3.1. Effect of Dephenolization on Protein Purity, Protein Recovery, and Amino Acid Profile
Proximate composition analyses were conducted on hazelnut meal samples. Protein, fat, ash, moisture, and total calculated carbohydrates of the hazelnut meal were 32.1 ± 1.7, 26.7 ± 1.3, 5.3 ± 0.01, 4.1 ± 0.04, and 31.8 ± 0.4%, respectively.
Dephenolization of hazelnut meal before protein isolation affected both protein recovery and the purity of the protein isolates. Indeed, dephenolization treatment led to a much higher purity of the isolates (98.4 ± 1.0%) than the non-dephenolized sample (78.7 ± 1.3%). However, the protein recovery was significantly lower in the dephenolized sample than that of the non-dephenolized sample (45.6 ± 4.5 vs. 76.7 ± 6.1%), which we assumed to be due to the protein loss during dephenolization treatment (there was a significant decrease in protein content after dephenolization) (
Table 1). The amino acid profile of hazelnut meal protein isolate is shown in
Table 2.
3.2. Average Particle Size, Size Distribution, and ζ-Charge
The average particle size, polydispersity index (PDI), and ζ-potential of the protein isolates with or without the complexation/presence with/of phenolics are shown in
Figure 2. The dephenolization of the protein isolates reduced their average size from 164.2 ± 2.9 nm (HPI) to 73.9 ± 5.6 nm (dHPI). The addition of HSE to dHPI at 0.05 mM resulted in particles of similar size to HPI (190.4 ± 2.9 nm,
p-value = 0.8911). Interestingly, the size increased tremendously with the increase of HSE concentrations (1840.7 ± 102.3; 2908.3 ± 45.6 and 7807.3 ± 570.6 nm for dHPI+HSE: 0.125; 0.25; and 0.5 mM, respectively), while dHPI–C resulted in complexes with a size similar to HPI at all tested concentrations (
p-values > 0.05). HPI showed a PDI of 0.4 ± 0.01, which increased upon dephenolization (dHPI) to 0.64 ± 0.1 and then decreased to the initial PDI after the addition of catechin (dHPI–C) at all concentrations (
p-values > 0.05) (
Figure 2A). The lowest PDI was reached by dHPI+HSE at the concentrations of 0.25 and 0.5 mM HSE, at 0.2 ± 0.0 and 0.2 ± 0.0, respectively.
All samples showed net negative surface charges at pH 7.4, as indicated by their ζ-potentials, with no significant differences (avg. ≈ −9.84 mV) except for dHPI+HSE (0.5) which showed higher absolute charge than the other samples (−12.6 ± 1.2 mV) (
Figure 2B).
As HSE concentration increases, phenolics may function as bridging agents between protein molecules, leading to the creation of greater protein–phenolic complexes [
6,
40]. However, catechin did not increase the size of HPI with the same significance as HSE, even though it was the main compound in HSE. This could be ascribed to the presence of other phenolics with larger molecular sizes than catechin in the extract (e.g., GCG, EGC, EGCG, phlorizin, Q3R, QUE, as shown in
Section 3.6.2 “HPLC-DAD Analysis of Phenolic Compounds”). Indeed, it has been reported that only phenolic compounds with sufficient size are able to interact with more than one site on the proteins, potentially forming cross-links between distant proteins, a process called the “
multidentate” mechanism of protein–phenolic interaction that leads to protein aggregation [
41]. In contrast, catechin, a relatively small molecule, would form a layer around the proteins without being able to link between them via the “
monodentate” mechanism. Similarly, Dai et al. reported that the saturating level of SPI and catechin binding occurred when the catechin concentration increased over a specific point [
42]. Since there was more catechin than SPI binding sites could accommodate, the complex particle size was not further increased.
The precipitation of these samples evidenced the aggregation of HPI+HSE samples after storage (
Figure S1 in supplementary data). It is quite predictable for proteins to have a negative net charge in a solution with pH (7.4) above their isoelectric point (pI = 4.5) [
43]. The slight increase in the surface charge of dHPI+HSE at the highest concentration of skin extracts could be inferred by this principle later due to the eventual saturation of protein particles with phenolics [
43]. Proteins and peptides need to have relatively high absolute ζ-potential in order to avoid their natural tendency to aggregate in the aqueous environment [
44]. This, together with the DLS results of the dHPI alone, strengthens the assumption that dHPI+HSE aggregated because of the presence of phenolics, not due to the pH of the medium or an eventual low net charge of the proteins. Bulkier phenolics in the extract had more bonding points (-OH groups) and were in galloylated form, resulting in more intense interactions [
10], which may have caused some charged groups (amino acids) to become buried inside the protein such that these groups could no longer contribute to surface charge.
3.3. Fluorescence Quenching
Changes in the intrinsic fluorescence intensities (FI) of dHPI in the presence of an increasing concentration of HSE are shown in
Figure 3.
A broad and slightly shouldered peak was observed in the hazelnut protein itself, independent of phenol. This type of fluorescence intensity peak indicates the presence of a high amount of tyrosine (Tyr) amino acid in addition to the tryptophan (Trp). When the amino acid composition of the HPI was controlled, the concentration of Tyr and Trp were calculated as around 19.9 mg/g-protein and 25.3 mg/g-protein, respectively.
The fluorescence emission of the proteins is dominated by Trp, which absorbs at the longest wavelength. In the presence of Trp, although there were phenylalanine (Phen) and Tyr amino acids in the protein, the energy they absorbed was mainly transferred to Trp. Protein fluorescence is generally excited at 280 nm, but Phen displays a structured emission with a maximum near 282 nm. Therefore, Phen, having a very small quantum yield, was not excited as in this study. The emission maximum of Tyr and Trp in water occurs at 303 nm and 350 nm, respectively. Thus, in
Figure 3, the observed emission peaks were due to the absorption of both Tyr and Trp at 280 nm. On the other hand, resonance energy transfers repeatedly occur from Tyr to Trp, so only a minor contribution of Try to the emission of most proteins can be observed.
To display the emission of Trp alone, the fluorescence intensity of the same samples was also excited at 295 nm, where the absorption is primarily due to Trp and no excitation of Tyr exists. In
Figure 4, the fluorescence intensity of dHPI obtained at λ
ex = 295 nm was much lower than 280 because only the Trp residues were excited at this wavelength, having a a lower quantum yield. This decreased fluorescence intensity at 295 nm was another indicator of the presence of Tyr in terms of the contribution of Tyr to the higher fluorescence intensity at 280 nm.
There was a continuous decrease in the fluorescence intensity of HPI when the phenol concentration increased from zero to 0.5 mM at all temperature conditions. This means that a fluorescence quenching of hazelnut proteins was induced by phenolic extracts. Fluorescence quenching is the decrease of the quantum yield of fluorescence from a fluorophore and is induced by a variety of molecular interactions with quencher molecules [
20]. For quenching, molecular contact is required between the fluorophore and the quencher. Due to molecular interaction or the change in the solvent composition, the excitation and emission behavior of fluorophores can be changed. For instance, Tyr is relatively insensitive to solvent polarity, while Trp is highly dependent upon polarity and/or the local environment [
21] due to its unique complexity in having two nearby isoenergetic transitions. In contrast, emission from tyrosine appears to occur from a single electronic state.
The quenching ability of HSE (0–0.5 mM) on the fluorescence emission of native hazelnut protein was investigated at different temperatures (298, 308, 318 K) (
Figure 3). The intensity of the fluorescence dropped considerably as the HSE concentration was increased at all temperatures.
When the effect of temperature on the changes in fluorescence intensity was compared in the presence of phenol extracts, no significant change was observed between 298 and 308 K (
Figure 3). However, the further increase in temperature to 318 K resulted in a rapid decrease in protein fluorescence intensity, indicating that the hazelnut protein was thermally unstable at this temperature.
Although the phenol induced a decrease in fluorescence intensity, λ
max for HPI was measured at 340 nm, which is the expected λ
max for the indole group of Trp alone (
Figure 4). Moreover, linear Stern–Volmer curves were observed at all temperature conditions (
Figure 5a). This usually means that all main fluorescing residues of the protein were equally accessible to the solvent. If two fluorophore populations are present, and one class is not accessible to the quencher, then the Stern–Volmer plots will deviate from linearity toward the x-axis.
The fraction of accessible fluorescence was calculated from Equation (9):
F0/ΔF was found linear to 1/[Q] so that the slope gave 1/(fa Ka), and from the intercept to ordinate, 1/fa was found. Then, the effective quenching constant for the accessible fluorophore (Ka) was calculated as the ratio of 1/fa to 1/(fa Ka). The linear lines intersected the y-axis at a value of around 1, meaning that all the fluorescence was due to quenchable Trp. These data, therefore, conform to the assumptions from the Stern–Volmer curves that all the main fluorescing tryptophan residues were quenchable and no others buried within the protein or unavailable to the quencher are expected.
3.3.1. Stern–Volmer Plots
Stern–Volmer plots were obtained in
Figure 5a by plotting the
F and
F0 values versus the phenolic extract concentration
[Q] at different temperatures. The linearity of the Stern–Volmer plot indicates that there is only one quenching mechanism, which can be static or dynamic. The maximum dynamic bimolecular quenching constant (
Kq) is 2 × 10
10 M
−1 s
−1, while the major quenching process is static if
Kq values are higher than this value [
45,
46].
Kq values of dHPI+HSE complexes were calculated as 34.716, 26.548, and 27.001 at 298, 308, and 318 K, respectively, indicating that the major quenching mechanism was static.
A non-fluorescent ground-state complex is formed between the fluorophore and quencher via static quenching [
21], while the Stern–Volmer constant (
Ksv) is lower at higher temperatures. The decrease in
Ksv values as the temperature increases is observed in the case of the static quenching mechanism [
22,
47] due to its negative effect on the stability of complex formation [
23,
48,
49].
Relatively high values were obtained from the
Ksv values determined for β-lactoglobulin/α-casein/β-casein–catechin/derivatives complexes in the literature [
50,
51]. A fluorophore buried in a macromolecule is usually inaccessible to water-soluble quenchers, so the value of
K is low. Larger values of
K are found if the fluorophore is free in solution or on the surface of a biomolecule [
21], and it is known that β-lactoglobulin has one Trp residue on the surface of the protein molecule and is accessible to the solvent [
52]. Moreover, it has been reported in some studies that galloylated forms have higher binding affinity compared to catechin [
50,
51]. Relatively larger phenolic compounds provide two or more aromatic rings and hydroxyl groups for interactions. It was also reported that phenols with galloylated monomers (epigallocatechin gallate, EGCG) have a higher affinity for whey protein or β-lactoglobulin binding than non-gallate ones (gallic acid, chlorogenic acid, ferulic acid) [
23,
45]. The molecular weight of phenols, their structural flexibility, and the number of OH groups in their structures have greatly affected the formation of multiple bonds, as well as an increase in hydrophobicity with size [
10]. Many of the phenolics found in HSE were in galloylated or glycosylated form, as well as a mixture of these. This explains the high binding affinity found in this study.
The number of binding sites (
n) was calculated from the double logarithm regression curve (
Figure 5b). The
n values at all temperature conditions were greater than 1 as a result of intermolecular cross-linking between proteins at all temperatures, indicating that at least one phenol molecule independently interacted with the protein, while the other phenolic compound may have formed a bridge between the proteins. In addition, the decrease in
Ka values with temperature indicates that the protein–phenolic interaction is exothermic, and the stability of the complex decreases [
53].
3.3.2. Thermodynamic Parameters
Thermodynamic parameters were calculated from Van’t Hoff plot (
Figure 5c) and are shown in
Table 3. Since ΔH < 0 and ΔS < 0, hydrogen bonding and van der Waals forces were the main interaction mechanism between the dHPI and HSE. Since ΔH > ΔS, dHPI+HSE interaction was induced primarily by enthalpy [
54]. Negative entropy (ΔS) resulted in an unfavorable increase in molecular order upon complex formation. The decrease in conformational mobility in the protein, as well as the exposure of hydrophobic surfaces to the solvent, both can contribute to unfavorable entropy [
55]. Negative enthalpy (ΔH) implies that the enthalpy of the protein–phenolic complex is lower than that of the reactants (protein and phenolics), which relates to (1) the increase in bond enthalpy, the number of the bonds, and bond strength, and (2) immobilization of the sufficiently counteracted ligand by liberation of bound water [
56]. Moreover, hydrogen bonding is an exothermic event [
55].
Favorable enthalpy contributions overcome unfavorable entropic contributions, and the reaction depends on temperature. Ultimately, molecular interactions must be strong enough due to complex formation to cover the loss of entropy. As temperature rises, hydrophobic interactions grow more powerful, and hydrogen bonds become weaker. A reduction in the binding constant with rising temperature implicates H-bonding as the main force [
53].
ΔG for all dHPI–HSE complexation was negative, and the interaction was spontaneous. Generally, protein–phenolic interactions were reported as spontaneous reactions (ΔG < 0). On the other hand, mainly hydrogen bonding and van der Waals forces were responsible for catechin interactions with human serum albumin [
56] and bovine serum albumin [
57] in the literature.
3.4. Fourier Transform Infrared (FTIR) Spectroscopy
To further investigate whether any structural change of dHPI was associated with phenolic extract interactions, FTIR spectra of dHPIs were recorded with and without the addition of HSE (0.05, 0.125, and 0.25 mM) and catechin (0.125 mM). There are basic FTIR spectrum regions specific to each sample: (1) functional group regions between 1500 and 4000 cm
−1 and (2) fingerprint regions between 400 and 1500 cm
−1. The fingerprint region is unique for each protein. The characteristics of the side chains, the specifics of the force field, and hydrogen bonding all play a role in the complexity of the fingerprint region bands. In this region, a peak was observed at 1462 cm
−1 originating from CH
2 bending vibrations and 1396 cm
−1 originating from CH
3 bending vibrations [
58].
Nine distinctive IR bands are produced by the amide groups of the protein backbone (amides A and B and amides I–VII) about the secondary structure and conformation of the protein backbone. The two most noticeable vibrational bands of the protein backbone are amides I (1600–1700 cm
−1) and II (1500–1600 cm
−1) [
59]. The absorbance of dHPI had a characteristic amide I band at 1631 cm
−1 (C=O stretching vibrations of peptide linkages), amide II at 1522 cm
−1 (C-N stretching, 40% and N-H bending, 60% of amino acids), a wide amide A band at 3273 cm
−1 (tensile vibration of intermolecular hydrogen bonding between O–H and N–H stretching occurring in the hydrogen bonds and intermolecular H bonding), and amide B at 2922 cm
−1 (CH
2 asymmetric stretch) and 2855 cm
−1 (CH
2 symmetric stretch) [
42,
58] (
Figure 6).
No new bands were formed in dHPI+HSE or dHPI+C complex formation. Thus, the protein–phenolic complex formation was mostly a physical process, and no covalent bonds were created between the matrix and the core [
60]. Further, dHPI+HSE or dHPI+C complexes caused differences in all amide bands of native hazelnut protein. The amide A band of dHPI+HSE complexes was slightly shifted to 3276 cm
−1, while dHPI+C complex had no shifts. The amide B band is attributable to the C–H tensile vibration of the CH
3 and CH
2 groups of protein [
42]. The amide B band of all dHPI–phenolic complexes was shifted to 2925 cm
−1.
Amide I and II bands were red-shifted from 1631/1522 to 1634/1525, 1637/1524, 1637/1524, and 1635/1525 cm
−1 for dHPI+HSE (0.05), dHPI+HSE (0.125), dHPI+HSE (0.25), or dHPI+C (0.125) complexes, respectively. Hence, the secondary structure of hazelnut proteins was altered upon extract or catechin complexation (
Figure 6). Any change in the intensity of the amide I and II bands was related to the protein C=O, C–N, and N–H groups with hydrogen and hydrophobic interactions [
50].
The secondary structure content of samples is given in
Table 4. The distinct decrease in the β-turn structure of dHPI+HSE (0.05) complex was supported by the transition to α-helix as with the dHPI+C (0.125) complex. A significant increase in the random coil was observed at the higher phenolic concentration compared to the native dHPI. There was a slight decrease in the α-helix and β-sheet structure of dHPI+HSE (0.25), although the number of random coils was increased. In the presence of phenolic at the lowest concentration (0.25 mM), regular structures (α-helix and β-sheet) increased, and the protein became more stable and rigid. The sum of the regular structures of the native protein was interrupted, and a more unfolded protein was formed via protein–phenolic interaction at medium to higher phenolic extract concentrations.
The protein/phenol ratio in the medium is of great importance for the protein–phenolic interaction. At low phenolic concentration (multisite ligand), the interaction regions of the proteins do not reach saturation (
multidentate mechanism), and phenolics must be large enough to form bridges between proteins. In these conditions, the irregular structure gradually increased in the presence of phenolic extract at 0.05- and 0.125-mM concentration via the
multidentate mechanism. As the phenolic concentration increased, several phenolics interact with one protein (
monodentate mechanism) [
41]. Most likely, the
monodentate mechanism was seen at higher HSE concentrations because it was very close to the saturation point of the protein–phenolic interaction. Eventually, their regular structure of the protein was altered upon phenolic binding. Similar to this study, changes in the amide I band density of complexes with α-/β-casein or β-lactoglobulin and tea phenolic (catechin, epicatechin, epigallocatechin, and epigallocatechin gallate) [
50,
51] and rice bran protein/catechin complexes [
61] were observed depending on the phenolic concentration, and the amide I band intensity increased at low phenol concentration and decreased as the concentration increased. Moreover, it was also reported that conformational changes were more noticeable for epigallocatechin and epigallocatechin gallate compared to catechin and epicatechin, since the bulkier and larger phenolics provided more perturbation [
50,
51].
3.5. Effect of Phenolics on the Digestibility of Hazelnut Proteins
The rate of hydrolysis of hazelnut proteins by pancreatin was fast during the first hour, then gradually decreased to become almost stagnant between 105 and 120 min, a typical proteolysis trend (
Figure 7). Non-dephenolized proteins showed a higher digestibility in comparison with all dephenolized samples for all reaction times (
p < 0.05), reaching a maximum of 32.6 ± 1.1% after 2 h. Within the dephenolized samples, the degree of hydrolysis was slightly reduced by the presence of HSE or catechin but progressively retrieved with time until no significant difference was observed after 120 min (DH ≈ 25%). Interestingly, the addition of HSE or catechin to the dephenolized hazelnut proteins did not restore the degree of hydrolysis achieved by non-dephenolized proteins.
Although phenolics may have mostly a negative effect on the digestibility of proteins, data in the literature showed a quite unpredictable effect of phenolic compounds on the enzymatic digestibility of proteins [
62]. Ni et al. showed a positive impact of salal fruit phenolic extracts on the degree of hydrolysis of whey protein by Flavourzyme
® [
63]. On the contrary, Wang and Tang reported an enhancement in the degree of hydrolysis, by trypsin, of buckwheat protein isolates upon dephenolization [
64]. The enhancement of hydrolysis by the presence of phenolics may be ascribed to an eventual partial unfolding of the proteins caused by phenolics, facilitating the access of the protease to catalytic sites in non-dephenolized hazelnut protein isolates, which may have enhanced their hydrolysis [
63]. This hypothesis is also supported by the decrease in the size of the dHP sample (
Figure 2), which could be due to an eventual folding upon dephenolization, making some catalytic sites inaccessible inside the three-dimensional structure of the protein. The fact that the addition of phenolics did not improve the digestibility of the proteins implies that whatever happened upon dephenolization was irreversible. Finally, the recovery of the degree of hydrolysis in dHPI+HSE and dHPI+C after 120 min in comparison with dHPI supports the fact that the interactions between dHPI and phenolic compounds were non-covalent and thus reversible, which is in total correlation with the fluorescence quenching study (
Section 3.2).
3.6. Simulated In Vitro Gastrointestinal Digestion
To determine the effect of in vitro gastrointestinal digestion on the phenolic extract and protein–phenolic solutions, total phenolic content and total antioxidant capacities were measured, and individual phenolic compounds were identified for each gastrointestinal digestion phase. The results vary in both extract and protein–phenolic solutions at each stage of in vitro gastrointestinal digestion due to their solubility at acidic or alkaline pH and interactions with enzymes.
3.6.1. Spectrophotometric Analyses
The changes in the total phenolic contents (TPC) of samples and recovery indexes (RI) of phenolic compounds at each digestion phase are shown in
Table 5. At the initial phase, the TPC of phenolic solution systems was determined to be in the following order: extract solutions > protein–phenolic solutions. The HSE solutions followed a similar trend as the initial phase in the gastric phase, whereas the TPCs of the c solutions were found as protein–phenolic solution extract solutions. The TPCs of the extract solutions were higher in the intestinal phase than the TPCs of the protein–phenolic solutions, just as they were in the initial phase.
TPCs of C and dHPI+C showed a significant increase in the intestinal phase compared with the initial phase, following a decreasing trend after gastric digestion. Similar trends were observed in the gastric and intestinal phases of HSE and dHPI+HSE samples, apart from the intestinal phase of HSE. After intestinal digestion, 136.1% and 78.2% of the phenolic compounds were present in the intestinal fraction for C and HSE extract solutions, respectively. However, 98.4% and 68.4% of these compounds were available in the dHPI+C and dHPI+HSE solutions, respectively.
In order to determine the effect of in vitro gastrointestinal digestion on total antioxidant capacity, DPPH and CUPRAC assays were performed. The total antioxidant capacities of initial solutions and gastric and intestinal phases are given in
Table 5. The total antioxidant capacities of C and HSE extract solutions throughout the digestive tract varied between 94.0 ± 5.9 to 143.0 ± 0.5 mg TE/g extract and 128.7 ± 0.4 to 181.0 ± 11.9 mg TE/g extract, respectively, whereas the total antioxidant capacities of dHPI+C and dHPI+HSE protein–phenolic solutions ranged from 128.7 ± 0.4 to 181.0 ± 11.9 mg TE/g extract and 62.9 ± 7.0 and 106.8 ± 10.6 mg TE/g extract, respectively, throughout the gastrointestinal tract, according to the DPPH method. The DPPH values of the extract solutions decreased from the initial phase to the intestinal phase. In contrast, the DPPH values of the protein–phenolic solutions increased in the intestinal phase following the decrease in the gastric phase.
The total antioxidant capacities of C and HSE extract solutions across the gastrointestinal system changed from 1067.6 ± 23.0 to 2285.6 ± 90.5 mg TE/g extract and 321.2 ± 22.7 to 705.0 ± 8.9 mg TE/g extract, respectively; on the other hand, the total antioxidant capacities of dHPI+C and dHPI+HSE protein–phenolic solutions varied between 1849.3 ± 35.7 to 2175.0 ± 195.2 mg TE/g extract and 364.5 ± 19.4 and 763.3 ± 8.1 mg TE/g extract, respectively, throughout the digestive tract, according to the CUPRAC method. The CUPRAC values of the extract solutions decreased from the initial phase to the intestinal phase (p < 0.05). In contrast, the DPPH values of the protein–phenolic solutions increased in the intestinal phase following the reduction in the gastric phase (p < 0.05). In addition, the CUPRAC values of the extract solutions showed a significant increase in the intestinal phase compared with the initial phase, following a decreasing trend after gastric digestion (p < 0.05). Similar trends were found in the dHPI+HSE protein–phenolic solution (p < 0.05).
3.6.2. HPLC-DAD Analysis of Phenolic Compounds
The major phenolic compounds of initial and digested extract solutions and protein–phenolic solutions were detected by HPLC-DAD analysis. Comparisons of the phenolic profiles of samples are shown in
Table 6. Up to 10 individual phenolic compounds were identified in the samples. Since the phenolic compounds identified in the HSE are mainly composed of catechins, the catechin standard was used as the control group in this study.
In the analysed HSE initial extract solutions, gallic acid (GA) and protocatechuic acid (PCA) were detected as the hydroxybenzoic acids; gallocatechin gallate (GCG), (-)-epigallocatechin (EGC), catechin (C), epicatechin (EC), and (-)-epigallocatechin gallate (EGCG) were identified as flavanols; quercetin 3-O-rhamnoside (Q3R) and quercetin (QUE) were detected as flavonols, and phlorizin was identified as chalcone. On the other hand, no phenolic compounds were detected in the initial phases of protein–phenolic solutions of both dHPI+C and dHPI+HSE. A lower amount of catechin was observed in the C solution’s gastric phase compared to the initial amount (p < 0.05). Moreover, the amount of C that reappeared in the gastric phase of the dHPI+C solution was significantly lower than the amount determined in the gastric phase of the C solution (p < 0.05). The amounts of detected phenolic compounds other than GA and GCG decreased in the HSE solutions’ gastric phase compared to their initial levels (p < 0.05).
Furthermore, QUE was found to be below the limit of detection for the gastric and intestinal phases of HSE solutions. The phenolic compounds identified in the HSE solution, except for QUE, were re-detected in the gastric phase of the dHPI+HSE solution. There were significant reductions in the quantities of the phenolic compounds, except for PCA and C, between the gastric phases of HSE and dHPI+HSE (p < 0.05). C was metabolized with digestion, and GCG, EGC, C, and EC were detected in the intestinal phases of C and dHPI+C solutions, whereas EGCG was only determined in the C solution. GC and EC were significantly higher in the intestinal phases of the dHPI+C solution than in the C solution (p < 0.05). On the other hand, GA, GCG, EGC, EC and EGCG, and Q3R had better accessibility in the intestinal phase of the dHPI+HSE solution than in the intestinal phase of the HSE solution after digestion (p < 0.05).
To the best of our knowledge, there is no study investigating the changes of phenolic compounds of hazelnut skin during in-vitro gastrointestinal digestion in the literature. In contrast, many studies examined their phenolic profile and antioxidant activities. However, since tea and green tea principally contain high amounts of catechins, the bioaccessibility of tea catechins and the effect of protein–catechin interactions on their digestibility is a trending topic [
50,
65,
66].
The inability to detect phenolic compounds from the initial phases of dHPI+C and dHPI+HSE complexes might be explained by concluding that as recovery lowered or disappeared, catechins and other phenolics bound to proteins more. These findings are in accordance with prior research on tea phenolics and milk proteins [
50,
66]. However, during the simulated gastric digestion, these bound phenolics and catechins were released from protein–phenolic complexes. The addition of hazelnut meal protein did not affect PCA and C recovery after gastric digestion. GA, GCG, and EGC might be more stable in acidic conditions compared to C, EC, and EGCG, but acid-catalyzed epimerization of EGCG might also occur [
65].
GCG and EC were more stable after intestinal digestion in the dHPI+C complex, while EGC, C, and EGCG were more accessible in C solution. Some studies have also reported that EGC and EGCG are unstable, while EC and ECG are relatively stable [
67,
68]. The difference in intestinal stability of these catechins might be attributable to the fact that three adjacent hydroxyl groups in EGC and EGCG at positions 3′, 4′, and 5′ are more susceptible to the formation of semiquinone free radicals at near-neutral pH when a proton was donated [
69]. Additionally, strong protein–flavonoid interactions are mediated by the hydrophobic binding of flavonoids with proline-rich regions of intact proteins and peptide fragments. According to the literature, larger polyphenols with a higher number of aromatic rings and hydroxyl groups are more effective in forming complexes with proline-rich peptides. Gallate catechins ECG, GCG, and EGCG have more aromatic and hydroxyl groups than non-galloylated catechins EC and EGC, resulting in higher binding activity with proteins. As a result, they become more stable during in vitro digestion, which explains GCG and EGCG’s increased bioaccessibility in HSE when complexed with hazelnut protein [
65,
66].