Patrick Giefer, Sabrina Bäther, Nadine Kaufmes, Helena Kieserling,
Anja Heyse, Wiebe Wagemans, Lars Barthel, Vera Meyer, Emanuel
Schneck, Udo Fritsching, Anja Maria Wagemans
Characterization of beta-lactoglobulin adsorption
on silica membrane pore surfaces and its impact
on membrane emulsification processes
Open Access via institutional repository of Technische Universität Berlin
Document type
Journal article | Accepted version
(i. e. final author-created version that incorporates referee comments and is the version accepted for
publication; also known as: Author’s Accepted Manuscript (AAM), Final Draft, Postprint)
This version is available at
https://doi.org/10.14279/depositonce-19435
Citation details
Giefer, P., Bäther, S., Kaufmes, N., Kieserling, H., Heyse, A., Wagemans, W., Barthel, L., Meyer, V., Schneck,
E., Fritsching, U., & Wagemans, A. M. (2023). Characterization of -lactoglobulin adsorption on silica
membrane pore surfaces and its impact on membrane emulsification processes. In Journal of Colloid and
Interface Science (Vol. 652, pp. 1074–1084). Elsevier BV. https://doi.org/10.1016/j.jcis.2023.08.103.
Terms of use
This work is protected by copyright and/or related rights. You are free to use this work in any way permitted by
the copyright and related rights legislation that applies to your usage. For other uses, you must obtain
permission from the rights-holder(s).
Characterization of β-lactoglobulin adsorption on silica membrane pore surfaces and its 1
impact on membrane emulsification processes 2
3
Patrick Giefer1,8+, Sabrina Bäther2+, Nadine Kaufmes2, Helena Kieserling2,3, Anja Heyse4, 4
Wiebe Wagemans5, Lars Barthel6, Vera Meyer6, Emanuel Schneck7, Udo Fritsching1,8, Anja 5
Maria Wagemans2*
6
+ These authors equally contributed to the manuscript 7
1: Leibniz Institute for Materials Engineering-IWT, Badgasteiner Straße 3, 28359 Bremen, 8
Germany 9
2: Technische Universität Berlin, Institute of Food Technology and Food Chemistry, 10
Department of Food Biosciences, Straße des 17. Juni 135, 10623 Berlin, Germany 11
3: Technische Universität Berlin, Institute of Food Technology and Food Chemistry, 12
Department of Food Chemistry and Analysis, Straße des 17. Juni 135, 10623 Berlin, 13
Germany 14
4: Technische Universität Berlin, Institute of Food Technology and Food Chemistry, 15
Department of Food Technology and Food Material Science, Straße des 17. Juni 135, 10623 16
Berlin, Germany 17
5: Fehrbelliner Straße 52a, 10119 Berlin, Germany 18
6: Technische Universität Berlin, Institute of Biotechnology, Department of Applied and 19
Molecular Microbiology, Straße des 17. Juni 135, 10623 Berlin, Germany 20
7: Technical University of Darmstadt, Department of Physics, 64277 Darmstadt, Germany 21
8: University of Bremen, Particles and Process Engineering, Bibliothekstraße 1, 28359 Bremen, 22
Germany 23
24
* Author of correspondence: 25
Anja Maria Wagemans 26
Department of Food Biosciences 27
Technische Universität Berlin 28
Königin-Luise-Str. 22, 14195 Berlin, Germany 29
Tel.: 0049-30-31471833; 30
Mail: wagemans@tu-berlin.de 31
ABSTRACT 32
Protein adsorption plays a key role in membrane fouling in liquid processing, but the specific 33
underlying molecular mechanisms of β-lactoglobulin adsorption on ceramic silica surfaces in 34
premix membrane emulsification have not been investigated yet. 35
In this study, we aimed to elucidate the β-lactoglobulin adsorption and its effect on the premix 36
membrane emulsification of β-lactoglobulin-stabilized oil-in-water emulsions. In particular, the 37
conformation, molecular interactions, layer thickness, surface energy of the adsorbed β-38
lactoglobulin and resulting droplet size distribution are investigated in relation to the solvent 39
properties (aggregation state of β-lactoglobulin) and the treatment of the silica surface 40
(hydrophilization). 41
The β-lactoglobulin adsorption is driven by attractive electrostatic interactions between 42
positively charged amino acid residues, i.e., lysin and negatively charged silanol groups, and is 43
stabilized by hydrophobic interactions. The strong negative charges of the treated silica surfaces 44
result in a high apparent layer thickness of β-lactoglobulin. Although the conformation of the 45
adsorbed β-lactoglobulin layer varies with membrane treatment and the solvent properties, the 46
β-lactoglobulin adsorption offsets the effect of hydrophilization of the membrane so that the 47
surface energies after β-lactoglobulin adsorption are comparable. The resulting droplet size 48
distribution of oil-in-water emulsions produced by premix membrane emulsification are similar 49
for treated and untreated silica surfaces. 50
KEY WORDS 51
Protein adsorption, conformation, molecular interactions, fouling, ceramic membranes 52
1 INTRODUCTION 53
Premix membrane emulsification (PME) is an emulsification technique for the preparation of 54
emulsions of sensitive biomaterials, such as oil-in-water (O/W) emulsions, which are relevant 55
in several industries. In industries such as food, pharmaceuticals, cosmetics and chemicals, 56
PME is used for the preparation of specific emulsions for consumer and technical products [1], 57
for the encapsulation and delivery of bioactive components [2], and for the synthesis of 58
micro−/nanomaterials [3]. In this technique, a coarse premix is forced through a porous 59
membrane by a pressure gradient. Glass membranes, such as borosilicate membranes, are often 60
used in PME, because the membrane properties – such as pore size, surface roughness and 61
polarity – can be easily modified [4–7]. Borosilicate membranes can be used to produce fine 62
emulsions with a narrow droplet size distribution [8]. Droplet size can be controlled by the 63
transmembrane pressure, pore size and number of cycles [9]. In addition, the droplet size can 64
also be influenced by the membrane surface properties (wettability), as the shear deformation 65
of oil droplets in the emulsion is influenced by the interactions between the oil phase, water 66
phase and membrane material, at the so-called three-phase contact line [7,10]. In addition, the 67
type of emulsifier also has a significant effect on the emulsion produced. High molecular weight 68
emulsifiers such as proteins are often preferred to over synthetic low molecular weight 69
surfactants such as monoglycerides [11]. 70
The stabilization process of O/W interfaces through proteins is complex , but can be described 71
as a three-stage process [12]: In stage (I), the protein migrates through the bulk water phase, in 72
stage (II), the protein adsorbs at the oil/water-interface, and in stage (III), the protein forms an 73
interfacial protein film. A well described model protein is the whey protein β-lactoglobulin (β-74
lg). In previous studies, we investigated the changes of the molecular interactions and the 75
interfacial properties of β-lg as a function of the β-lg aggregation state, which we modified 76
through the solvent properties, specifically using pH 7, pH 7 with 100 mM NaCl and pH 9 (pH 77
7, pH 7NaCl and pH 9) [13–17]. For pH values close to 7, an equilibrium of the monomeric and 78
dimeric forms could be observed [18]. Higher ionic strengths cause a screening of electrostatic 79
charges, shifting the equilibrium to the dimer side [18–20]. At higher pH values, deprotonation 80
leads to negatively charged β-lg, and, due to electrostatic repulsion, the monomeric form 81
dominates [21,22]. Because the solvent properties influence the sensitive monomer-dimer 82
equilibrium, they might also affect the β-lg adsorption leading to a mono- or multi β-lg layers 83
at the O/W interface [4]. This in turn, affects the stabilization process and the droplet size 84
distribution of the emulsion. 85
Besides adsorbing at the O/W interface, β-lg also adsorbs on the membrane surfaces within the 86
PME, which potentially leads to clogging of the pores and the formation of a layer on the 87
membrane surface. This, so-called, fouling of the membrane limits continuous processing and 88
upscaling for industrial applications [23,24]. Moreover, it can lead to a loss of β-lg to function 89
as emulsifier [25, 26]. Fouling is defined as a decrease in membrane performance due to an 90
increasing resistance over time, leading to s an undesired decrease in flux at constant 91
transmembrane pressure [27]. As consequence, membranes need to be regenerated through 92
cleaning procedures or need to be replaced, both lowering the overall efficiency [28]. 93
Fouling is determined by the affinity of the β-lg to the membrane. For adsorption of β-lg to 94
silica surfaces, relevant molecular interactions are electrostatic interactions, hydrogen bonds, 95
hydrophobic interactions as well as van der Waals interactions. For silicas, the silanol groups 96
are deprotonated above the pKa (pH 7.1) [29] and, therefore, exhibit negative charges at neutral 97
pH values [30]. Likewise, β-lg is negatively charged above its isoelectric point, which is pH 5.2 98
- 5.3 [21,31]. Consequently, at neutral pH values electrostatic repulsion between β-lg and the 99
membrane is very likely. However, the presence of local electrostatic attraction between silanol 100
groups and amino acid residues of lysin are also possible. In general, electrostatic interactions 101
can be altered by ions, which is especially true for the β-lg sample pH 7NaCl. Furthermore, 102
hydrophobic interactions can be relevant, since β-lg has hydrophobic residues that can be 103
exposed due to unfolding, for instance at pH 9 [14]. The silica surfaces can also exhibit 104
hydrophobic patches due to organic contaminants or non-oxidized silicon atoms [32], 105
facilitating attractive hydrophobic interactions between β-lg and the membrane. Additionally, 106
van der Waals forces lead to attractive interactions supporting the adsorption. The 107
hydrophilicity of the membrane can be increased through a treatment with piranha, which has 108
two effects. Firstly, it removes the organic contaminants and consequently decreases the 109
hydrophobic interactions [32, 33]. Secondly, the silicon atoms are oxidized and silanol groups 110
are formed, leading to an increase in the hydrophilicity and negative charges of the treated 111
membranes [30]. This in turn affects the molecular interactions between β-lg and the membrane, 112
influencing adsorption, the wetting behavior and consequently the oil droplet size distribution 113
in the O/W emulsions. 114
Although fouling is a well-described phenomenon in membrane-related processes, the 115
molecular interactions responsible for the adsorption of β-lg on silica surfaces have not been 116
discussed in detail in the literature. In particular, the adsorption of β-lg on silica surfaces as 117
function of the aggregation state has not been extensively investigated. Only, Elofsson et al. 118
investigated the layer thickness of adsorbed β-lg on silica surfaces as function of the ionic 119
strength [34]. Jachimska et al. studied β-lg adsorption on a gold surface with increasing ionic 120
strength and found a higher adsorption below the isoelectric point, indicating the importance of 121
electrostatic interaction [35]. Pérez-Fuentes et al. investigated the adsorption of β-lg on 122
hydrophobic latex surfaces using both a quartz crystal microbalance, AFM, and MD simulation 123
and reported values for a β-lg monolayer of approximately 2 nm [36]. With regard to PME, 124
Heyse et al. studied the adsorption of lipase on silica and found a reduced wetting of the silica 125
surface, implying a higher hydrophobicity of the membrane [37]. Molecular dynamic 126
simulations showed an increase in the tertiary structure (the radius of gyration) of the adsorbed 127
lipase, but the lipase activity and secondary structure were not affected by ad- and desorption 128
on the silica membrane. In their study, a lipase solution was used, highlighting the importance 129
of investigating the effects of fouling on the oil droplet break-up and the resulting distribution 130
within the PME. Habibi et al. investigated membrane fouling using the protein (bovine serum 131
albumin) and a small tripeptide (glutathione), but in the context of a different process as 132
microfiltration instead of PME was used [38]. 133
Therefore, the aim of this study is the elucidation of the β-lg adsorption and its effect on the 134
membrane surface and droplet size distribution within the premix membrane emulsification 135
process. To alter the conformational state of the β-lg, the solvent properties are varied (pH 7, 136
pH 7NaCl, pH 9). To study the effect of membrane hydrophilization, the silica surfaces are 137
selectively treated with piranha solution. In this study, we hypothesized the following: 138
• β-lg adsorption: β-lg adsorbs more at pH 7, due to electrostatic attractions between the 139
negatively charged silanol groups and the positively charged amino acid residues of 140
lysin. β-lg adsorption is less pronounced at pH 9 and/or for treated membranes (with a 141
higher negative net charge), due to a stronger repulsion between the β-lg and the silica 142
membrane. 143
• Membrane surface properties and droplet size distribution: The surface energy 144
(wettability) is affected by the hydrophilization of the membrane, because the molecular 145
interaction between β-lg and silica surface are changed. This leads to changes of the 146
conformation of the β-lg layer. As a consequence, the wettability changes with the 147
hydrophilization, which in turn influences the droplet size distribution of the O/W 148
emulsion within the PME. 149
In order to validate these hypotheses, β-lg adsorption on the membrane is characterized using 150
confocal laser scanning microscopy (CLSM) and molecular dynamics (MD) simulations 151
(conformation of β-lg at membrane), ellipsometry (β-lg thickness) and sessile drop analysis 152
using the Owens–Wendt–Rabel and Kaelble model (wettability). The latter two methods are 153
performed with a SiO2-coated wafer, which is equivalent to the borosilicate membrane surfaces 154
[7]. Finally, the droplet size distribution of β-lg stabilized emulsions within PME is 155
characterized using static light scattering. 156
2 EXPERIMENTAL SECTION 157
2.1 Sample Preparation 158
β-lg was isolated from whey protein isolate GermanProt 9000, provided by Sachsenmilch 159
Leppersdorf GmbH (>90 %, Leppersdorf, Germany) as described by Schestkowa et al. [13]. β-160
lg solutions (0.1 %wt) were prepared by dissolving β-lg in distilled water and stirring for 1 h at 161
22 °C and 500 rpm. Three β-lg solutions with different pH and ionic strength were made by 162
adjusting to pH 7.00 ± 0.01, pH 9.00 ± 0.01, and pH 7.00 ± 0.01 containing 0.1 M NaCl 163
(referred to as pH 7, pH 9 and pH 7NaCl) using 0.1 M NaOH, 0.1 M HCl and NaCl, all purchased 164
from Carl Roth GmbH (analytical-grade, Karlsruhe, Germany). 165
To remove free fatty acids and other interfacial active compounds, middle-chain 166
triacylglyceride oil (MCT-oil, >99.9 %, Witarix MCT 60/40, IOI Oleochemical, Hamburg, 167
Germany) was treated with Florisil (activated magnesium silica; MgO x 3.6 SiO2 x 1.53 OH, 168
100 %, Carl Roth GmbH, Karlsruhe, Germany), in accordance with Schestkowa et al. [14]. 169
The premix O/W emulsion was prepared by dispersing 5 % (w/w) purified MCT-oil (dispersed 170
phase) in a β-lg solution (continuous phase). Coarse emulsions were produced by using a rotor-171
stator homogenizer (Ultra Turrax T25 basic, 500 W power output, S25 KD-25 F dispersion 172
tool, IKA GmbH, Staufen, Germany) at 6500 min-1 for 15 s. Then, the coarse emulsions were 173
pushed through a silica membrane (borosilicate membranes type P4, membrane diameter 174
10 mm, membrane height 2.0 ± 0.2 mm median pore diameter 10 – 16 µm, ROBU Glasfilter-175
Geräte GmbH, Hattert, Germany) at a transmembrane pressure of 5 bar (effective membrane 176
area Aeff = 28.3 mm2). The scheme of the PME process is depicted in Figure 1A. A new 177
membrane was used for each emulsification process. 178
2.2 Membrane Treatment 179
Both, the silica membranes (borosilicate membranes type P4, ROBU Glasfilter-Geräte GmbH, 180
Germany) and the membrane equivalent SiO2-coated wafer (SIEGERT Wafer GmbH, Aachen, 181
Germany) were treated by soaking them in a piranha solution consisting of three parts (v/v) 182
sulfuric acid (98 %, Carl Roth GmbH, Germany) and one-part (v/v) hydrogen peroxide (35 %, 183
Carl Roth GmbH, Germany) for 30 min. Afterwards, the membranes and wafers were rinsed 184
with distilled water until the pH remained constant. The treatment result in a different amount 185
of silanol groups of silica backbone (Figure 1C). 186
187
Figure 1 Flow diagram of the premix membrane emulsification process (A): Compressed air 188
supply (1), pressure valve (2), safety valve (3), emulsification unit containing premix emulsion 189
(4), membrane (5), beaker containing fine emulsion (6), weight balance (7). Photograph of 190
borosilicate membrane (10 mm diameter, 2.8 mm height) and SiO2-coated wafers (B). 191
Schematic illustration of the untreated (contamination with particles, metals, chemicals) and 192
treated silica surfaces with functional silanol groups (piranha treatment consisting of sulfuric 193
acid and hydrogen peroxide) at pH > 7 (modified accordingly to [39]) (C). 194
2.3 Confocal Laser Scanning Microscopy 195
30 mL of a 0.1 wt% β-lg solution followed by 20 mL solvent (pH 7, pH 9 and pH 7NaCl) were 196
pushed through the membrane (using the premix membrane emulsification unit, Figure 1A). 197
Afterwards, the used membranes were stained with 80 μL of a Nile Red solution (0.02 wt% in 198
acetone, Carl Roth GmbH, Karlsruhe, Germany) for 10 min in dark and then, rinsed with 199
distilled water. 200
Changes of local polarity in immediate proximity of Nile Red binding sites cause a red shift in 201
the fluorescent emissions of Nile Red [40–42]. As a consequence, structural changes become 202
visible through a color change from blue (hydrophobic) to red (hydrophilic). 203
The microscopic image data were acquired using a Leica TCS SP8 laser scanning confocal 204
microscope (Leica Microsystems, Wetzlar, Germany). All images were recorded with a HC PL 205
APO CS2 20x/0.75 IMM objective. Sample excitation was performed with a 514 nm laser. 206
Detectors were set to a range of 550 nm – 650 nm (signal visualized in blue) and 207
651 nm – 750 nm (signal visualized in red). Z-stacks with a depth of approximately 150 µm 208
were recorded for all samples and visualized in a 3D model using LAS X software (Leica 209
Microsystems, Wetzlar, Germany). 210
To quantify the red and blue contributions to the detected signal for each pixel of the Z-stack 211
images, a custom Python script was used to calculate histograms of the number of pixels within 212
a Z-stack image with specific red/blue ratios, ranging from only blue to only red. To obtain a 213
symmetric range for the histogram, the red/blue ratios are calculated as atan(Ired/Iblue), with 214
Ired and Iblue the respective red and blue pixel signal, analogous to the hue value in the HSV 215
color model. 216
2.4 Molecular dynamic simulations 217
The molecular dynamic simulations were conducted based on the Gromacs simulation package 218
[43] using AMBER-99SB force field in explicit TIP3P water and periodic boundary conditions. 219
The β-lg structure has been taken from the protein data bank (code: 1BSQ) and the protonation 220
state of the titratable amino acids were adjusted to a pH of 7 using H++ web server version 3.2 221
[44]. The disulfide bonds were assigned manually. The structure has been solvated in a water 222
box with 8 Na+ ions to ensure charge neutrality. The system was energy minimized and 223
equilibrated for 20 ns within the canonical-, isothermal-isobaric-, and microcanonical 224
ensemble. The amorphous silica surface was modeled according to the work of Cole et al. [45] 225
using the interaction potentials for water-silica derived by Butenuth et al. [46]. The protonation 226
state of the terminal surface groups at pH 7 was adjusted to reproduce those of the 227
potentiometric titration experiments [47,48]. The system containing the equilibrated β-lg 228
structure and the 9 nm x 9 nm silica slab structure were packed using packmol [49] and solvated 229
in Gromacs adding 79 Na+ ions to ensure charge neutrality. The β-lg was placed 10Å above 230
the interface. During the simulation runs all surface atoms except the hydrogens were 231
constrained. All bonds involving a hydrogen atom were constrained using the LINCS algorithm 232
[50]. The system was energy minimized and equilibrated. Consecutively, the canonical 233
simulation runs were performed using a soft coupling time of 100 ps to ensure that the system 234
is not influenced by small temperature fluctuations. A timestep of 2 fs was chosen and the cutoff 235
distance for non-bonded interactions was set to 1.2 nm. The electrostatic interactions were 236
treated with a Particle Mesh Ewald method [51]. Result visualization was performed using 237
pymol [52] and the molecular dynamic trajectories also were analyzed within Gromacs. 238
2.5 Layer Thickness (Ellipsometry) 239
The ellipsometry method is based on the change in the polarization state of light upon reflection 240
from the surface at an incident angle θ. For a given refractive index n, the change depends on 241
the layer thickness and is quantified in terms of the phase difference Δ and the amplitude ratio 242
Ψ encoded in the ratio between the complex reflection coefficients Rs and Rp for s and p 243
polarizations, respectively [53]: 244
𝑅𝑅𝑃𝑃𝑅𝑅𝑆𝑆
⁄=tan Ψ 𝑒𝑒−𝑖𝑖Δ [1]
Ellipsometry measurements were performed at θ = 70° with an Optrel Multiskop ellipsometer 245
(Optrel GbR, Sinzing, Germany) working at a wavelength λ = 632.8 nm. 246
SiO2-coated wafer (105 nm SiO2 layer thickness, SIEGERT Wafer GmbH, Aachen, Germany) 247
with thermal oxide were used as substrates, either “treated” or “untreated” (see Section 2.2 248
Membrane Treatment). The time between membrane treatment and β-lg adsorption was kept 249
below 30 min in order to prevent a reversible structural change of the adsorbed β-lg. The wafer 250
remained in the β-lg solutions for 30 min. Afterwards, the wafers were briefly rinsed with 251
distilled water. 252
Three independent measurements were performed on different positions along the surface of 253
each wafer. Based on the refractive index of silicon nSi = 3.885 - 0.018i [54] and that of silicon 254
oxide, determined earlier for the same batch of silica surfaces [55], nSiOx = 1.468, a fit to the 255
measured values Δ untreated = (79.2 ± 0.4)° and Ψ untreated = (43.1 ± 0.3)° yielded an oxide 256
thickness of dSiOx_untreated = (102.7 ± 0.8) nm for the untreated wafer, where the numbers are 257
given as mean ± standard deviation. For the treated wafer, Δ treated = (79.2 ± 0.4)° and 258
Ψ treated = (43.4 ± 0.4)°. In the next step, the measurement values obtained after β-lg adsorption 259
and subsequent drying (averaged again over 3 positions) were analyzed while accounting for 260
the oxide layer properties obtained individually for the corresponding bare wafer. In this 261
procedure, we assumed nblg = 1.575, as determined earlier [56]. It should be noted that the 262
apparent layer thicknesses measured by ellipsometry characterize the adsorbed β-lg amount in 263
terms of the β-lg volume per unit area rather than the spatial extension of the individual β-lg 264
molecules in the direction perpendicular to the interface. The native layer thickness (with SiO2) 265
of each wafer was measured to deduce the difference between this and the layer thickness after 266
treatment with the β-lg solution on the real layer thickness of the β-lg film. In this way, 267
measurement errors due to variations in wafer quality could be excluded. 268
2.6 Membrane Wettability 269
The sessile drop method was used to determine the surface energy from the optical contact 270
angle using the OCA20 instrument (Dataphysics Instruments GmbH) equipped with a high-271
speed camera (IDS uEye CP, 123 frames per minute, Dataphysics Instruments GmbH, 272
Germany). The drop shape was captured after 16 s and the contact angle was calculated with 273
an algorithmic approximation based on the Young-Laplace differential equation (SCA20 274
software, Dataphysics Instruments GmbH, Germany). 275
The measurements were performed at a temperature of 22 °C using membrane equivalent SiO2-276
coated wafers (prepared as described before) as a model system for the membrane. To analyze 277
the membrane wettability, the surface free energy, which consists of a polar fraction and a 278
disperse fraction, was calculated using the OWRK method [57–59]. At least two liquids with 279
known disperse and polar parts of the surface tension are required to determine the surface free 280
energy of the solid, wherein at least one of the liquids must have a polar fraction > 0. Thus, 281
distilled water, ethylene glycol and a mixture thereof (1:1, wt%) were used as reference liquids 282
to determine the free surface energy and polar and disperse fractions of the wafers which were 283
produced for the ellipsometry measurement (2.4). 284
2.7 Droplet Size Measurement 285
The drop size distributions of the emulsions were measured with a laser light scattering 286
spectrometer (Horiba LA-950, Retsch Technology GmbH, Haan, Germany). The sample was 287
constantly circulated and stirred during the measurement. No additional ultrasound was used 288
for dispersion. The drop size determination is based on the Mie theory, using a refractive index 289
of 1.46. Each sample was analyzed in analytical triplets, and the mean average of the analytical 290
triplet has been used as one technical measurement. Furthermore, all measurements were 291
performed in duplicates. The measured droplet sizes of the emulsions were sorted into 32 292
classes within a size range from 0 to 200 µm and evaluated as a quantity-based cumulative 293
distribution function. 294
3 RESULTS & DISCUSSION 295
3.1 Characterization of β-lg adsorption 296
First, the conformational changes of β-lg as indicated through emission intensity shifts in 297
CLSM are covered. Second, the relevant molecular interactions between the different β-lg 298
samples and borosilicate surfaces are specified using MD simulations. Third, the β-lg film 299
thickness measured with ellipsometry is discussed. 300
For a better understanding of the adsorption within the PME of β-lg-stabilized emulsions, 301
CLSM has been used. Besides detecting the adsorption of β-lg-stabilized emulsions on the 302
membrane surface, the conformational change affecting the molecular interactions of adsorbed 303
β-lg can be examined. Nile Red is able to interact with β-lg through hydrophobic interactions 304
which causes an increase in Nile Red emission intensity accompanied by a shift to smaller 305
wavelengths (blue) in a hydrophobic environment [40] and a shifts to larger wavelengths (red) 306
in a hydrophilic environment. The β-lg conformational changes that facilitate those shifts can 307
either be caused by β-lg adsorption on the membrane, by solvent properties modifying 308
electrostatic interactions, or by a combination of both. Figure 2A and B were obtained in the 309
same CLSM measurement. Figure 2A shows the transmission light channel image, and Figure 310
2B the fluorescence image of an unused membrane, both obtained from the same sample. In 311
addition, CLSM images of a treated (Figure 2D) and untreated (Figure 2C) membrane used in 312
the PME of a β-lg stabilized emulsion at pH 7 are shown. Since the unused membrane that was 313
stained with Nile Red did not show any fluorescence (Figure 2B), the structures visible in 314
Figure 2C and D can be ascribed to adsorbed β-lg, MCT oil or both. Neither the color nor the 315
fluorescence pattern varied for the different membranes or solvent properties (data not shown). 316
This can be explained by the fact, that Nile Red interacts through hydrophobic interactions, not 317
only with β-lg, but also with MCT oil, which might superimpose the fluorescence signal caused 318
by conformational changes of the β-lg. 319
To better understand the conformational changes upon adsorption, aqueous β-lg solutions, 320
instead of β-lg-stabilized emulsions were used in further investigations to isolate the influence 321
of the adsorbed β-lg from the influence of the oil. Accordingly, Figure 3 shows CLSM 322
micrographs of borosilicate membranes after PME processing with β-lg solutions. For all 323
samples, adsorption was evident. The fluorescence pattern and intensity varied with the 324
membrane treatment and the solvents used. An influence of the fluorescence intensity and 325
emission spectra of Nile Red due to different pH values can be neglected because Nile Red is 326
pH-stable between pH 4 to 9 [40,41]. Nevertheless, the intensities of the micrographs in 327
different solvents should not be compared with each other, as the fluorescence intensity was 328
adjusted for each CLSM micrograph in order to obtain the highest brightness and intensity. To 329
get a qualitative measure of the blue shift, histograms of the number of pixels with specific 330
blue/red contributions within a Z-stack image were calculated. This enables the comparison 331
between the treated and untreated membrane for each solvent, where a high count of pixels with 332
a strong blue component can be linked to a hydrophobic environment and those with a high red 333
component to a hydrophilic environment. 334
At pH 9, the untreated membrane gave a relatively higher count of blue pixels (hydrophobic 335
environment), compared to the treated membrane, while the red count (hydrophilic 336
environment) was comparable for both membranes. In contrast, at pH 7 the untreated membrane 337
showed a more pronounced red shift, whereas the adsorption at the treated membrane led to a 338
higher blue count. For the electrostatically shielded β-lg at pH 7NaCl, similar red and blue counts 339
were obtained. The different adsorption behavior at different pH and ionic strength can be 340
explained by variation of the electrostatic interactions. Due to less electrostatic repulsion, the 341
adsorption of β-lg to the untreated membrane is more pronounced at pH 7 compared to pH 9. 342
The adsorbed β-lg layer determines the surface hydrophilicity, rather than the membrane 343
surface itself. As a result, the surface of the exposed β-lg layer at pH 7 appears more hydrophilic 344
due to the hydrophilic β-lg surface. For β-lg at pH 7NaCl, the higher ionic strength screened the 345
electrostatic charges, facilitating a more hydrophobic vicinity regardless of the membrane 346
treatment leading to comparable blue and red counts. In this context, the untreated membrane 347
at pH 7NaCl appeared more inhomogeneous with larger red areas compared to the treated 348
membrane. This can be ascribed to larger inhomogeneities due to organic contaminants or non-349
oxidized silicon atoms that locally altered the Nile Red’s environment. 350
351
Figure 2 CLSM transmitted light image (A) and CLSM fluorescence image (B) of an unused 352
borosilicate membrane (no signal). CLSM Z-stack images of an untreated (C) and piranha 353
treated (D) borosilicate membrane used in the PME process of an emulsion stabilized with β-354
lg at pH 7. Nile Red was used as fluorescence dye in all CLSM micrographs. 355
356
Figure 3 CLSM images and the normalized count of the pixel color (blue/red contribution) 357
deduced from the CLSM images of the untreated and treated borosilicate membranes used in 358
PME of β-lg (0.1 wt%) solutions at pH 7, pH 7NaCl and pH 9. 359
Although CLSM results indicate that the β-lg adsorption varies with the treatment and the 360
aggregation state, the involved molecular interactions are not clear and need to be identified 361
using MD simulation. The MD results in Figure 4 indicate that the adsorption of the β-lg can 362
be explained by the electrostatic interactions between the β-lg and the surface. Since the silica 363
interface is mainly negatively charged at pH 7, the β-lg adsorbs with the remaining positively 364
charged residues (blue) to the interface. The negatively charged side of the β-lg (blue) is facing 365
away from the interface into the bulk parallel and normal to the interface. 366
To identify the contact and binding residues, the binding time was analyzed and is depicted in 367
Figure 5. Since the formation of hydrogen bonds is possible within 3Å relative distance to the 368
interface, a vertical line is plotted each timestep a residue of the β-lg is within this distance. 369
This allows the identification of frequent and constant binders, as well as the time of adsorption. 370
The adsorption occurs after around 150 ns. The positive charged contact residues at pH 7, 371
LYS8, LYS14, and LYS100 are identified. At pH 7, lysin is protonated and, therefore, 372
positively charged, due to a pKa of 10.5 [60]. Furthermore, the non-charged hydrophobic 373
residue TYR99 shows a significant contact to the interface. Due to its neutral charge, the 374
proximity to the surface can be explained by the neighboring binding residue LYS100. After 375
the adsorption to the silica interface at around 100 ns, an exposure of the hydrophobic residues 376
can be noted, as shown in Figure 5. These results emphasize that the adsorption mechanism to 377
the silica interface is determined by electrostatic interactions, and then stabilized by 378
hydrophobic interactions being exposed during adsorption. 379
380
Figure 4 Electrostatic potential surface from –1 eV (red) to 1 eV (blue) of β-lg at silica-water 381
interface viewed from four different sides and the top, the silica interface is represented in grey. 382
383
Figure 5 Binding residues over time for 1BSQ at SiO2-water-interface (A), solvent accessible 384
surface (SAS) over time divided in hydrophobic and hydrophilic areas (B). 385
Ellipsometry was used to characterize the apparent layer thickness of the adsorbed β-lg. In 386
Fehler! Verweisquelle konnte nicht gefunden werden. the measured layer thickness is shown. 387
For all samples, values below 2 nm were reported. Overall, the layer thickness for the treated 388
surfaces were similar for the different conformational states all ranging between 0.3 - 0.6 nm. 389
In contrast, the layer thickness for the untreated surfaces varied. The highest layer thickness 390
was found for the untreated surface at pH 7 with 1.7 nm, followed by pH 7NaCl with 1.1 - 1.2 391
nm. At pH 9, the film thickness was the lowest (0.6 - 0.8 nm). 392
393
Figure 6 Film thickness (sample 1 and 2 of duplicate) analyzed by ellipsometry on the treated 394
and untreated SiO2-coated wafer after 30 min treatment of 0.1 wt% β-lg solution at pH 7, 395
pH 7NaCl and pH 9. 396
Considering the β-lg diameter of 2 - 3.6 nm [19,61], either the formation of a flattened β-lg 397
monolayer on the silica surfaces [35] or patches of adsorbed β-lg with uncovered areas in 398
between can be assumed. The latter might be more reasonable for the untreated silica, as the 399
surface is more heterogenic compared to the treated hydroxylated silica. Pérez-Fuentes et al. 400
[36] analyzed the layer thickness of uniform β-lg monolayers adsorbed to hydrophobic latex 401
surfaces using both a quartz crystal microbalance, AFM, and MD simulation and reported 402
values of approximately 2 nm. Their results are in line with our results, although the molecular 403
interactions might be different for β-lg adsorbing on silica and β-lg adsorbing on latex surfaces. 404
A high film thickness for the untreated surface can be explained by less electrostatic repulsion 405
between the β-lg and silica and more hydrophobic attraction between them, confirming the first 406
hypothesis of this work. For pH 7, in addition to hydrophobic interactions, local electrostatic 407
attraction between the negatively charged silanol groups and the positively charged amino 408
residues of lysine contribute to a greater adsorption. At pH 7NaCl, the electrostatic interactions 409
are screened through the counter ions, increasing the density of the ß-lg molecule and, therefore, 410
reducing the film thickness [15]. At pH 9, the strong electrostatic repulsion within the β-lg 411
molecules and between the β-lg and silica surface might hinder the adsorption. This is also 412
consistent with the CLSM micrographs of pH 9 in Figure 3. Much bigger and fewer 413
fluorescence patches were visible compared to the other solvents, suggesting the ß-lg rather 414
interacts with itself than the negatively charged surface. Even though the strong electrostatic 415
repulsion counteracts adsorption, hydrophobic interactions caused by residues of non-oxidized 416
silicon atoms or van der Waals interactions could facilitate attraction between the partially 417
unfolded β-lg at pH 9 and the silica surface [14]. This is supported by the film thickness of the 418
treated surfaces not changing with the solvent properties, which modify electrostatic 419
interactions. Besides attractive hydrophobic interactions, β-lg adsorption is thermodynamically 420
favorable, since counter ions are released leading to an entropy gain which can be the driving 421
force for β-lg adsorption at pH values above the isoelectric point [34,62]. The relevance of 422
electrostatic interaction upon β-lg adsorption on surfaces was also demonstrated by Jachimska 423
et al. [35]. In their study, the adsorbed mass of β-lg on a gold surface increased with the ionic 424
strength. Moreover, the β-lg adsorption was higher below the isoelectric point, where the β-lg 425
is positively charged, compared to pH values above the isoelectric point, where β-lg and the 426
gold surface are both negatively charged. This was also found by Elofsson et al. [34], where the 427
adsorbed mass increased with the ionic strength. In the study of Marsh et al. [63], the adsorbed 428
mass of β-lg was higher for hydrophobic surfaces, consisting of a monolayer of C12 chains, 429
compared to hydrophilic silica surfaces, since electrostatic repulsion did not counteract the 430
hydrophobic attraction. In their study, the measured amounts of adsorbed mass were lower than 431
the estimated amount for dense packed β-lg monomer layers on the surface, indicating that the 432
β-lg conformation was affected upon adsorption. 433
Overall, our first hypothesis could be verified as β-lg adsorption was found to be driven by 434
electrostatic attractions between the negatively charged silanol groups and the positively 435
charged amino acid residues of lysin at pH 7 and pH 9. Moreover, β-lg adsorption was found 436
to be less pronounced for the treated surfaces with a higher negative net charge and at pH 9, 437
due to a stronger repulsion between the β-lg and the silica membrane. Here, the electrostatic 438
repulsion impedes the approach of β-lg and therefore, the formation of local attractive 439
electrostatic interactions between lysin residues. In addition, the adsorbed β-lg layer 440
conformation varied in respect to the membrane treatment and solvent properties. Whether the 441
conformational changes and β-lg layer thickness influence surface properties and therefore the 442
droplet backup, is discussed in the following section. 443
3.2 Characterization of Membrane Surface Properties and Oil Droplet Distribution 444
In this section, the surface properties of silica surfaces modified through β-lg adsorption and 445
the resulting oil droplet distribution of β-lg stabilized emulsions produced using PME are 446
discussed. The sessile drop method was used to determine the surface energy from the contact 447
angle applying the Owens–Wendt–Rabel and Kaelble model (OWRK), allowing conclusions 448
about the wettability before and after β-lg adsorption [64]. This is relevant as the droplet 449
breakup within PME is directly correlated to the interactions at the three-phase contact line 450
(continuous and dispersed phase of the emulsion and the wall material) [10]. 451
In Table 1, the measured surface energy (with the polar and disperse fractions) of the treated 452
and untreated silica surfaces are listed. According to Owens–Wendt–Rabel and Kaelble, the 453
free surface energy of surfaces, can be investigated by measuring the contact angle of reference 454
liquids with a known surface tension and known polar and disperse fractions. In general, the 455
surface energy consists of a polar fraction (Coulomb interactions; interactions between 456
permanent dipoles and induced dipoles like H-bonds, Lewis acid/base interaction) and a 457
disperse fraction (mostly van-der-Waals-interaction). The more these fractions match between 458
two phases, the more they interact and the higher the adhesion and adsorption. The treated silica 459
surfaces before β-lg adsorption exhibited the highest polar fraction (60.1 J/m2) and the lowest 460
disperse fraction (5.6 J/m2) resulting in the highest surface energy (65.6 J/m2). In contrast, the 461
untreated silica surfaces show the lowest surface energy and polar fraction as well as the highest 462
disperse fraction with 43.3 J/m2, 23.1 J/m2 and 20.3 J/m2, respectively. This is in line with our 463
assumptions that the treatment with piranha removes the organic contaminants and silicon 464
atoms are oxidized, leading to an increase of the overall hydrophilicity of the treated 465
membranes. Comparing the surface energy of the treated membrane before and after β-lg 466
adsorption, it is evident that the surface energy due to β-lg adsorption decreases from 65.6 J/m2 467
to 50.4 J/m2, therefore resulting in a more hydrophobic membrane surface. After β-lg 468
adsorption, the surface properties of the treated and untreated membranes at the different 469
solvent properties are equalized. Here, the treated silica surfaces lead to slightly higher surface 470
energies (50.1 - 51.4 J/m2) compared to the untreated silica surfaces (46.6 - 49.7 J/m2). 471
Accordingly, the polar fraction of the treated surface after β-lg adsorption becomes slightly 472
higher and the disperse fraction slightly lower compared to the untreated surface. An effect of 473
the solvent is not evident. The balancing of the surface properties of the treated and untreated 474
wafers after β-lg adsorption can be explained by the formation of adsorbed layers with uniform 475
surface properties. Although the CLSM results indicate differences in the β-lg conformation of 476
the adsorbed β-lg, the outer surface of the monolayer adsorbed film might be similar. An 477
explanation could be that, regardless of the solvent properties and the surface treatment, β-lg 478
binds to the membrane (via local electrostatic attraction and/or hydrophobic interactions), and 479
exposes predominantly hydrophilic residues, since majority of the hydrophobic patches are 480
hidden in the calyx. This also is evident in the corresponding MD simulations. 481
Table 1. Membrane wettability characterized by surface energy with polar and disperse 482
fractions of the treated and untreated silica surfaces after exposure to 0.1 wt% β-lg solution at 483
pH 7, pH 7NaCl and pH 9 for 30 min. 484
Membrane
Solvent properties Surface energy [J/m2]
pH NaCl [M] Polar Disperse Surface energy
Untreated
– –
23.1
20.3
43.3
Treated
60.1
5.6
65.6
Untreated
7 0.0
31.7
14.8
46.6
Treated
35.6
14.8
50.4
Untreated
7 0.1
30.7
16.3
47.1
Treated
37.7
13.7
51.4
Untreated
9 0.0
33.5
16.2
49.7
Treated
35.5
14.7
50.1
485
The β-lg stabilized emulsion samples were produced with the custom-made PME device 486
(Figure 1A). Since the NaCl screens the electrostatic interactions between the oil droplets, 487
aggregation occurred at pH 7NaCl and, therefore, these results are not taken into consideration. 488
The oil droplet size distribution at pH 7 and 9 is presented in Figure 7. Comparing the 489
cumulative droplet size distribution of the emulsions prepared with PME, only small differences 490
are found, disproving the second hypothesis. The treated membrane at pH 9 leads to a slightly 491
smaller mean droplet size (~ 9 µm) in comparison to the other samples (~10 µm). This is in line 492
with the results of Kieserling et al. [15], where the same PME set-up was used. As the emulsion 493
is forced through the membrane, the pressure drop across the membrane causes the droplets to 494
deform, leading to the formation of necks and eventually, the breakup of the droplets. The 495
deformation and neck formation is highly dependent on the wettability of the membrane 496
[65,66], thus, a similar wettability of the membranes might produce similar droplet sizes. 497
Wollborn et al. [67] also produced β-lg stabilized O/W emulsions with oil droplet sizes in a 498
similar range using a comparable PME device and identical (untreated) borosilicate 499
membranes, although different pressures (2 and 8 bar instead of 5 bar) were used. The slightly 500
lower oil droplet sizes at pH 9 compared to pH 7 was explained by increasing negative charges 501
of the adsorbed β-lg at pH 9 at O/W interfaces resulting in increased unfolding of the β-lg at 502
the interface leading to a large interfacial area as demonstrated by Kieserling et al. [68]. In their 503
study, the electrostatic repulsion within the β-lg molecule led to a structural flexibility and a 504
high film stability, which in turn, might favor the formation of a smaller droplet sizes. The pH-505
dependent degree of β-lg unfolding at the water/air interface was also reported by Gochev et al. 506
[69]. 507
508
Figure 7 Effect of the membrane treatment and pH on the cumulative distribution function 509
(cdf Q0) of the oil droplet size ddrop of β-lg (0.1 wt%) stabilized emulsions produced by PME 510
for the untreated (UT, solid lines) and treated membrane (T, dashed lines), respectively. 511
4 CONCLUSIONS 512
This study clarifies underlaying mechanism of β-lactoglobulin adsorption on silica surfaces and 513
its effect on premix membrane emulsification of β-lactoglobulin-stabilized oil-in-water 514
emulsions, which have not been covered in literature before. In particular, the molecular 515
interactions responsible for the β-lactoglobulin adsorption on silica surfaces in porous ceramic 516
membranes, the determination of the conformation and layer thickness of the adsorbed β-517
lactoglobulin, the surface energy, and the resulting emulsification droplet size distribution of β-518
lactoglobulin stabilized oil-in-water emulsions within the premix-membrane emulsification 519
process have been considered. 520
The β-lactoglobulin adsorption upon silica surfaces was determined with confocal laser 521
scanning microscopy and the conformation of the adsorbed β-lactoglobulin was found to be 522
dependent on the surface hydrophilization and solvent properties (pH 7, pH 7NaCl and pH 9). 523
Molecular dynamics simulations demonstrated that the adsorption is driven by local 524
electrostatic attraction and stabilized through hydrophobic interactions, confirming the first 525
hypothesis that adsorption is due to local attractive interactions between the negatively charged 526
silanol groups and the positively charged amino acid residues of lysin. The relevance of 527
electrostatic interactions for β-lg adsorption on surfaces was also emphasized by Jachimska et 528
al. [35] and Elofsson et al. [34]. 529
Apparent layer thicknesses below 2 nm were found in our study, suggesting the formation of 530
either a flattened β-lactoglobulin monolayer or adsorbed β-lactoglobulin patches with 531
uncovered areas in between. The untreated membrane led to the highest layer thickness at pH 532
7 due to less electrostatic repulsion, while also forming attractive electrostatic interactions 533
between lysin residues and the silanol groups. The results are in line with Pérez-Fuentes et al. 534
[36], who also found values of approximately 2 nm. 535
The surface energies of the adsorbed β-lactoglobulin layers were similar for all samples, which 536
disproves the second hypothesis, where we stated that the hydrophilization is affecting the 537
surface energy of the adsorbed β-lactoglobulin layer. This can be explained by the hydrophilic 538
amino acid residues predominating the β-lactoglobulin surface, leading to an overall 539
hydrophilic surface regardless of the conformation. As a consequence, no substantial 540
differences for the droplet size distribution of the β-lactoglobulin stabilized oil-in-water 541
emulsions produced through premix membrane emulsification regarding different solvent 542
properties or the membrane treatment were found. To sum up, the influence of solvent 543
properties and membrane hydrophilicity were evident through differences in the conformation 544
of the adsorbed β-lactoglobulin film on a molecular level, however, on a macroscopic level, the 545
membrane hydrophilization with piranha solution did not significantly affect the emulsion 546
droplet size distribution. 547
The adsorption on the membrane might cause additional stress on the protein on a molecular 548
level. In that case, the intensity of the molecular interaction, and the consequent conformational 549
changes, influence the nativity and functionality. To study the effect of conformational changes 550
on the function of proteins in oil-in-water emulsions within premix membrane emulsification, 551
other proteins such as enzymes could be used. The investigation of the loss of enzymatic activity 552
due to structural changes upon ad- and desorption of lipase could provide more detailed 553
information about the sorption behavior und stress residence time of sensitive proteins within 554
the premix membrane emulsification process. 555
AUTHORS CONTRIBUTIONS 556
Patrick Giefer: Investigation; Methodology (MD simulation); Writing - Original Draft
557
Sabrina Bäther: Writing - Original Draft; Visualization; Writing - Review & Editing
558
Helena Kieserling: Visualization; Writing - Review & Editing
559
Anja Heyse: Visualization; Writing - Review & Editing
560
Nadine Kaufmes: Investigation (sample preparation, membrane treatment, CLSM, 561
ellipsometry, membrane wettability, droplet size measurement)
562
Wiebe Wagemans: Methodology (python script), Writing - Review & Editing 563
Lars Barthel: Methodology (CLSM), Writing - Review & Editing 564
Vera Mayer: Supervision 565
Emanuel Schneck: Methodology (ellipsometry) 566
Udo Fritsching: Supervision; Writing - Review & Editing 567
Anja Maria Wagemans: Conceptualization, Writing - Original Draft, Resources, Writing 568
- Review & Editing, Supervision, Project administration, Funding 569
acquisition
570
571
ACKNOWLEDGEMENTS 572
This work was supported by the German Research Foundation (DFG) within the priority 573
program, SPP1934 “DiSPBiotech – Dispersity, structural and phase modifications of proteins 574
and biological agglomerates in biotechnological processes”. Computational resources were 575
provided by the North-German Supercomputing Alliance (HLRN) under Grant No. hbi00037. 576
REFERENCES 577
[1] D.J. McClements, Food Emulsions: PRINCIPLES, PRACTICES, AND TECHNIQUES, 578
THIRD EDITION, Food Emuls. Princ. Pract. Tech. Third Ed. (2015) 1–676. 579
https://doi.org/10.1201/B18868/FOOD-EMULSIONS-DAVID-JULIAN-580
MCCLEMENTS. 581
[2] M. Chappat, Some applications of emulsions, Colloids Surfaces A Physicochem. Eng. 582
Asp. 91 (1994) 57–77. https://doi.org/10.1016/0927-7757(94)02976-8. 583
[3] N. Anton, J.P. Benoit, P. Saulnier, Design and production of nanoparticles formulated 584
from nano-emulsion templates-A review, J. Control. Release. 128 (2008) 185–199. 585
https://doi.org/10.1016/J.JCONREL.2008.02.007. 586
[4] M. Rabe, D. Verdes, S. Seeger, Understanding protein adsorption phenomena at solid 587
surfaces, Adv. Colloid Interface Sci. 162 (2011) 87–106. 588
https://doi.org/10.1016/j.cis.2010.12.007. 589
[5] K. Al-Malah, J. McGuire, V. Krisdhasima, P. Suttiprasit, R. Sproull, Ellipsometric 590
evaluation of .beta.-lactoglobulin adsorption onto low- and high-energy materials, 591
Biotechnol. Prog. 8 (1992) 58–66. https://doi.org/10.1021/bp00013a009. 592
[6] C.A.-C. Karlsson, M.C. Wahlgren, A.C. Trägårdh, β-Lactoglobulin fouling and its 593
removal upon rinsing and by SDS as influenced by surface characteristics, temperature 594
and adsorption time, J. Food Eng. 30 (1996) 43–60. https://doi.org/10.1016/S0260-595
8774(96)00045-3. 596
[7] G.T. Vladisavljević, J. Surh, J.D. McClements, Effect of emulsifier type on droplet 597
disruption in repeated shirasu porous glass membrane homogenization, Langmuir. 22 598
(2006) 4526–4533. https://doi.org/10.1021/la053410f. 599
[8] N. Hornig, U. Fritsching, Liquid dispersion in premix emulsi fi cation within porous 600
membrane structures, J. Membr. Sci. 514 (2016) 574–585. 601
https://doi.org/10.1016/j.memsci.2016.04.051. 602
[9] A. Nazir, K. Schroën, R. Boom, Premix emulsification: A review, J. Memb. Sci. 362 603
(2010) 1–11. https://doi.org/10.1016/j.memsci.2010.06.044. 604
[10] T. Wollborn, P. Giefer, H. Kieserling, A.M. Wagemans, S. Drusch, U. Fritsching, 605
Investigation of local and temporal interfacial shear stress distribution during membrane 606
emulsification, Can. J. Chem. Eng. 100 (2022) 1061–1078. 607
https://doi.org/10.1002/CJCE.24186. 608
[11] H. Kiserling, Influence of the Protein Structure on the Stabilization of Oil/Water-609
Interfaces, Berlin, 2021. 610
[12] S.G. Baldursdottir, M.S. Fullerton, S.H. Nielsen, L. Jorgensen, Adsorption of proteins at 611
the oil/water interface—Observation of protein adsorption by interfacial shear stress 612
measurements, Colloids Surfaces B Biointerfaces. 79 (2010) 41–46. 613
https://doi.org/10.1016/J.COLSURFB.2010.03.020. 614
[13] H. Schestkowa, S. Drusch, A.M. Wagemans, FTIR analysis of β-lactoglobulin at the 615
oil/water-interface, Food Chem. 302 (2020) 125349. 616
https://doi.org/10.1016/j.foodchem.2019.125349. 617
[14] H. Schestkowa, T. Wollborn, A. Westphal, A.M. Wagemans, U. Fritsching, S. Drusch, 618
Conformational state and charge determine the interfacial stabilization process of beta-619
lactoglobulin at preoccupied interfaces, J. Colloid Interface Sci. 536 (2019) 300–309. 620
https://doi.org/10.1016/j.jcis.2018.10.043. 621
[15] H. Kieserling, A. Pankow, J.K. Keppler, A.M. Wagemans, S. Drusch, Conformational 622
state and charge determine the interfacial film formation and film stability of 623
\textgreek{b}-lactoglobulin, Food Hydrocoll. 114 (2021) 106561. 624
https://doi.org/10.1016/j.foodhyd.2020.106561. 625
[16] H. Kieserling, P. Giefer, M.J. Uttinger, V. Lautenbacher, T. Nguyen, R. Sevenich, C. 626
Lübebrt, C. Rauh, W. Peukert, U. Fritsching, S. Drusch, A.M. Wagemans, Structure and 627
adsorption behavior of hydrostatic high pressure-treated beta-lactoglobulin, Submitt. to 628
J. Colloid Interface Sci. (2021). 629
[17] H. Kieserling, I.M. Alsmeier, A. Steffen-Heins, J.K. Keppler, R. Sevenich, C. Rauh, 630
A.M. Wagemans, S. Drusch, Interfacial film formation and film stability of high 631
hydrostatic pressure-treated β-lactoglobulin, Food Hydrocoll. (2021) 106746. 632
https://doi.org/https://doi.org/10.1016/j.foodhyd.2021.106746. 633
[18] C.S. Hundschell, S. Bäther, S. Drusch, A.M. Wagemans, Osmometric and viscometric 634
study of levan, β-lactoglobulin and their mixtures, Food Hydrocoll. 101 (2020). 635
https://doi.org/10.1016/j.foodhyd.2019.105580. 636
[19] P. Aymard, D. Durand, T. Nicolai, The effect of temperature and ionic strength on the 637
dimerisation of β-lactoglobulin, Int. J. Biol. Macromol. 19 (1996) 213–221. 638
https://doi.org/10.1016/0141-8130(96)01130-0. 639
[20] M. Verheul, J.S. Pedersen, S.P.F.M. Roefs, C.G. de Kruif, Association behavior of native 640
β-lactoglobulin, Biopolymers. 49 (1999) 11–20. https://doi.org/10.1002/(SICI)1097-641
0282(199901)49:1{\textless}11::AID-BIP2{\textgreater}3.0.CO;2-1. 642
[21] J.E. Kinsella, D.M. Whitehead, Proteins in Whey: Chemical, Physical, and Functional 643
Properties, in: J.E. Kinsella (Ed.), Adv. Food Nutr. Res., {Academic Press} and Elsevier, 644
San Diego, 1989: pp. 343–438. https://doi.org/10.1016/S1043-4526(08)60130-8. 645
[22] A.R. Madureira, C.I. Pereira, A.M.P. Gomes, M.E. Pintado, F. Xavier Malcata, Bovine 646
whey proteins -- Overview on their main biological properties, Food Res. Int. 40 (2007) 647
1197–1211. https://doi.org/10.1016/j.foodres.2007.07.005. 648
[23] A. Trentin, S. De Lamo, C. Güell, F. López, M. Ferrando, Protein-stabilized emulsions 649
containing beta-carotene produced by premix membrane emulsification, J. Food Eng. 650
106 (2011) 267–274. https://doi.org/10.1016/J.JFOODENG.2011.03.013. 651
[24] E. Piacentini, E. Drioli, L. Giorno, Membrane emulsification technology: Twenty-five 652
years of inventions and research through patent survey, J. Memb. Sci. 468 (2014) 410–653
422. https://doi.org/10.1016/J.MEMSCI.2014.05.059. 654
[25] P.J.M. Van den Oetelaar, I.M. Mentink, G.J. Brinks, Loss of Peptides and Proteins Upon 655
Sterile Filtration Due to Adsorption to Membrane Filters, 656
Http://Dx.Doi.Org/10.3109/03639048909040198. 15 (2008) 97–106. 657
https://doi.org/10.3109/03639048909040198. 658
[26] C.H. Suelter, M. DeLuca, How to prevent losses of protein by adsorption to glass and 659
plastic, Anal. Biochem. 135 (1983) 112–119. https://doi.org/10.1016/0003-660
2697(83)90738-8. 661
[27] M.A. Mohamed, J. Jaafar, A.F. Ismail, M.H.D. Othman, M.A. Rahman, Fourier 662
Transform Infrared (FTIR) Spectroscopy, Membr. Charact. (2017) 3–29. 663
https://doi.org/10.1016/B978-0-444-63776-5.00001-2. 664
[28] J.C. Te Lin, D.J. Lee, C. Huang, Membrane Fouling Mitigation: Membrane Cleaning, 665
Https://Doi.Org/10.1080/01496391003666940. 45 (2010) 858–872. 666
https://doi.org/10.1080/01496391003666940. 667
[29] M.L. Hair, W. Hertl, Acidity of surface hydroxyl groups, J. Phys. Chem. 74 (1970) 91–668
94. https://doi.org/10.1021/J100696A016/ASSET/J100696A016.FP.PNG_V03. 669
[30] B.M. Lowe, C.K. Skylaris, N.G. Green, Acid-base dissociation mechanisms and 670
energetics at the silica–water interface: An activationless process, J. Colloid Interface 671
Sci. 451 (2015) 231–244. https://doi.org/10.1016/J.JCIS.2015.01.094. 672
[31] C.M. Bryant, D.J. McClements, Molecular basis of protein functionality with special 673
consideration of cold-set gels derived from heat-denatured whey, Trends Food Sci. {\&} 674
Technol. 9 (1998) 143–151. https://doi.org/10.1016/S0924-2244(98)00031-4. 675
[32] Wafer Surface Cleaning, (n.d.). https://www.mksinst.com/n/wafer-surface-cleaning 676
(accessed September 21, 2021). 677
[33] S. Micciulla, Y. Gerelli, R.A. Campbell, E. Schneck, A Versatile Method for the 678
Distance-Dependent Structural Characterization of Interacting Soft Interfaces by 679
Neutron Reflectometry, Langmuir. 34 (2017) 789–800. 680
https://doi.org/10.1021/ACS.LANGMUIR.7B02971. 681
[34] U.M. Elofsson, M.A. Paulsson, T. Arnebrant, Adsorption of β-lactoglobulin A and B: 682
Effects of ionic strength and phosphate ions, Colloids Surfaces B Biointerfaces. 8 (1997) 683
163–169. https://doi.org/10.1016/S0927-7765(96)01319-7. 684
[35] B. Jachimska, S. Świątek, J.I. Loch, K. Lewiński, T. Luxbacher, Adsorption 685
effectiveness of β-lactoglobulin onto gold surface determined by quartz crystal 686
microbalance, Bioelectrochemistry. 121 (2018) 95–104. 687
https://doi.org/10.1016/j.bioelechem.2018.01.010. 688
[36] L. Pérez-Fuentes, C. Drummond, J. Faraudo, D. Bastos-González, Adsorption of milk 689
proteins (β-casein and β-lactoglobulin) and BSA onto hydrophobic surfaces, Materials 690
(Basel). 10 (2017) 1–25. https://doi.org/10.3390/ma10080893. 691
[37] A. Heyse, M.M. Gößmann, P. Giefer, P. Gäbel, U. Fritsching, A.M. Wagemans, S. 692
Drusch, Lipase adsorption during premix-membrane emulsification affects membrane 693
surface properties and structural conformation of lipase, J. Colloid Interface Sci. (2022) 694
under review. 695
[38] S. Habibi, M. Rabiller-Baudry, F. Lopes, F. Bellet, B. Goyeau, M. Rakib, E. Couallier, 696
New insights into the structure of membrane fouling by biomolecules using comparison 697
with isotherms and ATR-FTIR local quantification, Environ. Technol. (United 698
Kingdom). (2020) 1–18. https://doi.org/10.1080/09593330.2020.1783370. 699
[39] B. Narasimhan, S. Mallapragada, M.D. (Marc D.. Porter, Combinatorial materials 700
science, (2007) 233. 701
https://books.google.com/books/about/Combinatorial_Materials_Science.html?hl=de&i702
d=tRdvxlL7mL0C (accessed March 29, 2023). 703
[40] D.L. Sackett, J. Wolff, Nile red as a polarity-sensitive fluorescent probe of hydrophobic 704
protein surfaces, Anal. Biochem. 167 (1987) 228–234. https://doi.org/10.1016/0003-705
2697(87)90157-6. 706
[41] J. Hendriks, T. Gensch, L. Hviid, M.A. van der Horst, K.J. Hellingwerf, J.J. van Thor, 707
Transient Exposure of Hydrophobic Surface in the Photoactive Yellow Protein 708
Monitored with Nile Red, Biophys. J. 82 (2002) 1632–1643. 709
https://doi.org/10.1016/S0006-3495(02)75514-8. 710
[42] S.Y. Kim, A.N. Semyonov, R.J. Twieg, A.L. Horwich, J. Frydman, W.E. Moerner, 711
Probing the sequence of conformationally induced polarity changes in the molecular 712
chaperonin GroEL with fluorescence spectroscopy, J. Phys. Chem. B. 109 (2005) 713
24517–24525. https://doi.org/10.1021/jp0534232. 714
[43] M.J. Abraham, T. Murtola, R. Schulz, S. Páll, J.C. Smith, B. Hess, E. Lindah, 715
GROMACS: High performance molecular simulations through multi-level parallelism 716
from laptops to supercomputers, SoftwareX. 1–2 (2015) 19–25. 717
https://doi.org/10.1016/J.SOFTX.2015.06.001. 718
[44] R. Anandakrishnan, B. Aguilar, A. V. Onufriev, H++ 3.0: automating pK prediction and 719
the preparation of biomolecular structures for atomistic molecular modeling and 720
simulations, Nucleic Acids Res. 40 (2012) W537–W541. 721
https://doi.org/10.1093/NAR/GKS375. 722
[45] D.J. Cole, M.C. Payne, G. Csányi, S.M. Spearing, L.C. Ciacchi, Development of a 723
classical force field for the oxidized Si surface: Application to hydrophilic wafer 724
bonding, J. Chem. Phys. 127 (2007) 204704. https://doi.org/10.1063/1.2799196. 725
[46] A. Butenuth, G. Moras, J. Schneider, M. Koleini, S. Köppen, R. Meißner, L.B. Wright, 726
T.R. Walsh, L.C. Ciacchi, Ab initio derived force-field parameters for molecular 727
dynamics simulations of deprotonated amorphous-SiO2/water interfaces, Phys. Status 728
Solidi. 249 (2012) 292–305. https://doi.org/10.1002/PSSB.201100786. 729
[47] P.M. Dove, C.M. Craven, Surface charge density on silica in alkali and alkaline earth 730
chloride electrolyte solutions, Geochim. Cosmochim. Acta. 69 (2005) 4963–4970. 731
https://doi.org/10.1016/J.GCA.2005.05.006. 732
[48] J. Sonnefeld, Surface charge density on spherical silica particles in aqueous alkali 733
chloride solutions, Colloid Polym. Sci. 1995 27310. 273 (1995) 932–938. 734
https://doi.org/10.1007/BF00660370. 735
[49] L. Martinez, R. Andrade, E.G. Birgin, J.M. Martínez, PACKMOL: A package for 736
building initial configurations for molecular dynamics simulations, J. Comput. Chem. 30 737
(2009) 2157–2164. https://doi.org/10.1002/JCC.21224. 738
[50] B. Hess, H. Bekker, H.J.C. Berendsen, J.G.E.M. Fraaije, LINCS: A Linear Constraint 739
Solver for Molecular Simulations, J Comput Chem. 18 (1997) 14631472. 740
https://doi.org/10.1002/(SICI)1096-987X(199709)18:12. 741
[51] T. Darden, D. York, L. Pedersen, Particle mesh Ewald: An N⋅log(N) method for Ewald 742
sums in large systems, J. Chem. Phys. 98 (1998) 10089. 743
https://doi.org/10.1063/1.464397. 744
[52] L.L.C. Schrödinger, W. DeLano, PyMOL, (n.d.). http://www.pymol.org/pymol. 745
[53] R.M.A.-G. Azzam, N.M. Bashara, Ellipsometry and polarized light, 3rd ed., Elsevier, 746
Amsterdam, 1987. 747
[54] S. Adachi, Model dielectric constants of Si and Ge, Phys. Rev. B. Condens. Matter 748
Mater. Phys. 38 (1988) 12966–12976. https://doi.org/10.1103/physrevb.38.12966. 749
[55] I. Rodriguez-Loureiro, E. Scoppola, L. Bertinetti, A. Barbetta, G. Fragneto, E. Schneck, 750
Neutron reflectometry yields distance-dependent structures of nanometric polymer 751
brushes interacting across water, Soft Matter. 13 (2017) 5767–5777. 752
https://doi.org/10.1039/c7sm01066d. 753
[56] G.G. Gochev, E. Scoppola, R.A. Campbell, B.A. Noskov, R. Miller, E. Schneck, β-754
Lactoglobulin Adsorption Layers at the Water/Air Surface: 3. Neutron Reflectometry 755
Study on the Effect of pH, J. Phys. Chem. B. 123 (2019) 10877–10889. 756
https://doi.org/10.1021/acs.jpcb.9b07733. 757
[57] D.K. Owens, R.C. Wendt, Estimation of the surface free energy of polymers, J. Appl. 758
Polym. Sci. 13 (1969) 1741–1747. https://doi.org/10.1002/app.1969.070130815. 759
[58] D.H. Kaelble, Dispersion-Polar Surface Tension Properties of Organic Solids, J. Adhes. 760
2 (1970) 66–81. https://doi.org/10.1080/0021846708544582. 761
[59] W. Rabel, Einige Aspekte der Benetzungstheorie und ihre Anwendung auf die 762
Untersuchung und Veränderung der Oberflächeneigenschaften von Polymeren, Farbe 763
Und Lack. 77 (1971) 997–1005. 764
[60] W.H. Stein, AMINO ACID COMPOSITION OF β-LACTOGLOBULIN AND 765
BOVINE SERUM ALBUMIN, J. Biol. Chem. 178 (1949) 79–91. 766
https://doi.org/10.1016/s0021-9258(18)56936-2. 767
[61] M.J. Martinez, M.E. Farías, A.M.R. Pilosof, The dynamics of heat gelation of casein 768
glycomacropeptide – β-lactoglobulin mixtures as affected by interactions in the aqueous 769
phase, Int. Dairy J. 20 (2010) 580–588. https://doi.org/10.1016/j.idairyj.2009.11.016. 770
[62] W. Norde, Adsorption of proteins from solution at the solid-liquid interface, Adv. 771
Colloid Interface Sci. 25 (1986) 267–340. https://doi.org/10.1016/0001-8686(86)80012-772
4. 773
[63] R.J. Marsh, R.A.L. Jones, M. Sferrazza, Adsorption and displacement of a globular 774
protein on hydrophilic and hydrophobic surfaces, Colloids Surfaces B Biointerfaces. 23 775
(2002) 31–42. https://doi.org/10.1016/S0927-7765(01)00204-1. 776
[64] A. Kape, B. Ruick, S. Drusch, Characterisation of the work of adhesion of food grade 777
coating materials on a maltodextrin model surface, Chem. Eng. Res. Des. 110 (2016) 778
152–159. https://doi.org/10.1016/j.cherd.2016.01.023. 779
[65] S. Gehrmann, H. Bunjes, Influence of membrane material on the production of colloidal 780
emulsions by premix membrane emulsification, Eur. J. Pharm. Biopharm. 126 (2018) 781
140–148. https://doi.org/10.1016/j.ejpb.2016.11.006. 782
[66] G.T. Vladisavljević, Preparation of microemulsions and nanoemulsions by membrane 783
emulsification, Colloids Surfaces A Physicochem. Eng. Asp. 579 (2019). 784
https://doi.org/10.1016/j.colsurfa.2019.123709. 785
[67] T. Wollborn, M. Michaelis, L.C. Ciacchi, U. Fritsching, Protein conformational changes 786
at the oil/water-interface induced by premix membrane emulsification, J. Colloid 787
Interface Sci. 628 (2022) 72–81. https://doi.org/10.1016/J.JCIS.2022.07.132. 788
[68] H. Kieserling, A. Pankow, J.K. Keppler, A.M. Wagemans, S. Drusch, Conformational 789
state and charge determine the interfacial film formation and film stability of β-790
lactoglobulin, Food Hydrocoll. 114 (2021) 106561. 791
https://doi.org/10.1016/j.foodhyd.2020.106561. 792
[69] G.G. Gochev, V.I. Kovalchuk, E. V. Aksenenko, V.B. Fainerman, R. Miller, β-793
lactoglobulin adsorption layers at thewater/air surface: 5. adsorption isotherm and 794
equation of state revisited, impact of pH, Colloids and Interfaces. 5 (2021). 795
https://doi.org/10.3390/colloids5010014. 796
797
