Binding of �-Lactolobulin to Pectins Varying in their Overalland Local Charge Density

Bram L. H. M. Sperber,†,‡ Martien A. Cohen Stuart,‡ Henk A. Schols,†

Alphons G. J. Voragen,† and Willem Norde*,‡,§

Laboratory of Food Chemistry, Department of Agrotechnology and Food Sciences, Wageningen University,P.O. Box 8129, 6700 EV Wageningen, The Netherlands, Laboratory of Physical Chemistry and Colloid

Science, Department of Agrotechnology and Food Sciences, Wageningen University,P.O. Box 8038, 6700 EK Wageningen, The Netherlands, and Department of Biomedical Engineering,University Medical Center Groningen, Antonius Deusinglaan 1, 9713 AV Groningen, The Netherlands

Received July 17, 2009; Revised Manuscript Received October 27, 2009

The formation of complexes between proteins and polysaccharides is of great importance for many food systemslike foams, emulsions, acidified milk drinks, and so on. The complex formation between �-lactoglobulin (�-lg)and pectins with a well-defined physicochemical fine structure has been studied to elucidate the influence ofoverall charge and local charge density of pectin on the complex formation. Binding isotherms of �-lg to pectinare constructed using fluorescence anisotropy, which is shown to be an excellent technique for this purpose, asit is fast and requires low sample volumes. From the binding isotherms the maximal adsorbed amount, bindingconstant (kobs) and the cooperativity of binding are obtained at different ionic strengths. The Hill model is usedto fit the binding isotherms and is shown to be preferable over a Langmuir fit. At pH 4.25, kobs shows a maximumat an ionic strength of 10 mM when using a low methyl esterified pectin (LMP) due to the balance of attractiveand repulsive electrostatic forces between �-lg and pectin and �-lg neighbors. For two high methyl esterifiedpectins, one with a blockwise distribution of methyl esters (HMPB) and one with a random distribution (HMPR),this ionic strength maximum is absent and kobs decreases with increasing ionic strength. kobs is found to be largestfor LMP and HMPB and considerably lower for HMPR. A positive cooperativity is observed for both LMP (abovean ionic strength of 45 mM) and HMPR (above an ionic strength of 15 mM) but not for HMPB. Positive cooperativityis thought to be caused by a rearrangement of the pectin helix structure caused by binding of �-lg, thus creatingnew or binding sites with a higher affinity. To attain strong binding of �-lg to pectin it is preferable to use apectin with a blockwise distribution of methyl esters. When complex formation takes place in high ionic strengthmedia an LMP gives the best results, while at low ionic strength a high methyl esterified pectin with blockwisedistribution may give better results, due to reduced electrostatic repulsion between both pectin and �-lg and �-lgneighbors.

Introduction

Complex formation between anionic polysaccharides andproteins find their application in a diversity of fields like foodindustry, biotechnology, medicine, pharmacy, and cosmetics.1

In the food industry complexes of protein and polysaccharidesare used for the stabilization of acid dairy drinks, foams andemulsions, as fat substitutes, meat analogues, purification ofmacromolecules, or microencapsulation of ingredients.2-8

Complexes of proteins and anionic polysaccharides areformed because of attractive interactions between the twobiopolymers and can be of an electrostatic or hydrophobicnature.9 Hydrophobic interactions are limited in systems consist-ing of proteins and anionic polysaccharides as the alkyl sidechain needs to be at least 3-4 carbon atoms long beforesignificant hydrophobic interactions take place.10 Electrostaticinteractions occur because of opposite charges on both biopoly-mers. As in a food system, the charge of the polysaccharide isnegative, and complex formation takes place below and around

the isoelectric point of the protein. If complex formation occursabove the isoelectric point, it is usually attributed to locallypositively charged patches on the protein.11 Depending onsolution parameters, like pH, ionic strength, and concentrationof the two biopolymers, either soluble or insoluble complexesare formed. Soluble complexes exist as long as the complexesare sufficiently stabilized against aggregation by electrostaticrepulsion. Generally speaking, the phase boundary betweensoluble and insoluble complexes is found at the pH, ionicstrength, and mix ratio of biopolymers that cause the complexofproteinandpolysaccharide tocarrynonetchargeanymore.11-13

To compare the formation of complexes between anionicpolysaccharides and protein parameters like composition of thecomplex, association constant, and the cooperativity of theassociation are commonly considered. To obtain these param-eters isotherms for the binding of the protein to the polysac-charide need to be constructed without disturbing the delicateequilibrium between bound and free protein.14 For this, Dubinet al.15,16 developed frontal analysis continuous capillary elec-trophoresis (FACCE). This technique allows the determinationof free ligand concentration without disturbing the equilibriumconditions due to continuous injection of fresh sample into theseparation capillary. FACCE has been used successfully toconstruct binding isotherms for different combinations of

* To whom correspondence should be addressed. Tel.: +31 (0)317483540. Fax: +31 (0)317 483777. E-mail: willem.norde@wur.nl.

† Laboratory of Food Chemistry, Wageningen University.‡ Laboratory of Physical Chemistry and Colloid Science, Wageningen

University.§ University Medical Center Groningen.

Biomacromolecules 2009, 10, 3246–32523246

10.1021/bm900812x CCC: $40.75 2009 American Chemical SocietyPublished on Web 11/11/2009

protein-polyelectrolyte.17-22 The big advantage of this tech-nique is that it uses only small amounts of sample (<1 µL).There is one major drawback to the FACCE technique: To avoidbinding of the protein to the negatively charged silica capillary,FACCE can only be used at a solution pH well above the pI ofthe protein.14,15

Fluorescence anisotropy (FA) is a technique that measuresthe polarization of the emitted light after excitation of the samplewith polarized light. The fluorescence anisotropy of a fluoro-phore is, among others, dependent on the diffusion rate andtherefore the size of the fluorophore.23 It implies that thefluorescence anisotropy of a fluorophore-containing protein, boundto a polysaccharide, deviates from the anisotropy when it is freein solution. FA has been used successfully in the construction ofbinding isotherms for protein-DNA complexes.24-27 There aretwo big advantages of using fluorescence anisotropy overFACCE: because the measurement is performed directly in thesolution, there is no need of separating free and bound proteinand the measurement can be performed at any pH, providedthat an optically clear sample is measured.

In this work, complexes between pectin and �-lactoglobulin(�-lg) are studied. Pectin is commonly used to stabilize aciddairy drinks and to form gels. Commercially available pectinsare isolated from plant materials by means of an acid extractionafter which their predominant structure is R-(1f4)-linkedD-galacturonic acid.28 The functionality of these pectins ismainly determined by two parameters: The degree of methylesterification (DM) of the carboxyl group on C-6 and thedistribution of these methyl esters (DB). A pectin with a randomdistribution pattern of the methyl esters has a low value forDB, a pectin with a blockwise distribution has a high DB.29

The DM determines the overall charge of the pectin, while theDB determines the local charge density.

�-Lg is the predominant protein in whey and used in foodsfor its emulsifying, foaming and gelation properties30-32 anddominates the complex formation between whey protein isolateand polysaccharides.12 A better understanding of the bindingof �-lg to pectin will help to interpret the influence �-lg-pectincomplex formation has on the stability of food structures likefoams and emulsions.3,33

Materials and Methods

Pectin. Three pectins, a low methyl esterified pectin (LMP), a highmethyl esterified calcium sensitive pectin (HMPB) with a blockwisedistribution of methyl esters, and a high methyl esterified noncalciumsensitive pectin (HMPR) with a random methyl ester distribution, werekindly provided by CPKelco (Lille Skensved, Danmark). The pectinsoriginate from lemon and were characterized by Daas et al.34 on DMand DB. Detailed information for the different pectins can be found inthe Supporting Information. The pectins had a typical molecular weightof 150 kDa.35

The pectins contained 0.02-0.05 mol of calcium per mole ofgalacturonic acid (GalA), as determined by flame photometry. Thecalcium was removed by washing the pectins five times with 60%ethanol, containing 5% (v/v) hydrochloric acid. Next, the pectins werewashed with 60% ethanol until they were free of chlorine (tested byaddition of 0.2% (w/w) silvernitrate, until no white precipitate isformed). Pectins were subsequently washed with acetone and air-dried.This treatment for removing calcium was sufficient for LMP and HMPB,but HMPR required further treatment. HMPR was dissolved to 10 g/Land mixed with AG 50W-X4 200-400 mesh ion-exchange resin fromBio-Rad Laboratories (Hercules, CA). HMPR and ion-exchange resinwere separated by centrifugation and HMPR was lyophilized. Thisresulted in pectins that contained less than 0.001 mol of calcium permole of GalA.

Preparation of �-Lactoglobulin and Alexa Fluor 430 Labeled�-Lactoglobulin. �-Lg was purified from bovine milk using a nonde-naturing technique as described by De Jongh et al.36 and was over99% pure. �-Lg was fluorescently labeled with Alexa Fluor 430carboxylic acid, succinimidyl ester (A10169, Invitrogen, Breda, TheNetherlands). An amine reactive probe, chosen as the only availablefree thiol group (C121), induces a structural change of �-lg whenmodified.37 The labeling reaction was carried out according to the“Molecular Probes” protocol (http://www.probes.com). Typically, 10mg of �-lg was dissolved in 1 mL of 0.1 M sodium bicarbonate buffer(pH 8.3). A stock solution of Alexa Fluor 430 was prepared bydissolving 5 mg in 0.5 mL of DMSO. The labeling reaction wasperformed at room temperature by adding 10 µL of the Alexa Fluor430 stock solution to the stirred �-lg solution in 2 µL portions with a10 min interval. The reaction was performed in the dark. A total of 20min after the last addition of the Alexa Fluor 430 stock solution, thereaction was stopped by adding 0.1 mL of a 1.5 M sodium hydroxy-lamine solution (pH 8.5). The conjugate was separated from not reactedreagent using a Sephadex G-25 gel filtration column (17-0851-01,GE Healthcare, Uppsala, Sweden) equilibrated with a 0.1 M sodiumacetate buffer (pH 4.25). The degree of labeling was approximately0.2 mol dye per mol �-lg.

Weak Anion-Exchange High Performance Liquid Chroma-tography. Weak anion-exchange high performance liquid chromatog-raphy (WAX-HPLC) was performed as described in detail by Guillotinet al.38 In short, a Dionex Propac WAX-10 column (WAX; 4 × 250mm) was attached to an Akta purifier system. A total of 200 µL of a5 g/L pectin solution was injected and eluted (1 mL/min) with a lineargradient from 0 to 0.6 M of sodium phosphate (pH 6) for 25 min. UVdetection is at 215 nm. All elution profiles were baseline corrected foran injection of 200 µL of water.

Fluorescence Anisotropy. Fluorescence anisotropy measurementswere carried out on a TECAN Infinite F500 micro plate reader (TECAN,Austria) equipped with an excitation filter of 420 nm (10 nm bandwidth)and an emission filter of 540 nm (10 nm bandwidth) and a Xenon flashlamp with a frequency of 40 Hz. The measurement time after the flashis set at 20 µs, and per measurement 10 flashes are recorded. Sampleswere measured in black flat bottom 96-well Fluotrac 200 plates (655076,Greiner Bio-One, Alphen a/d Rijn, The Netherlands). Fluorescenceanisotropy is calculated as follows:

with A the fluorescence anisotropy and I| and I⊥ the intensity of thefluorescent light in the parallel and perpendicular direction, respec-tively.23

Construction of Binding Isotherms. Pectin is dissolved in a 20mM sodium acetate buffer of pH 4.25 to a concentration of 0.5 mg/mL. The ionic strength of the pectin solution is set with sodium chlorideto obtain the correct ionic strength (4-300 mM) after mixing with �-lgand buffer. �-Lg (nonlabeled) is dissolved in 20 mM acetic acid. Pectinand �-lg are kept overnight in the fridge to allow for full hydration ofthe molecules. Prior to experiments the pH is carefully adjusted with1 M sodium hydroxide to 4.25. Samples of pectin and �-lg (both labeledand nonlabeled) are mixed to obtain weight ratios (mg �-lg/mg pectin)of 0.25 up to 16, all with a pectin concentration of 0.1 mg/mL and anAlexa Fluor 430 labeled �-lg concentration of approximately 1 µg/mL. Labeled and nonlabeled �-lg are assumed to bind identically topectin. All points of the binding isotherm are averages of four samples.

Binding isotherms are constructed from the fluorescence anisotropymeasurements. The observed fluorescence anisotropy (Aobs) is theaverage of the fluorescence anisotropy of the free (Af) and bound (Ab)�-lg.23 ∆A is the difference between Aobs and Af

A )I| - I⊥

I| + 2I⊥(1)

∆A ) Aobs - Af ) (fbAb + ff Af) - Af (2)

Binding of �-Lactolobulin to Pectins Biomacromolecules, Vol. 10, No. 12, 2009 3247

with ff and fb the fraction of free and bound �-lg, respectively. As thefluorescence anisotropy is influenced by the viscosity of the solutionAf is determined in a sample that contains both pectin and �-lg and hasan ionic strength of 300 mM. At these conditions there is no complexformation between pectin and �-lg13 and the measured fluorescenceanisotropy is that of free �-lg. The fluorescence anisotropy of bound�-lg, Ab, is determined by taking the maximum value for thefluorescence anisotropy throughout all the experiments. The value forbound �-lg is comparable to the fluorescence anisotropy value forlabeled �-lg in the high viscosity medium glycerol. To obtain a molarconcentration of �-lg, it is assumed that �-lg adsorbs as a dimer topectin, as there are no indications that adsorption to pectin disturbs themonomer-dimer equilibrium of �-lg.

Analysis of Binding Isotherms. The binding isotherms are fittedwith a Langmuir and a Hill adsorption isotherm to obtain a bindingconstant and a measure for the cooperativity of the system (for theHill model). Equation 3 gives the Hill model used to fit the bindingisotherms

with θ the fractional occupation of binding sites (Γ/Γm with Γm for theappropriate ionic strength from Table 2, Supporting Information), kobs

the observed molar binding constant, n the Hill coefficient, and X themolar fraction of free �-lg.39 For n < 1 the binding is anticooperative,for n > 1 cooperative and for n ) 1 the binding is noncooperative andeq 3 reduces to the Langmuir model. The Langmuir model assumesreversibility of binding, identical and independent binding sites, andno interaction between the adsorbed molecules. The Hill adsorptionallows for interaction between the adsorbed molecules by means ofthe Hill coefficient.

The fitting procedure is performed using a nonlinear least-squaresfit based on the Levenbert-Marquadt algorithm in Origin 5.0 software(Microcal Software, Northampton, MA).

Results and Discussion

Size and Distribution of Non-Methyl-Esterified Gala-cturonic Acid Blocks Along the Pectin Backbone. Todetermine the block size and block distribution of methyl esterson the pectin backbone several parameters have been determinedby Daas et al.34 for the pectins used. These parameters are uronicacid content, DM, DB, distribution of mono-, di-, and tri-GalAafter endo-PG digestion, and the ratio between the peak area ofmethyl ester (Me) containing oligosaccharides and the peak areaof the non-methyl-esterified (n-Me) oligosaccharides after endo-PG digestion (Me/n-Me ratio). The details of these parameterscan be found in the Supporting Information. Figure 1 showsthe elution profile of the pectins on a WAX column to revealthe presence of populations of pectins with different character-istics in the same sample. The combination of all the parametersmentioned above provides a detailed picture of the fine structureof the three pectin samples.

DM determination shows LMP to be a low methyl esterifiedpectin and HMPB and HMPR to be high methyl esterified pectinwith virtually identical DM. The DB of HMPR is very low,indicating that only a very small fraction of HMPR is K. fragilisendo-PG digestible, while for both LMP and HMPB a consider-able amount of n-Me-GalA is present in blocks.

The digestion of pectin with endo-PG gives a distributionbetween mono-, di-, and tri-GalA oligomers released form thepectin as well as (partially) Me-GalA oligomers. If the distribu-tion of mono-, di-, and tri-GalA resembles that of the distributionof a poly galacturonic acid digest, the blocks can be considered

of “infinite length”, as is the case for LMP. HMPB has a reducedamount of di-, and tri-GalA, with an increased amount of mono-GalA. This is indicative for a reduced block length of n-Me-GalA residues. HMPR has an even bigger reduction of tri-GalAoligomers, mainly in favor of di-GalA. This large decrease intri-GalA shows that the n-Me-GalA blocks present on HMPR

are small, which is also indicated by HMPR being a noncalciumsensitive pectin.

The Me/n-Me ratio shows how close endo-PG degradableblocks are together. Endo-PG degradable blocks more than sixGalA residues apart will not be detected by this method.40 HMPR

has a Me/n-Me ratio of 0, implying that the blocks of GalA onHMPR are more than six GalA residues apart and can thereforebe considered isolated blocks. Also HMPB has a very low Me/n-Me ratio. Although the blocks are much larger than for HMPR,they are still virtually isolated because of the high DM of HMPB.LMP has a high Me/n-Me ratio revealing the close proximityof the blocks. This is mainly caused by LMP being a low methylesterified pectin, causing the blocks to be close together.

The elution on the WAX-column (Figure 1) reveals thepresence of different populations within the pectin sample, aswell as information about the size of the blocks in the pectin.38

Peaks in region I contain the material that does not bind to thecolumn. For commercial pectin samples, it was found that thispeak contains no GalA.38 Region II contains high methylesterified, random pectin. HMPR has a large peak in region II,HMPB a much smaller one, and for LMP no peak is observed.In region III, high to intermediate methyl esterified pectin witha blockwise distribution is found. A small peak is found forHMPR, but HMPB has a continuous peak throughout this region,indicating a distribution in the block length for HMPB. LMPonly starts showing a peak at the end of region III. Region IVis the “PGA”-region, where pectins with large (“infinite”) blocksof n-Me-GalA elute. This is the region where virtually all LMPelutes, as well as some HMPB, while HMPR has no peak in thisarea.

The three pectins can thus be characterized as follows: LMPhas the lowest DM and, therefore, the most n-Me-GalA groups.The blocksize of n-Me-GalA residues on LMP is large and theyare close together. LMP has a homogeneous blocksize, mainlybecause of its low DM, as shown by elution on the WAXcolumn. HMPB and HMPR have a comparable DM, but differin the distribution of the methyl esters. The blocks of n-Me-GalA on HMPB are large, albeit being somewhat smaller thanthose on LMP as shown by the Me/n-Me ratio. The blocks ofn-Me-GalA on HMPB are considered to be isolated. Thedistribution on the WAX column shows that HMPB is a

θ )kobs

n Xn

1 + kobsn Xn

(3)

Figure 1. WAX elution profile after background correction of LMP(black line), HMPB (gray line), and HMPR (black circles).

3248 Biomacromolecules, Vol. 10, No. 12, 2009 Sperber et al.

heterogeneous sample, consisting of many different pectins,ranging from HM-random pectin to PGA characteristics. HMPR

is a pectin that contains no large blocks of n-Me-GalA and onlyvery few isolated intermediate blocks of n-Me-GalA.

Binding Isotherms. Binding isotherms are measured toinvestigate the binding of �-lg to pectins having differentphysicochemical characteristics. The influence of the localcharge density and overall charge of the pectin on the complexformation with �-lg is expected to be reflected in the bindingisotherms, showing the mass of �-lg bound per unit mass ofpectin as a function of �-lg concentration in solution.

Figure 2 shows ∆A for �-lg and LMP at pH 4.25 and anionic strength of 4 mM. At low weight ratio (WR), ∆A isconstant at the maximum value, indicating that all �-lg is boundto LMP. When WR increases, ∆A decreases as LMP reachessaturation and not all of the �-lg is bound to LMP. The datafor ∆A is used to calculate the fractions of free (ff) and bound(fb) �-lg, using eq 2. In Figure 3 binding isotherms for �-lg toLMP, HMPB, and HMPR are shown. FA shows to be a reliableand fast technique for measuring binding isotherms betweenprotein and polysaccharide, with the advantage over FACCEthat it can also successfully measure binding isotherms belowand around the pI of the protein, providing that the sample isoptically clear. As the measurements are performed in microtiterplates, sample volumes can be kept at 1 mL or less and themeasurement of an entire binding isotherm takes only severalminutes.

�-Lg shows a very high affinity for LMP and HMPB at lowionic strength, as virtually all added �-lg binds to pectin. Onlywhen the pectin binding sites are close to saturation, theconcentration of free �-lg increases. For HMPR the initial high

affinity is lower, most likely by the virtual absence of largeblocks of n-Me-GalA. When the ionic strength of the solutionis increased, the binding of �-lg to pectin decreases. HMPR

shows the highest sensitivity to ionic strength: at an ionicstrength of 60 mM no binding of �-lg to HMPR takes placeanymore. At an ionic strength of 75 mM �-lg no longer bindsto HMPB and for LMP binding of �-lg is suppressed at an ionicstrength of 200 mM. This difference in sensitivity for ionicstrength was found before by means of dynamic light scattering(DLS),13 only at slightly higher ionic strength. The measure-ments with DLS are extremely sensitive to complex formation:at pH 4.25, the increase in the intensity of scattered light (Is) athigh ionic strength is below 1% of the increase in Is at lowionic strength. This high sensitivity of the DLS technique cannotbe matched by FA, leading to a slightly lowered maximum ionicstrength where complex formation is observed.

The maximum amount of �-lg adsorbed per gram of pectin(Γm) is the highest for LMP with an Γm of 138 µmol/g, HMPB

and HMPR bind similar quantities of �-lg with 84 µmol/g and82 µmol/g, respectively (see Supporting Information). Clearly,Γm increases when the DM becomes smaller. These Γm valuesare comparable to the Γm found by Girard et al.,17 who found,after recalculation to the same units, 122 µmol/g for a LMP(DM 28.3) and 62 µmol/g for a HMP (DM 73.4), even thoughGirard et al. measured at pH 4.

When Γm is calculated as mol �-lg per mol n-Me-GalA ofthe pectin, Γm is 44 mmol/mol for LMP, 58 mmol/mol forHMPB, and 64 mmol/mol for HMPR. The two HM-pectins bindconsiderably more �-lg than LMP. Gao and Dubin10 found thatthe minimum length of the alkyl side chain is 3-4 carbon atoms,before hydrophobic interactions start to influence binding ofprotein to a oppositely charged polymer. This observation issubstantiated by Girard et al.41 who found hydrophobic interac-tions not to be important for pectin-�-lg systems. The differencein adsorption of �-lg per unit of n-Me-GalA for the differenttypes of pectin is therefore not caused by different extents ofhydrophobic interactions between �-lg and pectin. The maxi-mum amount of �-lg that binds to pectin is found at or belowan ionic strength of 10 mM, where both attractive and repulsiveelectrostatic forces are largest. LMP has the largest blocks ofn-Me-GalA. This prevents all n-Me-GalA residues to participatein binding of �-lg due to electrostatic repulsion between �-lgneighbors, as �-lg molecules cannot be packed “shoulder toshoulder”. The blocks of n-Me-GalA on HMPB are somewhatsmaller but also more isolated than on LMP. This may reducerepulsion between pectin and local negatively charged patcheson �-lg, as the presence of Me-GalA residues at the edges ofthe n-Me-GalA blocks have no repulsive interactions with �-lg.HMPR binds slightly more �-lg than HMPB. Both have a similarDM, but the difference in DB allows HMPR the higher Γm: Themore even spread of n-Me-GalA on HMPR leads to less lateralrepulsion between �-lg molecules as they are not packed asclosely as on LMP or HMPB.

Γm may be estimated based on the dimensions of �-lg andthe GalA unit. The �-lg dimer is an ellipsoid with a length of6.9 nm and width of 3.6 nm,42 the GalA monomer has a lengthof 0.435 nm.43 Considering the GalA content and the DM ofthe pectins, and depending on which diameter of the �-lg dimeris used, Γm is estimated to be between 196 and 375 µmol �-lg/gLMP, Γm between 91 and 175 µmol �-lg/g HMPB, and Γm

between 81 and 155 µmol �-lg/g HMPR. Only HMPR reachesthe lower boundary of the estimation, LMP and HMPB bindless �-lg. Because Γm is measured at low ionic strengthelectrostatic repulsion between �-lg molecules will be high, thus

Figure 2. Fluorescence anisotropy of �-lg-LMP mixtures at pH 4.25,ionic strength is 4 mM.

Figure 3. Binding isotherm of �-lg to LMP (b,O), HMPB (1,3), andHMPR (9,0) at pH 4.25. Ionic strength is 10 mM for filled symbols,and 45 mM for open symbols. �-Lg is assumed to bind as a dimer.

Binding of �-Lactolobulin to Pectins Biomacromolecules, Vol. 10, No. 12, 2009 3249

lowering the packing density of �-lg on the pectin. Also, theestimation does not take the presence of Me-GalA into account.It is possible that not all n-Me-GalA is participating in thebinding of �-lg as their neighbors are of the Me-GalA kind,causing insufficient attraction between binding site on the pectinand �-lg. Because HMPR comes closest to the estimate, thepresence of n-Me-GalA residues in the binding site seems tohave less influence than the electrostatic repulsion caused bycrowding of �-lg molecules along the pectin chain.

Analysis of Binding Isotherms. To obtain binding charac-teristics from the binding isotherms, the reversibility of bindingof �-lg to pectin has to be established, as this is a key assumptionof most binding models. The reversibility of the binding of �-lgto pectin with respect to dilution was tested by mixing samplesof different WR and the same ionic strength. Samples of thesame WR, but of different ionic strength were mixed to establishthe reversibility with respect to ionic strength. The mixedsamples gave the correct values for free and bound �-lgaccording to their new WR or ionic strength (data not shown)from which it was concluded that the binding of �-lg to pectinis reversible.

The binding isotherms may be fitted with theoretical modelsto obtain information about the binding constant, kobs, and thecooperativity of binding. Previously, different models were usedto describe the binding of a protein to either charged syntheticpolymers or polysaccharides: a two class binding site Langmuirfit,14,15 the Hill model,14 and the McGhee-Von Hippel model44,45

for overlapping binding sites.16-18,46 The two class binding siteLangmuir model is used when two types of binding sites arepresent. The analysis of the pectin samples shows that especiallyHMPB, but also HMPR and LMP are not homogeneous in theirdistribution of methyl esters and hence the size of n-Me-GalAblocks. Additionally, it is difficult to define what is a bindingsite and how the presence of methyl esters influences the bindingof �-lg to pectin. Because of this, it is more appropriate to usean average binding site model, instead of assuming two bindingsites.

The McGhee-Von Hippel model was originally developedfor protein-DNA binding, but has also been applied toprotein-synthetic polyelectrolyte binding by Dubin et al.16,18,46

and for �-lg-pectin by Girard et al.17 This model takes intoaccount that a binding site consists of multiple polymer repetitiveunits. When the binding sites are being filled, the space inbetween two bound ligand molecules may be smaller than thebinding site of the ligand. The polymer units in between thesetwo ligands are thus inaccessible for a ligand molecule to bind.This inaccessibility of polymer units results in an apparentnegative cooperativity.44 The McGhee-Von Hippel modelassumes a lattice as a linear array of N identical units. Thisassumption will work well for a polyelectrolyte built ofcontinuously repeating units. Pectin’s repeating unit is GalA,which would suit this assumption as well, if it were not for themethyl esters on the carboxyl group that make pectin consistingof two types of units. This makes the McGhee-Von Hippelmodel poorly applicable to �-lg-pectin systems, as one of theparameters in the model is the number of lattice units that areoccupied by one ligand.

The Hill equation is derived from the Langmuir model toallow for cooperative binding and has been used to describethe binding of BSA to synthetic polyelectrolytes with varyinglength in alkyl side chain.14 As no homogeneous binding sitecan be identified on the pectin, only an average binding constantcan be obtained. The cooperativity parameter in the Hill modelhas an empirical character, only showing whether or not the

binding is cooperative or not. The empirical character of thecooperativity parameter makes it attractive for the �-lg-pectinsystem, where it is difficult to define what the binding site lookslike. In view of these considerations, we decided to fit thebinding isotherms with the Hill model and, for comparison, withthe one site Langmuir model.

Figure 4 shows the Hill fit, but also the fit of the Langmuirmodel (n ) 1) to the binding isotherms. As the Hill model isan extension of the Langmuir model, it is more flexible to fitthe experimental data and, therefore, may fit the data better thanthe Langmuir model. Figure 5 shows the observed bindingconstants from the Langmuir and Hill model, the Hill coefficientfor the Hill model can be found in the Supporting Information.The values found for kobs by fitting with either the Langmuir orthe Hill model are comparable. The additional informationobtained from the Hill model concerning the cooperativity ofbinding, gives it preference over the Langmuir model.

For LMP, kobs shows a maximum with increasing ionicstrength. Seyrek et al.18 also found a maximum in kobs withincreasing ionic strength for other protein-synthetic polyelec-trolyte combinations. The maximum of kobs is related to thedistribution of positively and negatively charged groups on theprotein molecule. The attractive and repulsive electrostatic forcesbetween protein and polyelectrolyte associated with thesecharges lead to an optimum for binding at a specific ionicstrength. For LMP this optimum is situated at 10 mM. Thecooperativity parameter, n, is just below unity for the ionicstrengths of 4 and 10 mM. When �-lg binds to LMP it will

Figure 4. Fit of Langmuir (solid lines) and Hill (dotted lines) bindingmodels to (a) LMP, (b) HMPB, and (c) HMPR at pH 4.25. Ionic strengthfor LMP: 10 mM (b) and 75 mM (O), HMPB: 10 mM (b) and 45 mM(O), and HMPR: 10 mM (b) and 30 mM (O).

3250 Biomacromolecules, Vol. 10, No. 12, 2009 Sperber et al.

occupy more than one GalA. McGhee and Von Hippel44,45

showed that this can lead to apparent negative cooperativity,because the amount of n-Me-GalA residues between two bound�-lg molecules is lower than the amount required for bindingof a �-lg molecule, making it harder to bind at higher θ. Atlow ionic strength, in this study 4 and 10 mM, the Debye lengthis longer, therewith increasing the repulsion between �-lgmolecules and, hence, making it more difficult for the next �-lgmolecule to adsorb in between two others. Binding isothermsof �-lg and a LM-pectin were obtained at pH 4 by Girard et al.(2003).17 Girard et al. reported a positive cooperativity at anionic strength of 5 mM by fitting with the McGhee and VonHippel model. Apart from the slight difference in pH, this ismost likely caused by the use of frontal analysis continuouscapillary electrophoresis (FACCE) and not by the choice of adifferent fitting model. When performing FACCE below the pIof the protein in a bare fused silica capillary, binding of theprotein to the negatively charged capillary wall is inevitable.47

In the initial region of the binding isotherm the concentrationof free protein is very low, because the majority of �-lg is boundto pectin. The little amount of �-lg that is not bound to pectinbinds to the capillary wall, leading to an underestimation offree �-lg at low concentrations of free �-lg. This was observedin our laboratory when trying to employ the FACCE techniqueto the �-lg-pectin system. Replacing the bare fused silicacapillary by a poly(vinyl alcohol) coated capillary did notimprove results. An underestimation of the free �-lg concentra-tion at low concentrations of free �-lg leads to an apparentpositive cooperativity, as �-lg seems to bind more profoundlyafter some �-lg is already bound to the pectin. The fluorescenceanisotropy measurements, used in this study, are not sensitiveto this problem.

The binding of �-lg to LMP shows n increasing slowly justabove unity for an ionic strength of 15-45 mM. At this higherionic strength, the repulsive electrostatic interactions betweenthe �-lg molecules are reduced, allowing a �-lg molecule toadsorb unhindered by its neighbors. Above 45 mM, n becomeslarger than 1.5. Positive cooperativity for the adsorption of �-lgto pectin can be the result of two effects: attraction between�-lg molecules or a change in the binding sites on pectin creatingnew binding sites or binding sites of a higher affinity.39 Pectinfolds into a helix structure in solution.48 It is possible thatbinding �-lg to pectin disrupts this helical structure, thuschanging the binding sites on pectin and allowing subsequent�-lg molecules to bind more easily. �-Lg molecules are knownto form higher oligomers between pH 4.0 and 5.2,42 theseaggregates are favored at low temperature (<10 °C) and low

ionic strength, both conditions that are not applicable to thecurrent experimental conditions. Attraction between �-lg mol-ecules is therefore unlikely to be the cause of the positivecooperativity.

For HMPB, there is no maximum observed in kobs. With theexception of 4 mM, kobs is lower for HMPB than for LMP at allionic strengths. As revealed by analysis of the size anddistribution of n-Me-GalA residues, blocks of n-Me-GalA alongthe HMPB backbone are relatively isolated, and they are smallerin size than the blocks on LMP. This may be the reason that, at4 mM, kobs is higher for HMPB than for LMP. When the size ofthe n-Me-GalA blocks on HMPB are of the optimal size for�-lg binding, the repulsion of unfavorable negative charges on�-lg and n-Me-GalA residues outside the binding site may bereduced as on HMPB these places are occupied by Me-GalAresidues. Also, as the binding site is isolated from others,unfavorable electrostatic repulsion between �-lg molecules willbe lower for binding to HMPB than to LMP. This is also a likelyreason for the absence of a maximum in kobs with increasingionic strength: there is not sufficient repulsion between HMPB

and �-lg and between �-lg neighbors to reach an optimum inattractive and repulsive forces with increasing ionic strength.When the ionic strength is subsequently increased theserepulsive interactions diminish, making the larger n-Me-GalAblocks on LMP favorable over the smaller blocks on HMPB.Increasing the ionic strength results in a decrease in kobs. Thereis some slight negative cooperativity at an ionic strength of 4and 10 mM, but at higher ionic strength the cooperativityparameter remains close to unity.

HMPR, like HMPB, shows no maximum in kobs probablybecause of similar reasons as outlined above for HMPB. kobs isalso considerably lower at the various ionic strengths. As onlya very few intermediate blocks of n-Me-GalA are present onHMPR, the binding sites on HMPR do not contain sufficientn-Me-GalA to tightly bind �-lg. A positive cooperativity isobserved above an ionic strength of 15 mM, most likely againdue to the disruption of the pectin helix by adsorption of �-lg.

Conclusions

Fluorescence anisotropy is an excellent technique to measureimmobilization of proteins and, hence, to determine bindingisotherms of protein to anionic polysaccharides. It has theadvantage over FACCE that it can be used at any pH, providingthe solution is optically clear. Because the measurements canbe performed in microtiter plates, the sample volumes can bekept low and an entire binding isotherm can be measured withina few minutes.

The physicochemical fine structure of the used pectins hasbeen elucidated in detail, revealing that LMP has large blocksof n-Me-GalA of a homogeneous block size that are closetogether. HMPB has large blocks of n-Me-GalA that are isolatedon the pectin backbone, with a heterogeneous distribution.HMPR has only very few isolated, relatively small blocks ofn-Me-GalA that are of a homogeneous size.

The Hill equation is preferred over the Langmuir modelbecause of the additional information about the cooperativityof binding. The binding constant, kobs, obtained from the Hillequation and the Langmuir model are comparable. For LMP,kobs shows a maximum at 10 mM, due to the balance betweenattractive and repulsive forces. Cooperativity is shown toincrease for binding of �-lg to LMP with increasing ionicstrength. At low ionic strength there is a slight negativecooperativity, caused by electrostatic repulsion between neigh-

Figure 5. Dependency of kobs on ionic strength of �-lg to pectin atpH 4.25 as obtained by the Langmuir (filled symbols) and Hill bindingmodels (open symbols): LMP (b,O), HMPB (1,3), and HMPR (9,0);lines are merely to guide the eye.

Binding of �-Lactolobulin to Pectins Biomacromolecules, Vol. 10, No. 12, 2009 3251

boring �-lg molecules. Above 45 mM, a positive cooperativityis observed, most likely caused by disrupting the helix fold ofpectin, which makes more binding places accessible to �-lg.For HMPB, there is no maximum in kobs, but at 4 mM, kobs

reaches a higher value than for LMP. This is caused by thenature of the blocks of n-Me-GalA on HMPB. The blocks aresmaller and of an isolated nature, compared to those on LMP.This leads to less electrostatic repulsion between �-lg and n-Me-GalA residues on pectin, but also between �-lg neighbors and,hence, a higher kobs. The cooperativety coefficient for bindingof �-lg to HMPB is close to unity, indicating noncooperativebinding over the entire range of ionic strengths. Binding of �-lgto HMPR gives the lowest kobs. The random distribution of n-Me-GalA leads to binding sites that have a reduced affinitycompared to the binding sites on the other two pectins, probablybecause of a smaller number of n-Me-GalA residues per site.HMPR shows positive cooperativity at an ionic strength above15 mM. The most likely explanation for this is again changesin the helix conformation of pectin, induced by the binding of�-lg.

To strongly bind �-lg to a pectin, it is best to choose a pectinwith a blockwise distribution of methyl esters. This leads tobinding sites with a higher affinity. When binding in media ofhigher ionic strengths is required, a low methyl esterified pectinis suited best. At low ionic strength it may be beneficial to usea high methyl esterified pectin with a large blocksize, becausethe isolated nature of the n-Me-GalA residues reduces theelectrostatic repulsion between pectin and �-lg as well asbetween �-lg neighbors.

Acknowledgment. The Dutch Ministry of Economic Affairssupported this research through the program IOP-IndustrialProteins.

Supporting Information Available. Tables containing thephysicochemical characteristics of the three pectins, Γm and theHill coefficients for the binding of �-lg to the three pectins.This material is available free of charge via the Internet at http://pubs.acs.org/.

References and Notes(1) Schmitt, C.; Sanchez, C.; Desobry-Banon, S.; Hardy, J. Crit. ReV. Food

Sci. Nutr. 1998, 38, 689–753.(2) Gharsallaoui, A.; Roudaut, G.; Chambin, O.; Voilley, A.; Saurel, R.

Food Res. Int. 2007, 40, 1107–1121.(3) Guzey, D.; McClements, D. J. J. Agric. Food Chem. 2007, 55, 475–

485.(4) Laurent, M. A.; Boulenguer, P. Food Hydrocolloids 2003, 17, 445–

454.(5) Sandrou, D. K.; Arvanitoyannis, I. S. Crit. ReV. Food Sci. Nutr. 2000,

40, 427–447.(6) Schmitt, C.; Palma da Silva, T.; Bovay, C.; Rami-Shojaei, S.; Frossard,

P.; Kolodzieczyk, E.; Leser, M. E. Langmuir 2005, 21, 7786–7795.(7) Vanderveen, J. E.; Glinsmann, W. H. Annu. ReV. Nutr. 1992, 12, 473–

487.(8) Wang, Y.-f.; Gao, J. Y.; Dubin, P. L. Biotechnol. Prog. 1996, 12,

356–362.(9) Tribet, C. Complexation between Amphiphilic Polyelectrolytes and

Proteins: From Necklaces to Gels. In Physical Chemistry of Polyelec-trolytes; Radeva, T., Ed.; Marcel Dekker, Inc.: New York, 2001; pp687-741.

(10) Gao, J. Y.; Dubin, P. L. Biopolymers 1999, 49, 185–193.(11) Park, J. M.; Muhoberac, B. B.; Dubin, P. L.; Xia, J. Macromolecules

1992, 25, 290–295.(12) Weinbreck, F.; De Vries, R.; Schrooyen, P.; De Kruif, C. G.

Biomacromolecules 2003, 4, 293–303.

(13) Sperber, B. L. H. M.; Schols, H. A.; Cohen Stuart, M. A.; Norde, W.;Voragen, A. G. J. Food Hydrocolloids 2009, 23, 765–772.

(14) Porcar, I.; Cottet, H.; Gareil, P.; Tribet, C. Macromolecules 1999, 32,3922–3929.

(15) Gao, J. Y.; Dubin, P. L.; Muhoberac, B. B. Anal. Chem. 1997, 69,2945–2951.

(16) Gao, J. Y.; Dubin, P. L.; Muhoberac, B. B. J. Phys. Chem. B 1998,102, 5529–5535.

(17) Girard, M.; Turgeon, S. L.; Gauthier, S. F. J. Agric. Food Chem. 2003,51, 6043–6049.

(18) Seyrek, E.; Dubin, P. L.; Tribet, C.; Gamble, E. A. Biomacromolecules2003, 4, 273–282.

(19) Hattori, T.; Kimura, K.; Seyrek, E.; Dubin, P. L. Anal. Biochem. 2001,295, 158–167.

(20) Porcar, I.; Gareil, P.; Tribet, C. J. Phys. Chem. B 1998, 102, 7906–7909.

(21) Østergaard, J.; Khanbolouki, A.; Jensen, H.; Larsen, C. Electrophoresis2004, 25, 3168–3175.

(22) Le Saux, T.; Varrenne, A.; Perreau, F.; Siret, L.; Duteil, S.; Duhau,L.; Gareil, P. J. Chromatogr., A 2006, 1132, 289–296.

(23) Lakowicz, J. R. Fluorescence Anisotropy. In Principles of FluorescenceSpectroscopy, 3rd ed.; Lakowicz, J. R., Ed.; Springer: New York, 2006;pp353-382.

(24) Maleki, S. J.; Royer, C. A.; Hurlburt, B. K. Biochemistry 2002, 41,10888–10894.

(25) Dragan, A. I.; Frank, L.; Liu, Y.; Makeyeva, E. N.; Crane-Robinson,C.; Privalov, P. L. J. Mol. Biol. 2004, 343, 865–878.

(26) Arosio, D.; Costantini, S.; Kong, Y.; Vindigne, A. J. Biol. Chem. 2004,279, 42826–42835.

(27) Datta, K.; LiCata, V. J. Nucleic Acids Res. 2003, 31, 5590–5597.(28) Voragen, A. G. J.; Pilnik, W.; Thibault, J. F.; Axelos, M. A. V.; Renard,

C. M. G. C. Pectins. In Food Polysaccharides and their Applications;Stephen, A. M., Ed.; Marcel Dekker, Inc.: New York, 1995; pp287-339.

(29) Daas, P. J. H.; Meyer-Hansen, K.; Schols, H. A.; De Ruiter, G. A.;Voragen, A. G. J. Carbohydr. Res. 1999, 318, 135–145.

(30) Dickinson, E. J. Chem. Soc., Faraday Trans. 1998, 94, 1657–1669.(31) Van Vliet, T.; Lakemond, C. M. M.; Visschers, R. W. Curr. Opin.

Colloid Interface Sci. 2004, 9, 298–304.(32) Dickinson, E. Colloids Surf., B 1999, 15, 161–176.(33) Ganzevles, R. A.; Zinoviadou, K.; Van Vliet, T.; Cohen Stuart, M. A.;

De Jongh, H. H. J. Langmuir 2006, 22, 10089–10096.(34) Daas, P. J. H.; Boxma, B.; Hopman, A. M. C. P.; Voragen, A. G. J.;

Schols, H. A. Biopolymers 2001, 58, 1–8.(35) Daas, P. J. H.; Voragen, A. G. J.; Schols, H. A. Biopolymers 2001,

58, 195–203.(36) De Jongh, H. H. J.; Groneveld, T.; De Groot, J. J. Dairy Sci. 2001,

84, 562–571.(37) Jayat, D.; Gaudin, J. C.; Chobert, J. M.; Burova, T. V.; Holt, C.;

McNae, I.; Sawyer, L.; Haertle, T. Biochemistry 2004, 43, 6312–6321.(38) Guillotin, S. E.; Van Loey, A.; Boulenguer, P.; Schols, H. A.; Voragen,

A. G. J. Food Hydrocolloids 2007, 21, 85–91.(39) Norde, W. Colloids and Interfaces in Life Sciences, 1st ed.; Marcel

Dekker Inc.: New York, 2003; pp 263-283.(40) Daas, P. J. H.; Voragen, A. G. J.; Schols, H. A. Carbohydr. Res. 2000,

326, 120–129.(41) Girard, M.; Turgeon, S. L.; Gauthier, S. F. Food Hydrocolloids 2002,

16, 585–591.(42) Verheul, M.; Pedersen, L. S.; Roefs, P. F. M.; De Kruif, C. G.

Biopolymers 1999, 49, 11–20.(43) Rees, D. A.; Wight, A. W. J. Chem. Soc. B 1971, 7, 1366–1372.(44) McGhee, J. D.; Von Hippel, P. H. J. Mol. Biol. 1974, 86, 469–489.(45) McGhee, J. D.; Von Hippel, P. H. J. Mol. Biol. 1976, 103, 679.(46) Hallberg, R. K.; Dubin, P. L. J. Phys. Chem. B 1998, 102, 8629–

8633.(47) Seyrek, E.; Hattori, T.; Dubin, P. L. Frontal Analysis Continuous

Capillary Electrophoresis for Protein-Polyelectrolyte Binding Studies.In Capillary Electrophoresis of Proteins and Peptides; Strege, M. A.,Lagu, A. L., Eds.; Humana Press: Totowa, NJ, 2004; Vol. 276, pp217-228.

(48) Perez, S.; Mazeau, K.; Herve du Penhoat, C. Plant Physiol. Biochem.2000, 38, 37–55.

BM900812X

3252 Biomacromolecules, Vol. 10, No. 12, 2009 Sperber et al.