m yield

d data®ter

l state

y

Journal of Colloid and Interface Science 288 (2005) 412–422www.elsevier.com/locate/jcis

Foaming and interfacial properties of hydrolyzedβ-lactoglobulin

J.P. Davis, D. Doucet, E.A. Foegeding∗

North Carolina State University, Department of Food Science, Box 7624, Raleigh, NC 27695-7624, USA

Received 26 January 2005; accepted 1 March 2005

Available online 5 April 2005

Abstract

β-lactoglobulin (β-lg) was hydrolyzed with three different proteases and subsequently evaluated for its foaming potential. Foastress (τ0) was the primary variable of interest. Two heat treatments designed to inactivate the enzymes, 75◦C/30 min and 90◦C/15 min,were also investigated for their effects on foamτ0. Adsorption rates and dilatational rheological tests at a model air/water interface aideinterpretation. All unheated hydrolysates improved foamτ0 as compared to unhydrolyzedβ-lg, with those of pepsin and Alcalase 2.4Lbeing superior to trypsin. Heat inactivation negatively impacted foamτ0, although heating at 75◦C/30 min better preserved this paramethan heating at 90◦C/15 min. All hydrolysates adsorbed more rapidly at the air/water interface than unhydrolyzedβ-lg, as evidenced bytheir capacity to lower the interfacial tension. A previously observed relationship between interfacial dilatational elasticity (E′) andτ0 wasgenerally confirmed for these hydrolysates. Additionally, the three hydrolysates imparting the highestτ0 not only had high values ofE′(approximately twice that of unhydrolyzedβ-lg), they also had very low phase angles (essentially zero). This highly elastic interfaciais presumed to improve foamτ0 indirectly by improving foam stability and directly by imparting resistance to interfacial deformation. 2005 Elsevier Inc. All rights reserved.

Keywords:Dilatational modulus; Dilatational elasticity; Adsorption; Interfacial rheology;β-lactoglobulin; Peptide; Hydrolysate; Foam; Yield stress; Whe

protein

ica-ve

ghed

itiesnd

osen-asede-

ing

teinnsialea-ly.ly

heir-nt isom-tion

es.s of-

1. Introduction

Proteins function as natural surfactants in many appltions that involve foam production. Egg white proteins hatraditionally served this role in the food industry, althousubstitution with other proteins, including those derivfrom bovine milk, is becoming more prevalent[1]. Commonmeans of evaluating foam surfactants include their capacto efficiently form foams (foamability), stabilize foams, aimpart specific foam rheological properties[2,3]. Foam rhe-ological studies have received far less attention than thpertaining to foamability and/or foam stability, although uderstanding the mechanisms responsible for protein-bfoam yield stress (τ0) has been a recent focus for our rsearch group[4–8].

* Corresponding author. Fax: +1 919 515 7124.E-mail address:allen_foegeding@ncsu.edu(E.A. Foegeding).

0021-9797/$ – see front matter 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.jcis.2005.03.002

Enzymatic hydrolysis is a common means for improvthe foaming potential of protein ingredients[9]. A commonapproach for evaluating the foaming performance of prohydrolysates is the whipping of dilute hydrolysate solutio(�0.05% w/v) in graduated cylinders, after which the initfoam height and its decrease with time are taken as msurements of foamability and foam stability, respectiveAccordingly, foamability and foam stability were markedimproved for a variety of hydrolysates as compared to tunhydrolyzed counterparts[10–14]. Chromatographic characterizations of these materials suggest this improvemepartially attributable to the reduced size of peptides as cpared to proteins, which promotes a more rapid adsorpat the air/water interface[10,11,14,15]. However, too muchhydrolysis may be detrimental for functionality purposFor example, hydrolysates containing high percentagelarge MW fragments (>7 kDa) were most correlated to im

proved foam stability in a comparative study of 44 differenthydrolysates[14]. There is also substantial evidence fromthe hydrolysate-emulsion literature suggesting that extensive

and

ted

hy-

delten-als.

on-re-s-)

ndasticvedcalegter-ofsistactin-tedrm

-ara-st-

fromtein

sis,learal

ol-

rd-

mes

siblentssti-monnc-

m

9),. P-urer-her

-ndandq-omofsingem.

)

/vted

tiothepH-s

been

forhe

torol-

edfi-

PAb-a-

two

J.P. Davis et al. / Journal of Colloid

hydrolysis can be detrimental to functionality[9]. The rela-tive hydrophobicity of hydrolysates has also been correlawith improved foamability and foam stability[10,11,14]. Asdiscussed shortly, very little has been reported on proteindrolysis as it affects foam rheology.

Direct characterization of hydrolyzed proteins at moair/water or oil/water interfaces has received far less attion than foaming or emulsifying tests of these materiTryptic peptides derived fromβ-lactoglobulin (β-lg) mosteffective at decreasing interfacial tension were those ctaining distinct zones of hydrophobic and hydrophilicgions within a minimum molecular weight allowing this ditribution [15]. Increasing levels of hydrolysis (up to 86%for β-lg variant A with a protease specific for glutamic aaspartic acid residues decreased the interfacial shear elity and viscosity of these materials, but resulted in improfoam overrun and stability as determined via a small sfoaming test[16]. This was surprising as improved foaminperformance is typically associated with increases in infacial rheological moduli. Dilatational rheological testsan amphipathic peptide isolated from a tryptic hydrolyof β-casein showed surface behavior similar to the inprotein [17]. There has been very little reported on theterfacial dilatational rheological behavior of unfractionamixtures of protein hydrolysates, which is the typical foof these ingredients.

The capacity to predict and control foamτ0 has considerable practical significance to the food industry, as this pmeter relates well to the empirical concept of foam robuness, and more robust foams are generally desirablea processing perspective. Several commercial whey prohydrolysates were found to improve foamτ0 as comparedto whey protein isolate (WPI) on an equal protein baalthough the mechanism for this improvement was unc[6]. A potential correlation between interfacial dilatationelasticity(E′) andτ0 has been noted for whey proteins subilized across a range of electrostatic conditions[7] andvarying degrees of polymerization[8]; however, this rela-tionship was untested for hydrolyzed whey proteins. Accoingly, hydrolysates ofβ-lactoglobulin, the primary wheyprotein, were prepared from three common food enzyto further test the validity of thisE′ vs τ0 relationship, andto investigate any other potential mechanisms responfor foamτ0. Furthermore, the effects of two heat treatmedesigned to terminate enzymatic activity were also invegated, as post-hydrolysis heating of hydrolysates is compractice on an industrial scale, yet its effects on foam futionality were unclear.

2. Materials

Bovineβ-lg (97% protein, dry basis) was obtained fro

Davisco Foods International, Inc. (Le Sueur, MN).β-lgandα-lactalbumin made up 93% and 5% of total proteins,respectively, as determined by the manufacturer. Trypsin

Interface Science 288 (2005) 412–422 413

-

(porcine pancreas, Type II-S, EC 3.4.21.4, No. T-740pepsin (from porcine stomach mucosa, EC 3.4.23.1, No7000), Tween 20 (SigmaUltra, P-7949), Hide Powder Az(No. H-6268) and Fibrin-Blue (No. F-5255) were all puchased from Sigma Chemicals Co. (St. Louis, MO). Ttrypsin contained 1800N -α-benzoyl-L-arginine ethyl este(BAEE) units/mg of trypsin activity and 2N -benzoyl-L-tyrosine ethyl ester (BTEE) units/mg of chymotrypsin activity as indicated by the manufacturer. Note that BAEE aBTEE are standard substrates for determining trypsinchymotrypsin activity, respectively. Alcalase 2.4L® (a liuid preparation from subtilisin Carlsberg) was obtained frNovozymes (Franklinton, NC). All other chemicals werereagent grade quality. Deionized water was obtained ua Dracor Water Systems (Durham, NC) purification systThe resistivity was a minimum of 18.2 M� cm.

3. Methods

3.1. Hydrolysis conditions and degree of hydrolysis (DH

β-lg was rehydrated to a protein concentration of 5% wfor all three hydrolysates. Trypsin (10% w/v) was rehydrain 1 mN HCl prior to addition ofβ-lg. Hydrolysis conditionswere as follows: 40◦C, pH 8.0, enzyme to substrate ra(E:S) 1:300 (w/w) based on the dry protein. The pH ofenzymatic reaction was kept constant at 8.0 using theStat technique[18]. The final degree of hydrolysis (DH) wa5.6% as determined by pH-Stat. These conditions havepreviously used to produce tryptic hydrolysates[19,20]. Al-calase hydrolysis conditions were as follows: 45◦C, pH 8.0,E:S 1:20 (v/w). The reaction was allowed to proceed3 h and the pH was not controlled. The final pH of thydrolysate was 6.2. Prior to pepsin hydrolysis, theβ-lg so-lution was heated at pH 3.0 at 85◦C for 15 min to improvedigestibility [21]. The solution was allowed to equilibrateroom temperature before enzyme addition. Pepsin hydysis conditions were as follows: 40◦C, pH 3.0, E:S 1:15(w/w) based on the dry protein. The reaction was allowto proceed for 3 h and the pH was not controlled. Thenal pH of the hydrolysate was 3.4. Theo-phthaldialdehyde(OPA) method was used to determineα-amino groups andto calculate the DH for all hydrolysates[22]. In this method,α-amino groups released by hydrolysis react with the Oreagent andβ-mercaptoethanol to form an adduct that asorbs strongly at 340 nm, quantification of which is indictive of DH.

3.2. Termination conditions

Enzymatic reactions were terminated by heating atdifferent time/temperature combinations: 90◦C/15 min[14]

◦

and 75 C/30 min. Preliminary experiments using Fibrin-Blue and Hide Powder Azure confirmed the efficacy of thesetreatments[23,24]. Enzymatic reactions were also placed

and

lesentsion.

1%,M7.0

hos-uffeandoomureaid-n a

ctron

byC).il-

ea

nm.owewith

col-s

he0%rane

e

.o-teinne-

me-

ntsding

tentsent

wastantd in

orate

r thethe

an-sed

mentini-

in-cial

un-ionsamropem-sedu-

y ain-od-asto as

414 J.P. Davis et al. / Journal of Colloid

on ice after hydrolysis to slow enzymatic activity. Sampwere frozen before all analyses except foam measuremwhich were done immediately after hydrolysate formatand equilibration to room temperature (heated samples)

3.3. Turbidity experiments

Samples were diluted to a protein concentration ofw/v with different buffers. McIlvain buffers of pH 3.0, 4.0and 5.0 were prepared using 0.1 M citric acid and 0.2dibasic sodium phosphate. Phosphate buffers of pH 6.0,and 8.0 were prepared from 0.1 M monobasic sodium pphate and 0.1 M dibasic sodium phosphate. Carbonate bof pH 9.0 was prepared with 0.1 M sodium carbonate0.1 M sodium bicarbonate. Samples were analyzed at rtemperature and in the presence of SDS (5% w/v) and(6 M). Solutions were equilibrated for 1 h before turbity measurements were made in triplicate at 500 nm oSpectronic 20 Genesys spectrophotometer (Thermo EleCorporation, Waltham, MA).

3.4. Molecular mass distribution profiles

Molecular mass distribution profiles were determinedhigh performance size exclusion chromatography (HPSEAnalyses were performed on a Waters HPLC system (Mlipore, Milford, MA) consisting of an injector (RheodynModel 7725i, Cotati, CA), two pumps (Model 515), andphotodiode array detector (Model 2996) adjusted at 220Data acquisition and analysis were made using the Empchromatography software. The analysis was performeda TSK-Gel G2000 SWXL column (0.78 i.d.× 30 cm) fromTosoHaas (Montgomeryville, PA) connected to a guardumn (0.6 i.d.× 4 cm) filled with the same matrix. Samplewere diluted to a concentration of 1% protein w/v with tmobile phase composed of 0.1% triflouroacetic acid in 7aqueous acetonitrile and then filtered on a 0.2-µm memb(PVDF). Elution was performed isocratically in the sammobile phase at 28◦C with a flow rate of 0.6 ml/min over30 min.

3.5. Foam generation

A Kitchen Aid Ultra Power Mixer (Kitchen Aid, StJoseph’s, MI) with a 4.5 qt (4.3 l) stationary bowl and rtating beaters was used for foam formation. 5% w/v prosolutions (225 ml) were whipped at speed setting 8 (platary rpm of 225 and beater rpm of 737) for 20 min.

3.6. Yield stress measurements

Foam yield stress was determined by vane rheotry as recently described[7,8]. A Brookfield 25xLVTDV-

ICP (Brookfield Engineering Laboratories, Inc. Middleboro,MA) viscometer was used at a speed of 0.3 rpm. The vanehad a 10 mm diameter and 40 mm length. Maximum torque

Interface Science 288 (2005) 412–422

,

,

r

r

response (M0) was documented for each of 3 measuremetaken per foam and used to calculate yield stress accorto published information[25]:

(1)τ0 = M0

[(h/d) + (1/6)](πd3/2),

where τ0 is the yield stress, andh and d are the heighand diameter of the vane. Three consecutive measurem(4 min max.) were taken per treatment, and each treatmwas replicated a minimum of 3 times. This measurementstable over the entire measurement time. We felt it importo produce foams at protein concentrations actually usefood industry, i.e.�5% w/v, as foam rheological behaviis expected to be quite different from the dilute hydrolysfoams described in the introduction.

3.7. Overrun

Overrun measurements were begun immediately aftefinal τ0 measurement. Foam was carefully scooped frombowl in a circular pattern with a rubber spatula, filling a stdard weight boat (100 ml) 3 times. The mean value was uto calculate overrun and air phase fraction according to[26]:

%Overrun

(2)= (wt 100 ml solution) − (wt 100 ml foam)

wt 100 ml foam× 100,

(3)Air phase fraction= φ = %overrun

%overrun+ 100.

Overrun measurements were stable over the measuretime (3 min max.). Each treatment was replicated a mmum of 3 times to determine the average overrun.

3.8. Interfacial measurements

The foaming solutions or their dilutions were used forterfacial measurements. The following method of interfadata collection has been recently described[7,8]. An auto-mated contact angle goniometer (Rame–Hart, Inc., Motain Lakes, NJ) was used for data collection and calculatin combination with the DROPimage computer progr[27]. The dynamic surface tension of a 16-µl capillary dwas initially monitored for 5 min with a 1-s resolution. Thcapillary drop was formed within an environmental chaber at room temperature, in which standing water increathe relative humidity to minimize evaporation effects. Sinsoidal oscillations of the drop’s area were then input bvolume amplitude of 0.5 µl, and the resulting change interfacial tension was used to determine the dilatational mulus. Preliminary work confirmed this strain amplitude wwithin the linear viscoelastic regime and correspondedrelative interfacial area change of∼2.3%. From the modulu

and from the phase angle between the surface area changeand surface tension response, the DROPimage software cal-culatedE′ andE′′, which are equivalent to and proportional

and

, re-de-

lesteshy-por-oth

ting-and

proW

d atatedtive

in

tessthy-

vitynceding-

m

en-nsere

m-ilar

-d ater-for

n in-/v

in-

ates

andto

theH oflop-t ifn-

h inS andtesree-the

bi-the

ho-re-

J.P. Davis et al. / Journal of Colloid

to the elastic and viscous components of the interfacespectively. The details for these calculations have beenscribed elsewhere[27].

4. Results

Fig. 1 shows the molecular mass distribution profifor Alcalase (A), trypsin (B), and pepsin (C) hydrolysaobtained by HPSEC prior to heat treatment. Alcalasedrolysates had the broadest distribution and highest protion of small peptides. Trypsin and pepsin hydrolysates bhad narrower molecular mass distribution profiles. Heaat 90◦C/15 min or 75◦C/30 min induced little to no detectable changes in the hydrolysate profiles for Alcalasetrypsin (data not shown). However, heating at 75◦C/30 mininduced considerable changes in the pepsin hydrolysatefile (Fig. 1D) as seen by the appearance of smaller Mpeptides. The profile of the pepsin hydrolysate heate90◦C/15 min also changed as compared to the nonhesample, although not as drastically (data not shown). Naβ-lg is presented inFig. 1E, none of which was detectedany hydrolysate profile.

The degree of hydrolysis for the different hydrolysais presented inTable 1. Alcalase hydrolysates had the moextensive hydrolysis as compared to trypsin and pepsin

Fig. 1. HPSEC profiles of various hydrolysates. A: Alcalase—no heat;B: trypsin—no heat; C: Pepsin—no heat; D: pepsin—75◦C/30 min;E: β-lactoglobulin.

Interface Science 288 (2005) 412–422 415

-

Table 1Degree of hydrolysis (%) as determined by OPA methoda

Hydrolysate Heat treatment

No heat 75◦C/30 min 90◦C/15 min

Alcalase 15.6 14.2 13.7Trypsin 6.7 4.6 4.8Pepsin 5.2 8.4 4.7

a Values are the average of three replications.

drolysates, which showed similar results. Residual actiwas observed for the nonheated hydrolysates as evideby their higher degrees of hydrolysis. Surprisingly, heatthe pepsin hydrolysate at 75◦C/30 min induced more hydrolysis than heating at 90◦C/15 min.

Fig. 2summarizesτ0 and overrun of foams prepared frothe three hydrolysates and the unhydrolyzed substrate,β-lg(pH 7.0). Note these solutions were all at a protein conctration of 5% w/v. Data for whey protein isolate solutioat protein concentrations of 5 and 10% w/v protein wincluded for comparison with previous work[4–8]. Heat-ing at 90◦C/15 min reducedτ0 for all hydrolysates, but thiseffect was minimal for the trypsin hydrolysate. The foaing behavior of the trypsin hydrolysates was more simto β-lg than either pepsin or Alcalase. Foamτ0 of hy-drolysate solutions heated at 75◦C/30 min was more similar to unheated hydrolysates than hydrolysates heate90◦C/15 min. Both heat treatments minimally affected ovrun, with a slight negative effect detectable, especiallythe pepsin hydrolysates. The decrease in overrun upocreasing the WPI protein concentration from 5 to 10% wis likely an artifact of the measurement resulting fromcreased drainage in the less concentrated solution[2,8]. Itis notable that in a screening study of various hydrolysprepared from WPI, of whichβ-lg typically makes up�70%of the protein, Alcalase hydrolysates had better overrunstability (5% w/v whipped protein foams) as comparedthose of trypsin and pepsin[28].

Heat treatments induced noticeable turbidity for bothAlcalase and pepsin hydrolysates at their respective p6.2 and 3.4. The aggregates causing this turbidity devement were insoluble, as they would eventually sedimengiven enough time. Turbidity was quantified by optical desity measurements at 500 nm as a function of pH, botthe presence and absence of the chaotropic agents, SDUrea (Fig. 3). All heated and unheated trypsin hydrolysadisplayed a sharp peak in turbidity at pH 4.0, in good agment with similarly prepared hydrolysates, as many ofgenerated peptides have pI near 4.0[20]. Both SDS andurea fully solubilized these aggregates. SDS fully solulized all samples regardless of conditions, except forAlcalase hydrolysates heated at 90◦C/15 min. This reflectsthe charged surfactant’s capacity to interact with hydropbic protein residues while simultaneously imparting a

pulsive charge[29]. Both Alcalase and pepsin hydrolysateswere more soluble thanβ-lg in the absence of heat (datanot shown), which behaved very similarly to trypsin hy-

ols ap

416 J.P. Davis et al. / Journal of Colloid and Interface Science 288 (2005) 412–422

Fig. 2. Yield stress and air phase fraction of foams prepared from various hydrolysates. Error bars are standard deviations of mean values. Symbpear inthe figure.

Fig. 3. Turbidity of various hydrolysates as a function of pH and in the presence of 5% w/v SDS and 6 M urea. Error bars are standard deviations of meanvalues. Symbols appear in the figure.

J.P. Davis et al. / Journal of Colloid and Interface Science 288 (2005) 412–422 417

(5% ar

ave

sugvedrac-

%

zingas-

ng ahy-atesatere

heatdeeir

y--

was

hichre-n

ntlym-her

f Al-

s are

n inehaseratioin-zedfor

ichm a

Fig. 4. Typical dynamic surface tension measurements of hydrolysatesin the figure.

drolysates, in agreement with previous findings that hshown hydrolysis to improve solubility[9]. 6 M urea re-duced turbidity for heated Alcalase and pepsin samples,gesting H-bonding is partially responsible for the obseraggregation, although the reduction of hydrophobic intetions cannot be excluded[30].

The interfacial adsorption rate of all hydrolysates (5w/v) was faster than unhydrolyzedβ-lg as determined bydynamic surface tension measurements (Fig. 4A). Note thatadsorption rates were qualitatively assessed by visualithe rate of surface tension change, and quantitativelysessed by the surface tension observed at 5 min. Heati90◦C/15 min did not change adsorption of the Alcalasedrolysates, decreased adsorption of the trypsin hydrolysand increased the adsorption rate of the pepsin hydrolys(Figs. 4A and 4B). Dynamic surface tension profiles wegenerally similar for hydrolysates heated at 90◦C/15 minand 75◦C/30 min (data not shown). Note theβ-lg was atneutral pH and the reduced adsorption observed uponing was a function of the molecule forming soluble, disulfilinked aggregates, which diffuse more slowly due to thlarger size[8].

Fig. 5 showsE′ of the various hydrolysates and unhdrolyzedβ-lg, all at a concentration of 5% w/v. All samples were aged 5 min and the frequency of oscillation0.04 Hz. In the absence of heat,E′ was notably improvedfor all hydrolysates as compared toβ-lg. Note again theβ-lgwas at neutral pH and the heat treatments, both of winduced the formation of soluble, disulfide linked agggates, slightly increasedE′. The effects of this phenomenoon WPI interfacial rheological behavior have been recedocumented[8]. E′ of the unheated hydrolysates was siilar for pepsin and Alcalase, and both were notably higthan that of trypsin. Both heat treatments reducedE′ for allhydrolysates, and this reduction was greater for those ocalase and trypsin as compared to pepsin hydrolysates.

Fig. 6 shows the phase angle of the various hydrolysatesand their dilutions as a function of surface pressure (Π ). Π isdefined as the net change in surface tension for the solven

w/v). A: No heat; B: 90◦C/15 min. Note thatβ-lg is at neutral pH. Symbols appe

-

t

,s

-

Fig. 5. Dilatational elasticity of hydrolysates andβ-lg (pH 7.0), all at aconcentration of 5% w/v. Lines are added to guide the eye. Error barstandard deviations of mean values. Symbols appear in the figure.

(water) upon surfactant adsorption at a given time (5 mithis case). Variations inΠ were a function of the hydrolysatconcentrations as listed in the figure legend. Note the pangle corresponds to the viscous and elastic responsewithin the dilatational modulus, with higher phase anglesdicative of a greater viscous response. Data for unhydrolyβ-lg at pH 7.0 is included with the pepsin hydrolysatescomparison with all hydrolysates. The concentration ofβ-lgwas 10, 5, 1 and 0.1% w/v, from highest to lowestΠ .

5. Discussion

Hydrolysates were characterized by HPSEC, in whcomponents are separated by selective exclusion fro

t

porous matrix based on differences in their hydrodynamicvolumes, with smaller molecules eluting last. Alcalase isa serine alkaline protease produced by a selected strain of

418 J.P. Davis et al. / Journal of Colloid and Interface Science 288 (2005) 412–422

Fig. 6. Interfacial dilatational phase angle of hydrolysates as a function of surface pressure. Lines are added to guide the eye.β-lg concentration values arective hydrol

il-ep-ino

b-inolu),ylfoades

ine

C

for

han

atedin-

han-

edtingnghis

r ton ofsionr to

sethe

neu-rm

more

ex-andon-cial

oredy-

ofpos-ich

dif-, anot

cedtes

imits areon-asi-

for-sin

hy-

10, 5, 1, and 0.1% w/v from highest to lowest surface pressure, respedilutions (% w/v) appear in the figure.

Bacillus licheniformis. Its main enzyme component, subtisin Carlsberg, has broad specificity, hydrolyzing most ptide bonds, preferentially those containing aromatic amacid residues[31]. Within whey proteins, Alcalase was oserved to have a high specificity for not only aromatic amacid residues (Phe, Trp, and Tyr) but also for acidic (Gsulfur-containing (Met), aliphatic (Leu and Ala), hydrox(Ser), and basic (Lys) residues[32]. The heterogeneity opeptides generated by Alcalase likely explains the brHPSEC distribution and high proportion of small peptid(Fig. 1A). Its high DH value supports this hypothesis (Ta-ble 1). Trypsin specifically cleaves at lysine and arginresidues for which there are 18 total sites in theβ-lg pri-mary sequence[33], explaining its relatively narrow HPSEpeptide profile (Fig. 1B).

Pepsin has a fairly broad specificity with a preferencecleaving after hydrophobic residues[34]. The HPSEC pro-file of unheated pepsin hydrolysate (Fig. 1C) and its DH(Table 1) suggests this hydrolysis was less extensive teither Alcalase or trypsin. As previously mentioned,β-lgis inherently resistant to pepsin digestion unless prehe[21], as were the current pepsin hydrolysates. Pepsinactivation at 75◦C/30 min increased the DH (Table 1), inagreement with its HPSEC profile, which was broader tits unheated counterpart (Figs. 1C and 1D). It seems that after the initial 3 h hydrolysis, heating at 75◦C/30 min allowedmore of the partially digestedβ-lg substrate to be accessby pepsin prior to its inactivation. We suspect that heaat 90◦C/15 min inactivates pepsin more rapidly, minimiziany further hydrolysis, as supported by the lower DH of thydrolysate (Table 1).

Solution pH of the hydrolysates was not adjusted prioheat treatments to simulate immediate heat inactivatiosuch enzyme based ingredients. Accordingly, the deciwas made to not to adjust the pH of the hydrolysates prio

the interfacial measurements in order to be consistent withfoaming measurements. The interfacial properties of pro-teins are well documented to be affected by the solution

ly. Error bars are standard deviations of mean values. Symbols for theysate

pH [7,35,36]. A prominent example is the typical increain adsorption rates and interfacial rheological moduli aspH approaches a protein’s isoelectric point, as the nettrally charged molecules adsorb more efficiently and fointermolecular interactions between adsorbed specieseffectively, due to decreased electrostatic repulsion[36]. Ac-cordingly, the pH of the current hydrolysates was alsopected to significantly influence the surface properties (foaming behavior) of these materials. Work is currentlygoing to investigate pH effects on the foaming and interfabehaviors of these hydrolysates.

Increased hydrolysate DH seemed to promote a mrapid adsorption at the air/water interface as indicated bynamic surface tension measurements (Fig. 4A andTable 1).Increasing the DH should 1) yield a higher percentagesmaller peptides and 2) increase the potential for exing previously buried hydrophobic residues, both of whshould promote a more rapid adsorption[3]. However, dueto the different specificities of each enzyme and henceferent distributions of peptides within each hydrolysatedefinitive statement relating DH and adsorption rates ispossible.

Heating at both time/temperature combinations induaggregation for the Alcalase and pepsin hydrolysa(Fig. 3). Enhanced aggregation in the bulk phase should ladsorption at the air/water interface as the two processeessentially competing for the peptides’ hydrophobic ctacts [37,38]. Adsorption of the trypsin hydrolysates wreduced upon heating (Fig. 4), even though it showed minmal heat induced aggregation near its pH of 8.0 (Fig. 3). Wetherefore suspect this reduced adsorption was due to themation of soluble, disulfide linked aggregates in the tryphydrolysate, as discussed previously forβ-lg. This phe-nomenon readily occurred in similarly prepared trypsindrolysates at ambient temperatures[19], and heating should

promote further this reaction. Adsorption of Alcalase hy-drolysates did not change upon heating at 90◦C/15 min,despite extensive aggregation (Figs. 3 and 4). This was

and

s ththede-tter

atesfileeata-

orp-ated

atsedly.edbe-ethe

einst

ntsim

thaner-ssesl re-s of

forare

ions

tes

tideson

atin-hy-

be-thatally

re pr

less

r,larlatetes;

c-

hy-ns at

ilityinoionatesd

sn

lit-

rent

cen-te,ated

y-einev-

tox-is-

andtioninods

bet-ates

turee tomesav-all

er

tandslyicad-for-

J.P. Davis et al. / Journal of Colloid

surprising as these aggregates were insoluble. Perhapremaining soluble peptides are more surface active, orinsoluble aggregates are somehow contributing to thecrease in interfacial tension. More work is necessary to beunderstand this phenomenon. Heating at 90◦C/15 min actu-ally increased the adsorption rate of the pepsin hydrolys(Fig. 4B). However, analysis of the pepsin HPSEC prosuggested a higher content of smaller peptides upon hing at 90◦C/15 min, which seemingly override any negtive aggregation effects by promoting a more rapid adstion. Indeed, adsorption of the pepsin hydrolysate heat 75◦C/30 min was even more rapid than that heated90◦C/15 min (data not shown), consistent with the increaDH observed for this hydrolysate as discussed previous

E′ primarily reflects the inherent rigidity of the adsorbsurfactant and the magnitude of the lateral interactionstween adsorbed surfactants[39]. Comparative studies havshown that rigid proteins better transmit forces acrossinterface upon deformation as compared to flexible protsuch asβ-casein[40,41]. One could imagine a very slighhydrolysis to potentially increase the flexibility ofβ-lg;however, no intactβ-lg was detected in any of the currehydrolysates. Hence, the generated peptides are of apler structure and able to sample fewer conformationsunhydrolyzedβ-lg. Once adsorbed at the air/water intface, these peptides might better transmit interfacial streas there would be less energy dissipated in structuraarrangements. This may partially explain the high valueE′ observed for many hydrolysates as compared toβ-lg;however, there is a broad range of peptides producedeach hydrolysate and their exact interfacial compositionsunclear. Furthermore, interfacial peptide–peptide attracthave not been considered, but will be discussed next.

E′ of the unheated Alcalase and pepsin hydrolysa(∼140 mN/m) were much greater thanβ-lg at pH 7.0(Fig. 5). We can speculate on the interfacial peptide–pepinteractions for the Alcalase hydrolysates by compariwith a similarly prepared material[29,32]. This earlier Al-calase hydrolysate was derived from WPI and gelledhigher concentrations (20% w/v) due to hydrophobicteractions between minimally charged peptides at thedrolysate’s natural pH of 6.0[29,32]. Comparison of the twoAlcalase hydrolysates revealed very similar interfacialhaviors (data not shown). Therefore, we hypothesizethese same hydrophobic associations between minimcharged peptides are also present at the interface, and amarily responsible for this hydrolysate’s high value ofE′.The unheated pepsin hydrolysates were considerablyhydrolyzed than those of Alcalase (Table 1), which mightpromote lower values ofE′ as discussed earlier; howeveagain this is ignoring the contribution from intermolecupeptide–peptide interfacial attractions. We cannot specutoo confidently on these interactions for pepsin hydrolysa

however, they are clearly strong. It is noteworthy that despitehaving drastically different HPSEC profiles and pH levels,the unheated pepsin and Alcalase hydrolysates behaved ver

Interface Science 288 (2005) 412–422 419

e

-

-

,

i-

similarly from a dilatational interfacial rheological perspetive.

Heat-induced aggregation in the Alcalase and pepsindrolysates seemed to hinder peptide–peptide interactiothe interface, resulting in the reduced values ofE′ (Fig. 5).As discussed previously, aggregation hinders accessibof functional groups, i.e. hydrophobic clusters of amacids that contribute to both the initial peptide adsorptand peptide–peptide interfacial attractions. The aggregformed forβ-lg upon heating at pH 7.0 were disulfide linkeand soluble and actually increasedE′. This was surprising aE′ of WPI, which is typically�70% β-lg, decreased upoformation of similar aggregates[8]. Similarly, E′ decreasedfor the heated trypsin hydrolysates which showed verytle aggregation near their pH value of 8.0 (Fig. 3) and wereexpected to have formed disulfide linkages[19]. More workis ongoing to better understand the effects of these diffeaggregate types on dilatational rheological behavior.

In addition to their high magnitudes ofE′, striking werethe very low phase angles observed at a protein contration of 5% w/v for the unheated Alcalase hydrolysaunheated pepsin hydrolysate and pepsin hydrolysate heat 75◦C/30 min (Fig. 6). Note the phase angle of unhdrolyzed β-lg steadily increased upon increasing protconcentration/Π , as has been previously observed for seral proteins[42]. This increased viscous response is dueincreasing interfacial relaxations, including diffusional echange with the bulk solution. Indeed, the dilatational vcosity (oil/water) of an isolated peptide derived fromβ-ca-sein increased with increasing peptide concentration,was speculated to be caused by a diffusional relaxamechanism[17]. However, it has been concluded thattime scales ranging from 1 to 1000 s, relaxation methwithin the interface must also be present[42]. These includestructural rearrangements of the surfactant, which wouldless likely at higher concentrations/�′s due to surfactancrowding at the interface[43]. This might explain the drastic drop in phase angle observed for these three hydrolysat higher concentrations, coupled with their loss of strucupon hydrolysis (as discussed previously). However, duthe wide range of peptides produced by different enzyand experimental conditions, a variety of interfacial behiors are not surprising. Accordingly, the phase angle ofhydrolysates did not show this dramatic decrease at highΠ

(Fig. 6).Frequency sweeps were conducted to better unders

hydrolysate interfacial behaviors, while simultaneoucomparing them toβ-lg and Tween20®, a typical nonionsmall molecular weight surfactant (SMWS). The rapidsorption and low values of surface tension observedmany of the hydrolysates (Fig. 4) suggested they might behave more like a typical SMWS, although the initialE′values suggested more of a protein-like behavior (Fig. 5).

y

Indeed, work with synthetic peptides revealed that minoramino acid substitutions can drastically alter a peptide’scapacity in transmitting uniaxial stretching forces at the in-

420 J.P. Davis et al. / Journal of Colloid and Interface Science 288 (2005) 412–422

Alcal bols for

deini-

ead-

or-de

ial

f, asons

omn-

. Itdis-ad

es.

ep-

tionands re-

like

ong-zednd

e,

tesor-mea-

PIal-

y-s

Fig. 7. Typical interfacial dilatational frequency sweeps ofβ-lg, unheatedthe dilutions (% w/v) appear in the figure.

terface, from much less to much greater thanβ-lg [44]. Datafor the unheated Alcalase hydrolysate is presented inFig. 7,along with β-lg and Tween20. Measurements were masequentially from lowest to highest frequency prior to antial 5 min aging. The low values ofE′ andE′′ observed forthe lowest concentrations ofβ-lg and Alcalase hydrolysatwere a function of not enough protein/peptide havingsorbed. Work with spread monolayers ofβ-lg suggests itundergoes non-diffusional relaxations occurring on theder of 20 s[45]. This timeframe is in the same magnituwith many of the frequencies tested inFig. 7, explaining thefrequency dependence ofE′ observed at higherβ-lg concen-trations, in addition to the increased likelihood of interfacdesorption upon increasing concentration[46]. In contrast,the frequency independence and much lower values oE′observed for Tween20 solutions are typical of a SMWSthese interfacial layers completely adjust to perturbatiwithin these measurement timeframes.

Increasing the Alcalase hydrolysate concentration fr1 to 5% w/v sawE′ become more frequency indepedent andE′′ drop to essentially zero (Fig. 7). This inter-face effectively behaved as a perfectly elastic 2-D solidseems that at 5% w/v, due to the decreased diffusiontances, these small, hydrophobic hydrolysate moleculessorb quite rapidly, forming very strong and elastic interfacInterfacial relaxations, which are manifest inE′′, seem to bereduced due to the simplified internal structure of the p

tides, and/or they may be occurring more rapidly than themeasurement timeframe. Unheated and 75◦C/30 min pepsinhydrolysates displayed similar responses to interfacial di-

ase hydrolysate and Tween20. Lines are added to guide the eye. Sym

-

latational frequency sweeps as a function of concentra(data not shown), despite their different compositionspH levels. Frequency sweeps of the trypsin hydrolysatevealed interfacial rheological behavior more similar toβ-lg.It seems all hydrolysate fractions do behave much moreproteins than typical SMWSs.

Theoretical analyses of foam rheology predict a strdependence on interfacial rheology[47]. Recent experimental observations for highly concentrated, protein-stabiliemulsions, which share many similarities with foams, foutheir dimensionless bulk elasticity,G′/(γ /r), to be posi-tively correlated withE′ [48]. Accordingly, previous workwith WPI foams (10% w/v protein) revealed a positivcurvilinear relationship in the plot ofE′ vs. τ0 [7,8].This trend was not as obvious for the current hydrolysa(Fig. 8A). However, there was still a general positive crelation between the two measurements. When thesesurements are combined with the data collected for W[7,8], the current values all fell on or above the previous vues (Fig. 8B). Foams of purifiedβ-lg typically have higherτ0 than those of WPI, andE′ of adsorbedβ-lg is gener-ally higher than that of WPI[7]. Therefore, the shift tohigher regimes seen inFig. 8B is partially explained by thesubstrate beingβ-lg and not WPI. The data for pepsin hdrolysates heated at 90◦C/15 min and Alcalase hydrolysateheated at 75◦C/30 min, strongly suggest that foamτ0 de-pends on other factor(s) besidesE′.

It is noteworthy that the three foams with the highestτ0

were formed from hydrolysates that hadbothhigh values ofE′ and very low phase angles (essentially zero). This highly

J.P. Davis et al. / Journal of Colloid and Interface Science 288 (2005) 412–422 421

tes. B r

tabi-

amflu-n

moses

ientsownityialmen

d

eoam

sate-o.for-ially

n-ting

entup-ce,Re-mareingnote-ilar

o-96,

00)

. A

er-

Food

agen,

gen,

gen,

Fig. 8. Dilatational elasticity vs foam yield stress. A: Current hydrolysain the figure.

elastic interfacial state may promote increased foam slization, which would indirectly promote higher foamτ0 byselecting for smaller bubbles[49,50]. This is assuming highvalues ofE′ and/or low phase angles do indeed retard fodestabilization mechanisms. However, a more direct inence on foamτ0 may be coming from the interfacial tensiogradients, which are manifest in the value ofE′. The in-fluence of these gradients on foam rheology becomesimportant in “dry” foams, i.e. foams with air phase volum(φ) above approximately 0.74[47]. Note theφ of all currentfoams is greater than 0.9. These interfacial tension gradimpart resistance to interfacial deformation and were shto be a primary contributor to a foam’s dilatational viscosin theory[47]. Accordingly, it seems logical these interfactension gradient effects are also present in these experital systems and are directly influencing foamτ0.

6. Conclusions

Hydrolysis of β-lg with Alcalase and pepsin producefractions that formed foams with significantly increasedτ0as compared to unhydrolyzedβ-lg. Trypsin hydrolysis onlyslightly improved foamτ0. Interfacial characterization of thhydrolysates revealed that samples that induced high fτ0 generally produced interfaces with highE′; however, thiswas not a prerequisite. Furthermore, the three hydrolyfractions imparting the highest values ofτ0 where all characterized by high values ofE′ andphase angles close to zerInterfacial rheological frequency sweeps confirmed themation of these extremely elastic interfaces with essentno viscous response. Heating the hydrolysates at 75◦C/30 min and 90◦C/15 min both successfully terminated ezymatic activity for the three hydrolysates. However, heaat 75◦C/30 min better preserved foamτ0 and this generally

′

reflected in higherE and/or a low interfacial phase an-gle. More work is ongoing to better understand some of theunique interfacial behaviors observed in these hydrolysates.

: Current hydrolysates plus WPI data from previous work[7,8]. Symbols appea

t

-

Acknowledgments

Paper No. 05-03 of the Journal Series of the Departmof Food Science, NCSU, Raleigh, NC 27695-7624. Sport from the North Carolina Agricultural Research ServiDairy Management, Inc., and the Southeast Dairy Foodssearch Center is gratefully acknowledged. We thank EmMcCrory for helping with foam data collection. Thanks aextended to Davisco Foods International, Inc., for supplyprotein. The use of trade names in this publication doesimply endorsement by the North Carolina Agricultural Rsearch Service of the products named nor criticism of simones not mentioned.

References

[1] E. Dickinson, Colloid Surf. B Biointerfaces 15 (1999) 161.[2] P.J. Halling, Crit. Rev. Food Sci. Nutr. 15 (1981) 155.[3] P.J. Wilde, D.C. Clark, in: G.M. Hall (Ed.), Methods of Testing Pr

tein Functionality, Blackie Academic & Professional, London, 19p. 110.

[4] C.W. Pernell, E.A. Foegeding, C.R. Daubert, J. Food Sci. 65 (20110.

[5] C.W. Pernell, E.A. Foegeding, P.J. Luck, J.P. Davis, Colloid SurfPhysicochem. Eng. Asp. 204 (2002) 9.

[6] P.J. Luck, N. Bray, E.A. Foegeding, J. Food Sci. 67 (2002) 1677.[7] J.P. Davis, E.A. Foegeding, F.K. Hansen, Colloid Surf. B Bioint

faces 34 (2004) 13.[8] J.P. Davis, E.A. Foegeding, J. Food Sci. 69 (2004) C404.[9] A. Kilara, D. Panyam, Crit. Rev. Food Sci. Nutr. 43 (2003) 607.

[10] P. Caessens, S. Visser, H. Gruppen, A.G.J. Voragen, J. Agric.Chem. 47 (1999) 2973.

[11] P. Caessens, W.F. Daamen, H. Gruppen, S. Visser, A.G.J. VorJ. Agric. Food Chem. 47 (1999) 2980.

[12] P. Caessens, H. Gruppen, S. Visser, G.A. van Aken, A.G.J. VoraJ. Agric. Food Chem. 45 (1997) 2935.

[13] P. Caessens, S. Visser, H. Gruppen, G.A. van Aken, A.G.J. Vora

Int. Dairy J. 9 (1999) 347.

[14] C. van der Ven, H. Gruppen, D.B.A. de Bont, A.G.J. Voragen, J. Agric.Food Chem. 50 (2002) 2938.

and

od

vist,

J.L.9.p-

gric.

od

36

airy

.lin.

ree-

99)

ood

ch-

p-

28.9)

00)

Sci.

r 19

108

04)

iller98,

.K.

73.n-

s.),ew

09

422 J.P. Davis et al. / Journal of Colloid

[15] S.L. Turgeon, S.F. Gauthier, D. Molle, J. Leonil, J. Agric. FoChem. 40 (1992) 669.

[16] R. Ipsen, J. Otte, R. Sharma, A. Nielsen, L.G. Hansen, K.B. QColloid Surf. B Biointerfaces 21 (2001) 173.

[17] J.M. Girardet, G. Humbert, N. Creusot, V. Chardot, S. Campagna,Courthaudon, J.L. Gaillard, J. Colloid Interface Sci. 245 (2002) 21

[18] J. Adler-Nissen, Methods in Food Protein Hydrolysis, Elsevier Aplied Science, London, 1986.

[19] P.E. Groleau, R. Jimenez-Flores, S.F. Gauthier, Y. Pouliot, J. AFood Chem. 50 (2002) 578.

[20] P.E. Groleau, P. Morin, S.F. Gauthier, Y. Pouliot, J. Agric. FoChem. 51 (2003) 4370.

[21] I.M. Reddy, N.K.D. Kella, J.E. Kinsella, J. Agric. Food Chem.(1988) 737.

[22] F.C. Church, H.E. Swaisgood, D.H. Porter, G.L. Catignani, J. DSci. 66 (1983) 1219.

[23] W.L. Nelson, E.I. Ciaccio, G.P. Hess, Anal. Biochem. 2 (1961) 39[24] H. Rinderknecht, M.C. Geokas, P. Silverman, B.J. Haverback, C

Chim. Acta. 2 (1968) 197.[25] J.F. Steffe, Rheological Methods in Food Process Engineering, F

man, East Lansing, MJ, 1992.[26] G.M. Campbell, E. Mougeot, Trends Food Sci. Technol. 10 (19

283.[27] R. Myrvold, F.K. Hansen, J. Colloid Interface Sci. 207 (1998) 97.[28] P.J. Althouse, P. Dinakar, A. Kilara, J. Food Sci. 60 (1995) 1110.[29] D. Doucet, S.F. Gauthier, D.E. Otter, E.A. Foegeding, J. Agric. F

Chem. 51 (2003) 6036.[30] P.P. Kamoun, Trends Biochem. Sci. 13 (1988) 424.[31] O.P. Ward, in: W.M. Fogarty (Ed.), Microbial Enzymes and Biote

nology, Elsevier Applied Science, London, 1983, p. 251.

[32] D. Doucet, D.E. Otter, S.F. Gauthier, E.A. Foegeding, J. Agric. Food

Chem. 51 (2003) 6300.

Interface Science 288 (2005) 412–422

[33] P.F. Fox (Ed.), Developments in Dairy Chemistry, vol. 4, Elsevier Aplied Science, London, 1989, p. 1.

[34] W. Konigsberg, J. Goldstein, R.J. Hill, J. Biol. Chem. 238 (1963) 20[35] M. Hammershoj, A. Prins, K.B. Qvist, J. Sci. Food Agric. 79 (199

859.[36] S. Roth, B.S. Murray, E. Dickinson, J. Agric. Food Chem. 48 (20

1491.[37] A.P.J. Middelberg, C.J. Radke, H.W. Blanch, Proc. Natl. Acad.

U.S.A. 97 (2000) 5054.[38] H.M. Zhu, S. Damodaran, J. Agric. Food Chem. 42 (1994) 856.[39] L.G.C. Pereira, O. Theodoly, H.W. Blanch, C.J. Radke, Langmui

(2003) 2349.[40] E.M. Freer, K.S. Yim, G.G. Fuller, C.J. Radke, J. Phys. Chem. B

(2004) 3835.[41] M.J. Ridout, A.R. Mackie, P.J. Wilde, J. Agric. Food Chem. 52 (20

3930.[42] J. Benjamins, E.H. Lucassen-Reynders, in: D. Mobius, R. M

(Eds.), Proteins at Liquid Interfaces, Elsevier, Amsterdam, 19p. 341.

[43] R. Wustneck, J. Kragel, R. Miller, V.B. Fainerman, P.J. Wilde, DSarker, D.C. Clark, Food Hydrocolloids 10 (1996) 395.

[44] D.B. Jones, A.P.J. Middelberg, Langmuir 18 (2002) 10357.[45] B.S. Murray, Colloid Surf. A Physicochem. Eng. Asp. 125 (1997)[46] F. MacRitchie, in: D. Mobius, R. Miller (Eds.), Proteins at Liquid I

terfaces, Elsevier, Amsterdam, 1998, p. 149.[47] D.A. Edwards, D.T. Wasan, in: R.K. Prud’homme, S.A. Khan (Ed

Foams—Theory, Measurements, and Applications, Dekker, NYork, 1996, p. 189.

[48] T.D. Dimitrova, F. Leal-Calderon, Adv. Colloid Interface Sci. 108–(2004) 49.

[49] H.M. Princen, A.D. Kiss, J. Colloid Interface Sci. 128 (1989) 176.[50] T.G. Mason, Curr. Opin. Colloid Interface Sci. 4 (1999) 231.