Food Chemistry 141 (2013) 2305–2313

Contents lists available at SciVerse ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Complexes between linoleate and native or aggregated b-lactoglobulin:Interaction parameters and in vitro cytotoxic effect

0308-8146/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.foodchem.2013.05.031

Abbreviations: ala, a-lactalbumin; blg, b-lactoglobulin; CLA, conjugated linoleicacid; CMC, critical micelle concentration; DMEM, Dulbecco’s modified Eaglemedium; FA, fatty acid; FAME, fatty acid methyl ester; FBS, foetal bovine serum;GC, gas chromatography; GP-HPLC, gel permeation high performance liquidchromatography; ITC, isothermal titration calorimetry; Ka, association constant;LA, linoleic acid; LCFA, long chain fatty acid; n, reaction stoichiometry; NATA, N-acetyl-tryptophanamide; PBS, phosphate buffered saline.⇑ Corresponding author at: AGROCAMPUS OUEST, UMR1253 STLO, 65 rue de

Saint Brieuc, F-35042 Rennes, France. Tel.: +33 223485927; fax: +33 223485350.E-mail addresses: Solene.Le.Maux@agrocampus-ouest.fr (S. Le Maux), Said.Bou-

hallab@rennes.inra.fr (S. Bouhallab), Linda.Giblin@teagasc.ie (L. Giblin), Andre.B-rodkorb@teagasc.ie (A. Brodkorb), Thomas.Croguennec@agrocampus-ouest.fr (T.Croguennec).

Solène Le Maux a,b,c, Saïd Bouhallab a,b, Linda Giblin c, André Brodkorb c, Thomas Croguennec a,b,⇑a INRA, UMR1253 STLO, 65 rue de Saint Brieuc, F-35042 Rennes, Franceb AGROCAMPUS OUEST, UMR1253 STLO, 65 rue de Saint Brieuc, F-35042 Rennes, Francec Teagasc Food Research Centre, Moorepark, Fermoy, Co. Cork, Ireland

a r t i c l e i n f o

Article history:Received 8 February 2013Received in revised form 2 April 2013Accepted 8 May 2013Available online 17 May 2013

Keywords:b-LactoglobulinLinoleateInteractionAggregationCytotoxicity

a b s t r a c t

The dairy protein b-lactoglobulin (blg) is known to form a complex with fatty acids (FA). Due to industrialprocessing, blg is often in its non-native form in food products, which can modify the FA/blg complexproperties. We investigated the interaction of bovine blg, in selected structural forms (native blg, a cova-lent dimer and as nanoparticles), with linoleate (C18:2). Using fluorescence and Isothermal Titration Cal-orimetry, linoleate was found to bind blg at two different binding sites. Regardless of the structural stateof blg, association constants remained in the same order of magnitude. However, the stoichiometryincreased up to 6-fold for nanoparticles, compared to that of native blg. The impact of these structuralchanges on linoleate uptake in vitro was measured by cytotoxicity assays on Caco-2 cells. The order ofcytotoxicity of linoleate was as follows: free > complexed to dimers > complexed to nanoparticles > com-plexed to native blg. Therefore, the in vitro cytotoxicity of linoleate could be modulated by altering thestate of blg aggregation, which in turn affects its binding capacity to the FA.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

b-Lactoglobulin (blg), the major whey protein in bovine milk, ispresent in a large number of food products. blg is a member of thelipocalin family, composed of 162 amino acids with a monomericmolecular weight of 18.4 kDa (Braunitzer, Chen, Schrank, & Stangl,1973). It contains nine b-strands, labelled from A to I, and a threeturn a-helices, that are arranged to form a globular protein struc-ture (Creamer, Parry, & Malcolm, 1983; Sawyer & Kontopidis,2000). Eight antiparallel b-strands are organised in a b-barrel,shaped into a hydrophobic calyx. Under physiological conditions,native blg exists as a non-covalent dimer/monomer in equilibrium.However, the blg structure is highly sensitive to processing condi-

tions used in the food industries, especially heat treatments, thatare applied during food manufacture to achieve specific food tex-tures or to reduce the microbial load (Considine, Patel, Anema,Singh, & Creamer, 2007; de Wit, 2009). Such treatments denaturenative blg, leading to the formation of non-native monomers andaggregates of blg in food products (de Wit, 2009).

blg is able to bind to small hydrophobic molecules, such as fattyacids (FA) (Sawyer and Kontopidis 2000), and the formation of suchcomplexes modifies FA digestion (Perez et al., 1992). It has beensuggested that native blg binds to hydrophobic ligands in its inter-nal calyx and on surface binding sites (Wu, Pérez, Puyol, & Sawyer,1999; Yang et al., 2008). However, FA binding to blg is sensitive tothe physicochemical conditions of the medium. Several studies re-lated the decrease of association constants between blg and bind-ing FA with a decrease in pH. Indeed, below pH 6.2, the calyxbinding site is closed by the EF loop region, decreasing interactionwith the hydrophobic components (Ragona et al., 2000). Addition-ally, Wang, Allen, and Swaisgood (1998) demonstrated that a de-crease in the proportion of native blg dimer increased the blgaffinity constant for the palmitate. A number of studies have as-sessed the interaction of ligands with heat treated blg (O’Neilland Kinsella, 1988; Yang et al., 2008). However, these differentstudies have shown inconsistent changes in the binding constantsof such ligands with heat treated blg compared to the native form.This may be due to the nature of the ligand, or to differences in theapplied heat treatments (O’Neill and Kinsella, 1988; Yang et al.,

2306 S. Le Maux et al. / Food Chemistry 141 (2013) 2305–2313

2008). In fact, the aggregates differ in the parts of protein exposedand therefore differ in how they react to heat (de Wit, 2009).

The essential long-chain fatty acid (LCFA) linoleic acid (LA, cis,-cis-9,12-octadecadienoic acid, n � 6, 18:2) constitutes 1–3% (w/w)of the total FA found in bovine milk fat (Jensen, 2002). LA serves asan essential precursor to a number of long chain metabolites(Mantzioris, James, Gibson, & Cleland, 1995; Russo, 2009). Itshealth benefits include anti-inflammatory effects, improvementsin serum lipoprotein profiles and a reduced risk of cardiovascularcoronary artery disease (Zhao et al., 2005; Zock & Katan, 1998).Furthermore, LA, at high concentrations, is cytotoxic to cancerouscells in vitro (Lu et al., 2010). However, the bioaccessibility of FAis altered according to the structure of the food matrix (Le Maux,Giblin, Croguennec, Bouhallab, & Brodkorb, 2012; Mu, 2008; Singh,Ye, & Horne, 2009). We previously demonstrated an interaction be-tween the water soluble form of LA, linoleate, and native blg (LeMaux et al., 2012). This binding alters the cytotoxicity of linoleateby decreasing its transport into the cell.

As blg is often in a non-native form in food products, the aim ofthis present work was to determine whether blg structural formsalter the blg/linoleate interaction and consequently linoleate cyto-toxicity, which is indication of its transport into the cell. Therefore,selected blg aggregates of controlled size, covalent dimers andnanoparticles, were formed. The binding properties of native blg,covalent dimers and nanoparticles with linoleate were measuredby both isothermal titration calorimetry and intrinsic fluorescence.The cytotoxicity of linoleate either, free in solution or in com-plexes, was measured for a better understanding of the proteinstructure impacts on the FA transport.

2. Materials and methods

2.1. Materials

blg (96% purity) was obtained from Davisco Foods International,Inc. (Eden Prairie, Minnesota) and sodium linoleate (purity P98%)from Sigma–Aldrich (St. Louis, MO). All other chemicals and solu-tions were purchased from Sigma–Aldrich unless stated otherwise.

2.2. Protein sample preparation and characterisation

2.2.1. Formation of b-lactoglobulin dimers and nanoparticlesCovalent dimers of blg were formed using the protocol reported

by Gulzar, Croguennec, Jardin, Piot, and Bouhallab (2009). Briefly,blg was dissolved in a 5 mM Bis–Tris buffer (pH 6.7), the final pro-tein concentration was 5 g/L. Copper chloride (CuCl2) was added tothe blg solution, at a Cu2+/blg molar ratio of 0.6. The solution washeated at 80 �C for 30 min to form covalent dimers, then cooledon ice. Covalent dimers were first dialyzed against 10 mM NaCl(dialysis baths were changed every hour for 4 h) and then againstdistilled water for 48 h (the water bath was changed twice). Sam-ples were then freeze-dried and stored at �20 �C prior toexperiments.

Nanoparticles of blg were formed according to the method ofSchmitt et al. (2009) Briefly, blg was dissolved in Milli-Q water(Millipore, Carrigtwohill, Ireland), to a final protein concentrationof 10 g/L. The pH of the protein solution was adjusted to 5.9 using1 M HCl, before heating the solution at 85 �C for 15 min, and thenrapidly cooling on ice. Samples were dialysed for 48 h against anexcess of distilled water, freeze-dried and stored at �20 �C priorto experimental use.

2.2.2. Characterisation of native b-lactoglobulin, covalent dimers andnanoparticles2.2.2.1. Quantification of b-lactoglobulin concentration in reconsti-tuted solutions. The concentration of native blg and covalent

dimers (expressed as monomer) were determined by optical den-sity using the extinction coefficient of blg at 278 nm(e278 = 0.96 L/g/cm).

For nanoparticles, the concentration of the blg monomers wasquantified on a reduced sample by the Bradford test followingthe manufacturer’s instructions (Sigma–Aldrich). For reduction,470 ll of the nanoparticle sample (1 mg of powder/ml) wasdissolved in phosphate buffered saline (PBS; 0.01 M phosphatebuffer, 2.7 mM KCl, 137 mM NaCl, pH 7.4), 5 ll of 10% SDS and25 ll b-mercaptoethanol, and the mixture was heated at 95 �Cfor 5 min.

2.2.2.2. Characterisation of b-lactoglobulin samples using gel perme-ation-HPLC. The proportion of monomers, dimers, oligomers andaggregates in blg samples were determined by gel permeation-HPLC (GP-HPLC) using a TSK G SW guard column (7.5 � 7.5 mm,Tosoh Bioscience GmbH, Stuttgart, Germany) and a TSK G2000SW column (7.5 � 600 mm, Tosoh Bioscience GmbH) connectedto an HPLC system, consisting of a Waters 2695 Separations Mod-ule (Waters, Milford, MA) and a Waters 2487 Dual k AbsorbanceDetector (Waters) working at 280 nm, using Empower Pro soft-ware (Waters), to acquire and analyse data. Solvent with 30% (v/v) acetonitrile (LabScan Analytical Sciences, Dublin, Ireland) and0.1% (w/v) trifluoracetic acid in Milli-Q water was used for proteinelution at a flow rate of 0.5 ml/min. The molecular-weight of thedifferent molecular entities, in the samples, was determined usinga protein molecular-weight standard calibration set (Sigma–Aldrich).

The molecular entities present in each blg sample were deter-mined as follows: solutions of native blg, covalent dimers andnanoparticles were prepared at 1 g/L in PBS. Nanoparticle solutionswere centrifuged at 12,000�g in order to separate the nanoparti-cles (pellet) from smaller molecular entities (supernatant). Solu-tions of native blg, covalent dimers and the supernatant ofnanoparticle solutions, were filtered (0.22 lm filter) prior to injec-tion onto the GP-HPLC. The proportions of monomers, dimers andhigher size oligomers of blg were determined from their relativeGP-HPLC chromatographic peak area obtained using Apex Trackintegration, and the sample total chromatographic peak area. Theproportion of monomers and aggregates in the nanoparticle sam-ples were determined from their chromatographic peak area inthe supernatant of the nanoparticle sample and the total chro-matographic area of a solution of native blg prepared at 1 g/L.The proportion of the different molecular entities for each of theblg samples (native blg, covalent dimers and nanoparticles) andof a-lactalbumin (ala, impurity) were calculated. The native blgsample contains 84.6 ± 1% monomers, 5.4 ± 0.5% dimers,5.4 ± 0.4% oligomers and 4.6 ± 0.4% of ala. The covalent dimer sam-ple has 74.4 ± 3.1% of dimers, 15.5 ± 1.4% of residual monomers,6.5 ± 1.6% of oligomers and 3.6 ± 0.4% of ala. The nanoparticle sam-ple has 77.6 ± 1.4% of aggregates and 22.4 ± 1.4% of monomers.

2.2.2.3. Mean hydrodynamic diameter of nanoparticles. To check thehomogeneity of the preparation, the mean hydrodynamic diameterof the aggregates in the nanoparticle sample was measured by dy-namic light scattering using a Zetasizer Nano ZS (Malvern Instru-ments Ltd., Malvern, Worcestershire, UK), equipped with a 4 mWhelium/neon laser at a wavelength output of 633 nm. Particle siz-ing was performed at 25 �C at 10 s intervals in a particle-sizing cell,using backscattering technology at a detection angle of 173�. Re-sults were the mean of 13 runs. The intensity of light, scatteredfrom the particles, was used to calculate the mean hydrodynamicdiameter (z-average mean), based on the Stokes–Einstein equation,assuming the particles to be spherical. The mean hydrodynamicdiameter of the aggregates (nanoparticles) was centered around130 nm (data not shown).

S. Le Maux et al. / Food Chemistry 141 (2013) 2305–2313 2307

2.3. Linoleate/b-lactoglobulin structure interaction

2.3.1. Isothermal titration calorimetryIsothermal titration calorimetry (ITC) was used to determine the

interaction parameters between the different forms of blg and lino-leate. ITC experiments were performed on a VP-ITC microcalorime-ter (Microcal, Northampton, MA). Solutions of blg (0.027 mM) andlinoleate (1.65 mM) in PBS were degassed under vacuum beforetitration experiments. The reference cell was filled with PBS, andthe sample cell (1.43 ml) was filled with blg solution. blg was ti-trated at 25 �C with 29 successive linoleate injections of 10 ll.The injection time was 20 s, and the time between injections wasfixed at 600 s to achieve thermodynamic equilibrium. During titra-tions, the solution in the sample cell was stirred at 310 rpm to en-sure complete mixing. The control measurement was obtained bytitrating sodium linoleate into PBS buffer, using the same injectionprocedure. The control measurement was subtracted from the blgtitration with linoleate and the first injection peak was systemati-cally ignored for the data analysis. The data were analysed usingMicroCal ORIGIN version 7.0 (Microcal). The integrated area of eachpeak was plotted versus the linoleate/blg monomer molar ratio. The‘‘two sets of binding sites’’ model was the best fit for all experi-ments, providing the binding parameters Ka1, Ka2, n1, and n2 (Ka

and n are the association constant and the stoichiometry, respec-tively). Each measurement was performed in triplicate.

2.3.2. Intrinsic fluorescenceIntrinsic fluorescence spectra were recorded at 345 nm, using

an excitation wavelength of 278 nm. For each titration, a fluores-cence spectrum was recorded from 300 to 450 nm in order to ob-serve deviation in the fluorescence properties of the protein.Experiments were performed at 25 �C on a SPEX 112 spectrofluo-rometer (Jobin–Yvon, Longjumeau, France), using 10 � 10 mmquartz cuvette. Excitation and emission slits were both set to5 nm. blg solutions in PBS (3 ml at 10 lM) were titrated with suc-cessive 3 ll injections of 5 mM linoleate, up to a linoleate/blg mo-lar ratio of 10. The solution was agitated and a 5 min equilibriumtime was respected prior to each measurement. An N-acetyl-tryp-tophanamide (NATA) blank was titrated following the same proce-dure in order to subtract the inner filter effect caused by the FA.NATA fluoresces similarly to tryptophan but does not bind FA (Co-gan, Kopelman, Mokady, & Shinitzky, 1976). The concentration ofNATA was chosen to have the same initial fluorescence (withoutFA) as the fluorescence of blg solutions. The fluorescence of NATAwas subtracted from fluorescence intensity measurements of theligand/protein complexes for all the linoleate/blg molar ratiostested. Each measurement was performed in triplicate. Fluores-cence data were fitted using two different methods.

In method 1, Lfree, Ltotal and Lbound represent the concentration offree, total and bound linoleate, respectively, Ptotal is the concentra-tion of blg, m is the fraction of linoleate molecules bound per moleof protein (m varies from 0 to n), n the number of linoleate bound toblg at saturation (number of sites), and fi the fraction of one site ofthe protein to be occupied by a ligand (fi varies from 0 to 1). Then:

Ltotal ¼ Lfree þ Lbound ð1Þ

m ¼ Lbound

Ptotal¼ nfi ð2Þ

Combining Eqs. (1) and (2) we deduce that:

Ltotal ¼ Lfree þ nPtotalfi ð3Þ

The value of fi is determined using the initial fluorescence inten-sity (F0), the fluorescence intensity at saturation (Fmax) and thefluorescence intensity at the ratio ligand/protein i (fi) as indicatedin Eq. (4):

fi ¼F i � F0

Fmax � F0ð4Þ

When Fmax was not reached experimentally, it was determinat-ed by fitting using an exponential phase decay model on Graph-Pad Prism software. The value of n was determined by plottingLtotal as a function of Ptotalfi. The data was fitted using a sequentiallinear regression in Graph-Pad Prism software 3.03 (GraphPadSoftware Inc., La Jolla, CA).

Method 2 is an adaptation of the Scatchard plot. In the Scat-chard plot described below, Ka is the association constant:

mLfree¼ nKa � mKa ð5Þ

Eqs. (3) and (5) can be rearranged as:

Ptotalð1� fiÞ ¼Ltotal

n1fi� 1

� �� 1

nKað6Þ

By fitting this equation using Graph-Pad Prism software, n and Ka

were determined.

2.4. Preparation of linoleate/b-lactoglobulin complexes for biologicalassay

2.4.1. Preparation of complexesLinoleate/blg complexes were prepared by mixing blg samples

with sodium linoleate according to Lišková et al. (2011) with mod-ifications as described in Le Maux et al. (2012). Briefly, 0.163 mMblg, in its native form, covalent dimers or nanoparticles, were dis-solved in PBS, and sodium linoleate was added to reach final lino-leate/blg molar ratios of 5, 7.5 or 10. Solutions containing nativeblg were heated at 60 �C for 30 min to facilitate blg/linoleate inter-action and rapidly cooled on ice. Solutions containing covalent di-mers or nanoparticles were mixed overnight at room temperature.Samples were dialysed against distilled water for 72 h with dialysisbags of nominal cut-off of 3500 Da. Samples were freeze-dried andpowders stored at �20 �C prior to experiments.

2.4.2. Determination of fatty acid content by gas chromatographyThe FA content of the complexes was determined by gas chro-

matography (GC) following a protocol adapted from Palmquistand Jenkins (2003) and Coakley et al. (2003) and described in detailpreviously (Le Maux et al., 2012). Briefly, the internal standard tri-decanoic acid (C13:0) was added to �4 mg of complexes. FA wereconverted to fatty acid methyl esters (FAME) and were analysedusing a CP-SELECT CB column for FAME (100 m, 0.25 mm,0.25 lm film thickness, Varian BV, Middelburg, The Netherlands),adaptated on a Varian 3400 GLC (Varian, Walnut Creek, CA) con-nected to a flame ionisation detector.

2.4.3. Complexes analysis by polyacrylamide gel electrophoresisSamples were analysed by sodium dodecyl sulphate polyacryl-

amide gel electrophoresis (SDS–PAGE). Mini-PROTEAN TGX precastGels (4–20% resolving gel, Bio-Rad Laboratories Inc., Hercules, CA)were used on a Mini Protean II system (Bio-Rad) according to themanufacturer’s instructions. Samples were prepared under non-reducing (in the absence of b-mercaptoethanol) and reducing (inthe presence b-mercaptoethanol) conditions. The protein was vis-ualised by staining with Coomassie blue (Bio-Safe Coomassie StainG-250, Bio-Rad). An Amersham Low Molecular Weight Calibrationkit (14.4–97 kg/mol, GE Healthcare UK Limited, UK) was used asmolecular weight standards.

2308 S. Le Maux et al. / Food Chemistry 141 (2013) 2305–2313

2.5. Cell Culture and cytotoxicity assay

The Caco-2 cell line was purchased from the European Collec-tion of Cell Cultures (collection reference: ECACC 86010202). Itwas derived from human colonic adenocarcinoma cells and canmimic the enterocytes of the intestine.

Cells cultures were maintained in a humidified 37 �C incubatorwith a 5% CO2 atmosphere. Cells were routinely grown in 75 cm2

plastic flasks in Dulbecco’s modified Eagle medium (DMEM), con-taining 4.5 g/L glucose and 0.584 g/L L-glutamine. Media for sub-culture was supplemented with 10% (v/v) foetal bovine serum(FBS), 100 U/ml penicillin and 100 mg/ml streptomycin. At 80%confluency, cells were trypsinised with 0.25% trypsin/EDTA, di-luted 1:6 in media and reseeded. The growth medium was changedthree times a week. All cells used in these studies were betweenpassage number 20 and 31.

The cytotoxicity of test samples on Caco-2 cell proliferation wasdetermined by an MTS assay, using a CellTiter 96 Aqueous OneSolution Cell Proliferation Assay according to the manufacturer’sinstructions (Promega Corporation, Madison, WI), as previously de-scribed in Le Maux et al. (2012). Briefly, 96-well plates were seededwith 2 � 104 Caco-2 cells/well, using serum-free media. After 24 h,the cells were treated with different concentrations of linoleate (0–200 lM) or linoleate/blg complexes (higher linoleate/blg complexwhich contained 0–200 lM linoleate as determined by GC) in ser-um-free media for 24 h. After the use of One Solution Cell Prolifer-ation reagent, viability was defined as the ratio of absorbance oftreated cells to untreated cells (cells exposed to serum-free Mediaonly) at 490 nm. Cells exposed to the different controls of blg weresubtracted to the corresponding samples. Each cell exposure wasperformed in triplicate.

The Lethal Dose 50 (LD50) values, the concentration required todecrease the cell viability by 50%, were determined using Graph-Pad Prism software 3.03 (GraphPad). The sigmoidal dose–responsewith a variable slope was used to fit the measured curves and cal-culate LD50.

2.6. Statistical analysis

Where appropriate, results were statistically analysed using theR software package version 2.15.1 (R Foundation for StatisticalComputing, Vienna, Austria) and the ANOVA system with a Tukey’sleast significant difference comparison. P-Values less than 0.05were deemed to be statistically significant.

3. Results

3.1. Binding properties of the different b-lactoglobulin forms withlinoleate

Binding parameters, determined at 25 �C using ITC and intrinsicfluorescence spectroscopy, were expressed on the basis of blgmonomeric units.

ITC data revealed an exothermic signal for the interaction be-tween linoleate and all the blg forms tested. Increasing the amountof linoleate in the titration cell resulted in a progressive decrease ofthe exothermic signal, due to the saturation of the binding sites(Fig. 1A). Regardless of the states of blg aggregation, the data wasbest fitted with two sets of binding sites model. The number ofbinding sites for each set of binding sites (n) and the correspondingassociation constant (Ka) could be determined from the fittedcurves (Table 1). Similar association constants were observed forall blg forms for each set of binding sites. The Ka values for the firstand second sets of binding sites were close to 106 M�1 and104 M�1, respectively. The molar ratio of linoleate bound to blg

monomer (n) varied for the first set of binding sites between0.53 ± 0.08 for covalent dimers and 0.92 ± 0.29 for nanoparticles.For the second set of binding sites, n varied somewhat more: nativeblg (6.79 ± 0.05), covalent dimers (8.64 ± 0.54) and nanoparticles(10.3 ± 1.64).

Intrinsic fluorescence titration is based on the change in theintensity of blg tryptophan fluorescence. The maximum emissionwavelength was 345, 353 and 350 nm for native blg, the covalentdimers and the nanoparticles, respectively; therefore aggregatedblg caused a red shift. However, the fluorescence spectra had asimilar shape for all the blg forms tested and the changes in fluo-rescence intensity consecutive to linoleate addition were corre-lated at the three wavelengths. Therefore the fluorescencechanges were followed at 345 nm, which is the wavelength ofmaximal fluorescence intensity of the native protein. In the titra-tion range used in this study, the change in fluorescence intensityreached a maximum of 10.5 ± 1.3%, 21.7 ± 1.6% and 32.2 ± 2.1%from the initial fluorescence intensity for native blg, the covalentdimers and nanoparticles, respectively (Fig. 1B). Increasing thelinoleate concentration in native blg samples induced an increasein the fluorescence intensity at 345 nm. This increase levels offwhen the linoleate/blg molar ratio reaches 3. In contrast, the fluo-rescence intensity of the covalent dimers and of the nanoparticlesdecreased continuously up to a linoleate/blg molar ratio of 10. Foreach titration, fluorescence data was fitted with two differentmodels.

In the first model, the total concentration of linoleate is plot-ted as a function of the total concentration of protein and varia-tion in fluorescence intensity (Ptotal�fi). It gave access to thenumber of binding sites (n), which are determined from the slopeof the graphical representation. For the entire titration, thegraphical representation can be fitted with two straight lines,indicating the presence of two sets of binding sites (Table 1).The number of binding sites varied according to the blg forms.From linoleate/native blg to linoleate/nanoparticles complexes,n1 increased from 2.38 ± 0.12 to 15.7 ± 0.55 and n2 from6.02 ± 0.29 to 40.7 ± 2.17.

The second model was an adaptation of the Scatchard plot, inwhich the maximum fluorescence (Fmax) was required for the plotconstruction. However, Fmax was not reached with a 10 linoleate/blg molar ratio for the complexes made with the covalent dimersand the nanoparticles. Therefore, the fit of the Scatchard plot wasobtained using the experimental Fmax for the complex made oflinoleate and native blg. An extrapolated Fmax was used for theScatchard plot of the linoleate/covalent dimers complex (Table 1).Unfortunately, the fluorescence data for the linoleate/nanoparti-cles complexes could not be fitted correctly using an extrapolatedFmax. The thermodynamic constants (Ka) for the two sets of bindingsites were 9.20 ± 2.65 � 105 M�1 and 0.62 ± 0.49 � 105 M�1 for thelinoleate/native blg complex and 14.7 ± 2.12 � 105 M�1 and0.37 ± 0.13 � 105 M�1 for the linoleate/covalent dimers complex.These values of association constants were in the same range asthose deduced from the ITC data. The stoichiometry n1 was2.45 ± 0.07 and 10.3 ± 0.05 while n2 was 5.27 ± 1.5 and15.3 ± 0.71 for the linoleate/native blg and linoleate/covalent di-mers complexes, respectively.

3.2. Changes in the structure of the linoleate/b-lactoglobulin complexes

Complexes of linoleate with native blg, covalent dimers andnanoparticles were analysed by SDS–PAGE and GP-HPLC in orderto identify changes in the aggregation state of blg following linole-ate interaction. Previously, we demonstrated that native blg aggre-gated into dimers and oligomers in the presence of linoleate (LeMaux et al., 2012). Fig. 2A confirms this observation with SDS–PAGE analysis of native blg, under non-reducing conditions, show-

Fig. 1. Binding association of linoleate/blg with different forms of blg obtained by ITC and fluorescence. (A) For the ITC experiments, linoleate was titrated in different forms ofblg (native, covalent dimers and nanoparticles) in PBS buffer (pH 7.4) at 25 �C. blg (0.027 lM) was titrated with increments of 10 ll linoleate (1.65 lM). Results represent theintegrated raw heat signals plotted against the linoleate/blg molar ratio. (B) For the intrinsic fluorescence experiments, linoleate (5 mM) was titrated in 10 lM blg (native,covalent dimers and nanoparticles) at 25 �C. Results represent the fluorescence at 345 nm corrected by the blank (NATA). , linoleate/native blg; , linoleate/covalentdimers; , linoleate/nanoparticles. Results represent mean ± SD (n = 3).

S. Le Maux et al. / Food Chemistry 141 (2013) 2305–2313 2309

ing a major band corresponding to the blg monomer with smallamount of dimers and trimers. The presence of linoleate increasesthe amount of blg dimers and oligomers at the expense of the blgmonomers. In contrast, the presence of linoleate had almost no ef-fect on the covalent dimers except a slight decrease in the intensityof the residual blg monomer band (Fig. 2B). A similar result is ob-tained for the SDS–PAGE of the complexes made with nanoparti-cles (Fig. 2C). In this latter case, the nanoparticles did not enterthe separation gel because of their high size. Under reducing con-ditions, SDS–PAGE for all the complexes and the blg controls (with-out linoleate) were similar. Fig. 2D is a representation of theseresults depicting nanoparticles and linoleate/nanoparticles com-plexes prepared at three different linoleate/blg molar ratio (5, 7.5or 10), under reducing conditions. Taking the non-reducing andreducing results together, linoleate induced aggregation of blg isstabilised by intermolecular disulphide bonds.

GP-HPLC chromatograms of the complexes formed with nativeblg, covalent dimers and nanoparticles were integrated and theproportion of blg monomers, dimers and oligomers (trimers andtetramers) as a function of the initial linoleate/blg molar ratio areshown in Table 2. A decrease in the concentration of monomersin the presence of linoleate were observed for all the blg forms,in agreement with the SDS–PAGE experiments. The monomericproportion decreased from 88.5 ± 5.2% to 51.1 ± 4.9% using nativeblg, from 16.3 ± 1.5% to 13.4 ± 0.6% for complexes using covalentdimers and from 22.4 ± 1.4% to 10.6 ± 1.9%, for complexes usingnanoparticles with an initial molar ratio of linoleate/blg varyingfrom 0 to 10. Concomitantly, an increase of the protein aggregationwas also observed. As predicted the difference in aggregation byincreasing the linoleate/blg molar ratio was more pronounced fornative blg than the other forms of blg assayed.

3.3. Cytotoxicity of linoleate bound to the different forms of b-lactoglobulin

The effect of linoleate (0–200 lM), bound to the different formsof blg, on Caco-2 cell viability was measured. For quantifying theeffect of the bound linoleate only, the complexes were dialysedto remove unbound linoleate. After dialysis, the exact stoichiome-try of the linoleate/blg complexes was determined from the freeze-dried complexes using GC (Fig. 3). The amount of linoleate boundto blg increased when the initial linoleate/blg molar ratio was in-creased. This increase varied depending on the blg form with morelinoleate binding increasing in the order of nanoparticles > cova-lent dimers > native blg. Only the complexes prepared with thehigher linoleate/blg molar ratio were used for cytotoxicity experi-ments (Fig. 4). No cytotoxic effect was detected for any of the blgforms used at the concentrations assayed when employed in theabsence of linoleate (data not shown). Free linoleate has a LD50

of 58.0 ± 4.2 lM (Le Maux et al., 2012). Comparatively, the LD50

of the complexes were all significantly different (p < 0.001). Thelinoleate/native blg complex was not cytotoxic to Caco2 cells atthe concentrations tested (LD50� 200 lM complex). The LD50

was 80.0 ± 3.1 lM for the linoleate/covalent dimers complex, and189 ± 4.1 lM for the linoleate/nanoparticles complex.

4. Discussion

The structural state of blg modified its binding properties tolinoleate. This was demonstrated using blg intrinsic fluorescenceand ITC measurements albeit the determined stoichiometry ofthe two techniques differed slightly. The number of binding sitesdetermined from ITC data for the interaction between linoleate

Table 1Binding constants of linoleate/blg with different forms of blg determined by ITC and fluorescence. For ITC, the association constant, Ka and molar ratio n of linoleate/blg werederived using a ‘‘two set of binding sites’’ model. For fluorescence, two methods of fitting were used. Association constants Ka and molar ratio n of linoleate/blg were determinedusing a modified Scatchard method. Experiments with linoleate binding to blg nanoparticles could not be fitted using the modified Scatchard method (non applicable, NA). Asequential linear regression (Ltotal = f(F � Ptotal)) model was used to determine n. n1 Ka1 and n2 Ka2 were the binding constant for the first and second binding sites, respectively.Results represent mean ± SD (n = 3). Ka1 and Ka2, and n1 and n2 of the same complex were significantly different, independent of the method and the blg form used, with a P-valueinferior to 0.01 and to 0.05, respectively; except for the Ka1 and Ka2 data obtained with the modified Scatchard of the linoleate/native blg complex (⁄).

ITC Fluorescence: modified Scatchad Fluorescence: Ltotal = f(F � Ptotal)

Linoleate/native blg

Linoleate/dimers

Linoleate/nanoparticles

Linoleate/native blg

Linoleate/dimers

Linoleate/nanoparticles

Linoleate/native blg

Linoleate/dimers

Linoleate/nanoparticles

n1 0.60 ± 0.01 0.53 ± 0.08 0.92 ± 0.29 2.45 ± 0.07 10.3 ± 0.05 NA 2.38 ± 0.12 9.8 ± 0.21 15.7 ± 0.55Ka1 � 105 M�1 18.0 ± 6.29 15.1 ± 9.53 15.8 ± 3.35 9.20 ± 2.65⁄ 14.7 ± 2.12n2 6.79 ± 0.05 8.64 ± 0.54 10.3 ± 1.65 5.27 ± 1.50 15.3 ± 0.71 6.02 ± 0.29 12.5 ± 0.76 40.7 ± 2.17Ka2 � 105 M�1 0.41 ± 0.05 0.50 ± 0.42 0.57 ± 0.22 0.62 ± 0.49⁄ 0.37 ± 0.13

2310 S. Le Maux et al. / Food Chemistry 141 (2013) 2305–2313

and native blg showed a lower n1 value, but a higher n2 value com-pared to those deduced from the intrinsic fluorescence data. How-ever, the total number of binding sites (n1 + n2) for linoleate tonative blg was similar (around 7.5–8 linoleate bound to the blg na-tive protein) regardless of technique and method used for data fit-ting. The binding parameters from blg intrinsic fluorescencetitration gave a higher number of binding sites for linoleate tocovalent dimers and to nanoparticles than the ITC data. This dis-crepancy may have resulted from (i) the intrinsic fluorescence datathat cumulates inner filter and non-specific quenching of the fluo-rescence spectrum of the complex under study and/or (ii) the ITCsignal complexity that includes all energetic changes occurringduring the titration such as structural changes of the protein, mod-ifications to the protein and/or ligand hydration (Bouchemal,2008). Similarly, Loch et al. (2012a) found a stoichiometry lowerthan 1 mol for lauric and myristic acids per mole of blg when theinteraction was studied by ITC while one FA was found in the calyxof native blg by Xray crystallography with a resolution of 1.9–2.1 Å.According to these authors, this may be related to the weak inter-action between the FA and blg. Spector and Fletcher (1970) demon-strated that stearic acid exhibited a secondary set of binding sitesto blg with the number of sites varying from 2 to 24, using thesame set of data analysed with different fitting parameters.

Comparative analysis of the fluorescence data show differencesin the fluorescence changes for the native blg experiments relativeto the aggregated blg experiments. The intrinsic fluorescence ofcovalent dimers and nanoparticles decreased the titration of lino-leate due to tryptophan quenching by the FA. Conversely, theintrinsic fluorescence of native blg increased in the presence oflinoleate. This can be explained by the compensation of the trypto-phan quenching effect by the denaturation of the protein caused bythe binding with linoleate, which reduced the tryptophan quantifi-cation by Cys–Cys disulphide bonds (Renard, Lefebvre, Griffin, &Griffin, 1998).

The number of linoleate bound per blg molecule increased withthe degree of aggregation (native blg < covalent dimers < nanopar-ticles) but the association constants for each sets of binding sitesremained similar. Several studies have demonstrated the impactof blg denaturation/aggregation for ligand binding, but weredependent on the type of ligands and/or the structure of the aggre-gates (Ron, Zimet, Bargarum, & Livney, 2010; Shpigelman, Israeli, &Livney, 2010). Hydrophobic ligands are able to bind native blg athydrophobic patches on the protein surface and in the internal ca-lyx if specific structural properties of the ligands are respected(Kontopidis, Holt, & Sawyer, 2004). The changes in binding param-eters are related to the structural changes of blg, which occur dur-ing heat denaturation/aggregation (de Wit, 2009). Heat-inducedprotein unfolding exposes internal hydrophobic patches (de Wit,2009) that constitute additional potential binding sites for hydro-phobic ligands. Even if they are usually of low specificity and lowaffinity, these hydrophobic patches could be responsible for the

higher ratio of linoleate bound per blg molecule in the covalent di-mers and nanoparticles compared to the native form of blg. Thehigher degree of aggregation in the nanoparticles, compared tocovalent dimers, could also create hydrophobic pockets, trappingmore ligands with a weak affinity. Indeed, nanoparticles are micro-gels, which have more hydrophobic binding sites available, com-pared to native blg, as shown by anilino naphthalene sulphonicacid (ANS) fluorescence (Schmitt et al., 2009). In addition, theinternal calyx of blg is modified during the heat-denaturationand aggregation of blg. Consequently the specific affinity to the li-gand at this site could be affected. The formation of covalent di-mers involves the displacement of the free Cys121 thatpotentially distorts the calyx, decreasing its affinity for the linole-ate. O’Neill and Kinsella (1988) showed that heat-denaturation ofblg (75 �C up to 20 min) increased the number of binding sitesfor 2-nonanone but decreased its association constant. Yang, Chen,Chen, Wu, and Mao (2009) found a weaker binding, with a lower n,when vitamin D3 was bound to heat denatured blg (100 �C for16 min) compared to native blg. Similar conclusions were reportedbySpector and Fletcher (1970) who found lower binding constantsbetween palmitate and blg when the protein was heat treated from55 to 80 �C. These different ligands were shown to specificallyinteract in the calyx of blg that is affected by the blg denaturation.Unlike these studies, conformational changes of blg do not lead to achange in the affinity for linoleate at the first set of binding sites.This is rather surprising, since the central cavity contains the bind-ing site with the strongest affinity for linoleate, as shown by 2.1 Åresolution crystallography (PBD ID: 4DQ4, Loch et al. (2012b)).However, it is possible that some specific protein structures are se-lected for crystal formation leading to different results when pro-tein in solid or liquid states are compared.

Cytotoxicity assays represent an excellent method for deter-mining changes in the bioaccessibility of FA to Caco-2 cells, sincethe linoleate must enter cells to be cytotoxic (Lu et al., 2010). Inthe present study, exposure of the cells to linoleate/blg complexesresulted in a decrease in cytotoxicity compared to free linoleate.Therefore, we can postulate that binding of linoleate to all theblg forms decreased the bioaccessibility of the FA. After a 24 hexposure period, linoleate bound to blg nanoparticles had a highercytotoxic effect compared to linoleate bound to native blg. Thiscould be explained by the higher binding capacity of nanoparticlesto FA compared to native blg:8.9 linoleate per blg nanoparticlesversus 3.3 linoleate per native blg. As only 0.6 or 0.9 linoleate isstrongly bound per 1 blg molecule in the nanoparticular or nativestate, respectively (ITC data), the fraction of linoleate bound with alower affinity is much higher for the nanoparticles. This may ex-plain the higher bioaccessibility of linoleate when bound to thenanoparticles. Spector and Fletcher (1970) demonstrated that pal-mitate bound to blg was taken up faster by Ehrlich ascites tumourcells compared to palmitate bound to bovine albumin because pal-mitate binds to bovine albumin with a higher affinity than to blg.

Fig. 2. SDS–PAGE profiles of the three linoleate/blg complexes. Non-reducingconditions were used for: (A) linoleate/native blg complexes, (B) linoleate/covalentdimers complexes, and (C) linoleate/nanoparticles complexes. Reducing conditionswere used for the gel (D) corresponding to the profile of the linoleate/nanoparticlesblg complexes (similar profiles were obtained for the two other complexes). Mw,molecular weight markers (14.4, 20.1, 30, 45, 66, 97 kDa); blg, blg control; lanes 5,7.5 and 10, complexes with an initial molar ratio of 5, 7.5 and 10 linoleate/blg,respectively.

Fig. 3. Stoichiometry of linoleate/blg with different forms of blg (native, covalentdimers and nanoparticles) as determined by GC after dialysis. Correlation of themolar ratios of linoleate/blg added to the starting solutions with the molar ratios oflinoleate/blg that were detected by GC analysis in the linoleate/blg samples afterextensive dialysis and freeze-drying. , linoleate/native blg; , linoleate/covalent dimers; , linoleate /nanoparticles.

S. Le Maux et al. / Food Chemistry 141 (2013) 2305–2313 2311

Consequently, the FA was more bioaccessible to the cells whenbound to blg than to bovine albumin. Interestingly, linoleate/cova-lent dimer complexes were more cytotoxic than linoleate/nanopar-ticles complexes, even though the amount of linoleate bound withhigher Ka was similar, as determined by ITC. As native blg protectsthe cells against the linoleate cytotoxicity, this difference in cellviability may be the result of the different proportions of blg mono-mers present in the test samples (22.4% blg monomers in the nano-particle sample compared to 16.3% blg monomers in the covalentdimer sample, prior to the addition of linoleate). In addition, to ob-

Table 2Protein proportion of linoleate/blg complexes with different forms of blg (native, covalent dmolar ratios of linoleate/blg. blg M, blg monomers; blg D, blg dimers; blg O, blg oligomers

Initial linoleate/blg 0 5

Linoleate/native blg blg M 88.5 ± 5.24 66.4blg D 6.64 ± 2.63 23.1blg O 4.92 ± 2.78 10.4

Linoleate/dimer blg M 16.3 ± 1.50 14.4blg D 78.3 ± 3.21 78.6blg O 5.43 ± 1.65 6.95

Linoleate/nanoparticle blg M 22.4 ± 1.36 15.8NanoP 77.6 ± 1.36 84.2

tain the same linoleate concentration, a higher quantity of complexwas needed for the linoleate/covalent dimers complex. The molarratios were 8.9 linoleate per blg in the nanoparticles versus 4.0linoleate per blg in covalent dimers. However, we have previouslydemonstrated that increasing blg concentration increased the lino-leate uptake by Caco-2 cells even if the kinetic of transport isslower than free linoleate (data not shown). Other studies have re-ported the opposite effect, with the binding of a given ligand to blgincreasing the ligand bioaccessibility. Indeed, Yang et al. (2009) ob-served that vitamin D3, which is practically insoluble in water, wastransported more effectively bound to blg than free vitamin D3 in amouse model. Proteins may affect differently the bioaccessibility ofthe ligand in function of the ligand solubility. The potential contri-bution of residual copper used to prepare blg covalent dimer in thecytotoxic effect of this oligomer cannot be excluded. Copper by it-self at concentrations up to 5 mg/L was not cytotoxic (data notshown). However, copper was reported to be a potent catalyst ofFA oxidation (Frémont, Belguendouz, & Delpal, 1999; Kleinveld,Hak-Lemmers, Stalenhoef, & Demacker, 1992). Peroxidated FA arereported to be more cytotoxic than FA (Alghazeer, Gao, & Howell,2008). Hence, the occurrence of a peroxidated form of linoleatewhich would increase its cytotoxicity cannot be ruled out.

This study has demonstrated that linoleate can bind to differentstructural states of blg (native, covalent dimmers and nanoparti-cles). Binding capacity but not affinity was affected by the proteinstructure. Stoichiometries increased with the size of the proteinaggregates. This is probably due to the exposure of hydrophobicsites during the protein denaturation and the formation of hydro-phobic pockets at the surface or in the inner structure of the aggre-gates. Changes in the binding properties modified the cytotoxicityof the complexes. Consequently, it is proposed that the in vitro bio-

imers and nanoparticles), obtained by GP-HPLC. 0, 5, 7.5 and 10 represents the initial; NanoP, blg nanopaticles. Results represent mean ± SD (n = 3).

7.5 10

± 8.24 60.4 ± 8.17 51.1 ± 4.95± 6.94 28.2 ± 9.57 34.9 ± 6.78± 4.51 11.4 ± 2.03 14.0 ± 3.97± 0.50 14.1 ± 0.19 13.4 ± 0.61± 4.23 79.6 ± 5.80 77.9 ± 6.48± 2.70 6.41 ± 3.74 8.72 ± 4.22± 2.48 12.0 ± 1.03 10.7 ± 1.95± 2.48 88.0 ± 1.03 89.4 ± 1.95

Fig. 4. Cytotoxicity of linoleate, free or bound to different forms of blg, using Caco-2 cells. Cell viability after 24 h on 2 � 104 Caco-2 cells compared to the control cells wasassessed using an MTS assay. Linoleate concentrations in the tested sample varied from 0 to 200 lM. , free linoleate; , linoleate/native blg; , linoleate/covalentdimers; , linoleate/nanoparticles.

2312 S. Le Maux et al. / Food Chemistry 141 (2013) 2305–2313

accessibility of linoleate can be modulated by changing proteinstructures, which subsequently modifies the ligand binding param-eters. This could be of interest in relation to optimising the designof food products from a sanitary, textural and health benefit per-spective. From a nutritional point of view, one question that arisesis how these various protein/FA complexes react to digestive en-zymes. Studies are in progress to determine the behaviour of thecomplexes under simulated gastro-intestinal in vitro digestionand the subsequent effect of digestion on FA cytotoxicity anduptake.

Acknowledgements

S. Le Maux is currently supported by a Teagasc Walsh Fellow-ship and the Department of Agriculture, Fisheries and Food (FIRMproject 08/RD/TMFRC/650). We also acknowledge funding fromIRCSET-Ulysses Travel Grant. The authors would like to expresstheir gratitude to Alan Hennessy for the GC analysis.

References

Alghazeer, R., Gao, H. L., & Howell, N. K. (2008). Cytotoxicity of oxidised lipids incultured colonal human intestinal cancer cells (caco-2 cells). Toxicology Letters,180(3), 202–211.

Bouchemal, K. (2008). New challenges for pharmaceutical formulations and drugdelivery systems characterization using isothermal titration calorimetry. DrugDiscovery Today, 13(21–22), 960–972.

Braunitzer, G., Chen, R., Schrank, B., & Stangl, A. (1973). Die sequenzanalyse des b-lactoglobulins. Hoppe-Seyler s Zeitschrift für physiologische Chemie, 354(2),867–878.

Coakley, M., Ross, R. P., Nordgren, M., Fitzgerald, G., Devery, R., & Stanton, C. (2003).Conjugated linoleic acid biosynthesis by human-derived Bifidobacteriumspecies. Journal of Applied Microbiology, 94(1), 138–145.

Cogan, U., Kopelman, M., Mokady, S., & Shinitzky, M. (1976). Binding affinities ofretinol and related compounds to retinol binding proteins. European Journal ofBiochemistry, 65(1), 71–78.

Considine, T., Patel, H. A., Anema, S. G., Singh, H., & Creamer, L. K. (2007).Interactions of milk proteins during heat and high hydrostatic pressuretreatments – A review. Innovative Food Science & Emerging Technologies, 8(1),1–23.

Creamer, L. K., Parry, D. A. D., & Malcolm, G. N. (1983). Secondary structure of bovinebeta-lactoglobulin B. Archives of Biochemistry and Biophysics, 227(1), 98–105.

de Wit, J. N. (2009). Thermal behaviour of bovine beta-lactoglobulin attemperatures up to 150 �C. A review. Trends in Food Science and Technology,20(1), 27–34.

Frémont, L., Belguendouz, L., & Delpal, S. (1999). Antioxidant activity of resveratroland alcohol-free wine polyphenols related to LDL oxidation andpolyunsaturated fatty acids. Life Sciences, 64(26), 2511–2521.

Gulzar, M., Croguennec, T., Jardin, J., Piot, M., & Bouhallab, S. (2009). Coppermodulates the heat-induced sulfhydryl/disulfide interchange reactions of beta-lactoglobulin. Food Chemistry, 116(4), 884–891.

Jensen, R. G. (2002). The composition of bovine milk lipids: January 1995 toDecember 2000. Journal of Dairy Science, 85(2), 295–350.

Kleinveld, H. A., Hak-Lemmers, H., Stalenhoef, A., & Demacker, P. (1992). Improvedmeasurement of low-density-lipoprotein susceptibility to copper-inducedoxidation: Application of a short procedure for isolating low-densitylipoprotein. Clinical Chemistry, 38(10), 2066–2072.

Kontopidis, G., Holt, C., & Sawyer, L. (2004). Invited review: Beta-lactoglobulin:Binding properties, structure, and function. Journal of Dairy Science, 87(4),785–796.

Le Maux, S., Giblin, L., Croguennec, T., Bouhallab, S., & Brodkorb, A. (2012). Beta-lactoglobulin as a molecular carrier of linoleate: Characterisation and effects onintestinal epithelial cells in vitro. Journal of Agricultural and Food Chemistry,60(37), 9476–9483.

Lišková, K., Auty, M. A. E., Chaurin, V., Min, S., Mok, K. H., O’Brien, N., et al. (2011).Cytotoxic complexes of sodium oleate with beta-lactoglobulin. European Journalof Lipid Science and Technology, 1207–1218.

Loch, J., Polit, A., Bonarek, P., Ries, D., Kurpiewska, K., Dziedzicka Wasylewska, M., &Lewi ski, K. (2012b). Bovine beta-lactoglobulin complex with linoleic acid. InURL (http://www.rcsb.org/pdb/explore/explore.do?structureId=4DQ4)DOI:10.2210/pdb4dq4/pdb, vol. (most recent, accessed 5.06.12)).

Loch, J., Polit, A., Bonarek, P., Olszewska, D., Kurpiewska, K., Dziedzicka-Wasylewska, M., et al. (2012a). Structural and thermodynamic studies ofbinding saturated fatty acids to bovine beta-lactoglobulin. International Journalof Biological Macromolecules, 50(4), 1095–1102.

Lu, X., He, G., Yu, H., Ma, Q., Shen, S., & Das, U. N. (2010). Colorectal cancer cellgrowth inhibition by linoleic acid is related to fatty acid composition changes.Journal of Zhejiang University Science B, 11(12), 923–930.

Mantzioris, E., James, M. J., Gibson, R. A., & Cleland, L. G. (1995). Differences exist inthe relationships between dietary linoleic and alpha-linoleic acids and theirrespective long-chain metabolites. American Journal of Clinical Nutrition, 61(2),320–324.

Mu, H. (2008). Bioavailability of omega-3 long-chain polyunsaturated fatty acidsfrom foods. Agro Food Industry Hi-Tech: Focus on Omega-3, 19(4), 24–26.

O’Neill, T., & Kinsella, J. (1988). Effect of heat treatment and modification onconformation and flavor binding by beta-lactoglobulin. Journal of Food Science,53(3), 906–909.

Palmquist, D. L., & Jenkins, T. C. (2003). Challenges with fats and fatty acid methods.Journal of Animal Science, 81(12), 3250–3254.

Perez, M. D., Sanchez, L., Aranda, P., Ena, J., Oria, R., & Calvo, M. (1992). Effect of beta-lactoglobulin on the activity of pregastric lipase. A possible role for this proteinin ruminant milk. Biochimica et Biophysica Acta (BBA), Lipids and LipidMetabolism, 1123(2), 151–155.

Ragona, L., Zetta, L., Fogolari, F., Molinari, H., Pérez, D. M., Puyol, P., et al. (2000).Bovine beta-lactoglobulin: Interaction studies with palmitic acid. ProteinScience, 9(7), 1347–1356.

Renard, D., Lefebvre, J., Griffin, M. C. A., & Griffin, W. G. (1998). Effects of pH and saltenvironment on the association of beta-lactoglobulin revealed by intrinsicfluorescence studies. International Journal of Biological Macromolecules, 22(1),41–49.

Ron, N., Zimet, P., Bargarum, J., & Livney, Y. (2010). Beta-lactoglobulin–polysaccharide complexes as nanovehicles for hydrophobic nutraceuticals innon-fat foods and clear beverages. International Dairy Journal, 20(10), 686–693.

Russo, G. L. (2009). Dietary n � 6 and n � 3 polyunsaturated fatty acids: Frombiochemistry to clinical implications in cardiovascular prevention. BiochemicalPharmacology, 77(6), 937–946.

Sawyer, L., & Kontopidis, G. (2000). The core lipocalin, bovine beta-lactoglobulin.Biochimica et Biophysica Acta (BBA), Protein Structure and Molecular Enzymology,1482(1-2), 136–148.

Schmitt, C., Bovay, C., Vuilliomenet, A. M., Rouvet, M., Bovetto, L., Barbar, R., et al.(2009). Multiscale characterization of individualized beta-lactoglobulinmicrogels formed upon heat treatment under narrow pH range conditions.Langmuir, 25(14), 7899–7909.

Shpigelman, A., Israeli, G., & Livney, Y. D. (2010). Thermally-induced protein-polyphenol co-assemblies: Beta lactoglobulin-based nanocomplexes asprotective nanovehicles for EGCG. Food Hydrocolloids, 24(8), 735–743.

Singh, H., Ye, A., & Horne, D. (2009). Structuring food emulsions in thegastrointestinal tract to modify lipid digestion. Progress in Lipid Research,48(2), 92–100.

Spector, A. A., & Fletcher, J. E. (1970). Binding of long chain fatty acids to beta-lactoglobulin. Lipids, 5(4), 403–411.

S. Le Maux et al. / Food Chemistry 141 (2013) 2305–2313 2313

Wang, Q. W. Q., Allen, J. C., & Swaisgood, H. E. (1998). Protein concentrationdependence of palmitate binding to beta-lactoglobulin. Journal of Dairy Science,81(1), 76–81.

Wu, S. Y., Pérez, M. D., Puyol, P., & Sawyer, L. (1999). Beta-lactoglobulin bindspalmitate within its central cavity. Journal of Biological Chemistry, 274(1),170–174.

Yang, M. C., Chen, N. C., Chen, C. J., Wu, C. Y., & Mao, S. J. T. (2009). Evidence for beta-lactoglobulin involvement in vitamin D transport in vivo-Role of the c-turn(Leu-Pro-Met) of beta-lactoglobulin in vitamin D binding. FEBS Journal, 276(8),2251–2265.

Yang, M. C., Guan, H. H., Liu, M. Y., Lin, Y. H., Yang, J. M., Chen, W. L., et al. (2008).Crystal structure of a secondary vitamin D3 binding site of milk beta-lactoglobulin. Proteins: Structure, Function, and Bioinformatics, 71(3), 1197–1210.

Zhao, G., Etherton, T. D., Martin, K. R., Vanden Heuvel, J. P., Gillies, P. J., & West, S. G.(2005). Anti-inflammatory effects of polyunsaturated fatty acids in THP-1 cells.Biochemical and Biophysical Research Communications, 336(3), 909–917.

Zock, P. L., & Katan, M. B. (1998). Linoleic acid intake and cancer risk: A review andmeta-analysis. American Journal of Clinical Nutrition, 68(1), 142–153.