B. Ribadeau-Dumas, R. Grappin To cite this version

HAL Id: hal-00929173https://hal.archives-ouvertes.fr/hal-00929173

Submitted on 1 Jan 1989

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Milk protein analysisB. Ribadeau-Dumas, R. Grappin

To cite this version:B. Ribadeau-Dumas, R. Grappin. Milk protein analysis. Le Lait, INRA Editions, 1989, 69 (5),pp.357-416. �hal-00929173�

https://hal.archives-ouvertes.fr/hal-00929173

https://hal.archives-ouvertes.fr

Laft(1989)69,357-416© Elsevier/INRA

357

Review

Milk protein analysis

B. Ribadeau-Dumas 1 and R. Grappin 2

1 INRA, station de recherches laitières, C.R.J., 78350 Jouy-en-Josas,2 INRA, station de recherches de technologie et analyses laitières, B.P. 94, 39800 Poligny, France

(received 7 October 1988, accepted 27 April 1989)

Summary - After a short description of bovine milk proteins, the various methods of current orpotential use for detecting and determining them in dairy products are reviewed. This includes, first,the determination of total protein from nitrogen analysis, dye-binding capacity, infra-red spectrometryand amino acid analysis. The methods that allow determination of sorne milk protein fractions ofinterest (whole casein, whey proteins, ~-Iactoglobulin) are then given. They include the Aschaffen-burg-Rowland procedure, dye-binding and infra-red methodologies. A description of the variousmethods (electrophoresis, column chromatography, immunochemical or enzymatic tests), that canbe used for detecting and individually quantitating the various caseins and whey proteins ispresented. Finally, sorne applications of various analytical procedures to the analysis of differentclasses of dairy products are given.

analytical techniques - milk products - proteins - peptides - amine acids

Résumé - Analyse des protéines du lait - Une revue. Après une brève description desprotéines du lait de vache, les différentes méthodes utilisées ou pouvant être utilisées pour mettreen évidence et doser ces protéines dans les produits laitiers sont passées en revue. Dans unepremière partie sont décrites les techniques permettant le dosage des protéines totales par dosaged'azote, fixation de colorants, spectrométrie infrarouge, puis celles qui permettent le dosage decertaines fractions protéiques (caséine entière, protéines du lactosérum, IHactoglobuline) :méthode d'Aschaffenburg-Rowland et méthodes qui utilisent la fixation de colorants ou laspectrométrie infrarouge. Sont examinées ensuite les diverses techniques permettant la détectionet le dosage individuel des principales protéines du lait, par électrophorèse, chromatographie surcolonne, par des tests immunochimiques ou enzymatiques.

Pour terminer, sont donnés des exemples d'applications de plusieurs méthodes à l'analyse deproduits laitiers divers.

techniques analytiques - produits laitiers - protéines - peptides - acides aminés

358 B. Ribadeau-Dumas and R. Grappin

INTRODUCTION

This review concerns ail types ofanalyses dealing with proteins in milk,dairy and a few non-dairy products. Onlythe methods that are in current use, orthose which have been recently publishedand might become of general use, will beconsidered. For ail of them, the reader willbe supplied with basic principles, sche-matie descriptions, references and criticalevaluations.

Compared to other food products, milkis a fairly simple fluid. It has beenthoroughly studied from the beginning ofthe 19 th century. Its composition and themain characteristics of its variousconstituents are now weil known. Inparticular, the amine acid sequence of its7 main protein components has beenelucidated. There is currently no otherfood product whose proteins are so weilcharacterized. This makes their analysisstraightforward in raw milk. However, assoon as technological treatments havebeen applied, any quantitative measure-ment, except nitrogen determination,becomes far more difficult. In particular,protein denaturation, which is not a one-step phenomenon, leads, for a givenprotein, to products which may differaccording to the treatment, often with anultimate transformation into insolubleaggregates. Furthermore, a number ofchemical reactions may occur during theprocessing of milk, dairy products andnon-dairy products which lead to covalentmodifications of proteins. In a number offood products, milk proteins have beenintentionally fragmented into peptides andamine acids by proteinases andpeptidases. Of course, it is not possible todetermine from which protein free amineacids originated. Theoretically, the originof any peptide with more than 5 amineacid residues, provided it can be isolated,

can usually be established if it is derivedfrom any milk protein. However, even therough characterization of a milk proteinhydrolyzate is a difficult and long taskwhich can only at the moment· beperformed in a few research laboratories.

As knowledge of the main charac-teristics of the various milk proteins isessential for those who wish to obtainreliable analytical data concerning them,we will first describe them, as weil as theirfate, in the various dairy products. Then,the procedures allowing their deter-mination as a whole and individually inthese products will be reviewed, as weilas those used for detecting anddetermining them in non-dairy products.Finally, we will describe the techniquesavailable for assessing their quality, indairy products. However, the methodsused to assess milk protein digestibility invitro or in vivo will not be described as theauthors are not specialists in this area.

Characteristics of milk proteins

The white appearance of raw milk is dueto Iight scattered by two types of particlesin suspension in lactoserum (whey), fatglobules and casein micelles.

Fat globules are large (diameter 2 to12 urn, average ca. 4 um), sphericaltriglyceride particles, surrounded by a unitmembrane composed mainly ofphospho-lipids and proteins such as xanthineoxidase and butyrophilin. High pressurehomogenization, as used for liquid milkmanufacture or infra-red (IR) determi-nation of the main milk components,disrupts these globules into much smallerparticles (average diameter : 1 um). Skimmilk proteins replace the destroyednatural membrane on these globules(Keenan et al., 1983).

Milk protein analysis 359

The white appearance of skim milkresults from light scattering by muchsmaller particles (diameter from 20 to 600nm; average ca. 100 nm), i.e. the caseinmicelles. These consist mainly of water(4 g/g whole casein), proteins ("wholecasein") and minerais (60 mg/g wholecasein), mainly phosphate and calcium(Schmidt, 1982). Milk contains ca. 33 gliproteins (27 9 whole casein, 6 9 wheyproteins). In addition, some non-proteinnitrogenous (NPN) substances (ca. 1.5g/l) occur in milk : urea, creatinine,ammonia, peptides, nucleotides, vitaminsetc.) (Table 1).

Lowering the pH of milk (ca. 6.7 forfresh milk) to 4.6· precipitates wholecasein, leaving, in solution, the other milkproteins, the NPN and the minerais whichwere associated with casein micelles.Treatment of milk with rennet (a calfabomasum extract containing two milkclottinq proteinases, chymosin and

Table 1.Average protein composition of cow milk.Composition protéique moyenne du lait de vache.

pepsin) is the first step in cheesemanufacture; it leads to milk clotting. Theclot (curd), after draining of whey, isconstituted of almost intact whole caseinto which the minerais of the caseinmicelles are still bound. One of the fourprote in components of whole casein, lC-

casein (ca. 12 % whole casein) has beensplit into two parts by rennet. The largerfragment (para-x-caseln, 2/3 of themolecule) remains in the clot, while thesmaller one (casein macropeptide, CMP)is present in the whey, together with NPNand the whey proteins. Thus, acid andrennet wheys differ only in that the lattercontains CMP and has a lower mineraicontent (Dalgleish, 1982).

Thecaseins

Casein micelles contain only 4 proteinspecies, uS1-' uS2-' ~- and x-caseins,

Protein Amount (gl/)

ClsrCaseinCls2-CaseinI3-Caseinx-Caseiny-CaseinsProteose-peptonesœ-LactalbuminI3-LactoglobulinSerum albuminImmunoglobulinsLactoferrinTransferrinMFGMMilk

10.02.69.33.30.80.81.23.20.40.80.10.10.4

33.0

From:Walstra,P. & Jenness,R. (1984) Oairy Chemistry and Physics, NewYork,Wiley-Interscience.Composition typical of milks from Lowland breeds. MFGM. proteins of milk fat globule membranes.O'après Walstra P. & Jenness R. Dairy Chemistry and Physics, New York, Wiley-Interscience, 1984.Composition typique du lait de troupeaux Lowland. MFGM, protéines de la membrane des globules gras.


bound together by amorphous calciumphosphate (the so-called colloïdalphosphate). A small proportion of ~- andCls2-Cnsoccurs, even in fresh milk, asfragments arising from partial digestion byan endogenous milk proteinase, plasmin.Some of these fragments are found inacid casein and in rennet curd. This is thecase for y1-, 12- and y3-Cns, derived fromB-On. Others, some of the so-calledproteose-peptones, also derived from~-Cn, and the fragments of Cls2-Cn,arefound in the whey (Swaisgood, 1982).

A small proportion of caseins are not inthe rnicellar state ("soluble casein"). Theproportion of soluble casein increaseswhen the temperature is lowered, and canreach up to 20% of whole casein whenmilk is kept at 4 oCfor a day or more. Thisphenomenon is reversible : heating milkreduces the proportion of soluble casein.Furthermore, addition of CaCI2 to freshmilk drives almost ail soluble casein intomicelles. Subsequently, high speedcentrifugation (ca. 90 000 g, 1 h) allowscomplete sedimentation of casein micel-les. ~-Cn represents a large proportion ofthe soluble casein (Reimerdes, 1982).

The 4 caseins display commonfeatures that are quite unusual. They arehydrophobie, phosphorylated proteins,always occurring as large, polydisperseaggregates in aqueous solution atambient temperature and neutral pH.Each displays a loose, highly hydratedtertiary structure. They are highly sus-ceptible to ail proteinases and exopep-tidases. Severe heat treatments (sterili-zation) do not affect them significantly.Antibodies against native caseins alsorecognize these proteins after heattreatments.

Caseins, proteose-peptones and CMPare negatively charged molecules abovepH 7, while para-x-On is positivelycharged. For this reason, electrophoresis

and ion exchange fractionation of caseinsare usually performed around pH 8.5. Theisoelectric pHs of caseinslie between 4.5and 5.5. Those of para-x-On and y-Cnsare higher, while those of proteose-peptones and CMP are lower (Dalgleish,1982; Swaisgood, 1982).

Since caseins associate with eachother in solution, mainly through hydro-phobie interactions, separation of theisolated molecules by electropheresis orion exchange chromatography requiresthe addition of an agent, usually urea,able to disrupt these interactions.Furthermore, as Cls2- and x-Ons eachcontain 2 cysteine residues, completedissociation necessitates the addition of areducing agent such as 2-mercapto-ethanol to c1eave any S-S bridge(Swaisgood, 1982).

The primary structures of the 4 bovinecaseins are known (Swaisgood, 1982). Aliare single chain proteins. ClSl-Cnhas 199amino acid residues, 8 phosphate groupsIinked to serine residues (a small fractionof ClS1-Cn,the so-called Clso-Cn,has 9phosphate groups). ~-Cn, 209 residueslong, has 5 phosphate groups. Together,these 2 caseins, whose proportions inmilk are approximatively similar, constituteabout 70% of whole casein. Cls2- andx-Ons each represent ca. 13 % of wholecasein. The former, 207 residues long,occurs in milk as 4 molecular species atleast, which differ only in their number ofphosphate groups (10 to 13). The 4species were once called Cls2-'Cls3-'Cls4-and Cls6-Cns,ClS5-Cnbeing a dimer Cls3-Cls4(Eigel et al., 1984).

x-casein, 169 residues long, is cleavedby chymosin and pepsin betweenresidues 105 and 106. Fragment 1-105 ispara-x-caseln, a highly hydrophobie,positively charged molecule at neutral pH.The fragment 105 -169 is CMP, which ishighly hydrophilic, carries one (or two)


phosphate groups and various numbers ofserine- and threonine-linked oligosaccha-ride side-chains of various lengths, corn-posed of N-acetylgalactosamine, galac-tose and N-acetylneuraminic acid(NANA). Thus, J(-Cn consists of apopulation of closely related moleculeswhich can be partially resolved into morethan 7 components by electrophoresis orion exchange chromatography, since theircharges at alkaline pH differ according totheir NANA and phosphate content(Swaisgood, 1982).

The solubility of caseins displays somecharacteristic features. lt is weil knownand understood that any protein has asolubility minimum at its isoelectric pH.The caseins are almost completelyinsoluble at pH 4.6, e.g., commercial "acidcasein" is insoluble. However, it can bebrought into solution by bringing adispersion of casein in water to a pH of 7or above by adding an alkali until it isdissolved. Commercial sodium andcalcium caseinates are prepared in thisway by using sodium and calciumhydroxides, respectively. However, Caions (neutralization of acid caseinate withcalcium hydroxide or addition of CaCI2 tosodium caseinate at constant, neutral pH)induce the formation of casein micelles.Among the 4 caseins, 3 of them, aS1' as2and p, are very insoluble in the presenceof millimolar concentrations of Ca2+, whilethis ion does not affect the solubility ofJ(-Cn(this is also the case for p-Cn at lowtemperatures). When x-On is added toneutral solutions of aS1-' as2- or p-Cns, ora mixture of them, it stabilizes thesecaseins against calcium precipitation.Lowering the pH of such mixtures, as weilas that of milk, progressively removescalcium, and calcium-free caseinsprecipitate at their isoelectric pH. Thestabilization of caseins by J(-Cn towardscalcium precipitation is abolished whenrennet releases CMP.

Common genetic variants of caseinsare as1 Band C, as2 A, pA1, A2 and C, lC

A and B. Monomeric caseins have asimilar Mr, ranging fram ca. 20 000 to25 000 (Swaisgood, 1982).

Whey proteins

Unlike the caseins, the whey proteins arec1assic globular proteins with a tighttertiary structure, occurring in milk asmonomers or oligomers. Two of them aredominant, p-lactoglobulin (P-Lg) anda-lactalbumin (a-La) which represent ca.50 and 12% of total whey proteins,respectively. Several other prateins occurin milk : serum albumin (BSA), immuno-globulins (Igs), lactoferrin (LF), andenzymes such as lactoperoxidase (LP),xanthine oxidase, etc. (Jenness, 1982;Kitchen, 1985). Except for B8A andxanthine oxidase, ail milk proteinsmentioned above have been prepared atleast on a pilot scale. More details onsome biologically active whey proteins(LF, Igs and enzymes) will be given in alater section. Determination of the activityof several of them is especially useful, inparticular, for assessing the severity ofheat treatments.

p-Lg occurs in milk as a dimer (Mr36 000) of two identical subunits. Theprimary and tertiary structures of themonomer are known (Papiz et et., 1986).According to the conditions of pH, ionicstrength etc., monomeric, dimeric, tetra-meric or octameric p-Lg can be found.This pratein has a free SH group which isinvolved in interactions with otherproteins, such as J(-and as2-Cns, throughthe formation of disulfide bridges onheating. In its native state it possessestwo such bridges. Irreversible denatu-ration of p-Lg in milk may occur duringdrastic heat treatments. Two common

362 B. Ribadeau-DumasandR.Grappin

genetic variants of I3-Lg(A and B) occurwith similar frequencies; therefore, both ofthem are found in bulk milk. They can beeasily separated by electrophoresis orcolumn ion exchange at alkaline pH. Theirisoelectric pHs are between 5.3 and 5.5.I3-Lg is very resistant to proteolyticenzymes (Swaisgood, 1982; Papiz et el.,1986).

a-La in milk exists in the monomericstate (Mr 14000). It is a metallo protein towhich a Ca ion is strongly bound, and itcontains 2 disulfide bridges. Its primaryand tertiary structures are known (Stuartet al., 1986). Around pH 3.5, theconformation of the protein changes andthe Ca ion is released. The "acid form", A,has a slightly more open structure thanthose of the native form (N) and theapo-œ-La (Ca-free, neutral pH).Conversion, from one of these forms tothe other, is reversible, while denaturation(heat or Gu. HCI treatments) results in anirreversibly expanded structure. It isirreversibly denatured in milk by severeheat treatments. Ils isoelectric pH is ca.4.3 (Swaisgood, 1982; Pfeil, 1987).

Both whey proteins lose their nativeantigenic determinants when they areirreversibly denatured. Thus, theirreaction with antibodies raised against thenative proteins can be used to assess theextent of their denaturation.

Whey proteins as a whole (4-7 g/l) arethe most interesting components of whey,a by-product of cheese and caseinmanufacture, which mainly containslactose (44 to 53 g/l). Whey also containsminerais (5 to 8 9 ash/l, according to themethod of manufacture) and ca. 0.3 g/lfat, mainly phospholipids. One of theoldest methods used to recover wheyproteins involves heat denaturation atacid pH. The resulting so-called"Iactalbumin" is insoluble and lacks thefunctional properties of the nativeproteins.

Another commercially available proteinproduct derived from milk is "copreci-pitate". Its manufacture involves heatingof milk to ca. 90 oC and precipitation ofcaseins plus whey proteins at 65 "O, byacidification to pH 4.6-5.6, or by additionof soluble calcium salts.

An increasing proportion of whey istreated each year by ultrafiltration for themanufacture of whey protein concentrates(WPC) which have high protein contents(30-70%).

Finally, modified whey powders areavailable, which have low contents inlactose or minerais or both. These areobtained by lactose crystallization and/orwhey demineralization by electrodialysisor ion-exchange (Muller, 1982; Marshall,1982).

Several laboratories and firms areattempting to separate I3-Lgfrom a-La onan industrial scale. One can see tworeasons for this operation : first, it wouldallow exploitation of' their ditferentfunctional properties in food manufacture.Second, it is considered that I3-Lg isdetrimental to human neonates fed cowmilk, or "humanized milk" in which theratio whole casein-whey proteins islowered to attain the value observed inhuman milk. This assumption is based onthe allergenicity of I3-Lg, which seemssomewhat higher than that of the otherbovine milk proteins. The absence of I3-Lgfrom human milk is also proposed as areason for eliminating this protein frominfant formulas.

A method was published by Pearce(1983b) which, in our opinion, is quitepromising. By moderate heat treatment(65 "C for less than 30 min) of Cheddarcheese whey retentate brought to pH 4.1-4.3, ail whey proteins precipitated exceptI3-Lg. This separation has been quitec1ear-cut in our experience. An improvedmodification of this technique has been

Milkproteinanalysis 363

developed by Pierre & Fauquant (1986).However, nobody knows whether theresulting precipitate, which besides a-Lacontains a fairly large amount of BSA:IgGs and LF, is less detrimental than p-Lgto babies allergie to cow milk.

Characteristics of sorne biologicallyactive pro teins

Ali milk proteins are likely to have abiological function in the cow or/and in thecalf. They ail supply the latter with theamine acids it needs. This is especiallythe case for caseins which are quiteaccessible to the gastrointestinal proteo-Iytic enzymes. In addition, the micellarsystem provides the calf with largeamounts of calcium and phosphate;clottinq of milk in the stomach regulatesintestinal transit. a-La is part of theenzyme lactose synthetase which cata-lyzes the last step in the biosynthesis oflactose (Kuhn, 1983). I3-Lg has recentlybeen found to be a retinol-binding protein(Papiz et al., 1986). LF, IgGs and LP areinvolved in the protection of the mammarygland and of the calf's gut againstpathogenic microorganisms. We do notknow whether the many enzymes that arepresent in milk have any particularfunction. Most of them occur and havedefinite functions in the body fluids of thecow and/or they may simply occur through"Ieakages" from blood. Indeed, milkplasminogen (Pg), the precursor ofplasmin, an active proteinase, has beenshown to originate from blood (Eigel etel., 1979), in which it has an importantfunction (Iysis of fibrin clots).

We will focus here on those biologicallyactive milk proteins whose activity can beexploited in dairy technology, or must beavoided : LP, IgGs, LF, plasmin, catalaseand alkaline phosphatase (AP).

Lactoperoxidase

This is the most abundant enzyme in milk(ca. 30 mg/I). It is virtually completelyinactivated by pasteurization of milk at78 -c for 15 s (Griffiths, 1986). LP (EC1.11.1.7) catalyzes the oxidation byhydrogen peroxide of a number ofunsaturated organic compounds (e.g.,o-dianisidine, pyrogallol, guaiacol) andsorne halides or pseudo-halides (Br, 1-,SCN-). The oxidation of SCN- and 1-is ofspecial interest, as weil as that of sorneorganic compounds that can be used forcolorimetrie determination of LP inprocessed milk to assess pasteurizationefficiency. The preparation of LP from milkor whey (e.g., Prieels & Peiffer, 1985;Mailliart & Ribadeau-Dumas, 1987) isfairly easy because of its high isoelectricpH (ca. 9.5) and molecular weight (ca. 78kDa). It is now commercially availablefrom a few dairy firms. Carbohydrates anda heme group are attached to the proteinmolecule. The latter is essential foractivity and gives solutions of the enzymea greenish color, with an absorbancemaximum at 412 nm. For the purestpreparations of LP, the ratio E 412 nm/E280 nm reaches a value of ca. 1.0. On ionexchange chromatography, LP appears tobe heterogeneous, possibly due todifferent degrees of glycosylation orpartial proteolysis.

LP catalyses the iodination of free orprotein-bound tyrosine, according to thereaction:

LP ~-t H20 + OH- + HO - P -CH2 - C\ - CO

1 NH/


This reaction is widely used for theradiolabelling of proteins, e.g., with 1251.The specifie activity of LP for iodination ismore than 100-fold higher than that ofhorseradish peroxidase (HP) (Pruitt &Tenovuo, 1985).

LP, as weil as HP and AP, is used inenzyme-Iinked immunosorbent assays(ELISA). The ELISA techniques can beemployed to determine any antigen Ag(e.g., protein, microorganism, etc.) incomplex media. In the procedure usinginhibition of antibody binding, antibodies(Ab1) are raised against Ag in, say,rabbits. Other antibodies (Ab2) are raisedagainst Ab1 in another animal species,purified and covalent complexes LP-Ab2prepared. A known amount of pure Ag isinsolubilized and fixed in an plastic weiland a suboptimal amount of Agi is addedtogether with an aliquot of the medium inwhich Ag is present at a concentration tobe determined. This soluble Ag partiallyinhibits the bindingof Ab1 to the insolubleAg. After washing, LP-Ab2 is added inexcess and the excess is washed-out.The amount of fixed LP-Ab2, measuredfrom the LP-activity, is inverselyproportional to the amount of addedsoluble Ag to be determined. The extremesensitivity of ELISA techniques is due tothe fact that several Ab1 molecules bindto one Ag molecule. Similarly, severalmolecules of LP-Ab2 bind to eachmolecule of Ab1. Therefore, a largenumber of LP molecules corresponds to asingle Ag molecule.

lt seems that LP, which is present inmost mucosal secretions (milk, saliva,cervical mucus), has a biological functionboth in the mammary gland and in thedigestive tract of the calf. In vitro and inmilk, it constitutes part of a bactericidalsystem (LP-system) that involves thefollowing reaction :

LPSCN- + H202 - OSCN- + H20

This system is active against a number ofGram+ and Gram- bacterial strains.

The hypothiocyanite ion, OSCN-, orhigher oxiacid derivatives, is responsiblefor the antibacterial effect which isbelieved to be due to oxidation of SHgroups, NADH and NADPH, that areessential for bacteria. The SCN- level inmilk is significant for cows fed on grass orhay. However, there is no source of H202in normal fresh milk. Xanthine oxidase, anenzyme of the fat globule membrane, isable to liberate H202, but there is notenough substrate (such as xanthine,aldehydes) in milk for this enzyme. TheLP-system may be activated in themammary gland in case of mastitis, H202being produced by bacteria orphagocyting leucocytes. Similar activationcould occur in the gut of the calf (Pruitt &Tenovuo, 1985).

Activation of the LP-system by adding10-12 ppm thiocyanate and 8-10 ppmH202, has been used as a means ofprotecting milk in tropical countries fromearly spoilage (Pruitt & Tenovuo, 1985).The same system can be used in non-dairy products (or milk products in whichLP has been destroyed by heattreatments), provided the 3 componentsof the system (LP/SCN-/H202) are added.Data from one of the LP-producing firms(Oleofina, Brussels) seem to show thatsupplementation of the diet with both LPand LF can prevent microbial diarrhoea incalves.

Immunoglobulins

ln the adult cow, lymphocytes-B areactivated by foreign Ag macromolecules(e.g., from bacteria), and each proliferatesand differentiates to a plasmocyte thatsecretes into the blood IgGs (or IgMs),specifie for the antigenic determinant (a


part of the Ag) which has caused theinitial activation. IgGs are large moleculesmade up of two sets of two identicalpeptide chains, light (L) and heavy (H).The light chains (Mr 22500) can be of twotypes (K or À). Similarly, in the cow, thereare two types of heavy chains, y1 and 'Y 2,giving the sub-c1asses IgG1 and IgG2.Sorne carbohydrates are bound to the Hchains. The four chains are Iinkedtogether by several disulfide bridges. It isnot possible in this review to give adetailed description of the mechanism ofaction of IgGs. Let us just say that, whenthe blood stream has previouslyencountered a bacterium, for example,the organism possesses a number oflymphocytes B able to produce a myriadof plasmocyte clones, each is able tosecrete one type of IgG specifie for one ofthe many determinants of the samebacterium. These IgGs each bind to thecorresponding determinant. IgG fixationinduces the activation of "complement", adozen serum proteinases or proteinswhich activate each other to ultimatelyperforate the bacterial membrane andthus destroy the microorganism.

This means that a large population ofsimilar, but not identical, IgGs circulate inthe blood stream. The calf, which is borndevoid of any circulating IgGs, mustreceive them from colostrum, which ishighly charged with IgGs, and the milk ofits mother. The cow mammary glandselectively transfers IgG1 from bloodserum to colostrum (ca. 60 g/llgs, 80% asIgG1). The other Igs found in colostrumand milk are IgMs (whose function issimilar to that of IgGs) and secretory IgAs(slgAs) that are specifie to externalsecretions (milk, saliva, gastrointestinalmucus, etc.). The latter have a localaction against microorganisms. In the gut,they aggregate those which have inducedtheir production (as for IgGs), thuspreventing their attachment to the

intestinal epithelium. In mature cow milk,the total Ig content is much lower (Ca 0.8g/C) than in colostrum and the proportionof IgAs is higher : 70% Ig G1, 17% slgA(Butler, 1974). Preparation of Igs, fromcolostrum or milk, is now feasible on apilot scale (e.g., Peyrouset, 1982). Theresulting products contain a huge varietyof molecular species directed against anumber of unknown microorganisms,unless the cow has been vaccinatedagainst the known ones. In that case, partof the IgGs found in colostrum and milkwill be active against the latter afier theyhave passed into the calf's blood serum.Igs prepared from milk or colostrum wheyare mainly used in veterinary medicine forsupplementing calf diet.

Lactoferrin

LF consists of a single peptide chain of Mr80 000. The primary structure of humanmilk LF is known. The protein has 703amino acid residues and 16 disulfidebridges. It bears 2 carbohydrate chains.The molecule is formed of 2 symmetricalhomologous parts, each carrying a Fe3+

binding site (Metz-Boutigue et al., 1984).For each ferric ion bound to such a site,one bicarbonate anion is concomitantlybound, and ca. 3 protons are released.Iron and carbonate alone cannot bindtightly. Bovine colostrum and milk containapproximately 6 and 0.2 g/l LF,respectively, which is found exclusively inwhey. Although the degree of ironsaturation of bovine LF in milk is a matterof controversy, LF preparations show adegree of saturation of ca. 20%. BovineLF is commercially available. Its highisoelectric pH (ca. 8) makes itspreparation from whey by ion exchangefairly easy (e.g., Peyrouset, 1982; Prieels& Peiffer, 1985; Monsan et al., 1985;Mailliart & Ribadeau-Dumas, 1987).

Catalase, Alkaline Phosphatase,Plasmin


Conventional pasteurization does notdenature it extensively.

Like LP and IgAs, LF occurs in mostexocrine secretions (bronchial mucus,saliva, tears, gastric juice, duodenalmucus, etc.). lt is also present inneutrophil granulocytes.

Iron saturation and depletion can beeasily performed without altering theprotein molecule. The saturated form ispink, with a broad absorbance maximumat 460 nm. When the pH is lowered, LFbegins to lose iron around pH 4.5. At pH4.4, the iron saturation level is still 75%.

Although many investigations havebeen concerned with the biological role ofLF,the situation is still not clear. Two mainputative functions have been assigned tothis milk protein. Together with IgAs, LPand lysozyme, it could act as anantibacterial agent in exocrine secretions.Indeed, it has been known for a long timethat unsaturated LF prevents, in vitro, thegrowth of a number of bacterial strains bychelating the iron they need for growth.This bacteriostatic effect is enhanced byIgs. It may protect the mammary gland; inthis organ the LF concentration increasesmarkedly during mastitis and involution. Itmay also protect the digestive tract of thecalf against pathogenic microorganisms.However, this has not been clearlydemonstrated. As mentioned above, aproducer firm claims a beneficial action ofLF + LP in preventing-diarrhoea in calves.

Also, LF could play a role similar tothat of blood serum transferrin, a verysimilar protein, by making iron moreavailable for absorption in the guI. Indeed,receptors for LF have been found inhuman and monkey brush border mem-brane (Huebers & Finch, 1987).

These enzymes, which are quite different,have been placed together here becausedetermination of their activities in milk isused or could be used as quality criteria.

High levels of catalase (and N-acetylglucosaminidase) in milk usually resultfrom mastitis. The absence of AP (or LP;see above) in pasteurized milk indicatesthat the heat treatment has been highenough to destroy pathogenic germs.Finally, a high level of plasmin activity inraw milk may be detrimental for furtherprocessing (cheese manufacture, UHTtreatment).

Gatalase (EG 1.11.1.6)

This enzyme is found in both skim milkand cream, and appears to be associatedwith membrane material in both of thesefractions. Its level in milk is correlated withthe somatic cell count and thus, themeasure of its activity can be used todetect mastitis (Kitchen, 1985). Theenzyme has been crystallized frombuttermilk (Ito & Akusawa, 1983). It isfound in nearly ail aerobic cells, and invivo, is partially responsible for protectingcells against the toxic effects of H202.Mammalian catalases consist of 4identical subunits of Mr 60000, eachbearing a heme group (Fe 111-protoporphyrin IX). The known amino acidsequences of catalase from bovine liverand erythrocytes are identical (Quan etal., 1986). Catalase from milk is likely tobe the same molecule. The enzymecatalyses the reaction :

2H202 - 2H20 + O2


lts optimum pH and temperature are8.0 and 20 oC, respectively. It is one of themore heat-Iabile enzymes occurring inmilk, with most of its activity beingdestroyed at 72 oC for 15 s (Griffiths,1986).

High level of N-acetyl-~-D-glucos-aminidase (EC 3.2.1.30) in milk is alsofound during mastitis (Kitchen, 1985).

Alkaline Phosphatase (EG 3. 1.3. 1)

Like the former, this enzyme is foundmainly in membrane structures of bothskim milk and cream. It has been shownthat at least 3 genes code for 3 differentAPs : the placental, intestinal and liver/bone/kidney (LBK) APs (Weiss et al.,1986).

Complete primary structure of 3 APsare known : that of the E. coli enzyme,and those of the human placental andLKB APs. Ali three display homologies,with 52 % amino acid identity between the2 human enzymes. Human milk APappears to be quite similar to the liverenzyme. Both enzymes probably differonly in their carbohydrate moieties(Hamilton et al., 1979; Weiss et al., 1986).The human LBK enzyme has a single524-residue-long chain. It contains 5potential sites of glycosylation. Accordingto recent data (Weiss et al., 1986), theseenzymes are anchored to the luminal sideof cells, either by aC-terminai hydro-phobie stretch of ca. 23 amino acids, or bycovalent binding to phosphatidyl inositol.

The bovine milk enzyme appears to bea dimer of two identical subunits (Mr -85000). Several Zn atoms seem to beessential for activity. The enzyme hasbeen purified from buttermilk. It is able tohydrolyse most phosphomonoesters :AMP, glycerophosphates, free andprotein-bound threonine and serine

phosphate. However, it does not seem tobe active in milk towards caseins,probably because of inhibition byinorganic phosphate, lactose and ~-Lg(Kitchen, 1985).

The determination of AP is applied todairy products to determine whetherpasteurization has been perlormedproperly and also to detect the possiblecontamination of pasteurized milk by rawmilk. An almost complete loss of activity isobtained by heating milk at temperaturesabove 70 -c for 15 s (Griffiths, 1986).

Plasmin (EG 3.4.21.7)

Bovine plasminogen (Pg) is a single-chainblood serum protein composed of 786amino acid residues with a number ofdisulfide bridges and 2 carbohydratemoieties (Schaller et al., 1985). In blood,Pg is involved in the Iysis of fibrin clot.Blood vessel endothelium synthesizes thetissue Pg-activator (t-PA), a serineproteinase, which is able, by splitting thesingle peptide bond 557-558, to convertPg into plasmin, another serineproteinase, consisting of a heavy and aIight chain linked by 2 disulfide bridges(Schaller et al., 1985). A small proportionof Pg is transferred to milk (Eigel et al.,1979) in which both Pg and plasmin arefound in a ratio of ca. 9:1. They seem tobe bound to casein micelles and the milkfat globule membrane (MFGM) (Hofmannet al., 1979). Pg activation occurs in themammary gland and in milk duringstorage, activation being faster inpasteurized than in raw milk (De Rham &Andrews, 1986). Plasmin is the mainproteinase in milk. lt is known to digestmainly ~- and Us2-Cns, as indicatedearlier, and appears to have asignificance in the processing technologyof milk and dairy products. In someproducts, such as Gouda and Swiss-type


cheese, its presence may be beneficial,while in others, such as UHT milk andsome casein products, it may bedeleterious (Richardson, 1983). Indeed,plasmin is fairly resistant to heat treat-ments (Driessen & Van der Waals, 1978).Finally, is is useful to know that milkcontains not only a PA (probably t-PA),but also plasmin inhibitors, such as a2-antiplasmin, and, probably, Pg-activatorinhibitor which, like t-PA, is secreted bythe epithelium of blood vessels.

DETERMINATION OF TOTAL PROTEININ MILK AND DAIRY PRODUCTS

As in many other fields, the nature andnumber of analyses performed to assessthe composition of milk and dairyproducts have changed dramaticallyduring the past 20 years. Fat, which wasconsidered to be the most valuablecomponent in milk since the verybeginning of the dairy industry, is nowbeing replaced by proteins which havehigher nutritional and economic values. Inmost countries with highly developeddairy industry, protein content is nowincluded in milk quality payment schemesand breeding programs.

Progress in milk protein research, asweil as the applications of protein testingto dairy husbandry and quality control inthe dairy industry, were for a long timehampered because no rapid and accuratemethod of analysis was available. A majorbreakthrough occurred with the intro-duction of the dye-binding methods in the1960's, followed by the development ofthe infra-red (IR) techniques which hadthe advantage of measuring directly ailthe major milk compounds : fat, proteinand lactose.

This chapter will first deal with newdevelopments concerning the determin-ation of nitrogen, mainly by the Kjeldahlmethod, then the two most importantindirect methods currently in use in dairylaboratories, dye-binding and IR methods,will be described. Because most of thestudies in this field have been devoted tomilk protein testing, relatively littleinformation is available on their appli-cation to dairy products. Extensivereviews on the numerous methods formeasurement of the protein content inmilk have been published by Bosset et al.(1976), and Guillou et al. (1986).

"Direct" methods (nitrogen determin-ation)

ln milk, as weil as in other foodstuffs,nitrogen is the element which essentiallycharacterizes proteins. As a conse-quence, nitrogen determination hasalways been used as a standard methodfor the estimation of the protein content offood.

The Kjeldahl method

This method is now internationallyrecognized as the reference method formeasuring the protein content in milkproducts, and is listed as such in theCodex Alimentarius.

Principle

ln 1883, Kjeldahl discovered that, byheating organic compounds in concen-trated sulfuric acid, nitrogen is convertedquantitatively into ammonium sulfate andcan subsequently be estimated as


ammonia by distillation and titration afteraddition of sodium hydroxide.

Ouring digestion, carbon is transform-ed into CO2 and hydrogen into H20.Oetailed information concerning thevarious reactions involved duringdigestion can be found in the review ofMcKenzie and Murphy (1970).

Development

Since its discovery, this method has beenstudied extensively and the procedurerevised periodically to improve both thedigestion rate and the accuracy. Theobjective is to convert, as quickly aspossible, the totality of the organicnitrogen, even the most refractorycompounds, into NH..j without loss ofnitrogen by pyrolytic decomposition ofammonia. It is now weil established that ahigh digestion rate and good nitrogenrecovery (over 99%) can be achieved onlywith a suitable ratio of K2S04/H2S04 toreach a high boiling temperature, and withthe addition of a catalyst. Severalconcentrations of K2S04, types of catalyst(selenium, mercury, copper) and oxidizingagent (H202) have been tested. If, withhard-to-digest compounds, mercury iscertainly the most efficient catalyst, it nolonger tends to be used because ofcurrent environmental concerns. Coppersulfate, which was first proposed byRowland (1938) for milk analysis, seemsto be a good alternative.

Following a study by Rexroad andCathey (1976), recent studies onfoodstuffs (Kane, 1984) and on milk (IDF,1986) have shown that HgO and CUS04give identical results.

ln the latest version of lOF Standard 20A/1986 (1986), for the determination ofnitrogen in milk by the macro-Kjeldahlmethod (5 ml of milk), the following ratios

are prescribed K2S04/H2S04 = 15g/25ml; CUS04' 5 H20/H2S04 = 0.05 g/25 ml.Compared to the original CUS04concentration recommended by Rowland(1938) (i.e., 0.2 giS ml), a 12-fold lowerconcentration was adopted to prevent theformation of ammoniacal complexes withcopper, which may lead to a relativeunderestimation of about 1 % of thenitrogen content.

It is important to bear in mind that acatalyst cannot be changed withoutreconsidering the whole procedure, andespecially the K2S04/H2S04 ratio and thetotal mineralization time. It is alsorecommended to consider the totalmineralization time given in a Standard asa minimum. The heating time should notbe reduced if the clearing time (when thedigest become clear) is short, as forinstance with low-fat milk samples.

On the other hand, for samples with ahigh fat or protein content, the arnount ofH2S04 has to be increased becauseorganic material consumes H2S04, andthe total mineralization tirne may have tobe extended if the clearing time is longerthan that given in the Standard.

For routine analysis, the c1assicalKjeldahl flask and gas burner or electricheater with separate distillationapparatus, are progressively beingreplaced by block digestion and steamdistillation apparatus, which give resultssimilar to those obtained with thestandard method. The traditional titrationusing an indicator solution can be carriedout with automatic pH titration or specificion electrode (Pailler, 1982). An automaticdevice developed by Foss Electric (OK),the Kjelfoss, which combines in a singleinstrument ail the steps of the procedure,i.e., mineralization, distillation andtitration, allows the determination of 20sampies per hour with good precision andaccuracy (Grappin & Jeunet, 1976).


Ana~ücalperlormance

When performed correctly, the Kjeldahlmethod is assumed to give the truenitrogen content of milk. However, manyreports of collaborative studies haveshown that large discrepancies betweenlaboratories frequently occur. To improvethe reproducibility, the new lOF standardrequires that two accuracy tests beperformed regularly. First, one test basedupon the analysis of tryptophan or asimilar refractory compound, e.g.,phenacetin, to check the mineralizationefficiency, with a percent of N recoverybetter than 98%. Lysine-HCI is analternative, but difficulties have beenreported by Kane (1984); nicotinamide,which is also a difficult-to-digest material,does not give an excellent recovery withcopper sulfate. Second, a test based onthe analysis of an ammonium salt (sulfateor oxalate) is performed to check thedistillation and the titration steps, with apercent of N recovery between 99 and100%.

To check their procedure, laboratoriesmay take advantage of using referencematerials like an NH4 H2 P04 solution ora milk powder with a certified N contentsupplied, respectively, by the AmericanNational Bureau of Standards, and theEuropean Bureau Communautaire deRéférence (Brussels).

A collaborative study on the lOFStandard 20 AJ1986 (Grappin & Horwitz,1988), carried out with the participation of24 reporting laboratories representing 12countries, gave, for the normal range ofvariation of nitrogen in milk, arepeatability relative SO of 0.5% and areproducibility relative SO of 1%.

The Dumas and related methods

ln these methods, organic and inorganicnitrogen is converted into nitrogen gaswhich is determined by gas chromato-graphy, volumetrically or by thermalconductivity. Commercially available areautomatic apparatuses able to performfast and reliable analyses (e.g.,the CarloErba NA 1 500 machine determinesnitrogen in 3 min). One machine wastested in 1974 by Lunder on casein andwhole dried milk and gave reliable data.Recently, two instrument manufacturers(Leco Corp, St Joseph, Michigan, USA,and Heraeus GmbH, Hanau, FRG) havedeveloped automated instruments for themeasurement, in 3 to 7 min, of nitrogen bythermal conductivity.

Conversion factor - Terminology

To estimate the amount of protein in milkand milk products, it is necessary toconvert nitrogen into protein, bymultiplying the nitrogen content by afactor called the Kjeldahl conversionfactor. A value of 6.38 for this factor,originally proposed a century aga byHammarsten and Sebelien on the basesof the nitrogen content (15.67%) ofpurified acid-precipitated casein, isgenerally accepted and was confirmed inthe latest lOF Standard.

However, this mode of determining theprotein content raises two importantquestions. First, the terminology "proteincontent" is not fully correct since theproportion of non-protein nitrogen (NPN),within and between dairy products, variesfrom 3-8% in milk and up to 25-30% inwhey. To avoid confusion, the term "cru deprotein" has to be used to express the


Table Il. Protein content and Kjeldahl factors of milk (Karman & Van Boekel, 1986).Teneur en protéines et facteurs Kjeldahl du lait (Karman & Van Boekel, 1986).

Amount (g/I)Protein Without carbohydrateN% Kjeldahl

factors

With carbohydrateN% Kjeldahl

factors

asl-CN 10.0 15.77 6.34as2-CN 2.6 15.83 6.30~-CN 9.3 15.76 6.34lC-CN 3.3 16.26 6.15y-CN 0.8 15.87 6.30~-Lg 3.2 15.68 6.38a-La 1.2 16.29 6.14SA 0.4 16.46 6.07Ig 0.8 16.66 6.00PP5, 8F, 8S 0.5 15.30 6.54PP3 0.3 16.97 5.89Lactoferrin 0.1 17.48 5.72Transferrin 0.1 17.00 5.88MFGM 0.4 15.15 6.60

Milk 33.0 15.87 6.30

15.67 6.38

16.14 6.20

15.27 6.5516.29 6.1416.10 6.2114.13 7.08

15.76 6.34

Table III. Experimental Kjeldahl factors for isolated milk protein (Karman & Van Boekel, 1986).Facteurs Kjeldahl expérimentaux pour des protéines du lait isolées (Karman & Van Boekel, 1986).

Protein % Non-protein Corrected Experimental Theoreticalash' N% Kjeldahl Kjeldahl

factor factor

2.16 15.55 6.43 6.333.78 16.41 6.10 6.342.11 14.84 6.74 6.381.40 15.62 6.40 6.34

12.60 14.97 6.68 6.38

as-Casein~-Caseinx-CaseinTotal casein~-Lactoglobulin

1. The difference between the ash content and the sum of P04 and S04 contents taken as non-protein ash (theS04 content was taken for I3-Lactoglobulin only).2. Corrected for fat and water content and non-protein ash.t, La différence entre la teneur en cendres et la somme des teneurs en P04 & 504 a été considérée commecendres non protéiques (seule la teneur en 504 a été prise pour la B-Iactoglobuline).2. Corrigé pour les teneurs en matière grasse et en eau et pour les cendres non protéiques.

does not measure directly the componentit is intended to quantify, but insteadmeasures one or more quantities orproperties which are functionally linked tothat component. For instance, the AmidoBlack method is an indirect methodbecause it does not directly determine theprotein content, but measures the amountof dye that is bound to proteins. The sameholds for instrumental methods such as IRtechniques. As a consequence, indirectmethods require standardization against areference method to convert the instru-mentai signal into component concen-tration. We may point out that, followingthis definition, the Kjeldahl method mayalso be considered an indirect method,even though it is accepted as a referencefor the determination of the proteincontent in milk products.


nitrogenous matter in milk. Its quantitativeexpression is represented by the amountof total nitrogen multiplied by 6.38, and isexpressed in 9 per 100 9 (or kg or liter) ofmilk or milk product.

Second, the conversion factor is notconstant, but highly dependent on theamine acid composition of the proteinfraction. Recently, using the primarystructure of milk proteins, Karman andVan Boekel (1986) have shown that forcow's milk, the conversion factor shouldbe 6.34 instead of 6.38, and differentfactors shoud be used for casein (6.34),paracasein (6.29), rennet whey proteins(6.45), acid whey proteins (6.30), andNPN (6.30). Recently, these figures wereslightly corrected by Van Boekel andRibadeau-Dumas (1987). For individualproteins fractions, the variability of thefactor is even greater (Table Il). In theirstudy, they demonstrated that experi-mental determination of the Kjeldahlfactor on (pure) protein fractions leads tosubstantial discrepancies from thetheoretical values obtained from aminoacid sequences (Table III), mainlybecause it is difficult to obtain purefractions and to accurately measure theash content.

Even though the term "crude protein"and the conversation factor of 6.38 wererecently confirmed in the Codex Alimen-tarius for milk and milk products, a moreaccurate definition and methodology formeasuring the "true" protein contentshould be promoted.

The indirect methods

Definition and characteristics

Definition

According to Grappin (1984), a method ofanalysis is designated "indirect" when it

Analytical attributes and calibration

Because of the influence of samplematrix, the quantities or propertiesmeasured by indirect methods are notusually exactly proportional to the compo-nent concentration, therefore introducingsystematic errors in the final test result.

Figure 1 illustrates the various criteriainvolved in the overail accuracy of anindirect method (IDF, 1987). If we exceptrandom errors (i.e., instrument repeat-ability), systematic errors can be dividedinto two components :

- the exactness of calibration, whichrelates to the closeness of agreement, ateach concentration level, between theindirect method value, represented by theequation of the observed line, and theestimated mean of the true values of ailthe individual samples at the corres-ponding levaI.

- the accuracy of estimate, or simplyaccuracy, which relates to the closeness


of agreement between individual testresults which are obtained with theinstrument exactly calibrated and with thereference method. The ellipse in Figure 1,which represents the 95% confidenceinterval within which the true value isexpected to lie for any result of theindirect method, illustrates the accuracy ofthe indirect method. The mathematicalexpression of the accuracy ls given by theresidual standard deviation Sy,x. Thisimportant attribute of an indirect methodgives the best estimate of the predictingvalue of the indirect method.

To counteract the influence of known orunknown, which are common to a wholepopulation of sampi es, the indirectmethod will have to be calibrated. Bydefinition (IDF, 1987), the calibration of aninstrument concerns the adjustment of theinstrument signal so that, at each level of

Re.ferencemethod

373

the component concentration, the meanof individual test result given by theinstrument is closely approximate to thetrue value of the component concen-tration. Practically, the calibration of anindirect method can be achieved bycalculating the relationship between thecomponent concentration obtained by thereference method on a set of samples orby reference materials, and the instrumentsignal.

The dye-binding methods

Following the fundamental work ofFraenkel-Conrat and Cooper (1944) onthe use of dyes for the determination ofacid and basic groups in proteins,quantitative determination of milk proteinby dye-binding was introduced in 1956 by

Accuracy

Indirectmethod

Fig. 1. Breakdown of the overail accuracy of an indirect method (IDF, 1987). Xi = single instrumentalvalue; Xi = arithmetic mean of several determinations of sampie S with the instrument; y = estimatedmean reference value for ail the samples with an instrument level Xi ; Yi = "true value" of thecomponent to be measured for sampie S Sy, x = standard deviation from the regression. Themathematical model of the components of the total error on x is :

(Xi- yi)(x- y;) = (Xi- x;) +~ accuracy of the mean

overall accuracy repeatability '" (Xi - Yi) + (y - Y;)exactness of accuracy calibration

Décomposition de la précision globale d'une méthode indirecte (lOF, 1987) x = valeur instrumentalesingulière; Xi = moyenne arithmétique de plusieurs dosages effectués sur l'échantillon S avec/'instrument; y = valeur de référence moyenne estimée avec un niveau instrumental Xi; Yi = «valeurexacte» correspondant au composé à doser dans l'échantillon S; S y, x = écart type de larégression. Le modèle mathématique des composants de l'erreur totale sur x est:

(Xi -Yi) l

(x - Yi) (Xi - ~i) + / précision de la moyenneprécision globale répétabilité ',. (Xi - y) + (y - Yj)

exactitude de précision calibration

374

excess of dye should be present insolution to get both a complete reactionbetween dye and proteins and aproportionality between milk proteincontent and the amount of dye whichprecipitates. Similarly, if the dye/proteinratio is too high, the DBC increases (Alaiset al., 1961; Ashworth & Chaudry, 1962).Orange G seems to have a betterstoichiometric reaction than Amido Black(Dolby, 1961; Ashworth & Chaudry, 1962).The various milk proteins or nitrogenfractions do not react in a similar way withthe dye. Several authors have foundsimilar results concerning the DBC of milkprotein fractions (Alais et al., 1961;Ashworth & Chaudry, 1962; Tarassuk etal., 1967; McGann, 1978). In Table IV arereported the findings of Tarassuk et al.(1967) and Ashworth and Chaudry (1962).It is important to point out that wheyproteins have a higher DBC (ca. 27%more) than the caseins and that NPNdoes not bind any dye, rather, it remainsin the supernatant (Alais et al., 1961;Tarassuk et al., 1967).

B. Ribadeau-Dumas and R. Grappin

Schober and Hetzel and by Udy. Thismethod was the first routine methodavailable for breeding programs andpayment schemes, and is still used forcontrol in the dairy industry. A recent briefreview of the method was published in1987 by Van Reusel and Klijn in the IDFmonograph on rapid indirect methods formeasurement of the major components ofmilk.

Principle

The dye-binding procedures used for milkprotein testing employ different azo-sulfonic acid dyes : CI Acid Black 1 (orAmido Black), CI Acid Orange 12, and CIAcid Orange 10 (or Orange G). Themechanism of the reaction between thedye and the proteins is now relativelyweil, if not fully, understood. In acid buffersolution below the isoelectric point of theproteins, the positive charges of both theterminal amino groups, and the histidine,arginine and lysine residues, combine in astoichiometric interaction with thenegative sulfonic group of the dye to forman insoluble dye-protein complex. Inaddition, Lakin (1974) has shown thathydrophobie interactions occur betweenfree and bound anionic dyes.

The differences between dyes, withrespect to their sensitivity (change ofabsorbancy per unit change of proteinweight), are partly associated with theirabilities to form hydrophobie interactions.Amido Black has the highest sensitivityfollowed by Acid Orange 12 and AcidOrange 10.

The quantity of dye bound per unitweight of protein, called the dye-bindingcapacity (DBC) of proteins, is dependenton the ratio of dye over proteins. Severalworkers (Fraenkel-Comat & Cooper,1944; Dolby, 1961) have shown that an

Analysis

The measurement of protein content isbased upon the determination of theamount of unbound dye after removal ofthe dye-protein precipitate. Provided thatthe ratio of dye to protein remains withincertain limits, a linear relationship existsbetween the protein content and thequantity of unbound dye. .However,McGann and Murphy (quoted in McGann,1978) have demonstrated that, with AcidOrange 12, the relationship betweenabsorbance and protein content is bestgiven using a quadratic equation. Asimilar observation was made recently inour laboratory (unpublished data) for theAmido Black method with different kindsof milk samples individual milk,reconstituted or ultrafiltrated milk sampies.


Table IV. Dye-binding capacity of milk protelns, mg Amido Black (AB) or Orange G (OG) per 9protein. Within brackets : ratio to casein.Capacité de fixation du colorant par les protéines du lait, en mg de Noir-Amido (AB) ou Orangé G(OG) par g de protéines. Entre parenthèses: rapporté à la caséine.

Tarassuk et al. (1967) Ashworth and Chaudry (1962)AB OG AB

Milk 350 (1.08) 179 (0.95) 354 (1.04)Casein 325 189 339Whey proteins 410 (1.2G) 247 (1.31) 447 (1.32)Proteose-peptones 350 (1.03)

Caseinalpha 345 (LOG) 199 (1.05) 373 (1.10)beta 290 (0.89) 170 (0.90) 318 (0.94)kappa 310 (0.95) 193 (1.02) 359 (LOG)

Whey proteinsalpha lactalbumin 440 (1.35)beta lactoglobulin 425 (1.31 ) 252 (1.33) 45G (1.35)serum albumin 430 (1.32)immunoglobulins 320 (0.98)

The procedure of the dye-bindingmethod involves four steps (Mc Gann,1978) :- The sample (between 0.5 and 1.0 ml)is thoroughly mixed with a constantvolume of dye solution (about 20 ml) at apH of 2.2-2.4, containing phosphate-citrate buffer for the Amido Black methodand phosphate-acetate-propionate bufferfor the Udy method using Acid Orange 12.- The second stage involves a rapidpartial equilibrium (ca. 5 s) between dyeand proteins to react and form aninsoluble complex.- A c1ear supernatant is obtained byfiltration or centrifugation.- The absorbance of the supernatant ismeasured in a spectrophotometer, usingusually a short path-Iength flow-throughcuvette at a fixed wavelength. between550 and 610 nm for Amido Black, and 480nm for Acid Orange 12. To obtain theprotein content, the instrument has to be

calibrated against the reference methodaccording to the principle given p. 372,and a linear regression line is normallyobtained within the normal range of milkprotein content (2.5-4.5%).

To improve the reproducibility of resultsbetween laboratories, a centralizedsystem of calibration using reference milksamples has been used, for instance, inFrance (Grappin & Jeunet, 1976), Poland(Michalak & Deskowicz, 1975) andAustralia (Clarke, 1979). Internationalstandards for dye-binding methods arenow available from AOAC and lOF.

Instrumentation

To perform a protein test by dye-binding,no specifie instrument is necessary excepta centrifuge and a spectrophotometer. Forroutine determination in a central testinglaboratory, fully automated equipmentusing Amido Black was first developed in

376

throughout the year and also betweengoat and cow milk (Lakin, 1970; Grappin& Jeunet, 1976; Grappin et al., 1979).However, for individual goat milk Grappinand Jeunet (1979) have shown than 67%of the differences between Kjeldahl trueprotein contents and Amido Blackreadings are explained by the variation ofthe ratio whey protein/true protein in milk(Fig. 3). As expected, a slight butsignificant influence of mastitis onaccuracy has been found by Grappin etal. (1970) and Waite and Smith (1972).For instance, on quarter milk samples, anincrease of 1 log somatic cell countincreases the Amido Black result by arelative 0.47% (Grappin et al., 1970).


Holland in 1960 and then in several otherEuropean countries, particularly inFrance. The Dutch apparatus used in1976 by the laboratory of Zutphen, wascapable of analyzing as many as 10 000samples per hour. This equipment hasbeen replaced gradually by IRtechniques.

Specifie manual equipment has beenmarketed since 1960 : the Udy system(Udy Analyser Co., Boulder CO., USA),using Acid Orange 12, and the Pro-MilkMKII (Foss Electric, Hillerod, DK) andProt-o-Mat (Funke-Gerber, Berlin, DE),using Amido Black.

Factors affecting the accuracy of proteindetermination in milk

Various factors affecting the selection ofthe dye to be used, the methodology andthe reaction conditions : have beenreviewed by Sherbon (1978) andTarassuk et al. (1967). Great care shouldbe taken to ensure the purity of the dye,especially Amido Black which sometimescontains a large proportion of sodiumchloride (Lakin, 1970) and traces of otherdyes. Amido Black 10 B is generallyrecommended for electrophoresis. On theother hand, Orange 12 can be purified(see AOAC standard).

Experience has shown that, despitethe fact that the various milk proteinfractions have a slightly different DBC(Table IV), the main source ofdiscrepancy between dye-binding andKjeldahl methods is the variation of NPNcontent. In a population of individual milksamples, the accuracy SD (Sy,x) islowered by 60% when true protein is usedfor reference instead of crude protein(Fig. 2). Variations in the ratio NPNlTotalN in milk almost entirely explain thedifferences in calibration observed

40

35

35 '"'"'"

..;

'" "..; 30zQ.

Z Z>-- .!:

30 t:

25

25

.0 0.2 0.3Absorbance

0.40.1

Fig. 2. Determination of true protein contentand cru de protein of individual milk sampi es bythe Amido Black dye-binding method (lOF,1987).True protein : r = - 0.994 Sy, x = 0.31 g/kgCrude protein : r = - 0.985 Sy. x = 0.51 g/kg.Détermination de la teneur vraie et de lateneur brute en protéines d'échantillonsindividuels de lait par la méthode au NoirAmido (lOF, 1987).Teneur vraie en protéines : r = -0.994 Sy.x =0,31 g/kgTeneur brute en protéines: r = -0.985 Sy, x =0,51 g/kg.


Ana~ücalperiormance

Precison : several collaborative studieshave been undertaken to evaluated theprecision parameters of the dye-bindingmethods for milk protein testing (Sherbon,1975; Grappin et al., 1980). Arepeatability value of 0.03 g/100 9 caneasily be achieved by experiencedlaboratories. Concerning reproducibility,the relatively poor result of R = 0.215g/100 g, found by Grappin et al. (1980),can be dramatically improved (R = 0.068)when standard materials are used forinstrument calibration (Grappin, 1984).

Accuracy : according to the definitiongiven p. 372, milk protein content can beestimated with a SD (xy,x) varying fram0.03 g/100 9 to 0.06 g/100 g, accordingthe origin of the milk (individual or herd

4 ./]' J<, .2 "'"~

.1"

ai 0

~ -,~ -2~ -3-e'e5 -4

~ u ~ ~ u n ~ ft H ~ n n% whey proteins

Fig, 3, Relationship between the differences :(Amido Black readings-true protein Kjeldahlvalues) (y axis) and the percentage of wheyprotein in true proteins (x axis) of 81 individualgoat milk samples (Grappin & Jeunet, 1979).Coefficient of correlation: r = 0.82.Relation entre les différences : (lectures Noir-Amido - teneur vraie Kjeldahl) (axe des y) etle pourcentage de protéines du lactosérumdans les protéines vraies (axe des x) pour 81échantillons de lait individuel de chèvre(Grappin & Jeunet, 1979).Coefficient de corrélation: r = 0,82.

milk) and the method used (manual orautomatic).

Application to dairy products

The dye-binding methods are alsoapplicable to a large variety of dairypraducts, like cheeses (Steinsholt, 1976),chocolate drinks, cultured buttermilk andhalf and half (Sherbon & Luke, 1968),non-fat dry milk and ice cream (Sherbon &Luke, 1969) and wheys (Roeper & Dolby,1971). It is c1ear that, for each type ofproduct, the procedure has ta be adaptedand that a specific calibration is requiredto reflect both the modification in theprocedure (volume ratio of testsample/dye solution) and the possiblealteration of the DSC of the proteinmixture following heat treatment orprateolysis.

The infra-red method

Basic principle of IR measurements

The IR spectrumThree regions characterize the IRspectrum according ta the wavelength orwavenumber of the radiation: the near IRregion (NIR) from 0.7 urn to 2.5 urn (from14285 crrr ' to 4 000 cm-1), the mid IRregion (MIR), from 2.5 urn to 25 um (4000cm-1 ta 400 crrr t) and the far IR region,from 25 urn up to 100 um (from 400 cm-1

to 100 cm-1). The energy of electro-magnetic radiation is proportional to thefrequency of the radiation, i.e., inverselyproportional ta the wavelength. When amolecule is submitted to IR radiation,energy will be absorbed only if thefrequency of the radiation corresponds tothe frequency of one of the fundamentalvibrations of the molecule (stretchingvibrations at high frequencies andbending deformations at low frequencies).

378 B. Ribadeau-DumasandR. Grappin

The vibration energy which characterizesa chemical group (e.g. G-H, G-H, C=O)is dependent upon both the bondstrength, and the masses of the twoatoms of the group. Fundamentalvibrations of the molecules take placemainly in the MIR region, and absorptionin the NIR occurs at wavelengths corres-ponding either to harmonie frequencies.or to combination frequencies of thefundamental vibrations.

These NIR absorption bands aregenerally quite broad, allowing use ofrather large spectral pass bands formeasurements. Their intensities areusually weak compared to the signaisobtained in the MIR region fromfundamental vibrations of the molecules.However, they are sufficiently important toallow quantitative analysis. Because NIRbands are less specifie for a givenmolecule than wavelengths in MIR, onecan anticipate that quantitative analysis ofa specifie component will require morewavelengths to obtain accurate measure-ments.

Quantitative analysis

The measurement of the IR energyabsorbed by a sampie can be madeeither in the transmission mode, if theproduct is in solution or is sufficientlytransparent to IR radiation, or in thereflection mode for opaque or solidsamples. Most of the MIR equipmentdesigned for liquid samples measures thetransmitted light directly, while NIRinstruments are usually built to measurethe IR light which is either diffuselyreflected from the product surface forsolid samples or from the surface of thesample holder for liquids, or transmittedthrough a cuvette for liquids.

With a single component solution, theamount of IR energy that is absorbed bythe sample (absorbance) is exponentially

proportional tothe component concen-tration and follows the Beer-Lambert law.If la is the intensity of the incident beamand 1 the intensity of the transmitted orreflected beam, the absorbance is AÀ =log la Il = EÀ.C./, where the ratio 1110represents either the transmittance or thereflectance, E the extinction coefficient ofthe component, C the concentration, À. themeasurement wavelength and / the pathlength of the cell, for transmittancemeasurement.

The absorbance of n absorbingcomponents is then expressed by theequation:Ax= Ext' Ct + Exz· Cz+ ..· + Exn' Cnwhere Ct, Cz ... Cn are the concentrationsof the n components,and Ext' Exz ... Exn the absorptivities ofthe n components at the wavelengths Xt,

Xz, ... Xn·

ln conventional MIR instruments, tocorrect any variation. of the systemresponse (source brightness, temperatureof sample, soil on the cuvette, etc.) and to

: amine arid residue 1, 1

R, : R2 11 1 1 :

- [H - [0 +NH- [H - [0 -r- NH- Peptide chainl ,

o ~ H

H a tll b 1 bite<, 1 ,./" [~ /[............ 9N............ ~::i~'ed peptide

[ N 1 [

Ite H Ilt a

R, 1 H 1

1 1

:-[0- NH-:l '1 1:Peptide bond'

o1 11 1

:-[0 -NH-:1 11 1, 11Peptide bond

Fig. 4. Mid-IRAbsorptionbandsof the peptidebond.-, directionof the Iight-induceddipole;R1, R2, aminoacidsidechains.Bandes d'absorption de la liaison peptidiquedans /'IR moyen. _. direction du dipôle induitpar la lumière; RI' R2• chaînes latérales desacides aminés.

Milkproteinanalysis

diminish, as much as possible, theinfluence of interfering components likewater, the measurements are made byreference to the amount of IR energyabsorbed, either by water at the samewavelength as the measuring wavelength,or by the sample at a close wavelength atwhich there is only a slight absorption bythe component being measured. For NIRinstruments using reflectance, thereference is usually obtained by theintensity of the incident beam reflected bya ceramic disk.

Analysis of proteins by MIR transmission

Since the work of Goulden in 1964, whofirst studied the analysis of milk by MIRabsorption, a large number of papersassessing new instruments, or dealingwith more fundamental works, have beenpublished.

ln the IDF monograph on indirectmethods for milk analysis (1987), Biggset al. have reviewed the current status ofknowledge concerning this technique.Much of the information given in thischapter is drawn from this monograph. Acomprehensive work on this method hasbeen prepared by Sjaunja (1982).

100

9080.. 70

".. 60

1 50

>- 40302010

03.5

379

Principle

ln MIR, there is a strong absorption band,called "amide Il'', at ca. 6.46 um (1 550cm-1) by the peptide bond. Thisabsorption originates from G-N stretchingvibration (40%), and from N-H bendingdeformation (60%) (Fig. 4). The peptidebond also shows other absorption bandsat near 1 650 cm-1 (6.1 urn, "amide ll",due mainly to C = 0 stretching vibration)and at 3 300 cm-1 (3.0 urn, N-Hstretching vibration). As illustrated byFigure 5, the determination of proteinconcentration in milk is based upon the"specifie" absorption of the peptidelinkages at 6.46 urn. Although proteinsare the major absorbing compounds atthis wavelength, the absorbance isinfluenced to a greater or lesser extent bythe other major milk compounds (fat andlactose) and by minor soluble elements.As most of the instruments are designedto make simultaneous determination offat, protein and lactose (and sometimeswater), the best accuracy for proteindetermination will be obtained if theprirnary protein signal is corrected for theabsorption by fat and lactose at theprotein wavelength. With moderninstruments, automatic correction isachieved by setting internai correction

5.7 6.5 9.6

Wavelelength (~ml

Fig.5. The infra-redtransmissionspectrumof milk versus water (courtesyof MultispecLtd, IDF,1987).Spectre de transmission IR du lait contre l'eau (courtoisie de Multispec Ltd; IDF, 1987).

380

narrow band filters and microprocessors,simpler, more reliable, automatic andaccurate instruments are now available.Ali modern instruments are now equippedwith a single cell and optical filters. Theymay utilize either a single or a doublebeam optical system, and may use eitherelectronic ratioing or a servo system toestimate the 1/10ratio. Figure 6 iIIustratesdifferent optical systems used in IRinstruments and Table 5 gives the maincharacteristics of commercial instrumentscurrently available for the analysis of milkand milk products.

Another kind of instrument using aninterferometer for wavelength discrimi-nation, and known as Fourier Transforminfra-red spectrometer (FT-IR), is currentlybeing evaluated in our laboratory, as weilas in other research centers. Compared toether instruments, it has the advantage ofproviding high signal/noise ratio, highenergy, accurate frequency measurementand rapid scan, making spectral subtrac-tion a useful feature and allowing the useof a full spectrum, instead of a fewwavelengths.

B. Ribadeau-DumasandR.Grappin

factors for each component at eachwavelength. To minimize the light scat-tering influence of fat globules, thesampie has to be homogenized prior toanalysis to reduce the globule size below1 urn,

AnalysisA warmed sampie (40 OC), thoroughlymixed, and if necessary blended and/ordiluted, is pumped through a 1, 2 or 3stage valve homogenizer or an ultrasonichomogenizer. The specific wavelengthsare focused through the cell containingthe sample and the infra-red radiation ischopped by a half-section mirror toproduce alternating sampie and referencesignais. The difference between thereference and the sample signais is thencollected by an IR detector and the signaloutput is processed by a log/lin converterand corrected for interference. Compo-nent concentration is finally displayed onthe instrument. Before analysis, theinstrument must be calibrated accordingto the principle given p. 372. Oetailedcalibration procedures are given byinstrument manufacturers and a provi-sional international standard (lOF, 1988)has been produced by an IOF/ISO/AOACgroup of experts.

Instrumentation

Following the pioneering work ofGoulden, the first commercially availableIR instrument allowing the directmeasurement of fat, protein and lactosein milk, was a conventional double beamspectrophotometer with two cells (areference cell with water and a samplecell), with a servosystem and a diffractinggrating. This instrument, named Infra-RedMilk Analyzer (IRMA), was manufacturedby Grubb Parsons (UK). With theoutstanding progress made during thelast 20 years in electronics, interference

Factors affecting the accuracy of proteindetermination in milk

Physico-chemical factors : if we excludethe influence of instrumental factors liketemperature, linearity, water vapor in theoptical console, homogenization effi·ciency, etc. and if we assume that theinstrument is correctly calibrated (includ-ing the correction for fat and lactosecontents), the accuracy of milk proteintesting will be mainly influenced by thevariation in the proportion of NPN and bythe presence of carboxylic acids. Therelationship between infra-red absorptionat 6.46 urn and true protein concentrationmeasured by Kjeldahl is relatively inde-pendent of the amino acid composition ofthe protein, since the ratio N content

Manufacturer Instrument Year No of Beam Measu- Wavelength Homoge- Components Notescell system ment system nization

system stages

Grubb Parsson, UK IRMA 1965 2 Double Servo Grating - Fat, protein, lactose ObsoleteMini IRMA 1978 2 - Ratio Optical filter - Fat, protein, lactose,

waterFoss Electric, DK Milko-Sacn 300 1974 1 - Servo - - 1 Fat, protein

Milko-Scan 203 1975 1 - - - - 1 Fat, protein, lactoseMilko-Scan 1021101 1979 1 Single Ratio - - 2 Fat, protein, lactose,104 water ~Milko-Scan 203B 1979 1 Double Servo 2 Fat A, protein, lactose 5f- - -0Milko-Scan 103B 1980 1 Single Ratio - - 2 Fat B, protein, lactose 0Milko-Scan 104 NB 1961 1 - - - - 2 Fat A, fat B, protein éD

5"lactose Pl

Milko-Scan 104 Auto 1983 1 3 See Milko-Scan 104::J- - - - Pl

Milko-Scan Speed 1983 1 3 See Milko-Scan 104 -<- - - - Ul

Milko-Scan 605 1984 1 - 3 Fat A, fat B, protein,Ci,- - -

lactoseMilko-Scan 133/133N 1985 1 - - - - 1 Fat A or/and B, protein, Recriculating133B lactose homogeneization

REIL Milko-Scan 104 NB 1986 1 Single Ratio - - 4 Fat A, fat B, protein, lactoseMultispec, UK Multispec 1977 1 Double Servo - - 2 Fat A, fat B, protein, lactose

Multispec Auto 1978 1 - - - - 2 Fat A, fat B, protein, lactoseMicro-null 1984 1 - - - - 3 Fat A, fat B, protein, lactoseMultispec Il 1984 1 - - - - 2 Fat A, fat B, protein, lactoseDairy Lab 1986 1 Single Ratio - - 2 fat A, fat B, protein lactose

(with the technical assistance of Foss Electric) and others

tù

~

Table V. Major characteristics of mid infra-red instruments marketed since 1965 (IDF, 1987)

382

- Optical system of HILKO-SCAN 2001300:nstruments Idouble beam, one cett. servesystem, cptical filters 1.

B. Ribadeau-Oumas and R. Grappin

-Optical system of IRMA instruments 1 dQuble be<lm, h~o relis. servo system diffr<lction grating )

-Op tic al s.ystem of MJLJ(O-SCAN 100 instruments ( single beam, one cett, etectrcnrc ratiaing,«ptk al ütters l

Fig. 6. Optical systems of different mid-IR instruments (lOF, 1987).Systèmes optiques des différents instruments opérant dans /'IR moyen (lOF, 1987).

Infraredsource

Detecter

-octtcat system of HUL T1SPEC instrumentsidouble beam, one ceü, ser-ve system,cpfical filtersl

Infraredseurre

""""uChopper

/number of peptide linkages is relativelyconstant. On the other hand, because theNPN fraction is not measured by infra-red, any variation in the proportion ofNPN will influence the accuracy of aninstrument calibrated in crude protein(total N x 6.38).

The presence of ionized carboxylgroup, COo-, gives rise ta an absorptionband at the protein wavelength. The mainindigenous source of such groups iscitrate. Gaudillere and Grappin (1982)and Sjaunja and Anderson (1985) have

Püter wneet

'" _.-::c-~----~"""-~S~ûetecrcr

(hopper

filter

shawn that natural variations in the citratecontent of individual milks explainbetween 40% and 60% of the differencebetween IR and Kjeldahl true proteinresults, and an increase of 0.01% unit ofcitric acid changes the protein readingby + 0.075 g/100 g.

The formation of carboxylic acids byfermentation of lactose may also causeinterference absorption at the proteinwavelength (Goulden, 1964). Similarly, ithas been demonstrated that lipolysisinduces an increase of the protein signal


by 0.01 percent unit per millimole of freefatty acid (Sjaunja, 1982). Grappin andJeunet (1979) have c1earlydemonstratedthat in fact most of the interferingcompounds are present in the milk solublephase (Fig. 7).- Biological factors: any biological factor(e.g., stage of lactation, mastitis, breed,species, feeding, season, etc.), that isknown to influence one of the physico-chemical characteristics mentionedabove, will in its turn cause systematicerrors in protein measurement by IRmethods. According to the review of Biggset al. (1987), only the influence of fewfactors has been demonstrated c1early.Goat milk, which has a lower proportion ofcitrate, requires a different calibration fortrue protein than cow milk (Grappin et al.,1979). Season/feeding, as weil as species(goat vs. cow) or breed (Jersey vs. others)

"

"-: .. ... ........ ..

, ' ,

, ,

-08 -06 -04 -02 002 .04 .06Milk anaiysis IMilko-Sc.n)- (Kj.ld.hl) 1 g/kgl

Fig, 7. Relationship between the differencesMilkoscan protein readings-true proteinKjeldahl values obtained on 81 individual goatmilk samples (x) and on the correspondingwhey samples (y) (Grappin & Jeunet, 1979).Coefficient of correlation r = 0.65.Relation entre les écarts taux de protéinesMilkoscan - taux de protéines vraies Kjeldahlpour 81 échantillons individuels de lait dechèvre (x) et pour les échantillons correspon-dants de lactosérum (y) (Grappin & Jeunet,1979).Coefficient de corrélation: r = 0,65.

383

which influence the proportion of NPN inmilk, will have a significant influence onthe accuracy of the method when theapparatus is calibrated for crude protein(Grappin & Jeunet, 1979; Sjaunja, 1982).Whenever possible, adjustment of theinstrument calibration will therefore benecessary.

Analytical performance

Since the first instrument tested byGoulden in 1964, real progress in theperformance of instruments has beenachieved (Biggs, 1979).

Repeatability

With modern instruments, an excellentrepeatability SO of 0.02 g/100 g, for milkfat, protein and lactose testing is nowcommonly obtained.

Accuracy

Following the definition given p. 372, milkprotein content can be estimated with aSy,x varying from 0.08 9 pour 100 9 forindividual milk to 0.04 g for 100 g for herdmilk.One can assume than with silo milk,accuracy will be closely related ta therepeatability of both the Kjeldahl and theIR methods.

Analysis of proteins by NIR reffection

Rapid measurement of the chemicalcomposition of food products by NIRreflection spectroscopy has been studiedand used mainly in the field of cereals andplant analysis. Despite the fact that theapplication of this technique to milkproduct analysis was investigated byGoulden as early as in 1957, it hasreceived limited attention from scientists.

384

A different approach was followed byRobert et al. (1987), who applied multi-variate analysis (principal componentanalysis and correspondance factorialanalysis). On a population of 38 milksamples from cow, goat, ewe, colostrumand mastitic animais, they found thatwithin the limits at 2 030 nm - 2350 nm,where there is no intense absorption bandfor water, two wavelengths at 2 050 nmand 2 180 nm could be assigned to milkproteins.

ln a one-year study, Launay et al.(1986) have found that, amongst the 13filters which were selected during thestudy by multiple regression analysis, fourwavelengths were constantly used : 2 180nrn, 2 100 nm, 1 820 nm and 1 450 nm.

These bands were also selected byJeunet and Grappin (1985) in a study onindividual milks. Two wavelengths, 2 100nm and 1 820 nrn, which have negativecoefficients, are not specifie to protein.

Table VI gives an example of thewavelengths which have been assignedby different authors for NIR proteintesting.


With the development of reliablecommercial instruments, the dairyindustry is now increasingly using NIRmethods to monitor the quality of dairyproducts : e.g., moisture of milk powder,moisture, protein and fat content of freshcheese and yogurt (Egli & Meyhack,1984).

Principle

Milk has an NIR spectrum which is verysimilar to water, and the determination ofwavebands which are "specific" to milkproteins can be achieved only bymathematical treatment of a collection ofmilk spectra. From the reflectance valuesrecorded by the instrument at the differentwavelengths available (for instance up to19 with the Technicon Infra-Analyzer 450)and the corresponding componentconcentration measured by a referencemethod, the most common way ofselecting wavelengths is to calculate astep-by-step multiple regression using theequation:

C = K1 log (1/R1) + K210g (1/R2) + ...+ Kn log (lIRn) + Ka

where C is the component concentration,K is a constant equivalent to a coefficientof reflectance at the corresponding wave-length, Ko is the bias, 1/R is thereflectance at the corresponding wave-length, and n is the number of selectedwavelengths.

As many wavelengths give redundantinformation, the optimum number ofwavelengths is obtained when theaddition of a new filter does not improvethe residual standard deviation of themultiple regression.

Analysis and instrumentationMost of the research and practical workdone on milk and milk product analysishas been carried out with the Infraalyzerinstrument (Technicon, Tarry Town, NY)and in some cases with PacifieScientific/Neotec, or Dikey-John instru-ments.

Sample preparation requires differentapproaches according to the nature of theproduct. For milk analysis, after homo-genization to limit the light scatteringeffect of fat globules, the sampie is placedinto a temperature-controlled (40 ± 0.1 OC)holder with a quartz window and a


Table VI. Waveband assignment for NIR protein analysis of dairy products.Affectation de longueurs d'onde proche IR pour les dosages des protéines dans les produitslaitiers.

Products Dried milk Liquidmilk Dried milk Caseinand Liquid milkdried cheese

Jeunet andGoulden (1957) Grappin (1985) Baeret al. Frank and Robert et al.

Launay et al. (1983) Birth (1982) (1987)(1986)

1180 1170 (*)1290 (*)

1450-1600 1450 15001730 1730 1700 (*)1820 1820 (*) 18201930 1980

2100 (*) 2050 20502050-2150 2180 2190 2175 2180

22802320 2310

Authors

Wavelength(nm)

ceramic diffuse reflector. Powder samplesare simply placed in a sample holder andpressed against a quartz window. Solidand pasty products, like cheese, shouldhave a uniform surface, and are placed inan open holder.

Wavelengths can be selected either bya holographie grating for research work(e.g., Technicon Infraalyzer 500), or bynarrow band interference filters in routineanalysis. An intermediate solution isoffered by tilting filters with a variableangle of incidence which provides someflexibility in the wavelength setting.

Reflected Iight originating from theirradiated sampie is measured either byIwo lead sulphide detectors placed in anintegrating sphere (Infraalyzer, Technicon)to collect as much as possible of thereflected energy, or by 4 detectors placedat an angle of 45° above the sampie(Pacifie Scientific Instrument).

The incident or reference beamintensity is measured automatically by the

detectors after rotation of the mirror foreach sampie measurement. A ceramicdisk is used as an external reference andthe reflectance values obtained at thedifferent selected wavelenghts are storedand processed to give the componentconcentration.

Analytical attributes and factors affectingthe accuracy of protein testing

Compared to MIR techniques, very littlework has been done to thoroughlyevaluate the physico-chemical andbiological factors which may influence theresponse of NIR analysis. On individualsamples of cow and goat milks, Jeunet &Grappin (1985) have found that Iipolysisdoes not interfere and only speciessignificantly influences the protein results.Conversely to MIR, the accuracy SO (Sy,x)is slightly lower (0.021 vs. 0.025) whenthe instrument is calibrated tor crudeinstead of true protein. Similarly, a betterestimate was obtained by Baer et al.

386

Amino acid composition of milk anddairy products


Table VII. Near infra-red ret/eelanee data for prolein in some dairy products (Frankhuisen &Van der Veen, 1985).Données concernant le dosage des protéines dans certains produits laitiers par réflexion dans leproche IR (Frankhuisen & Van der Veen, 1985).

Product Calibration Prediction

ne r2 SEc rangec np r2 SEp rangepc p

Skim milk powder 159 0.97 0.27 32.9-41.2 19 0.95 0.20 34.0-36.8Butter milk powder 84 0.97 0.21 28.9-34.6 20 0.97 0.21 29.2-34.6Skim milk powder +Buttermilk powder 254 0.98 0.34 28.9-41.2 20 0.97 0.46 30.4-40.6Denalured milk powder 92 0.98 0.34 27.1-40.1 20 0.95 0.42 28.5-39.7Milk powder wilh 61 1.00 0.18 17.6-26.0 10 0.96 0.25 18.4-25.4non-milk falProeessed eheese 40+ 50 0.90 0.31 20.0-24.0 10 0.89 0.35 19.9-23.7

(1983) on non-fat dry milk when theKjeldahl method was used on areference, rather than the dye-binding,melhod. A repeatability SD of 0.023 g/1009 was reported for the Infraalyzer Dairy400 (Jeunet & Grappin, 1985).

Application to dairy products

NIR techniques have been eva1uatedforthe analysis of moisture, fat and proteinmainly for various milk powders (Baer etal., 1983; Frankhuisen & Van der Veen,1985) and cheese (Frankhuisen & Vander Veen, 1985; Frank & Birth, 1982). Anexample of the findings of Frankhuisenand Van der Veen (1985), concerningprotein determination on milk powdersand processed cheese, is given in TableVII.

Direct amino acid analysis after acidhydrolysis can give a good estimation ofthe protein content of any food product ifthe content of peptides and free amineacids ls negligible. This condition can beconsidered as fulfilled for dairy productssuch as industrial caseins and caseinates,whey protein concentrates (WPC{),lactalbumins, coprecipitates, and partia(lyfulfilled for good quality milks, wheys, milkand whey powders. However, it Iscertainly not the best method as far asrapidity (At least 1 h for one sample) andaccuracy are concerned. An indication ofthe accuracy of protein determination bythis technique can be deduced from theresults of inter- and intra-Iaboratory

n: number of sampi es; r2: coefficient of determination; SE: standard deviation of estimate.n, nombre d'échantillons; r2, coefficient de détermination; SE, écart type individual.


analyses performed in 1983 on a singlecasein sample (Sarwar et sl., 1983). Eachlaboratory analyzed the sampie for drymatter and nitrogen by AOAC procedures(1975). Three hydrolyses (6N HCI,performic acid + 6N HCI, and 4.2N NaOH)were carried out in duplicate fordetermination of stable amine acids, sulfuramine acids and tryptophan, respectively.Table VIII shows a good correlationbetween the nitrogen content (Kjeldahl)and the recovery of nitrogen calculatedfrom quantitative analysis of 18 amineacids plus ammonia.

Amino acid composition of a foodproduct is most useful for nutritionists. Itcan sometimes be used for the detectionof adulteration, e.g., detection of addedWPC in non-fat dry milk (Greenberg &Dower, 1986). Therefore, some infor-mation will be given below on the aminoacid analysis of milk and dairy products,although the methodology does not differsignificantly from that used for other foodproducts and has not changed greatly formany years.

Sam pie preparation

Sampling is easy for most dairy products(Iiquids and powders). For cheese it is

387

convenient to freeze-dry and grind a fairlylarge sampie in order to obtain anhomogeneous powder.

Many amine acid analyzers separatethe amine acids from the hydrolyzedsample by ion exchange. This techniquedoes not necessitate very c1eanproducts.ln particular, carbohydrates, even in largeproportion and salts, do not causeexaggerated problems. However, the lipidcontent must be low. Centrifugai skimmingis quite sufficient for Iiquid dairy products,unless rnilk has been homogenized. Inother cases, the products must be driedand ether-extracted. The levels of saltsand Iipids in the samples for analysis byHPLC, after pre-column derivatization,must be negligible.

Although amino acid analysis afterdirect acid hydrolysis can give a fairlyaccurate determination of cysteine +cystine and methionine, provided ailcysteine has been converted to cystine,most people prefer to transform theseamine acids to more stable derivatives.For food products, this is usually achievedby performic oxidation. Part of the sampleis treated with a mixture of formic acid andhydrogen peroxide at low temperatureand the excess of reagents removed byevaporation. This converts cystine and

Table VIII. Recovery of nitrogen from amino acid analysis of casein carried out by 7 laboratories.Rendement en azote estimé à partir d'analyses d'acides aminés d'un échantillon de caséine.

8etween laboratories Within laboratories

Mean % CV% CV%SE

99.4 2.9 2.9 0.8

The protein content (N x 6.38) of the sample was 96.7 ± 0.8 (mean + SEM of the 7 laboratories). SEM, standarderror of the mean; SE, standard error; CV, coefficient of variation (from Sarwar et al., 1983).La teneur en protéines IN x 6,38) de l'échantillon était de 96,7 ± 0,8% (moyenne ± SEM des résultats de 7laboratoires). SEM, erreur standard de la moyenne; SE, erreur stendsra; Cv. coefficient de variation (d'aprèsSarwaret al., 1983).

388

(e.g., 22, 44, 88 h at 110 oC;4, 8 and 12 hat 140 OC).The values for the amino acidsof the former group are extrapolated to atime, while only the longest time is takeninto account for Val and Ile.

For high sensitivities, gaseous 5.7N/HCI is frequently used for proteinhydrolysis because of the lower back-ground obtained and the lower amount ofacid needed (Bidlingmeyer et al., 1984).

Hydrolysis for determination of Trp on,an amine acid analyzer is performed atalkaline pH since this amino acid isdestroyed during acid hydrolysis. Twomethods for the determination of Trp infoods and feeds, including casein, havebeen published recently and the condi-tions for hydrolysis have been evaluated(Sato et el., 1984; Nielsen & Hurtell,1985). The results were similar. About 50mg protein were hydrolyzed- in a pyrextube or a polypropylene liner contained inthe pyrex tube (the 2 methods gavesimilar results) with 4 ml of 4.2 N NaOH.The addition of partially-hydrolyzed starch(0.17 g) improved Trp recovery. One dropof 1- octanol (which could also be used foracid hydrolysis) prevented foaming duringevacuation of the air, which could beperformed with a tap-water vacuum pump.The sealed tube was placed at 110°C for20 h. After cooling, the hydrolysate wasadjusted to pH 4.0 with precooled 6 N HCIand diluted to 20 ml with a pH 4.2-4.3citrate buffer. This solution could beconserved at least 2 weeks at 5 oCwithout further Trp loss. Some destructionof Trp occurred during hydrolysis and thiscould be corrected based up on therecovery of an internai standard. In theinter- and intra-Iaboratory studymentioned above (Sarwar et et., 1983), itwas shown that inter-Iaboratory variationof Trp (CV up to 24 %) was greater thanthat for ail other amino acids (CV up to10%). The values for Trp obtained by one


cysteine into cysteic acid (2 mol/molcystine) and methionine into methioninesulfone (Macdonald, 1985). As thisprocedure totally destroys tryptophan andpartiaily destroys tyrosine and histidine,complete analysis of a sample requires 3different determinations carried out on 3hydrolysates : acid hydrolysates of thenative and oxidized sample, and alkalinehydrolysate for tryptophan determination.

Hydrolysis

Food products are usually hydrolyzed at110°C for 22 or 24 h in the presence of6N HCI (for stable and sulfur amino acids)at a ratio of mg protein/ml acid, varyingbetween 0.4 and 4.0 (Sarwar et al.,1983). The lower ratios are used forcarbohydrate-rich products. Hydrolysis isperformed in tubes or flasks, eithersealed, or evacuated and sealed, orevacuated, nitrogen-flushed and sealed,or under reflux. After hydrolysis, the acidis evaporated and the residue taken withpH 2.2 citrate buffer for conventionalamino acid analyzers. Filtration through a0.45 urn filter is performed. Someanalysts add phenol (50 ~I to 1 ml/100 ml)to the 6 N HCI to reduce tyrosine losses.The authors do not think that this is reallyuseful. Recently, comparison has beenmade between conventional acidhydrolysis, as above, and hydrolysis at145 oCfor 4 h for foods and feeds (Lucas& Sotelo, 1982; Gehrke et al., 1985). Forcasein, results were quite similar, with aslightly higher value for Ile and a lowervalue for Ser and Thr at highertemperature. A slightly higher recovery oftotal amino acids was obtained at 110°C.

ln order to obtain more precise data forsensitive or more difficult-to-releaseamino acids (Thr, Ser, Met, Tyr; Val andIle when bonds involving these 2 residuesoccur), 3 hydrolysis times are selected


laboratory using the colorimetric methodsof Spies and Chamber (1948) werecomparable to those using a methodclose to that described above.

It seems that destruction of Trp duringhydrolysis with HCI is due to the presenceof chlorine in HCI. Hydrolysis of proteinswith sulfonic acids in a reducing medium(e.g., with mercaptoethane sulfonic acid,MESA) is considered to preserve Trp fromdestruction. However, this procedure islittle used; even under these conditions,the presence of carbohydrates affects Trprecovery. Furthermore, MESA is notvolatile.

Amino acid analysis

Several new methods for amine aciddetermination after acid hydrolysis haveappeared during the last 10 years. Theypresent 2 main advantages over theconventional ion exchange method follow-ed by the ninhydrin reaction : somewhatfaster analyses and higher sensitivity.They include, in particular, ion exchangewith new post-column derivatizationreagents and reversed-phase highperformance liquid chromatography (RP-HPLC) with pre-column derivatization.

Conventional techniques

The conventional ion-exchange/ninhydrinprocedure is by far the most employed foramine acid analysis on foods.Improvements include automaticsamplers of various types, narrow boreshorter columns (made of glass or metal)with resin particles of smail diameter (forincreased sensitivity and reduced bufferconsumption, with concomitant higherpressure), more general use of a singlecolumn with stepwise elution (3 or 4different buffers), shorter reaction times

with ninhydrin at a higher temperature(e.g., 120 OC),sophisticated colorimetersand computerization of data collection.

Ion exchange with new reagents for post-column derivatization

Fluorescamine, and 0 -phtaldialdehyde(OPA) +2-mercaptoethanol, react withprimary amine acids only, givingfluorescent derivatives which permit highsensitivity. The latter is now the mostpopular reagent. It provides a hundred-fold increase in sensitivity over ninhydrin.However, the detection of secondaryamine acids (e.g., proline) requires anadditional pump and a post-columnreactor to convert the imino acids intoprimary amine acids with chloramine T orsodium hypochlorite (Bôhlen & Schroeder,1982).

Reversed-phase HPLC (RP-HPLC)

RP-HPLC with pre-column derivatizationwith OPA (Cooper et al., 1984), dansylchloride (Wilkinson, 1978), dabsyl chloride(Vendrel & Aviles, 1986) or phenylisothio-cynate (PITC) are most popular at presentin protein chemistry since they allow highsensitivity (picomolar level) and fastanalyses. OPA, of course, presents herethe same drawback as mentioned above :determination of secondary aminesrequires an oxidation step prior to reactionwith OPA. Furthermore, OPA derivativesof Gly, Lys and hydroxyLys decomposeeasily. The reaction time must be shortand accurately controlled.

The other reagents mentioned above,especially PITC which converts ail aminoacids, including proline, into fairly stablephenylthiocarbamyl (PTC) derivatives, arebeing used increasingly. For pure proteinsarnples, separation of PTC-amino acid

390

methyl chloroformate (FMOC). Theresulting amine acid derivatives are quitestable, unlike PTC-amino acids, anddisplay a high fluorescence which givesextremely high sensitivity. Furthermore, itseems that derivatization is not influencedby the presence of salts in the hydroly-zate.

RP-HPLC, with direct fluorometricdetection, can be used to determinetryptophan in food productsafter alkalinehydrolysis (Nielsen & Hurrell, 1985).


Table IX. Precision of casein amino acid analysis carried out by seven laboratories.Fidélité des analyses d'acides aminés effectuées par 7 laboratoires.

Aminoacid Mean Between laboratories Within Laboratories

g/16g.N SE CV1% CV2% CV%

Arg 3.71 0.34 9.1 8.9 1.8His 2.97 0.18 6.0 5.2 1.6Ile 5.36 0.26 4.8 3.8 1.3Leu 10.16 0.44 4.3 2.5 0.7Lys 8.44 0.42 5.0 4.3 0.6Met 3.02 0.13 4.2 3.1 0.8Cys 0.47 0.08 17.6 17.0 2.6Phe 5.47 0.39 7.1 5.8 1.5Tyr 6.04 0.42 6.9 5.6 1.2Thr 4.64 0.32 7.0 5.4 1.5Trp 1.31 0.19 14.3 14.2 1.1Val 6.85 0.30 4.4 2.6 1.4Ala 3.30 0.19 5.8 4.3 1.8Asx 7.71 0.31 4.0 1.9 0.9Glx 24.00 2.53 10.5 10.1 1.3Gly 2.00 0.13 6.4 5.5 1.3Pro 11.72 0.78 6.6 5.0 1.9Ser 6.10 0.22 3.6 2.1 1.6NH3 1.98 0.19 9.6 10.7 5.9

SE. standard error; CV, coefficient of variation; CV1, between laboratories; CV2, CYs for adjusted arnino acid data(corrected to a total nitrogen recovery of 100% in each laboratory (from Sarwar et al., 1983).SE, écart type; Cv. coefficient de variation; CVI, entre laboratoires; CV2, CV pour les données ajustées(corrigées pour obtenir un rendement d'azote total de 100% dans chaque laboratoire) (d'après Sarwar el al.,(983).

on C18 reversed-phase columns, 3.9 mmx 15 cm, with particles of 3 um andgradient elution, requires ca. 12 min. Thetotal duration of the analysis, includingcolumn washing and equilibration,requires only ca. 20 min (Bidlingmeyer etal., 1984).

Quite recently, the VARIAN company(Hicksville, NY) has developed a newautomatic system for amine acid analysisof protein hydrolyzates by RP-HPLC afterpre-column derivatization with 9-f1uorenyl


Precision of data from amino acidanalysis

As an example, we will give more detailsof an inter-Iaboratory assay, alreadymentioned above, carried out in 1983 byseven North-American laboratories(Sarwar et al., 1983) on 7 food products(casein, egg white, beef, soy isolate,rapeseed concentrate, pea flour andwheat flour), using conventionaltechniques. The results for casein areshown in Table IX. The inter-Iaboratoryvariation (estimated as CV) for ail amineacids except cystine (17.6%), tryptophan(14.3%) and glutamic acid (10.3%) wasless than 10% (3.6-9.1%), whereas, theintra-Iaboratory variation for ail amineacids was lower than 3% (0.6-2.6%).Adjustment of amine acid data to constantnitrogen recovery was reported to improvethe comparability between laboratories.Application of a similar correction to thedata of Table IX shows (CV1 vs. CVbetween laboratories) reduced inter-laboratory variability for ail amine acids,but not for ammonia.

SEPARATE DETERMINATION OFWHOLE éASEIN, WHEV PROTEINS,J3-LACTOGLOBULIN AND NPN IN MILKAND DAIRV PRODUCTS

The Aschaffenburg-Rowland proce-dure

As early as 1938, S.J. Rowland publisheda method for quantitative determination ofdifferent protein fractions, whole casein,albumins, globulins and proteose-peptones in cow's milk. In 1959,

Aschaffenburg and Drewry improved thismethod on the basis of paperelectrophoresis used as an analyticaltechnique, and proposed a procedure toquantitate J3-Lg (an albumin), residualalbumins (mainly a-La), globulins (mainlyimmunoglobulins) and proteose-peptones.lt must be recalled that the latterconstitutes a heterogeneous group ofpeptides or proteins which are not madeinsoluble by heating milk or whey at pH4.6, but are precipitated by 12% TCA.Both methods use several precipitatingagents. Nitrogen is determined in milk andsupernatants by the Kjeldahl technique,the protein contents being obtained byusing the 6.38 coefficient (see above).The Aschaffenburg and Drewry procedurewas examined (Farah, 1979) on the basisof polyacrylamide gel electrophoresis ofthe fractions obtained in the differentsteps, as far as the non-casein-proteins ofmilk were concerned. This study indicatedthat casein precipitation with Na acetateand acetic acid, as performed by Aschaf-fenburg and Drewry, indeed gives asupernatant devoid of caseins. On theother hand, it is weil known that wholecasein, so prepared, provided it is washedat pH 4.7, is only slightlycontaminated by whey proteins (5%). Butthis is not true for human casein (Brignonet sl., 1984). This study also showed thatAschaffenburg and Drewry's fraction "J3-Lgplus NPN", which remains soluble at pH2.0 in the presence of 20 9 Na2S04!'100ml milk, indeed contains J3-Lgas the onlyprotein. As it has been known for a longtime that the NPN fraction contains most ifnot ail the low molecular weightnitrogenous molecules present in milk(amino acids, small peptides, nucleotides,vitamins, urea, creatinine, ammonia etc.),the following fractions can be safelydetermined according to Aschaffenburgand Drewry (1959) with the new recentlydetermined conversion factors (Karman &

presence of dyes or IR radiation, thedirect estimation of the concentration ofeach of these two groups of componentsis not feasible. For their determination,separation of the caseins, either by acidprecipitation or by rennet coagulation,followed by at least two measurements, isrequired.


Van Boekel, 1986; Van Boekel &Ribadeau-Dumas, 1987) :- protein content of milk : (nitrogencontent) x 6.35- content of milk non-protein nitrogenouscompounds : (nitrogen content of the 12%TCA supernatant) x 3.60- whole casein content of milk : (nitrogenof milk minus that of the pH 4.7supernatant) x 6.36- whey protein content of milk : (nitrogencontent of the above supernatant minusthat of the 12% TCA supernatant of milk)x 6.28- ~-Lg content of milk : (nitogen contentof the Na2S04 supernatant, pH 2, of milkminus that of the 12% TCA supernatant) x6.29.

However, according to the data givenby Farah (1979), the other proteinfractions, which could be determined bythe Aschaftenburq-Drewry method,appear to be meaningless.

The detailed procedure has beenclearly described by Aschaffenburg andDrewry (1959), taking into account, for thecalculations, the dilutions, the volume ofthe precipitates and the change of volumecaused by salt addition.

Note that Kjeldahl factors, mentionedabove, give protein contents whichinclude bound phosphate andcarbohydrates. Therefore, the figuresobtained should be slightly higher thanthose deduced from amino acid analysis(except for ~-Lg, which contains neitherphosphate nor carbohydrate).

Determination of casein and wheyproteins by indirect methods

Because in milk, both whole casein andserum proteins behave similarly in

Dye binding methods

- McGann et al. (1972) developed aprocedure for measuring casein in milkafter acid precipitation (acetic acid-sodiumacetate) of the non-casein fraction, andseparation by filtration. Determinationinvolves three measurements : milk, milk+ 1 ml of water, and milk + 1 ml of thenon-casein filtrate. A correction factor of0.04 has to be used to get an accuratecasein reading when using a Pro-milkinstrument calibrated for crude milkprotein analysis. A standard deviation of0.022 g/mg was obtainèd with bulk milksamples by comparison with the Kjeldahlmethod.- A different procedure using rennetcIotting has been developed by severalauthors (Kristoffersen et al., 1974;Mickelsen & Shukri, 1975). This methodinvolves initial separation of the rennetcoagulable casein fraction by filtration orcentrifugation after a small amount ofrennet has been added to the milk,followed by analysis on milk and whey.Whey protein content can be measureddirectly using the same procedure andsame calibration as for milk proteinanalysis. The rennet casein content isthen obtained by subtracting the wheyprotein content from the milk proteincontent; a correction factor might benecessary to obtain the exact rennetcasein value (Kristoffersen etaI., 1974). Intheir experimental work on goat milk(Jeunet & Grappin, 1985), and in other


unpublished experiments, Jeunet andGrappin have used a slightly differentprocedure. The whey protein content isassayed by using a test sample volumeequivalent to three times the volume usedfor milk testing in order to obtain a proteinconcentration in the reaction mixture thatis similar to that of milk.

For cows' milk, the whey true proteincontent (WP) is obtained using theformula: WP = 0.36 x (AB) - 2.10, where(AB) is the Amido Black reading (in g/kg)given by the instrument calibrated for themeasurement of true protein in milk. Toobtain the correct whey protein content inmilk, the test result On whey has to bemultiplied by a factor close to 0.95(Grappin & Jeunet, 1979), which takesinto account the volume of the precipitateobtained during clotting. The rennetcasein content of milk is then obtained bysubtracting the corrected whey trueprotein content from the true proteincontent of milk.

Infra-red methods

For both MIR and NIR techniques, onlythe procedure measuring rennet caseincan be used because with the acidprecipitation technique the presence of alarge arnéunt of acid in the non-cassinfiltrate interferes at the proteinwavelength(s).

To rneasure the protein content inwhey, the IR instrument has to becalibrated by reference to the Kjeldahlmethod. If the 6.46 urn wavelength in MIRis suitable for both milk and wheyanalysis, a specific calibration involvingdifferent wavelengths and differentcoefficients may be necessary with theNIR technique (Jeunet & Grappin, 1985).

ln each case, the rennet caseincontent is obtained by subtracting the

393

whey protein content from the milk proteinCOntent.

Analytical performances

For the Amido Black method and IRtechniques, the repeatability standarddeviation for the measurement of proteinin whey is close to the values found formilk testing. Because of the lower proteinconcentration in whey, Onemight expect ahigher coefficient of variation. Table Xshows examples of the accuracy of themeasurement of whey proteins andrennet casein by Amido Black and IRinstruments, using the Kjeldahl method asa reference.

The excellent accuracy of the IRmethods for measuring rennet casein willbe appreciated when compared to theother methods. One reason might be thatmost of the non-protein componentswhich interfere at the proteinwavelength(s), for both milk and wheyanalysis, are in the soluble phase (seeFig. 7 for MIR) and their influence iseliminated by subtracting the whey proteintest result from the milk protein test result.

SEPARATE DETERMINATION ANDCHARACTERIZATION Of INDIVIDUALPROTEINS IN MILK AND DAIRYPRODUCTS

Since the technological and nutritionalbehaviour of a protein is unique, anincreasing number of people would Iike toobtain quantitative and qualitativeinformation (e.g., the content in thevarious proteins, genetic variants) Onsome or each main protein species infood products. Such information can nOW

394

Electrophoresis


Table X. Accuracy standard deviation (Sy,x in g/100 g) of the determination of milk true protein, wheytrue protein and rennet casein in milk using the Amido Black, Milko-Scan and Infraalyzer techniques.Déviation standard (Sy.x en gl100 g) du dosage des protéines vraies du lait, des protéines vraies dulactosérum et de la caséine présure dans le lait en utilisant le Noir Amido, le Mi/ka-Scan et/'Infraalyzer.

Milk Whey Difference milk-wheyReference Methods

True protein True protein Rennet casein

81 individual Amido Black 0.040 0.018 0.040goat milks 1 Milko-Scan 300 0.035 0.026 0.025

40 individual Amido Black 0.035 0.024 0.028cow milks 2 Milko-Scan 104 0.044 0.032 0.018

20 individual Amido Black 0.043 0.015 0.059cow milks 3 Milko-Scan 104 0.035 0.026 0.024

Infraalyzer 400 0.025 0.023 0.018

1 Grappin & Jeunet, 19792 Michalak & Oczkowicz, 19753 Jeunet & Grappin, 1985

be obtained, at least in sorne laboratories,on raw milk and sometimes on manysimple dairy products, such aspasteurized and sterilized milk, caseins,caseinates, whey and WPC. The situationis more complex in fermented dairyproducts (cheese, yoghurt) in whichpartial degradation of proteins hasoccurred. However, information on theoccurrence and amount of a milk proteincan provide useful information ontechnological treatments and fraudulentadulterations (e.g., mixtures of milks ofdifferent species).

The main techniques that have beenemployed during the last 10 years tocharacterize and determine the mainproteins in milk and sorne dairy productsare electrophoresis, liquid column chro-matography and immunochemicalmethods. For the milk enzymes ofinterest, determinations are usually madeby activity measurements.

Electrophoresis without SOS

This is the most widely used procedure forphenotyping individual cows, and themost rapid one for determining the mainproteins in raw milk.

Dairy species phenotyping

Electrophoresis has been used tophenotype milk proteins from the cow,water buffalo, goat and ewe. We will onlyconsider here what has been done onbovine milk.

Electrophoresis is performed in eitherstarch, polyacrylamide, or polyacrylamide-agarose. The presence of urea and areducing agent (2-mercaptoethanol ordithiothreitol) is required for satisfactory


separation of caseins. The former reagentallows dissociation of intermolecularaggregates while the latter splits inter-molecular disulfide bridges in as2- and K-Cns.

Genetic variants of a protein arise fromheritable modifications involving the genecoding for il. As these are rare events,they lead only to a few modificationswithin a species. The most commonmodifications are point mutations : asingle base in the gene DNA is replacedby another. This may lead to thesubstitution of one amine acid residue byanother in the peptide chain. Thissubstitution can be detected byelectrophoresis if it involves at least oneamino acid carrying an electric charge,that is, in milk proteins, Asp, Glu, P-Ser,P-Thr, Arg, Lys or His. However, Asp andGlu take a negative charge only ab ove thepK of their lateral carboxyl groups, i.e.,above pH 3 to 4, while the positive chargeof His appears below the pK of theimidazole ring NH group, i.e., below pH 6.P-Ser and P-Thr are negatively chargedand Arg and Lys are positively chargedthroughout the entire pH range, which canbe used in electrophoresis. In otherwords, substitutions of P-Ser, P-Thr, Argand Lys by a neutral amino acid can bedetected both at alkaline and acid pHvalues (usually pH 8.6 and 3) onelectrophoresis, while and acid pH mustbe used when His is involved.Unfortunately, only ca. one third of themutations involves charged ami no acidsand there is no simple method fordetecting the other two-thirds. Studies ongenetic variants of milk proteins havereceived renewed interest during the last10 years; indeed, correlations have beenfound between the occurrence of certainvariants and some interesting technolo-gical features of milk.

Among the common variants of milkproteins, a-La B, [)-Lg A and B, aS1-Cn B

and C, aS2-Cn A, [)-Cn A 1, A2 and B, only[)-Cns A 1 and A2, which differ by ahistidine residue, have to be differentiatedby electrophoresis at acid pH.

Methods have been published for thesimultaneous characterization of thevariants of the main whey proteins (a-La,[)-Lg) and caseins (as1' aS2' [) and K) byelectrophoresis at alkaline pH. Unfor-tunately, this does not allow differentiationof [)-Cn A 1 and A2. Furthermore, theelectrophoretic patterns so obtained arenot easily interpretable. Nowadays, mostpeople perform 3 different electrophoreticseparations for each sample to be tested :at alkaline and acid pH values on milk orpH 4.7 -precipitated casein in thepresence of urea and reducing agent, fortyping caseins; at alkaline pH on milk oracid whey, without the latter two reagents,for typing whey proteins.

Electrophoresis is usually run for 4 h(or overnight) at alkaline pH and overnightat acid pH. A large number of sampi escan be analyzed simultaneously. Stainingis performed using Coomassie Blue (R orG) or Amido Black.

Information concerning the chromato-graphie procedures that can be used forphenotyping or quantifying bovine milkproteins, the frequencies of the variousalleles of each main milk protein indifferent breeds and the association ofgenetic variants with some milk charac-teristics can be found in the publicationsby McKenzie (1971), Swaisgood (1982),Grosclaude (1979, 1988), Ng-Kwai-Hanget al. (1984, 1987), Ng-Kwai-Hang andKroeker (1984), MeLe an et al. (1982,1984) and Gripon et al. (1975).

Quantitation of milk pro teins

This has mainly been carried out onindividual fresh bovine milk for genetic


purposes. The techniques are identical tothose mentioned above. However, asdensitometric scanning is required, acryl-amide or acrylamide-agarose gels arepreferred to starch gels, which are nottransparent. Separate electrophoresis atalkaline pH are carried out on pH 4.7-precipitated casein (with urea andreducing agent) and whey. When verticalslab gels are used, para-x-On, whichoften occurs in milk in small amounts,cannot be determined since it migratesupwards at alkaline pH. Amido Black 10 Bor Coomassie Blue G 250 are used tostain proteins. For determination of theabsolute amount of proteins followingscanning densitometry, which gives areasproportion al to the amounts, it isnecessary to know their dye bindingcapacities. These are: 375, 295, 315,425, 440, 430 and 320 mg dye/g proteinwith Amido Black for as-' ~-, x-Ons, ~-Lg,a-La, BSA and Igs, respectively(Tarassuk et al., 1967); as-Cn being thesum as1 + as2' With Coomassie Blue G250, the following relative figures havebeen determined : as1 B = 1.00; as1 C,0.94; aS2' 0.98; ~A1, 0.69; ~A2, 0.66; ~B,0.71; para-x, 0.76 (Mc Lean et al., 1982,1984). In the latter case, whey proteinswere determined by radial immuno-diffusion. Furthermore, as horizontal gelswere used, the concentrations of as1-:as2- and ~-Cns were determined insamples not treated with chymosin (CV =5% expressed as percent of total casein)and those of y- and x-Ons weredetermined in chymosin-treated samples(CV = 14%).

On electrophoresis, aS1-' as2 and K-

Cns give 2, 4, 1 major and several minorbands, respectively. The several K-Cnbands are transformed by chymosin intoone (or two) para-x-On bands. When K-

Cn is determined directly on vertical gels(without chymosin treatment), only themajor band is taken into consideration.

Electrophoresis in the presence of SOS

SOS binds strongly to proteins, mainlythrough hydrophobic interactions. Theamount, which is fixed, is approximatelyproportional to the weight of protein : ca.1.4 9 SOS/g protein (Reynolds & Tanford,1970). Thus, any protein molecule willbind a large number of SOS molecules,each of which carries a negative charge.The indigenous net charge of the proteinat any pH is thus made negligible. Anyprotein should migrate at the samevelocity towards the anode in free flowelectrophoresis in the presence of SOS.However, in zone electrophoresis, inparticular in acrylamide gels, the bigger aprotein, the lower its electrophoreticmobility due to the sieving action of thegel.

This technique is widely used todetermine the Mr of proteins as thelogarithm of Mr is proportional to thedistance of migration (Shapiro et al.,1967). Mr is easily obtained by running anelectrophoregram of the protein(s) ofinterest and a mixture of proteins ofknown molecular weights, and by plottinglog Mr versus distance of migration.

By using polyacrylamide gels with anacrylamide concentration increasinglinearly from cathode to anode, it ispossible to coyer a wider range of Mrsth an with homogeneous gels. In this case,a linear relationship exists between log Mrand log (distance of migration) (Rodbardet el., 1971).

Contrary to what would be expectedfrom their molecular weights, the 4 bovinecaseins can be separated by SDS-PAGE in the presence of a reducing agent(Green & Pastewka, 1976; Creamer &Richardson, 1984; Trieu-Cuot, 1981;Miranda, 1983), giving 4 distinct bandscorresponding, in the order of increasingmobility, to : aS1-' aS2-' ~- and x-Ons,

Milkproteinanalysis

Furthermore, the following other proteins,ail of which have mobilities higher thanthose of the caseins, can also bedistinguished by SDS-PAGE of milk :y 1-Cn, ~-Lg, a-La + para-lC-Cn,y 2- + y 3-Cns in the order of increasing mobilities(Miranda, 1983). To our knowledge, thismethod, which is widely used in ourlaboratory for other purposes, has neverbeen employed for quantitative determi-nation of milk proteins. However, it couldgive, by scanning densitometry of a singlegel, the absolute amounts of the mainproteins present in individual or bulkmilks.

Isoelectric focusing (IEF)

Although IEF was applied to caseinfractions in 1969, the first c1earidentification of the various proteincomponents which can be separated fromwhole casein by this technique, appearedin 1981 in a paper by Trieu-Cuot & Gripon(1981). IEF was performed in 1-mm thickpolyacrylamide gels containing ampho-lytes, 7 M urea and 0.1% 2-mercapto-ethanol. The following components wereidentified, in order of decreasingisoelectric pHs : y -Cns (the 3 knowncomponents), x-On (2 components), as2-Cn (the 4 known components), ~-Cn andas1-Cn (several components, includingasa). IEF was used recently for pheno-typing, in a single run, ail milk proteins inultrathin-Iayer polyacrylamide gels (Sei-bert et el; 1985). The following variantscould be detected : asrCns A, B, C; as2-Cn B; ~-Cns Al, A2, A3, B, C; x-On, A, B,a-La B; and, ~-Lgs A, B, C. Table XI givesthe isoeletric pHs that havebeenmeasured (Trieu-Cuot & Gripon, 1981;Seibert et al., 1985).

The method described by Seibert etal. appears quite promising. It isparticularly short: 15 min prefocusing and

397

30 min focusing. A number of sampiescan be treated simultaneously. Prior tofocusing, defatted milk was diluted 1 : 11with a solution of 8 M urea in distilledwater containing 3% (vlv) 2-mercapto-ethanol.

Two-dimensional electrophoresis

This technique is especially useful forqualitatively analyzing complex mixturesof proteins, by taking advantage of twodifferent criteria simultaneously, generallyisoelectric pH (or electrophoretic mobility)and Mr• It was applied to caseins byTrieu-Cuot & Gripon in 1981. They usedisolectric focusing (pH 4 to 9) in the firstdimension and PAGE (fram 1% to 28%acrylamide) in the presence of 0.1% SDSand 4.9 M urea. Good resolution wasobtained of as1-Cn (2 fractions with sameMrS), as2-Cn (several fractions with sameMrs), ~-Cn, x-On (3 fractions with thesame MrS),y2 and y3 fractions (the sameMrs)andy1.

ln 1983, Miranda showed that two-dimensional electrophoresis of milkproteins (fraction insoluble in 12% TCA)separated the caseins mentioned above,as weil as para-x-On, ~-Lg and a-La. UHTmilk gave a similar pattern. However,x-Cn, .œ-La and ~-Lg produced fainterspots. With sterilized milk, only Us1-and13-Cns. remained visible. A high Mrcomponent, which could not penetrate thegel,appeared. It may correspond to apolyrner produced by Maillard reactionbetween milk proteins and lactose(Andrews & Cheeseman, 1971).

The main interest of two-dimensionalelectrophoresis is its high resolution whichcan be exploited to study complexhydrolysates of milk proteins. Thus,Trieu-Cuot & Gripon (1981); Miranda

; (1983), used it to follow cheese ripening,


Table XI. pHi values of milk proteins from isoelectric focusing.Valeurs des points isoélectriques des protéines du lait obtenues par focalisation isoélectrique.

Protein Interval at probability levelof 5%, after cotrectiotw

Seibertet al. (1985) Trieu-Cuot and Gripon (1981)

lXsl-Cn A 4.16 - 4.40B 4.23 - 4.47 4.44 - 4.76C 4.27 - 4.49

lXs2-Cn A 4.83 - 5.13f3 -Cn Al 4.68 - 4.96

A2 4.60 - 4.84 4.83 - 5.07A3 4.50 - 4.74B 4.78 - 5.10C 4.97 - 5.29

K -Cn A 5.43 - 5.81 5.45 - 5.77B 5.54 - 6.12

œ -La B 4.66 - 4.89f3 -Lg A 4.64 - 4.90

B 4.72 - 4.98C 4.77 - 5.13

Y 1 5.55 - 5.87Y 2 6.38 - 6.72Y 3 6.01 - 6.29

a, correction for pH shift, due to urea, was made by subtracting the pH difference between an ampholyte solutionand the same solution plus urea, ampholytes and urea being atthe same concentration as in the gel.a, correction pour le déplacement du pH du à l'urée. Obtenue en soustrayant la différence de pH observée entreune solution d'ampholytes et la même solution additionnée d'urée, les ampholytes et l'urée étant à la mêmeconcentration que dans le gel.

and proteolyticproteins duringrespectively.

breakdownin vivo

of milkdigestion,

Column chromatography

A review dealing with the chromato-graphie separation of milk proteins waspublished in 1971 by Yaguchi and Rose.At that time, only gel filtration and ionexchange were used in open columns.Nowadays, the tremendous advances inthe technology of chromatographiestationary phases and equipment have

led to an almost complete superseding ofopen columns by HPLC columns in whichstationary phases composed of spherical,small-diameter particles are used at highpressure. Fast analyses at high sensitivitycan thus be obtained. This is especiallyinteresting in the analytical field. Further-more, new procedures, such as reversed-phase chromatography and chromato-focusing have appeared.

Gel Filtration

Direct analysis of skim-milk in phosphatebutter, pH 7.0 at 4 oC on Sephadex G


200, separated caseins into 3 fractions (K+ (Xs+ ~, (Xs> ~, ~ > (Xs)·~-Lg and (X-Lafollowed caseins in this order. Resolutionwas poor (Yagushi & Tarassuk, 1967).Several attempts at separating skim milkproteins on high performance gel filtrationcolumns have been made (Oimenna &Segall, 1981; Shimazaki & Sukegawa,1982; Gupta, 1983). It seems to us thatthe best results for analytical purposeswere obtained by Oimenna & Segall(1981) using a tandem of 2 TSK columnsfrom Toyo-Soda (Biorad, Richmond, CA)(Fig. 8). Gupta (1983) studied thefractionation of skim milk proteins on aTSK 3000 SW column, either in theirnative state, or afier denaturation by SOSand 2-mercaptoethanol. According to thisauthor, the first procedure allowed

4 5

,---;--=r,-r, ---,,1'-,-.,o 10 20 30 40 60

Time Imin.l

Fig.8. FPLC Gel filtrationof skim milk on atandemof 7.5x 300 mm Toyo-Soda3000SWplus 2000 SW columns (Dimenna& Segan,1981).10III injected;flow rate : 0.5 ml/min;eluent :0.05MphosphatebufferpH6.80containing0.1M Na2S04:detectionat280nm.1,Cns;2, IgGs;3, BSA; 4, ~-Lgs;5, (X-La.Noproteinin6-8.Gel filtration (FPLC) de lait écrémé sur 2colonnes Toyo-Soda de 7,5 x 300 mm, 3 000SW et 2 000 S\tY, en série (Dimenna & Segall,1981).Volume injecté : 10 ul; débit : 0,5 ml/min;éluent: tampon phosphate 0,05 M, pH 6,80,contenant Naz S04 0,1 M; détection à 280 nm.1, Cns; 2, IgGs; 3, BSA; 4, f3-Lgs;5, a-La. Pasde protéines dans les fractions 6-8.

quantitation of the individual wheyproteins within 10%. A method of calcu-lation was given for determining, by usingthe second procedure, the concentrationof individual proteins in a mixture ofcasein and whey proteins.

Gel permeation is not suitable foranalyzing skim milk and whole casein. Inmilk, caseins occur as micelles of varioussizes together with sorne soluble casein(aggregates). Removal of calciumdissociates the micelles into large poly-disperse aggregates containing variousproportions of the different caseins.Finally, in the presence of urea andreducing agent, the monomeric caseinsare obtained. Their MrS are too close toeach other and too close to those of sorneof the whey proteins (~-Lg, Igs lightchains) for a satisfactory separation in gelpermeation.

However, whey proteins are quite weilseparated by such a system, which allowstheir quantitation (Oimenna & Segall,1981; Shimazaki & Sukegawa, 1982;Gupta, 1983; Humphrey, 1984; Andrewset al., 1985). In a study by Andrews et al.,(1985) using a Superose 12 column(Pharmacia, Uppsala, Sweden), asatisfactory separation of the main wheyproteins and orotic acid was obtained(Fig. 9). Area measurements led toconcentration values of Igs = 0.51, BSA =0.56, ~-Lg (B + A) = 3.25, (X-La = 1.25mg/ml, in good agreement with literaturevalues.

Ion exchange

No satisfactory result has been obtainedfor fractionating skim milk proteins directlyby ion exchange. However, both wholecasein and whey proteins can beseparated in conditions suitable forquantitative analyses.

400

(Davies & Law, 1977a, 1977b, 1980). Theagreement between duplicate analysesand the average recovery (95.66 ± 1.3%)were both satisfactory, but long runs (ca.20 h) were necessary. Quite similarfractionations were recently obtained fromreduced non-alkylated whole casein byhigh performance anion exchange onDEAE-TSK-5PW (Toyo-Soda, Japan) orMono Q HR SIS (Pharmacia, Uppsala,Sweden) columns, under conditionssimilar to those mentioned above, but withthe addition of a reducing agent to butters,with elution times of ca. 1 h (Humphrey &Newsome, 1984; Visser et al., 1986;Guillou et al., 1987). Limited quantitativedata are given in a paper by Humphreyand Newsome (1984); in the sample ofcasein analyzed, aSl-casein constituted39% (w/w) and 13-casein36% (w/w) of thetotal proteins. Protein recovery deter-mined for these two caseins was > 96%.Fig. 10 shows the separation of anindividual casein sample on a mono Qcolumn in the presence of urea andreducing agent. In this case, sorne geneticvariants were separated (Guillou et el.,1987).

Although it has been known for a longtime that satisfactory separation of wheyproteins can be obtained on DEAE-cellulose, no quantitative data are avail-able on the protein composition of wheydeduced from chromatography on openanion exchange columns. By anionexchange HPLC, with Toyo-Soda DEAE-5PW or Pharmacia Mono Q HR, SIScolumns good separations were obtainedfor the main whey proteins (Humphrey,1984; Andrews et al., 1985; Humphrey &Newsome, 1984; Manji et al., 1985) (Fig.11). Concentrations determined fromchromatography of a sampie of acid wheyon mono Q column were 2.1, 1.3 and 0.8mg/ml for 13-Lg A, 13-Lg B and a-La,respectively (Humphrey & Newsome,1984).


0.12

0.10 3

0.08 4

CJ~

0.06

~<> 0.04

0.02 V A0 J...k ~.10 15 20 25

(mU

Fig.9. FPLCseparationof whey proteins bygel filtration on a Superose 12 column(Andrewset al., 1985).50 J.l1 fresh acid whey injected;flow rate : 0.5ml/min;eluent: 0.1 M Tris-HCI butterpH 7.0containing0.5 M NaCI and 10 mM NaNs.1,Igs;2, BSA;3, ~-Lgs;4, a-La;5, orotieaeid.Gel filtration (FPLC) des protéines dulactosérum sur une colonne Superose 12(Andrews et al., 1985).Volume injecté : 50 pl de lactosérum acidefrais; débit : 0,5 ml/min; éluent : tamponTris-HCI, 0,1 M pH 7,0, contenant NaCI 0,5Met NaN3 10 mM. 1, Igs; 2, BSA; 3, f3-Lgs;4,a-La; 5, acide orotique.

With open columns, caseins wereseparated on anion or cation exchangers(e.g., Mercier et al., 1968; Annan &Manson, 1969) in the presence of ureaand 2-mercaptoethanol. Anion exchan-gers are now by far the most frequentlyemployed. Using DEAE-eellulose, aprocedure was developed by Davies andLaw (1977a) for quantitative analysis ofthe components of reduced and alkylated(iodoacetamide) whole casein. Satis-factory separation of 'Y 2-, 'Y1-, 'Y 3-, x-, 13-,as2- and asl-Cns, eluted in that order bya NaCI gradient in Tris butter, pH 8.6,containing 6 M urea, was obtained. Theprotein content of each casein fractionwas determined by a micro-biuretmethod, from the specific extinction coef-ficients of the corresponding copper-protein complexes. This procedure wasused to determine the proteins mentionedabove in whole casein from a number ofindividual, herd bulk and creamery milks


Chromatography on hydroxyapatite(HA)

This stationary phase is especiallyinteresting for fractionating caseins since

0.043 4

0.32H--,,r ,

,,E "

,c: 0.03 ,

<>,

00 ,,N ,-:;; ,,

1 ,'" 1

c: 0.02 /'2 016 H ';il..'f! z:Ji

.Q...0,01

10 20 30 40 50 60min.

Fig. 10. FPLC separation of individual wholecasein (K NB, [3 C/A', C1.s2A,C1.s,B)by anionexchange on a Mono Q column (Guillou et al.,1987).Casein sample in 5.10-3 M Tris-HCI, 4.5 Murea butter pH 8.0, 8.10-4 M dithiothreitol; flowrate : 1 ml/min; 40 oC; elution with the samesolution as above with a 0-0.32 M NaCIgradient.1, KO-Cn B; 2, KO-Cn A; 3, [3-Cn C; 4, [3-Cn A';S, C1.s2-' C1.s,-, C1.so-Cns.Overloading allowsc1ear visualisation of the K-Cn fractions, butdecreases resolution between C1.s2- (Ieft part ofthe main peak), C1.s'- (main peak), C1.so- (rightpart of main peak) Cns.Séparation (FPLC) d'une caséine entièreindividuelle (1( AlB, {3 C/At, us2 A, aS1 B) paréchange d'anions sur colonne Mono Q(Guillou et al., 1987).Echantillon de caséine dissous dans dutampon Tris-HGI 5.10-3 M, urée 4,5 M, pH8,0, dithiothréitol 8.10'-4M; débit: 1 ml/min;40 °G; élution avec la même solution queprécédemment avec un gradient de 0 à 0,32Men NaCI.1, IdJ-Gn B; 2, IdJ-Cn A; 3, {3-Cn C; 4, {3-CnA 1; 5, uS2-' ast", uso-Gns.La surcharge met clairement en évidence lesfractions de caséine x, mais diminue larésolution entre les caséines us2 (partiegauche du pic principal), as1 (pic principal) etuso(partie droite du pic principal).

it separates them according to theirphosphate content. A number of caseinsamples from herd milks were analyzedon open columns of HA by Barry &Donnelly (1980). Whole casein wasresolved into 5 fractions: y + para-x-Ons,K-Cn + unidentified chymosin-resistantcomponents, unidentified minor protein, 13-Cn, Cls1 + Cls2-Cns. The quantitativecontribution of chymosin-resistant compo-nents to the K-Cn fraction was deducedfrom chromatography of chymosin-treatedwhole casein. The analysis time was ca.10 h. HA-HPLC on a column of Bio-GelHPHT (Biorad, Richmond, CA) gave quitesimilar results with a better separation inca. 45 min. Partial separation of K A and K

20 40Time {min.!

Fig. 11. FLPC separation of acid whey byanion exchange on a Mono Q column(Humphrey & Newsome, 1984). .Acid whey sample; flow rate : 0.5 ml/mm;elution in 0.02 M piperazine butter pH 6.0 with0-0.4 M NaCI gradient. 1, orotic acid; 2, œ-lact-albumin; 3, [3-lactoglobulin B; 4, [3-lactoglobulinA.Séparation (FPLC) de lactosérum acide paréchange d'anions sur une colonne mono Q(Humphrey & Newsome, 1984).Echantillon de lactosérum acide; débit : 0,5ml/min; élution en tampon pipérazine 0,02 M;pH 6,0 avec un gradient de 0 à 0,4 Men N~CI.1, acide orotique; 2, a-Iactalbumme;3, {3-lactoglobulineB; 4, {3-lactoglobulineA.

402

non-polar groups (such as C8 or C18 alkylchains) similarly attached. Unsubstitutedresidual silanol groups are subsequently"end-capped" by groups such as trimethyl-silyl; in theory, RP-HPLC stationaryphases can therefore interact with solutesonly by hydrophobic interactions.

Whole casein and whey proteins wereeach separated by the Chaplin procedure(1986) on a phenyl-Superose HR 5/5 HIcolumn fram Pharmacia. With the former,elution was performed with a 0.8-0.05M,pH 6.0, sodium phosphate gradient,containing 3.75 M urea. For wheyproteins, in a 1.5-0.05M gradient of


B variants was observed. aS2-Cn waspooriy separated from as1-Cn (tailing partof the asrcasein peak), but a differentelution gradient could produce betterseparation (Visser et al., 1986; Kawasakiet al., 1986). The main problem appearsto be the life-time of HA-HPLC columns.

Chromatofoeusing

To our knowledge, this powerful techniquehas been employed only for analyticaland preparative separations of wheyproteins on an open column of PolybufferExchanger 94 (Pharmacia, Uppsala,Sweden). With a pH gradient from 5.2 to4.2, excellent separation of the mainwhey proteins was obtained from acidwhey (Pearce & Shanley, 1981, Fig. 12).ln our opinion, fast and resolutiveseparation could be obtained with theMono P column supplied by Pharmacia.

Hydrophobie interaction and reversed-phase ehromatography (HI-, RP-HPLC)

Although both techniques rely uponhydrophobic interactions between astationary phase and the solutes to befractionated, their applications are quitedifferent. With the former, fixation of thesolutes on the phase is carried out inaqueous solution at high ionic strength,and elution is achieved by lowering theionic strength of the mobile phase. InRP-HPLC, fixation occurs in aqueoussolution of low ionic strength, and elutionis obtained by increasing the hydropho-bicity of the mobile phase (increasing theproportion of a low-polarity solvent). In HI-chromatography, the stationary phaseconsists of a polar material to which non-polar groups (such as phenyl) arecovalently attached, while the phasesused in RP-HPLC are made of silica with

pH

6.0 lê0~....

5.5 -:;~~

5.0.....~~

4.5

4.010 20 30 40 50

fraction number

Fig. 12.Chromatofocusingseparationof acidwhey(Pearce& Shanley,1981).Sampie of milk serum concentrated anddialyzed against starting buffer; columns ofPolybufferExchanger94 equilibratedin 0.025M histidine/HCI buffer pH5.2; 5 ml ofPolybuffer74 (dil.1 : 10 in waterandadjustedto pH 4.2) were run on the columnbeforethesampie was loaded.Elution with Polybuffer/HC!.Flow rate : 0.32 ml/min.1, BSA;2, ~-LgB;3, ~-LgA;4, a-La.Séparation de lactosérum acide parchromatofocalisation (Pearce & Shanley,1981).Echantillon concentré de lactosérumdialysé contre le tampon de départ; colonne.de Polybuffer Exchanger 94 équilibrée ehtampon histidine 0,025 M-HCI, pH 5,2; 5 ml dePolybuffer 74 (dil. 1110dans l'eau et pH ajustéà 4,2) ont été passés sur la colonne avant le.dépôt de l'échantillon. Elution avecPolybuffer/HCI. Débit: 0.32 ml/min. 1, BSA; 2,lJ-LgB; 3, {J-LgA;4, a-La.


ammonium sulfate in 0.05 M sodiumphosphate, pH 7.0, was used. Run timeswere ca. 50 and 30 min, respectively. Inour opinion, resulting the poor resolutionprevents analytical applications.

A few studies have been made on thebehaviour of whole casein (Visser et el.,1986; Caries, 1986) and whey proteins(Pearce, 1983a; Humphrey, 1984) in RP-HPLC. For caseins, Visser et al. (1986)compared 2 columns from Biorad

1.5 1 2 4 100,5 ,,

7 ,6 8 ,,, <Il

,, co1 '") ~10 '"

1

50 ~,,

0.5 ,,,

-' '<1 \J "vJ..,

15 30 45 60Time lmin.!

Fig. 13. RP-HPLC separation of individualwhole casein on a u-Bondapak C 18 column(Caries, 1986).Casein sample : 1.25 mg in 50 III of 10 mM Naphosphate buffer pH 7.2, 1% 2-mercapto-ethanol, 0.1% NaNs; flow rate : 1.5 ml/min;solvent A : 10 mM Na phosphate, 10 mM SDS,pH 7.2; solvent 8 : 2-propanol/solvent A (211;vlv), 10 mM SDS, pH 7.2. 1, NaNs andphosphate; 2, mercaptoethanol; 3, 4, noprotein; 5, as1-casein; 6, 7, (3-caseinvariants(substitution Pro in 6 Leu in 7); 8, as2-casein;9, 10, x-casein.Séparation (RP-HPLC) de caséine entièreindividuelle sur une colonne Ji-Bondapak C18(Caries, 1986).Echantillon : 1,25 mg de caséine dans 50 ulde tampon phosphate de Na 10 mM, pH 7,2,2-mercaptoéthanol 1%, NaN3 0,1%. Débit;1,5 ml/min. Solvant A ; phosphate de Na 10mM, SOS 10 mM, pH 7,2; solvant B ;2-propanol/solvant A (211; vlv), 10 mM SOS,pH 7,2. 1, NaN3 et phosphate; 2, mercapto-éthanol; 3, 4, pas de protéines; 5, caséine cx.sl;6, 7, variants de caséine {3 (substitution de Prodans 6 par leu dans 7); 8, caséine cx.s2; 9, 10,caséine K.

403

(Richmond, CA) : a 25 cm RP-318 (C18-alkylated, silica-based, 33 nm pores, 5 umpartiele size) and a 7.5 cm TSK-PhenylRP (phenyl hydroxylated-polyether, 100nm pore size, 10 urn particle size). Wholecasein sample in a pH 7 buffer solution,containing urea and 2-mercaptoethanol,was left standing for ca. 1 h and theninjected. Solvent A was 0.1% TFA in 10%CHa CN; solvent B was 0.1% TFA in 90%CHaCN. Elution was achieved by a Iineargradient from A to B, with recording at220 nm. The best resolution was obtainedwith the RP-318 columns, the caseinsbeing eluted in the following order : le (3peaks), (Xs2' (Xsl' P + 'Y (several peaks), inca. 30 min. Starting from reduced wholecasein, Caries (1986) (Fig. 13) obtained a

10 15 20 25 30Elution time (min.!

Fig. 14. RP-HPLC separation of defattedcheese whey on Spherisorb C6 column(Pearce, 1983a).Sampie of whole, defatted cheese wheyadjusted to pH 2.1; flow rate: 1 ml/min; solvantA : 0.15 M NaCI/HCI pH 2.1; solvent B : aceto-nitrile. Elution by multistage linear gradientfrom 0 to 48% 8. 1; 8SA; 2, a-La; 3, (3-LgB;4,(3-LgA.Séparation (RP-HPLC) de lactosérum defromagerie dégraissé sur une colonneSpherisorb C6 (Pearce, 1983a).Echantillon de lactosérum de fromageriedégraissé ajusté à pH 2, 1. Débit ; 1 ml/min.Solvant A ; NaCI 0, 15 M/HCI, pH 2, 1; solvantB ; acétonitrile. Elution par gradients linéairesmultiples de 0 à 48% de B. 1, BSA; 2, a-La; 3,{3LgB; 4;,{3-LgA.


good resolution on a 30 cm u-BondapakC18 column at 40 oC, with SOS as acounter-ion. The caseins were eluted inthe following order : aS1' 13, aS2' K. Usingthis procedure, two genetic variants off3-casein, differing by a substitutionPro-Leu, were separated and characte-rized from an individual casein sample.

Figure 14 shows a fractionation ofcheese whey on a C6 column (Pearce,1983a).

Immunologieal techniques

Whey proteins behave like any mono-meric or oligomeric globular proteins :they are highly antigenic and theirdetermination by usual techniques suchas Mancini's double diffusion (1963),rocket-immunoelectrophoresis (Laurell,1966), enzyme-linked immunosorbentassay (ELISA) (Ruitenberg et el; 1976),and radioimmunoassay (Bennet & Mohla,1976) is straightforward. Although quitefeasible, the determination of the differentcaseins by immunological techniques ismore difficult : these proteins are poorantigens and they occur as aggregates(self aggregation or aggregation witheach other).

A few examples of the use of immuno-logical techniques are given below.McLean et al. (1984) used radial immuno-diffusion for the determination of a-La andf3-Lgin a study of the effect of milk proteingenetic variants on milk yield andcomposition. The same technique wasextensively used for assaying Igs(antisera against bovine IgG1 and IgG2are now commercially available) (e.g.,Norman & Hohenboken, 1981). Ig deter-mination is useful for detecting colostrumin milk. LF was determined in milk byradial immunodiffusion, rocket immuno-

electrophoresis (e.g., Welty et el., 1976),and by ELISA in human serum(Hetherington et al., 1983). A radioimmu-noassay was developed for a-La (Beck &Tucker, 1977). f3-Cn was quantified in anumber of individual human milk samplesby rocket immunoelectrophoresis (Chtou-rou et al., 1985). Using the same techni-que and specific antisera against the fourgoat caseins (as1' aS2' 13, K), Grosclaudeet al. (1987), quite recently, found aMendelian polymorphism underlyingquantitative variations of goat aS1-Cn.

Immunological techniques are by farthe most specific tools for determiningproteins, provided they are used carefully.Their use in dairy technology will probablyincrease in the near future.

Determination of some indigenousmilk enzymes

For many years, the determination of milkenzymes has been and is still used toassess milk quality (detection of mastitis,heat treatment evaluation). The readershould refer to Kitchen (1985) for a reviewon milk indigenous enzymes.

As far as the detection of mastitis isconcerned, Kitchen found in 1976 that thisdisease induced an elevated N-acetyl-f3-O-glucosaminidase activity in milk. Rapidand simple diagnostic tests have beendeveloped for routine assay in mastitismonitoring programs. As indicated earlier,the level of catalase in milk varies with thesomatic cell count, and thus a measure ofits activity has been also used to detectmastitis (Kitchen, 1985). This is oftenperformed using a "Catalasemeter". Thisdevice measures the time taken for a tilterpaper disc soaked in the sampie andimmersed in H202 solution to produceenough oxygen to bring the disc to the


surface. The catalase activity is inverselyproportional to this time (Dubois et al.,1982). However, the indications given bythis assay, although correlated to sorneextent to the hyginic quality of milk(microorganisms also have catalaseactivities), are often difficult to interpret :indigenous catalase activity is highlydependent upon individuals, breed,feeding and lactation period. Its level ishigh in colostrum and in late lactationmilk.

The efficiency of pasteurization ismonitored routinely by the AP test whichwas discovered in 1933. Ali milks containbovine AP. Il appears that this enzyme ismore heat-resistant than non-sporeforming pathogens which can be presentin milk. A number of methods formeasuring AP activity in milk and dairyproducts have been published (Richard-son, 1985; Seng Kwee, 1985). Most arebased on the release by the enzyme ofphenol from disodium phenyl phosphateor paranitro phenyl phosphate, orphenolphtalein from phenolphtalein mono-phosphate. Phenol is measuredcolorimetricaliy after reaction with 2,6-dichloroquinone chloroimide, which givesIndophenol Blue. Phenolphtalein isdetected by NaOH addition.

These tests must be performedcarefuliy since they can easily give falseresponses : occurrence of phenol inchemicals and glassware, interferingcoloured materials, "reactivation". Milkscan give negative phosphatase testresults immediately after heating andcooling, or when stored at 4 "C or below.However, the same milks can givepositive test results when not adequatelystored, e.g., at or above 10°C forextended periods. Such "reactivated" APcan be differentiated from residual AP :storage at 34 "C in presence of Mg saltsgives 4 to 10 times more AP activity from

405

reactivation than the same product storedwithout Mg salts (Richardson, 1985).

From the recent work of Griffiths(1986), it appears that LP determinationoffers the most promising method fordetecting heat treatments of the order76 -c for 15 s. This enzyme was formerlydetermined by colorimetric measurementof the reaction products from guiacol oro -dianisidine. The best substrate of LPnow appears to be ABTS (Boehringer,FRG) which can be used as indicated byShindler . and Bardsley (1975). It wasreported that, depending on the tempe-rature of inactivation, sorne LP activitymay be restored after storage. The moresevere the heat treatment, the smalier thedegree of subsequent regeneration.However, storage at 4 oC for 24 h yieldedno restoration of activity foliowing heatingat 65 oCto 80 oCwith a holding time of 15s. LP activity does not vary greatly amongraw milk samples (Griffiths, 1986).

As indicated earlier, plasmin activity ofmilk appears to be beneficial in sornedairy products, detrimental in others. Thismeans that a number of dairy manu-facturers should be interested in simpletests for assessing the plasmin activity inmilk. A rapid, sensitive assay for plasminin dairy products is available (Richardson,1983). However both the substrate (N-succinyl-L Ala-L Phe-L Lys-7-amino-4methyl coumarin), and the required equip-ment (fluorometer), are expensive.Furthermore, as mentioned previously,milk contains plasmin, Pg activator,plasmin inhibitors and, likely Pg-activatorinhibitor. Thus, the plasmin content is ofIittle significance, since Pg, alwayspresent in milk in higher arnounts, cangive plasmin during milk processing. Pgdetermination is quite feasible : plasminassays are performed before and after Pgactivation by urokinase. But the fate of Pgand plasmin during processing is

fairly resistant to such treatments. Asmentioned above, the extent of inactiv-ation of some milk enzymes can indicatethe extent of heat treatments. The effectsof UHT and HTST pasteurization of milkon the properties of its proteins have beenstudied. The solubility of' caseins isreduced by UHT treatment. Whey proteinnitrogen analysis shows significant proteindenaturation. No significant losses innutritive values were found, anddifferences in viscosity and emulsification·capacity were small (Douglas et al.,1981). It has been suggested that residualplasmin activity (this enzyme is fairly heat-resistant) in UHT milk may be one otthefactors of late gelation during storage.


unknown : Pg could be converted intoplasmin, but Pg activator and plasmincould also be inactivated.

SOME APPLICATIONS Of MILKPROTEIN ANALYSIS

A number of such applications have beengiven throughout this review. We willc1assifythese applications here, and addsome that have not been mentionedpreviously.

Milk protein analysis in dairy products

Raw milk

Milk is analyzed routinely in countrieswhere it is paid for, according to itscomposition and hygienic quality. In thiscase, protein determination is especiallyimportant. In addition, the presence ofcolostrum, which may cause problems incheese manufacture, is often sought(detection of Igs by immunologicalmethods). Information on the levels ofsome enzymes, indicating the occurrenceof mastitic milk (N-acetylglucosaminidase,catalase) or potential proteolytic activity(Pg, plasmin) is often useful.

A number of individual milks havebeen, and still are, analyzed to findrelationships between genetic variantsand technological properties.

Processed fluid milk

Heat treatments affect both milk microbialflora and whey proteins, caseins being

Dry milk products

Protein analyses are extensively used forclassification and evaluation of productssuch as milk powders, non-fat dry milks(NFDM), caseins, caseinates and wheyprotein concentrates (WPC). In particular,the extent of heat denaturation of wheyproteins is often measured. Most methodsdetermine the proportion of soluble wheyproteins under various conditions. Wheyproteins are considered to be in theirnative state when they remain solubleafter heat treatments. Harland & Ashworth(1947) developed a turbidimetric methodfor routine estimation of undenaturedwhey proteins in heat-treated milk andNFDM. lt was standardized by Kurarnotoet al. (1959) and modified by Leighton(1962), who introduced the whey proteinindex (WPI) for NFDM. WPI is defined asmilligrams of whey protein nitrogensoluble in saturated NaCI solution pergram of NFDM. Quantitative analysis bySDS-PAGE was recently compared withthese procedures. The described methodappears to be effective for obtaining theratio of casein to whey protein in dairy


products and detecting adulteration ofNFDM (or fluid milk) with WPC (Basch etal., 1985). This type of adulteration canalso be detected fram amine acid analysis(Greenberg & Dower, 1986). Von Kneifeland Ulberth (1985) compared methods forthe evaluation of heat treatment andresidual undenatured whey proteinnitragen content of NFDM, using differentprocedures to precipitate casein anddenatured whey proteins. They found acoefficient of correlation of 0.99 (P < 0.01)between HPLC (gel filtration) and nitrogenanalysis of the soluble fraction. Greineret al. (1985), developed a rapid immuno-turbidimetric method for the screening ofwhey proteins in NFDM and buttermilk,using anti-whole whey serum. Thecoefficient of variation was ca. 4% and theminimum detectable level was 3% wheypratein. A higher sensitivity (detectionlevel : less than 1% whey protein added)was obtained by Olieman & Van denBeden (1983) by HPLC-gel permeationdetermination of the CMP, a peptidereleased from K-casein by rennet (seeabove). However, the latter is more time-consuming.

Cheese and milk protein hydrolyzates

As indicated earlier, electrophoresis andelectrofocusing are convenient methodsfor following the extent of proteolysisduring cheese manufacture. Furthermore,these methods often allow theidentification of the proteinases whichhave been involved during ripening. In astudy of Camembert cheese ripening,Trieu-Cuot & Gripon (1982) showed fromthe identification of several degradationproducts, that rennet, aspartyl- andmetallo-proteinases from P. caseicolum,and plasmin were involved. A number ofmilk protein hydrolyzates (from cheese orother sources) are studied in our

laboratory. Good information is usuallyobtained by treating the sampie in 2%TCA and analyzing the insoluble fractionby electrophoresis or ion-exchange FPLCand the soluble fraction by RP-HPLC.

ln any case, the degree of proteolysiscan be estimated by using TNBS(trinitrobenzene sulfonic acid), a reagentspecific for free NH2 groups, andcolorimetry (Adler-Nissen, 1979).

Detection of bovine milk in milk (orcheese) from other species

Methods are necessary for detectingfraudulent adulteration of human, goat,sheep and water buffalo milks with bovinemilk, which is cheaper. Such methods doexist and are based either on immuno-logical procedures or, more often, onelectrophoresis (e.g., Addeo, 1984;Sanchez et al., 1984).

Detection of non-rnilk proteins in dairyproducts

This is usually possible only when theoccurrence of a known product issuspected. Detection of soybean proteinsin milk products was studied by Hâhnel(1984) using SDS-PAGE, IEF and immu-nodiffusion with antisera to native heat-denatured soya proteins. The highestsensitivity was obtained with the latter.

Milk protein analysis in non-dairyproducts

Sorne food-stuffs (e.g., milk chocolate)contain milk proteins in quantities usuallyabove a certain Iimit, sorne contain none(e.g., sorne meat products). Although

408 B. Ribadeau-Oumas and R. Grappin

electrophoresis can be used in sorneinstances, immunological techniques aremost interesting in both cases. Theirpotential applications and limitations havebeen studied by Willner (1984). It must berealized that the antisera used must beable to recognized denatured (or bothnative and denatured) milk proteins.

CONCLUSION

As far as milk protein analysis isconcerned, the present status and futuretrends can be described briefly :

- The whole protein content of milk andmany dairy products is widely determinedby IR techniques, often with localreference to a dye-binding procedure, andnational or international reference to theKjeldahl method. It seems to us that theDumas method could supersede the latterwhen technical problems, due to watercontent and sam pie heterogeneity, aresolved.

- The content in sorne protein fractions(e.g., casein in cheese milk, whey proteinin WPC) is demanded by many manu-facturers. Asthese fractions cannot beeasily distinguished by their physicalcharacteristics, thelr separation willprobably always be required prior todetermination by IR or dye-bindingmethods. Furthermore, NPN assaysnecessitate nitrogen determination.

- Among the various methods availablefor the characterization and determinationof individual milk proteins, electrophoresisor isoelectric focusing are, no doubt, mostsuitable for routine analysis of a numberof sampi es. However, more care shouldbe taken when they are used for quanti-tative measurements. Immunologicalprocedures will certainly be employed

more frequently in the near future, whengood and inexpensive polyclonal ormonoclonal antibody preparations arecommercially available.

- Column chromatography does notappear to be suitable, as yet, for analysisof individual milk proteins on a large scalebecause of their slowness (ca. 30 min persample). However, automatic equipmentcapable of working day and night isavailable. RP-HPLC is the most powerfultechnique for characterizing peptides andprotein hydrolyzates. Quite recent studiesshow that separation of peptides andproteins by RP- or IE-HPLC can beperformed in seconds or minutes, usingpellicular, non-porous stationary phases(Kalghatgi & Horvath, 1987; Kato et al.,1987).

APPENDIX : ABBREVIATIONS ANDSYMBOLS USED IN THIS ARTICLE

AbABTS

AgAMPAOAC

APBSACMPCnOBCELISAGu.HCIHPHPLC

HTSTlOFIEFIg

AntibodyAmmonium 2,2'-azino-di(3-ethylbenzothiazoline)-6-sulfonateAntigenAdenosine monophosphateAssociation of Official AnalyticalChemistsAlkaline phosphataseBovine serumalbuminCaseino macropeptideCaseinDye binding capacityEnzyme Iinked immunosorbent assayGuanidinium chiorideHorse radish peroxidaseHigh performance liquidchromatographyHigh sterilization short timeInternational Oairy FederationIsoelectric focusingImmunoglobulin


IRISO

Infra-redInternational StandardizationOrganizationœ-LactalburninLactoferrin (Iactotransferrin)Lactoperoxidase(3-LactoglobulinWavelengthMilk fat globule membraneMid-infra-redRelative molecular weightNon-fat dry milkNear-infra-redNon-protein nitrogen, or non-proteinnitrogenous substancesFrequency v = CFA,with C = lightvelocity, llÂ.= wavenumber

OPA OrthophtaldialdehydetoPA Tissue plasminogen activatorPAGE Polyacrylamide gel electrophoresisPg PlasminogenPITC PhenylisothiocyanateR Reproducibility,or 1IR : reflectancer Coefficient of correlationRP-HPLC Reversed phase HPLCSD or S Standard deviationSOS Sodium dodecylsulfateTNBS Trinitrobenzene sulfonic acidWP Whey proteinsWPC Whey protein concentrateWPI Whey protein index

<x-LaLFLP(3-LgÂ

MFGMMIRMrNFDMNIRNPN

v

REFERENCES

Addeo F. (1984) Detection and determination ofbovine milk in Pecorino cheese. Latte 9,37-44Adler-Nissen J. (1979) Determination of thedegree of hydrolysis of food protein hydro-Iyzates by trinitrobenzene sulfonic acid. J.Agric. Food Chem. 27, 1256-1262Alais C., Ribadeau-Dumas B. & Saint-Lebe L.(1961) Etude de la réaction des protéines dulait avec le Noir Amido. Application au dosagerapide des matières azotées du lait. Lait 41,241-273

409

Andrews A.T. & Cheeseman G.C. (1971)Properties of aseptically packed UHT milk :casein modification during storage and studieswith model systems. J. Dairy Res. 38, 193-207Andrews A.T., Taylor M.D. & Owen A.J. (1985)Rapid analysis of bovine milk proteins byHPLC. J. Chromatogr. 348, 177-185Annan W.D. & Manson W. (1969) A fractio-nation of the <Xs-caseincomplex of bovine milk.J. Dairy Res. 36, 259-268AOAC (1975) Official Methods of Analysis(12th ed.) Association of Official AnalyticalChemists, Washington, OCAschaffenburg R. & Drewry J. (1959) Newprocedure for the routine determination of non-casein proteins of milk. XVth Int. Dairy Congr.London,3,1631-1637Ashworth U.S. & Chaudry MA (1962) Dye-binding capacity of milk proteins for AmidoBlack 10B and Orange G. J. Dairy Sei. 45, 952-957Baer R.J., Frank J.F. & Loewenstein M. (1983)Compositional analysis of non-fat dry milk byusing near infrared diffuse reflectance spectro-scopy. J. Assoc. Off. Anal. Chem. 66, 858-863Barry J.G. & Donnelly W.J. (1980) Caseincompositional studies. 1. The composition ofcasein from Friesian herd milks. J. Dairy Sci.47,71-82Basch J.J., Douglas F.w., Procino L.G.,Hoisinger V.H. & Farrell H.M. (1985) Quanti-fication of casein and whey proteins ofprocessed milks and whey protein concen-trates, application of gel electrophoresis andcomparison with Harland-Ashworth procedure.J. Dairy Sci. 68, 23-31Beck N.F.G. & Tucker HA (1977) Relation-ships between radioimmunoassays of alphalactalbumin and prolactin in bovine skim milk.J. Dairy Sci. 60, 542-545Bennet R.M. & Mohla C. (1976) A solid-phaseradioimmunoassay for the measurement oflactoferrin in human plasma : variations withsex, age, and disease. J. Lab. Clin. Med. 88,156-166Bidlingmeyer B.A., Cohen SA & Tarvin T.L.(1984) Rapid analysis of amino acids usingpre-column derivatization. J. Chromatogr. 336,93-104Biggs DA (1979) Performance signification forinfra-red milk analysis. J. Assoc. Off. Anal.Chem. 62,1211-1214


Biggs DA, Johnsson G. & Sjaunja L.O. (1987)Analysis of fat, protein, lactose and total solidsby infra-red absorption. Bull. IDF208, 21-30

Bëhlen P. & Schroeder R. (1982) High-sensi-tivity amino acid analysis : methodology for thedetermination of amino acid composition withless than 100 picomoles of peptides. Anal.Biochem. 126, 144-152Bosset J.O., Blanc B. & Plattner E. (1976) Ledosage des protéines du lait et de sesprincipaux dérivés : méthodes et appareillagestirés de la littérature parue entre 1964 et 1974.Trav. Chim. Alim. Hyg. 67, 226-261Brignon G., Chtourou A. & Ribadeau-Dumas B.(1984) Contamination of human whole caseinby whey proteins. Quantitative aspects. Lait 64,537-543Butler J.E (1974) Immunoglobulins of themammary secretion. In : Lactation, A Compre-hensive Treatise, vol. III (B.L. Larson & V.A.Smith, eds), Academie Press, New York, pp.217-255Caries C. (1986) Fractionation of bovinecaseins by reversed-phase high-performancechromatography : identification of a geneticvariant. J. Dairy Res. 53, 35-41Chaplin L.C. (1986) Hydrophobie interactionfast protein Iiquid chromatography of milkproteins. J. Chromatogr. 363, 329-335Chtourou A., Brignon G. & Ribadeau-Dumas B.(1985) Quantificaton of f3-caseinin human milk.J. Dairy Res. 52, 239-247Clarke T. (1979) Standard reconstituted milksfor Pro-Milk and Kjeldahl tests. Aust. J. DairyTechnol. 34, 121-123Cooper J.D.H., Lewis M.T. & Turnell D.C.(1984) Pre-column O-phtalaldehyde derivati-zation of amino acids and their separationusing reversed-phase high-performance Iiquidchromatography. J. Chromatogr. 285, 490-494Creamer L.K. & Richardson T. (1984) Anoma-lous behaviour of bovine cxsl" and f3-caseinsongel electrophoresis in sodium dodecyl sulfatebuffer. Arch. Biochem. Biophys. 234, 476-486Dalgleish D.G. (1982) The enzymatic coagul-ation of milk. In : Developments in DairyChemistry.l. Proteins (P.F. Fox, ed.), AppliedScience Publishers, London, pp. 157-187Davies D.T.& Law A.J.R. (1977a) An improvedmethod for the quantitative fraetionation ofcasein mixtures using ion-exchange chromate-graphy. J. Dairy Res. 44, 213-221

Davies D.T. & Law A.J.R. (1977b) Thecomposition of whole casein from the milk ofAyreshire cows. J. Dairy Res. 44, 447-454Davies D.T.& Law A.R.J. (1980) The contentand composition of protein in creamery milks insouth-west Scotland. J. Dairy Res. 47, 83-90De Rham O. & Andrews A.T. (1986) The rolesof native milk proteinase and its zyrnoqen :during proteolysis in normal bovine milk. NestléResearch News 1984/85, Nestec Ltd., 191-194Dimenna G.P. & Segall H.J. (1981) High-performance gel-permeation of bovine skimmilk proteins. J. Liq. Chromatogr. 4, 639-649Dolby R.M. (1961) Dye-binding methods forestimation of protein in milk. J. Dairy Res. 28,43-55Douglas F.W.,Greenberg R., Farrell H.M. &Edmonson L.F. (1981) Effect of ultra-hightemperature pasteurization on milk proteins. J.Agric. FoodChem. 29,11-15Driessen F.M.& Van Der Waals C.B. (1978)Inactivation of native milk proteinase by heattreatment. Neth. Milk Dairy J. 32, 245-254Dubois G., Charbonneau R. & Gagnon M.(1982) Dosage automatisé de la catalase dansle lait. Sei. Aliments 2, 245-249Egli H.R. & Meyhack U. (1984) Measurementsof the principal constituents of solid and Iiquidmilk products by means of near infra-redanalyses. In : Challenges to ContemporaryDairy Analytical Techniques. The Royal Societyof Chemistry, London, pp. 103-116Eigel W.N., Butler J.E, Ernstrom CA, FarrellH.M., Harwalkar V.R., Jenness R. & Whitney R.M L. (1984) Nomenclature of proteins of cow'smilk : fifth revision. J. Dairy Sei. 67, 1599-1631Eigel W.N., Hofmann C.J., Chibber B.AK,Tomich J.M., Keenan T.W. & Mertz ET. (1979)Plasmin-mediated proteolysis of casein inbovine milk. Proe. Nat!. Aead. Sei. USA 76,2244-2248Farah Z. (1979) Examination of Aschaffenburgand Drewry procedure for determination of non-casein proteins of milk by discontinuouspolyacrylamide electrophoresis. Z. Lebensm.Unters. Forseh. 168, 394-396Fraenkel-Conrat H. & Cooper M. (1944) Theuse of dyes for the determination of acid andbasic groups in proteins. J. Biol. Chem. 154,239-246


Frank J.F. & Birth G.S. (1982) Application ofnear infra-red reflectance spectroscopy tocheese analysis. J. Dairy Sei. 65, 1110-1116Frankhuizen R. & Van Der Veen N.G. (1985)Determination of major and minor constituentsin milk powder and cheese by near-infra-redreflectance spectroscopy. Neth. Milk Dairy J.39,191-207Gaudillère T. & Grappin R. (1982) Influence del'acide citrique sur le dosage des protéines dulait par la méthode infrarouge. Mémoire de find'études, SPILAB, DouaiGehrke C.W., Wall L.L., Absheer J.S., KaiserF.E. & Zumwalt R.W. (1985) Sampie prepar-ation for chromatography of amino acids : acidhydrolysis of proteins. J. Assoc. Off. Anal.Chem. 68, 811-821Goulden J.D.S. (1957) Diffuse reflexion spectraof dairy products in the near infra-red region.J. Dairy Res. 24, 242-251Goulden J.D.S. (1964) Analysis of milk byinfrared absorption. J. Dairy Res. 31,273-284Grappin R. (1984) Assessment and opti-mization of indirect instrumental methods fortesting major constituents in milk and dairyproducts. In : Challenges to ContemporaryDairyAnalytical Techniques. The Royal Societyof Chemistry, London, pp. 77-90Grappin R. & Horwitz W. (1988) Determinationof nitrogen content in milk by the Kjeldahlmethod using copper sulfate : interlaboratorystudy. J. Assoc. Off. Anal. Chem. 71, 893-897Grappin R. & Jeunet R. (1976a) Essai del'appareil Kjel-Foss utilisé pour le dosaged'azote dans le lait. Rev. Lait Fr. 346, 591-597Grappin R. & Jeunet R. (1976b) Calibrage desméthodes de dosage de routine des protéinesdu lait. Rev. Lait Fr. 347, 663-667Grappin R. & Jeunet R. (1979) Méthode deroutine pour le dosage de la matière grasse etdes protéines du lait de chèvre. Lait 59, 345-360Grappin R., Jeunet R. & Le Doré A. (1979)Determination of the protein content of cows'and goats' milk by dye-binding and infra-redmethods. J. Dairy Sei. 62, (suppl. 1) 38Grappin R., Jeunet R. & Roguinsky M. (1970)Influence de la mammite sur la composition dela matière azotée des laits de quartiers et surles dosages de matière azotée par la méthodeNoir Amido et l'appareil Infra-Red MilkAnalyser. Lait 50, 491-510

411

Grappin R., Packard V.S., Ginn R.E. & MellemaJ. (1980) Precision of the Pro-Milk method inroutine determination of protein in dairy testinglaboratories. J. Food Proto 43, 52-53Green M.R. & Pastewka J.V. (1976) Molecularweights of three mouse milk caseins by sodiumdodecyl sulfate-polyacrylamide gel electro-phoresis and K-Iike characteristics of a fourthcasein. J. Dairy Sei. 59, 1738-1745Greenberg R. & Dower H.J. (1986) Detection ofadded whey protein concentrate in non-fat drymilk by amino acid analysis. J. Agric. FoodChem. 34, 30-32Greiner S.P., Kellen G.J. & Carpenter D.E.(1985) A rapid immunoturbidimetric method forwhey proteins in non-fat dry milk andbuttermilk. J. Food Sei. 50, 1106-1109Griffiths M.S. (1986) Use of milk enzymes asindices of heat treatment. J. Food Proto 49,696-699Gripon J.C., Desmazeaud M.J., Le Bars D. &Bergère J.L. (1975) Etude du rôle desmicroorganismes et des enzymes au cours dela maturation des fromages. Il. Influence de laprésure commerciale. Lait 548, 502-512Grosclaude F. (1979) Polymorphism of milkproteins : sorne biochemical and geneticalaspects. Proc. 18th Int. Conf. Anim. BloodGrps 1, 54-92Grosclaude F. (1988) Le polymorphismegénétique des principales lactoprotéinesbovines. INRA Prod. Anim. 1, 5-17Grosclaude F., Mahé M.F., Brignon G.,Di Stasio L. & Jeunet R. (1987) A Mendelianpolymorphism underlying quantitative variationsof goat uS1-caseins.Genet. Sel. Evol. 19, 399-412Guillou H., Miranda G. & Pelissier J.P. (1987)Analyse quantitative des caséines dans le laitde vache par chromatographie liquide rapided'échange d'ions (FPLC) Lait67, 135-148

Guillou H., Pelissier J.P. & Grappin R. (1986)Méthodes de dosage des protéines du lait devache. Lait66, 143-175

Gupta B.B. (1983) Determination of native anddenatured milk proteins by high-perlormancesize-exclusion chromatography. J. Chroma-togr. 282, 463-475Hâhnel W.C. (1984) The detection of soybeanproteins in milk products. Thesis, Ludwig-Maximilian-Universitat, Munich, FRG


Hamilton TA, Gornicki S.Z. & Sussman H.H.(1979) Alkaline phosphatase from human milk.Comparison with isoenzymes from placentaand Iiver. Biochem. J. 177, 197-201

Harland HA & Ashworth U.S. (1947) A rapidmethod for estimation of whey proteins as anindication of baking quality of nonfat dry-milksolids. Food Res. 12, 247-251

Hetherington S.V., Spitznagel J.K. & Quie P.G.(1983) An ELISA for measurement oflactoferrin. J. Immunol. Methods 65,183-190

Hofmann C.J., Keenan T.W. & Eigel W.N.(1979) Association of plasminogen with bovinemilk fat globule membrane. Int. J. Biochem. 10,909-917

Huebers HA & Finch CA (1987) Thephysiology of transferrin and transferrinreceptors. Physio/. Rev. 67, 520-528

Humphrey R. (1984) Whey Proteins. In :Handbook of HPLC for the Separation ofAmino Acids, Peptides, and Proteins, vol. 2,(W.S. Hancock, ed.), CRC Press, Boca Raton,pp. 471-478

Humphrey R.S. & Newsome L.J. (1984) Highperformance ion-exchange chromatography ofthe major bovine milk proteins. N.Z.J. DairySei. Technol.19, 197-204

IDF (1986) Provisiona/ standard 20 A. Mi/kdetermination of nitrogen content (Kjeldahlmethod) and calculation of crude proteincontent, International Dairy Federations,BrusselsIDF (1987) Monograph on rapid indirectmethods for measurement of the majorcomponents of milk, Bulletin no. 208, Inter-national Dairy Federation, BrusselsIDF (1988) International provisional standard141. Whole mi/k. Determination of mi/k fat,protein and lactose content. Guide for theoperation of mid-infra-red instruments,International Dairy Federation, BrusselsIto O. & Akusawa R. (1983) Purification,crystallization and properties of bovine milkcatalase. J. Dairy Sei. 66, 967-973

Jenness R. (1982) Interspecies comparison ofmilk proteins. In : Deve/opments in DairyChemistry-1-Proteins (P.F. Fox, ed.), AppliedScience Publishers, London, pp. 87-114

Jeunet R. & Grappin R. (1985) Evaluation del'infralyser Dairy pour le dosage des principauxconstituants du lait. Tech.Lait 1003, 53-58

Kalghatgi K. & Horvath C. (1987) Rapidanalysis of proteins and peptides by reversed-phase chromatography. J. Chromatogr. 398,335-339

Kane P.F.(1984) Comparison of HgO andCUS04 as digestion catalysts in manualKjeldahl determination of crude protein inanimal feeds : collaborative study. J. Assoc.Off. Anal. Chem. 67, 869-877

Karman A.H. & Van Boekel MAJ.S. (1986)Evaluation of the Kjeldahl factor for conversionof the nitrogen content of milk and milkproducts to protein content. Neth. Milk Dairy J.40,315-336

Kato Y., Kitamra T., Mitsui A. & Hashimoto T.(1987) High performance ion exchangechromatography of proteins on non-porous ionexchangers. J. Chromatogr. 398, 327-334

Kawasaki T., Ikeda K., Takahashi S. & KubokiY. (1986) Further study of hydroxyapatite high-performance liquid chromatography using bothproteins and nucleic acids, and a newtechnique to increase chromatographicefficiency. Eur. J. Biochem. 155, 249-257

Keenan T.W., Dylewski D.P.,Woodford TA &Ford R.H. (1983) Origin of milk fat globules andthe nature of the milk fat globule membrane.ln : Developments in Dairy Chemistry-2-Lipids(P.F. Fox, ed.), Applied Science Publishers,London, pp. 83-118

Kitchen B.J. (1985) Indigenous milk enzymes.ln : Developments in Dairy Chemistry-3-Lactose and Minor Constituents (P.F.Fox, ed.),Applied Science Publishers, London, pp. 239-279

Kjeldahl J. (1883) En ny Methode tilKvaelstofbestemmelse i organiske Stoffer. Z.Anal. Chem. 22, 366-382

Kneifel W. & Ulberth F. (1985) Zur Methodik derHitzklassifizierung von Magemilchpulver unterBerücksichtigung der HPLC-Analyse. Mi/ch-wissenschaft 40, 265-269

Kristoffersen T., Koo K.H. & Slatter W.L. (1974)Determination of casein by the dye method forestimation of cottage cheese curd yield. Cult.Dairy Prod. J. 9, 12-14

Kühn N.J. (1983) The biosynthesis of lactose.ln : Biochemistry of Lactation (T.B. Mepham,ed.), Elsevier,Amsterdam, pp. 159-176

Kuramoto S., Jenness R., Coulter S.T. & ChoiP. (1959) Standardization of the Harland-


Ashworth test for whey protein nitrogen. J.Dairy Sci. 42, 28-38Lakin AL. (1970) Methods, apparatus : newproduct research, process development anddesign. Chem. Ind. 38,1227-1229Lakin A.L. (1974) A rapid method for estimatingtotal protein in milk. Public. Short Commu-nication. lOF Congr. New Delhi, 303-304

Launay D., Devaux M.F., Bertrand D. & RobertP. (1986) Etude méthodologique de l'analysedu lait par spectroscopie infrarouge et procheinfrarouge. Rapport Adrian, NantesLaurell C.B. (1966) Quantitative estimation ofproteins by electrophoresis in agarose gelscontaining antibodies. Anal. Biochem. 15, 45-52Leighton F.R. (1962) Determination of wheyprotein index of skim milk powder. Aust. J.Dairy. Technol. 17, 186-188Lucas B. & Sotelo A (1982) Amino aciddetermination in pure proteins, foods and feedsusing Iwo different acid hydrolysis methods.Anal. Biochem. 123, 349-356Lunder T.L. (1974) Determination of totalnitrogen in foodstuffs according to Dumas, bymeans of the "Micro Rapid N" automaticanalyzer. Lab. Pract. 23, 170-172MacDonald J.L., Krueger MW. & Keller J.H.(1985) Oxidation and hydrolysis determinationof sulfur amino acids in food and feedingredients : collaborative study. J. Assoc. Off.Anal. Chem. 68, 826-829McGann T.C.A. (1978) Automated physico-chemical methods for the analysis of milk. Areview of major advances (1960-1978) Ir. J.Food Sci. Technol. 2, 141-156McGann T.CA, Mathiassen A & O'ConnellJA (1972) Applications of the Pro-Milk MK II.III. Rapid estimation of casein in milk andprotein in whey. Lab. Pract. 21, 628-650McKenzie HA (1971) Methods for zoneelectrophoresis of milk proteins. In : MilkProteins, vol. Il (HA McKenzie, ed.) AcademiePress, London, pp. 483-508McKenzie HA & Murphy W.H. (1970) Generalmethods and elemental analysis. Determinationof total nitrogen. In : Milk Proteins, vol. 1 (HAMcKenzie, ed.), Academie Press, New York,pp. 154-161

McLean D.M., Graham E.R.B. & McKenzieHA (1982) Estimation of casein componentsby gel electrophoresis.21st Int. Dairy Congr.Moscow l, 221

McLean D.M., Graham E.R.B. & Ponzoni RW.(1984) Effect of milk protein genetic variants onmilk yield and composition. J. Dairy Res. 51,531-546Mailliart P. & Ribadeau-Dumas B. (1987)Procédé pour l'obtention de lactoperoxydaseactive et de lactoferrine à partir de lactosérumet les substances obtenues par ce procédé.French patent 87 4896Mancini G., Vaerman J.P., Carbonera AO. &Heremans J.F. (1963) A single radial-immuno-diffusion method for the immunological quan-titation of proteins. Protides Biol. Fluids Proc.11,370-384Manji B., Hill A., Kakuda Y. & Irvine D.M. (1985)Rapid separation of milk whey proteins byanion exchange chromatography. J. Dairy Sci.68, 3176-3179Marshall K.R. (1982) Industrial isolation of milkproteins : whey proteins. In : Developments inDairy Chemistry-1-Proteins. (P.F.Fox, ed.),Applied Science Publishers, London, pp. 339-373Mercier J.C., Maubois J.L., Poznanski S. &Ribadeau-Dumas B. (1968) Fractionnementpréparatif des caséines de vache et de brebispar chromatographie sur DEAE-cellulose enmilieu urée et 2-mercaptoethanol. Bull. Soc.Chim. Biol. 50, 521-530Metz-Boutigue M.H., Jolies J., Mazurier J.,Schoentgen F., Legrand D., Spik G., MontreuilJ. & Jolies P. (1984) Human lactotransferrin :amino acid sequence and structural cornpa-rison with other transferrins. Eur. J. Biochem.145,659-676Michalak W. & Oczkowicz H. (1975) The appli-cation of a central system of controllingapparatuses for determining the main milkcomponents and its effects on thecomparability of results obtained. Pr. Mater.Zootech. 8,103-112Mickelsen R. & Shukri NA (1975) Measuringcasein by dye-binding. J. Dairy Sei. 58, 311-312Miranda G. (1983) Etude cinétique de laprotéolyse in vivo des lactoprotéines bovines


dans l'estomac du rat : effet du traitementthermique. Thesis, University of Paris VIIMonsan P.F., Thibault PA, Brossard C. &Bruvier C.S. (1985) Procédé d'Extraction desProtéines du Lait, Produits, Appplications duProcédé, et Compositions Pharmaceutiques.French patent FR 25 84 727Muller L.L. (1982) Manufacture of casein,caseinates and coprecipitates. In : Develop-ments in Dairy Chemistry-1-Proteins (P.F.Fox,ed.), Applied Science Publishers, London, pp.315-377Ng-Kwai-Hang K.F., Hayes J.F., Moxley J.E. &Monardes H.G. (1984) Association of geneticvariants of casein and milk serum proteins withmilk fat, and protein production by dairy caille.J. Dairy Sei. 67, 835-840Ng-Kwai-Hang K.F., Hayes J.F., Moxley J.E. &Monardes H.G. (1987) Variation in milk proteinconcentration associated with genetic poly-morphism and environmental factors. J. DairySei. 70, 563-570Ng-Kwai-Hang K.F.& Kroeker E.M. (1984)Rapid separation and quantification of majorcaseins and whey proteins of bovine milk bypolyacrylamide gel electrophoresis. J. DairySei. 67, 3052-3056Nielsen H.K. & Hurrell R.F. (1985) Tryptophandetermination of food proteins by HPLC afteralkaline hydrolysis. J. Sei. Food Agric. 36, 893-907Norman L.M. & Hohenboken W.D. (1981)Genetic differences in concentration of immu-noglobulins G1 and M in serum and colostrumof cows and in serum of neonatal calves. J.Anim. Sei. 53,1465-1472Oielman C. & Van Den Beden J.W. (1983) Asensitive HPLC method of detecting andestimating rennet whey total solids in skim milkpowder. Neth. Milk Dairy J. 37, 27-36Pailler F. (1982) Dosage après minéralisationdes protéines par électrode spécifique de "ionammonium. Comparaison avec la méthodeKjeldahl. Application aux denrées alimentaires.Ann. Falsif. Expert. Chim. Toxicol.75, 431-439Papiz M.Z., Sawyer L., Eliopoulos E.E.,North AC.T., Findlay J.B.C., Sivaprosadaro R.,Jones T.A, Newcomer M.E. & Kraulis P.J.(1986) The structure of 13-lactoglobulinand itssimilarity to plasma retinol-binding protein.Nature 324, 383-385

Pearce R.S. (1983a) Analysis of whey proteinsby high performance Iiquid chromatography.Aust. J. Dairy Technol.38,114-117Pearce R.J. (1983b) Thermal separation of13-lactoglobulin and a-Iactalbumin in bovinecheddar cheese whey. Aust. J. Dairy Technol.38,144-149Pearce R.J. & Shanley R.M. (1981) Analyticaland preparative separation of whey proteins bychromatofocusing. Aust. J. Dairy Technol. 36,110-114Peyrouset A. (1982) Process of extraction oflactoferrin and immunoglobulins of milk.Société Nationale Elf-Aquitaine. UK patentapplication, GB 2 098 998APfeil W. (1987) Is thermally denatured proteinunfolded? The example of a-Iactalbumin.Biochim. Biophys. Acta 911, 114-116Pierre A & Fauquant J. (1986) Principe pour unprocédé industriel de fractionnement desprotéines du lactosérum. Lait 66, 405-419Prieels J.P. & Peiffer R. (1985) Procédé depurification des protéines du lait et de sesdérivés. French patent FR 2 5767 52Pruill K.M. & Tenovuo J.O. (1985) TheLactoperoxidase System : Chemistry andBiological Significance M. Dekker Inc., NewYorkQuan F., Korneluk R.G., Tropak M.B. & GravelRA (1986) Isolation and characterization ofthe human catalase gene. Nucleic Aeids Res.14,5321-5333Reimerdes E.H. (1982) Changes in the proteinsof raw milk during storage. In : Developmentsin Dairy Chemistry-1-Proteins (P.F. Fox, ed.),Applied Science Publishers, London, pp. 271-288Rexroad P.R. & Cathey R.D. (1976) Pollution-reduced Kjeldahl method for crude protein. J.Assoc. Off. Anal. Chem. 59,1213-1217Reynolds JA & Tanford C. (1970) The grossconformation of protein-sodium dodecyl sulfatecomplexes. J. Biol. Chem. 245, 5161-5165Richardson B.C. (1983) Variation of theconcentration of plasmin and plasminogen inbovine milk. NZ.J. Dairy Sci. Technol.18,247-252Richardson G.H. (1985) Standard Methods forthe Examination of Dairy Products.15th ed.,American Public Health Association, WashigtonOC


Robert P.,Bertrand O., Devaux M.F. & GrappinR. (1987) Multivariate analysis applied to near-infra-red spectra of milk. Anal. Chem. 59, 2187-2191

Rodbard O., Kapadia G. & Chrambach A.(1971) Pore gradient electrophoresis. Anal.Biochem. 40, 135-157

Roeper J. & Dolby R.M. (1971) Estimation ofthe protein content of wheys by the AmidoBlack method. N.Z.J. Dairy Sei. Technol.6, 65-68

Rowland S.J. (1988) The determination of thenitrogen distribution in milk. J. Dairy Res. 9, 42-46

Ruitenberg E.J., Steerenberg PA, Brosi B.J.M.& Buys J. (1976) Reliability of the enzymeIinked immunosorbent assay (ELISA) for theserodiagnosis of Trichinella spirales infectionsin conventionally raised pigs. J. Immunol.Methods 10,67-83

Sanchez Ramos M. & Olano A. (1984)Quantification of cow's milk in goat's milk bypolyacrylamide-gel electrophoresis using <Xs1-

casein as internai standard. J. Dairy Sei. 67,(suppl. 1) 72

Sarwar G., Christensen DA, Finlayson A.J.,Friedman M., Hackler L.R., MacKenzie S.L.,Pellett P.L. & Tkachuk R. (1983) Inter- andintra-Iaboratory variation in amino acid analysisof food proteins. J. Food Sei. 48, 526-531

Sato H., Seino T., Kobayashi T., Murai A. &Yugari Y. (1984) Determination of thetryptophan content of feed and feedstuffs byion exchange Iiquid chromatography. Agric.Biol. Chem. 48, 2961-2969

Schaller J., Moser P.W., Dannegger-MullerG.A.K., Rosselet S.J., Kampfer U. & Rickli E.E.(1985) Complete amino acid sequence ofbovine plasminogen. Comparison with humanplasminogen. Eur. J. Biochem. 149, 267-278

Schmidt D.G. (1982) Association of casein andcasein micelle structure. In : Developments inDairy Chemistry-1-Proteins (P.F.Fox, ed.),Applied Science Publishers, London, pp. 61-86

Schober R. & Hetzel H.F. (1956) Uber eineinfache, kolorimetrische Bestimmung vonMilchproteinen. Milchwissenschaft 11, 123-126

Seibert B., Erhardt G. & Senft B. (1985)Procedure for simultaneous phenotyping ofgenetic variants in cow's milk by usingisoelectric focusing. Anim. Blood GroupsBiochem. Genet. 16, 183-191

415

Seng Kwee W. (1985) An improvedAschaffenburg and Mullen alkaline phospha-tase test for coloured milk products. Aust. J.Dairy Technol.40, 27-30

Shapiro A.L., Vinuela E. & Maizel J.V. (1967)Molecular weight estimation of polypeptidechains by electrophoresis in SDS-polyacry-lamide gels. Biochem. Biophys. Res.Commun. 28, 815-820

Sherbon J.w. (1975) Collaborative study of thePro-Milk method for the determination ofprotein in milk. J. Assoc. Off. Anal. Chem. 58,770-776

Sherbon J.W. (1978) Recent developments indetermining protein content of dairy productsby dye-binding. J. Dairy Sei. 61, 1274-1278

Sherbon J.W. & Luke HA (1968) Collaborativestudy of the dye-binding method applied tochocolate milk drinks, cultured buttermilk andhalf-and-half. J. Assoc. Off. Anal. Chem. 51,811-816

Sherbon J'w. & Luke HA (1969) Comparisonof the dye-binding and Kjeldahl methods forprotein analysis of non-fat dry milk and icecream. J. Assoc. Off. Anal. Chem. 52,138-142

Shimazaki K. & Sukegawa K. (1982) Chroma-tographic profiles of bovine milk wheycomponents by gel filtration. J. Dairy Sei. 65,2055-2062

Shindler J.S. & Bardsley W.G. (1975) Steady-state kinetics of lactoperoxidase with ABTs aschromogen. Biochem. Biophys. Res.Commun. 67,1307-1312

Sjaunja L.O. (1982) Studies on milk analysesof individual cow milk samples. Report 56,Swedish University of Agricultural Sciences,UppsalaSjaunja L.O. & Andersson 1. (1985) Laboratoryexperiments with a new IR milk analyzer, theMilko-Scan. Acta Agric. Scand. 85, 345-352

Spies J.R. & Chambers D.C. (1948) Chemicaldetermination of tryptophan. Anal. Chem.20,30-39

Steinsholt AK (1976) Reasons for reducedbinding capacity of Amido Black 10 B in milkproducts, Report 193, Dairy Research 1nstitute,Agricultural University of NorwayStuart 0.1., Acharya K.R., Walker N.P.C., SmithS.G., Lewis M. & Phillips D.C. (1986) (X-

lactalbumin possesses a novel calcium-bindingloop. Nature 324, 84-87


Swaisgood H.E. (1982) Chemistry of milkprotein. In : Developments in Dairy Chemistry-1-Proteins (P.F.Fox, ed.), Applied SciencePublishers, London, pp. 1-59Tarassuk N.P.,Abe N. & Moats W.A. (1967)The dye-binding of milk proteins. Tech. Bull.US Dep. Agric. 1369, Washington, DCTrieu-Cuot P. (1981) Etude des caséines et deleurs produits de dégradation par méthodesélectrophorétiques. Application à l'étude del'affinage du camembert. Thesis, University ofParis VIITrieu-Cuot P. & Gripon J.C. (1981) Electro-focusing and two-dirnensional electrophoresisof bovine caseins. J. Dairy Res. 48, 303-310Trieu-Cuot P. & Gripon J.C. (1982) A study ofproteolysis during Camembert cheese ripeningusing isoelectric focusing and two-dimensionalelectrophoresis. J. Dairy Res. 49, 501-510Udy D.C. (1956) A rapid method for estimatingtotal protein in milk. Nature 178, 314-315Van Boekel MAJ.S. & Ribadeau-Dumas B.(1987) Addendum to the evaluation of theKjeldahl factor for conversion of the nitrogencontent of milk and milk products to proteincontent. Neth. Milk Dairy J. 41, 281-284Van Reusel A. & Klijn C.J. (1987) Automatedmethods for routine analysis of raw milk. Thedye-binding method for determination of theprotein content of milk. Bull. IDF 208, 17-20Vendrel J. & Aviles F. (1986) Complete aminoacid analysis of proteins by dansylderivatization and reversed-phase liquidchromatography. J. Chromatogr. 358, 401-413

Visser S., Slangen K.J. & Rollema H.S. (1986)High performance Iiquid chromatography ofbovine caseins with the application of variousstationary phases. Milchwissenshaft 41, 559-562

Waite R. & Smith G.M. (1972) Measurement ofthe protein content of milk from mastiticquarters by the Amido Black method. J. DairyRes. 39, 195-201

Weiss M.J., Henthorn P.S., Lafferty MA,Siaughter C., Raducha M. & Harris H. (1986)Isolation and characterization of a cDNAencoding a human liverlbone/kidney-type alka-line phosphatase. Proc. Nat!. Acad. Sei. USA83,7182-7186

Welty F.K., Smith K.L. & Schanbacher F.L.(1976) Lactoferrin concentration during invo-lution of the bovine mammary gland. J. DairySei. 59, 224-231

Wilkinson J.M. (1978) The separation of dansylamino acids by reversed-phase high perfor-mance Iiquid chromatography. J. Chromatogr.Sei. 16,547-552

Willner 1. (1984) Potential applications andlimitations of immunological determination ofmilk proteins in foods and animal feeds. Thesis,Technishe Universitât, München, FRG

Yaguchi M. & Rose D. (1971) Chromatographieseparation of milk proteins : a review. J. DairySei. 54,1725-1743

Yaguchi M. & Tarassuk N.P. (1967) Gel filtrationof acid casein and skimmilk on Sephadex. J.Dairy Sei. 50, 1985-1988

Documents

B. Ribadeau-Dumas, R. Grappin To cite this version