187–194. Azuma, N., Nara, K. and Kanno, C. (1991). Enzymic Coolbear, K.P., Elgar, D.F. and Ayers, J.S. (1996). Profiling modification of as1-casein with peptidylarginine deimi- of genetic variants of bovine k-casein macropeptide nase: preparation of less acid-coagulable and less cal- by electrophoretic and chromatographic techniques. cium-sensitive casein. J. Dairy Res. 58, 421–429. Int. Dairy J. 6, 1055–1068. Baumy, J.J. and Brule, G. (1988). Effect of pH and ionic Corradini, C. (1969). Distribution of the genetic variants strength on the binding of bivalent cations to b-casein. as1-, b-, and k-caseins in milk from Jersey cows in The Lait 68, 409–418. Netherlands. Neth. Milk Dairy J. 23, 79–82. Bech, A.M. and Kristiansen, K.R. (1990). Milk protein poly- Creamer, L.K., Richardson, T. and Parry, D.A.D. (1981). morphism in Danish dairy cattle and the influence of Secondary structure of bovine as1- and b-casein in genetic variants on milk yield. J. Dairy Res. 57, 53–62. solution. Arch. Biochem. Biophys. 211, 689–696. Berry, G.P. and Creamer, L.K. (1975). The association of Creamer, L.K., Plowman, J.E., Liddell, M.J., Smith, M.H. bovine b-casein. The importance of the C-terminal and Hill. J.P. (1998). Micelle stability: k-casein struc- region. Biochem. 14, 3542–3545. ture and function. J. Dairy Sci. 81, 3004–3012. Bingham, E.W., Farrell, H.M., and Carroll, R.J., Jr. (1972). Dalgleish, D.G. and Parker, T.G. (1980). Binding of Properties of dephosphorylated as1-casein. Preci- calcium ions to bovine as1-casein and precipitability pitation by calcium ions and micelle formation. of the protein-calcium ion complexes. J. Dairy Res. Biochem. 11, 2450–2454. 47, 113–122. Bonsing, J., Ring, J.M., Stewart, A.F. and MacKinlay, A.G. Dalgleish, D.G., Paterson, E. and Horne, D.S. (1981). (1988). Complete nucleotide sequence of the bovine Kinetics of aggregation of as1-casein/Ca2+ mixtures: beta-casein gene. Aust. J. Biol. Sci. 41, 527–537 charge and temperature effects. Biophys. Chem. 13, 307–314. Bouniol, C., Printz, C. and Mercier J.-C. (1993). Bovine aS2-casein D is generated by exon VIII skipping. Gene Darewicz, M., Dziuba, J., Mioduszewska, H. and 128, 289–293. Minkiewicz, P. (1999). Modulation of physico-chemical properties of bovine b-casein by non-enzymatic gly- cation associated with enzymatic dephosphorylation. Acta Aliment. 28, 339–354. De Kruif, C.G. and Grinberg, V.Y. (2002). Micellisation of b-casein. Colloid. Surface. A 210, 183–190. De Kruif, C.G. and Holt, C. (2003). Casein micelle struc- ture, functions and interactions, in Advanced Dairy Chemistry 1: Proteins, 3rd edn. P.F. Fox and P.L.H. McSweeney, eds., Kluwer Academic/Plenum Publishers, New York. pp. 233–276.

156 T. Huppertz De Kruif, C.G. and May, R.P. (1991). k-Casein micelles: Farrell, H.M., Jr., Jimenez-Flores, R., Bleck, G.T., Brown, structure, interaction and gelling studied by small- E.M., Butler, J.E., Creamer, L.K., Hicks, C.L., Hollar, angle neutron scattering. Eur. J. Biochem. 200, C.M., Ng-Kwai-Hang, K.F. and Swaisgood, H.E. 431–436. (2004). Nomenclature of the proteins of cows’ milk: sixth revision. J. Dairy Sci. 87, 1641–1674. De Kruif, C.G., Tuinier, R., Holt, C., Timmins, P.A. and Rollema, H.S. (2002). Physicochemical study of k- Farrell, H.M. Jr., Malin, E.L., Brown, E.M. and Mora- and b-casein dispersion and the effect of cross-linking Gutierrez, A. (2009). Review of the chemistry of as2- by transglutaminase. Langmuir 18, 4885–4891. casein and the generation of a homolgous molecular model to explain its properties. J. Dairy Sci. 92, Dong, C. and Ng-Kwai-Hang, K.F. (1998). Characterization 1338–1353. of a nonelectrophoretic genetic variant of b-casein by peptide mapping and mass spectroscopic analysis. Gagnard, S., Zuev, Y., Gaudin, J.-C., Fedotov, V., Choiset, Int. Dairy J. 8, 967–972. Y., Axelos, M.A.V., Chobert, J.-M. and Haertle, T. (2007). Modifications of the charges at the N-terminus Dosaka, S., Kimura, T., Taneya, S., Sone, T., Kaminogawa, of bovine b-casein: consequences on its structure and S. and Yamauchi, K. (1980). Polymerization of as1- its micellisation. Food Hydrocolloid 21, 180–190. casein by calcium ions. Agric. Biol. Chem. 44, 2443–2448. Garnier, J. (1966). Conformation of b-casein in solution. Analysis of a thermal transition between 5 and 40°C. Ecroyd, H., Koudelka, T., Thorn, D.C., Williams, D.M., J. Mol. Biol. 19, 586–590. Devlin, G., Hoffman, P. and Carver, J.A. (2008). Dissociation from the oligomeric state is the rate- Garnier, J., Osguthorpe, D.J. and Robinson, B. (1978). limiting step in fibril formation by k-casein. J. Biol. Analysis of the accuracy and implications of simple Chem. 283, 9012–9022. methods for predicting the secondary structure of globular proteins. J. Mol. Biol. 120, 97–120. Ecroyd, H., Thorn, D.C., Lui, Y. and Carver, J.A. (2010). The dissociated form of k-casein is the precursor to its Gouldsworthy, A.M., Leaver, J. and Banks, J.M. (1996). amyloid fibril formation. Biochem. J. 429, 251–260. Application of a mass spectrometry sequencing tech- nique for identifying peptides present in Cheddar Eigel, W.N., Butler, J.E., Ernstrom, C.A., Farrell, H.M., cheese. Int. Dairy J. 6, 781–791. Jr., Harwalkar, V.R., Jenness, R. and Whitney, R.M. (1984). Nomenclature of proteins of cow’s milk: fifth Graham, E.R.B., Malcolm, G.N. and McKenzie, H.A. revision. J. Dairy Sci. 67, 1599–1631. (1984). On the isolation and conformation of bovine b-casein A1. Int. J. Biol. Macromol. 6, 155–161. Erhardt, G. (1993). A new as1-casein allele in bovine milk and its occurrence in different breeds. Anim. Genet. Griffin, M.C.A., Price, J.C. and Martin, S.R. (1986). 24, 65–66. Effect of alcohols on the structure of caseins: circular dichroism studies of k-casein A. Int. J. Biol. Macromol. Erhardt, G. (1996). Detection of a new k-casein variant in 8, 367–371. the milk of Pinzgauer cattle. Anim. Genet. 27, 105–107. Groenen, M.A.M., Dijkhof, R.E.M., Verstege, A.J.M. and Evans, M.T.A. and Phillips, M.C. (1979). The conforma- Van der Poel, J.J. (1993). The complete sequence of the tion and aggregation of bovine beta-casein A. II. gene encoding bovine aS2-casein. Gene 123, 187–193. Thermodynamics of thermal association and the effects of changes in polar and apolar interactions on Grosclaude, F. (1988). Le polymorphisme genetique des micellization. Biopolymers 18, 1123–1140. principales actoproteines bovines. Relations avec la quantite, la composition et les aptitudes fromageres du Farrell, H.M. Jr., Kumosinski, T.F., Thompson, M.P. and lait. INRA Prod. Anim. 1, 5–17. Pulaski, P. (1988). Calcium induced associations of the caseins: a thermodynamic linkage approach to Grosclaude, F., Mercier, J.-C. and Ribadeau-Dumas, B. precipitation and resolubilization. Arch. Biochem. (1969). Sur la localisation, dans la séquence COOH- Biophys. 265, 146–158. terminale de la caseine alpha-S1 bovine, de la sub- stitution GLU-GLY différenciant les variants Farrell, H.M. Jr., Kumosinski, T.F., Cooke, P.H., King, G., génétiques B et C. CR Acad. Sci. D: Sci. Natur. 268, Hoagland, P.D., Wickham, E.D., Dower, H.J. and 3133–3136. Groves, M.L. (1996). Particle size of purified k-casein: metal effect and correspondence with predicted three- Grosclaude, F., Mahe, M.-F., Mercier, J.-C. and Ribadeua- dimensional molecular models. J. Prot. Chem. 15, Dumas, B. (1970). Localisation, dans la partie NH2- 435–445. terminale de la caséine as1 bovine, d’une délétion de 13 acides aminés différenciant le variant a des variants Farrell, H.M. Jr., Wickham, E.D., Unruh, J.J., Qi, P.X. and B et C. FEBS Lett. 11, 109–112. Hoagland, P.D. (2001). Secondary structure studies of bovine caseins: temperature dependence of b-casein Grosclaude, F., Mahe, M.-F., Mercier, J.-C. and Ribadeau- structure as analyzed by circular dichroism and Dumas, B. (1972). Caracterisation des variants genet- FTIR spectroscopy and correlation with micellization. iques des caseines as1 et b bovines. Eur. J. Biochem. Food Hydrocolloid 15, 341–354. 26, 328–337. Farrell, H.M. Jr., Cooke, P.H., Wickham, E.D., Piotrowski, Grosclaude, F., Mahe, M.-F. and Ribadeau-Dumas, B. E.G. and Hoagland, P.D. (2003). Environmental (1973). Structure primaire de la caseine as1- et de la influences of bovine k-casein: reduction and conver- caseine b-bovine. Eur. J. Biochem. 40, 323–324. sion to fibrillar (amyloid) structures. J. Prot. Chem. 22, 259–273. Grosclaude, F., Mahe, M.-F. and Mercier, J.-C. (1974a). Comparison du polymorphisme genetique des lacto-

4 Chemistry of the Caseins 157 proteines du zebu et des bovines. Ann. Genet. Sel. calcium chloride. Biochim. Biophys. Acta 379, Anim. 6, 305–329. Grosclaude, F., Mahe, M.-F. and Voglino, G.-F. (1974b). 638–644. Le variant bE et le code de phosphorylation des caseins bovines. FEBS Lett. 45, 3–5. Jann, O., Ceriotti, G., Caroli, A. and Erhardt, G. (2002). Grosclaude, F., Mahe, M.-F., Mercier, J.-C., Bonnemaire, J. and Teissier, J.H. (1976). Polymorphisme des A new variant in exon VII of the bovine b-casein lactoproteines de bovines Nepalais. Polymorphisme des caseine as2-mineurs. Ann. Genet. Sel. Evol. 8, gene (CSN2) and its distribution among European 461–479. Grosclaude F., Joudrier, P. and Mahe, M.-F. (1978). cattle breeds. J. Anim. Breed. Genet. 119, 65–68. Polymorphisme de la caseine aS2 bovine: Etroite liaison du locus aS2-Cn avec les loci as1-Cn, b-Cn et Jenness, R., Larson, B.L., McMeekin, T.L., Swanson, k-Cn; mise en evidence d’une deletion dans le variant aS2-Cn D. Ann. Genet. Sel. Anim. 10, 313–327. A.M., Whitnah, C.H. and Whitney, R.M. (1956). Grosclaude, F., Mahe, M.-F. and Accolas, J.P. (1982). Note sur le polymorphisme genetique des lactopro- Nomenclature of the proteins of bovine milk. J. Dairy teines de bovins et de yaks Mongols. Ann. Genet. Sel. Anim. 14, 545–550. Sci. 39, 536–541. Groves, M.L. (1969). Some minor components of casein and other phosphoproteins in milk. A review. J. Dairy Jimenez-Flores, R., Kang, Y.C. and Richardson, T. (1987). Sci. 52, 1155–1165. Groves, M.L., Dower, H.J. and Farrell, H.M. Jr. (1992). Cloning and sequence analysis of bovine b-casein Reexamination of the polymeric distributions of k-casein isolated from bovine milk. J. Prot. Chem. 11, cDNA. Biochem. Biophys. Res. Comm. 142, 617–621. 21–27. Groves, M.L., Wickham, E.D. and Farrell, H.M. Jr. (1998). Kajiwara, K., Niki, R., Urakawa, H., Hiragi, Y., Donkai, Environmental effects on disulphide bonding patterns of bovine k-casein. J. Prot. Chem. 17, 73–84. N. and Nagura, M. (1988). Micellar structure of Han, S.K., Shin, Y.C. and Byun, H.D. (2000). Biochemical, molecular and physiological characterization of a new b-casein observed by small-angle X-ray scattering. b-casein variant detected in Korean cattle. Anim. Genet. 31, 49–51. Biochim. Biophys. Acta 955, 128–134. Herskovits, T.T. (1966). On the conformation of casein. ORD properties. Biochem. 5, 1018–1026. Kumosinski, T.F., Brown, E.M. and Farrell, H.M. Jr. Ho, C. and Waugh, D.F. (1965). Interactions of bovine as- casein with small ions. J. Am. Chem. Soc. 87, (1991). Three-dimensional molecular modeling of 110–117. Hoagland, P.D., Unruh, J.J., Wickham, E.D. and Farrell, bovine caseins: k-casein. J. Dairy Sci. 74, 2879–2887. H.M. Jr. (2001). Secondary structure of bovine as2- casein: theoretical and experimental approaches. Kumosinski, T.F., Brown, E.M. and Farrell, H.M. Jr. J. Dairy Sci. 84, 1944–1949. Holland, J.W., Deeth, H.C. and Alewood, P.F. (2004). (1993). Refined molecular modeling of bovine Proteomic analysis of k-casein microheterogeniety. Proteomics 4, 743–752. caseins: a refined energy-minimized k-casein struc- Holland, J.W., Deeth, H.C. and Alewood, P.F. (2005). Analysis of O-glycosylation site occupancy in bovine ture. J. Dairy Sci. 76, 2507–2520. k-casein glycoforms separated by two-dimensional gel electrophoresis. Proteomics 5, 990–1002. Leclerc, E. and Calmettes, P. (1997a). Interactions in Holland, J.W., Deeth, H.C. and Alewood, P.F. (2006). Resolution and characterization of multiple isoforms micellar solutions of b-casein. Physica B 234–236, of bovine k-casein by 2-DE followed by reversible cysteine-tagging enrichment strategy. Proteomics 6, 207–209. 3087–3095. Holt, C. and Sawyer, L. (1988). Primary and predicted Leclerc, E. and Calmettes, P. (1997b). Interactions in secondary structures of the caseins in relation to their biological functions. Protein Eng. 2, 251–259. micellar solutions of b-casein. Phys. Rev. Lett. 78, Holt, C., Parker, T.G. and Dalgleish, D.G. (1975). The thermochemistry of reactions between as1-casein and 150–153. Leclerc, E. and Calmettes, P. (1998). Structure of b-casein micelles. Physica B 241–243, 1141–1143. Leonil, J., Henry, G., Jouanneau, D., Delage, M.-M., Forge, V. and Putaux, J.-L. (2008). Kinetics of fibril formation of bovine k-casein indicate a conforma- tional reaarangement as a critical step in the process. J. Mol. Biol. 381, 1267–1280. Mahe, M.-F. and Grosclaude, F. (1982). Polymorphisme de la caseine a des bovines: characterization du vari- S2 ant C du yak (Bos grunniens). Ann. Genet. Sel. Anim. 14, 401–416. Mahe, M.-F., Miranda, G., Queral, R., Bado, A., Souvenir- Zafidrajaona, P. and Grosclaude, F. (1999). Genetic polymorphism of milk proteins in African Bos taurus and Bos indicus populations characterization of variants aS1-Cn H and k-Cn. J. Genet. Sel. Evol. 31, 239–253. Malin, E.L., Brown, E.M., Wickham, E.D. and Farrell, H.M. Jr. (2005). Contributions of terminal peptides to associative behavior of as1-casein. J. Dairy Sci. 88, 2318–2328. Manson, W., Carolan, T. and Annan, W.D. (1977). Bovine as0-casein; a phosphorylated homologue of as1-casein. Eur. J. Biochem. 78, 411–417. Mariani, P. and Russo, V. (1975). Varianti genetiche delle a protein del latte nella razza. Rendena. Riv. Zootec. Vet. 3, 345–348. Mariani, P., Summer, A., Anghinetti, A., Senese, C., Di Gregorio, P., Rando, P. and Serventi, P. (1995). Effects

158 T. Huppertz of the as1-CN G allele on the percentage distribution of k-casein and glycomacropeptide. Biochim. Biophys. caseins as1-, as2-, b-, and k- in Italian Brown cows. Acta. 911, 318–325. Ind. Latte 31, 3-13. Osta, R., Marcos, S. and Rodellar, C. (1995). A MnlI Mercier, J.-C. (1981). Phosphorylation of the caseins, polymorphism at the bovine as2-casein gene. Anim. present evidence for an amino acid triplet code Genet. 26, 213. posttranslationally recognized by specific kinases. Parker, T.G. and Dalgleish, D.G. (1981). Binding of Biochimie 63, 1–17. calcium ions to bovine b-casein. J. Dairy Res. 48, Mercier, J.-C., Grosclaude, F. and Ribadeau-Dumas, B. 71–76. (1971). Structure primaire de la caseine as1 bovine. Parry, R.M. and Carroll, R.J. (1969). Location of k-casein in Sequence complete. Eur. J. Biochem. 23, 41–51. milk micelles. Biochim. Biophys. Acta 194, 138–150. Mercier, J.-C., Brignon, G. and Ribadeau-Dumas, B. Payens, T.A.J. and Heremans, K. (1969). Effect of pres- (1973). Structure primarie de la caseine k-bovine B. sure on the temperature-dependent association of Sequence complete. Eur. J. Biochem. 35, 222–235. b-casein. Biopolymers 8, 335–345. Mikheeva, L.M., Grinberg, N.V., Grinberg, V.Ya., Payens, T.A.J. and Schmidt, P.G. (1965). The thermody- Khokhlov, A.R. and De Kruif, C.G. (2003). namic parameters of the association of as1-casein C. Thermodynamics of micellization of bovine b-casein Biochim. Biophys. Acta 109, 214–222. by high-sensitivity differential scanning calorimetry. Payens, T.A.J. and Schmidt, D.G. (1966). Boundary Langmuir 19, 2913–2921. spreading of rapidly polymerizing as1-casein B and C Minkiewicz, P., Slangen, C.J., Lagerwerf, F.M., during sedimentation. Arch. Biochem. Biophys. 115, Haverkamp, J., Rollema, H.S. and Visser, S. (1996). 136–145. Reversed-phase high-performance liquid chromato- Payens, T.A.J. and Van Markwijk, B.W. (1963). Some graphic separation of bovine k-casein macropeptide features of the association of b-casein. Biochim. and characterization of isolated fractions. J. Chrom. Biophys. Acta 71, 517–530. A 743, 123–135. Payens, T.A.J., Brinkhuis, J.A. and Van Markwijk, B.W. Miranda, G., Anglade, P., Mahe, M.-F. and Erhardt, G. (1969). Self-association in non-ideal systems. (1993). Biochemical characterization of the bovine Combined light scattering and sedimentation measure- genetic kappa-casein C and E variants. Anim. Genet. 4, ments in b-casein solutions. Biochim. Biophys. Acta 27–31. 175, 434–437. Molle, D. and Leonil, J. (1995). Heterogeneity of the Pepper, L. and Farrell, H.M. Jr. (1982). Interactions bovine k-casein caseino-macropeptide, resolved by leading to formation of casein submicelles. J. Dairy liquid chromatography on-line with electrospray Sci. 65, 2259–2266. ionization mass spectrometry. J. Chrom. A 708, Pisano, A., Packer, N.H., Redmond, J.W., Williams, K.L. 223–230. and Gooley, A.A. (1994). Characterization of O-linked Ng-Kwai-Hang, K.F. and Grosclaude, F. (2003). Genetic glycosylation motifs in the glycopeptides domain of polymorphism of milk proteins, in Advanced Dairy bovine k-casein. Glycobiology 4, 837–844. Chemistry 1: proteins, 3rd edn., P.F Fox and P.L.H. Prinzenberg, E.-M., Hiendleder, S., Ikonen, T. and Erhardt, McSweeney, eds., Kluwer Acedemic/Plenum G. (1996). Molecular genetic characterization of new Publishers, New York. pp. 739-816. bovine k-casein alleles CSN3-F and CSN3-G and Ng-Kwai-Hang, K.F., Hayes, J.F., Moxley, J.E. and genotyping by PCR-RFLP. Anim. Genet. 27, Monardes, H.G. (1984). Association of genetic vari- 347–349. ants of casein and milk serum proteins with milk, fat, Prinzenberg, E.-M., Krause, I. and Erhardt, G. (1999). and protein production by dairy cattle. J. Dairy Sci. SCCP analysis of the bovine CSN3 locus discrimi- 67, 835–840. nates six alleles corresponding to known protein Niki, V.R., Takase, K. and Arima, S. (1977). Über die variants (A, B, C, E, F, G) and three new DNA gestalt and gröbe des temperaturabhängigen assoziats polymorphism (H, I, A1). Anim. Biotechnol. 10, von b-casein. Milchwissenschaft 32, 577–582. 49–62. Noelken, M. and Reibstein, M. (1968). The conformation Qi, P.X., Wickham, E.D. and Farrell, H.M. Jr. (2004). of b-casein B. Arch. Biochem. Biophys. 123, 397. Thermal and alkaline denaturation of bovine b-casein. O’Connell, J.E., Grinberg, V.Ya. and De Kruif, C.G. Prot. J. 23, 389–402. (2003). Association behavior of b-casein. J. Colloid Qi, P.X., Wickham, E.D., Piotrowski, E.G., Fagerquist, Interf. Sci. 258, 33–39. C.K. and Farrell, H.M. Jr. (2005). Implication of Ono, T., Kaminogawa, S., Odagiri, S. and Yamauchi, K. C-terminal deletion on the structure and stability of (1976). A study on the binding of calcium ions to as1- bovine b-casein. Prot. J. 24, 431–444. casein. Agric. Biol. Chem. 40, 1717–1723. Raap, J., Kerling, K.E.T., Vreeman, H.J. and Visser, S. Ono, T., Yahagi, M. and Odagiri, S. (1980). The binding (1983). Peptide substrates for chymosin (rennin): of calcium to k-casein and para-k-casein. Agric. Biol. conformational studies of k-casein-related oligo- Chem. 44, 1499–1503. peptides by circular dichroism and secondary struc- Ono T., Yada, R., Yunati, K. and Nakai, S. (1987). ture prediction. Arch. Biochem. Biophys. 221, Comparison of the comformations of k-casein, para- 117–124.

4 Chemistry of the Caseins 159 Rasmussen, L.K., Hojrup, P. and Petersen, T.E. (1992). Sulimova, G.E., Sokolova, S.S., Semikozova, O.P., Nguet, Localization of two interchain disulfide bridges in L.M. and Berberov, E.M. (1992). Analysis of DNA dimers of bovine aS2-casein. Eur. J. Biochem. 203, polymorphisms of clustered genes in cattle: casein 381–386. genes and genes of the major histocompatibility complex (BOLA). Tsitol. I Genetika 26, 18–26. Rasmussen, L.K., Hojrup, P. and Petersen, T.E. (1994). Disulphide arrangement in bovine caseins: localiza- Swaisgood, H. E. (1982). Chemistry of milk proteins, in, tion of intrachain disulphide bridges in monomers of Developments in Dairy Chemistry 1: Proteins. P.F. Fox, k- and as2-casein from bovine milk. J. Dairy Res. 61, ed., Applied Science Publishers, New York. pp. 1–59. 485–493. Swaisgood, H.E. (1992). Chemistry of the caseins, in, Ribadeau-Dumas, B., Grosclaude, F. and Mercier, J.-C. Advanced Dairy Chemistry 1: Proteins, 2nd edn. P.F. (1970). Localization in the peptide chain of bovine Fox, ed., Elsevier Applied Science, New York. beta casein of the His-Gln substitution differentiating pp. 63–110. the A2 and A3 genetic variants. CR Acad. Sci. D: Sci. Natur. 270, 2369–2372. Swaisgood, H.E. (2003). Chemistry of caseins, in, Advanced Dairy Chemistry 1: Proteins, 3rd edn. P.F. Ribadeau-Dumas, B., Brignon, G., Grosclaude, F. and Fox and P.L.H. McSweeney, eds., Kluwer Academic/ Mercier, J.-C. (1972). Structure primaire de la casein b Plenum Publishers, New York. pp. 139–201. bovine. Eur. J. Biochem. 25, 505–514 Swaisgood, H.E. and Timasheff, S.N. (1968). Association Rollema, H.S., Brinkhuis, J.A. and Vreeman, H.J. (1988) of as1-casein C in the alkaline pH range. Arch. 1H-NMR studies of bovine kappa-casein and casein Biochem. Biophys. 125, 344–361. micelles. Neth. Milk Dairy J. 42, 233–248. Swaisgood, H.E., Brunner, J.R. and Lillevik, H.A. (1964). Saito, T. and Itoh, T. (1992). Variations and distributions Physical parameters of k-casein from cow’s milk. of O-glycosidically linked sugar chains in bovine Biochem. 3, 1616–23. k-casein. J. Dairy Sci. 75, 1768–1774. Syme, C.D., Blanch, E.W., Holt, C., Jakes, R., Goedert, Sawyer, L. and Holt, C. (1993). The secondary structure M., Hecht, L. and Barron, L.D. (2002). A Raman opti- of milk proteins and their biological function. J. Dairy cal activity study of rheomorphism in casein, synucleins Sci. 76, 3062–3078. and tau. New insight into the structure and behavior of natively unfolded proteins. Eur. J. Biochem. 269, Schmidt, D.G. (1969). On the Association of as1-Casein, 148–156. Ph.D. Thesis, University of Utrecht. Tai, M.-S. and Kegeles, G. (1984). A micelle model for Schmidt, D.G. (1970a). Differences between the associa- the sedimentation behavior of bovine b-casein. tion of the genetic variants B, C and D of as1-casein. Biophys. Chem. 20, 81–87. Biochim. Biophys. Acta 221, 140–142. Takase, K., Niki, R. and Arima, S. (1980). A sedimenta- Schmidt, D.G. (1970b). The association of as1-casein B at tion equilibrium study of the temperature-dependent pH 6.6. Biochim. Biophys. Acta 207,130–138. association of bovine b-casein. Biochim. Biophys. Acta 622, 1–8. Schmidt, D.G. (1982). Association of caseins and casein micelle structure, in, Developments in Dairy Chemistry Talbo, G.H., Suckau, D., Malkoski, M. and Reynolds, 1, P.F. Fox, ed., Elsevier Applied Science, London. E.C. (2001). MALDI-PSD-MS analysis of the phos- pp. 61–86. phorylation sites of caseinomacropeptide. Peptides 22, 1093–1098. Schmidt, D.G. and Payens, T.A.J. (1972). The evaluation of positive and negative contributions to the second Talbot, B. and Waugh, D.F. (1970). Micelle-forming char- virial coefficient of some milk proteins. J. Coll. acteristics of monomeric and covalent polymeric Interface Sci. 39, 655–662. k-caseins. Biochem. 9, 2807–2813. Schmidt, D.G. and Van Markwijk, B.W. (1968). Further Tanford C. (1962). Contribution of hydrophobic interac- studies on the associating subunit of as1-casein. tions to the stability of the globular conformation of Biochim. Biophys. Acta 154, 613–614. proteins. J. Am. Chem. Soc. 84, 4240–4274. Senocq, D., Molle, D., Pochet, S., Leonil, J., Dupont, D. Tauzin, J., Miclo, L., Roth, S., Molle, D. and Gaillard, and Levieux, D. (2002). A new bovine b-casein genetic J.-L. (2003). Tryptic hydrolysis of bovine as2-casein: variant characterized by a Met93→Leu93 substitution identification and release kinetics of peptides. Int. in the sequence A2. Lait 82, 171–180. Dairy J. 13, 15–27. Smyth, E., Syme, C.D., Blanch, E.W., Hecht, L., Vasak, Thompson, M.P. and Farrell, H.M. Jr. (1974). Genetic M. and Barron, L.D. (2001). Solution structure of variants of the milk proteins, in, Lactation, Vol. III, native proteins with irregular folds from Raman B.L. Larson and V.R. Smith, eds., Academic, New optical activity. Biopolymers 58, 138–151. York. pp. 109–134. Snoeren, T.H.M., Van Markwijk, B. and Van Montfort, R. Thompson, M.P., Kalan, E.B. and Greenberg, R. (1967). (1980). Some physicochemical properties of bovine Properties of caseins modified by treatment with as2-casein. Biochim. Biophys. Acta 622, 266–276. carboxypeptidase A. J. Dairy Sci. 50,767–769. Stewart, A.F., Bonsing, J., Beattie, C.W., Shah, F., Willis, Thorn, D.C., Meehan, S., Sunde, M., Rekas, A., Gras, I.M. and Mackinlay, A.G. (1987). Complete nucle- S.L., MacPhee, C.E., Dobson, C.M., Wilson, M.R. and otide sequences of bovine as2- and b-casein cDNAs: Carver, J.A. (2005). Amyloid fibril formation by comparisons with related sequences in other species. bovine milk k-casein and its inhibition by the Mol. Biol. Evol. 4, 231–241.

160 T. Huppertz molecular chaperones as- and b-casein. Biochem 44, Vreeman, H.J., Both, P., Brinkhuis, J.A. and Van der Spek, C. 17027–17036. (1977). Purification and some physicochemical Thorn, D.C., Ecroyd, H., Sunde, M., Poon, S. and Carver, properties of bovine k-casein. Biochim. Biophys. Acta J.A. (2008). Amyloid fibril formation by bovine milk 491, 93–103. as2-casein occurs under physiological conditions yet is prevented by its natural counterpart, as1-casein. Vreeman, H.J., Brinkhuis, J.A. and Van der Spek, C. Biochem. 47, 3926–3936. (1981). Some association properties of bovine SH-k- Thurn, A., Burchard, W. and Niki, R. (1987). Structure of casein. Biophys. Chem. 14, 185–193. casein micelles. I. Small angle neutron scattering and light scattering from b-casein and k-casein. Colloid Vreeman, H.J., Visser, S., Slangen, C.J. and Van Riel, Polymer Sci. 265, 653–666. J.A.M. (1986). Characterization of bovine k-casein Toma, S.J. and Nakai, S. (1973). Calcium sensitivity and fraction and the kinetics of chymosin-induced molecular-weight of casein-as5. J. Dairy Sci. 56, macropeptide release from carbohydrate-free and 1559–1562. carbohydrate-containing fractions determined by Trieu-Cuot, P. and Gripon J.-C. (1981). Electrofocusing high-performance gel-permeation chromatography. and two-dimensional electrophoresis of bovine Biochem. J. 240, 87–97. caseins. J. Dairy Res. 48, 303–310. Visser, S., Slangen, C.J., Lagerwerf, F.M., Van Dongen, Whitney, R.M., Brunner, J.R., Ebner, K.E., Farrell, H.M. W.S. and Haverkamp, J. (1995). Identification of a Jr., Josephson, R.V., Morr, C.V. and Swaisgood, H.E. new variant of bovine beta-casein using reversed- (1976). Nomenclature of the proteins of cow’s milk: phase high-performance liquid-chromatography and fourth revision. J. Dairy Sci. 59, 795–815. mass-spectrometric analysis. J. Chrom. A 711, 141–150. Yamuachi, K., Takemoto, S. and Tsugo, T. (1967). Vreeman, H.J. (1979). The association of bovine SH-k- Calcium-binding property of dephosphorylated casein. casein at pH 7.0. J. Dairy Res. 46, 271–276. Agric. Biol. Chem. 31, 54–63. Yan, S.-C.B. and Wold, F. (1984). Neoglycoproteins: in vitro introduction of glycosyl units at glutamines in b-casein using transglutaminase. Biochem. 23, 3759–3765.

Higher Order Structures 5 of the Caseins: A Paradox? H.M. Farrell Jr, E.M. Brown, and E.L. Malin 5.1 Introduction and Historic data from our laboratory demonstrated a lack of Views of Casein Structure a-helix in the caseins, and since that was all that could be measured at the time, caseins were con- One of the fundamental theorems of modern sidered to be the model for random-coil proteins protein chemistry, the Anfinsen hypothesis, is (Farrell, 1988). However, as sequences became that the vast majority of protein architecture in available and circular dichroism (CD) was biological systems arises from the primary employed as a tool for protein analysis (Creamer sequences of the proteins (Anfinsen, 1973). While et al., 1981), the possibility of periodic structure the most recent discoveries of the action of chap- was considered. Swaisgood (1982) was perhaps eronins have indicated a role for these proteins in the first to suggest that the caseins were neither the kinetics of protein folding, the fundamental globular nor random-coil proteins and that they theorem still applies. A major corollary to this could be composed of rather distinct functional theorem also appears to be time tested: biological domains. The next important step on the road to function arises from protein or nucleic acid the understanding of casein structure may be the structure. concept of Holt and Sawyer (1993) who sug- gested that caseins are rheomorphic in nature. In From all of the concepts regarding casein this instance, the “formed under flow” hypothesis structure-function which have been set forth over suggests that casein structure is not fixed at all in the years, two fundamental functions of casein the absence of calcium. In its extreme, this can be envisioned: hypothesis may be considered as the “spaghetti 1. The effective transport of calcium plate” hypothesis, in that no regular structures 2. The self-associations which lead to the colloi- occur until aggregates are formed in response to calcium-phosphate binding. Supporting this dal state (Farrell et al., 2002a, 2006a) hypothesis was the observation of Paulsson and This review will concentrate on the latter func- Dejmek (1990) that pure caseins, when studied tion. Although casein has been studied for many by differential scanning calorimetry (DSC), years, the molecular structural basis for its func- showed flat endotherms on heating. The coopera- tion in self-association reactions in milk has been tive unfolding of native globular proteins always elusive. Historically, optical rotatory dispersion yields a rather characteristic Gaussian pattern when studied by this methodology. However, H.M. Farrell Jr (*) • E.M. Brown • E.L. Malin Paulsson and Dejmek (1990) suggested an alter- U.S.D.A., Eastern Regional Research Center, Wyndmoor, native view, i.e., caseins exhibit no peak because PA 19038, USA they contain heat-stable structures. e-mail: [email protected] P.L.H. McSweeney and P.F. Fox (eds.), Advanced Dairy Chemistry: Volume 1A: Proteins: Basic Aspects, 161 4th Edition, DOI 10.1007/978-1-4614-4714-6_5, © Springer Science+Business Media New York 2013

162 H.M. Farrell Jr et al. The concept that caseins have little or no fixed 1994a). At a later time, it was discovered that aS2- structure seemed quite appealing, but data which casein was structurally similar to the chloride we had collected in collaboration with the late channel (CLIC) proteins which have known crys- Heino Susi appeared to be at odds with this con- tal structures (Farrell et al., 2009). This discovery cept (Byler et al., 1988). We had applied Susi’s allowed us to produce a 3D structure for aS2- Raman methodologies, which he had developed casein by homologous modeling techniques for globular proteins, to purified caseins in the (Farrell et al., 2009), and it is also shown in absence of calcium. Inspection of the data Fig. 5.1. revealed that all of the caseins had intricate amide I profiles, similar to those obtained for globular 5.2 New Views of Protein Structure proteins, with a moderate distribution of various periodic secondary structures; FTIR analyses of Historically, proteins were thought to fold or the caseins are in agreement with these data. The unfold in a concerted fashion as shown in Eq. 5.1. patterns of all purified caseins and their mixtures The idea that a protein may unfold or fold through showed bands characteristic of reasonable a multistep process has in turn led to the “New amounts of b-turns and b-sheet and a modest View” of protein folding. Interestingly, much of amount of a-helix. It has, therefore, been difficult the early data which led to this theory were from to reconcile the concept of fixed structure giving the milk protein, a-lactalbumin (Xie et al., 1991; rise to function with either the random-coil or the Farrell et al., 2002b). rheomorphic hypotheses. Native Intermediate Unfolded (5.1) Starting about 1990, we began to conduct a series of three-dimensional (3D) molecular mod- According to the “New View,” during folding, eling experiments on caseins. In these studies, we a protein chain may “sample” a significant attempted to derive structures for caseins from amount of conformational space before settling the basic Anfinsen hypothesis. Because an infinite into a selective energy minimum. Indeed, several number of potential structures are available in false minima may lie quite close or even some- conformational space to proteins the size of the what remotely removed from the true global caseins (»200 residues), we attempted to arrive at minimum (Farrell et al., 2006b). Such an inter- working models by constraining the computer mediate area has been postulated to be the mol- experiments to predicted secondary structures ten globule state as shown in Fig. 5.2 for the derived from primary sequence data (Garnier theoretical energy landscape of a “minimally et al., 1978). We further constrained the global frustrated” heteropolymer, as folding is viewed structure by requiring that it conform to Raman from top to bottom. The parameter, Q, is the and FTIR limits. Thus, for example, the limit for global order parameter and represents a value of a-helix was no more than 10% and the limit on 1.0 for all interactions in the native state, and 0.0 extended structure 30%. Finally, the number of for the completely unfolded protein. Reflection turns was increased from algorithm predictions of the interactions between Q, E, and S results in to correlate the spectroscopic data with the rela- a three-dimensional funnel. In the “New View” tively high abundance of proline in the caseins. then the intermediate state of Eq. 5.1 is now This was done because proline, while a structure- defined as a multiplicity of states which includes breaking residue for helix and sheet, can be not only the molten globule as defined in Fig. 5.2, instrumental in the formation of turns in peptides but also a number of confomationally defined and proteins (Ananthanarayanan et al., 1984; states which reside above the molten globule Cohen et al., 1986). Using these principles in (MG) region (Uversky, 2002). These new states, conjunction with force field calculations, we working upward from the molten globule state arrived at refined energy-minimized working of Fig. 5.2, have been previously defined as models for k-, b-, and aS1-caseins which are shown in Fig. 5.1 (Kumosinski et al., 1993a, b,

5 Higher Order Structures of the Caseins: A Paradox? 163 Fig. 5.1 Three-dimensional molecular models of the potential with red being negative, light being neutral and/or hydrophobic, and blue positive. These models represent a caseins obtained from sequence-based secondary structural working view of casein structure and are subject to change as future experimentation progresses (Kumosinski et al., predictions and aligned with spectroscopic data. Top: left 1993a, b, 1994a; Farrell et al., 2009) aS1-casein, right aS2-casein; bottom: left b-casein and right k-casein. The models display a pseudo-charged surface pre-molten globule (PMG) and natively unfolded energy states (see Fig. 5.2). This results from (NU) based upon their physical and chemical three general properties of these proteins: properties (for a review, see Farrell et al., 2006b). 1. They contain high contents of proline and glu- A different classification by means of biological function has also been attempted (Tompa, 2002; tamine leading to segments of polyproline II Tompa and Kalmar, 2010) and under the general conformation (PPII), which produce extended heading of intrinsically unstructured protein structures with high hydration (IUP), five functional definitions of these new 2. They have a high net charge which prevents states have been suggested. The question now close approach of segments of the protein arises, how can this “New View” be applied to molecules casein structure? 3. They usually have a low hydrophobicity which does not allow for hydrophobic collapse into a The major hallmark of the IUP or NU proteins highly folded structure is that while they contain significant amounts of With regard to the classification of the caseins defined secondary structures, they do not fold in this scheme, they do have high amounts of pro- and remain trapped in conformational energy line and glutamine and have been shown to con- states above that of the MG state depicted in tain significant amounts of PPII (Farrell et al., Fig. 5.2, whereas native molten globules (which 2001; Syme et al., 2002), they have high levels of represent a subset of intrinsically disordered pro- hydration (Kumosinski et al., 1988), and they teins) are trapped in the MG-like conformational have high net negative charges, resulting in part

164 H.M. Farrell Jr et al. Fig. 5.2 A schematic representation for the energy land- pact but still well within the MG levels and all are scape for a minimally frustrated heteropolymer during clearly not ordered proteins. protein folding. Enat is the minimum potential energy for the native state. From Onuchic et al. (2000); reprinted For the NU, PMG, and MG states, the experi- courtesy of Academic Press, San Diego, CA mentally determined volume of a monomeric form of the protein is always much greater than from their phosphoserine residues (Farrell et al., the volume of a compact globular protein of the 2004). However, in contrast to many other IUP or same chain length. Uversky (2002) has developed NU proteins, they are highly hydrophobic, and it a series of equations to classify proteins by this is this latter property which undoubtedly leads to method. Using these equations, k-casein (with a their aforementioned propensity for self-associa- reducing agent added) appears to be the most tion. Using these three criteria alone, it is some- unfolded while the other three appear to be in the what difficult to assign the caseins to their place PMG state (Table 5.1). An additional volume- within this “New View” as shown in Fig. 5.2. based calculation can be made for b- and However, Farrell et al. (2006c) have examined in k-caseins. Because their polymeric states have detail each casein and its relationship to a variety been well described in terms of molecular weight of criteria developed by Uversky (2002) for and size, it is possible to calculate the volume of classifications within these states. The overall the b- and k-casein monomers within their results of these analyses are given in Table 5.1. respective polymers and compare these using the Uversky (2002) equations. This analysis places The hydropathy plots and PONDR® analyses both the b- and k-casein monomers solidly within are sequence-based binary predictors of disorder the MG state. in proteins. These programs were used by Farrell et al. (2006c) to analyze 11 aS1-CNs, 9 aS2-CNs, The 3D models as shown in Fig. 5.1 can be 13 b-CNs, and 20 k-CNs with known sequences. analyzed in terms of their size and shape as well, Using these methods, the aS1- and aS2-caseins are and the radius of gyration can be calculated from set apart from the b- and k-caseins, primarily the monomer molecular volume. Applying the because of their high net negative charges. The size/volume calculations to the 3D models yields b- and k-caseins are predicted to be more com- a prediction for their states within Fig. 5.2 as shown in Table 5.1. Here, aS1-casein is the most extended and predicted to be in the NU state, while aS2-casein is more compact in the PMG state. However, the 3D models predict the MG state for the b- and k-caseins, but these predic- tions most likely again represent the characteris- tics of the monomers within their respective polymers rather than an independent monomer state. Indeed, this was a prediction made during the original modeling of b-casein (Kumosinski et al., 1993a). Overall, then it may be concluded that aS1- casein may occur in the NU state, aS2-casein is more compact and falls into the PMG state, while both b- and k-caseins as monomers within their polymers reside in the MG state. It is important to remember that none of these states is random coils and as noted above contain significant amounts of defined secondary structures, but do not, for various reasons, fold into compact globu- lar proteins. Moreover, all of these predictions

5 Higher Order Structures of the Caseins: A Paradox? 165 Table 5.1 Comparison of classification “New View” schemes for aS1-, aS2-, b-, and k-caseins. Classification methodology Casein Hydropathy plotsa Monomer size Monomer size 3D model sized by GPCb from polymer datac aS1- NU PMG MG NU aS2- NU PMG PMG b- MG PMG MG k- MG NU MG MG NU natively folded, PMG pre-molten globule, GPC gel permeation chromatography aFarrell et al. (2006c) for hydropathy and PONDR® analyses bPepper and Farrell (1982) cDynamic light scattering and sedimentation equilibrium (Farrell et al., 2006c) dKumosinski et al. (1994a, 1993a, b) for aS1-, b-, and k-CN; Farrell et al. (2009) for aS2-CN are for the caseins in the absence of added cal- self-association (Schmidt and Payens, 1976); this cium or phosphate. The caseins, then, are solidly region encompassing residues 140–190 is in the realm of these new protein classes and are extremely hydrophobic in nature and is found a part of the rapidly emerging study area termed between 10 and 7 o’clock in the 3D model of aS1- “unfoldomics” by Dunker et al. (2008). For the casein (Fig. 5.1). A second site for aggregation caseins, these open and more hydrophobic states may occur in the region 14–25, as the deletion of lead to their intrinsic property of self-association. this region in the aS1-casein A genetic variant It could be said, then, that the native states of the leads to altered forms of self-association aS1-, aS2-, b-, and k-caseins are the states in which (Kumosinski and Farrell, 1991, 1994); this region they exist when fully either immersed in the also contains a chymosin-sensitive bond, cleav- casein micelle or self-associated with themselves age of which leads to changes during cheese mat- or other caseins where the hydrophobic regions uration (Tunick et al., 1997). In contrast to the of the caseins are intermingled. In the next sec- molecule as a whole, the N-terminal region (1–26 tion we will examine several examples of how and 4 o’clock in the 3D model) is positively predicted secondary structures for the casein charged. Finally, Horne (1998) has suggested that monomers lead to their unique polymeric states. these two segments represent surface-active regions of aS1-casein. 5.3 Molecular Modeling of aS1- 5.3.1 Hydrophobic Dimers and Casein: Interactions and Oligomers of aS1-Casein Support from Experimental Data Schmidt and Payens (1976) and Schmidt (1982) summarized light scattering studies on variants To exemplify how persistent secondary structures of aS1-casein under a variety of environmental in the caseins may lead to aggregation and self- conditions from which a stoichiometry of the aS1- association, we will first focus on the primary casein self-association is obtained at selected calcium-binding protein of milk: aS1-casein ionic strengths and at 20°C. From these results, it (Farrell et al., 2004). The molecular model, pre- can be concluded that aS1-casein undergoes a sented for aS1-casein in Fig. 5.1, shows two concentration-dependent, reversible, hydropho- potential sites for self-association. The first is the bically controlled association from monomer to C-terminal portion (residues 130–199) which may dimer and then to tetramer, hexamer, and octamer, be responsible for pronounced salt-accelerated

166 H.M. Farrell Jr et al. Table 5.2 Weight-average molecular weights of selected caseins and mixtures by analytical ultracentrifugation at 37oCa Casein or mixture Wt.-Avg. molecular weight Average polymeric size Rotor speed aS1-Casein 56,000 Dimer 12,000 b-Casein 1,250,000 52-mer 3,000 RCM k-casein 3,040,000 160-mer 3,000 1.5 aS1-:1 RCM k- 316,000 15-mer 3,000 4 aS1-:1 RCM k- 92,400 Tetramer 6,000 4b-:1 RCM k- 1,010,000 43-mer 3,000 1b-:1aS1- 213,000 Nonamer 3,000 aAll data were obtained at 37°C, pH 6.75 in 25 mM disodium piperazine-N,N¢-bis(2-ethane sulfonic acid) with 80 mM KCl to mimic milk salt conditions in the mammary gland in the absence of calcium; the rotor speeds were appropriate to the weight-average MW as previously described. RCM reduced and carboxymethylated. Modified from Farrell et al. (2006a) and even higher levels if the ionic strength is rigidity to this site. A dimer can be formed easily increased to 0.6 M (Thurn et al., 1987). These if two of these sheets are docked in an antiparallel extended polymers are thought to be rodlike and fashion (Fig. 5.3), and this structure, following may involve head-to-head (1–25) and tail-to-tail energy minimization, displays a stabilizing (140–190) condensations of the molecule shown energy of −520 kcal mol−1 residue−1 (Kumosinski in Fig. 5.1. Studies of the whole aS1-casein mol- et al., 1994a). ecule and its derivative, produced by carboxy- peptidase activity (f1–197), which lacks Trp199, 5.3.2 Hydrophobic Dimers appeared to confirm these two areas as the pri- and Oligomers of aS1-Casein mary interaction sites (Alaimo et al., 1999a). Fragment f136–196 However, recent work (Malin et al., 2005) has shown that at 37°C and at physiological salt con- All of the above data appear to suggest that for centrations, the higher order polymers are broken aS1-casein, preformed structural elements sur- down and that aS1-casein is essentially a dimer rounding Pro147 participate in the formation of under physiological conditions (Table 5.2). It has dimers. A better understanding of these interac- also been shown (Malin et al., 2005) that this tions was brought about by investigating the dimerization involves residues 136–158 of each hydrophobic portion of the protein, namely, the monomer (10 o’clock in the aS1-model of aS1-casein (f136–196) peptide. Isolation of this Fig. 5.1). These experiments demonstrate that for sizable portion of the C-terminal half of the pro- aS1-casein, preformed structural elements sur- tein was achieved by cyanogen bromide cleavage rounding Pro147 participate in the formation of and purification as described by Alaimo et al. dimers at conditions simulating physiological (1999a, b). The global secondary structure of aS1- temperature, pH, and ionic strength, and neither casein (f136–196), as estimated by far-UV CD, is the 1–25 region nor tryptophan 199 is involved. given in Table 5.3. At 27°C and pH 6.75 in a low ionic strength buffer, the results for aS1-casein To mimic this hydrophobic dimerization, (f136–196) are consistent with the putative 3D Kumosinski et al. (1994a) constructed an energy- structure in which all six proline residues are minimized dimer from the large-stranded b-sheets involved in turns. Here the peptide may be which occur at residues 136–158. The side chains expected to be a dimer in equilibrium with its are predominately hydrophobic, and limited hydrogen bonding of the sheet structure lends

5 Higher Order Structures of the Caseins: A Paradox? 167 Fig. 5.3 Double ribbon structure of aS1-casein dimer action, W199 C-terminal tryptophan, R1 N-terminal argi- constructed by docking two large hydrophobic sheets in nine, P147 proline 147, P168 proline 168 (Kumosinski et al., 1994a) an antiparallel fashion interaction; sites are noted (1, 2 indicate molecules 1 and 2). Hb hydrophobic site of inter- Table 5.3 Comparison of secondary structural estimates for the peptide aS1-casein (f136–196) by three methods Method Temperature (°C) b-Sheet (%) Turns (%) Unspecified (%) a-Helix (%) FTIRa 25 49 ± 3b 22 ± 1 23 ± 1 5±2 CDc 10 64 ± 2 28 ± 1 5±1 3±1 27 58 ± 1 31 ± 1 8±1 3±1 50 49 ± 2 28 ± 1 18 ± 1 5±1 3D model 70 49 ± 2 29 ± 1 18 ± 1 3±1 in vacuo 57 28d 15 0 aAverage of three determinations in PIPES-KCl aqueous at pH 6.75 ionic strength 124 mM (Alaimo et al., 1999b) bFor FTIR—includes 310-helix, bent strand, and extended b-sheet (Alaimo et al., 1999b) cFor CD-average fits, one determination (six accumulations at each temperature) ionic strength 4 mM dFor molecular models—includes I-P-N-P-I loop at residues 182–186 component monomer and tyrosines 144, 146, and this analysis were in good agreement with the CD 154 served as reporter groups for this interaction. data (Table 5.3). However, as the peptide is Global secondary structure was also estimated heated, there is an increase in the calculated from the FTIR spectra of a 1.5% aqueous solu- amount of unspecified (random-coil) structure tion of aS1-casein (f136–196) and at higher con- and loss of b-sheet (Table 5.3). Malin et al. (2005) centrations and ionic strengths where higher showed that in computer-generated molecular order aggregates occur at 25°C. The results of dynamics simulations of this peptide, the

168 H.M. Farrell Jr et al. sheet-turn-sheet motif directed by Pro147 was sta- Fig. 5.4 Schematic representation of monomer and poly- ble in an aqueous environment while others were mer models for aS1- or b-caseins; (a) monomer, (b) not. For the peptide (f136–196) we propose that tetramer, and (c) planar representation of a rosette-shaped the turn region about Pro147 is the initial site of spherical polymer. Waugh (1970), reprinted courtesy of hydrophobic self-association (dimerization) at Academic Press, San Diego, CA physiological conditions. Moreover, interactions centering on Pro168, involving Trp164 as a reporter at 0.1 M ionic strength, pH 7, and 20°C to about group, indicate that this region is involved in the 52 at 0.110 M ionic strength, pH 6.7, and 37°C higher order polymers found for the peptide (Table 5.2). Perhaps the best representation of (f136–196) at higher ionic strengths and at 25°C this process was given by Waugh (1970) origi- (Alaimo et al., 1999a, b; Malin et al., 2005). By nally for aS1-casein, but by extension to b-casein. extension, the regions seen at the top in Fig. 5.3 As shown in Fig. 5.4, the phosphopeptide is rep- would represent sites for the formation of the resented as a ring and the remainder of the mol- higher order polymers found for the whole aS1- ecule as a torpedo-like structure. Overall, the casein molecule at high ionic strengths (Thurn geometry of the b-casein polymers as given in et al., 1987). Thus, the sheet-turn-sheet second- Fig. 5.4 is still in accord with data elucidated over ary structure motif is most likely the basis for the the past 40 years. However, as noted above, the self-association behavior of this casein and as intrinsic volume of an individual b-casein within will be seen below for its vital interactions with the polymer is significantly lower than the mono- other caseins. mer, so Waugh’s monomer still represents the monomer within the polymer in a molten glob- 5.4 Molecular Modeling of b-Casein: ule-like state exemplified by the 3D model for Interactions and Support from b-casein (Fig. 5.1). b-Casein has one site which Experimental Data is particularly sensitive to chymosin (189– 190/192–193) found at about 7 o’clock in the b-Casein is the most hydrophobic casein and has b-model of Fig. 5.1. As shown by Qi et al. (2005) the largest regions of high hydrophobicity (55–90 loss of the C-terminal peptide (193–209) by chy- and 130–209) with a very acidic N-terminal mosin cleavage substantially reduces but does region of 24 amino acids. The acidic N-terminal not eliminate the association-dissociation equi- region is at the right top of the model in Fig. 5.1. librium. At pH 6.75, 37°C, and ionic strength There are two regions where plasmin readily 50 mM, the weight-average MW drops from cleaves the protein (at bonds 28–29 and 105– 495,000 to 90,222 Da (Table 5.4). The latter rep- 106/107–108, respectively; 3 and 4 o’clock in the resents a tetramer as the monomer molecular b-casein model of Fig. 5.1). Cleavage at these weight for f1–192 of b-casein is 22160 Da (Qi sites by plasmin yields the fragments previously et al., 2005). Thus, the peptide f193–209 is a req- known as g1- and g2-caseins, respectively (Farrell uisite for the normal self-association of b-casein. et al., 2004). The self-association of this protein is (deter- gent) micelle-like (Qi et al., 2005), and both ionic strength and temperature increase the quantity of polymer present (i.e., increased association con- stant) and the degree of association (n): nβ − CN ↔ β − CNn (5.2) The number of monomer proteins in these nearly spherical polymers ranges from about 15

5 Higher Order Structures of the Caseins: A Paradox? 169 Table 5.4 Weight-average molecular weights, apparent association constants of selected caseins and mixtures by analytical ultracentrifugation at 37°C Analytical ultracentrifuge aS1-Casein MWb Protomer MWc kad (L g−1) na Rotor speed (rpm) b-Caseinf 56,000 56,000 1.09 × 10-1 4 12,000 495,000 24,000 6,000 4.38 × 1014 29 b-Caseinf (f 1–192) 90,200 22,000 4.68 × 10−3 18 10,000 RCM k-casein 4,140,000 505,000 1.35 × 1011 16 3,000 4 aS1-:1 RCM k- 92,400 69,800 nde nde 6,000 4b-:1 RCM k- 1,010,000 711,000 5.82 × 10−6 4 3,000 1 b-:1aS1- 213,000 100,000 7.36 × 10−3 4 6,000 aAll data except where noted were obtained at 37°C, pH 6.75 in 25 mM disodium piperazine-N,N¢-bis(2-ethane sulfonic acid) with 80 mM KCl to mimic milk salt conditions in the mammary gland in the absence of calcium; the rotor speeds were appropriate to the weight-average MW as previously described. RCM reduced and carboxymethylated. Modified from Farrell et al. (2006a) bWeight-average MW in Da, three determinations ± 5% cThe protomer is the lowest molecular weight found at the top of the cell at equilibrium; it represents the kinetically active species which participates in association reactions as described in Equation (5.2) dka and na as described in Equation (5.2) eThese quantities were not determined fThese b-casein experiments were conducted at a lower salt concentration (I = 50 mM) to accentuate the differences between the intact molecule and f1–192 (Qi et al., 2005) The question now as stated above is, does this logical function. A notable example is tailspike segment contain any persistent periodic second- protein produced by phage 22 which infects ary structures which contribute to this self- Salmonella spp. The tailspike protein through its association? C-terminal region binds to the surface of the bac- terial cells. As seen in its crystal structure (pro- CD and FTIR spectral analysis indicates that tein data bank # 1TYW; Steinbacher et al., 1994), at 6°C, b-casein has a relatively low level of this region contains a section of the sheet-turn- a-helix (~15%), an intermediate level of turn- sheet motif (residues 632–666) as shown in like structure (~29%), and a similar level of Fig. 5.5 (top). Interestingly, residues 636–655 of extended or b-sheet (~30%) in dilute, low ionic the tailspike protein have a 64% homology with strength, neutral solutions. However, there are f193–209 of b-casein. Using the crystal structure significant increases in (CD) ellipticity at 220 nm coordinates of this segment of the tailspike pro- when the solution temperature is increased from tein, it is possible to construct a homologous 6 to 37°C. b-Casein is predicted to contain a molecular model for f193–209 of b-casein. To do significant amount (30%) of polyproline II struc- this the peptides were aligned as described by Qi ture, but these structures have been shown to et al. (2005); four residues found in the center of occur at about 20% in b-casein by Raman spec- a b-sheet in tailspike protein but not in b-casein troscopy, FTIR and CD (Farrell et al., 2001; were deleted as were seven residues at the Syme et al., 2002). Thus, it is possible that C-terminal end. The selection was then submitted increases in both b-strand and polyproline II to the ExPASy web site for the production of a structure are responsible for this temperature- homologous model, as described by Farrell et al. dependent structural change. (2009) for aS2-casein. A representation of the homologous structure is shown in Fig. 5.5 (bot- It has been suggested for a growing number of tom) and it retains the sheet-turn-sheet motif of proteins (Qi et al., 2005) that reactions at their C-terminal regions are important for their bio-

170 H.M. Farrell Jr et al. Fig. 5.5 On the top is the C-terminal portion of the tailspike protein (phage22) residues 657–688 with its C-terminal at the top and the arrowheads represent the direction of travel of the chain from N- to C-terminal with a stable sheet-turn-sheet motif (protein data bank # 1TYW, Steinbacher et al., 1994). On the bottom is the homologous model for the C-terminal peptide of b-casein f193–209 as generated from the tailspike peptide. This may also represent a model for casecidins 15 and 17 (Birkemo et al., 2009) the original model, but with weaker hydrogen Measuring the volume of the extended tail as bonding. This model may be of use in that the 588 Å3, about 37 tails could be placed into this sequence corresponds to casecidins 15 and 17 cavity area. It should be remembered that chy- (Birkemo et al., 2009). These two peptides have mosin cleavage of this sheet-turn-sheet area antimicrobial activity toward E. coli and repre- dramatically reduces b-casein self-association. sent f192–209 and f192–207 of b-casein. It is So as noted above, these residues may well interesting to speculate that the antimicrobial coalesce to anchor the b-casein monomers to their action of the peptides is mechanistically and spherical polymer. This provides another example structurally related to the action of the phage22 of how a sheet-turn-sheet motif in a casein may protein on Salmonella. lead to its characteristic mode of self-association. Next, the coordinates obtained for the b-casein 5.5 Molecular Modeling of k-Casein: peptide f192–209 were transferred to the whole Interactions and Support from b-casein model of Fig. 5.1 and following the regi- Experimental Data men of Farrell et al. (2009), this leads to the extension in space of the original C-terminal sec- k-Casein, which constitutes 10–12% of whole tion of the molecule as now given for b-casein in casein, plays a crucial role in stabilizing the Fig. 5.6. Here the extended C-terminal region casein micelles in milk and, after enzymatic would be free to interact with other b-caseins and cleavage, destabilizing the colloidal casein sys- to self-associate into the large polymers shown in tem (Farrell et al., 2004). The enzymatic cleav- Fig. 5.4 for Waugh’s model. In the sleek repre- age that brings about this transformation is sentation in Fig. 5.4, no such tails are shown and important for the nutrition of the suckling young a void is depicted at the center of the cross sec- and for the production of many cheese varieties. tion. When Kumosinski et al. (1993a) constructed This is achieved by the molecule having two dis- polymers of b-casein from the 3D model of tinctly different domains. As seen at the bottom Fig. 5.1, similar voids occurred. The void in Fig. 5.4 can be calculated to be about 22,000 Å3.

5 Higher Order Structures of the Caseins: A Paradox? 171 Fig. 5.6 The original model of b-casein as shown in cleavage site (residues 28–29), the blue region indicates Fig. 5.1 is rotated 90° toward the viewer and has had its the second plasmin cleavage site (residues 105–107), the psi, phi, and omega angles of residues 192–209 changed to white colored portion represents the bulk of the hydropho- conform to those of the peptide of Fig. 5.5. The red color bic body, and the yellow represents the C-terminal section represents the N-terminal phosphopeptide region, the pink (f193–209) modeled after the tailspike peptide and a region of lower net charge containing the first plasmin extended in space for potential polymerization reactions portion of the k-model in Fig. 5.1, the N-terminal site; most recent work has clearly demonstrated domain (residues 1–95) carries a net positive that a b-sheet structure is required for the chy- charge, is very hydrophobic, and interacts mosin active site (Palmer et al., 2010). strongly with the other casein molecules. The Accordingly, the k-casein wire model shown in C-terminal domain (residues 113–169) carries a Fig. 5.7 has a seven-residue segment of b-sheet in net negative charge and contains a preponderance this region, and Phe105 and Met106 are clearly vis- of polar residues (top portion of the k-model in ible. In the altered model the proline residues that Fig. 5.1). These two domains are joined by a pep- precede and follow this segment of b-sheet put a tide (residues 96–112, 10 o’clock in the k-casein significant strain on the region, and hydrolysis model of Fig. 5.1) that carries a net positive results in increased disorder (entropy) in silico. charge, is predicted to be a b-strand, and is gener- ally well conserved in most species (Palmer et al., As purified from milk, k-casein occurs as a 2010). This region contains a motif that is readily series of intermolecular disulfide-bonded aggre- recognized by chymosin and is rapidly and gates (Groves et al., 1992). Farrell et al. (1996) specifically cleaved to give the two domains produced a 3D model of a k-casein octamer com- noted above. The peptide f106–169 is called posed of two disulfide-bonded tetramers to caseinomacropeptide (CMP) or glycomacropep- explain the overall properties of k-casein as tide (GMP) because about half of the k-casein purified from milk. This octamer is shown in molecules are posttranslationally glycosylated Fig. 5.8 (left) and displayed with the same and phosphorylated (Farrell et al., 2004). pseudocharges of Fig. 5.1; it is deep red (nega- Additionally, all of these sites are surface orien- tive). It is generally accepted that the “hairy” tated in the k-casein model (Fig. 5.1). GMP (Horne, 1998, 2006) provides steric hin- drance and prevents casein micelles from coalesc- The original model of k-casein (Fig. 5.1) con- ing in normal milk. When the GMP is removed tained a segment of alpha helix for the chymosin by chymosin, it is thought to bring about

172 H.M. Farrell Jr et al. Fig. 5.7 The original monomer model of k-casein as blocked. The molecular model thus allows for the forma- shown in Fig. 5.1 is modified here to show a segment of tion of dimers (center) and tetramers (right). The further b-sheet for the chymosin cleavage site at 10 o’clock (left) propagation of the polymeric structures from the reduced and the phenylalanine-methionine residues are in the center monomer is responsible for the formation of the high of the b-sheet. In this model the sulfhydryl groups are molecular weight amyloid structures (Farrell et al., 2003b) Fig. 5.8 Molecular model for a k-casein octamer com- k-casein octamer and is colored red representing a high posed of two disulfide-bonded tetramers. This model can be negative electrostatic potential. When the negatively charged thought of as a representation of a poly k-casein area on the GMP is removed in silico, there is a complete charge rever- surface of a micelle. The model on the left represents intact sal (right) where blue represents positive charge aggregation of the casein micelles in milk because Circular dichroism (CD) and FTIR spectral of loss of this hindrance. However, as seen in the analysis indicates that for k-casein there is a rela- 3D model for an octamer of k-casein, when the tively low level of a-helix (15%), an intermediate negatively charged GMP is removed in silico, level of turn-like structure (~25%), and a higher there is a complete charge reversal in Fig. 5.8 level of extended or b-sheet (~30%). The above (right) where blue represents positive charge. molecular models for k-casein are in accord with Thus, the creation of surface positive charge may these measurements. In addition, all of these enhance the aggregation of casein micelles fol- structures appear to be thermostable, as shown in lowing chymosin action through charge–charge Table 5.5 (Farrell et al., 2003b). interactions with the abundance of surface nega- tive charges on other caseins. As noted above, k-casein exists as a series of intermolecular disulfide-bonded aggregates, and

5 Higher Order Structures of the Caseins: A Paradox? 173 Table 5.5 Comparison of secondary structural estimates for k-casein by three methods Method Temperature (°C) b-sheet (%) Turns (%) Unspecified (%) a-helix (%) FTIRa 25 35 ± 3b 25 ± 2 23 ± 4 17 ± 2 CDc 25 40 ± 1 26 ± 1 24 ± 1 9±1 10 36 ± 2 28 ± 1 24 ± 1 12 ± 1 3D model 50 36 ± 2 27 ± 1 24 ± 1 14 ± 1 70 36 ± 2 24 ± 1 20 ± 1 19 ± 1 in vacuo 30 32 30 10 aAverage of three determinations in PIPES-KCl aqueous at pH 6.75 (Farrell et al., 1996) bFor FTIR—includes 310-helix, bent strand, and extended b-sheet (Farrell et al., 1996) cFor CD-average fits, one determination (six accumulations at each temperature) (Farrell et al.,2003b) these disulfide bonds may be formed after the ever, instead of self-association this polymeriza- casein micelles have been assembled in the epi- tion is more properly classified as an aggregation. thelial cells (Farrell et al., 2006a). These disulfide Another example in k-casein is the b-sheet struc- aggregates range from dimers to octamers and ture which is surrounded by proline turns and above (Groves et al., 1992). However, about 5% which contains the chymosin-sensitive bond. The of the cysteine residues in total are not disulfide- accurate hydrolysis of this bond initiates the first bonded and may react with other proteins or par- step in the digestive process and again represents ticipate in oxidation reduction reactions, e.g., how the defined persistent secondary structure of k-casein aggregation with heat (Groves et al., a casein leads to ultimate biological function. 1998) or formation of complexes with b-lacto- globulin in heated products (Haque et al., 1987; 5.6 Molecular Modeling of aS2- Douglas et al., 1981). As a reduced isolated pro- Casein: Interactions and Support tein, at 20°C k-casein self-associates in a similar from Experimental Data manner to b-casein to a 600,000 Da molecular weight polymer. However, the reduced car- aS2-Casein is the least hydrophobic and the most boxymethylated (RCM) protein at 37°C forms highly and variably phosphorylated of the caseins. amyloid bodies which were first discovered in There are three phosphopeptide regions (5–18, this casein (Farrell et al., 2003b). These amy- 49–68, and 126–145) in the casein sequence (left loids have a soluble molecular weight of over side of the model of aS2-casein in Fig. 5.1) and a 3 × 106 Da (Table 5.2). The b-sheet structures at large central hydrophobic region (90–120) with 6 o’clock of k-casein monomer in Figs. 5.1 and very little charge (region of aS2- protruding from 5.7 are thought to promote amyloid formation. the center of the model to 6 o’clock in Fig. 5.1). In both CD and FTIR studies there is little change There is a second large hydrophobic region (160– in the overall content of secondary structure dur- 207), but it has a number of positively charged ing fibril formation. This type of sheet-turn-sheet residues, which represent the highest positively interaction was predicted for the model as shown charged area for any casein (Fig. 5.1). In milk, in Fig. 5.7 (Farrell et al., 2003b). In subsequent the majority (~90%) of the protein occurs with an studies, X-ray diffraction data were found to be internal disulfide bond between cysteine residues in accord with the model presented in Fig. 5.7 36 and 40 forming a small loop in the structure. (Thorn et al., 2005). In addition a small proportion of this protein exists as a disulfide-bonded dimer as well as Thus for k-casein, like b- and aS1-caseins, a polymers with k-casein (Farrell et al., 2009). defined sheet-turn-sheet secondary structure leads to protein–protein interactions. Here, how-

174 H.M. Farrell Jr et al. Circular dichroism (CD) and FTIR spectral (Waugh, 1970; Farrell et al., 2006a). A summary analysis indicates that for aS2-casein there is a of changes in weight-average MW for selected relatively high level of a-helix (30–40%), an casein–casein interactions is given in Table 5.2. intermediate level of turn-like structure (~20%), and a similar level of extended or b-sheet 5.7.1 Mixed Associations (~20%). Prediction methods indicate that likely of the Caseins positions for helix formation are near both the acidic N-terminal and the basic C-terminal More in-depth analyses of the above interactions regions (Hoagland et al., 2001). This protein is were attempted and the results are given in readily hydrolyzed by plasmin and trypsin at a Table 5.4. With regard to Table 5.4, it is impor- number of sites primarily in the afore noted tant to understand the meaning of the term “pro- C-terminal region, so that at neutral pH these tomer.” In the independent self-association positively charged residues are primarily at the reactions of aS1-casein at low ionic strength and surface and could actively participate in the for b-caseins (Alaimo et al., 1999a; Farrell et al., binding of inorganic phosphate. In addition, on 2001; Qi et al., 2005), the protomer is the lowest proteolysis, this area gives rise to a number of molecular weight species found near the top of biologically active peptides with defined struc- the equilibrium cell and usually represents the tures (Farrell et al., 2009). species which then goes on to yield polymers. For b-casein (Eq. 5.2) and its chymosin fragment, Association studies at 20°C show that mono- the protomers are the monomers which is also meric aS2-casein behaves in a very similar way to true for aS1-casein at low ionic strength and at aS1-casein except that there is a maximum asso- 25°C (Alaimo et al., 1999a). However, as seen in ciation at salt concentrations above 0.2 M Table 5.4, aS1-casein at 37°C and physiological (Snoeren et al., 1980). However, at 37°C this pro- ionic strength is essentially a dimer and there is tein has been found to form elongated amyloid little self-association beyond this (ka = 10−1), so structures similar to those formed by k-casein that the protomer approximates the weight-aver- (Thorn et al., 2008). These interactions are age MW (Malin et al., 2005); no monomer is thought to occur through associations of the cen- apparent. For RCM k-casein at 37°C, the pro- tral hydrophobic core (residues 90–120) noted tomer leading to amyloids is 505,000 Da; again above. This section of aS2-casein has a strong this is the smallest species present in the experi- homology with the portion of k-casein responsi- ment. In the case of the b-casein-RCM k-casein ble for amyloid production. Again, for this pro- mixtures, the protomer and weight-average MW tein, amyloid formation occurs through appear to be large and similar, but considerably sheet-turn-sheet structural motifs. This represents smaller than k-casein alone; also the association another example of aggregations driven by constants are substantially reduced from >1011 to defined structural states. 10−6. It should be noted that bovine RCM k- and human b-caseins form similar complexes and go 5.7 Molecular Modeling of Casein– on to form casein micelles, in vitro, with added Casein: Interactions and Support calcium and phosphate (Sood et al., 2006). from Experimental Data For 1:1 aS1–b-casein mixtures, the high molec- In contrast to the number of excellent studies on ular weight polymers of b-casein are reduced in the interactions of individual casein species in size; the association constant for b-casein is solution, there have been only a few studies of reduced from 1014 to 10−3 at 6,000 rpm in mixed associations involving primarily binary Table 5.4. The equations predict a protomer of mixtures. Historically, such studies have shown 100,000 Da, with a weight-average MW of that, in the absence of polyvalent cations and at 213,000 Da. These latter values are considerably 37°C, aS1- and k-caseins associate most strongly smaller than those found at 3,000 rpm; studies at

5 Higher Order Structures of the Caseins: A Paradox? 175 9,000 and 12,000 rpm gave still smaller numbers. to act as a single species, while a protomer Such behavior indicates a case of pressure-depen- molecular weight of 69,800 Da (possibly repre- dent aggregation which can be a hallmark of senting two aS1-+ one k-) was found, fitting to the highly hydrophobically driven interactions progressive association equations yielded poor (Hummer et al., 1998; Farrell et al., 2002c); results with n and k not determined accurately extrapolation to atmospheric pressure yields a (Table 5.4); however, a small amount of high value of 247,000 for the weight-average particle molecular weight amyloid could skew the data. of the aS1–b-casein mixture, possibly a decamer. As was the case for aS1-b-casein interaction, Here aS1-casein acts as a molecular detergent to increasing the speed of the rotor greatly reduced reduce greatly the size of the b-casein polymers the weight-average MW indicating pressure-de- reducing the weight-average MW from over one pendent hydrophobic interactions. Extrapolation million to the extrapolated value of 247,000 Da to atmospheric pressure yielded a weight-average (see Table 5.2 for comparable ionic strengths). MW of 117,000 Da, which approximates a 4:1 Using the latter molecular weight and a Stokes ratio. So, using amyloid repression and the above radius from Pepper and Farrell (1982), the molec- data as a guideline, overall the 4:1 ratio appears ular volume reflects a mixture of the NU volume best but perhaps not unique. Interestingly, once of aS1-casein and the MG volume of b-casein. b-casein is added to these mixtures and the This represents a departure from the model of concentrations are increased to those found in Waugh (1970) who believed that in the mixed milk, more uniform complexes occur, as a result polymers of aS1- and b-casein shown in Fig. 5.4, of up to 6–8 association reactions, and sodium the individual aS1- and b-caseins were nearly caseinate-like particles are found (Slattery and identical in shape and interchangeable without an Evard, 1973; Schmidt and Payens, 1976). effect on particle size. In the “New View” of protein structure noted Early studies on the weight ratios for the above, aS1-casein can be considered to be a formation of aS1-:k- particles gave estimates natively unfolded (NU) “assembler” in that it is from 1:1 to 10:1; later values centered on 4:1 able to break down aggregates of b- and k-casein (Waugh, 1970; Kumosinski et al., 1994b). These and lead to successful transit through the mam- earlier experiments were conducted with whole mary secretory system prior to the addition of k-casein which we now know represents a mix- calcium in the Golgi apparatus. Indeed, its ture of disulfide-bonded aggregates. The clearest absence in the caprine homozygous null aS1- example of this problem was demonstrated by allele (Chanat et al., 1999) leads to the accumu- Pepper (1972) using gel chromatography. In the lation of large protein particles in the latter experiments, aS1-casein clearly reduced endoplasmic reticulum resulting in reduced the size of the k-casein aggregates, but the inter- secretion. Although, as noted above b-casein action peaks demonstrated large polydispersity. can repress k-casein amyloid formation, b-casein The weight-average MW of RCM k- is reduced knockout genes in rats do not appreciably alter by aS1-casein at a 1.5:1 ratio to 316,000 Da secretion. b-Casein appears to be phosphory- (Table 5.2). Thorn et al. (2005) showed that lated and to enter into the Golgi secretory amyloid formation was not repressed at this ratio granules at a later time than aS1-casein. This and required a weight ratio of nearly 4:1 for makes a strong case for aS1-casein as the pri- nearly complete repression. The inability of aS1- mary force in casein micelle secretion (Chanat casein at low weight ratios to depress fibril for- et al., 1999; Le Parc et al., 2010). Recently, it mation argues against these 1:1 complexes as a has been shown that the amyloid bodies formed driving force for micelle formation as previously by aS2-casein can be disassembled by aS1-casein suggested (Waugh, 1970). For an aS1-:k- ratio of but not by b-casein (Thorn et al., 2008). It is 4:1, where amyloid production is repressed also interesting that in vitro caseins can aid in (Thorn et al., 2005), the molecular weight is enzyme folding acting in a limited way as chap- reduced (Table 5.2). The particle does not appear erones (Farrell et al., 2006a).

176 H.M. Farrell Jr et al. 5.7.2 Sodium Caseinate 5.7.3 3D Models for Sodium Caseinate (Particles in Solution) Sodium caseinate (whole casein in the absence of divalent cations) is an excellent ingredient and Interpretations of all of the physical chemical finds many applications in food processing. This data collected on sodium caseinate lead to a poin- product can have somewhat varying properties tillistic or smeared view of the particles. To depending on its method of preparation (Douglas develop further a molecular basis for structure- et al., 1981). Heating sodium caseinate at 140°C function relationships of the sodium caseinate brings about a conversion to larger aggregates, system, an energy-minimized 3D structure of a due only partly to the free sulfhydryl groups of putative casein submicelle (sodium caseinate k-casein (Chu et al., 1995). Properties, such as particle) was constructed consisting of one viscosity, of casein solutions are concentration- k-casein, four aS1-casein, and four b-casein mol- dependent and a 15–20% casein solution is very ecules (Kumosinski et al., 1994b). The models viscous because of the associations of the mole- for the individual caseins were from the 3D struc- cules with one another. A number of emulsion and tures depicted in Fig. 5.1 and the primary interac- foaming studies have been done using either tions used were those given above: two various commercial whole proteins, e.g., total milk hydrophobically driven dimers of aS1-casein protein (TMP) or milk protein concentrate (MPC) interact with the b-sheet “legs” of k-casein, thus or whole casein materials. However, it must be preventing amyloid formation (a 4:1 ratio); sec- stressed that at the core of the predictable func- ondarily, two dimers of b-casein, held together tionality of the caseins lies the strong, rather selec- by their C-terminal peptides, interact at a later tive, protein–protein interactions discussed above. time with the k- and aS1- complexes. The con- struction of the model was described in detail in Laboratory preparations of sodium caseinate the previous version of this chapter (Farrell et al., through mixed associations of the individual 2003a) and will not be repeated here. This mod- caseins discussed above bring about the forma- eling yielded two energy-minimized 3D struc- tion of a rather stable polymer with an average tures for the putative casein submicelle and the diameter of 18 nm and a MW of about 280,000 Da more favored one is shown in Fig. 5.9. This at room temperature. As viewed by electron refined k-casein-based submicellar structure was microscopy, these particles appear to be rather tested by generating theoretical small-angle uniform in shape and size (Kumosinski et al., X-ray scattering (SAXS) curves and comparing 1996). However, at 37°C, these polymers can dis- them with experimental data (Kumosinski et al., sociate into monomers and smaller polymers at 1994c). Excellent agreement between experi- concentrations below 1% (Pepper and Farrell, mental and theoretical curves was found. The 1982). An apparent dissociation constant can be global shapes and sizes of these putative submi- calculated to be 6.2 mg/mL from Pepper and cellar structures were further tested by compari- Farrell (1982). Thus, for the casein found in the son of their computer-generated van der Waals endoplasmic reticulum, prior to any addition of dot structures with actual transmission electron calcium and phosphate (Chanat et al., 1999), in micrographs (EM) of reduced and carboxymeth- Golgi apparatus during phosphorylation (Farrell ylated whole casein (Kumosinski et al., 1996). et al., 2006a) and in milk (about 27 mg/mL), These comparisons, as shown in Fig. 5.10, dem- strong casein-casein associations are favored, as onstrate quite good agreement between the mod- the concentrations are threefold greater than the els and the EM. In actuality then the precise dissociation constant. In all studies of lactating molecular coordinates of the 3D models can yield mammary tissue, sodium casein-like particles back the pointillistic views from which we (putative submicelles) are abundant in electron started; the work has gone full cycle. micrographs of the mammary secretory process (Farrell et al., 2006a; Chanat et al., 1999; Very little data are available concerning the Schmidt, 1982). role of aS2-casein in sodium caseinate structure

5 Higher Order Structures of the Caseins: A Paradox? 177 Fig. 5.9 Energy-minimized casein asymmetric submi- bones without side chains; k-casein B in green, aS1-casein celle structure, i.e., one k-casein, two aS1-casein dimers, in white, b-casein A2 in red, and bound water in yellow and two b-casein symmetric dimers. Space-filled back- (Kumosinski et al., 1994b) and formation. However, the aS2-casein molecule ture than previously thought. More research is has several similarities with k-casein that are necessary to prove this speculation but almost all data now point to k-casein as occurring as worth noting. First, residues 77–118 have a strong disulfide-bonded polymers and so some particles in the rather uniform fields seen for sodium homology with residues 42–84 of k-casein and caseinate may be k-casein-rich while others may contain no k-casein. Such a theoretical particle is both proteins form amyloid tangles on heating; shown in Fig. 5.11. That sodium caseinate might contain particles of different composition was the k-casein amyloid centers on residues 26–85, suggested by Slattery and Evard (1973) based upon mixed associations of the caseins in sedi- while amyloid fragments of aS2-casein found in mentation velocity studies at elevated ionic mammary gland begin with residue 81 (Niewold strengths. Interestingly, the above model offers a reply to the critical question (Horne, 2006) as to et al., 1999). Secondly, in their monomer states how to account for submicelles of varying composition. (Fig. 5.1) both have central hydrophobic cores If all of the caseins contain persistent struc- apparently available for hydrophobic interactions tures, such as those seen above, then can they also be open and flexible, as their physical data which contain the latter noted regions. suggest? Additionally, aS1-casein inhibits amyloid forma- 5.7.4 FTIR Studies of Sodium tion at a ratio of 4:1 for k-casein, and a ratio of Caseinate (Casein Submicelles) 2:1 for aS2-casein. Hoagland et al. (2001), prior Curley et al. (1998) studied the effects of Ca2+ to the model for aS2-casein, speculated that on the and Na+ or K+ on the FTIR spectra of sodium basis of structural similarity, perhaps aS2-casein could compete with k-casein as a primary inter- actant in the sodium caseinate particle. The new aS2-casein model can do precisely that as demon- strated in Fig. 5.11 for the asymmetric submicelle model; note the area of positive charge on the surface of this model. This model has not been rigorously tested in the same fashion as the k-casein-based submicelle model (Farrell et al., 2003a) and serves only to suggest that aS2-casein could play a more important role in casein struc-

178 H.M. Farrell Jr et al. Fig. 5.10 Comparison of matched shape and dimensions Waals dot surface model, (c) enlargement of image- of the asymmetric submicelle 3D model with photograph- enhanced micrograph, (d) van der Waals of (a) rotated 90° ically enlarged image enhanced representations of submi- about y-axis, and (e) enlargement of image-enhanced celles (TEM bar = 10 nm, molecular model = 5 nm). (a) representation of the submicelle particle (Kumosinski Backbone structure for asymmetric model, (b) van der et al., 1996) caseinate. They concluded that electrostatic bind- led to apparent decreases in large loop or helical ing of Ca2+ to casein resulted in a redistribution of structures at 37°C with concomitant increases in the protein components of the infrared spectra. the percentage of structures having greater bond Addition of Ca2+ in salt solutions of K+ and Na+ energy, such as turns and extended helical

5 Higher Order Structures of the Caseins: A Paradox? 179 Fig. 5.11 Energy-minimized casein asymmetric submi- brown for two aS1-casein dimers and purple for two celle structure; here the one k-casein is replaced by one b-casein dimers. The aS2-casein is in red except for the aS2-casein. The model shows backbone atoms only in basic C-terminal which is in blue structures. The effects of Ca2+ on global protein All of the above data point to the loop-helix- structure with micelle formation have been sug- loop motif as being flexible and subject to con- gested by several physical chemical studies. In formational change on ligand binding and particular, studies using SAXS (Kumosinski conversion to aggregate structures. This is in con- et al., 1988) predicted a swelling of the outer trast to the apparent rigidity of the hydrophobic shell of casein submicelles as they are incorpo- peptides studied in Sections 5.3, 5.4, and 5.5. rated into reformed micelles. This swelling repre- How then can we reconcile these divergent views sented a 30% increase in hydration (with a of different portions of the same molecule and concomitant decrease in the electron density of in the overall framework structure of sodium proteins). Holt and Sawyer (1993) suggested that caseinate aggregates (submicelles) and micelles? a recurrent motif in ruminant caseins is helix- loop-helix in which the loop region is typically 5.8 The Tensegrity Hypothesis phosphorylated. Studies of the molecular dynam- and Resolution ics of the aS1-casein phosphopeptide also sug- gested that the swelling of these loop structures The architectural world has often borrowed struc- accommodated the increased hydration that tural forms from biological shapes. Now it accompanied Ca2+ binding (Kumosinski and appears that an architectural contrivance may Farrell, 1994). Thus, presumably the Ca2+ bind- help us to understand biological forms. Tensegrity ing and the incorporation of Ca caseinates into structures were originally constructed by Snelson micelles could deform a-helical elements and and popularized by Buckminster Fuller as the extend loop elements because they are spatially geodesic dome (Ingber, 1998). In architectural adjacent. Hence, the changes in the 1,655 ± 5 cm−1 parlance, the forces of compression and tension region of the FTIR spectra upon Ca2+ addition balance, so that rigid struts bear compression could be due to loop-helix alterations with move- while more flexible elements stretch. Tensegrity ment of the resonances to higher wave numbers structures offer the maximum space (openness) and with the higher bond energies needed to but- for a minimal amount of building material. A tress the swelling and extension of the polypep- simple toy shown in Fig. 5.12 best illustrates the tide chains.

180 H.M. Farrell Jr et al. Fig. 5.12 Typical tensegrity structure illustrating the principles of rigidity and flexibility, published by permission of Manhattan Toy, Minneapolis, MN, 55401 concepts of tensegrity structures which are of turn-sheet motif also appears quite heat stable in interest to us. Donald Ingber has pioneered the k-casein as well (Table 5.5) and in isolation it application of tensegrity concepts to biological gives rise to amyloid structures. Note also that structures ranging from the cytoskeleton of cells Graham et al. (1984) and Farrell et al. (2001) to microtubules, to viral envelopes, and to pollen demonstrated a good deal of heat stability for grains (Ingber, 1998). We propose, here, that a sheet and turns in b-casein by similar CD studies. type of tensegrity structure may account for the In addition, the tailspike peptide is another exam- overall properties of sodium caseinate aggregates ple of sheet-turn-sheet interactions translated into (submicelles) and casein micelles. b-casein. In the tensegrity structural analogy we propose that these heat-stable sheet-turn-sheet In the putative submicelle framework shown motifs of aS1-, b-, and k-caseins interact as shown in Fig. 5.9, the protein–protein interactions which in Sects. 5.3, 5.4, and 5.5 and represent the solid lead to this structure occur primarily via sheet- struts or pre-compressed modules (Fig. 5.12). turn-sheet interactions. In computer experiments, These structures then provide the framework for a good deal of structural integrity was gained casein–casein interactions. For b-casein, these through these interactions (Kumosinski et al., interactions are more hydrophobic and in the pre- 1994c). As we have seen, these structures, which vious 3D models (Kumosinski et al., 1994b, c), follow the sheet-turn-sheet motif, occur in aS1- the b-casein was fitted into the framework of aS1- casein monomers and dimers and are quite ther- /k-casein. This also accommodates the ability of mally stable in solution (Alaimo et al., 1999b) b-casein to dissociate from aggregates and and under molecular dynamic simulations (Malin micelles at 5°C (Downey and Murphy, 1970). et al., 2005). For contrast, consider the high degree of flexibility which occurs for the helix- On the other hand, the helix-loop-helix por- loop-helix motif relative to the sheet-turn-sheet tions of the aS1- and b-caseins represent more motif, once the aS1-dimer has been formed. flexible structures and are analogous to the k-Casein also contains this same rigid motif cables of tensegrity structures. Here, changes in (Fig. 5.7) between residues 14 and 64; this sheet- structure with the degree of phosphorylation, or

5 Higher Order Structures of the Caseins: A Paradox? 181 Fig. 5.13 Backbone asymmetric structure of a casein red, b-casein in magenta, oxygen from droplet waters in submicelle with water molecules from droplet algorithm, cyan (Kumosinski et al., 1994c) i.e., 2,823 water molecules: k-casein in blue, aS1-casein in with calcium or proton binding, may yield con- remain open. In contrast the highly flexible formational changes (Huq et al., 1995; Curley regions such as the loop-helix-loop regions define et al., 1998) or increased conformational their ability to bind and effectively transport cal- flexibility (Kumosinski and Farrell, 1994). cium as casein micelles. Halfmann and Lindquist (2010) have pioneered the concept that binding to and modification of Finally, a remembrance of an old casein struc- flexible proteins (prions) may lead to significant ture idiom: all good models hold water (M.P. changes in biological function. The tensegrity Thompson). The openness of the tensegrity struc- analogy easily accounts for the swelling of sub- tures and the 3D models can readily accommo- micellar structures through flexible elements as date extremely high water content in the interior they are incorporated into reformed micelles space. The water content of micelles and submi- (Kumosinski et al., 1988). Once again, the prin- celles varies from 1 to 8 g H2O g−1 protein ciple of maximum space with minimum build- (Kumosinski et al., 1988). In a series of studies ing material is a tensegrity concept. on casein–water interactions, Mora-Gutierrez et al. (1997) used 17O NMR to probe and enumer- For casein, then, persistent secondary struc- ate sources of bound, trapped, and preferentially tures, such as the sheet-turn-sheet motifs, define absorbed water molecules. These cavities and their self-association reactions, but because the voids have been correlated with the 3D model of tension and flexibility compromise, no hydro- Kumosinski et al. (1994c) as shown, partially phobic compression occurs and the proteins hydrated, in Fig. 5.13. Thus, the tensegrity

182 H.M. Farrell Jr et al. hypothesis can account for the physical and Dunker A.K., Oldfield, C.J., Meng, J., Romero, P., Yang, chemical properties of caseins, while explaining J.Y., Chen, J.W., Vacic, V., Obradovic, Z. and Uversky, how the molecules can be both rigid and flexible V.N. (2008). The unfoldomics decade: an update on at the same time and exhibit highly hydrated intrinsically disordered proteins. BMC Genomics 9 backbones. Suppl 2, S1. References Farrell, H.M., Jr. (1988). Physical equilibria: proteins, in, Fundamentals of Dairy Chemistry, 3rd edn., N. Wong, Alaimo, M.H., Wickham, E.D. and Farrell, H.M., Jr. ed., Van Nostrand, Reinhold, New York. pp. (1999a). Effect of self-association of aS1-casein and its 461–510. cleavage fragments on aromatic circular dichroic spec- tra: comparison with predicted models. Biochim. Farrell, H.M., Jr., Kumosinski, T.F., Cooke, P.H., King, Biophys. Acta. 1431, 395–409. G., Hoagland, P.D., Wickham, E.D., Dower, H.J. and Groves, M.L. (1996). Particle sizes of purified k-ca- Alaimo, M.H., Farrell, H.M., Jr. and Germann, M.W. sein: metal effect and correspondence with predicted (1999b). Conformational analysis of the hydrophobic 3D models. J. Protein Chem. 15, 435–445. peptide aS1-casein (f 136–196). Biochim. Biophys. Acta. 1431, 410–420. Farrell, H.M., Jr., Wickham, E.D., Unruh, J.J., Qi, P.X. and Hoagland, P.D. (2001). Secondary structural anal- Ananthanarayanan, V.S., Brahmachari, S.K. and ysis of bovine caseins: temperature dependence of Pattabiraman, N. (1984). Proline-containing b-turns in b-casein structure as analyzed by circular dichroism peptides and proteins: analysis of structural data on and FTIR spectroscopy and correlation with micelliza- globular proteins. Arch. Biochem. Biophys. 232, tion. Food Hydrocoll. 15, 341–354. 482–495. Farrell, H.M., Jr., Kumosinski, T.F., Malin, E.L. and Anfinsen, C. (1973). Principles that govern the folding of Brown, E.M. (2002a). The caseins of milk as calcium protein chains. Science. 181, 223–230. binding proteins, in, Methods in Molecular Biology, Vol. 172, Calcium Binding Protein Protocols, Vol. 1, Birkemo, G.A., O’Sullivan, O., Ross, R.P. and Hill, C. H.J. Vogel, ed., Humana Press, Totowa, NJ. pp. (2009). Antimicrobial activity of two peptides caseci- 97–140. din 15 and 17, found naturally in bovine colostrum. J. Appl. Microbiol. 106, 233–240. Farrell, H.M., Jr., Qi, P.X., Brown, E.M., Cooke, P.H., Tunick, M.H., Wickham, E.D. and Unruh J.J. (2002b). Byler, D.M., Farrell, H.M., Jr. and Susi, H. (1988). Raman Molten globule structures in milk proteins: implica- spectroscopic study of casein structure. J. Dairy Sci. tions for potential new structure-function relation- 71, 2622–2639. ships. J. Dairy Sci. 89, 459–471. Chanat, E., Martin, P. and Olivier-Bousquet M. (1999). Farrell, H.M., Jr., Qi, P.X., Wickham, E.D., and Unruh, aS1-Casein is required for the efficient transport of b- J.J. (2002c). Secondary structural studies of bovine and k-casein from the endoplasmic reticulum to the caseins: structure and temperature dependence of Golgi apparatus of mammary epithelial cells. J. Cell b-casein phosphopeptide (1–25) as analyzed by circu- Sci. 112, 3399–3412. lar dichroism, FTIR spectroscopy, and analytical ultra- centrifugation. J. Protein Chem. 21, 307–321. Chu, B., Zhou, Z., Wu, G. and Farrell, H.M., Jr. (1995). Laser light scattering of model casein solutions: effects Farrell, H.M., Jr., Brown, E.M., Hoagland, P.D. and Malin, of high temperature. J. Coll. Interf. Sci. 170, E.L. (2003a). Higher order structures of caseins: a 102–112. paradox? in, Advanced Dairy Chemistry, 3rd edn., Vol. 1, Proteins Part A, P.F. Fox and P.L.H. McSweeney, Cohen, F.E., Abarbanel, R.M., Kuntz, I.D. and Fletterick, eds., Kluwer Academic, Plenum Publishers, New R.J. (1986). Turn prediction in proteins using a pat- York, NY. pp. 203–231. tern-matching approach. Biochemistry. 25, 266–275. Farrell, H.M., Jr., Cooke, P.H., Wickham, E.D., Piotrowski, Creamer, L.K., Richardson, T. and Parry, D.A.D. (1981). E.D. and Hoagland, P.D. (2003b). Environmental Secondary structure of bovine aS1- and b-caseins in influences on bovine k-casein: reduction and conver- solution. Arch. Biochem. Biophys. 211, 689–696. sion to fibrillar (amyloid) structures. J. Protein Chem. 22, 259–273. Curley, D.M., Kumosinski, T.F., Unruh, J.J. and Farrell, H.M., Jr. (1998). Changes in the secondary structure Farrell, H.M., Jr., Jimenez-Flores, R., Bleck, G.T., Brown, of bovine casein by FTIR: effects of calcium and tem- E.M., Butler, J.E., Creamer, L.K., Hicks, C.L., Hollar, perature. J. Dairy Sci. 81, 3154–3162. C.M., Ng-Kwai-Hang, K.F. and Swaisgood H.E. (2004). Nomenclature of the proteins of cows’ milk Douglas, F.W., J., Greenberg, R., Farrell, H.M., Jr. and sixth revision. J. Dairy Sci. 87, 1641–1674. Edmondson, L.F. (1981). Effects of ultra-high-temper- ature pasteurization on milk proteins. J. Agr. Food Farrell, H.M., Jr., Malin, E.L., Brown, E.M. and Qi, P.X. Chem. 29, 11–15. (2006a). Casein micelle structure: what can be learned from milk synthesis and structural biology? Curr. Op. Downey, W.K. and Murphy, R.F. (1970). The temperature Coll. & Interf. Sci. 11, 135–147. dependent dissociation of b-casein from bovine casein micelles and complexes. J. Dairy Res. 37, 361–372. Farrell, H.M., Jr., Qi, P.X. and Uversky V.N. (2006b). New views of protein structure: implications for poten- tial new protein structure-function relationships, in,