HAL Id: hal-00900572https://hal.archives-ouvertes.fr/hal-00900572

Submitted on 1 Jan 2005

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.

Focus on the supramolecular structure of milk fat indairy products

Christelle Lopez

To cite this version:Christelle Lopez. Focus on the supramolecular structure of milk fat in dairy products. Reproduc-tion Nutrition Development, EDP Sciences, 2005, 45 (4), pp.497-511. �10.1051/rnd:2005034�. �hal-00900572�

497Reprod. Nutr. Dev. 45 (2005) 497–511© INRA, EDP Sciences, 2005DOI: 10.1051/rnd:2005034

Original article

Focus on the supramolecular structure of milk fat in dairy products

Christelle LOPEZ*

UMR Science et Technologie du Lait et de l’Œuf, INRA-Agrocampus, 65 rue de Saint-Brieuc, 35042 Rennes Cedex, France

Abstract – Bovine fat is dispersed in raw milk as natural milk fat globules, with an average diameterof 4 µm, which are enveloped in a biological membrane, the milk fat globule membrane (MFGM).However, dairy processes modify the supramolecular structure and the surface composition of milkfat. Thus, milk fat is present in many dairy products under various forms. In this study, we focusedon the fact that natural milk fat globules are rarely consumed in their native state, i.e. in fresh rawmilk. In most drinking milks, fat globules are homogenised in order to avoid their rising at the surfaceof the products. Furthermore, fat globules are heat treated to avoid the growth of micro-organisms.As a consequence of the technological process applied, the volume-weighted average diameter offat globules in drinking milks is in the range 0.2–0.5 µm. Homogenisation of fat globules led to thepartial disruption of the MFGM and to the adsorption of milk proteins. Moreover, this study showedthat in cheeses, milk fat can be dispersed as (i) fat globules with the MFGM, (ii) aggregates of fatglobules, (ii) homogenised fat globules, (iii) free fat and (iv) a combination of different phases andstructures. The knowledge of the supramolecular structure of milk fat in dairy products is of primaryimportance regarding its technological, sensorial and nutritional properties.

milk fat / microstructure / confocal laser scanning microscopy

1. INTRODUCTION

Milk fat is widely consumed in differentkinds of products, i.e., milk, cream, ice-cream, yoghurt, cheese, butter and as anhy-drous milk fat (AMF). Its concentration andchemical composition are easily deter-mined and thus well-known. However,information on the supramolecular struc-ture of milk fat in dairy products is scarce.

Milk, which is the secretion of the mam-mary gland, is a staple food for all youngmammals [1]. It is important to highlightthat directly consumed from the mother to

the newborn, all the constituents of milk,mainly proteins and fat, are in their nativestate. Depending on the species considered(humans, bovine, sheep, goats, etc.), milkhas special characteristics, i.e. compositionand structure of the constituents [2, 3].Although milk is an important product,regarding the high nutritive value of its con-stituents, mammals usually do not consumeit post weaning. However, bovine milk anddairy products are widely used for humanconsumption.

Bovine milk consists of 87% water, [4].Fat is a major component of fresh raw milk

* Corresponding author: Christelle.Lopez@rennes.inra.fr

Article published by EDP Sciences and available at http://www.edpsciences.org/rnd or http://dx.doi.org/10.1051/rnd:2005034

498 C. Lopez

that exists as small droplets called the milkfat globules. The size of natural milk fatglobules ranges from 0.2 to 15 µm with avolume-weighted diameter of about 4 µmdepending on cow breed and season [5].They are composed mainly of 98% triacyl-glycerols, at least 200 of which have beenidentified [3, 6]. Natural milk fat globulesare surrounded by a biological membrane,the milk fat globule membrane (MFGM).The native MFGM consists of a complexmixture of proteins, glycoproteins, enzymes,phospholipids, cholesterol and other minorcomponents [7]. The structure in trilayer,thickness (∼ 10 nm) and composition of thenative MFGM depends on the secretionprocess [8–10]. The MFGM acts as a natu-ral emulsifying agent that prevents aggre-gation and coalescence of milk fat and alsoprotects the fat against enzymatic action,mainly lipolysis [11].

During the processing of milk, severaltreatments may alter the natural structure offat globules and the MFGM: (i) mechanicaltreatments, i.e. pumping, agitation and homog-enisation, (ii) thermal treatments, i.e. refrig-eration and heating (pasteurisation, sterili-sation) and (iii) enzymatic treatments, i.e.lipolysis. It is well known that heat treat-ments and homogenisation produce the great-est changes in the milk fat globules. Heattreatment of milk causes changes in theMFGM by promoting interactions betweenmilk proteins and native MFGM components[12, 13]. Homogenisation of milk causes areduction of fat globule size and a concur-rent increase in the milk fat surface area,which alters the native MFGM and modi-fies the composition of the membrane [11,12, 14].

However, the effects of physico-chemi-cal and physical treatments applied duringthe manufacture of dairy products on thechanges in milk fat globule supramolecularstructure have not been fully studied. Fewstudies exist on the organisation of fat incheese [15–19].

In France, the consumption of milk, but-ter and cheese is an everyday act with a

strong cultural dimension. Thus, milk anddairy products are of great economicalimportance. About 23 million tons of milkand 1 million tons of milk fat are producedeach year [20]. Considering the productionof milk and the concentration of fat in milk,we can calculate that in 2001, milk fat wasmainly present in the form of butter andanhydrous milk fat (360 000 t), cheese(340 000 t) and cream (100 000 t). Frenchpeople consume about 13.5 kg milk fat/per-son/year: mainly as butter (6.6 kg fat/per-son/year) and ripened cheese (3.6 kg fat/person/year). Regarding the consumptionof cheese, France is the second country inthe world with 24.5 kg/person/year. Theexceptional variety of cheese produced inFrance is accompanied by a diverse array ofcheese-making techniques.

Milk fat is widely consumed. However,health authorities in most Western nationshave drawn links between fat consumptionand chronic conditions such as heart diseaseand obesity, resulting in widespread recom-mendations for reductions in the consump-tion of saturated fats and cholesterol [21].Regarding the role of milk fat in dairy prod-ucts, Jaros et al. [22] found that the texture,flavour and physico-chemical properties ofcheese are greatly governed by the milk fat.Thus, the structure of fat in dairy productsis of tremendous importance regarding tech-nological, functional, sensorial and nutri-tional properties.

The objective of this work was to showthat the supramolecular structure of milk fatin dairy products can be drastically differ-ent, depending on the technological processapplied during the manufacture. We focusedon the organisation of fat in raw milk, drink-ing milks and on the supramolecular struc-ture of fat in different kinds of cheeses.

2. MATERIAL AND METHODS

2.1. Samples

The raw whole milk (3.8 g fat/100 g) usedin this study was obtained from a local dairy

Organisation of fat in dairy products 499

plant (Triballat, Noyal-sur-Vilaine, France).Full-fat UHT milk (3.6 g fat·100 g–1), semi-skimmed UHT milk (1.55 g fat·100 g–1),skimmed UHT milk (0.1 g fat·100 g–1),full-fat pasteurised milk (3.5 g fat·100 g–1)and semi-skimmed pasteurised milk (1.5 gfat·100 g–1) were commercial products, pur-chased from a local supermarket (Géant,Rennes, France).

Camembert and ultrafiltration cheeses(soft cheeses) and whipped cream cheesewere commercial cheeses, obtained as sam-ples submitted for routine quality controlanalysis.

The samples characterised during the man-ufacture of Emmental cheese (hard cookedcheese) were obtained as follows: cheesemilk was coagulated at 31 °C, pH 6.62, therennet-induced gel was cut, stirred and heatedto 51 °C for 20 min, before being drainedin the vat, resulting in curd grains of aboutthe same size as rice: length 4–6 mm. Thesecurd grains were maintained under pressure(0.4 kPa), at 47 °C, during 4 h. Then, thecurd was demoulded and placed in a coldbrine bath for 48 h. The cheese was ripenedduring 52 d [23].

2.2. Milk fat globule size measurements

The fat globule size distribution wasmeasured by laser light scattering using aMastersizer 2000 (Malvern, UK), with twolaser sources. The refractive indexes usedwere 1.458 and 1.460 for milk fat at 633 and466 nm, respectively, and 1.33 for water.The samples of milk (about 0.2 mL) werediluted in 100 mL of water directly in themeasurement cell of the apparatus in orderto reach 10% obscuration. The casein micelleswere dissociated by adding 1 mL of 35 mMEDTA/NaOH, pH 7 buffer to the milks, inthe apparatus. In order to determine the sizeof fat globules in the renneted gel, 1 g of thesample was dissociated with 5 mL of disso-ciation buffer ( 6 mol·L–1 urea, 100 mmol·L–1

EDTA, 20 mmol·L–1 imidazole bufferpH 6.6), and stirred for 30 min at room tem-perature prior to measurement. The size dis-

tribution of fat globules were characterisedby the volume-weighted average diameterd43 defined as Σnidi

4/Σnidi3, where ni is the

number of fat globules of diameter di. Thisparameter is very sensitive to the presenceof small amounts of large particles. Thespecific surface area S = 6.ϕ/d32, where ϕis the volume fraction of milk fat and d32 isthe volume/surface average diameter definedas Σnidi

3/Σnidi2, was calculated. The size

distribution width, Span = (dv0.9 – dv0.1)/dv0.5, where dv0.9 is the diameter belowwhich lie 90% of the globule volume, andrespectively 10% for dv0.1 and 50% for dv0.5was calculated. The fraction Φ (expressedin percent) of the milk fat globule surfacethat is covered by milk proteins afterprocessing of commercial milks was calcu-lated as follows: Φ = [(Sf – Si)/Sf]·100,where Sf is the specific surface area of milkfat globules after processing and Si is thenatural milk fat globule specific surfacearea calculated from the fat globule size dis-tribution of raw milk.

The fat globule size distribution wasobserved by light microscopy using anOlympus B× 51 microscope.

2.3. Microstructure of cheese

The supramolecular structure of milk fatat different stages during the manufactureof Emmental cheese and in different kindsof cheeses was examined using confocallaser scanning microscopy (CLSM). Thinslices of samples, measuring approximately5 mm × 5 mm × 3 mm thick, were preparedfrom the freshly cut samples, using a scal-pel. The fat was stained using a lipid-solu-ble Nile Red fluorescent dye (Sigma-Aldrich, St Louis, USA). The protein net-work was stained using Acridine Orangefluorescent dye (Aldrich Chemical Com-pany, Inc., Milwaukee, USA). Each slice ofsamples was placed between a microscopeslice and a cover slip. The sample sliceswere thus incubated with the stains for30 min in the dark at 4 °C. Microstructuralanalysis was made using a confocal Leica

500 C. Lopez

TCS NT microscope (Leica, Microsystems,Heidelberg, Germany), which employed anargon/krypton laser in a dual-beam fluores-cent mode, with excitation wavelengths of568 nm and 488 nm for fat and protein,respectively. The two-dimensional imageshad a resolution of 1024 × 1024 pixels andthe pixel scale values were converted intomicrometers using a scaling factor.

3. RESULTS

3.1. On the size distribution of fat globules in fresh raw milk and drinking milks

Figure 1 shows the size distribution ofmilk fat globules in full-fat raw milk, full-fat pasteurised milk and full-fat UHT milkdetermined by laser light scattering andobserved by light microscopy. Particle sizedistribution parameters of the fat globulesare presented in Table I. The size of naturalfat globules dispersed in raw milk ranges

from 0.03 to about 11 µm, with two smallpeaks at 120 and 500 nm and a main peakcentred at 4 µm. Thus, the size distributionof native milk fat globules is polydispersedand multimodal. The volume-weighted diam-eter of natural fat globules in raw milk is4.07 µm. In the full-fat UHT and full-fatpasteurised milks, the size of milk fat glob-ules range from 0.03 to 2–3 µm with a vol-ume-weighted diameter around 0.4–0.5 µm(Tab. I). The shape of the size distributionchanged markedly between raw and heat-treated milks, with the disappearance of thepeak centred at about 4 µm. Furthermore, thelatter results show that the size of fat glob-ules was lower in drinking milks comparedto the size of natural milk fat globules in rawmilk. For a given concentration of fat in themilk, decreasing the diameter of fat glob-ules from 4 to 0.4–0.5 µm (10 to 12 fold),corresponded to an increase of the numberof fat globules with a multiplicative factorof 1000 to 1700. Thus, the specific surfacearea (S) of fat globules was larger for full-fat drinking milks than for raw milk; S

Figure 1. Milk fat globule size distribution in full-fat raw milk (ο), full-fat UHT milk (∆) and full-fat pasteurised milk (◊), measured by laser light scattering. The light micrographs corresponding tothe different milks are identified on the figure.

Organisation of fat in dairy products 501

increased from about 7 to 28–33 m2·g–1 fat(Tab. I). Furthermore, the fraction Φ of themilk fat globule surface that is covered bymilk proteins after processing of commer-cial milks was calculated; it correspondedto 74.5% for UHT full-fat milk and 78.6%for pasteurised full-fat milk (Tab. I), mean-ing that about 25% of the newly formed sur-face of milk fat globules may have beencovered by the MFGM. This increase in thesurface of milk fat globules and the changein the composition of the interface areimportant considering chemical and enzy-matic reactions which occur at the interface,such as lipolysis of triacylglycerols andthus the digestibility of fat globules.

Figure 2 shows the size distribution of fatglobules in UHT and pasteurised milks asa function of fat concentration. Particle sizedistribution parameters of the fat globulesare presented in Table I. Regarding UHTmilks, the size distribution of fat globulesranged from 0.03 to 2–2.5 µm in full-fatand semi-skimmed milks, with a volume-weighted diameter of 0.5–0.7 µm respec-tively (Fig. 2A). In the skimmed milk, whichcontained 0.1 g fat·100 g–1, the size of fatglobules ranged from 0.03 to 2 µm with avolume-weighted diameter of 0.17 µm(Fig. 2A). In the pasteurised milks, the sizeof fat globules ranged from 0.03 to 1 µm for

semi-skimmed milk and to 2 µm for full-fatmilk (Fig. 2B). The difference observedbetween full-fat and semi-skimmed milksmay have been induced by physical treat-ments applied during the process, i.e. skim-ming and standardisation in fat, which mayhave led to coalescence of fat globules.

3.2. On the supramolecular structure of milk fat in cheese

This study shows that the supramolecu-lar structure of milk fat in cheese is greatlyinfluenced by the technological processapplied. We focused on the impact of theheat and mechanical treatments on theorganisation of milk fat during the manu-facture of the most widely consumed hardcheese in France: Emmental cheese. Fur-thermore, the microstructures of differentkinds of cheeses were characterised. Thesupramolecular structure of fat in a com-plex matrix such as cheese was observedusing confocal laser scanning microscopy(CLSM).

3.2.1. Influence of heat and mechanical treatments on the organisation of milk fat in a complex matrix

Changes in the structure of milk fat glob-ules due to heat treatment of the curd grains

Table I. Fat concentration and physical characteristics of raw and commercial milks. Parameters ofthe size distribution of natural and processed milk fat globules determined from raw and commercialmilks: d43 = volume-weighted average diameter; Span = size distribution width; S = specific surfacearea. Fraction (Φ) of the milk fat globule surface that is covered by milk proteins after processing ofcommercial milks.

Samples

Heat treatment Fat content

Fat concentration

g·100 g–1

Size distribution parameters Φ(%)d43

§ (µm)

Span S§

(m2·g–1 fat)

Raw milk Full fat 3.8 4.07 ± 0.01 1.66 7.1 ± 0.3 –Pasteurised milk Full fat 3.5* 0.37 ± 0.01 2.43 33.2 ± 0.4 78.6Pasteurised milk Semi-skimmed 1.5* 0.26 ± 0.01 2.21 41.3 ± 0.3 82.8UHT milk Full fat 3.6* 0.49 ± 0.02 2.61 27.9 ± 0.3 74.6UHT milk Semi-skimmed 1.55* 0.71 ± 0.01 2.17 20.9 ± 0.6 66.0UHT milk Skimmed 0.1* 0.17 ± 0.02 1.77 56.0 ± 0.4 87.3

* Value indicated in the package of commercial milk. § n = 3.

502 C. Lopez

during the manufacture of Emmental cheesewere particularly studied (Fig. 3).

Figure 3A shows the microstructure ofthe rennet-induced gel formed during theearly stage of the manufacture of Emmentalcheese. Considering the physico-chemicalcharacteristics of the gel, the total solid con-tent was 116 g·kg–1, with 28.5 g·kg–1 of fatand 33.6 g·kg–1 of total protein. The micro-graph shows the continuous network formedby casein micelles after their coagulationand reveals the porous structure of the gelin which the fat globules are entrapped. Thefat globules are spherical individual parti-cles. The size of fat globules in the gel wasmeasured after dissociation of the caseinmatrix, as detailed in the material and meth-ods section. The size of fat globules observedby CLSM was in agreement with laser lightscattering experiments which calculated amean volume-weighted diameter of 4.47 ±

0.06 µm (Fig. 4). The size of the milk fatglobules entrapped in the gel was largerthan that of native milk fat globules in rawmilk, meaning that coalescence occurredduring the preparation and coagulation ofthe milk (Tab. I). Milk fat globules aremainly located in the serum pores of thecasein network. It is important to highlightthat such a rennet-induced gel is the startingpoint in the manufacture of many kinds ofcheeses.

After coagulation, the rennet-inducedgel characterised in Figure 3A was cut andheated from 32 °C to 51 °C during 20 min.Figure 3B shows the microstructure of thecurd grain after heating. The micrographclearly shows that heating of the rennet-induced gel induced the aggregation of milkfat globules. The aggregates observed onthe micrograph were formed by 2 to 4 fatglobules. Their size did not exceed 10 µm.

Figure 2. Milk fat globule sizedistribution in (A) full-fat (∆)semi-skimmed (◊) skimmed (+)UHT milks and (B): full-fat (∆)and semi-skimmed (◊) pasteur-ised milks.

Organisation of fat in dairy products 503

Aggregation of fat globules may favourtheir coalescence, e.g. fusion between fatglobules, by increasing the time in whichthey are in contact.

The comparison between the size distri-bution of fat globules before and after theheat treatment is presented in Figure 4. Theincrease in fat globule size was quantified

Figure 4. Milk fat globule size distribution in rennet-induced gel before heating (ο) and in a curdgrain after heating (∆), measured by laser light scattering.

Figure 3. Confocal laser scanning micrographs taken during the manufacture of Emmental cheese:(A) rennet-induced gel before heating, (B) curd grain after heating (51 °C, 20 min). FG: fat globules.The fat globule aggregates are circled.

504 C. Lopez

by laser light scattering. The volume-weighteddiameter increased from 4.47 ± 0.06 µm to5.18 ± 0.1 µm.

This study shows that the heat treatment(51 °C) of a rennet-induced gel modifies thesupramolecular structure of milk fat glob-ules by inducing their aggregation and theircoalescence. As all oil in water emulsions,milk is submitted to physical instabilities.

Many different mechanical treatmentsmay be applied to milk, coagulum and curdgrains during the manufacture of cheeses.In this study, we focused on the pressingstage applied to curd grains during the man-ufacture of hard cheeses.

Figure 5A shows the microstructure of aheated curd grain. In this micrograph, thefat appears as bright areas, the protein net-work as grey and the serum phase as black.The fat globules are entrapped in the caseinmatrix. They do not correspond to individ-ual spherical particles but rather to aggre-gates of milk fat globules. Figure 5B showsthe microstructure of Emmental cheese afterpressing. This mechanical treatment resultsin large changes in the composition and thestructure of the curd grains. The total solid

content increased from 401.9 ± 4.8 to 569.8 ±2.6 g·kg–1 (+30%), as a result of wheyremoval. Compared with Figure 5A, thedensity of the protein network and fatincreased. This can be explained as follows:after pressing of curd grains, the volume ofEmmental cheese decreased as a result ofthe compaction of curd grains. Moreover,dry matter increased from 402 to 570 g·kg–1.Briefly, after the pressing of curd grains, theEmmental cheese microstructure consistedof cavities containing fat globules and aserum phase, surrounded by protein strands.This micrograph clearly shows the disrup-tion of milk fat globules under pressure.The MFGM rupture and the absence of theMFGM around fat were not directly meas-ured in this study, but the change in theshape and the increase in the size of fat glob-ules observed may correspond to the forma-tion of free fat, which is the fat not protectedby the MFGM. Thus, after pressing, fat ispresent in the casein matrix of Emmentalcheese under different surpamolecular forms:(i) individual milk fat globules, which havea similar size as in cheese milk (∼ 4 µm),(ii) aggregates of fat globules, (iii) coa-lesced fat globules and (iv) non-globular

Figure 5. Confocal laser scanning micrographs of (A) the heated curd grains before pressing and(B) the curd grains of Emmental cheese after pressing (47 °C, 0.4 kPa, 4 h). Fat appears as brightareas, protein network appears as grey levels.

Organisation of fat in dairy products 505

fat, also called free fat. The parametersimplicated in the deformation, rupture andcoalescence of milk fat globule fat underpressure are discussed in Lopez et al. [23].

3.2.2. Comparison of the supramolecular structure of fat in different kinds of cheeses

The microstructure of cheese is influ-enced by the chemical composition and the

technological process applied during themanufacture. Figure 6 shows the microstruc-ture of different kinds of cheeses observedusing CLSM. In all the micrographs, the fatis coloured in red and the proteins appear ingrey levels.

Figure 6A corresponds to the micro-structure of the most widely manufacturedhard cheese in France, which is Emmentalcheese. Three main supramolecular structures

Figure 6. Confocal laser scanning micrographs of different kinds of cheeses: (A) Emmental cheese(B) Camembert cheese (C) whipped cream cheese (D) soft cheese made with ultrafiltration techno-logy. Fat is coloured in red, proteins are in grey levels. Black areas correspond to serum or gas holes.

506 C. Lopez

of milk fat are observed: (i) large non-glob-ular inclusions of fat resulting from the dis-ruption of MFGM and allowing free tria-cylglycerols to fill voids in the proteinmatrix, (ii) aggregates of partially disruptedfat globules and (iii) small fat globuleswhich may be enveloped by the MFGM(23). The casein network separating the fatinclusions may form impermeable barrierspreventing the coalescence of fat. The blackhole corresponds to a pocket of gas, whichis characteristic of propionic acid fermen-tation, leading to the formation of the“eyes” in Emmental cheese. The chemicalcomposition of the Emmental cheese char-acterised in this study was: 298 g·kg–1 fat,276 g·kg–1 protein and 628 g·kg–1 dry matter.

Figures 6B and 6D show the microstruc-ture of two different soft cheeses. Figure 6Bshows the microstructure of Camembertcheese, manufactured with traditional tech-nology. Fat is dispersed as fat globules witha size distribution larger than the size of nat-ural milk fat globules. These results meanthat the heat and mechanical treatmentsapplied during the manufacture of Camem-bert cheese induced some coalescence of fatglobules. Furthermore, aggregates of fatglobules were observed. Figure 6D showsthe microstructure of a low-fat soft cheesemade with the ultrafiltration (UF) technol-ogy. Fat is dispersed as small fat globules,∼ 0.5–1 µm, meaning that homogenisationwas used during the process. Homogenisa-tion of milk induced (i) a decrease of milkfat globule size and (ii) changes in the com-position of the fat globule surface with theadsorption of milk proteins, mainly caseins.As a consequence, the fat globules mayinteract with the protein phase during coag-ulation. Pockets of serum phase, whichappear as black areas, are also observed inthe micrograph. The gross chemical com-position of this product was the following:120 g·kg–1 fat, 210 g·kg–1 protein and380 g·kg–1 dry matter.

Figure 6C shows the microstructure of awhipped cheese. The composition of thiscommercial cheese was 250 g·kg–1 fat,

65 g·kg–1 protein and 385 g·kg–1 dry mat-ter. Two main supramolecular structures ofmilk fat were observed. They correspond to(i) small fat globules (∼ 0.5 µm) entrappedin protein aggregates, (ii) fat globules dis-persed in the serum. The very small fatglobules entrapped in the protein aggre-gates led to the superposition of white andred colours corresponding respectively tothe staining of protein and fat. In this kindof product, the phase which appears inblack in the micrograph may correspond toboth the air and serum phases.

This study shows that the supramolecu-lar structure of milk fat in cheese dependson the technology applied. Milk fat can beorganised as (i) native fat globules, (ii) aggre-gates of fat globules, (iii) small fat globuleswhich may be the result of homogenisation,(iv) coalesced fat globules, (v) free fat, or(vi) a combination of different phases andstructures (Fig. 7).

4. DISCUSSION

Our results were similar to those previ-ously obtained by Walstra [5] regarding thesize (d43 = 4.07 µm) and the specific surfacearea (S = 7.1 m2·g–1 fat) of natural milk fatglobules in raw milk. Mulder and Walstra[8] estimated that small natural fat globuleswith d < 1 µm comprise about 80% of thenumber of fat globules, but only a few per-cent of the total fat volume. Comparing thesize of milk fat globules in raw milk anddrinking milks, the results obtained in thisstudy show important differences. Theaverage fat globule sizes measured for UHTand pasteurised milks were lower than0.7 µm, whatever the concentration of fat.This value was lower than the size of nativemilk fat globules; about 4 µm. Thus, theresults obtained in this study confirmed thatpasteurised and UHT milks are high-pres-sure homogenised products.

In many parts of the world, consumerscan buy bovine milk from the farmer or insupermarkets as certified raw milk. How-ever, fresh raw milk which comes straight

Organisation of fat in dairy products 507

from the cow is hard to find nowadays. Milkis mainly transformed into drinking milks,which are at least homogenised and heattreated, or processed to obtain dairy prod-ucts. Drinking milks, with their wide diver-sity (heat-treated or not, skimmed, semi-skimmed or full fat, guaranteed vitamincontent, etc.), are the widely consumeddairy product in France [20].

Milk, directly from the cow, is rarelyconsumed due to fears regarding possiblemicrobiological risks and also the percep-tion of increased fat intake, due to the fatlayer that develops on the milk surfacewhen left to stand. Thus, milk undergoestechnological treatments. They are heattreated and homogenised to prevent theirmicrobiological and physical instability,respectively.

Homogenisation is of special interest inthe dairy industry in order to reduce the sizeof fat globules in milk and creams, and toprevent creaming and coalescence duringlong shelf-storage [11]. Homogenisationincreases the interfacial surface area of fatin solution by increasing the number of fatglobules. The increase in the fat surfacearea measured in this study was about 4- to8-fold (Tab. I). Such a tremendous increasealters the membrane composition whichcan no longer cover the fat globules entirely

and which may be partially lost in the serumphase. Thus, the newly formed interfacesuffers from a lack in membrane material tocover the surface. Adsorption of new mate-rial at the oil-water interface occurs to coverthis increase in the surface area. The frac-tion Φ of the milk fat globule surface thatis covered by milk proteins after processingof commercial milks was quantified in thisstudy; it corresponded to 66–87 %. (Tab. I).The amount of fat globules with a diametercentred at about 100 and 500 nm in rawmilk, increases in processed drinking milks.Part of these fat globules may be damagedduring homogenisation, but these fat glob-ules may also be newly formed “synthetic”milk fat globules. After homogenisation,the new membrane which surrounds fatglobules, called synthetic FGM, consists ofnative MFGM plus adsorbed milk proteins,with casein as the dominant group and wheyproteins, with mainly β-lactoglobulin [11,12, 14, 24, 25]. Cano-Ruiz and Richter [25]reported that caseins represent about 70%of the proteins in the synthetic FGM andthat approximately 10% of the proteinswere from the native MFGM after homog-enisation of the milk. These results wereconsistent with the results of Keenan et al.[14] who found that about 10% of the sur-face of homogenised fat globules in milk is

Figure 7. Schematic representation of the supramolecular structure of milk fat in dairy products(not to scale).

508 C. Lopez

covered by their native MFGM. McPhersonet al. [26], who analysed the composition ofmembrane material isolated from commer-cial pasteurised homogenised milks andUHT milks, mainly found caseins and wheyproteins. Their results are in agreement withstudies performed with heat treated andhomogenised milks. Considering the organ-isation of proteins, it has been shown thatthe milk fat globule surfaces were coveredwith (i) intact casein micelles, (ii) partiallyspread casein micelles, (iii) smaller parti-cles, which probably result from the desin-tegration of casein micelles and (iv) a thinlayer of proteins, which was probably wheyproteins [27, 28]. The sequence of homog-enisation and heating changes the structureand the composition of the synthetic FGM[25]. Houlihan et al. [29] found that heat-denaturated serum proteins interact withnative MFGM proteins.

The synthetic FGM formed after homog-enisation of milk have different character-istics and properties compared to the nativeMFGM. The composition and structure ofthis new membrane may be responsible forsome of the observed differences betweenhomogenised and non-homogenised milks.These include changes in the susceptibilityof the product to Cu and light-induced fla-vour deterioration, a reduction in the cream-ing rate and changes in the colour and fla-vour characteristics [8].

Considering the characterisation of thesupramolecular structure of fat in a com-plex matrix, the development of adequateprotocols and the choice of non destructivemethods is essential. It is important to high-light that the destabilisation of fat, whichmay occur during the manufacture ofcheese, and the formation of free fat, whichis the fat not protected by a membrane, doesnot allow the determination of the size of fatglobules by laser light scattering. Thus, wechose the CLSM, which is a useful tech-nique to characterise the supramolecularstructure of fat in a complex matrix. CLSMis a powerful tool to penetrate the surface ofa sample and to visualise thin optical sec-

tions. The use of CLSM provides an oppor-tunity to characterise milk fat globules incheeses, without disturbing the internal struc-ture [15–17, 30].

Regarding the structure of fat in dairyproducts, some elements are assumed. Inmilk, cream and yoghurt, fat is consumed asa direct emulsion with the fat globules beingdispersed in the plasma phase: fat globulesare covered by the MFGM, tiny homoge-nised fat globules are mainly covered bycaseins. In butter, fat is consumed as areverse emulsion in which water dropletsare dispersed in the continuous partiallycrystallised fat phase. In cheese, the fat isdispersed in a protein network, but itssupramolecular structure depends greatlyon the process. In spite of the great signifi-cance of the physical structure of milk fatand the interactions between fat and pro-teins, regarding lipase activity and the func-tional properties, only a few authors havestudied the organisation of fat in complexproducts.

This study showed that the milk fat glob-ules are submitted to all the physical insta-bilities of oil in water emulsions, e.g.creaming, aggregation, coalescence. As aconsequence, milk fat may be present inmany products in different structures, aspreviously detailed in this study. Theseresults were in agreement with Walstra [5],who found that many dairy products con-tain homogenised, recombined or destabi-lised fat, rather than natural milk fat glob-ules. The surface of fat globules may be(i) the MFGM, (ii) caseins, (iii) serum pro-teins, (iv) amphiphilic molecules such asphospholipids and MFGM proteins, (v) acombination of MFGM and milk proteins.Considering more precisely the cheeses,milk fat may have similar physico-chemi-cal properties as in raw milk, i.e. globuleswith d ∼ 4 µm and the presence of theMFGM, in soft cheeses such as Camem-bert.

Depending on the composition of theinterface, e.g. MFGM, monolayer of phos-pholipids, caseins, serum proteins or other

Organisation of fat in dairy products 509

proteins, milk fat interacts differently withthe protein matrix in which it is entrapped,leading to specific rheological properties ofgels [31–33]. Furthermore, both the supramo-lecular structure of fat, e.g. non-globular orglobular with various sizes, and the compo-sition of the interface, e.g. MFGM, caseins,may influence the diffusion of molecules(minerals, lactacte, …) in the cheese matrix.

The role of fat globules on the function-ality of different kinds of cheeses, such asCheddar and Mozzarella cheese has beenexplored but is still not fully understood.Gunasekaran and Ding [15] examined,using CLSM, the three dimensional charac-teristics of fat globules in one month oldCheddar cheese of varying fat contents.Guinee et al. [16] also examined the micro-structure of Cheddar cheeses. The size ofthe fat globules and their dispersion in thecasein matrix of Mozzarella cheese hasbeen shown to be related to meltability andfree-oil formation [34]. Furthermore, it hasrecently been shown in our laboratory thatthe size of native milk fat globules affectsthe physico-chemical and functional prop-erties of Emmental cheese [19] and Camem-bert cheese [18]. Furthermore, foaming andwhipping are strongly affected by proper-ties of the milk fat globules and the proper-ties of high-fat products may especially bedominated by those of the milk fat [5].

Sensory properties of milk fat have longbeen recognised and considered of impor-tance. Fat content and emulsion character-istics have a marked influence on the tex-ture, melting and mouthfeel, of most dairyproducts [21]. Moreover, milk fat influ-ences the flavour of dairy products. Themajority of flavour compounds are at leastpartially soluble in fat, suggesting that thefat acts as a reservoir for flavour com-pounds. Thus, the supramolecular structureof fat may facilitate flavour generation byproviding a fat-protein interface for fla-vour-producing reactions to occur. Further-more, the perception of fat in dairy productsand the release of flavour compounds maydepend on (i) the concentration of fat in the

product, (ii) the supramolecular organisa-tion of fat in the product, e.g. free fat orglobular fat and (iii) the composition andthe structure of the surface covering the fat.

Microstructural and physicochemicaldynamics of fat globules in cheese appearto influence the localisation and retention ofstarter lactococci in cheese [35]. However,to date, no detailed scientific investigationhas been undertaken to elucidate the mech-anism of accessibility of fat in cheese forlipolysis [36]. The enzymatic hydrolysis offat, e.g. lipolysis, may be influenced by(i) the interfacial surface area of fat and(ii) the accessibility of fat to enzymes.Since lipases are active on emulsified sub-strates, the specific surface area of the milkfat globules plays an essential role in lipol-ysis. The interfacial surface area of fat in adairy product depends on (i) the supramo-lecular organisation of fat: non-globular fat,globular fat and on (ii) the size of the milkfat globules: for a defined concentration offat, the lowest is the size of fat globules, thelargest is the interfacial area. Reducing thesize of fat globules by homogenisation makesthe fat more susceptible to fat hydrolysis bymeans of lipase enzymes; Homogenisationis widely used to make blue cheese in orderto enhance lipolysis. Moreover, the acces-sibility of fat to enzymes may depend on thecomposition of the surface covering the fat.It is known that the MFGM, in which areenveloped natural milk fat globules, is anatural physical barrier against lipolysis.Coalesced fat globules, resulting from thefusion of natural milk fat globules of smallersize are enveloped by the MFGM. Homog-enised fat globules are enveloped by theMFGM and milk proteins, mainly the caseins.The free fat is not protected by the MFGMbut the interface may be constituted byamphiphilic molecules such as (i) the pri-mary membrane of the MFGM, (ii) frag-ments of the MFGM, (iii) phospholipidscoming from the MFGM and (iv) proteins.

Dairy products are widely consumedduring meals and considering the enzy-matic hydrolysis of fat during the digestion

510 C. Lopez

process, it may be possible that the supramo-lecular structure of milk fat and its environ-ment (minerals, proteins) in dairy productsmay affect its digestibility and nutritionalproperties.

5. CONCLUSION

Bovine raw milk is the milk comingstraight from the farm before it is processed.The supramolecular structure of fat in rawmilk is the milk fat globule (∼ 4 µm), envel-oped in the specific biological membrane(the MFGM). Native milk fat globules dis-persed in fresh raw milk are the startingpoint for many dairy products but are rarelyconsumed.

At the dairy factory, raw milk is madeinto dairy products including drinking milk,creams, ice cream, yoghurt, butter, dairydesserts and all kinds of cheeses. The rela-tive fat content of dairy products varies enor-mously, from less than 0.5% in skimmedmilk and low-fat products through upwardsof 82% in butter. Due to dairy processes ofmilk, the structure of fat is greatly modifiedand the MFGM is damaged or absent inmost dairy products. High pressure homog-enisation of milk, which is the most widelyused mechanical process, induces a decreasein the size of fat globules and changes in thecomposition of the membrane stabilisingthe milk fat globules.

In this study, CLSM permitted the inves-tigation of the organisation of fat in a com-plex protein matrix, such as cheese. Weshowed that milk fat can be organised as:(i) fat globules covered by the MFGM,(ii) coalesced fat globules covered by theMFGM, (iii) tiny homogenised fat glob-ules, with d < 4 µm, mainly covered bycaseins, (iv) aggregates of fat globules,(v) non-globular fat (free fat). Natural milkfat globules, “synthetic” fat globules andnon-globular fat have different physico-chemical membrane composition. Due todifferences in the composition and struc-ture of the interface, milk fat may have dif-

ferent properties regarding the interactionswith the protein matrix and the retention ofwater. Thus, the supramolecular structureof fat in dairy products may influence sen-sorial, functional and nutritional properties.

ACKNOWLEDGEMENTS

The author would like to thank R. Primault(Microscopy department, University Rennes I)and M.-N. Madec for technical assistance inCLSM and light microscopy, respectively.A. Ayerbe and ARILAIT Recherches areacknowledged for the analysis done on the con-sumption of milk fat in France as well as for val-uable discussions.

REFERENCES

[1] Gurr MI. Nutritional significance of lipids. In:Fox PF (Ed), Advanced dairy chemistry,Vol 2, Lipids, 2nd ed, Chapman & Hall, 1995,p 349–402.

[2] Jensen RG, Blanc B, Patton S B. Particulateconstituents in human and bovine milk. In:Jensen R.G. (Ed), Handbook of milk fat com-position, Academic Press, New-York, 1995,p 50–62.

[3] Christie WW. Composition and structure ofmilk lipids. In: Fox PF (Ed), Advanced dairychemistry, Vol 2, Lipids, 2 nd ed, Chapman &Hall, 1995, p 1–32.

[4] Jensen RG, Newburg DS. Bovine milk lipids.In: Jensen RG (Ed) Handbook of milk com-position, Academic Press, Inc, New-York,1995, p 542.

[5] Walstra P. Physical chemistry of milk fatglobules. In: Fox PF (Ed), Advanced dairychemistry, Vol 2, Lipids, 2nd ed, Chapman &Hall, 1995, p 131–178.

[6] Gresti J, Burgaut M, Maniongui C, Bezard J.Composition of molecular species of triacylg-lycerols in bovine milk fat. J Dairy Sci 1993,76, 1850–1869.

[7] McPherson AV, Kitchen BJ. Reviews of theprogress of dairy science; the bovine milk fatglobule membrane. Its formation, composi-tion, structure, and behaviour in milk anddairy products. J Dairy Res 1983, 50: 107–133.

[8] Mulder H, Walstra P. In: The milk fat globule,Center for Agricultural publishing and docu-mentation, Wageningen, The Netherlands,1974.

Organisation of fat in dairy products 511

[9] Keenan TW, Dylewski DP. Intracellular ori-gin of milk lipid globules and the nature andstructure of the milk lipid globule membrane.In: Fox PF (Ed), Advanced dairy chemistry,Vol 2, Lipids, Chapman & Hall, London, UK,1995, p 89–130.

[10] Danthine S, Blecker C, Paquot M, InnocenteN, Deroanne C. Evolution des connaissancessur la membrane du globule gras: synthèsebibliographique. Lait 2000, 80: 209–222.

[11] Walstra P, Jenness R. In: Dairy Chemistry andPhysics, John Wiley & sons Publ Inc, NewYork, NY, 1984.

[12] Sharma SK, Dalgleish DG. Interactionsbetween milk serum proteins and synthetic fatglobule membrane during heating of homog-enized whole milk. J Agric Food Chem 1993,1407–1412.

[13] Kim HHY, Jimenez-Flores R. Heat-inducedinteractions between the proteins of milk fatglobule membrane and skim milk. J Dairy Sci1995, 78: 24–35.

[14] Keenan TW, Moon TW, Dylewski DP. Lipidglobules retain globule material after homog-enisation. J Dairy Sci 1983, 66:196–203.

[15] Gunasekaran S, Ding K. Three dimensionalcharacteristics of fat globules in cheddarcheese. J Dairy Sci 1999, 82: 1890–1896.

[16] Guinee TP, Auty MAE, Fenelon MA. Theeffect of fat content on the rheology, micro-structure and heat-induced functional charac-teristics of Cheddar cheese. Int Dairy J 2000,10: 277–288.

[17] Rowney MK, Roupas P, Hickey MW, EverettDW. The Effect of Compression, Stretching,and Cooking Temperature on Free Oil Forma-tion in Mozzarella Curd. J Dairy Sci 2003, 86:449–456.

[18] Michalski MC, Gassi JY, Famelart MH,Leconte N, Camier B, Michel F, Briard V. Thesize of native milk fat globules affects phys-ico-chemical and sensory properties of Cam-embert cheese. Lait 2003, 83: 131–143.

[19] Michalski MC, Camier B, Briard V, LeconteN, Gassi J., Goudedranche H, Michel F,Fauquant J. The size of native milk fat glob-ules affects physico-chemical and functionalproperties of Emmental cheese. Lait 2004, 84:343–358.

[20] Centre National Interprofessionnel de l’Écon-omie Laitière, L’économie laitière en chiffres,éd 2004, Paris, 2004.

[21] Mela DJ, Raats MM. In: Fox PF (Ed),Advanced Dairy Chemistry, Vol 2, Lipids,Chapman & Hall, 1994, p 403–432.

[22] Jaros D, Petrag J, Rohm H, Ulberth F. Milk fatcomposition affects mechanical and rheolog-ical properties of processed cheese. ApplRheol 2001, 11: 19–25.

[23] Lopez C, Camier B, Gassi JY. Evolution of themilk fat microstructure during the manufac-ture and ripening of Emmental cheese observedby confocal laser scanning microscopy. Sub-mitted to Int Dairy J, December 2004.

[24] Darling DF, Butcher DW. Milk-fat globulemembrane in homogenised cream. J DairyRes 1978, 45: 197–208.

[25] Cano-Ruiz ME, Richter RL. Effect of homog-enization pressure on the milk fat globulemembrane proteins. J Dairy Sci 1997, 80:2732–2739.

[26] McPherson AV, Dash MC, Kitchen BJ. Isola-tion and composition of milk fat globule mem-brane material. II. From homogenized andultra heat treated milks. J Dairy Res 1984, 51:289–297.

[27] Sharma R, Singh H, Taylor MW. Compositionand structure of fat globule surface layers inrecombined milk. J Food Sci, 1996, 61: 28–32.

[28] Hillbrick GC, McMahon DJ, McManus WR.Microstructure of indirectly and directlyheated ultra-high temperature (UHT) proc-essed milk examined using transmission elec-tron microscopy and immunogold labelling,Lebensm Wiss Technol 1999, 486–494.

[29] Houlihan AV, Goddard PA, Kitchen BJ,Masters CJ. Changes in structure of the bovinemilk fat globule membrane on heating wholemilk. J Dairy Res 1992, 59: 321–329.

[30] Blonk JCG, van Aalst H. Confocal scanninglight microscopy in food research. Food ResInt 1993, 26: 297–311.

[31] Xiong YL, Kinsella JE. Influence of fat glob-ule membrane composition and fat type on therheological properties of milk based compos-ite gels. II. Results. Milchwissenschaft 1991,46: 207–212.

[32] Lopez C, Dufour E. The composition of themilk fat globule surface alters the structuralcharacteristics of the coagulum. J ColloidInterface Sci 2001, 233: 241–249.

[33] Michalski MC, Cariou R, Michel F, GarnierC. Native vs. damaged milk fat globules:membrane properties affect the viscoelasticityof milk gels. J Dairy Sci 2002, 85: 2451–2461.

[34] Rowney M, Roupas P, Hickey M, EverettDW. Milkfat structure and free oil in mozza-rella cheese. Aust J Dairy Technol 1998, 53:110.

[35] Laloy E, Vuillemard JC, El Soda M, SimardRE. Influence of the fat content of Cheddarcheese on retention and localization of start-ers. Int Dairy J 1996, 6: 729–740.

[36] Collins YF, McSweeney PLH, WilkinsonMG. Lipolysis and free fatty acid catabolismin cheese: a review of current knowledge. IntDairy J 2003, 13: 841–866.