Non-covalent interactions between whey protein isolate and taxifolin and their potential application in yogurt

Abstract

This study investigated the interaction mechanism of non-covalent complexes between whey protein isolate (WPI) and taxifolin (Tax) and subsequently applied the complexes into yogurt to evaluate their effects on the quality and functionality of yogurt during storage. The results demonstrated that the WPI and Tax can form granular aggregates with average particle sizes between 184.57 and 287.03 nm. As Tax concentration increases, WPI exhibits enhanced UV–visible absorption alongside intensified static fluorescence quenching, primarily driven by conventional hydrogen bonding and π-alkyl interactions between the two compounds. And the incorporation of the WPI + Tax complex into yogurt significantly improves its water-holding capacity and reduces syneresis, with the WPI + 40 μM group showing the optimal effect. The complexes also enhanced yogurt's antioxidant activity and inhibitory effects against α-glucosidase and α-amylase, which reached the highest levels on day 14 of storage, while maintaining good sensory acceptability.

Highlights

• •WPI and taxifolin (Tax) form a stable non-covalent complex via conventional hydrogen bonds and π-alkly interactions.
• •The addition of Tax statically quenches the intrinsic fluorescence of WPI and alters its secondary structure.
• •The addition of the WPI+Tax complex significantly improved yogurt's WHC and reduced syneresis.
• •The WPI+Tax complex demonstrates significant application potential in the development of functional yogurt.

1. Introduction

A by-product of making cheese, whey protein isolate (WPI) is mostly made up of α-lactalbumin (α-LA, 20%) and β-lactoglobulin (β-LG, 60%), with trace amounts of bovine serum albumin (BSA), lactoferrin, and other proteins. With its excellent emulsifying, gelling, and film-forming properties, WPI has been widely utilized in the food industry, such as in infant formula, yogurt, and beverages (Królczyk et al., 2016). The addition of WPI to yogurt not only improves the protein content and total solids index but also significantly improves the product texture, enhances water-holding properties, and effectively inhibits whey precipitation (Ali et al., 2023). But it is worth noting that despite the outstanding processing performance of WPI, it still has some limitations in terms of immunomodulation, antioxidant, and other biofunctional activities, which makes it difficult to meet the increasingly important health needs of modern consumers.

Taxifolin (Tax) is a highly promising polyphenolic compound of the dihydroflavonol group, mainly derived from larch tree roots. Its molecular structure contains five free hydroxyl groups and a conjugated double bond system, and this unique structure endows Tax with powerful biological activities, including antioxidant, anti-inflammatory, and lipid-lowering properties. In addition, Tax has the potential to regulate postprandial blood glucose (Rehman et al., 2019). The European Food Safety Authority has approved its use as a food ingredient that can be used in functional foods such as non-alcoholic beverages, yogurt, chocolate confectionery, etc. (Sunil & Xu, 2019).

In recent years, the non-covalent interactions of proteins and polyphenols have been a hot research topic in the field of food science (Ebrahimi et al., 2025). For example, L. Li, Yu, et al. (2025) demonstrated that the complex system formed by WPI and chlorogenic acid through non-covalent interactions can serve as a novel multifunctional antioxidant. Similarly, hydrophobic interactions and hydrogen bonding are the primary forces mediating the interaction between lactoferrin and procyanidin (C. Li et al., 2021). Beyond the mechanism of interaction, these protein-polyphenol complexes can also lead to improved functional properties. For instance, the soybean protein-catechin complexes exhibit improved storage stability and antioxidant activity, along with improved bioaccessibility of catechin (Dai et al., 2023). Despite these advances, current research has primarily focused on elucidating the interaction mechanisms of protein-polyphenol complexes, with limited exploration of their practical applications in the food industry. Yogurt is a fermented dairy product with a long history and represents an ideal functional food due to its unique flavor, texture, and rich nutritional profile, including high-quality protein, calcium, and B vitamins, making it one of the most widely consumed functional dairy products worldwide. However, traditional yogurt suffers from defects such as whey separation and contains a relatively limited amount of bioactive substances, so research on adding natural functional ingredients to yogurt has received extensive attention from both the scientific community and the industry.

In this study, we investigated the interaction mechanism of non-covalent complexes between WPI and Tax through multi-spectral analysis, molecular docking, and molecular dynamics simulations. Furthermore, the resulting complexes were applied to a yogurt fermentation system to systematically evaluate their effects on the quality and rheological properties of yogurt during storage. This study provides a theoretical foundation for the application of WPI and Tax in functional foods.

2. Materials and methods

2.1. Materials

WPI (purity>90%) was purchased from Macklin Biochemical Technology Co., Ltd. (Shanghai, China). Taxifolin was obtained from Aladdin Chemical Co., Ltd. (Shanghai, China). α-amylase (from hog pancreas) and α-glucosidase (from Saccharomyces cerevisiae) were obtained from Sigma-Aldrich Co., Ltd. (Shanghai, China). The commercial starter YO-MIX PRIME 800 LYO 200 DCU was supplied by Danisco Co., Ltd. (Shanghai, China). Milk containing 3.0% protein, 3.8% fat, and 11% total solids was purchased from a local supermarket. All other supplementary reagents were analytical grade.

2.2. Preparation of WPI + Tax complex

The Tax stock solution was gradually added to the WPI solution in order to generate the WPI + Tax complex. In short, 50 mL of WPI stock solution (1 mg/mL) was prepared using deionized water, agitated for 4 h at 25 °C, and then maintained at 4 °C for 12 h. The Tax stock solution was prepared at a concentration of 10 mM by dissolving it in 30% anhydrous ethanol. Gradually adding the Tax stock solution to the WPI stock solution resulted in the formation of WPI + Tax complexes at Tax concentrations of 0, 10, 20, 30, 40, 50, and 60 μM. All samples were prepared at room temperature. The complex was then freeze-dried, crushed, and stored for future use. The complex system should have received no more than 2% (v/v) of ethanol overall to minimize protein denaturation.

2.3. Measurement of particle size, zeta potential, and polydispersity index (PDI)

The measurement of particle size, PDI, and zeta potential of the WPI and WPI + Tax complex was determined by the dynamic light scattering instrument (Nano-ZS, Malvern Instruments Ltd., UK). Before measurement, all samples were filtered using the 0.22 μm filter membrane. Three measurements were performed for each group of samples at 298 K with a scattering angle of 173°.

2.4. Measurement of ultraviolet-visible (UV–vis) absorption spectroscopy

The UV–vis spectra of the WPI and WPI + Tax complex with different concentrations of Tax (0, 10, 20, 30, 40, 50, and 60 μM) were collected using UV–vis absorption spectroscopy (UV-2550, Shimadzu, Japan). The approach previously described by Xiao et al. (2025) was used to measure the UV spectrum. Deionized water was used as a blank control. The UV absorption spectra of the samples were collected in the 240–400 nm region with a resolution of 0.1 nm.

2.5. Measurement of fluorescence spectroscopy

Using a fluorescence spectrophotometer (F-7100, Hitachi Corporation, Japan), the fluorescence spectra of the WPI and WPI + Tax complex were acquired at different temperatures: 298 K, 304 K, and 310 K with varying concentrations of Tax (0, 10, 20, 30, 40, 50, and 60 μM). The excitation wavelength was 280 nm, while the emission wavelengths ranged from 290 to 450 nm. The scanning rate was 1200 nm/min, and the slit width was 5 nm.

2.6. Determination of fluorescence quenching mechanism and binding constant

The fluorescence quenching mechanism between WPI and Tax was further investigated using the Stern-Volmer eq. (1):

$$\frac{F_0}{F}=1+K_{\mathit{sv}}\left[Q_{\mathit{Tax}}\right]=1+{\mathrm{\kappa }}_q{\tau }_0\left[Q_{\mathit{Tax}}\right]$$ (1)

where the fluorescence intensities of WPI with and without Tax are shown by $F_0$ and F, respectively; the quenching constant is $K_{\mathit{sv}}$; the concentration of Tax is $\left[Q_{\mathit{Tax}}\right]$; the quenching rate constant is ${\mathrm{\kappa }}_q$; and the average lifetime of the WPI is ${\tau }_0$ (10⁻⁸ s).

Following the establishment of the fluorescence quenching mechanism, the binding constant ($K_a$) and the number of binding sites (n) were calculated using the double logarithmic eq. (2) in order to clarify the interaction mechanism between WPI and Tax:

$$\log \frac{F_0-F}{F}=\log K_a+n\log \left[Q_{\mathit{Tax}}\right]$$ (2)

where the fluorescence intensities of WPI with and without Tax are shown by $F_0$ and F, respectively, the concentration of Tax is $\left[Q_{\mathit{Tax}}\right]$.

2.7. Determination of thermodynamic parameters

The enthalpy change ($∆H$), entropy change ($∆S$) and Gibbs free energy ($∆G$) were computed using the Van't Hoff eq. (3) and the Gibbs function eq. (4) to determine the primary driving factor between WPI and Tax:

$$\ln K_a=-\frac{∆H}{\mathit{RT}}+\frac{∆S}{R}$$ (3)

$$∆G=∆H-T∆S$$ (4)

Where the gas constant is R (8.314 J•mol⁻¹•K⁻¹) and the absolute temperature is T.

2.8. Measurement of synchronous fluorescence spectroscopy

The synchronous fluorescence spectra were collected on the same equipment in section 2.5. at Δλ = 15 nm and Δλ = 60 nm, the emission fluorescence spectra in the 200–500 nm range, the scanning rate to 1200 nm/min, and a response time of 2.0 s.

2.9. Measurement of three-dimensional fluorescence spectroscopy

The three-dimensional fluorescence spectra of the WPI and WPI + Tax complex were obtained using the same equipment in section 2.5 with three-dimensional scan measurement. With increments of 2 nm, the wavelengths of the excitation and emission are 200 to 500 nm and 200–500 nm, respectively.

2.10. Measurement of circular dichroism (CD) spectroscopy

The secondary structure of the proteins in the WPI and WPI + Tax complex was collected using a CD spectropolarimeter (Bio-Logic Science Instrument, MOS-500, Claix, France) between the range of 190 to 250 nm (Jianhua Zhao et al., 2025). The final protein concentration is 0.1 mg/mL. A DICHROWEB server was used to analyze the CD data in order to look at the secondary structure of the WPI.

2.11. Molecular docking

Understanding the interaction between the WPI and the Tax can be aided by the application of molecular docking. The crystal structures of α-La (PDB ID: 1hfz), β-Lg (PDB ID: 3npo), and BSA (PDB ID: 3v03) were provided from the Protein Data Bank. All crystallographic ligands, non-protein heteroatoms, water molecules, and ions were removed. The Chem3D 15.0 software was used to improve the 3D structure of Tax (CID: 439533), which was obtained from the PubChem Database. The optimized structure was then saved in PDB format and converted to PDBQT. All rotatable bonds were treated as flexible, and Gasteiger charges were assigned. Molecular docking was then carried out using the open-source program AutoDock Vina 1.1.2 under pH 7.0 conditions, which has been widely utilized in protein-ligand interaction investigations and offers advantages including considerable computational efficiency and excellent prediction accuracy. And lonizable residues were assigned to their dominant protonation states at pH 7.0 (histidine tautomers treated explicitly). Each system underwent ten distinct docking experiments, and the stable binding mode was determined by selecting the conformation with the lowest binding energy. Finally, the protein-ligand interaction was subsequently visualized using PyMOL software (https://www.pymol.org).

2.12. Molecular dynamics simulations

A 100 ns molecular dynamics simulation of the protein-ligand combination was performed using AMBER 18. FF14SB was utilized to handle α-La, β-Lg, and BSA, while Tax was parameterized using GAFF with AM1-BCC partial charges generated via Antechamber. Protonation states of ionizable residues were assigned for pH 7.0. Each complex was solvated in a periodic box of the TIP3 water model, neutralized by counterions, and injected with sodium and chloride ions to mimic ionic screening. Subsequently, the steepest descent approach was used to minimize energy. Following this, each system was allowed to equilibrate in the isothermal and isobaric simulation (NPT) ensembles for 0.05 ns. Finally, molecular dynamics simulations of the α-La, β-Lg, and BSA proteins complexed with Tax were performed for 100 ns. Additionally, the production simulation was run with a time step of 2 fs for a total length of 100 ns. The simulation trajectory was collected every 20 ps for later examination.

2.13. Preparation of yogurt

Yogurt was made and fermented according to our previous study (Bao et al., 2025) with slight modifications. The raw milk was divided into eight treatment groups: (a) a control group without addition, (b) a group with 2% (w/w) WPI, and (c-h) WPI + Tax complex groups (the corresponding Tax preparation concentrations were 10, 20, 30, 40, 50, and 60 μM, with an addition rate of 2% w/w). After adding the sugar, the combined milk was heated at 90 °C for 8 min. And then cooled to 43 °C. 3% of YO-MIX starter culture was then inoculated. All of the inoculated yogurt milk was fermented at 43 °C until the pH reached 4.6. The fermented yogurt was then allowed to mature at 4 °C for a full day.

2.14. Measurement of pH, water holding capacity (WHC), and syneresis of yogurt

The pH value of yogurt samples was measured using a pH meter (Leici PHSJ-4F, Shanghai, China). During storage, WHC and syneresis were measured based on the method described by Hu et al. (2025). Briefly, 8 g of yogurt was centrifuged at 4000 ×g for 15 min in a 50 mL tube. The WHC and syneresis were then calculated using eqs. (5), (6), respectively.

$$\mathit{WHC}\left(\%\right)=\frac{W_1}{W}\times 100$$ (5)

$$\mathit{Syneresis}\left(\%\right)=\frac{W_2}{W}\times 100$$ (6)

where W denotes the initial weight of the yogurt, W₁ denotes the weight of the precipitate after centrifugation, and W₂ denotes the weight of the supernatant.

2.15. Determination of free radical scavenging activity of yogurt

The antioxidant activity of yogurt during storage was evaluated using DPPH and ABTS radical scavenging methods. First, 20 g of yogurt samples were centrifuged at 4000 ×g for 15 min, and the supernatant was collected for later use. The measurements of supernatant primarily reflect the functional properties of the soluble/releasable fraction of yogurt rather than the total activity of the whole gel system.

The following is how the DPPH radical scavenging activity was determined: 200 μL of the supernatant was combined with 800 μL of DPPH (0.1 mmol/L in anhydrous ethanol). The combination was allowed to stand for 30 min at room temperature in the dark. A spectrophotometer was used for measuring absorbance at 517 nm, with anhydrous ethanol serving as the control.

The following is how the ABTS radical scavenging activity was measured: ABTS (7.4 mmol/L) was mixed with potassium persulphate (2.45 mmol/L), and the resulting ABTS radical cation stock solution was kept in the dark at room temperature for 14 h. A mixture of 50 μL of supernatant and 150 μL of ABTS radical cation stock solution was incubated for 30 min in the dark.

The following formula (7) was used to calculate the yogurt's DPPH and ABTS radical scavenging activities:

$$\mathit{DPPH\; radical\; scavenging\; activities}\left(\%\right)=1-\frac{A_{\mathit{sample}}}{A_{\mathit{control}}}\times 100\%$$ (7)

where A_sample represents the absorbance of the yogurt and A_control represents the absorbance of the control group.

2.16. Inhibition activity of yogurt on α-amylase and α-glucosidase

The yogurt supernatant was prepared according to the method described in Section 2.15.

The inhibition activity of the yogurt on α-amylase was determined from Zheng et al. (2020) with some modification. Briefly, 200 μL of α-amylase (1.0 mg/mL) was mixed with 200 μL of supernatant, then incubated at 37 °C for 10 min. Following this, 300 μL of a 1% (w/v) soluble starch solution was then added, and the mixture was kept at 37 °C for 20 min. 2 mL of ethanol was added to stop the process. After centrifugation at 5000 rpm for 8 min. A d-glucose kit (GOPOD, Megazyme International Ireland, Wicklow, Ireland) was then used to determine the amount of glucose that α-amylase had liberated from starch.

The inhibition activity of the yogurt on α-glucosidase was evaluated using the approach of W. Li, Cao, et al. (2025) with a few minor modifications. 30 μL of a 0.3 U/mL α-glucosidase solution, 20 μL of supernatant, and 50 μL PBS (0.01 M, pH = 0.68) were added to a 96-well enzyme plate and kept for 10 min at 37 °C. A mixture was incubated at 37 °C for 15 min after 80 μL of pNPG solution (2.5 mM) was added. The reaction was halted by adding 60 μL Na₂CO₃ (0.2 M), and the absorbance values were measured at 405 nm.

2.17. Sensory evaluation

15 professionals make up the sensory evaluation panel. The sensory evaluation for this study was conducted at Jilin University and did not require approval from an ethics committee. All members were told about the purpose and relevant details of the sensory evaluation and voluntarily signed informed consent forms. The entire study was conducted in strict accordance with the protocol to safeguard the rights, safety, and privacy of the participants. All personal information was kept strictly confidential, and the data was fully anonymized prior to analysis. A 9-point hedonic scale, with 1 denoted as “dislike extremely” and 9 denoted as “like very much,” was used for evaluations. The flavor, taste, and overall acceptability of the yogurt were assessed by the participants. Three replicate assessments were performed on each sample group, and the average score was utilized for data analysis.

2.18. Measurement of rheological properties of yogurt

The DHR-1 rheometer (TA Instruments, USA) was used to examine the rheological properties of yogurt. Testing employed a 40 mm parallel plate fixture with a geometric gap set at 1 mm. The apparent viscosity was determined at 25 °C with shear rates varying from 0.01 to 100 s⁻¹ under steady shear conditions. The storage modulus (G′) and loss modulus (G′′) of yogurt were measured dynamically utilizing oscillatory measurements inside the linear viscoelastic domain using frequency sweeps from 0.01 to 10 Hz at a constant strain of 1% and 25 °C.

2.19. Statistical analysis

IBM SPSS Statistics 19 was used for the statistical analysis, and Duncan's multiple range test (P < 0.05 threshold) was used to assess the significance of difference and mean comparisons. The mean ± standard deviation was used to display the experiment data. OriginPro 2021 software and Excel 2019 were used to create the graph and charts.

3. Results and discussion

3.1. Particle size, zeta potential, and polydispersity index (PDI) analysis

The particle size distribution curves of the WPI and WPI + Tax complex are shown in Fig. 1A. It can be observed that the peak position of the WPI + Tax complex shifted towards larger particle sizes as the concentration of Tax increased. This occurred probably because the complex aggregated as Tax molecular bound to WPI. Fig. 1B displays the zeta potential of WPI and the WPI + Tax complex. All sample have an electronegative zeta potential, and as the concentration of Tax rose, so did the complex's zeta potential's absolute value. Notably, the zeta potential value was 54.94% higher in the WPI + 60 μm Tax than in the WPI alone. Tax binding changes the surface charge distribution of WPI, increasing intermolecular repulsion and enhancing the stability of the WPI + Tax complex in solution, as indicated by the increase in the absolute value of the zeta potential. Additionally, the WPI's average particle size was 184.56 nm (Fig. 1C), consistent with a previous study (Xu et al., 2022). The average particle size of the complex gradually increased with increasing Tax concentration, and the PDI marginally dropped after binding with the Tax, going from 0.418 to 0.240.

Fig. 1.

The particle size distribution (A), zeta potential (B), and average particle size with PDI values (C) of the WPI and WPI + Tax complex. UV–vis absorption spectra of Tax in deionized water (D) and the WPI + Tax complex (E). The plot of the maximum UV–vis absorption spectra and corresponding wavelengths of the WPI + Tax complex (F). C_WPI = 1 mg/mL, C_Tax = 0, 10, 20, 30, 40, 50, and 60 μM, respectively.

3.2. UV–vis absorption spectroscopy analysis

UV–vis absorption spectroscopy is widely employed to deduce the interaction mechanisms between proteins and biologically active small compounds, along with elucidating structural transformations within complexes. Generally speaking, it has been established that proteins with aromatic amino acids can absorb UV–visible light (Xie et al., 2017). Fig. 1D displays the UV–vis absorption spectra of Tax in deionized water, while Fig. 1E shows the spectra of WPI and the WPI + Tax complex. All samples exhibited highest peaks between 277 and 283 nm. The absorbance of the WPI was lowest at 277 nm. While the absorption peak of the WPI + Tax complexes consistently rose and red-shifted from 277 nm to 283 nm with the increase of Tax concentration (Fig. 1F). This could be a result of Tax's interaction with the hydrophobic amino acid residues in WPI, which increases the chromophoric groups' exposure to the solvent around them and improves UV absorption. The study by Wang et al. (2024) on naringenin and soybean protein isolate showed similar trends, with the complex's UV absorbance peak showing an increasing trend as the content of naringenin increased. To rule out the possibility that the spectral changes in the complex were due to the simple superposition of the spectra of the two components, this study further calculated the difference spectrum (Fig. S1). The results of the difference spectrum indicate that there is indeed an interaction between WPI and Tax. Furthermore, the UV–visible absorption changes, when considered together with fluorescence and thermodynamic ananlyses, suggesting that WPI interacts with Tax through a predominantly static quenching mechanism (X. Yu et al., 2022).

3.3. Fluorescence emission spectroscopy analysis

One method for evaluating the mechanisms of interaction between small compounds and macromolecular proteins is fluorescence spectroscopy. The fluorescence spectroscopy of the WPI and WPI + Tax complex was recorded at a temperature of 298 K. As seen in Fig. 2A, WPI showed a significant fluorescence peak at 342 nm, in line with a previous study (Lu et al., 2022). The fluorescence intensity of the WPI + Tax complex reduced as the concentration of Tax increased; this trend suggested that Tax has a quenching impact on WPI fluorescence. In Fig. 2D, compared to WPI (342 nm), a fluorescence peak at 346 nm was obtained for WPI + 60 μM Tax complex, causing a 4 nm red shift. The interaction between the Trp and Tyr residues within the WPI and Tax leading to the formation of a stable complex and, eventually, the quenching of the fluorescence, was the main cause of this occurrence (Meng et al., 2023). Additionally, a concentration-dependent impact between WPI and Tax was identified; the fluorescence quenching effect in WPI is more noticeable at greater Tax concentrations. As the concentration of Tax increases within the range of 0–60 μM, more hydrophobic residues of WPI are exposed to a hydrophilic area. Fluorescence is quenched as a result of the energy transfer and release caused by this. This finding is comparable to the research conducted by Z. Li et al. (2026).

Fig. 2.

Fluorescence spectra of the WPI and WPI + Tax complex at 298 K (A), 304 K (B), and 310 K (C), respectively. Corresponding maximum fluorescence intensities and wavelength shift of the WPI + Tax complex at 298 K (D). The Stern-Volmer equation fit plot (E) and double logarithmic fit plot (F) of WPI by Tax at 298 K, 304 K, and 310 K. Quenching constants (K_sv, K_q, and K_a), number of binding sites (n), and thermodynamic parameters (ΔG, ΔH, and ΔS) (G). C_WPI = 1 mg/mL, C_Tax = 0, 10, 20, 30, 40, 50, and 60 μM, respectively.

3.4. Fluorescence quenching mechanism

The quenching impact of small compounds on protein fluorescence can generally be divided into two categories: static and dynamic quenching (D. Yu et al., 2025). To better understand the fluorescence mechanism between WPI and Tax, the fluorescence intensity fluctuations of WPI and the WPI + Tax complex were recorded at various temperatures (298 K, 304 K, and 310 K), as illustrated in Fig. 2A-C. The experiment data was fitted using the Stern-Volmer equation.

As seen in Fig. 2E, the Stern-Volmer curves displayed a trend of significant linear correlations at 298 K, 304 K, and 310 K (R² > 0.98). K_sv is shown in Fig. 2G. During the quenching process, K_sv increased from 1.01 × 10⁴ L/mol at 298 K to 1.24 × 10⁴ L/mol at 310 K, indicating the presence of hydrophobic interactions. Meanwhile, the calculated minimum K_q was determined to be 1.01 × 10¹² L/mol/s⁻¹, which is significantly higher than the maximum dynamic burst constant of 2.0 × 10¹⁰ L/mol/s⁻¹. This showed that the quenching of WPI by Tax was dominated by a static quenching process involving ground-state complex formation instead of diffusion-controlled collisional quenching, which was consistent with the findings of Wu et al. (2025).

The binding constant (K_a) and the number of binding sites (n) in a static quenching process can be determined using the double-logarithm equation. The results were shown in Fig. 2F and G, where the double-logarithm curves of the WPI and WPI + Tax complex exhibited a good linear relationship (R² > 0.96). The intensity of the interaction between the fluorescent molecule and the quenching agent can be evaluated using K_a. The observed drop in K_a from 12.88 × 10⁴ L/mol at 298 K to 0.22 × 10⁴ L/mol at 310 K could be ascribed to the fact that WPI and Tax have a greater diffusion coefficient in the reaction system at higher temperatures. This causes the WPI + Tax system to become inherently unstable, which lowers K_a. Furthermore, the calculated binding site values of 'n' for the WPI + Tax complex at three temperatures were 1.27, 1.20, and 0.83, respectively. These values were consistently close to 1, suggesting WPI and Tax interacted in a 1:1 ratio. Notably, the n value somewhat dropped as the temperature rose, which might be explained by the lower binding affinities for both at higher temperatures.

3.5. Thermodynamic parameter

To determine the binding driving force, the thermodynamic parameters were calculated from the Van't Hoff and thermodynamic equations (Fig. 2G). When ΔH < 0, ΔS < 0, the principal forces at work are van der Waals and hydrogen bonding; when ΔH > 0, ΔS > 0, the predominant force is hydrophobic; and when ΔH < 0, ΔS > 0, the dominant force is electrostatic. It can be observed that the ΔG values at temperatures of 298 K, 304 K, and 310 K were all negative, suggesting that the WPI + Tax complex formed as a result of a spontaneous reaction. Meanwhile, both of the ΔH (−258.35 kJ•mol⁻¹) and ΔS (−764.87 J•mol⁻¹•K⁻¹) were negative, implying that the van der Waals and hydrogen bonding were the primary forces for the binding between WPI and Tax. Similarly, J. Zhang et al. (2025) investigated the mechanisms of interaction between anthocyanin-3-O-glucoside and β-casein and found that hydrogen bonding and van der Waals were the primary binding forces at pH = 4.

3.6. Synchronous fluorescence spectra analysis

Changes in the microenvironment of protein amino acid residues can be examined using synchronous fluorescence spectrometry. The microenvironment modifications of Tyr and Trp residues are usually described at wavelengths of Δλ = 15 nm and Δλ = 60 nm, respectively (Liang et al., 2023). Fig. 3 displayed the WPI and WPI + Tax complex's synchronized fluorescence spectra. The fluorescence at Δλ = 60 nm (Trp residues) was clearly higher than that at Δλ = 15 nm (Tyr residues); this is in line with the report of Qu et al. (2025) and indicates that the Trp residues were primarily responsible for WPI's fluorescence. When the wavelength difference was Δλ = 15 nm, the Tyr residue of the WPI and WPI + Tax complex showed the highest fluorescence intensity at 300.6 nm (Fig. 3A). There was no red or blue shift (Fig. 3C), indicating that there was no discernible change in the microenvironment around the residue. At a wavelength of Δλ = 60 nm, there was a slight shift in the maximum emission wavelength was collected with increasing Tax concentration (Fig. 3B and D), indicating that the polarity surrounding Trp residue was changed. Moreover, the fluorescence intensity of the WPI + Tax complex steadily decreases as the concentration of Tax rises at both wavelength intervals, revealing that Tax had a quenching impact on WPI. The results of studies employing UV–vis and intrinsic fluorescence are in agreement with this observation.

Fig. 3.

The synchronous fluorescence spectra were collected at excitation wavelengths of 280 nm (A) and 295 nm (B) for the WPI and WPI + Tax complex. C_WPI = 1 mg/mL, and C_Tax = 10, 20, 30, 40, 50, and 60 μM, respectively. Maximum fluorescence intensity and corresponding wavelength at 280 nm (C) and 295 nm (D).

Furthermore, when the concentration of Tax in the complex reached 60 μM, the fluorescence intensity of Tyr residues fell by 51.05%, while that of Trp residues reduced by 61.72%. Suggesting that the binding site between WPI and Tax may be more closely located to the Trp residue, as indicated by the larger decrease in Trp fluorescence intensity when compared to Tyr. This trend was consistent with the results by Fu et al. (2020).

3.7. Three-dimensional fluorescence spectra analysis

Protein complex fluorescence intensity variations can be detected using three-dimensional fluorescence spectroscopy, which can also record conformational and microenvironmental changes in these proteins. The excitation spectra of both WPI and the WPI + Tax complex revealed a distinct characteristic peak in the range of 300–425 nm (Fig. 4), which corresponds to the spectral features of tyrosine and tryptophan residues. The other scattered small peaks (λ_ex = λ_em) are caused by Rayleigh scattering (Yan et al., 2024). As the Tax concentration increased, the intensity of the fluorescence peak decreased significantly, and the emission wavelength underwent a significant red shift. This implies that the combination of WPI and Tax causes a conformational change in protein, exposing hydrophobic amino acid residues such as tryptophan to a polar environment, thereby causing fluorescence quenching (Jiale Zhao et al., 2022). The results are aligned with the intrinsic fluorescence spectrum analysis reported before (Allahdad et al., 2019).

Fig. 4.

Three-dimensional fluorescence spectra of the WPI and WPI + Tax complex. C_WPI = 1 mg/mL (A), C_Tax = 10 (B), 20 (C), 30 (D), 40 (E), 50 (F), and 60 μM (G), respectively.

3.8. CD analysis

The change in the molecular structure of proteins is in its basic shape, which was characterized by CD spectroscopy. Table 1 displays the alterations of the α-helix, β-sheet, β-turn, and random coil in the WPI and WPI + Tax complex. Notably, compared to WPI alone, the relative content of α-helix of WPI + Tax increased, while that of β-turn decreased (p < 0.05); the rise in α-helix implies a stabilization of the hydrogen bonding network between WPI and Tax. Additionally, it also indicates the binding of Tax makes the secondary structure of WPI tend to be changed. And the relative contents of the β-sheet and the random coil remained largely unchanged.

Table 1.

CD spectroscopy analysis of the WPI and WPI + Tax complex.

Secondary structure of protein	Relative content (%)
	WPI	WPI + 30 μM Tax	WPI + 60 μM Tax
α-helix	65.97 ± 0.34	66.76 ± 0.12	70.78 ± 0.19
β-sheet	8.90 ± 0.14	7.83 ± 0.24	8.53 ± 0.11
β-turn	12.00 ± 0.40	11.30 ± 0.03	9.47 ± 0.27
Random coil	13.47 ± 0.13	13.10 ± 0.42	13.60 ± 0.16

3.9. Molecular docking

A visual technique for determining the binding energy, locations, and force between proteins and small compounds is molecular docking simulation. It should be noted that AutoDock Vina's scoring function has inherent limitations and may not fully capture precise binding thermodynamic parameters. Therefore, predicted binding energies should be regarded as estimates of relative binding affinity rather than absolute thermodynamic values. Fig. 5 displays the docking analysis of the main proteins in WPI (α-LA, β-LG, and BSA) and Tax. The binding energy can be used to measure the stability of the binding between the ligand and receptor; the lower the binding energy, the stronger the interaction between the two and the more stable the complex that forms (Kong et al., 2020). Tax had binding free energies of −6.1 kcal/mol with α-LA, −7.5 kcal/mol with β-LG, and − 8.2 kcal/mol with BSA (Table 2). All of these binding free energies were less than −5.0 kcal/mol, revealing that the three primary components of WPI may combine with Tax to form stable complexes under the docking model.

Fig. 5.

Molecular docking analysis of α-LA, β-LG, and BSA with Tax.

Table 2.

The interaction categories between the main proteins in WPI and the small-molecule Tax, as well as the involved amino acid residues.

Main proteins in WPI	Binding free energy (kcal/mol)	2D structure of WPI + Tax	Hydrogen bonds	Hydrophobic interactions
α-LA	−6.1		Asn56, Glu49, Gln54, His32, Thr33, Ala106, Leu105	Tyr103, Trp104
β-LG	−7.5		Asn88, Asn109, Ser116, Met107, Pro38, Lys69	Ala118, Glu108, Leu117, Leu39, Lys60, Ile71, Asn90
BSA	−8.2		Arg144	Arg458, Ala193, Tyr147, Ser192, Arg196, Pro146, His145, Asp108, Pro110, Glu424, Leu189, Thr190, Ser428, Ile455

Table 2 displays the results of the molecular docking analysis between Tax and three whey proteins. For α-LA, it can be observed that the Tax primarily binds within a hydrophilic surface groove, with the stability of its binding conformation maintained by an extensive network of conventional hydrogen bonds. This network involves multiple polar residues like His32, Glu49, and Gln54. Among these, the hydroxyl/oxygen functionalities in the Tax molecule can act as donors or acceptors to form conventional O-H···O/N hydrogen bonds with the carboxylate oxygen atoms of Glu49 (H···O = 2.88 Å), along with the imidazole heteroatoms of His32 (H···O = 2.95 Å) and the amide group of Gln54 (H···O = 3.31 Å). Additionally, the aromatic side chains of Tyr103 and Trp104 provide further non-covalent stabilization by interacting with Tax via π-π stacking or CH-π interactions. For β-Lg, Tax's binding pocket exhibits both hydrophilic and hydrophobic characteristics. In the hydrophilic region, Tax forms conventional hydrogen bonds with several polar residues (like Asn88, Asn109, Ser116, Met107, Pro38, and Lys69). Specifically, the terminal ammonium group of Lys69 forms a hydrogen bond with either the hydroxyl or carbonyl oxygen of Tax, with a bond length of 2.94 Å. Asn88, Asn109, and Ser116 participate in the hydrogen bond network formation through their side chain amide groups or hydroxyl groups. There is a π-alkyl interaction between Tax's aromatic ring and Leu39 of β-Lg. Furthermore, Tax forms non-covalent interactions with BSA by being embedded in the cleft between subdomains. In partically, the phenolic hydroxyl group of Tax forms a hydrogen bond with the Arg144 residue (with a bond length of 3.22 Å), while its aromatic backbone engages in π-alkyl interactions with Arg458 and Arg196.

Notably, compared to the α-LA-Tax and β-Lg-Tax complexes, the BSA-Tax complex forms fewer hydrogen bonds yet exhibits a significantly higher binding free energy (−8.2 kcal/mol). This discrepancy primarily stems from three factors. First, although α-LA-Tax and β-Lg-Tax have more conventional hydrogen bonds, the BSA binding cleft provides a more favorable electrostatic environment and larger hydrophobic and π-bond-related stabilizing interactions. Second, the α-LA groove is more hydrophilic and open, increasing ligand solvent exposure and reducing the net benefit of stacking interactions, whereas the BSA cleft encourages tighter form complementarity. And the polar contact point (Arg144) of BSA can be locked with the ligand orientation to produce significant π-alkyl and stacking interactions. The aggregate effect of several weaker or more solvent-exposed polar contacts within the surface groove would be significantly less than that of these interactions. Additionally, the use of stiff receptor for docking and the removal of crystals and ions to maintain system consistency may have an effect on absolute energy estimates. Therefore, docking is used solely to predict binding modes, with stability in solvent environments subsequently evaluated via molecular dynamic simulations.

According to the information obtained from molecular docking, conventional hydrogen bonding, π-π stacking and π-alkyl can enhance the affinity between WPI and small molecules like Tax, significantly improving the stability of the WPI + Tax complex. This outcome is in line with the findings of the fluorescence quench process and thermodynamic analysis.

3.10. Molecular dynamic simulations

To investigate the interaction mechanism of the complex in depth, molecular dynamics simulations were conducted on the complex of α-LA, β-LG, BSA, and Tax bound to the most advantageous binding sites. The root mean square deviation (RMSD) is a key indicator for evaluating protein structural stability after protein-ligand binding (X. Zhang et al., 2022). Generally, the simulated system is said to have attained equilibrium when its average RMSD changes by less than 0.1 nm. As shown in Fig. 6(A-C), the RMSD of three whey proteins showed different behaviors. The RMSD of α-LA-Tax and β-LG-Tax complexes was in the range of 0.15–0.25 nm in 100 ns with a certain magnitude but eventually trended to be stable, demonstrating that the system found equilibrium during the simulation process and that stable complexes could be formed between three proteins and Tax. And the RMSD values of the three protein-Tax complexes were significantly lower than those of single proteins in the range of 0–100 ns. This situation highlights that the binding of Tax with protein may improve the complex's overall structural stability by limiting the protein's degree of conformational freedom.

Fig. 6.

Results of molecular dynamics simulation. RMSD (A, B, and C) and RMSF (D, E, and F) of α-LA, β-LG, and BSA with Tax complexes. The binding free energy components (G) and the binding free energy of key amino acid residues (H) of α-LA, β-LG, and BSA with Tax complexes.

The root mean square fluctuation (RMSF) shows how much each amino acid residue in a protein molecule deviates from its typical conformational location. As illustrated in Fig. 6(D—F), the RMSF value of α-LA-Tax, β-LG-Tax, and BSA-Tax complexes were lower than those of single proteins, which indicates that the Tax could stably bind to α-LA, β-LG, and BSA. Additionally, α-LA-Tax exhibits greater flexibility in the regions of residues 12–20, 44–46, and 113–114, while β-LG-Tax exhibits greater flexibility in the regions of residues 50–52, 62–64, 85–90, 110–115, and 127–130; similarly, the BSA-Tax complex showed higher RMSF values in residues 31–39 and 48–60, suggesting that these areas were more flexible. This phenomenon demonstrated that Tax modulates local conformational flexibility through hydrophilic interactions with residues in WPI.

The MMPBSA method can determine the binding free energy of a system and decompose the energy contribution at the residue level by analyzing molecular dynamics simulation trajectories. It can be observed that the α-LA-Tax, β-LG-Tax, and BSA-Tax complexes had a negative value of binding free energy (E_total) of −31.12, −34.12, and − 50.12 kJ/mol from Fig. 6G, respectively, suggesting that the Tax could bind to α-LA, β-LG, and BSA to form stable complexes spontaneously. The values of E_vdW, E_ele, and E_surf were all negative, whereas the value of E_gb was positive. These results showed that E_ele, E_vdW, and E_surf promoted the binding of Tax and α-LA, β-LG, and BSA, while E_gb hindered the combination. Notably, BSA-Tax complex exhibits a significantly high electrostatic contribution (more negative E_ele), which corresponds with the arginine-rich environment in the binding cleft, in contrast to α-LA-Tax, which has lower net stabilizing impact due to its more exposed hydrophilic groove characteristics. These findings are in line with the molecular docking studies. Following a statistical examination of the residues' energy contribution, the amino acid residues that are necessary for Tax to bind to the three proteins (α-LA, β-LG, and BSA) were subsequently selected out. As shown in Fig. 6H, Asn102, Tyr103, and Leu105 were the essential residues for Tax binding to α-LA, accounting for less than −1 kJ/mol of the energy. Leu39, Ile71, Ile84, Met107, and Glu108 were the main residues that contributed less than −1.5 kJ/mol to the energy of Tax binding to β-LG. And the Tax binding to BSA was mostly facilitated by the residues Glu186, Leu189, Ile455, and Arg458, which contributed less than −2.0 kJ/mol.

3.11. pH, WHC, and syneresis of yogurt

The final pH value is a key indicator for determining whether yogurt fermentation is finished. As illustrated in Fig. 7A, the final pH value of all yogurt samples stabilized within the range of 4.48 and 4.62, demonstrating the completeness of the fermentation process (Ran et al., 2024).

Fig. 7.

The final pH of yogurt (A). The WHC (B) and syneresis (C), DPPH radical scavenging activity (D), ABTS radical scavenging activity (E), and inhibition activities of α-amylase (F) and α-glucosidase (G) of yogurt during storage. The flavor (H), taste (I), and overall acceptability (J) of yogurt during storage.

WHC and syneresis are the primary indicators of yogurt quality. As shown in Fig. 7B and C, the yogurt samples supplemented with WPI or the WPI + Tax complex exhibited a higher WHC than the control group from the first day of storage, suggesting that the addition of the WPI and WPI + Tax complex significantly improved water immobilization with the gel matrix of yogurt while reducing its syneresis value. Among these, yogurt containing WPI + 40 μM Tax complex demonstrated the most pronounced effect, showing the highest WHC (44.27%) and the lowest value of syneresis (53.36%). This indicates that the WPI + Tax complex enhances the WHC of yogurt by building the formation of the gel network structure. As the Tax concentration in the WPI + Tax complex increased to 50 and 60 μM, the WHC value of yogurt steadily dropped but remained higher than that of the control group and samples containing WPI alone. This implies that the functional effect of the complex does not increase linearly with Tax concentration but rather exhibits an optimal concentration around 40 μM. During storage, the WHC of all yogurt samples peaked on day 14 after an initial increase, followed by a decline. This indicates that the gel network continued to rearrange and densify during the early refrigeration period, thereby enhancing water retention. The subsequent decline in WHC may be attributed to gel shrinkage and whey separation caused by post-acidification.

3.12. Free radical scavenging activity of yogurt

This study used two different antioxidant models, DPPH and ABTS, to evaluate antioxidant properties of the yogurt during 21 days of storage at 4 °C. As depicted in Figs. 7D and E, on the first day of storage, yogurt containing the WPI and the WPI + Tax complex exhibited higher antioxidant activity compared to the control group, with the effect increasing in a dose-dependent manner with the concentration of Tax. The WPI + 60 μM Tax group showed the highest radical scavenging activity among them, with DPPH and ABTS values of 74.76% and 62.26%, respectively-corresponding to a 2.97-fold and 2.43-fold increase over the control group. This is explained by the fact that the Tax contains multiple phenolic hydroxyl groups and an aromatic that can donate electrons or hydrogen atoms to neutralize reactive oxygen species (Kılıçaslan, 2025). Higher Tax concentration provides more redox active sites; simultaneously, the non-covalent binding between WPI and Tax enhances Tax dispersion and stability within the yogurt gel matrix, thereby enhancing the antioxidant properties of yogurt. These results are consistent with earlier studies that polyphenol-protein complexes generally exhibit stronger antioxidant activity than single proteins (Yang et al., 2025).

During storage, the DPPH and ABTS free radical scavenging activity of all samples increased, peaked on day 14, then progressively decreased. This effect could be attributed to two factors. On the one hand, the continuous restructuring of the yogurt gel network during the initial storage period promotes a more uniform distribution of Tax within the matrix; on the other hand, after Tax binds to proteins via hydrogen bonds and π-type interactions, its own oxidative degradation is effectively inhibited. In summary, the WPI + Tax complex holds potential for application as a natural antioxidant in a yogurt system.

3.13. Inhibition of α-amylase and α-glucosidase

This study assessed yogurt's anti-diabetic potential by measuring α-amylase and α-glucosidase inhibition. Figs. 7F and G showed the inhibition activities of α-amylase and α-glucosidase of the yogurt during storage. On the first day of storage, the inhibitory rates of the control group against α-amylase and α-glucosidase were 18.79% and 27.78%, respectively. Following the addition of WPI, the corresponding inhibition rates rose to 32.55% and 20.47%, while the addition of the WPI-Tax complex significantly improved these inhibitory effects, which showed a clear dose-dependent increase within the Tax concentration range of 10–40 μM. And the inhibition rates tended to plateau when the Tax concentration exceeded 40 μM. The enzyme-inhibitory activity of all samples peaked on day 14 during storage, and it then slowly dropped after that. The enhanced accessibility of the complex in the yogurt during early storage, which encourages more effective enzyme binding, may account for this change.

Furthermore, the inhibitory effect on α-glucosidase inhibition was significantly stronger than α-amylase. This selective inhibition pattern minimizes undigested carbohydrate accumulation in the lower gastrointestinal tract, thereby reducing microbial fermentation-induced abdominal discomfort and flatulence (Dong et al., 2012).

3.14. Sensory evaluation

Sensory evaluation is a key indicator for determining customer acceptability. In this study, we assessed the flavor, taste, and overall acceptability of yogurt during storage, as displayed in Figs. 7H-J. On the first day of storage, yogurt containing the WPI + Tax complex exhibited slight advantages in taste and overall acceptability. Among these, the WPI + 40 μM group showed the highest taste score, whereas the WPI + 60 μM Tax group showed the lowest overall acceptability, indicating that high Tax concentrations of Tax may have a negative impact on the product's sensory quality. During storage, the WPI + Tax complex group at a Tax concentration of 10–40 μM outperformed both the control group and the WPI alone group in terms of flavor, taste, and overall acceptability. This result indicates that the WPI + Tax complex can improve the flavor, taste, and overall acceptability of yogurt, with a 40 μM addition concentration showing the best results.

3.15. Rheological properties of yogurt

Rheological properties are the fundamental characteristics of fluid stability as well as critical indicators for evaluating the texture and sensory attributes of yogurt. Based on the above discussion and analysis, this study selected yogurt samples from the control group, WPI group, and WPI + 40 μM Tax group stored for 14 days for rheological property testing. Fig. S2 displays rheological properties of all yogurt samples. All yogurt samples showed typical shear thinning behavior, with their apparent viscosity dramatically decreasing as the shear rate rose (Zhu et al., 2025). The addition of WPI and WPI + Tax complexes increased the yogurt's apparent viscosity compared with the control (Fig. S2A), which could be attributed to the formation of a reinforced gel network through synergistic interactions between whey proteins and casein micelles during fermentation (Laiho et al., 2017). The G′ and G′′ are key indicators for evaluating the viscoelastic behavior of yogurt, as shown in Figs. S2B and C. It was observed that there was no overlap between G′ and G′′ at any frequency (0.1 Hz–10 Hz), with G′ consistently exceeding G′′, a characteristic of weak gels, suggesting that the yogurts possessed a stable gel structure (Huang et al., 2021). It is worth noting that all samples showed a downward trend in G′ value at high frequencies, indicating that excessive oscillatory shear may disrupt the original structure of the yogurt. The G′ value of WPI and WPI + Tax complex yogurts was significantly higher than those of the control, indicating that both WPI and WPI + Tax complexes improved yogurt's viscoelastic properties. However, there is no apparent difference between the G' values of WPI and WPI + Tax yogurts, demonstrating that the WPI + Tax complex did not adversely affect the rheological properties of yogurt. This finding provides a foundational basis for future food applications involving Tax.

4. Conclusion

This study investigated the interaction mechanism between WPI and Tax using multi-spectral analysis, molecular docking, and molecular dynamics simulation and subsequently applied the resulting complexes into yogurt to evaluate their effects on the product's physical, functional, and rheological properties during storage. The results showed that the complex formation increased particle size, decreased PDI, and raised absolute zeta potential to build larger, more homogeneous, and colloidally stable assemblies. Multi-spectroscopic analysis demonstrated that the incorporation of Tax leads to conformational changes in WPI, including static quenching of its intrinsic fluorescence (tryptophan and tyrosine residues), exposure of hydrophobic domains, and reconfiguration of its secondary structure. The thermodynamic parameters revealed that the hydrogen bonding and van der Waals forces are the main forces driving and maintaining this binding process. Molecular docking and molecular dynamics simulations further validated these results. The addition of the WPI + Tax complex significantly improved yogurt's WHC and reduced syneresis. And it markedly enhanced yogurt's free radical scavenging capacity of yogurt and reinforced the inhibition of .α-amylase and α-glucosidase, indicating that the complex could contribute not only quality but also to the functional value of the final product. Rheological results indicated that both the WPI and WPI + Tax complex enhanced yogurt's apparent viscosity and viscoelasticity. Among these, the sample WPI + 40 μM Tax performed particularly well across all parameters and received the highest overall acceptability score. This work supports the unilization of protein-polyphenol complexes as functional components in dairy systems and elucidate the molecular mechanism of WPI + Tax interaction.

CRediT authorship contribution statement

Chenligen Bao: Writing – review & editing, Writing – original draft, Investigation, Data curation, Conceptualization. Mi Yan: Investigation, Conceptualization. Fangyu Li: Investigation. Liqing Jiang: Data curation. Tiehua Zhang: Writing – review & editing, Supervision, Software, Project administration.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.