Lait (1998) 78, 391-400© InraJElsevier, Paris

391

Original article

Rapid separation of bovine whey proteinsby membrane convective liquid chromatography,

perfusion chromatography, continuous bedchromatography, and capillary electrophoresis

Jean-Michel Girardet-", Franck Saulnier", Guy Linden",Gérard Humbert"

a Laboratoire des biosciences de l'aliment, unité associée à l'Inra, Faculté des sciences,université Henri-Poincaré Nancy 1, BP 239, 54506 Vandœuvre-lès-Nancy cedex, France

b Service commun de séquences des protéines, Faculté des sciences, université Henri-PoincaréNancy l, BP 239, 54506 Yandœuvre-Iès-Nancy cedex, France

(Received 19 September 1997; accepted 26 November 1997)

Abstract - Membrane convective liquid chromatography is a technique based on porous cellulosemembranes designed for the separation of biomolecules in few minutes at high flow-rates and low back-pressures. Bovine whey proteins are separated in less than JO min, at pH 8.5, with a flow-rate of5.6 ml-min"! and with a 0-0.2 mol-L'! NaCllinear gradient. Three other rapid methods are alsoproposed. With the ion-exchange perfusion liquid chromatography based on beads with large poresand with the continuous bed chromatography based on a polymer matrix, separations are achieved inonly JO min. Capillary zone electrophoresis using an untreated fused-silica capillary allows the sepa-ration of whey proteins in a single run of 8 min without the presence of polymerie additives. Theserapid methods are suitable in the quality control of wheys and could be applied in dairy industry orin research. © InraJElsevier, Paris

membrane liquid chromatography / MCLC / perfusion chromatography / continuous bedchromatography / capillary electrophoresis / whey protein

Résumé - Séparation rapide des protéines du lactosérum bovin par chromatographie liquidesur membranes à flux convectif, par chromatographie par perfusion, par chromatographieen lit continu et par électrophorèse capillaire. La chromatographie liquide sur membranes à fluxconvectif est une technologie utilisant des membranes de cellulose poreuses destinées à la séparationdes biomolécules en quelques minutes à des débits élevés et des contre-pressions faibles. Les protéinesdu lactosérum bovin sont bien séparées en moins de JO min à pH 8,5, avec un débit de 5,6 mlzmin "et un gradient linéaire de 0 à 0,2 mol-L:' de NaCI. Trois autres méthodes rapides sont également

* Correspondence and reprints. Jean-MicheI.Girardet@scbiol.u-nancy.fr

392 J.-M. Girardet et al.

proposées. La chromatographie d'échange d'ions par perfusion, utilisant des billes à larges pores, etla chromatographie en lit continu utilisant une matrice polymérisée donnent des séparations bienrésolues en seulement JO minutes. L'électrophorèse capillaire utilisant un capillaire en silice fonduenon traité permet de séparer les protéines lactosériques principales en une seule étape et en 8 minutes,sans adjonction de polymères. Ces méthodes rapides sont idéales pour le contrôle de la qualité des lac-tosérums et pour l'étude de l'hydrolyse ou de l'agrégation des protéines lactosériques. © InralElse-vier, Paris

chromatographie liquide sur membranes / MCLC / chromatographie par perfusion /chromatographie en lit continu / électrophorèse capillaire / protéine de lactosérum

1. INTRODUCTION

Membrane convective liquid chroma-tography (MCLC) is a biomolecule sepa-ration method designed to overcome thelimitations of conventionalliquid chroma-tography or fast protein liquid chromato-graphy (FPLC). Unlike in tradition al beadbased soft gel and HPLC or FPLC columns,mass transport of the molecules being sepa-rated is achieved with cellulose membranesof 1.2 um pore size through convectiverather than diffusive flow [4]. This allowsthe flow to be increased and separations tobe achieved in a few minutes.

Perfusion chromatography was also deve-loped for the rapid separation and purifica-tion of proteins. This technology utilizespacked columns, in which particules havea bidisperse porous structure composed of600-800 nm through-pores transecting theparticules and 50-150 nm diffusive-poresthat line the through-pores [1].

The newly introduced continuous bedtechnology was developed from the pre-vious work of Hjerten et al. [6] and is basedon the polymerization of advanced mono-mers and ionomers directly in the chroma-tographic column. The polymer chainsaggregate into a dense network of nodulesconsisting of micro-particles with an ave-rage diameter of 200 nm. The channels bet-ween the nodules are large enough (7 to15 um diameter) to permit a high hydrody-namic flow. The first successful results inthe development of a new continuous matrix

were obtained with an amphipathic macro-porous gel plug, consisting of a copolymerof acrylic acid and N,N' -methylenebis-acrylamide [6].

On the other hand, capillary zone elec-trophoresis (CZE) is a technique that takesmuch less time and labour, and requiressmaller sample and buffer volumes than gelelectrophoresis and HPLC or FPLC. AsMCLC or perfusion chromatography, CZEis applicable for rapid separation and mea-sure ment of proteins [7, 16].

Rapid and easy analyses for whey pro-teins, 13-lactoglobulin variant A and variantB (I3-LG A and B), œ-lactalbumin (a-LA),bovine serum albumin (BSA), and immu-noglobulins mainly class G (IgG) are ofmajor interest for the dairy industry for thequality control of wheys and the productionof whey protein concentra tes and forresearch on hydrolysis and aggregation ofwhey proteins. Classical chromatographieand electrophoretic techniques have beenwidely used for the isolation of milk pro-teins (for review, see [13]). Recently, rapidseparation and determination of the majorwhey proteins have been performed byreverse-phase perfusion HPLC [14] andCZE [3, 9, 10].

In this paper, a rapid separation of themajor whey proteins was optimized usingan ion-exchange membrane connected to anFPLC system and compared to separationsachieved by ion-exchange perfusion chro-matography, continuous bed chromatogra-phy or CZE. In the case of MCLC, effects of

Rapid separation of whey proteins

pH, flow-rate, mass loading and concentra-tion of NaCI used for protein elution on theretention times of the major whey proteinswere studied.

2. MATERIALS AND METHODS

2.1. Whey protein samples

For MCLC, perfusion chromatography andcontinuous bed chromatography, whey proteinswere prepared from bovine raw skim milk afterisoelectric precipitation of the whole casein and,then, dialysed and freeze-dried. For CZE, wheywas diluted twice with pure water before directinjection onto the capillary. For the four separa-tion techniques used (MCLC, perfusion chro-matography, continuous bed chromatographyand CZE), proteins were identified with ~-LG A,~-LG B, a-LA, and BSA purified by anion-exchange FPLC with a MonoQ HR5/5 column(Pharmacia) according to Andrews et al. [2].

2.2. MCLC

The diethylaminoethyl (DEAE) MemSep1000 (0.5 x 1.9 cm ID; lA mL bed volume) ion-exchange cartridge (Millipore Co., Bedford, MA,USA) was made of cellulose membrane dise fil-ters placed on top of each other in a polypropy-lene housing. Weak anion exchangers (DEAEgroups) were linked on the surface of each mem-brane. The MemSep unit was connected to anFPLC system (Pharmacia-LKB Biotechnology,Uppsala, Sweden) which included a Model GP-250 gradient controller, two Model P-500 pumps,and aV -7 injection valve. The MemSep effluentwas monitored at 280 nm by a Model UV-Idetector and an Omniscribe B-5000 recorder(Houston Instrument, Austin, TE, USA). Theelution buffer was 0.02 mol·L-l Tris, adjusted todifferent pH ranging from 7.5 to 9.5 with HCIand retention of the major whey proteins wasstudied according to the pH. Flow-rate variedfrom lA (one bed volume per min) to7.0 mlzmirr'' (five bed volumes per min) in orderto evaluate the efficiency of the DEAE Mem-Sep 1000 in linear gradient chromatography.Volumes of 100 ul, containing 1, 2, 3, or 4 mg ofwhey proteins and a volume of 200 ul, containing8 mg of proteins were loaded onto the MemSepcartridge in order to study the effect of the massloading on the performance of the MemSep. Four

393

linear gradients (10, 20, 40, and 80 rnmol-L"!Nafll-min'") were tested to examine the effectof salt on protein retention time with a constantgradient volume of 56 mL.

The chromatography conditions varied accor-ding to the parameters studied as follows:- effect of pH: Flow-rate 5.6 ml.unirr '. massloading 3 mg, linear elution gradient20 mmol-L:! NaCl-mirr":- effect of flow-rate: Elution buffer at pH 8.5,mass loading 3 mg, linear elutient gradient20 mrnol-L"! NaCI·min-l;- effect of mass loading: Elution buffer at pH 8.5,flow-rate 5.6 ml.unin ", linear elution gradient20 mmol-L:! Naf.l-rnin"! (in the se conditions,the gradient time was kept constant but not thegradient volume);- effect of linear elution gradient: Elution bufferat pH 8.5, flow-rate 5.6 mL· mirr ', mass loading3mg.

2.3. Perfusion chromatographyand continuons bed chromatography

Perfusion chromatography of the major wheyproteins was achieved with a Poros 20HQ(10 x 1 cm ID; 7.9 mL bed volume) column (Per-Septive Biosystems, Framingham, MA, USA)connected to the FPLC system. A volume of100 fIL containing 3 mg of whey proteins wasadjusted to pH 7 with HCI. The flow-rate was5 ml-min:" and a linear gradient of NaCI ran-ging from 0 to 0.35 mol-L." was applied with agradient volume of 30 mL.

To fractionate the whey proteins by conti-nuous bed chromatography, the anion-exchangeUNO Q-I (3.5 x 0.7 cm ID; 1.3 mL bed volume)column (Bio-Rad Laboratories, Hercules, CA,USA) was connected to the FPLC system. Theconditions were slightly different from thoseused with the Poros 20HQ column. Flow-ratewas 3 ml-min"! and the Iinear gradient was ran-ged from 0 to 0.30 mol-L'" with the same gra-dient volume.

2.4.CZE

Capillary zone electrophoresis was carriedout with an untreated fused-silica capillary(72 cm x 50 um ID; 49 cm from injector to detec-tor) connected to a Model 270A-HT system(Applied Biosystems, Foster City, CA, USA).Diluted whey was introduced for 1 s by hydro-

394 J.-M. Girardet et al.

static injection onto the capillary filled with elec-tro-elution solution of 0.18 mol-L:' phosphoricacid at pH 3.6. Proteins were separated at 37 "Cat a constant potential difference of 22 kV anddetection was monitored at 214 nm. Data wereprocessed with the Model 600 Data AnalysisSystem version 1.0.2 software (Applied Biosys-tems).

3. RESULTS AND DISCUSSION

Bovine acid whey contains ~-LG A and B(3.2 g-L"), a-LA (1.2 gL:'), IgG (0.75 g-L"),and BSA (004 gL-l) as major proteins andsmall amounts of more than 30 other pro-teins and peptides also called proteose pep-tones [9, Il].

In the present work, rapid separation ofwhey proteins was performed by three chro-matography methods (MCLC, perfusionchromatography and continuous bed chro-matography) and by capillary zone electro-phoresis (CZE). With the three chromato-graphy techniques, the flow-rate could easilyincrease without loss of resolution causedby small pore diffusion typical of conven-tional porous beads. In a classical bead basedcolumn, the molecules (whey proteins) tra-vel only by diffusion from the inters titi alfluid, i.e., the volume of buffer betweenbeads, to the adsorption sites located mainlyinside the porous beads. Thus, flow-ratesare slow (l ml.imin"! with the MonoQcolumn; [2]). Resistance to diffusive trans-port has been minimized with chromato-graphie methods based on large porositymaterials, where the mass transfer is mainlyachieved through convective flow.

The competition adsorption of a-LA andBSA to a sulfopropyl ion-exchange mem-brane (SP MemSep 1010) was studied [17].Indeed, these ion-exchange membranes qua-litatively display local-equilibrium beha-viour and are an useful tool for examiningcompetitive protein sorption behaviour.

In our study, the major whey proteinswere eluted with the DEAE MemSep 1000cartridge using smaller salt concentrations

than those required with the MonoQ column.With classical bead-based columns, modi-fications of salt concentration performed inthe interstitial fluid need sorne time to beeffective inside the porous beads where 90 %of the adsorbed proteins are localized. Withmembrane-based cartridges, however, thesalt concentration remains the same in thebulk fluid and inside the pores due torninirnization of the diffusion phenomenon.Under convective flow conditions, elution ofproteins does not require mobile phases asconcentrated in sodium chloride as thoseused in bead-based chromatography tech-niques.

On the MCLC profiles of whey proteins,IgG were not or weakly retained onto theMemSep cartridge at pH 8.5. The elutionorder remained the same as that observedwith bead-based colurnns for a-LA, ~-LG Band A. BSA, however, was eluted last andcorresponded to a broad rninor peak. In clas-sical anion-exchange FPLC, IgG are weaklyadsorbed onto the MonoQ column [2] andtheir retention capability increases with pH[5]. BSA is co-eluted with a-LA at pH 6.3and co-eluted with ~-LG B at pH 8.0. Atneutral pH, BSA is identified by an indivi-dual broad peak between the two peaks of a-LA and ~-LG B [5]. The other geneticvariant of ~-LG is eluted at the end of thelinear gradient.

Study of the effect of pH showed a wea-ker adsorption of proteins onto the Mem-Sep cartridge than onto the MonoQ columnsince a-LA was not retained at pH valueslower than 8.0-8.5 (figure 1). To fractionatethe major whey proteins, more basic condi-tions (pH ~ 8.5 instead of pH = 7.0 for theMonoQ column) were needed with theMemSep cartridge. We have observed asimilar phenomenon with a classical beadbased column called Bio-Scale Q (Bio-RadLaboratories), where a-LA was not retai-ned at a pH lower than 8.0 (result notshown). According to the manufacturer, thematrix of this strong anion-exchanger ismore hydrophilic than the Monobeads ™

Rapid separation of whey proteins 395

7 7.5 8 8.5 9 9.5 10

pH

Figure 1. Variation of retention times of œ-Iactalbumin (0), ~-lactoglobulin A (0), and ~-lacto-globulin B (e) as a function of pH. A volume of 100 JlL containing 3 mg of whey proteins wasinjected onto a diethylaminoethyl MemSep 1000 cartridge and separation was performed using a1inear elution gradient of 20 mmol-L:' NaCl·min-1 in 20 mmol-L:' Tris buffer at various pH valuesfor 10 min with a flow-rate of 5.6 mls-mirr".Figure 1. Évolution des temps de rétention de l'a-lactalbumine (0), de la ~-lactoglobuline A (0) etde la ~-lactoglobuline B (e) en fonction du pH. Un volume de 100 JlL contenant 3 mg de protéineslactosériques a été injecté dans une cartouche MemSep 1 000 porteuse de groupements diéthylami-noéthyles. La séparation a été réalisée avec un gradient linéaire de 20 mmol-L:' NaCl·min-1 en tam-pon Tris 20 mmol-L:' à différents pH pendant 10 minutes, avec un débit de 5,6 mL·min-l.

constituting the MonoQ stationary phase.For a given pH value, the adsorption of a-LAwas dependent of the hydrophilic or hydro-phobie nature of the matrix of the column. Inthe case of the MemSep 1000, the cellulosemembranes seemed to be more hydrophilicthan the Monobeads' and was not able toretain a-LA at a pH lower than 8.5.

An increasing linear gradient slope cor-responded to decreasing retention times forthe three principal proteins (figure 2). Curvesof the retention times of the different pro-teins versus the linear gradient slopeapproximated to hyperbolic functions,except in the case of a-LA which separa-ted with a linear gradient slope lower than20 mmol-L:' NaCl-mirr '. The behaviour ofa-LA toward the MemSep was complex. Infact, when the linear gradient slope wasincreased from 10 to 20 mmol-L -1 NaCl-mirr'',this protein was adsorbed onto the mem-brane more strongly. With a slope of10 mmol-L:' Nafll-mirr '. only a few of the

protein's interaction sites interacted effec-tively with the DEAE ligand due to a spe-cifie folding of the protein in the presence ofa small salt concentrations. Conforma-tional changes were possible when salt wasadded to the solution and adsorption of a-LA improved. Indeed, it was reported thataddition of ions induces the importantconformation al changes of a-LA observedby fluorescence studies [8]. Thus, due to theparticular behaviour of a-LA on the Mem-Sep membranes, i.e., a weak affinity for thecellulose matrix and an ionie strength-sen-sible spatial conformation, good separationsbetween a-LA and p-LG were obtained withlinear gradients of la or 20 mrnol- L-INaf'l-mirr '.

After pH and linear gradient optimization,influence of the flow-rate was tested(figure 3). The difference in retention timesof a-LA and p-LG was improved by increa-sing the flow-rate from lA ml-min"! (onebead volume) to 7.0 ml-rnirr" (five bed

396 J.-M. Girardet et al.

12

10'2]QI 8S

:;:l~0 60:g~QI~ 4

20 20 40 60 80 100Linear gradient slope (rnmol VI NaCl min-I)

Figure 2. Variation of retention times of œ-lactalbumin (0), ~-Iactoglobulin A (0), and ~-Iacto-globulin B (e) as a function of the linear elution gradient. A volume of 100 JlL containing 3 mg ofwhey proteins was injected onto a diethylaminoethyl MemSep 1000 cartridge and separation was per-formed using various Iinear elution gradients of NaCI in 20 mrnol-L"! Tris buffer at pH 8.5 forJO min with a flow-rate of 5.6 ml.vmirr '.Figure 2. Évolution des temps de rétention de l'a-lactalbumine (0), de la ~-Iactoglobuline A (0), etde la ~-Iactoglobuline B (e) en fonction du gradient linéaire d'élution. Un volume de 100 JlL conte-nant 3 mg de protéines lactosériques a été injecté dans une cartouche MemSep 1000 porteuse degroupements diéthylaminoéthyles. La séparation a été réalisée avec différents gradients linéaires deNaCI en tampon Tris 20 mmol-L:' à pH 8,5 pendant 10 minutes, avec un débit de 5,6 rnlr-rnirr".

volumes). In our conditions, resolution couldalso be slightly improved by changing thegradient time but not the gradient volume.Efficiency of the MemSep 1000 cartridgewas determined by reporting the plot of thetheoretical plate height (h) versus the flow-rate (figure 3; [15]). The factor h is definedby the relation h = Un, where L is the heightof the stack of the cross-linked cellulosemembranes and n the theoretical plate num-ber calculated from the different peaks ofa-LA obtained at each flow-rate tested. Theplate height was on the order of 8 um andwas relatively insensitive to flow-rate. Thus,dispersive contributions from fluid flow irre-gularities were minimal in the MemSep1000 cartridge under our conditions of flow-rate of 1.4-7.0 ml.smin ". Similarly, theplate heights determined for human trans-ferrin and cytochrome c were constant underoperating conditions of 1-10 ml-min"! [4].A flow-rate of 5.6 ml-rnirr ' offered, howe-ver, a good compromise between a short

run time and a high level of back pressuregenerated on the MemSep cartridge. Theback pressure does not exceed 7 bar. In thecase of the DEAE MemSep 1000 cartridge,a flow-rate range of 2 to 6 ml.cnin " isrecommended by the manufacturer.

The mass loading could be increased from1 to 8 mg without significantly changing theresolution. The retention times were, howe-ver, increased by about 0.5-1 min (figure 4).A slight increase of the back pressure wasobserved during runs where the loaded pro-tein mass was increased from 2 to 4 mg. Thiswas probably the result of a local increasein the mobile phase viscosity inside theMemSep cartridge that extended the retentiontimes of each whey protein.

Finally, a satisfactory separation of eachmajor proteins could be achieved in a shorttime « 10 min) at pH 8.5, with a flow-rateof 5.6 ml-mirr ' and with a 20 mmol-L:'N eCl-rnirr ' linear gradient for 10 min(figure 5a). The recent work of Splitt et al.

Rapid separation of whey proteins 397

Flow-rate (mL min-I)

lO'~_9/g 8.<: 7

6o 2 4 6 8

Flow-rate(mLmin-1)

10

9

~8

" 7S+ll::: 60+ll:::

" 5...,"~

4

30 2 4 6 8

Figure 3. Variation of retention times of œ-lactalbumin (0), ~-Iactoglobulin A (0), and ~-Iacto-globulin B (e) as a function of the flow-rate. The inset shows the Van Deemter plot [15] represen-ting the theoretical plate height (h) calculated for œ-Iactalbumin as a function of the flow-rate. Avolume of 100 ul, containing 3 mg of whey proteins was injected onto a diethylaminoethyl MemSep1000 cartridge and separation was performed using a linear elution gradient of 20 mmol-Lr' NaCI·min-!in 20 mrnol-L:' Tris buffer at pH 8.5 during 10 min with various flow-rates.Figure 3. Évolution des temps de rétention de l'n-lactalbumine (0), de la ~-Iactoglobuline A (0), etde la ~-Iactoglobuline B (e) en fonction du débit. L'encart montre la courbe de Van Deemter [15] repré-sentant la hauteur de plateau théorique (h) calculée pour l'œ-lactalbumine en fonction du débit. Unvolume de 100 ul, contenant 3 mg de protéines lactosériques a été injecté dans une cartouche Mem-Sep 1000 porteuse de groupements diéthylaminoéthyles. La séparation a été réalisée avec un gradientlinéaire de 20 mmol-L:' Naûlmirr ' en tampon Tris 20 rnrnol-Lr! à pH 8,5 pendant 10 min, avecdes débits variables.

[12] reported the preparation of major wheyproteins from whey and permeate obtainedduring lactose production in modem dairiesusing Sartorius MA Q 15 ion-exchangemembranes. The authors obtained a partialresolution of the two variants A and B of~-LG. The resolution can be, however,improved by coupling two MA Q15 unitsin series. Moreover, separation of BSA isachieved using these two consecutivemodules with local isocratic elution condi-tions. Separation of whey proteins is thusachieved by judicious module coupling toge-ther with a fine tuned gradient.

MCLC was compared with ion-exchangeperfusion chromatography, continuous bedchromatography, and CZE. With thePoros 20HQ and UNO Q-l columns, masstransport was achieved through convectiveflow. The elution order and resolution ofthe different whey proteins were similar to

those obtained with a classical MonoQcolumn (figure 5b, c). Unlike the MemSep1000 cartridge, the two matrixes constitu-ting Poros 20HQ and UNO Q-l, have abehaviour similar to that of the MonoQanion-exchanger, and whey proteins, espe-cially a-LA, were retained onto the columnseven at a pH lower than 8.0. Separationswere achieved in 8 to 10 min (at high flow-rates) instead of 20-25 min in the case ofconvention al FPLC. A perfusion reversed-phase HPLC method was recently developedby Torre et al. [14] for the rapid separationof the major whey proteins in less than 3 minwith a Porosl lOR (5 x 0.21 cm ID) column,with a bed volume of 0.17 mL instead ofthe 7.9 mL of our Poros 20HQ column.However, perfusion ion-exchange chroma-tography is more suitable for preparativefractionations than the reverse-phase methoddue to its lower protein denaturation effect.

398 J.-M. Girardet et al.

8

~ :7ê:gQ) 6S~c0 5~.sQ)

~

I:l:: 4

3

0 2 4 6 8 10Loaded protein (mg)

Figure 4. Variation of retenti on times of œ-lactalburnin (0), ~-Iactoglobulin A (0), and ~-Iacto-globulin B (e) as a function of the mass of proteins loaded onto the MemSep cartridge. A volume of100 ul, containing various masses of whey proteins was injected onto a diethylaminoethyl MemSep1000 cartridge and separation was performed using a linear elution gradient of 20 mmol-L:' NaCI·min-!in 20 rnmol-L:' Tris buffer at pH 8.5 during 10 min with a flow-rate of 5.6 rnl.i-mirr". To load 8 mgof proteins, an injection volume of 200 mL was required.Figure 4. Évolution des temps de rétention de l'a-lactalbumine (0), de la ~-Iactoglobuline A (0), etde la ~-Iactoglobuline B (e) en fonction de la quantité de protéines chargée dans la cartouche Mem-Sep. Un volume de 100 ul, contenant des quantités variables de protéines lactosériques a été injectédans une cartouche MemSep 1000 porteuse de groupements diéthylaminoéthyles. La séparation a étéréalisée avec un gradient linéaire de 20 mmol·L-1 NaCI rnirr ' en tampon Tris 20 mmol-Lr! à pH 8,5pendant JO minutes, avec un débit de 5,6 ml.imirr '. Pour déposer 8 mg de protéines, un volumed'injection de 200 ul, a été nécessaire.

The most satisfactory resolution was obtai-ned with continuous bed chromatography.According to Hjerten et al. [6], the reasonsfor this are that a gel plug has more homo-geneous structure than a packed bed of beads,e.g., Porous 20HQ, and that the gel plug wascompressed, which has a favourable effecton the resolution. It is also likely that thecontinuous bed is non-porous, i.e., the 'walls'of the channels in the gel are impermeableto proteins, which in combination with com-pression of the bed gives good resolutionsindependently of flow-rate [6]. Thus, conti-nuous bed chromatography combines theadvantages of cellulose membrane basedchromatography, i.e., the ease ofuse and theefficient resolutions independent of flow-rate, and the advantages of bead based per-fusion chromatography, i.e., the high adsorp-tion capability of the polymer matrix.

CZE was utilized differently for the sepa-ration of whey proteins with an untreated

fused-silica capillary. OUe et al. [9] showthat, in phosphate buffers at pH 7.0 or moreand with a potential difference of 22 kV,these conditions are not suitable for the sepa-ration of BSA and /3-LG, although separa-tion of the two genetic variants of /3-LG isachieved. According to these authors, BSAand /3-LG are well-separated at low pHvalues (pH = 2.5) but not the A and Bvariants which co-elute. Injection of stan-dard proteins shows that IgG are electro-eluted last in a broad peak at an inje ctor todetector migration time on min. Under theconditions used by Recio et al. [10], a-LA,BSA, and /3-LG B and A are well-separatedin borate buffer at pH 8.2 with a run time of12 min and a potential difference of 7 kV.In our conditions, at pH 3.6 and with a poten-tial difference of 22 kV, the electro-elutionorder we determined was BSA, /3-LG A, /3-LG B, a-LA and IgG (figure 6). IgG wasnot easily detectable when whey was directly

Rapid separation of whey proteins

(X-LAo.e

~0.,..;;~~=..of0

'"~

0

0 4 8 12Retention time (min)

(c)

(X·LA

0.5

II-LGAIgG

0II-LGB::l

'il8 '" r ;.= '"..

. ~ '0

~ '" !"'B8A ë3

0g

0 4 12Retention time (min)

399

(a) 0.5 (X·LA (b)

~II-LGA0ee.. IgG

]~;;Il U~J/

~~ â '0'0 of !! ~

~ §§ ..~ ~

0

0 4 8 12Retention lime (min)

Figure 5. Separation of the major whey proteins,immunoglobulins c1ass G (lgG), œ-lactalbumin (a-LA), p-Iactoglobulins B and A (P-LG B and P-LG A,respectively), and bovine serum albumin (BSA) by(a) membrane convective Iiquid chromatography(MCLC), (b) ion-exchange perfusion Iiquid chroma-tography, and (c) continuous bed chromatography.For details, see the Materials and Methods section.Figure 5. Séparation des protéines lactosériques prin-cipales, immunoglobulines de classe G (IgG), a-lac-talbumine (a-LA), p-Iactoglobulines B et A (P-LG Bet P-LG A, respectivement), et sérumalbumine bovine(BSA) (a) par chromatographie liquide sur mem-branes à flux convectif (MCLC), (b) par chromato-graphie d'échange d'ions par perfusion et (c) par chro-matographie en lit continu. Pour plus de détails, sereporter à la section « Matériel et méthodes »,

injected onto the capillary. Separation ofBSA, p-LG A and p-LG B was achieved ina single run of 8 min. In another work,Cifuentes et al. [3] separated p-LG A+B,a-LA and BSA by protein size in 10 minusing a capillary of a 20-cm effective length.The separation by size is carried out at 16 kVwith a Tris-borate buffer at pH 8.6 contai-ning polymerie additives (sodium dodecylsulfate and polyethylene glycol 8 000). Theseparation of the two genetic variants ofp-LG is, however, not achieved.

4. CONCLUSION

MCLC, perfusion chromatography, conti-nuous bed chromatography, and CZE arefour rapid suitable methods applicable tothe separation and determination of the

major whey proteins and could be applied inresearch on hydrolysis or aggregation ofwhey proteins. Continuous bed chromato-graphy based on a polymer matrix has notyet been used for the separation of milk pro-teins. The use of the UNO Q-I column issuitable for the separation of the major wheyproteins and also of caseins (results notshown). The MemSep cartridge offered agood separation efficiency of whey proteinsonly at a pH greater than 8.0. This was dueto the weak adsorption capability of the cel-lulose matrix, compared to the Poros 20HQ,UNO Q-I and MonoQ matrix. Due to theparticular technology based on stackedmembranes, MCLC seemed to us easier tohandle than perfusion chromatography orcontinuous bed chromatography, especiallyfor scale-up in preparative fractionations.

400

0.04

9~'";,;"§

of0

~

0

Il-LGAIl-LGB

BSA

J.-M. Girardet et al.

a·LA

6 7

Retention time (min)

Figure 6. Separation of the major whey proteins,immun0l;,lobulins c1ass G (IgG), œ-lactalbumin(a-LA), p-lactoglobulins B and A (P-LG BandP-LG A, respectively), and bovine serum albu-min (BSA) by capillary zone electrophoresis(CZE). For details, see the Materials andMethods section.Figure 6. Séparation des protéines lactosériquesprincipales, immunoglobulines de classe G (IgG),a-lactalbumine (a-LA), p-Iactoglobulines B et A(P-LG B et P-LG A, respectivement), et séru-malbumine bovine (BSA), par électrophorèsecapillaire. Pour plus de détails, se reporter à lasection « Matériels et méthodes ».

REFERENCES toglobulin by capillary zone electrophoresis,Neth. Milk Dairy J. 48 (1994) 81-97.

[1] Afeyan N.B., Gordon N.F., Mazsaroff 1., Varady [10] Recio 1., Molina E., Ramos M., de Frutos M.,L., Fulton S.P., Yang Y.B., Regnier F.E., F1ow- Quantitative analysis of major whey proteinsthrough particules for the high-performance by capillary electrophoresis using uncoatedliquid chromatographie separation of biomole- capillaries, Electrophoresis 16 (1995) 654-658.cules: perfusion chromatography, J. Chroma-togr. 519 (1990) 1-29. [II] Ribadeau Dumas B., Physicochimie et biochimie

[2] Andrews A.T., Taylor M.D., Owen A.J., Rapid des protéines du lait. Données récentes, Lait 71analysis of bovine milk proteins by fast prote in (1991) 133-139.liquid chromatography, J. Chromatogr. 348 [12] Splitt H., Mackenstedt 1., Freitag R., Prepara-(1985) 177-185. tive membrane adsorber chromatography for

[3] Cifuentes A., de Frutos M., Diez-Masa J.C., the isolation of cow milk components, J. Chro-Analysis of whey proteins by capillary electro- matogr. 729 (1996) 87-97.phoresis using buffer-containing polymerie addi-tives, J. Dairy Sci. 76 (1993) 1870-1875. [13] Strange E.D., Malin E.L., Van Hekken D.L.,

[4] Gerstner J.A., Hamilton R., Cramer S.M., Mem- Basch J., Chromatographie and electrophoretic

brane chromatographie systems for high- methods used for analysis of milk proteins, J.

throughput protein separations, J. Chromatogr. Chromatogr. 624 (1992) 81-102.

596(1992) 173-180. [14] Torre M., Cohen M.E., Corzo N., Rodriguez[5] Girardet J.M., Pâquet D., Linden G., Effects of M.A., Diez-Masa J.C., Perfusion liquid chro-

chromatographie parameters on the fractiona- matography of whey proteins, J. Chromatogr.tion of whey proteins by anion exchange FPLC, 729 (1996) 99-111.Milchwissenschaft 44 (1989) 692-696.

[15] Van Deemter 1.1., Zuiderweg F.J., Klinkenberg[6] Hjertén S., Liao J.L., Zhang R., High-perfor- A., Longitudinal diffusion and resistance to massmance Iiquid chromatography on continuous transfer as causes of nonideality in chromato-polymer beds, J. Chromatogr. 473 (1989) graphy, Chem. Eng. Sci. 5 (1956) 271-289.273-275.[7] Lindeberg J., Capillary electrophoresis in food [16] Watzig H., Dette c., Capillary electrophore-

analysis, Food Chem. 55 (1996) 73-94. sis - A review. Strategies for method develop-[8] Musci G., Berliner L.J., Probing different confor- ment and applications related to pharmaceuti-

mational states of bovine œ-lactalbumin: fluo- cal and biological sciences, Pharmazie 49 (1994)rescence study with bis-ANS, Biochemistry 24 83-96.(1985) 3852-3856. [17] Weinbrenner W.F., Etzel M.R., Competitive

[9] Olle J.A.H.J., Kristiansen K.R., Zakora M., Qvist adsorption of œ-lactalbumin and bovine serumK.B., Separation of individual whey proteins albumin to a sulfopropyl ion-exchange mem-and measurement of œ-lactalbumin and ~-Iac- brane, J. Chromatogr. 662 (1994) 414-419.