Controlling Agglomeration of Protein Aggregates for Structure Formation in Liquid Oil: A Sticky Business.

Proteins are known to be effective building blocks when it comes to structure formation in aqueous environments. Recently, we have shown that submicron colloidal protein particles can also be used to provide structure to liquid oil and form so-called oleogels ( de Vries , A. J. Colloid Interface Sci. 2017 , 486 , 75 - 83 ) . To prevent particle agglomeration, a solvent exchange procedure was used to transfer the aggregates from water to the oil phase. The aim of the current paper was to elucidate on the enhanced stability against agglomeration of heat-set whey protein isolate (WPI) aggregates to develop an alternative for the solvent exchange procedure. Protein aggregates were transferred from water to several solvents differing in polarity to investigate the effect on agglomeration and changes in protein composition. We show that after drying protein aggregates by evaporation from solvents with a low polarity (e.g., hexane), the protein powder shows good dispersibility in liquid oil compared to powders dried from solvents with a high polarity. This difference in dispersibility could not be related to changes in protein composition or conformation but was instead related to the reduction of attractive capillary forces between the protein aggregates during drying. Following another route, agglomeration was also prevented by applying high freezing rates prior to freeze-drying. The rheological properties of the oleogels prepared with such freeze-dried protein aggregates were shown to be similar to that of oleogels prepared using a solvent exchange procedure. This Research Article provides valuable insights in how to tune the drying process to control protein agglomeration to allow for subsequent structure formation of proteins in liquid oil.

Introduction

It has long been recognized that a diet rich in saturated and trans fats is associated with an increase in the amount of LDL cholesterol (low density lipoprotein) at an expense of HDL cholesterol (high density lipoprotein),[2] which is related to a higher risk of developing coronary artery disease. On the other hand, a diet rich in cis-unsaturated fatty acids decreases these risks.[3,4] However, food reformulation is not straightforward as the use of saturated and trans fats has technological benefits such as providing texture and oxidative stability to food products. One interesting alternative, which has gained much attention over the recent years, is the use of so-called “oleogels”.[5−8] The purpose of designing edible oleogels is to be able to provide a solid-like structure to liquid oil at room temperature other than by the conventional use of saturated and trans fatty acids.

The most common gelling agents of organic liquids, called in a general term “organogelators”, are found within the class of low molecular weight organogelators (LMOG). Their network formation relies on the complex self-assembly of the components by noncovalent bonds such as dipolar interactions, π-stacking, and intermolecular hydrogen bonds into crystals, tubules or fibrillar structures responsible for the solid-like behavior.[9−11] These organogels have been studied for different applications, such as oil spills,[12,13] electronics,[14] and drug delivery.[15−17] For food purposes, many different edible oil soluble LMOG’s have been studied for their gelling properties, such as lecithins,[18,19] monoacylglycerides,[20,21] fatty acids, alcohols,[22] sterols,[23,24] or waxes.[25−27]

In contrast to the large diversity of low molecular weight compounds, only a limited amount of biopolymers have been studied to structure oil. Because of their predominant hydrophilic nature, these compounds usually do not dissolve in oil. A well-studied exception is the cellulose derivative ethyl cellulose (EC).[28] EC dissolves in liquid oil at high temperature and upon cooling, the polymer chains interact to form solid structures. The resulting mechanical properties of the gels, depending on polymer–polymer interactions, can be tailored by changing the molecular weight of the polymer chain or by adding surfactants.[28,29] Another example of a biopolymer studied for its oil structuring ability is the polysaccharide chitin. Modifying crude chitin into “nanocrystals”[30] or hydrophobic “whiskers”[31] was necessary to efficiently provide a structure to liquid oil. To overcome the problem of low dispersibility of biopolymers, other researchers have adopted a foam- or emulsion-template approach. In a first step, hydrophilic polymers like hydroxyl propyl methylcellulose, gelatin and xanthan are used to stabilize an oil–water or air–water interface. Subsequently, the water is removed and by shearing the dried product into the liquid oil, gelled structures are obtained.[32,33] However, there is limited control over the network formation as the particle size of the building blocks might be difficult to tune, and shear forces may disrupt the network. A better control over the network formation of biopolymers is desirable to tune specific rheological characteristics of such gelled oils.

Recently, we reported a new method to create oleogels using whey proteins.[1,34] Whey proteins are extensively used as ingredients in foods due to their high nutritional value and functional properties and are therefore an ideal candidate as a structuring agent for liquid oil. First, we developed a method in which protein oleogels were prepared in a more direct way.[34] To create oleogels, we prepared heat-set whey protein hydrogels, and a solvent exchange procedure was applied to include the oil in the interstitial areas of the protein network. In this process, the water is first replaced by an oil-miscible solvent, acetone, which is then replaced by liquid oil. To enhance the control over the network formation, we then prepared whey protein aggregates of colloidal size (∼150 nm) as initial building blocks.[1] In this Research Article, we showed that the solvent exchange procedure was able to prevent agglomeration of preformed heat-set aggregates and therefore enabled gel formation due to sufficient protein–protein interactions. Alternatively, when the protein aggregates were freeze-dried to remove the water before they were dispersed in oil, irreversible agglomeration of the initial particles was obtained. This agglomeration led to increased particle size in oil, which was related to an observed poor stability against sedimentation and poor gel formation. However, it is currently not precisely known how the solvent exchange procedure prevents such agglomeration.

The aim of the current paper is to elucidate on the enhanced stability against agglomeration of whey protein aggregates present in oil after applying a solvent exchange procedure. To this end, we formulate two possible mechanisms: (1) Prevention of stresses, such as capillary pressure, resulting from drying processes. By keeping the particles “solvated” by the different solvents, starting with water, followed by acetone and finally liquid oil, the wet conditions may avoid strong forces arising from the formation of a solvent–air interface. (2) Conformational changes of the proteins as a function of the properties of the surrounding solvent. Solvents with electronegative atoms, such as oxygen, are able to interfere with intramolecular hydrogen bonding,[35,36] which could lead to the exposure of a larger fraction of hydrophobic groups to the solvent. Potentially, this could result in more favorable protein–solvent interactions and enhanced stability against agglomeration.

To gain insight in the effect of the different solvents on possible agglomeration effects and conformational changes, we have used several solvents of different polarity in the solvent exchange procedure. In addition to acetone, we have included propanol as a solvent with the ability to form hydrogen bonds, and as an alternative for a hydrophobic liquid oil, we have used volatile apolar solvents such as hexane, decane and heptane. To analyze the protein composition and conformation, the protein material was dried from the solvent by evaporation, and agglomeration effects were tested by examining the dispersibility of the dried material in oil. The results were compared to aggregates freeze-dried from water. Understanding the mechanisms involved to prevent irreversible agglomeration could provide insights on how to design a route to obtain a dry protein material that is dispersible in liquid oil and allows for direct network formation without the need for a solvent exchange. This could be beneficial from a product development point of view, since the large amount of solvent needed for a solvent exchange procedure limits practical applicability.

Materials and Methods

Materials

Whey protein isolate (WPI, BiPro) was obtained from Davisco Foods International (Le Sueur, MN, USA). The protein concentration was 93.2% (N × 6.38) and was used as received. Acetone and n-hexane were supplied by Actu-All Chemicals (Oss, The Netherlands). Hydrogen chloride, sodium hydroxide, sodium dodecyl sulfate (SDS), and n-decane were purchased from Sigma-Aldrich (Steinheim, Germany). 1-Propanol and n-heptane were purchased form Merck (Damstadt, Germany). N,N-Dimethyl-6-propionyl-2-naphthylamine (PRODAN) was obtained from Sigma-Aldrich. Refined sunflower oil (Vandermoortele NV, Breda, The Netherlands) was bought at a local supermarket and was used without further purification. All chemicals used were of analytical grade. Demineralized water was used throughout the experiments.

Methods

Preparation of WPI Aggregates

To prepare a protein stock solution, WPI powder (4% w/w) was dissolved in demineralized water under continuous stirring at room temperature for 2 h. Afterward, the stock solution was stored overnight at 4 °C to ensure complete protein hydration. The next day, the pH of the stock solution was adjusted to 5.7 using a 1 M HCl solution. The resulting solution was heated in 50 mL plastic tubes with screwcaps at 85 °C for 15 min using a temperature controlled water bath to induce protein denaturation. After cooling in ice water, a weak protein gel was obtained. This weak gel was easily broken into the smaller aggregates by hand shaking and vortexing. The resulting WPI aggregate dispersion was homogenized using a rotor stator homogenizer (Ultra Turrax, T25, IKA Werke, Germany) at 13.000 rpm for 3 min. The protein aggregates were then collected as a pellet by centrifugation at 3904 g (Hermle Z383 K, Hermle Labortechnik GmbH, Wehingen, Germany) for 20 min at 20 °C. After collection, the pellet was redispersed and centrifuged twice with demineralized water to remove remaining soluble protein material. Afterward, the sample was homogenized using a laboratory scale homogenizer (Labhoscope, Delta Instruments, Drachten, The Netherlands) at 200 bar (3 passes). The final pH of the WPI aggregate suspension was 8.0.

Preparation of the Protein Oleogels Using a Solvent Exchange

To prepare the protein oleogels, the WPI aggregates were transferred to the oil phase using a solvent exchange procedure, which was based on a method described in detail previously.[1] In this procedure, the polarity of the solvent was changed gradually to remove the surrounding water from the WPI aggregates and replace the continuous phase for oil. In short, 15 g of aqueous pellet, containing the WPI aggregates, was redispersed in 150 mL of acetone, and mixed thoroughly using rotor stator homogenization. Afterward, the sample was centrifuged at 3904g for 20 min at 20 °C. Excess acetone was removed by decanting and the pellet, containing the protein aggregates, was collected. The procedure of redispersing and centrifugation was repeated once more using acetone to ensure water removal. The pellet was then redispersed twice in liquid oil. The obtained pellet of WPI aggregates in oil was diluted in a ratio of 1:10 with sunflower oil and left overnight under continuous stirring to allow for evaporation of the remaining acetone. The next day, the suspension was centrifuged at 4000g for 20 min at 20 °C to increase the concentration of the protein aggregates and induce gel formation.

Preparation of Dried WPI Aggregates

Evaporation

To determine various properties of the WPI aggregates during the solvent exchange procedure, we dry the aggregates by evaporation from acetone, 1-propanol, hexane, heptane, and decane. To produce protein aggregate suspensions in the different solvents, the water was exchanged for acetone or 1-propanol, as described above. The suspensions in hexane, heptane and decane were prepared using acetone as an intermediate solvent during the solvent exchange. To easily collect the protein material from its solvent, the protein suspensions were centrifuged at 3900g for 20 min and the resulting pellet was placed in an aluminum tray (ø = 5 cm) and dried in a fume hood for 16 h at room temperature. After drying, the powder was collected and grinded using a pestle and mortar until no further reduction in particle size was observed. To investigate the effect of the drying conditions, instead of drying from a concentrated pellet, a 1 wt % protein aggregate suspension was dried from water, acetone or hexane by evaporation. The solvent was evaporated in the same way as described above for the drying method using the concentrated pellets.

Freeze-Drying

After the aqueous suspension was washed twice with demineralized water, the resulting suspension was homogenized using a lab scale homogenizer (Labhoscope) at 200 bar (3 passes) followed by centrifugation. The pellet was frozen at −20 °C in a freezer for 16 h. Thereafter, the sample was freeze-dried (Christ alpha 2–4 LD plus, Martin Christ Gefriertrocknungsanlagen GmbH, Osterode am Harz, Germany) for 48 h to remove all water. In another experiment, the liquid suspension of WPI aggregates (1 wt %) was added dropwise into liquid nitrogen (−195 °C) to rapidly freeze the material. Thereafter, the frozen sample was freeze-dried as described above.

Composition

Protein Content

The nitrogen content was determined using Dumas (Dumas Flash EA 1112 Series, N Analyzer, Thermo Scientific). After they were weighed, the samples were dried overnight in an oven at 60 °C before analysis. To calculate the protein content, a nitrogen conversion factor of 6.38 was used.

Water Content

The water content in the oleogel was determined by dry matter determination. Aluminum cups (ø = 5 cm) were first heated to 105 °C in an oven (Venticell, BMT Medical Technology, Brno, Czech Republic) to remove any water contamination. Afterward, approximately 1 g of oleogel sample was added to the cup, and its weight was recorded before and after drying for 4 h at 105 °C. Water content in the dried powders was determined by Karl Fisher titration. Measurements were performed in duplicate.

Chemical Stability

To assess the internal bonds involved in the stabilizing mechanism of the WPI aggregates, several denaturants were added to a 1 wt % of WPI aggregate dispersion. The denaturants used were 10 M urea to examine disruption of hydrogen bonds, 140 mM sodium dodecyl sulfate (SDS) for hydrophobic interactions, and 50 mM dithiothreitol (DTT) for disulfide interactions.[37] Several combinations of these denaturing agents were tested, and whenever DTT was used, heat treatment was applied at 70 °C for 15 min.

Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE)

The protein composition of the freeze-dried aggregates, the acetone-dried aggregates, the hexane-dried aggregates, as well as the protein composition in the supernatant after centrifugation was analyzed under reducing conditions by SDS-PAGE using the Novex NuPAGE gel system (Invitrogen, Thermo Fischer Scientific). Samples were prepared by the addition of NuPAGE LDS sample buffer (4×) and NuPAGE Reducing agent (10×) to a final protein concentration of 2 mg/mL. Thereafter, samples were vortexed and heated at 75 °C for 10 min in a water bath. After cooling, samples were loaded into the wells of a NuPage 4–12% Bis-Tris gel. As a running buffer, NuPAGE MES SDS (20×) was used, with antioxidant in the cathode chamber. Electrophoresis was performed by applying a constant voltage of 200 V for 40 min. Afterward, gels were stained using coomassie blue (SimplyBlue). The apparent molecular weight of the proteins present in each sample was determined by comparing the position of the bands to a reference sample with proteins of various molecular weights (Mark12 unstained standard, Invitrogen). The gels were scanned in a densitometer (Gelscanner GS-900, Bio-Rad, Hercules, CA, USA) with Image Lab software, which allows for the identification of the proteins present in each sample.

Surface Hydrophobicity

The surface hydrophobicity of the aggregates obtained through the solvent exchange method and freeze-drying was measured against the native WPI by means of the fluorescent probe method, with N,N-dimethyl-6-propionyl-2-naphthylamine (PRODAN) as the binding probe. The procedure to determine the surface hydrophobicity was similar as reported by Haskard et al.[38] Five different concentrations of protein were prepared ranging from 0.04 to 0.2 mg/mL and analyzed in duplicate. PRODAN was dissolved in acetone at a concentration of 0.0041 M and stored in a freezer (−20 °C) protected from light and evaporation. Ten microliters of PRODAN solution was added to each 4 mL sample and vortexed well. The relative fluorescence intensity (RFI) was measured using a fluorimeter (PerkinElmer luminescence spectrometer LS50B) after 15 min of reaction of the PRODAN with the proteins in the dark at room temperature, using disposable acrylic cuvettes (Sarstedt, Nümbrecht-Rommelsdorf, Germany). The measurement settings were set to an excitation wavelength of 365 nm, emission and excitation slit widths of 5 nm, emission scan from 300 to 600 nm, and a scan speed of 200 nm/min. The net RFI values of each sample (protein with PRODAN) was obtained by subtraction of the protein blank sample from the measured RFI value. The slope of the net RFI values taken from the maximum emission spectra of the bounded protein-PRODAN were plotted versus the protein concentration. The resulting slope of this linearization is used as a measure for the protein surface hydrophobicity.

Attenuated Total Reflectance Fourier Transform Infrared Resonance (ATR-FTIR)

To directly analyze protein conformation of the dried protein aggregates, samples were analyzed using infrared spectroscopy. Dried protein material was placed directly on the crystal using an ATR-FTIR spectrometer (Platinum Tensor, Bruker Optics, Coventry, UK). The IR spectrum was recorded from 4000 to 600 cm–1 and for each sample, 64 scans with a resolution of 4 cm–1 were averaged. After averaging the scans, the spectrum was cut to obtain the amide I and II region at 1400–1750 cm–1, baseline corrected, and vector normalized using the OPUS software to analyze the amide I, II, and III region. To assess the protein conformation, the second derivative of the amide I region (1600–1700 cm–1) was taken and smoothened using the OPUS software. All samples were measured in duplicate.

Particle Size Analysis

The particle-size distribution of WPI aggregates was determined by static light scattering (Mastersizer 2000, Malvern Instruments, Worcestershire, UK) with either sunflower oil or demineralized water as the continuous phase. The refractive index of water was set to 1.33 and for sunflower oil to 1.469. The refractive index of the protein aggregates was set to 1.45 for aqueous protein samples and 1.54 for protein samples in sunflower oil to correct for the change in refractive index upon dehydration. The particle size distribution was determined as an average of three measurements.

Scanning Electron Microscopy (SEM)

A scanning electron microscope (Phenom G2 Pro, Phenom-World BV, Eindhoven, The Netherlands) was used to visualize the structure of the different protein powders and used to analyze any structural differences between the different samples. To this end, a small sample was taken and fixated using carbon tabs on aluminum stubs (SPI Supplies/Structure Probe Inc., West Chester, USA). Conveniently, because of the low voltage used (5 kV), sample pretreatment was not necessary, and the appearance of the powders could be visualized directly.

Rheology

Oscillatory rheology was performed on the Oleogel made with WPA aggregates obtained via either the solvent exchange procedure or via freeze-drying. Both samples were standardized to 10 wt % protein and 1.2 wt % water to allow for comparison between the samples. Before the measurement, the samples were homogenized using rotor-stator homogenization (13.5000 × rpm) for 180 s. Afterward, samples were degassed using a vacuum pump for 30 min and loaded into a stress-controlled rheometer (AP 502, Anton Paar GmbH, Graz, Austria) between sandblasted parallel plates (ø = 49.978 mm) to prevent slip phenomena. The temperature was controlled for all measurements at 20 °C. Before any measurements were performed, the samples were allowed to equilibrate for 60 min at a frequency of 1 Hz and a strain (γ) of 0.01% (which was within the linear viscoelastic region). Frequency sweeps were performed by increasing the frequency logarithmically from 0.01 to 50 Hz at γ = 0.01%. Amplitude sweeps were performed by increasing the strain logarithmically from 0.001 to 100% at 1 Hz. All measurements were performed in triplicate.

Results and Discussion

Characteristics of Whey Protein Aggregates

To understand possible changes in the protein aggregates during the solvent exchange procedure, we first examined the characteristics of the protein aggregates prepared in aqueous conditions. Figure A shows the particle size distribution of the whey protein isolate (WPI) aggregates after heat treatment. The major fraction of the WPI aggregates had a particle size around 150 nm, which is of comparable size to what has been reported in other studies.[39] Upon heating a protein solution above the protein denaturation temperature, heat-set aggregates are formed that are stabilized through physical and chemical interactions. The strength of these interactions determine the stability of the protein aggregates against external forces such as those resulting from shear and drying processes. To assess which interactions are involved that lead to the stability of the protein aggregates, several denaturants were added to an aqueous 1% WPI aggregate suspension. After adding SDS, urea, or a combination of both, we noticed that the suspensions remained turbid, suggesting that hydrogen bonds and hydrophobic interactions were not the only interactions responsible for the stabilization. However, when DTT was added, the suspension turned completely transparent as a result of disintegration of the aggregates. This suggests that the structure of the WPI aggregates is partially stabilized by internal covalent disulfide bonds and is in agreement with other studies using heat-set WPI aggregates prepared at similar conditions.[40]

Figure 1

Particle size distribution of dispersions of WPI aggregates. (A) Particle size in water measured either directly after homogenization (○) or redispersed after freeze-drying (□). (B) Particle size of aggregates in sunflower oil using freeze-drying (□) or a solvent exchange procedure (○).

After freeze-drying the WPI aggregates, the resulting powder was redispersed in demineralized water. The resulting particle size distribution was comparable before freeze-drying, as is presented in Figure A. As already discussed in our previous research,[1] the freeze-dried aggregates do not readily disperse in oil and particles of more than 100 μm were obtained, as is depicted in Figure B. Alternatively, when a solvent exchange procedure was used to disperse the aggregates in the oil, a particle size of 150 nm was found (open circles in Figure B). Although these results show that the solvent exchange procedure prevents agglomeration of the protein aggregates, the underlying mechanism is not yet understood. To gain more insight in the effect of the solvents during the solvent exchange process, hexane was used as an alternative nonpolar solvent in the following sections to easily isolate the aggregates during the solvent exchange process by evaporation. Hereafter, we studied the properties of the dried aggregates in terms of composition, conformation, and dispersibility in liquid oil.

Protein Composition

Since WPI is a mixture of several proteins, the composition of the proteins present in the aggregates might be different as a result of exposure to different solvents. To exclude these differences, the freeze-dried, acetone-dried and hexane-dried WPA aggregates samples were analyzed using SDS-PAGE. After the heating step in aqueous conditions, only 80% of the protein material was found to be included in the WPI aggregates (i.e., in the pellet), and 20% was still present in the supernatant. For this reason, also the supernatant was analyzed for its composition. Native WPI was analyzed for comparison. The SDS-PAGE electrophoretograms are shown in Figure . Lane 1 shows the electrophoretogram of native WPI and the major whey protein fractions can easily be recognized. Major bands found around 66, 18, and 14 kDa correspond to bovine serum albumin (BSA), β-lactoglobulin (β-lac), and α-lactalbumin (α-lac), respectively. The major protein fraction, as indicated by the higher band intensity, is β-lac. Comparable to the native sample, the freeze-dried WPI aggregates (lane 2) showed bands at all major protein fractions. However, the intensity of the band corresponding to α-lac (14 kDa) seems to be lower, as will be discussed in more detail below. The acetone- and hexane-dried WPI aggregates (lane 3 and 4) showed a high similarity with the freeze-dried sample. The protein composition of the supernatant, that is, soluble protein material after heat treatment and centrifugation, showed a distinctly different electrophoretogram (lane 5). The band corresponding to BSA is not detected, the band intensity of β-lac is less intensive and the band intensity of α-lac has increased compared to the samples containing the aggregates.

Figure 2

Scans of SDS-PAGE electrophoretograms of native WPI isolate (1), freeze-dried WPI aggregates (2), acetone-dried WPI aggregates (3), hexane-dried aggregates (4), and supernatant after centrifuging the WPI aggregates (5). M: molecular weight markers.

To estimate the relative content of α-lac and β-lac in each sample, we have taken the band intensity of β-lac as an internal standard per electrophoretogram, and determined the ratio α-lac/β-lac for the different samples (Figure ). The α-lac/β-lac ratio of the aggregates was found to be lower compared to the native WPI, whereas for the supernatant, this ratio was much higher. Our result suggest that the disulfide cross-linked aggregates contain a higher amount of BSA and β-lac compared to its native protein composition, and that α-lac is to a lesser extent incorporated into the aggregates. Taking the gelation mechanism of the different proteins into account, the results can be explained since both BSA and β-lac have a higher gelation rate and have a free thiol group available to from disulfide bonds.[41,42]

Figure 3

Ratio α-lactalbumin/β-lactoglobulin in the different protein samples. FD, AD, and HD represent freeze-dried, acetone-dried, and hexane-dried aggregates, respectively. Supernatant was taken after centrifuging the WPI aggregates.

From these results, we can conclude that changing the aqueous solvent to acetone or hexane had no apparent influence on the internal protein composition of the aggregates. This indicates that the protein aggregates were stable during the solvent exchange procedure, regardless of the drying method or solvents used. Therefore, changes in oil dipsersibility as a result of the solvent exchange, is not caused by a change in protein composition of the aggregates.

Particle Morphology

The obtained powders, dried by either freeze-drying from water or evaporation from organic solvents, were analyzed by scanning electron microscopy (SEM) to determine the particle morphology. As can be seen in Figure , the freeze-dried powder (Figure A1 and A2) contained large particle agglomerates (∼50–100 μm), but appear to have an open, porous structure. The morphology of the powder when evaporated from acetone (Figure B1 and B2) had a similar appearance as the freeze-dried sample, that is, large particle agglomerates with similar porosity. In contrast, the hexane-dried protein powder is absent of any large agglomerates (Figure C1 and C2), and the powder consists of smaller, porous agglomerates (10–20 μm). It shows that the nature of the solvent from which the aggregates are dried had a large influence on the agglomeration and powder morphology. The packing density of the powders was estimated by weighing 1 mL of dried material in a graded cylinder. The packing densities were found to be approximately 0.26 g/cm3 for the freeze-dried powder, 0.14 g/cm3 for the acetone-evaporated powder, and 0.07 g/cm3 for the hexane-evaporated powder. Clearly, when WPI aggregates are dried from solvents with low polarity, the powder morphology changed from large agglomerates to agglomerates of smaller size consisting of more loosely packed particles.

Figure 4

SEM micrographs of dried WPI aggregates: (A) freeze-dried from water, (B) evaporated from acetone, and (C) evaporated from hexane. Numbers 1 and 2 refer to a different magnification of the same sample.

The effects of different drying conditions are dependent on the hydrophobicity of the solvent, which seem to change the interactions between the protein aggregates. By removal of the solvent by evaporation, initially the solvent evaporates from the bulk but will eventually form a liquid bridge between two particles, as schematically depicted in Figure . This causes a capillary force (Fc) across two particles as a result of a curved interface governed by the interfacial tension, γ. When the distance between two particles approaches zero, Fc becomes[43]where R is the particle radius and θ is the contact angle between the liquid and the solid, which in our case is the protein aggregate (Figure ). When the capillary force is sufficiently high, this may result in large attractive forces between the particles. Therefore, in the production of nanometer-sized ZrO2, Al2O3, or TiO2 particles, solvents with a low surface tension were used during drying to reduce their agglomeration and increase their specific surface area.[44,45] Similarly, in the case of aerogels, it has been shown that large capillary forces leads to a collapse of the initial microstructure. In order to prevent this undesired effect, often alternative solvents or supercritical drying methods are used to change the wetting angle and interfacial tension.[46,47] In our case, the solvent evaporates from the interstitial spaces between and from the surface of the aggregates, which may lead to increased capillary forces between the aggregates. When the solvent has a high surface tension and a small contact angle, capillary pressure facilitates particle agglomeration. The contact angle is related to the difference in polarity between the particle and the solvent.[48] In the case of water, with a high surface tension of 73 mN/m and an estimated low contact angle (θ < 90°), the resulting high capillary force would lead to a large degree of agglomeration. However, in the case of hexane, the lower interfacial tension (19 mN/m) and a higher contact angle, given the low polarity of hexane, leads to much lower values of the capillary force and therefore a low degree of agglomeration, in accordance with our experiments. In the case when θ > 90°, the resulting force could even lead to and effective repulsion between two protein aggregates.

Figure 5

Schematic drawing of a liquid bridge between two spherical particles during evaporation of the solvent.

To visualize the effect of the drying conditions more clearly, we dried a liquid suspension of WPI aggregates from the solvents water, acetone, or hexane and observed the appearance and microstructure of the film formed. Figure shows the difference in the appearance of the dried protein material as well as the microstructure of the films at a smaller length scale. When the drying medium was water, the resulting film was hard and almost translucent. The SEM micrograph shows a tight packing of the protein aggregates. Drying from acetone resulted in a more brittle film, which was more difficult to handle than the water-dried protein film. The microstructure shows more cracks or gaps between the protein aggregates. When the protein aggregate suspension was dried from hexane, film formation was inhibited and the sample was very brittle, resulting from limited interactions between the protein aggregates. In addition, the resulting material was opaque (Figure C). The limited interactions can be seen more clearly in Figure F, where a very porous structure can be observed with larger distances between the individual aggregates. This is consistent with the observed powder morphology in Figure , where drying from hexane prevents agglomeration. Prevention of strong capillary forces during drying thus seems to reduce agglomeration between the particles.

Figure 6

Appearance of dried protein suspensions from water (A), acetone (B), or hexane (C) at ambient conditions and corresponding SEM micrographs (D–F).

Dispersibility of Dried Whey Protein Aggregates in Liquid Oil

The dried protein aggregates obtained from different drying methods were tested for their dispersibility in liquid oil. To this end, a 1% w/w dispersion was prepared by adding the dried powders to sunflower oil, and the resulting particle size was measured after homogenization. As can be seen in Figure A, the acetone-dried sample showed poor dispersibility. A large increase in particle sizes (10–300 μm) compared to the sizes observed in the original aqueous dispersion is indicative of irreversible particle agglomeration during drying. Even though during drying the surface tension and wettability of acetone is lower than that of water, apparently the capillary forces are not yet decreased to such extent that agglomeration is prevented. In contrast, good dispersibility was obtained when the drying medium was hexane (Figure B), since a large amount of small (<1 μm) particles was obtained. Although agglomeration was not fully prevented as shown by the presence of a smaller peak at 10–100 μm, the major fraction of the protein aggregates retained their small initial size, shown by the large peak at ∼150 nm. Since the difference in particle agglomeration was observed by SEM, it seems that drying-induced agglomeration was largely irreversible. Mixing the agglomerated powders into the oil by shear hardly seems to break any formed agglomerates and leads to poor dispersibility in oil.

Figure 7

Particle size distribution of dispersed WPI aggregates in liquid oil. (A) size distribution of aggregates dried from 1-propanol (○) and acetone (△). (B) Particle size distribution of aggregates dried from hexane (△), heptane (□), or decane (○).

To test which physical properties of the solvent have an effect on particle agglomeration and resulting dispersibility in oil, 1-propanol, heptane, and decane were also used as the suspending solvents during drying. These solvents differ in surface tension and dielectric constant as can be seen in Table . Here, we use the dielectric permittivity as a measure for the polarity. As can be seen in Figure , drying from 1-propanol led to a similar size distribution as acetone, whereas decane and heptane led to a similar particle size distribution as hexane. Interestingly, although the surface tension of decane and 1-propanol is similar, the resulting dried aggregates showed a very different size distribution when dispersed in oil. A low surface tension does not seem to be the most important prerequisite to prevent irreversible agglomeration. Particle agglomeration seems to be better related to the dielectric permittivity, that is, the polarity of the solvent. The low polarity causes a low wettability with the mainly hydrophilic proteins and subsequently a large contact angle. In turn, this results in a lower capillary pressure across two protein aggregates during solvent evaporation, which leads to less irreversible agglomeration and better dispersibility of the aggregates into oil.

Table 1

Properties of the Solvents Used for Drying WPI Aggregates

	dielectric permittivity, ε (298 K)	surface tension mN/m (293 K)
water	80.1	72.8
acetone	21.4	25.2
1-propanol	19.4	23.7
n-hexane	1.8	18.4
n-heptane	1.8	20.1
n-decane	1.8	23.8

Protein Conformation

The previous results show that prevention of agglomeration by choosing a solvent with a low polarity is an important factor for the increased dispersibility of protein aggregates. However, changes within the proteins aggregates may also add to this effect. Since proteins are subjective to structural reorientation as solvent conditions are changed, it is possible that the protein conformation is altered upon contact with solvents like acetone or hexane. Possibly, this contributes to increased protein–solvent interactions by structural reorientations. To determine if the protein conformation differed among WPI aggregates as a result of the type of solvent it was dried from, we analyzed the different powders by attenuated total reflectance Fourier transform infrared resonance (ATR-FTIR). This technique is capable to determine the protein conformation (i.e., secondary structure) as well as the hydration level.[49] The amide I region (1600–1680 cm–1) is mostly due to C=O stretching and is closely related to changes in protein conformation. The amide II region (1480–1560 cm–1), caused by NH bending and CN stretching is closely related to protein hydration and less sensitive to conformational changes.[50] Conveniently, we can probe the conformation as well as the hydration level of the obtained dried samples directly. Figure A shows the amide I and II region of different WPI aggregate samples as well as the native WPI sample. We display the results for aggregates dried from hexane (HD) and acetone (AD), as these solvents gave a large difference in morphology and particle size, and compared these samples to aggregates dried via freeze-drying (FD). Looking at the amide I region, no clear changes can be observed. Changes in the shape of the amide II region are more extensive between the samples than changes in the amide I region. This may indicate a change in hydration level of the proteins, which can be probed by comparing the intensity ratio of the peak at 1535 and 1541 cm–1 to the peak at 1515 cm–1. The vibration at 1515 cm–1 is related to tryptophan and is insensitive to hydration,[51] which makes this a valuable internal standard. These intensity ratios were used by other researchers to assess the hydration level during film drying of β-lactoglobulin films.[52]Figure B shows that the I1535/I1515 and I1541/I1515 ratios are higher for all dried aggregates compared to the native WPI powder. Aggregates obtained from freeze-drying showed lower hydration levels than the aggregates dried from both acetone and hexane, but no differences were seen between acetone or hexane dried aggregates. The lower hydration levels were confirmed by Karl Fisher titration, where the measured water content was 6.3% (±0.1) for native WPI, 7.0 (±0.1) for the freeze-dried aggregates, 7.9 (±0.1) for acetone-dried aggregates and 7.9 (±0.3) for hexane-dried aggregates. Most likely, the amount of water present in the solvent-dried samples (either acetone or hexane) was slightly higher because of the hygroscopic nature of the protein powder, attracting moisture from the air. This effect can be enhanced for powders with low density and large contact area. However, the increased dispersibility and the small particle size of the hexane-dried aggregates compared to acetone-dried protein aggregates in oil is not related to these differences in water content.

Figure 8

FTIR results for native WPI powder, freeze-dried aggregates (FD), acetone-dried aggregates (AD), and hexane-dried aggregates (HD). (A) Amide I and II IR-spectra and (B) band intensity ratio of selected band/1515 cm–1 (I/I1515). Gray bars: 1535 cm–1. Black bars: 1541 cm–1. (C–E) Second derivative of amide I band of dried whey protein aggregates (solid lines). In each figure, native WPI powder was added as a reference (dotted line).

To determine the protein conformation, the second derivative of the amide I region was obtained for native WPI and the WPI aggregates and displayed in Figure C–E. Since the solvent exchange did not affect protein composition of the sample, this allows for direct comparison between the aggregates. In all graphs, the second derivative of native WPI was added as a reference (dotted line). Compared to the signal from the native WPI, the aggregates did not show major differences. Only a broadening of the major band is noticeable at 1630 cm–1, assigned to an increase of intermolecular β-sheet formation, and is indicative of aggregation.[53,54] Between the WPI aggregates, however, regardless of the drying method, no obvious differences were seen. The only noticeable difference is the slight increase in the intensity around 1640 cm–1 for the samples dried from acetone and hexane compared to the freeze-dried sample, which might be related to the small difference in hydration. Although small changes are observed between the aggregates dried from the different solvents, we suspect that these differences are not large enough to account for significant changes in the structure of the proteins. Note that also the spectra of the second derivative of WPI aggregates dried from 1-propanol, heptane, and decane did not differ from those shown in Figure . Moreover, we observed no changes in the second derivative of the amide I region when WPI aggregates were suspended in oil using a solvent exchange procedure (data not shown). The increased dispersibility of the aggregates as a result of the solvent exchange procedure thus seems to be unrelated to any changes in protein conformation since the results from the FTIR measurements for these samples show a high level of similarity.

Surface Hydrophobicity

Though FTIR was unable to detect differences in conformation, we checked the surface hydrophobicity of the freeze-dried and the hexane-dried WPI aggregates. Using PRODAN, a fluorescent probe, the relative fluorescence intensity (RFI) was measured as a function of protein concentration and the results are displayed in Figure . The slope of RFI versus protein concentration, is used as a measure of hydrophobicity. We have added the results of native WPI as a comparison. As can be seen, the affinity for PRODAN increased as a result of applying a heat treatment, as a higher slope was measured for the aggregates than for the native proteins. This increase in hydrophobicity is expected as the heating process leads to exposure of hydrophobic groups normally buried within the native folded structure of proteins.[55] Between the two WPI aggregate samples, however, there was no clear difference between the freeze-dried and hexane-dried sample, as the slope shows a high similarity. From this, we conclude that the hydrophobicity does not change resulting from drying from different solvents, consistent with the results discussed before. However, these results have to be taken with care, since these measurements were performed in aqueous environments and therefore only irreversible changes as affected by the different drying method can be measured.

Figure 9

Relative fluorescence intensity (RFI) as a function of protein concentration for native WPI (○), freeze-dried WPA (□), and hexane-dried WPA (△).

In summary, from the results obtained by SDS-PAGE, FTIR, and hydrophobicity measurements, we can conclude that no differences in protein composition, conformation or hydrophobicity occurred as affected by the presence of different types of solvents. This shows that structural changes on a molecular level are therefore most likely not responsible for the enhanced dispersibility of WPI aggregates in oil. A slight difference in water content was found as result of the different drying methods, but this does not seem to explain the difference in dispersibility between the acetone- and hexane-dried aggregates. Instead, other factors than water content or conformational changes seem to dominate the dispersibility. We propose that prevention of capillary forces between the aggregates during drying is most likely the cause for the increased dispersibility of the aggregates in oil. It suggests that an optimized drying process to avoid particle agglomeration could be an alternative for the solvent exchange procedure. Since a solvent exchange requires a large amount of solvent, an alternative method would provide many advantages from a practical point of view. Therefore, we examined the process of freeze-drying more closely by considering different rates of freezing.

Effect of Freeze-Drying Conditions

During the process of freeze-drying, water is removed from the sample by sublimation of ice at low pressure. Although freeze-drying is considered a mild drying technique, freezing effects can play a significant role in the agglomeration of particles. Ice formation in a colloidal suspension, such as our suspension of aggregates, typically expels the particles from the frozen areas, effectively increasing the particle concentration locally. The formed ice crystals pack the particles close to one another with high forces that can overcome repulsive forces and thus induce (irreversible) particle agglomeration. When conditions are carefully chosen, in a process called “freeze casting”, directional ice formation can even occur, leading to the formation of a layered pattern.[56] When the freezing rate is increased, however, the formed ice crystals are small, and tight packing of the particles is prevented to some extent and instead, the particles remain more evenly distributed.[57] Therefore, as an alternative to slow freezing at −20 °C, we investigated the effect of a higher freezing rate by dripping a 1 wt % aqueous WPI aggregate suspension directly into liquid nitrogen. The temperature difference, and subsequently the freezing rate, was thus roughly increased by a factor 5. After the material was freeze-dried, the dried powder was dispersed directly into oil by homogenization. The resulting particle size distribution was measured and as can be seen in Figure , the sample contains two main size populations as is noticeable by the appearance of two distinct peaks. One population having a size of approximately 150 nm, the other a broad range of larger particle sizes (∼10–500 μm). The resulting size distribution was different from the sample frozen at −20 °C (Figure B), where only large agglomerates were observed. The peak at 150 nm shows that irreversible agglomeration of the aggregates was prevented to some extent by the process of fast freezing and subsequent freeze-drying. Even though the average particle size was reduced, we found that agglomeration into larger agglomerates was still unavoidable. These larger aggregates (>100 μm) were removed by centrifugation at low speeds (500g). The particle size distribution of the supernatant (Figure , open squares) shows that the peak at smaller particle sizes became more prominent. Even though a bimodal distribution can be seen, the major fraction of the particles was now below 1 μm. Comparing the size as a surface weighted diameter, d3,2, we found an average of 140 nm for the aggregates obtained with the solvent exchange procedure, and 220 nm for the WPI aggregates in the supernatant obtained with the freeze-drying method using rapid freezing. This shows that a high freezing rate can prevent particle size agglomeration to a large extent.

Figure 10

Particle-size distribution of freeze-dried aggregates in sunflower oil using a high freezing rate (○) and the supernatant of the same sample after centrifugation at 500g (□).

Rheology of WPI Oleogels Prepared via Solvent Exchange and Freeze-Drying

From a practical point of view, freeze-drying as an alternative for a solvent exchange is desirable since much less solvent is needed to transfer the protein aggregates to the oil phase and the dried protein powder can be added to the oil directly. To assess the capability of the freeze-dried WPI aggregates to form a network in liquid oil, the supernatant was centrifuged at higher speeds (4000g, 40 min) to collect the protein aggregates as a dense pellet. The rheological properties of the protein oleogel prepared from freeze-dried aggregates were compared to those of the oleogel prepared using the solvent exchange procedure. To allow for comparison between the different oleogels, the composition of the samples was standardized for protein (10 wt %) and water content (1.2 wt %) as determined by Dumas and dry matter analysis, respectively. Both samples were paste-like gels and the results of the frequency dependence can be seen in Figure A. Both oleogels show a high degree of similarity as G′ was roughly an order of magnitude higher than G″ for both samples, indicating an elastic network had formed. Since G′ was only slightly dependent on frequency and the complex viscosity decreased linearly with the applied frequency, it shows that for both gels, the viscoelastic response was not significantly affected by the rate of deformation. Figure B displays for both oleogels the rheological response to an increased strain amplitude of deformation. It can be seen that both samples had a similar linear viscoelastic region, as the G′ deviated from linearity at roughly the same strain value. Furthermore, the overshoot in G″ can be noticed in both samples, which indicated fast rearrangements in the network structure during deformation.[58] Even though the rheological response is highly similar, the magnitude of G′ was somewhat lower for the sample with freeze-dried WPI aggregates. This small difference can be explained by the larger particles present in the freeze-dried sample, which were not observed by applying a solvent exchange procedure. Since larger particles are less effective in creating a network structure, given the lower surface area available, this leads to a less efficient network formation and lower values for the moduli. Nonetheless, by preventing severe particle agglomeration due to the high freezing rate, the increased surface area available for protein–protein interactions resulted in effective gel formation. Given that the results are similar for both type of oleogels, we can conclude that the solvent exchange and the freeze-drying method both lead to effective network formation of the protein aggregates in liquid oil. Therefore, tuning the conditions during drying, such as a fast freezing process, could be an effective strategy to produce a protein powder which is well dispersible in oil and directly capable of forming a gelled structure.

Figure 11

Frequency sweeps (A) and strain sweeps (B) of 10 wt % protein oleogels prepared by solvent exchange (□) or by freeze-drying (○). (A) frequency sweep and (B) strain sweep. Error bars of triplicate measurements were typically not larger than the symbols and were left out for clarity.

Conclusions

The aim of the current paper was to elucidate on the enhanced stability against agglomeration of whey protein isolate (WPI) aggregates in oil after applying a solvent exchange procedure. To this end, heat-set WPI aggregates were transferred from water to several solvents differing in polarity. We have shown that drying protein aggregates by evaporation from solvents with a low polarity (e.g., hexane) resulted in a low density powder, which showed good dispersibility of the aggregates into liquid oil. When the aggregates were dried from more hydrophilic solvents, such as acetone, 1-propanol, or water, the drying process resulted in agglomeration of the protein aggregates, and poor dispersibility in oil. No change in protein composition, protein conformation, or surface hydrophobicity was observed as a result of the solvent exchange procedure. Therefore, we concluded that reduced agglomeration is dominated by a reduction of attractive capillary forces between the protein aggregates during drying. Nonpolar solvents such as hexane, having a low surface tension and low wettability, prevent agglomeration by avoiding a capillary pressure build up during drying. This result suggests that the drying conditions can be tuned to minimize the degree of irreversible agglomeration of the protein aggregates. Indeed, we were able to show that by increasing the freezing rate prior to freeze-drying the water, irreversible agglomeration was prevented to a large extent. The resulting particle size distribution of the freeze-dried WPI aggregates after fast freezing showed to be close to that of the solvent exchange sample. For both methods, the small aggregates were effective in forming a gel network where G′ > G″. This research has shown that by carefully designing a drying process, irreversible agglomeration of WPI aggregates can be prevented to obtain a dried protein material that can be used directly for structure formation in liquid oil.