Quantitative Determination of Complex Carbohydrates in Bovine Milkand in Milk-Based Infant Fonnulas

JEAN-RICHARD NEESER, MIRElLLE GOLLIARD, and SIMONE DEL veDOVONestec Ltd.

Nestle Research CentreVers-chez-Ies-Blanc

PO Box 44CH-1000. Lausame 26, Switzer1and

ABSTRACT

Quantitative determination of allstructural families of complex carbohy­drate micronutrients was perfonned onbovine milk samples, milk-based infantfonnulas, and whey-based manufacturingraw materials. Differences found be­tween fonnulas depended mainly ontheir whey:casein ratios. A solvent sepa­ration procedure was required for quan­titative estimation of the gangliosidesand neutral glycolipids within the fatfraction. All infant formulas except onecontained slightly more gangliosidesthan bovine milk. Complex carbohy­drates were consistently higher in thenonfat fraction. By gel permeation chro­matography, an oligosaccharide subfrac­tion was separated from a glycopeptideone. Oligosaccharide content of infantformulas increased as a function of thewhey:casein ratio, and glycopeptideswere found only in fonnulas made withwhey components. Neuraminic acidsfrom infant fonnuIas were associated pri­marily with the glycoprotein fraction, ex­cept in hydrolysate-based preparations inwhich "precipitable" glycoproteins wereconverted into "soluble" glycopeptidesby trypsin treatment. Because whey­based raw materials are very rich in allbovine milt glycoconjugates and oligo­saccharides, their increased use will re­sult in high contents of these micro­nutrients in modem formulas.(Key words: complex carbohydrates,glycoconjugates, infant formulas)

Received October 3. 1990.Accepted April 8. 1991.

1991 1 Dairy Sci 74:2860-2871

Abbreviation key: DWP = demineralizedwhey powder, IF =infant formulas, MFGM =milk fat globule membrane, NeuNAc =neura­minic acid, WPC =whey protein concentrate.

INTRODUCTION

Among mammalian milks, human milk. isunique with regard to the number and struc­tural diversity of its complex oligosaccharides(20), the concentration of which decreases aslactation progresses (33). Human milk is alsovery rich in glycoconjugates (glycoproteinsand glycolipids) that carry complex carbohy­drate moieties. Biological significance of thesecarbohydrate micronutrients is not understood.Human milk oligosaccharides have been sus­pected of being growth-promoting factors forLActobacillus bifidus, the predominant intesti­nal flora of breast-fed infants (31); however, aprecise relationship has not been established(3). The complex carbohydrates might protectthe breast-fed baby against infectious diseases.Indeed, in vitro assays have shown that similarmolecules competitively inhibit microbialadhesion and enterotoxin binding by acting asreceptor analogues (2, 7, 8, 14, 23, 27). Thehigh content of neuraminic acid (NeuNAc) inhuman milk oligosaccharide and glycoproteinfractions has been observed to decline ex­ponentially over the first 2 mo of lactation (5).Highly sialylated carbohydrates in human milkmight contribute to the increased concentrationof NeuNAc present in cerebral and cerebellarglycoconjugates of the breast-fed baby (5).

Compared with human milk, bovine milkcontains very little free saccharides other thanlactose, and neuraminyllactose is the onlycomplex oligosaccharide present in both thesemilks (22). However, bovine milk containsnumerous complex carbohydrates that are at­tached to protein or lipid backbones asgIycoconjugates (20). Infant fonnulas (IF)

2860

MILK GLYCOCONJUGATES IN INFANT FORMULAS 2861

TABLE 1. Macronutrients. compositions and ingredients characterizing the infant formulas analyzed in this sbldy.1

Proteins Lipids Carbohydrates

Whey:casein Milk:vegetable Lactose:maltodextrins

Formulas ratio lDgredients fat Jatio Ingredients ratio

AL 110 0:100 Casein 80:20 Milk fat 0:100Com oil

LACTOGEN (LA 20) 18:82 Whole milk 80:20 Milk fat 100:0Com oil

NAN (NA 60) 60:40 DWP 80:20 Milk fat 100:0Whole milk Com oil

NAN/LP (NA 64) 60:40 DWP 80:20 Milk fat 100:0Whole milk Com oilr caseinate

NAN/LP (NA 3549) 60:40 DWP 0:100 Vegetable oils 100:0Sldmmed milkr caseinate

BBBA HA. NAN H. A.(NAB 2) 100:0

NATIVA H. A. (NAH 3572) 100:0

HydrolyzedWPC 0:100 Vegetable oils

Hydrolyzed 60:40 Milk fatWPC Vegetable oils

70:30

100:0

lIP =: Low phosphate. H.A. =: hypoallergenic. DWP =: deminc:ralized wiley powder. WPC =: whey proteinconcentrate.

made with various bovine milk fractions willcontain these carbohydrate micronutrients.Quantitative analysis of these molecular spe­cies in formulas has been limited (4, 5. 23)although questions have been asked concern­ing the specific benefits for the bottle-fed in­fant of having complex carbohydrates of b0­vine origin in their food (6. 23). Detenninationof NeuNAc in IF established that mostsialylated species were in the glycoproteinfraction (4) and that content of NeuNAc in theoligosaccharide fraction was dependent on thewhey:casein ratio of the formula (5). TheNeuNAc-containing glycolipids (gangliosides)were detennined also in a whole milk-based IF(23). However, no comprehensive study hasbeen conducted to detennine quantitatively allthe molecular families of carbohydratemicronutrients present in the various kinds offormulas manufactUred.

The aim of the present work was to analyzethe main milk constitutive fractions for theiroligosaccharide and glycoconjugate compo­nents upon their isolation from three differentbovine milk samples (freshly collected. pas­teurized, and UHf-treated). eight milk-basedIF(incwdingh~~~genichYWu~~te~ased

formulas), and two whey-based raw materials.Human milk samples were also included forsome quantitative comparisons.

MATERIALS AND METHODS

Milk samples, Formulas,and Dairy Raw Materials

Two pools of human milk aliquots weremade from samples collected from a uniquemother (blood group O. secretor H) during d 4to 14 postpartum and the 4th mo of her lacta­tion. Fresh bovine milk was obtained immedi­ately after collection; pasteurized and UHT­treated milk samples were purchased in a su­permarket The IF of varied composition werefrom Nestle (Vevey, Switzerland) (Table 1).Demineralized whey powder (DWP) and wheyprotein concentrate (WPC) were obtained froma NestI6 factory. Pure samples of lactosyIcera­mide and ganglioside On3 were purchasedfrom BioCarb (Lund, Sweden). The neu­raminyllactose reference compound was ob­tained from Sigma Chemical Company (St.Louis, MO), and the caseinoglycomacropep­tide was prepared as previously descnbed (25).

Journal of Dairy Science Vol. 74. No.9. 1991

2862 NEESER ET AL.

Analytical Methods

Total neutral sugars (9) and total NeuNAc(17) were estimated quantitatively by colorim­etry. Neutral and amino sugar compositionswere analyzed by capillary gas-liquid chroma­tography (26). High perfonnance thin layerchromatography of neutral glycolipids wasperf.:lnned by using CHQ3-MeOH-H20 (65:

25:4, vol/vol/vol). Spots were visualized bydipping the plates (12). High perfonnance thinlayer chromatography of gangliosides was per­fonned by using CHCI3-MeOH-aqueous .25%CaCl2 (65:35:8, vol/vol/vol) in two consecu­tive runs. Spots were visualized by sprayingthe plates with a solution of resorcinol reagent(28).

Milk or reconstitutedformula / ingredient

solvent extraction

I

milk fatnonfat

fraction

partition

1

Ifat

fractiongangliosidefraction

liquidchromatography

neutralglycolipids

Figure 1. Scheme of solvent extraction of milk fat, ganglioside partition, and purification of neutral glycolipids.

lournal of Dairy Science Vol. 74, No.9, 1991

MILK GLYCOCONJUGATES IN INFANT FORMULAS 2863

Solvent Extraction

The procedure developed by Tettamanti etal. (30) was used on dialyzed milk and recon­stituted powder samples. Starting volumes ofhuman milk samples from early and late lacta­tion were 22 and 220 mI, and bovine milksample volumes were 500 mI. For fonnulareconstitution, 64 g of dry powder were dis­solved in 500 mI of water; for reconstitution ofwhey-based raw materials, samples equivalentto 17.5 g of protein were dissolved in 500 mIof water.

All preparations were dialyzed (molecularweight cutoff: 6000 to 8000 Da) against waterand freeze-dried prior to fat extraction. Frac­tionation steps were perfonned as described(30). Lyophilized material was homogenized(20%, wt/vol) into 10 roM potassium phos-

phate buffer (pH 6.8). Tetrahydrofuran (eightvolumes) was added to the suspension prior tocentrifugation at 1500 x g for 10 min at IYC.Pellet was re-extracted three times with a mix­ture of phosphate buffer and tetrahydrofuran(1:4, vol/vol), and the supernatants were com­bined. The residual pellet (containing the non­fat fraction) was suspended repeatedly in waterand concentrated under reduced pressure toremove traces of tetrahydrofuran prior tolyophilization. To the pooled supernatants, .3volumes of ethyl ether were added The result­ing organic and aqueous layers were separatedbefore the fonner was washed with .1 volumeof water. Combined aqueous layers (containingthe ganglioside fraction) were concentrated anddialyzed (cutoff = 1000) prior to lyophiliza­tion. The organic layer (containing the fat frac-

Milk or reconstitutedformula / ingredient

Idefatting

.--__1__----,

skimmed milk

TCA-precipitation

cream

solvent extraction

pelletSupernatant

liquidchromatography

nonfatfraction

"'-....--11111¥.,.--/glycoproteins I

fatfraction

I.part~t~on

Figure 2. Scheme of preparation of the subfractiODS derived from cream and skim layers.

JoumaI of Daily Science Vol. 74, No.9, 1991

2864 NEESER ET AL.

A) A)

00to

200

00

10

400 600 ml

.!\~-)\_,,\

200 400 600 ml

Figure 3. Separation of glycopeptides and oligosaccha­rides from milk and infant formula samples. Liquid chro­matography was pelformed on gel Sephadex G-50. Totalneutral sugars (e) were measured on aliquots of 50 IJl andtotal NeuNAc (0) on aliquots of 500 1Jl. A) The TCAsupernatant obtained from bovine skim milk (freshly col­lected). B) The TCA supernatant obtained from skim NA64 formula.

Physical Milk Defatting andAnalysis of Subfractlons

To obtain milk. fractions suitable for furthersubfractionation, the milk and reconstitutedpowder samples were centrifuged at 37'C for40 min at 2100 x g prior to cooling in an icebath for 2 h. The pellet [material from milk fatglobule membranes (MFGM)] was combinedwith the upper layer (cream), and the liquidlayer (skim milk) was collected separately. The

\j \

"400 600 ml

B)

10

00

Figure 4. Separation of glycopeptides and oligosaccha­rides from demineralized whey powder (DWP) and wheyprotein concentrate (WPC). Liquid chromatography wasperformed on gel Sephadex G-50. A) The TCA superna­tant obtained from DWP. Total neutral sugars (e) weremeasured on aliquots of 50 IJl and total neuraminic acid(NeuNAc) (0) on aliquots of 500 1Jl. B) The TCA super­natant obtained from WPC. Total neutral sugars (e) weremeasured on aliquots of 200 IJl and total NeuNAc (0) onaliquots of 50 1Jl.

cream fraction, when subjected to the solventextraction procedure, yielded a residue(MFGM proteins and glycoproteins) and a fatfraction from which the gangliosides were par­titioned (Figure 2).

To the skim milk layer, an equal volume ofTCA solution (24%, wt/vol) was added drop­wise before the resulting mixture was centri­fuged for 15 min at 4200 x g. Prior to lyophili­zation, the supernatant was dialyzed againstwater until it reached pH 5. An aliquot of thislast material was applied onto a Sephadex G­50 column and eluted with 100 mM aceticacid. Column fractions were analyzed for totalneutral sugars and NeuNAc, and the NeuNAc­positive fractions were collected (Figure 2).Glycopeptide fractions recovered from IF thatcontained maltodextrins were digested withamyloglucosidase to remove glucose-oligomercontaminants. Digestion was performed at6O'C in acetate buffer (100 roM, pH 5) with

lao ml400200

B)

DO!1.0!-

tion) was concentrated under reduced pressure.A 3-g aliquot of this fat was dissolved in 100ml of dichloromethane and applied onto al00-g silicic acid column (to). Neutral lipidswere eluted with 1 L of dichloromethane, fol­lowed by neutral glycolipids with 400 ml ofacetone and methanol (9:1, voWol), and phos­pholipids with 500 ml of methanol. The lastliter of eluant was collected in loo-ml frac­tions that were monitored for glycolipid con­tent by thin layer chromatography. Theglycolipid-containing fraction was concen­trated under reduced pressure. The overall frac­tionation procedure is summarized in Figure 1.

Journal of Daily Science Vol. 74, No.9, 1991

MILK OLYCOCONJUOAlES IN INFANT FORMUlAS 2865

200 400 600 ml

addition of enzyme portions occurring at thebeginning and middle of the 2-h reaction peri­od. Resulting digest was repurified by liquidchromatography on the same G-50 Sephadexcolumn.

glycopeptides, and glycoproteins. Carbohy­drate micronutrients present in these three frac­tions were estimated quantitatively by colorim­etry. Specifically, total neutral sugars weredetermined in the neutral glycolipid fraction,NeuNAc in the ganglioside fraction, and bothmeasurements were used to quantify the com­plex carbohydrates within the nonfat fraction.Across fractions, contents of NeuNAc and neu­tral sugars within complex carbohydrates werehigher in human than in bovine milk samples(fable 2). Levels were higher also during theearly relative to late human lactation period.Bovine milk samples exhibited very small dif­ferences compared with those observed amongIF (fable 2).

Heat treatment of bovine milk did not causea loss of either ganglioside-bound or nonfatfraction NeuNAc (fable 2). All IF except AL110 contained more gangliosides than the b0­vine milk samples. Elevated amounts of gan­gliosides were found in both whey-based man­ufacturing materials, DWP and WPC. Asexpected (11), upon thin layer chromatogra­phy, lactosylceramide was identified as themajor neutral glycolipid in all bovine milk andIF samples (results not shown).

Nonfat fractions contained 20 to 100 timesmore NeuNAc than the ganglioside fractions(fable 2). Similarly, 40 to 500 times moreneutral sugars were present in nonfat complexcaIbohydrates than in neutral glycolipids (fa­ble 2). The NeuNAc content of nonfat materialrecovered from IF increased with the whey:casein ratio. For example, formulas made withcasein (AL 110) or with whole milk (LACfO­GEN) contained about half the amount of non­fat NeuNAc found in bovine milk samples,whereas formulas like NAN and NANILP, inwhich DWP was added to increase the Whey:casein ratio, contained the same amount ofnonfat NeuNAc as bovine milk. Hypoaller­genic formulations prepared with only WPCand DWP (NAN H.A. and NATIVA H.A.)contained still higher levels (fable 2).

600 ml200 I----l 400

Figure 5. Separation of glycopeptides and maltodex­trins from BEBA H.A. formula. Liquid chromatographywas performed on gel Sephadex 0-50. Total neutral sugars(e) were measured on aliquots of 50 J.1l and total NeuNAc(0) on aliquots of 250 J.1l A) The TCA supernatant ob­tained from the skim formula. B) Olucoamylas~tofthe NeuNAc-positive fractions collected under A) andpooled.

00

20

B)1.6

\2

.8

.4

002.0

A)

,.

'2

.8

.4

RESULTS

Complex Carbohydrates fromthe Lipid, Ganglioside, andTotal Nonfat Milk Fractions

Separation procedure (Figure 1) yielded pu­rified fractions of gangliosides and neutralglycolipids plus a nonfat fraction that con­tained a mixture of complex oligosaccharides,

Ollgosaccharldes, Glycopeptldes,and Glycoprotelns

Precipitation of proteins from skim milksamples yielded supernatants that containedsialylated molecular species. After dialysis, ali­quots of these supernatants were subjected togel permeation chromatography (Figure 2).Typical elution patterns are shown in Figures 1to 3. The TCA supernatant from fresh bovine

Journal of Daily Science Vol. 74, No.9, 1991

2866 NEESER ET AL.

skim milk: contained only one fraction positivefor NeuNAc. Based upon its elution close tothe lactose peak (Figure 3A), this sialylatedspecies was identified as neuraminyllactose.Similar elution profiles we.re obtained frompasteurized and UHf-t.reated bovine milk sam­ples and from LAcrOOEN IF. The superna­tant obtained from AL 110 did not containsialylated complex carbohydrates. By contrast.the TCA supernatant originating from NA 64IF contained two distinct fractions positive forNeuNAc (Figure 3B). The first fraction wasretained moderately by the 0-50 Sephadex,and the second corresponded to the neu­raminyllactose just described. The fonner frac­tion exhibited all the characteristics of thecaseinoglycomacropeptide, based upon its lackof precipitation by TCA, apparent molecularweight, and general occurrence in formulasmade with a whey component.

To ascertain the origin of this "formula­caseinoglycomacropeptide", elution profilesobtained from DWP and WPC were examined(Figure 4). The TCA supernatant of DWP ex­hibited the same two NeuNAc-containing frac­tions as the supernatant from NA 64 formula

(compare Figures 3B and 4A). When derivedfrom WPC, the TCA nonprecipitable materialcontained very high amOWlts of casein­oglycomacropeptide (Figure 4B); however,neurarninyllactose no longer occurred in thissupernatant. Finally, when performed on thehypoallergenic formula BEBA H.A., this ana­lytical procedure was complicated by the pres­ence of maltodextrins (Figure SA). Fortunate­ly, the glycopeptide fraction eluted after highmolecular weight maltodextrins but beforetheir low molecular weight analogues. Thisglycopeptide subfraction was purified after iso­lation by digesting residual maItodextrins withglucoamylase. Subsequent liquid chromatogra­phy removed the glucose released by this en­zymatic hydrolysis (Figure SB).

All bovine milk. samples contained between50 and 70 mgIL borne by NeuNAc theoligosaccharide neurarninyllactose (fable 3).This level was significantly lower in the wholemilk-based LA 20 formula but maintained inwhey-containing formulas (NA 60 and NAH2). As already noted, DWP was rich in neu­raminyllactose, whereas this oligosaccharidewas absent from WPC (fable 3). Amount of

NeutralNomat fraction glyoolipids Gangliosides

Neutral sugars Total NeuNAc Neutral sugars NeuNAc

3419 933 6.8 ND2

1006 270 3.6 10.4220 166 5.2 2.0222 190 4.7 2.2232 192 5.8 2.4

ND 92 ND <.5290 100 2.5 4.6298 204 6.0 4.4318 172 4.5 3.2390 168 3.9 4.2

ND 284 ND 3.6ND 250 72 6.2

TABLE 2. Neutral sugars and IICUl'IID1inic acids (NeuNAc) associated with complex carbohydrates in. fat and nonfatfractions derived from various milk samples, infant formulas, and manufacturing materials.!

Milk sampleHuman (lactation d 4 to 14)Human (4th lactation month)Bovine (freshly collected)Bovine (pastewized)Bovine (UHT treated)

Infant formutas3AL 110LACfOOEN (LA 20)NAN (NA 60)NAN/lP (NA 64)NAN/lP (NA 3549)NAN H.A. (NAB 2)NATIVA H.A. (NAB 3572)

Manufacturing materials4

DWP 735 480 ND 8.8WPC 665 685 10.4 10.4

IBxpressed in milligrams glucose equivalent or NeuNAc by liter of milk sample or reconstituted material.

2ND = Not determined.

3RecoosUtured at 12.8% (dry Wf/vol).

4Reconstituted at 3.5% (protein wt/vol), DWP = demineralized whey powder, WPC = whey protein concenttate.

Journal of Da.iJy Science Vol. 74, No.9, 1991

MILK GLYCOCONJUGATES IN INFANT FORMULAS 2867

TABLE 3. Quantitative determination of neuxaminic acidsl bome by complex carbohydrates in subtractions obtainedfrom milk samples, infant fonnulas. and manufacturing materials.

Skim fraction Cream

Glycoproteins2Gangliosides

Oligosaccharides Glycopeptides (recovery)

Milk sampleND3Bovine (freshly collected) 71 95 1.9 (95%)

Bovine (pasteurized) 59 NO 131 1.8 (82%)Bovine (UHT treated) 53 NO 139 1.1 (46%)

Infant formulas4

AL 110 ND NO 92 S.sLAcrOGEN (LA 20) 28 NO 72 .6 (13%)NAN (NA 60) 49 42 113 1.9 (43%)NAN H.A. (NAH 2) 69 213 2 2.9 (81%)

Manufacturing ma~DWP 88 96 296WPC 9 244 432

IExpressed in milligrams of neuraminic aCid by liter of milk sample or reconstituted material.

2calculated by differeoce between the total nonfat neuraminic acids (from Table 2) and those occurring in theoligosaccharide and glycopeptide fractions (columns 1 and 2 from this lable).

~ =Not detected.

"Reconstituted at 12.8% (dry wt/vol).

~econstituted at 3.5% (protein wt/vol), DWP = demineralized whey powder, WPC = whey protein concentrate.

NeuNAc measured as a probe for glycopep­tides was especially high in WPC and in theNAN H.A. formula, which is based on a DWP:WPC mixture (fable 3). Monosaccharidesbome by these glycopeptide molecules wereNeuNAc, galactose, and N-acetylgalactosamine(fable 4), as expected for the caseinoglycoma­cropeptide (1). In addition, amounts of fucose,mannose, and N-acetylglucosamine were alsomeasured in the nonprecipitable glycopeptidesfrom BEBA H.A. (fable 4), These last constit­uents originated from the whey glycoproteincarbohydrate chains normally precipitated byTCA, but, in hypoallergenic formulas, thesecarbohydrate units were recovered as non­precipitable glycopeptides due to enzymatichydrolysis of the proteins.

The TCA precipitates contained milk pro­teins and glycoproteins. The low relative per­centage of glycoproteins allowed measurementof less than 1% total neutral sugars in thesepellet fractions. Moreover, the TCA precipita­tion procedure caused extensive proteindenaturation, which yielded pellets not readilyavailable for colorimetric measurement ofNeuNAc. Finally, these skim milk glycopro­teins did not represent the total glycoprotein

content of milk and IF samples, becauseglycoproteins associated with MFGM wouldbe recovered with the cream layer. Conse­quently, NeuNAc borne by the glycoproteinsubfraction was estimated by calculating thedifference between the total nonfat NeuNAc(see Table 2) and the NeuNAc associated withboth the oligosaccharide and the glycopeptidesubfractions (fable 3). The hypoallergenic IF(NAN H.A.) contained almost no moreglycoproteins, although its manufacturing ma­terials (DWP and WPC) were very rich inthese molecular species (fable 3). This fmdingagain shows that in hydrolysate-based IF,glycoprotein cleavage yielded carbohydratechains that occur on TeA-soluble glycopep­tides.

Recovery of MFGM ComponentsIn the Cream Layer

Generally, the MFGM components are stud­ied after their isolation from a cream fractionprepared by physical methods (23). To allowdata comparison, it was necessary to study thebehavior of the material derived from MFGMduring this physical milk defatting process. For

Joumal of Dairy Science Vol. 74, No.9, 1991

2868 NEESER ET AL.

TABLE 4. Monosaccharide composition of the glycopeptides found in the TeA-soluble fraction obtained from skimformulas and manufacturing ingredienls.

Monosaccharides

Fucose Mannose Galactose GicNAc1 GalNAc NeuNAc

---------- (%. wt/Wt) ----------NAN (NA 60) <.5 <.5 3.9 <.5 5.9 14.7NANILP (NA 64) <.5 <.5 6.5 <.5 6.2 14.7BEBA H.A. (NAH 2) 0.7 3.2 8.3 2.9 8.0 11.2DWJ>2 <.5 <.5 6.5 <.5 7.6 16.0WPC <.5 <.5 5.7 <.5 7.9 17.4

IGlcNAc =N-Acetylglncosamine. Ga1NAc =N-acetylgaJactosamine; NeuNAc = neuraminic acid.

2DWP =Demineralized whey powder, WPC = whey protein concentrate.

this purpose, gangliosides served as probesbecause they are typical of polar lipids fromcell membranes. Starting with the cream layers(Figure 2), the procedure of Tettarnanti et al.(30) yielded both a "cream-glycoprotein" frac­tion and a "cream-ganglioside" fraction. Quali­tative analysis of these gangliosides was per­formed by thin layer chromatography. Results(not shown) confirmed that Go3 was the majorbovine milk ganglioside (19) and establishedthat this was true for all milk-based IF ana­lyzed. The NeuNAc contained in the "cream­ganglioside" fractions was estimated quantita­tively (Table 3) for comparison with the valuesobtained for gangliosides when whole milksamples were partitioned (Table 2). Ratios be­tween these values represented the gangliosiderecoveries in the cream layers (Table 3). Asexpected. freshly collected bovine milk yieldeda cream layer made of intact milk fat globulesbecause 95% of the total gangliosides wererecovered in this layer. Notably, for bovinemilk samples and for the whole milk-based LA20 formula, ganglioside recovery in creamdecreased as a function of the heat treatmentHowever, when the whey:casein ratio in­creased in the formulation, the gangliosiderecovery in cream increased also (Table 3).

DISCUSSION

To perform a comprehensive quantificationof gangliosides, neutral glycolipids, and thenonfat species (glycoproteins, glycopeptides,and complex oligosaccharides), two separationprocedures that involved two distinct aliquotsof each sample were used. Methodology deve­loped by Tettamanti et al. (30) for fractionating

10umal of Dairy Science Vol. 74, No.9, 1991

brain glycoconjugates was used to achieve thecomplete solvent separation of gangliosidesand the purification of neutral glycolipids frommilk fat (Figure 1). In addition, physical milkdefatting was performed to yield skimmed andcream layers, which were subfractionated(Figure 2). The first procedure quantitativelyyielded gangliosides and neutral glycolipids,and the second permitted distinction betweenthe various species of nonfat complex carbohy­drates.

With one exception, level of ganglioside­boMdNeuNMwashi~~infurmulasthanin

bovine milk samples (Table 2). Gangliosidesare important components of the MFGM (19,24, 29). However, when IF ingredients areprepared from homogenized milk, these polarglycolipids associated with disrupted MFGMcan migrate into various milk fractions, i.e.,cream, milk SNP, or whey-based components.Upon combining these IF ingredients, resultingcontent of gangliosides mi~t exceed the initialamount fOMd in bovine milk.

It has been shown that fat layers obtainedby thermal defatting had lower levels of gan­gliosides than those reported herein (23). Gan­glioside recovery from these cream layers isnot complete, especially if those layers havebeen heat treated (Table 3). Present resultsindicated that quantitative determination ofgangliosides in industrially processed milksamples absolutely requires use of the solventpartition procedure (Figure 1).

Content of IF gangliosides is not alwaysdependent on the origin of the fat component(Table 2). Although both AI... 110 and LACfO­GEN are made with a fat component of milkorigin, the AI... 110 does not contain ganglio-

MILK GLYCOCONJUGATES IN INFANT FORMULAS 2869

sides, but LACfOOEN contains a high levelof these molecular species (Table 2). If themilk. fat ingredient used for manufacturing anIF is cream, these complex caIbohydrates willoriginate partly from this component. Howev­er, if butter oil is used instead of cream,glycoconjugates will not occur in this fat in­gredient because MFGM was removed withthe butter milk. (15). Fat fractions of NA 60and NA 3549 fonnulas contained the sameganglioside level, but the former was madewith milk fat and the latter with vegetable fat(Table 2). Apparently, these gangliosides didnot originate from the formula fat component,but rather from the material used to increasethe whey:casein ratio, namely DWP. The highNeuNAc content of the DWP ganglioside frac­tion strongly supported this hypothesis (Table2). Qualitative analyses confirmed that Go3was the major bovine milk ganglioside (13, 19,29) and that all formulas (except AL 110)exhibited a ganglioside pattern identical to thatfrom bovine milk (23).

Neutral glycolipids are also important com­ponents of the MFGM (11). Reported levels of70 mg of lactosylceramide and 38 mg ofglucosylceramide per 10 L of bovine milk. arein agreement with current results (Table 2).The NeuNAc are monosaccharides commonlyfound in complex carbohydrates of animal ori­gin, but they do not occur in all glycocon­jugates or oligosaccharides. Among the milkcomplex carbohydrate families, neutralglycolipids alone do not carry NeuNAcresidues. Considering the low concentration ofmilk. glycolipids relative to nonfat glycocon­jugates and oligosaccharides (Table 2) and thegeneral occurrence of NeuNAc in milk com­plex carbohydrates, content of total nonfatNeuNAc is an appropriate probe for these milkcarbohydrate micronutrients. Consequently, aclear relationship appeared between the levelof nonfat NeuNAc in IF and the amount ofDWP and WPC used for their manufacture(Tables 1 and 2), because both contained veryhigh levels of nonfat NeuNAc (Table 2). Otherreports (4, 5) support this observation.

Neuraminyllactose is the only complex hu­man milk oligosaccharide also found in milkof other mammalian species (22). It was identi­fied as the major sialylated free oligosaccha­ride in formulas (4). Formulas manufacturedby mixing whey protein and casein in a 60:40

ratio have been reported to contain fivefoldmore NeuNAc in their oligosaccharide fractionthan that found in this fraction from cow'smilk and formulas made with a 18:82 ratio [72vs. 14 mg/L, (5)]. In this last study, neu­raminyllactose was not conclusively identified,because the fractionation procedure used toobtain these "oligosaccharides" did not sepa­rate them from the caseinoglycomacropeptide,a glycopeptide that, upon cleavage from K­

casein by rennin during cheese making, occursin whey and whey-based raw materials (1). Toresolve this analytical problem, gel permeationliquid chromatography was used herein to sep­arate the glycopeptide and oligosaccharidefractions (Figures 3 to 5). Quantitative estima­tions clearly showed that IF content of neu­raminyllactose increased with the whey:caseinratio, but not in the proportion reported (5).Glycopeptide fraction occurred only in formu­las made with one or more whey components(Table 3). The DWP material contained similaramounts of neuraminyllactose and caseio­oglycomacropeptide, but WPC had a highglycopeptide content but no oligosaccharide(Table 3). The ultrafiltration or diafiltrationstep used to process the WPC resulted in reten­tion of the whey glycopeptides, and wheyoligosaccharides were lost with lactose.

Milk proteins can be divided in two maincategories: caseins and whey proteins. Amongthe various casein subcomponents, only lC­casein has been confirmed to be a glycoprotein(20). Regarding the carbohydrate chains frombovine mature milk lC-casein, three structureshave been established: NeuNAca2 ~ 3Gal131~ 3GalNAc, Gal~1 ~ 3-(NeuNAca2 ~

6)GalNAc and NeuNAca.2 ~ 3Gal131 ~

3(NeuNAca.2 ~ 6)-GalNAc (16, 32). By con­trast, numerous whey glycoproteins have beendescribed (20). Glycoproteins are not onlyfound in skim milk but also in the milk fatfraction, more precisely on the MFGM (18,21). Among the monosaccharide constituentscommonly occurring in milk. glycoprotein car­bohydrate chains, only sialic acids have beenquantified in IF (4, 5). Present primarily onglycoproteins (4), NeuNAc values of 56 to 67mg/L were found in milk.-derived formulas (5),whereas a value of 146 mg/L was obtained forcows' milk. (5). Present data (Table 3) con­firmed that most of the sialic acid residuesfrom milk. glycoconjugates are borne by

Journal of Dairy Science Vol. 74, No.9, 1991

2870 NEESER ET AL.

glycoproteins. The case of hypoallergenic f?r­mulas is an exception, because glycoprotemsare converted into TCA-soluble glycopeptidesby trypsin hydrolysis. An analysis of the com­position of the various glycopeptide fractionsfrom IF corroborated this observation (Table4); fucose, mannose, and N-acetylglucosaminewere only found in the glycopeptide fraction ofBEBA KA.

CONCLUSION

Most milk-based infant formulas containedcomplex carbohydrates in amounts similar tothose found in bovine milk. In whey-based rawmaterials, elevated levels of all bovine milkglucoconjugates and oligosaccharides wereevident. Consequently, modem formulas pre­pared from these raw materials will containincreased amounts of these microoutrients.

REFERENCES

1 Alais, C., and P. Jom~s. 1961. Etude compam: descaseino-glycopeptides formes par action de Ia pr6suresur les cas6incs de vache, de brebis et de c~vre. n.Etude de Ia partie non-pcptidique. Biochim. Diophys.Acta 51:315.

2 Andersson, D., O. Ponas, L. A.. Hanson, T. Lagergard,and C. SvlUJborg-Eden. 1986. Inhibition of attachmentof Streptococcus pnewnonioe and Haemcphilus in­fluenza hy human milk and receptor oligosaccharidcs.I. Infect Dis. 153:232. .

3 Braun, O. H. 1981. Effect of consumption of humanmilk and other formulas on inleStinal bacterial flora ininfants. Page 247 in Textbook of gastroenterology andnutrition in infancy. E. Lebenthal. cd. Raven Press,New York, NY.

4 Cabezas, J. A., and A. Carrion. 1969. Acidos sialicos.X. Contenido en acidos acilneuraminicos de variosproductos die~ticos (leches matemizadas). Rev. Es­pan. Pisiol. 25:141.

5 Carlson, S. E. 1985. N-AcetylDeuraminic acid coucen­trations in human milk oligosaccharides andglycoproteins during lactation. Am. J. Clin. Nutr. 41:720.

6 Cleary, T. G., J. P. Chambers, and L. K. Pickering.1985. Protection of sucldiDg mice from heat-stableenterotoxin of Escherichia coli by infant formulas. J.Pediatr. Gastrocnterol. Nutr. 4:125.

7 Coppa, G. V., O. Gabrielli, P. Giorgi, C. Catassi, M.P. Montanari, P. E. Varaldo, and B. L. Nichols. 1990.Preliminary study of breast feeding and bacterialadhesion to urocpithelial cells. Lancet 335:569.

8 Cravioto, A., A. Tello, H. VilIafan, J. Ruiz, S. DelVcdovo, and J.-R. Necser. 1991. Inhibition of local­ized adhesion of enteropathogenic Escherichitz coli toHEp-2 cells hy immunoglobulin and oligosaccharidefractions of human colostrum and breast-milk. J. In­fect Dis. 163(6).

Ioumal of Dairy Science Vol. 74, No.9, 1991

9 Dubois, M., K. A. Gilles, J. K. Hamilton, P. A.Rebcrs, and F. Smith. 1956. Colorimetric method fordelCrmination of sugars and related substances. Anal.Chem. 28:350.

10 Esselman, W. J., R. A. Laine, and C. C. Swecley.1972. Isolation and characterization of glycosphin­golipids. Page 140 in Methods in enzymology. VoL28, Part B. V. Ginsburg, cd. Academic Press, NewYork, NY.

11 Fujino, Y., M. Nakano, and T. SacIri. 1970. Thechemical structure of glycolipids of bovine milk.Agric. BioI. Cbcm. 34:442.

12 Hansen, S. A. 1975. Thin-Iayer chromatographicmethod for identification of oligosaccharides in starchhydro1ysates. J. Chromatogr. 105:388.

13HaU1tccOcur, B~ S. Sonoino, and R. Ghidoni. 1985.Characterization of two molecular species 003 gan­glioside from bovine buttermilk. Diochim. Biophys.Acta 833:303. •

14 Holmgren, J., A.-M. Svenncrholm, and C. Ahr6n.1981. Nonimmunoglobulin fraction of human milkinhibits bacterial adhesion (hemagglutination) and en­terotoxin binding of Escherichia coli and VibriocMlerae. Infect. Immun. 33:136.

15 Huang, R.T.C. 1973. Isolation and characterization ofthe gangliosides of butter milk. Biochim. Biophys.Acta 306:82.

16 Jo11es, P., and A.-M. Piat. 1979. The carbohydrateportions of milk glycoproteins. J. Dairy Res. 46:187.

17 Jourdian, G. W., L. Dean, and S. Roseman. 1971. Thesialic acids. XI. A periodalc-resoreinol method for thequantitative estimation of free sialic acids and theirglycosides. J. BioI. Chem. 246:430.

18 Kanno, C. 1986. Receptor proteins for concanavalin Aand wheat genn agglulinin of bovine milk fat globulemembrane probed hy affinity chromatography in thepresence of sodium dodecy1 sulfate. Agric. BioI.Chem. 50:2997.

19 Keenan, T. W. 1974. Composition and synlhesis ofgangliosides in mam.mary gland and milk of the b0­vine. Biochim. Biophys. Acta 337:255.

20 Kobata, A. 1977. Milk glycoproteins and oligosaccha­rides. Page 423 in The glycoconjugates. Vol. I. M I.Horowitz and W. Pigman, cd. Academic Press, NewYork, NY.

21 Kobylka, D. and K. L. Carraway. 1972. Proteins andg1ycoproteins of the milk fat globule membrane. Bio­chim. Biophys. Acta 288:282.

22Kuhn, R.. and R. BrosSlIlCl'. 1956. Ueber o-Acetyl­Iactamins1ure-lactosc aus Kuh-Colostrum und ihreSpaltberlreil durch Influenza-Virus. Olem. Ber. 89:2013.

23 Laegreid, A., A.-B. K. Otnaess, and J. Fuglesang.1986. Human and bovine milk. Comparison of gan­glioside composition and enterotoxin-inhibitory activ­ity. Pediatr. Res. 20:416.

24 McI1hinney, RAJ., S. Patel, and M. E. Gore. 1985.Monoclonal antibodies recognizing epitopcs carriedon both glycolipids and glycoproteins of the humanmilk fat globule membrane. Bioc1lcm. J. 227:155.

25 Neeser, I.-R., A. Chambaz, S. Del Vedovo, M.-J.Prigent and B. Guggenheim. 1988. Specific and non­specific inhibition of adhesion of oral Actinomycesand Streptococci to erythrocytes and polystyrene by

MILK GLYCOCONIUGATES IN INFANT FORMULAS 2871

caseinoglycopeptide derivatives. Infect. Immun. 56:3201.

26 Neeser, I.-R., and T. P. Schweizer. 1984. A quantita­tive determination by capillaIy gas-liquid chromatog­raphy of neutral and amino sugars (as o-methyloximeacetates), and a study on hydrolytic conditions forgIycoproteins and polysaccharides in order to increasesugar recoveries. Anal. Biochem. 142:58.

27 Newburg, D. S., L. K. Pickering, R. H. McCluer, andT. G. Cleary. 1990. Pucosylated oligosaccharides ofhuman milk protect suckling mice from heat-stableenterotoxin of Escherichia coli. 1. Infect Dis. 162:1075.

28 Svennerbolm, L. 1957. Quantitative estimation ofsialic acids. n. A colorimetric resorcinol-hydrochloricacid method. Biochim. Biophys. Acta 24:604.

29 Takamizawa, K., M. lwamori, M. Mutai, and Y.Nagai. 1986. Gangliosides of bovine butter milk. Iso­lation and characterization of a novel monosialogan-

glioside with a new branching structure. 1. BioI.Chem. 261:5625.

30 Tettamanti, G., P. Bonali, S. Marchesini, and V. Z3m­botti. 1973. A new procedure for the extraction, puri­fication and fractionation of brain gangliosides. Bio­chim. Biophys. Acta 296:160.

31 Tissier, H. 1990. Recherches sur la flare intestinalenormale et pathologique du nourisson. Ph.D. Diss.,Univ. Paris, Paris, France.

32 Van Halbeek, H., L. Dorland, I.F.G. Vliegenthart, A.­M. Fiat, and P. Ioll~. 1980. A 360-MHz IH-NMRstudy of three oligosacchar.ides isolated from cow IC­

casein. Biochim. Biophys. Acta 623:295.33 Viverge, D., L. Grimmonprez, G. Cassanas, L. Bardel,

H. Bonnet, and M. Solm. 1985. Variations of lactoseand oligosaccharides in milk from women of bloodtypes secretor A or H, secretor Lewis, and secretor HInonsecretor Lewis during the course of lactation. Ann.Nutr. Metab. 29: 1.

Iournal of Daily Science Vol. 74, No.9, 1991