3.1. Solubility of Resveratrol in Bulk Oils and Protein Aqueous Solutions
The solubility of resveratrol was ranked in the order of peppermint oil > MCT > fish oil > sunflower oil (
Table 1). Fish oil is rich in long-chain unsaturated fatty acids. The content of unsaturated fatty acids is 85–91% in sunflower oil [
19]. MCT is a saturated fatty acid with a carbon chain length of 8–12. Peppermint oil is a mixture of alcohols, ketones, esters and terpenes, with menthone and menthol accounting for more than 60% of the total [
20]. The solubility of resveratrol in the oils is consistent with the oil dielectric constants (
Table 1). It is easier for the more polar groups in the oil phase to induce dipole–dipole interactions between hydroxyl groups on resveratrol and fatty acid polar groups [
21].
The solubility of resveratrol in the aqueous solutions increased as the concentrations of WPI, hWPI or SC increased from 0.5% to 2% (
Figure 1). The solubility of resveratrol in the presence of SC or hWPI was greater than that in the presence of WPI at 0.5%, while the polyphenol solubility in protein solutions ranked in the order of SC > hWPI > WPI at higher concentrations. The increase in the solubilization of resveratrol may be due to the polyphenol binding to proteins. The interaction between proteins and resveratrol was mainly driven by hydrogen bond and hydrophobic interactions, and the polyphenol binding constants with WPI and SC were, respectively, 1.2 × 10
5 M
−1 and (3.7–5.1) × 10
5 M
−1 [
22,
23]. The loading efficiencies of resveratrol in SC particles were reportedly greater than those in WPI particles when the protein concentration was 1% [
24]. Thermal denaturation caused exposure of more hydrophobic residues, improving the affinity of resveratrol to β-lactoglobulin, a major whey protein [
25].
3.2. Interfacial Protein in Emulsions
The interfacial protein percentage decreased as the concentration of WPI (
Figure 2A), hWPI (
Figure 2B) or SC (
Figure 2C) increased from 0.5% to 2% in emulsion, which was not significant for sunflower oil emulsions with 1% and 2% proteins and MCT emulsions with 0.5% and 1% proteins. The interfacial percentages of WPI, hWPI and SC at the same concentration were similar in fish oil and peppermint oil emulsions (
Figure 2). In sunflower oil emulsions, the interfacial percentages of WPI, hWPI and SC were similar at 0.5%, while the interfacial percentages of SC (
Figure 2C) were greater than those of WPI (
Figure 2A) and hWPI (
Figure 2B) at higher protein concentrations. These results are consistent with 20% soya oil emulsions stabilized by WPI and caseinate at pH 7, where WPI and caseinate adsorbed to the oil–water interface at the same extent at low concentrations, but caseinate adsorbed in preference to WPI with an excess of proteins [
26]. In MCT emulsions, the interfacial percentages of SC (
Figure 2C) were greater than those of WPI (
Figure 2A) and hWPI (
Figure 2B) at 0.5–2%. The protein layer at the oil–water interface is in a dynamic equilibrium which could be affected by the structure and intermolecular interaction of proteins [
27]. The protein layer will undergo reversible collapse when the amount of protein exceeds the maximum molecule density, and the reform of the interfacial membrane was driven by the attraction force of protein at the oil–water interface [
28]. The peppermint oil emulsions with 2% proteins were not analyzed, since they separated into the creaming layer and the aqueous phase upon preparation with SC and after 2 days with WPI and hWPI.
Oil type affects the adsorption of proteins at the oil–water interface (
Figure 2). When the protein content was 0.5%, the interfacial percentages of WPI or SC in fish oil and sunflower oil emulsions were greater than those in MCT and peppermint oil emulsions (
Figure 2A,C), while the percentage of hWPI decreased in the sequence of fish oil, sunflower oil, peppermint oil and MCT emulsions (
Figure 2B). At 1% proteins, the interfacial percentages of WPI, hWPI or SC ranked in the order of fish oil > sunflower oil ~ MCT > Peppermint oil (
Figure 2). In the case of 2% proteins, the percentage of WPI in fish oil emulsions was greater than those in sunflower oil and MCT emulsions (
Figure 2A); the percentage of hWPI ranked in the order of fish oil > sunflower oil > MCT (
Figure 2B), and the percentages of SC in fish oil and sunflower oil emulsions were greater than that in MCT emulsions (
Figure 2C). The polarity of four oils was ranked in the order of peppermint oil > MCT > fish oil ~ sunflower oil (
Table 1). It has been reported that protein adsorption was slower on the surface of more polar oils. The expansion and adsorption of proteins were higher on the surface of hydrophobic oils, while proteins were adsorbed in a random orientation at a polar oil–water interface with lower interfacial tensions [
29].
In 50% walnut oil emulsions stabilized by 4% SC, the loading content of SC at the surface of oil droplets was not affected by 2 mM resveratrol but improved by 4 and 6 mM resveratrol [
11]. In this study, the concentration of resveratrol is 130 μg/mL (~0.6 mM). The interfacial percentages of WPI, hWPI and SC were not basically affected by addition of resveratrol at the protein concentration of 1% (
Figure S2).
3.5. Mechanism of Resveratrol Partition in Emulsions
In the emulsions stabilized by low-molecular-weight surfactants, there was a basic assumption that antioxidants distribute among the oil phase, the aqueous phase and the interfacial region according to their solubilities in each region [
35]. The transfer of hydrophilic caffeic acid or catechin from the aqueous phase to the interfacial region was reportedly spontaneous in corn oil emulsions stabilized by Tween 20, when 4-hexadecylbenzenediazonium was used as a chemical probe in the interfacial region [
36,
37]. It was found that more than 85% of resveratrol located in the interface, and a small fraction in the oil and aqueous regions of corn oil emulsions were stabilized by Tween 20 [
38]. In comparison, the interfacial percentages of resveratrol were less in protein-stabilized emulsions (
Figure 5).
Multiple linear regression analysis was thus performed to clarify the combined effect of resveratrol solubility in bulk oil (
Table 1) and in the aqueous solution of proteins (
Figure 1), and protein partition (
Figure 2) on resveratrol partition (
Figure 4,
Figure 5 and
Figure 6) in emulsions. The greater the absolute value of the standardized regression coefficient (β), the stronger the dependence on the variables [
39]. In fish oil, sunflower oil, MCT and peppermint oil emulsions, there is a negative correlation (
p < 0.01) between the aqueous percentage of resveratrol with the percentage of interfacial protein (
Pi) and the solubility of resveratrol in bulk oils (
Ro,
Table 2). The β value of
Ro is greater than that of
Pi, suggesting that the polyphenol solubility in the oil phase is more important for removing resveratrol in the aqueous phase. The aqueous percentages of resveratrol can be calculated using the optimized Equation (5), where 80.3% of the variability could be accounted for by
Ro and
Pi. There was a good correlation between the predicted value by Equation (5) and the experienced value (
Figure S5A).
In fish oil, sunflower oil, MCT and peppermint oil emulsions, there is a positive correlation (
p < 0.01) between the oily percentage of resveratrol with
Pi,
Ro and protein concentration in emulsions (
Pt) in
Table 3. According to the β values, the importance of the variables was ranked in the order of
Ro >
Pi >
Pt. The oily percentages of resveratrol can be calculated using the optimized Equation (6), where 77.9% of the variability could be accounted for by
Ro,
Pi and
Pt. The content of proteins at the oil–water interface can be calculated by multiplying
Pi and
Pt. It is thus suggested that the accessibility of protein to the oil–water interface contributes to the transfer of resveratrol from the aqueous phase into the inner oil phase. When fish oil emulsions were excluded, the correlation between the predicted and experienced values (
Figure S5B,D) was improved. It can be seen that the oily percentage of resveratrol is only correlative to Ro (
Table 4). The oily percentages of resveratrol in sunflower oil, MCT and peppermint oil emulsions can be calculated using the optimized Equation (7), where 91.2% of the variability could be accounted for by
Ro. These results suggest that resveratrol solubility in the oil phase drives its transfer from the aqueous phase into the phase of sunflower oil, MCT and peppermint oil in emulsions, while the combination of resveratrol solubility in the oil phase with interfacial protein contributes to the polyphenol transfer from the aqueous phase into the phase of fish oil in emulsions.
The oil–water interface has higher interfacial stress in apolar than polar oils, provoking stronger hydrophobic interactions between the oil components and hydrophobic residues of proteins [
40]. The polarity of fish oil is the lowest of all oils (
Table 1), resulting in a stronger hydrophobic interface and greater adsorption of proteins (
Figure 2). The adsorbed protein at the oil–water interface improves the accessibility of protein-loaded resveratrol to the oil phase, contributing to the polyphenol transfer into the inner oil phase. Therefore, the percentages of resveratrol in the fish oil phase were greater than those in the MCT phase of emulsions (
Figure 6), although the solubility of resveratrol in MCT was greater than that in fish oil (
Table 1). Moreover, the greater loading of resveratrol by SC than by WPI and hWPI (
Figure 1) corresponded to greater transfer of resveratrol from the aqueous phase into the oil phase of peppermint oil emulsions (
Figure 4 and
Figure 6).
In fish oil, sunflower oil, MCT and peppermint oil emulsions, there is a positive correlation regarding the interfacial percentage of resveratrol with
Pi and
Ro, but a negative correlation between the interfacial percentage of resveratrol with protein concentration in emulsions (
Table 5). Protein concentration in emulsions and
Pi could be considered as one variable since they had a significant negative correlation (
Table S1). The interfacial percentage of resveratrol can be calculated using the optimized Equation (8), where R
2 indicates that 52.1% of the variabilities could be accounted for by
Pi and
Ro (
Table 5). The correlation between the predicted and experienced values was improved when sunflower oil emulsions were excluded (
Figure S5C,E). The interfacial percentages of resveratrol can be calculated using the optimized Equation (9), where 71.5% of the variability could be accounted for by
Ro and
Pi (
Table 6).
Ro and
Pi have close β value values, suggesting both factors are important for the interfacial partition of resveratrol in emulsions.
The complexation with resveratrol had no impact on the adsorption of proteins at the interface (
Figure S2). Additionally, the interfacial resveratrol (
Figure 5) is greater than the interfacial proteins (
Figure 3), of which the difference increased as the polyphenol solubility in bulk oils (
Table 1) increased. These results suggest that the transfer of resveratrol from the aqueous phase into the oil phase improves the polyphenol accumulation in the protein membrane at the oil surface. Therefore, there is about 50% resveratrol at the interface of peppermint oil emulsions (
Figure 5). Although the polyphenol solubility in MCT was greater than that in sunflower oil (
Table 1), the oily percentages of resveratrol in MCT emulsions are similar to those in sunflower oil emulsions (
Figure 6). It is suggested that the transfer of resveratrol from the aqueous phase into the MCT phase was withheld by the interfacial proteins, due to the complexation or encapsulation by proteins. When curcumin was added from the oily phase at the polyphenol concentration below its solubility in MCT, β-lactoglobulin at the oil–water interface of emulsions had a better capability of lowering the interfacial tension compared with protein alone, suggesting that the curcumin could accumulate at the protein layer at the interface [
41]. Moreover, the curcumin transfer to the oil–water interface was reported to form the polyphenol–protein complex in the soybean oil emulsion stabilized by WPI [
42]. Therefore, the oil–water interface provides the microenvironment for the enrichment of resveratrol by proteins (
Figure 5).