Functional properties and Pickering emulsion stability of whey protein isolate–inulin complexes enhanced by covalent modification of proanthocyanidins

Abstract

BACKGROUND

Ultrasound‐assisted covalent modification offers a promising approach for fabricating protein–polysaccharide–polyphenol ternary complexes as advanced Pickering emulsion stabilizers, yet facile preparation strategies remain underexplored. This study aimed to develop an ultrasound‐assisted covalent modification strategy for preparing whey protein isolate–inulin–proanthocyanidin (WPI‐I‐PC) ternary complexes and evaluate their functional performance, particularly as Pickering emulsion stabilizers.

RESULTS

The WPI‐I‐PC complexes were prepared by pH adjustment and wet‐heating method with or without ultrasonic assistance. The incorporation of PC markedly decreased the particle size and surface hydrophobicity while altering the secondary structure of WPI. Compared with WPI‐PC binary complexes (covalent or non‐covalent) and WPI‐I‐PC ternary non‐covalent complexes, the WPI‐I‐PC covalent complexes prepared with ultrasound‐assisted wet‐heating method at pH 10 exhibited superior thermal stability, emulsifying activity, and antioxidant capacity. Notably, ultrasound‐assisted preparation for 72 min (WPI‐I‐PC (c)) achieved comparable functionality to the 4 h wet‐heating method (WPI‐I‐PC (4 h)), with a smaller particle size (384.50 versus 476.03 nm). In Pickering emulsions, the WPI‐I‐PC covalent complexes markedly improved emulsion performance, as reflected by a reduction in flocculation index (from 32.00% for WPI to 7.92% for WPI‐I‐PC (4 h) and 8.24% for WPI‐I‐PC (c)) and an increase in interfacial protein adsorption content (from 76.80% to 84.92% and 84.37%, respectively). Furthermore, WPI‐I‐PC (c) demonstrated outstanding stability and enhanced antioxidant activities.

CONCLUSION

Ultrasound‐assisted covalent modification enabled rapid fabrication of WPI‐I‐PC ternary complexes with enhanced structural and functional properties. The resulting Pickering emulsions demonstrated outstanding stability and antioxidant activity, offering an efficient strategy for developing novel food‐grade emulsion stabilizers with potential applications in functional food delivery systems. © 2026 Society of Chemical Industry.

INTRODUCTION

Whey protein isolate (WPI), a nutritionally valuable mixture of globular proteins derived from whey protein, ¹ exhibits limited functionality and solubility near its isoelectric point (IP) or at elevated temperatures, necessitating modification for broader application.2, 3 The Maillard reaction can enhance protein functionality by forming covalent complexes with polysaccharides. Conventional thermal methods, however, are inefficient and prone to causing protein denaturation. Ultrasonic processing presents a more efficient alternative, using cavitation to accelerate the reaction and produce complexes with superior thermal stability, foaming, and antioxidant properties.4, 5 Inulin is a natural polysaccharide composed of β‐(2→1)‐linked d‐fructofuranose units, predominantly sourced from Asteraceae tubers and roots. As a prebiotic, it regulates gut microbiota and offers multiple health benefits, such as modulating metabolic syndrome, lowering colon cancer risk, enhancing nutrient absorption, and ameliorating mood disorders. ⁶ Previous work from our laboratory demonstrated that whey protein isolate–inulin (WPI‐I) complexes prepared via the ultrasound‐assisted wet‐heating method altered the secondary structure of WPI and significantly improved its functional properties, including thermal stability, emulsifying activity index (EAI), emulsion stability (ES), and antioxidant capacity. ⁷

Polyphenols and proteins can form complexes via non‐covalent or covalent interactions. The former primarily involves hydrogen bonding, hydrophobic effects, and van der Waals forces, while the latter is typically induced by enzymatic reactions, alkaline conditions, or free radical methods. ⁸ Among these, covalent bonds are generally more stable owing to electron pair sharing. Under alkaline conditions, polyphenols are easily oxidized to form quinone intermediates, which can covalently react with nucleophilic amino acid residues of proteins, offering a simple and effective strategy to prepare stable protein–polyphenol conjugates without the need for exogenous reagents. ⁹ Consequently, protein–polyphenol complexes often demonstrate enhanced functional activities compared to proteins, particularly in terms of antioxidant capacity. ¹⁰ Proanthocyanidins (PCs) contain abundant phenolic hydroxyl groups and exhibit strong free radical scavenging capacity.11, 12 The conjugation of PC with WPI leads to enhanced antioxidant activity, foaming, and emulsifying properties, along with a concurrent reduction in surface hydrophobicity (H ₀). ¹³ Building upon research into binary systems, protein–polysaccharide–polyphenol ternary complexes have attracted increasing interest. Existing studies have confirmed that the myofibrillar protein–dextran–epigallocatechin gallate (EGCG) covalent conjugate improved the thermal stability and antioxidant capacity of emulsions. ¹⁴ Similarly, the WPI–pectin–chlorogenic acid (CA) complex could optimize emulsion stability. ¹⁵ These results demonstrated the promising application of protein–polysaccharide–polyphenol ternary complexes as innovative natural emulsifiers. Nevertheless, further research is required to explore effective modification strategies for WPI to strengthen its emulsifying performance.

Pickering emulsions are stabilized by solid particles, improving the processability and storage stability of perishable oils. However, as thermodynamically metastable systems, their stability is critically dependent on the performance of the emulsifier. An effective emulsifier must sufficiently lower interfacial tension while preventing droplet aggregation, flocculation, or coalescence. Single‐component stabilizers (proteins or polysaccharides) often perform poorly, highlighting the need for novel composite emulsifier systems. ¹⁶ Stable emulsions have been successfully produced from protein–polysaccharide or protein–polyphenol complexes in prior research. Pickering emulsions stabilized with pea protein isolate–dextran conjugates (prepared via wet‐heating method) exhibited superior viscosity and dispersion stability compared to those stabilized with pea protein isolate alone. ¹⁷ Similarly, emulsions stabilized by ellagic acid/EGCG–walnut protein isolate conjugates showed smaller average droplet sizes, increased surface charge, and significantly improved pH, thermal, and storage stability relative to those stabilized by walnut protein isolate. ¹⁸

Therefore, this study builds upon the pre‐formed WPI‐I binary complexes and selected samples subjected to wet‐heating treatment for 2 h (WPI‐I (2 h)), 4 h (WPI‐I (4 h)), and ultrasound‐assisted wet‐heating treatment (WPI‐I (c)). By regulating the pH, whey protein isolate–inulin–proanthocyanidin (WPI‐I‐PC) ternary complexes (including both non‐covalent and covalent forms) with different grafting degrees were prepared. A series of key structural and functional indicators was systematically evaluated to compare the covalent and non‐covalent complexes, thereby providing theoretical and technical foundations for the application of protein–polysaccharide–polyphenol ternary complexes. Furthermore, Pickering emulsions were prepared using WPI‐I complexes (wet‐heating treatment for 2 h, 4 h, and ultrasonic‐assisted wet‐heating treatment) and WPI‐I‐PC complexes (pH 10, wet‐heating treatment for 2 h, 4 h, and ultrasonic‐assisted wet‐heating treatment) as emulsifiers. In addition, the structural properties, physicochemical stability, and extent of lipid oxidation during storage of the Pickering emulsions were evaluated, thereby offering experimental support for developing natural emulsion stabilizers. This study aligns with the Sustainable Development Goals (SDGs) by promoting the utilization of renewable natural biopolymers to enhance the stability and oxidative resistance of emulsion‐based food systems, thereby contributing to reduced food waste and improved food quality.

MATERIALS AND METHODS

Experimental materials

WPI was obtained from Hilmar Ingredients International Inc. (Hilmar, CA, USA); Inulin (as Orafti® GR) was purchased from Beneo Company (Mannheim, Germany); PC (extracted from Lycium ruthenicum Murr., purity 90%) was supplied by Shanghai Yuanye Biotechnology Co., Ltd (Shanghai, China); soybean oil was obtained from a local supermarket in Wuhan, China; O‐phthalaldehyde (OPA, purity ≥ 98%), ethylenediaminetetraacetic acid (EDTA), and 8‐aniline‐1‐naphthalenesulfonic acid (ANS) were obtained from Shanghai Macklin Biochemical Co., Ltd (Shanghai, China). Analytical‐grade chemicals were used for all other compounds.

Sample preparation

Preparation of non‐covalent/covalent WPI‐I‐PC ternary complexes

WPI‐I complexes were prepared following a previously established method. ⁷ Briefly, WPI and inulin stock solutions (100 mg mL⁻¹) were prepared separately in ultrapure water and stirred continuously at room temperature for 2 h to ensure complete dissolution.

For WPI‐I (2 h) and WPI‐I (4 h) the 1:1 (v/v) mixture of WPI and inulin was adjusted to pH 7.0, hydrated at 4 °C overnight, and then heated at 80 °C for 2 h or 4 h, respectively. For WPI‐I (c) the 3:4 (v/v) mixture of WPI and inulin was adjusted to pH 8.2, hydrated at 4 °C overnight, and treated ultrasonically (JY92‐IIDN ultrasonic cell disruptor (Scientz Biotechnology Co., Ltd, Ningbo, China), 24% power, 72 min, 50% duty cycle (2 s on/2 s off), 8 mm probe). The reaction was terminated by ice‐bath cooling, followed by 48 h dialysis and freeze‐drying.

For non‐covalent/covalent WPI‐PC, WPI‐I‐PC (2 h), WPI‐I‐PC (4 h), and WPI‐I‐PC (c) complexes a PC solution was mixed individually with WPI, WPI‐I (2 h), WPI‐I (4 h), and WPI‐I (c) solutions (all at 50 mg mL⁻¹ concentration). The mixtures were adjusted to pH 7.0, 9.0, 10.0 or 11.0, stirred at room temperature for 24 h, dialyzed for 48 h, and freeze‐dried to obtain WPI‐PC binary and WPI‐I‐PC ternary non‐covalent (pH 7.0) and covalent complexes (pH 9.0, 10.0, and 11.0).

All freeze‐dried samples were stored at −18 °C.

Preparation of Pickering emulsion

The sample solution (10 mg mL⁻¹) and soybean oil were combined in a 9:1 ratio and homogenized using a ULTRA‐TURRAX® homogenizer (IKA, Staufen, Germany) at 11 400 rpm for 3 min. This was followed by ultrasonication for 14 min (JY92‐IIN ultrasonic cell disruptor, 30% power, 1 s on/1 s off cycle).

Determination of polyphenol content

Following the addition of 2.5 mL fresh Folin‐Ciocalteu reagent to 0.5 mL of sample (0.2 mg mL⁻¹), the mixture was kept in the dark for 5 min. Subsequently, 2 mL of a 7.5% (w/v) sodium carbonate (Na₂CO₃) solution was added, and the mixture was allowed to react for 2 h in darkness before measuring the absorbance at 760 nm on an ultraviolet (UV)‐visible spectrophotometer (UV‐1800; Shimadzu, Kyoto, Japan). ⁸ The total phenolic content of each sample was determined using a standard curve of PC.

Determination of sulfhydryl content

Ellman's reagent (2.5 mL), prepared in a Tris‐glycine buffer containing 0.01 mol L⁻¹ EDTA and 8 mol L⁻¹ urea, was added to 0.5 mL of the sample solution (2.0 mg mL⁻¹), and the mixture was heated in a boiling water bath for 15 min to denature the protein. After addition of 20 μL Ellman's reagent (4 mg mL⁻¹ in Tris‐glycine buffer), the system was maintained at room temperature for 60 min. The absorbance at 412 nm was recorded with a UV‐1800. The sulfhydryl content (in μmol g⁻¹) was calculated using Eqn (1):

$$\text{Sulfhydryl content}=\frac{73.53A_{\mathrm{SH}}D}{C_{\mathrm{S}}}$$ (1)

where, A _SH denotes the absorbance of the solution at 412 nm, D represents the dilution factor of total sulfhydryl (6.05), and C _S denotes the concentration of sample.

Determination of free amino group content

The sample solution (2.0 mg mL⁻¹, 100 μL) and OPA reagent (4 mL) were mixed and allowed to react for 2 min at 25 °C. The absorbance at 340 nm was immediately recorded on a UV‐1800 spectrophotometer. ⁷ Results were calculated based on an l‐leucine standard curve.

Particle size

Sample solutions prepared at a concentration of 2.0 mg mL⁻¹ were analyzed for average particle size using a NanoZS nanoparticle analyzer (Malvern Instruments Ltd, Malvern, UK), with triplicate measurements.

Surface hydrophobicity

The H ₀ was measured according to the method described by Cui et al. ¹⁹ with minor modifications. The sample solution was diluted to 0.05–0.25 mg mL⁻¹ with 10 mmol L⁻¹ phosphate‐buffered saline (PBS, pH 7.0), followed by addition of 20 μL of the fluorescent probe ANS (8 mmol L⁻¹ in 10 mmol L⁻¹ PBS, pH 6.7). The mixture was kept in the dark at room temperature for 15 min after mixing. Fluorescence spectra were acquired using an F‐4600 spectrophotometer (Hitachi High‐Tech (Shanghai) Co., Ltd, Shanghai, China) at excitation and emission wavelengths of 390 and 470 nm, respectively. The H ₀ of WPI was determined from the slope of the linear regression between fluorescence intensity and protein concentration.

Fourier‐transform infrared (FTIR) analysis

Freeze‐dried powders were combined with potassium bromide (KBr) and compressed into pellets. The Fourier‐transform infrared (FTIR) spectra were then recorded across the 4000–400 cm⁻¹ range with a 4 cm⁻¹ resolution (IS50; Thermo Fisher Scientific, Shanghai, China), using pure KBr for measurement background.

PeakFit v4.12 was utilized to quantify protein secondary structures by performing second‐derivative analysis and multi‐Gaussian fitting of the amide I band in the acquired spectra. ²⁰ The spectral bands for protein secondary structures were assigned as follows: 1646–1664 cm⁻¹ (α‐helix), 1615–1637 and 1682–1700 cm⁻¹ (β‐sheet), 1664–1681 cm⁻¹ (β‐turn), and 1637–1645 cm⁻¹ (random coil). ²¹

Thermal stability

A freeze‐dried sample weighing 3.0 mg was placed in a sealed aluminum crucible, with an empty crucible employed as the blank control. The temperature was increased from 20 to 200 °C at 10 °C min⁻¹ under a 50 mL min⁻¹ nitrogen purge (differential scanning calorimetry (DSC), 204 F1; Netzsch Gerätebau GmbH, Selb, Germany). From the thermograms, the initial denaturation temperature (T _m), peak denaturation temperature (T _p), and enthalpy change (ΔH) were determined.

Emulsifying properties

EAI and ES indices were determined by turbidimetry following a previously established method. ⁷ Briefly, a sample solution (2 mg mL⁻¹) was homogenized with soybean oil (5 mL) at 11 400 rpm for 3 min. At 0 and 10 min after homogenization, 50 μL of the emulsion was diluted with 0.1% (w/v) sodium dodecyl sulfate (SDS, 5 mL), and the absorbance at 500 nm was measured using a UV‐1800 spectrophotometer. EAI (in m² g⁻¹) (Eqn (2)) and ES (in minutes) (Eqn (3)) were calculated as follows:

$$\begin{array}{c}\mathrm{EAI}=\frac{\left(2\times 2.303\times A_0\times N\right)}{\left(\rho \times \phi \times \left(1-\theta \right)\times 10000\right)}\end{array}$$ (2)

$$\begin{array}{c}\mathrm{ES}=\frac{A_0}{A_0-A_{\mathrm{t}}}\times \mathrm{t}\end{array}$$ (3)

where, N denotes the dilution factor (N = 101); ρ represents the protein concentration (in g mL⁻¹); $\phi$ is the cuvette path length (in centimeters); θ is the oil volume fraction (θ = 0.20); A ₀ and A _t denote the sample absorbance recorded at 0 min and t min (t = 10).

Antioxidant capacity

ABTS free radical scavenging capacity

ABTS (2,2'‐azino‐bis(3‐ethylbenzothiazoline‐6‐sulfonic acid)) radical scavenging capacity was determined following a previously described methodology by Jiang et al. ²² Briefly, the ABTS free radical scavenging capacity was calculated as follows (Eqn (4)), based on the absorbance at 734 nm recorded after the reaction of 0.4 mL sample (1–5 mg mL⁻¹) with 3 mL ABTS reagent in the dark for 30 min.

$$\begin{array}{c}\text{ABTS free radical scavenging capacity}\left(\%\right)=\left(1-\frac{A_1}{A_0}\right)\times 100\end{array}$$ (4)

where, A ₁ and A ₀ denote the absorbance of the sample and distilled water, respectively.

DPPH free radical scavenging capacity

DPPH (2,2‐diphenyl‐1‐picrylhydrazyl) radical scavenging capacity was determined following Brand‐Williams et al. ²³ with minor adjustments. A 0.5 mL sample (1–5 mg mL⁻¹) was mixed with 2.5 mL of 0.1 mmol L⁻¹ DPPH (in anhydrous ethanol) and reacted in the dark at room temperature for 30 min. Absorbance was measured at 517 nm (UV‐1800 spectrophotometer), and scavenging capacity was calculated using Eqn (5):

$$\begin{array}{c}\text{DPPH free radical scavenging capacity}\left(\%\right)=\left(1-\frac{A_{\mathrm{S}}-A_{\mathrm{C}}}{A_{\mathrm{B}}}\right)\times 100\end{array}$$ (5)

where, A _S, A _C, and A _B represent the absorbance of the sample, control group (0.5 mL sample + 2.5 mL anhydrous ethanol), and blank group (0.5 mL anhydrous ethanol + 2.5 mL 0.1 mmol L⁻¹ DPPH solution), respectively.

Flocculation Index (FI)

Flocculation index (FI) is calculated based on the particle size values obtained in distilled water and 1% (w/v) SDS solution, with the following equation (Eqn (6)):

$$\begin{array}{c}\mathrm{FI}\left(\%\right)=\left(\frac{D_{\mathrm{W}}}{D_{\mathrm{S}}}-1\right)\times 100\end{array}$$ (6)

where, D _W and D _S denote the particle size of the emulsion measured in distilled water and 1% (w/v) SDS, respectively.

Adsorbed interface protein content

Following centrifugation of 2 mL of the emulsion at 10 528 × g for 20 min, the aqueous phase was harvested and filtered (0.22 μm). Interfacial protein adsorption was quantified using the protein content determined in the filtrate by the Bradford method, ²⁴ with the following equation (Eqn (7)):

$$\begin{array}{c}\mathrm{AP}\left(\%\right)=\frac{\left[\left(C_0-C_{\mathrm{f}}\right)\times 100\right]}{C_0}\end{array}$$ (7)

where, C ₀ denotes the initial protein concentration of the aqueous phase (in g mL⁻¹); C _f represents the unadsorbed protein concentration in the lower aqueous layer (in g mL⁻¹).

Rheological properties

Samples were loaded onto a 40 mm test plate with a 1000 μm gap, and apparent viscosity was determined over a shear rate range of 0.1–100 s⁻¹ (DHR2 rheometer; TA Instruments, New Castle, DE, USA). The linear viscoelastic (LVE) region was first determined by strain sweep measurements at a constant frequency of 1 rad s⁻¹ over a strain range of 0.01% to 10%. A strain amplitude of 1.0% was then selected to ensure that all subsequent dynamic oscillatory tests were conducted within the LVE region. Frequency sweeps were performed from 0.1 to 100 rad s⁻¹ at 25 °C, and the frequency‐dependent variations of the storage modulus (G') and loss modulus (G″) were recorded.

Confocal laser scanning microscopy (CLSM)

Stock solutions of FITC (fluorescein isothiocyanate, aqueous phase) and Nile Red (oil phase) were separately prepared in anhydrous ethanol at a concentration of 10 mg mL⁻¹ for fluorescent staining. For confocal laser scanning microscopy (CLSM) analysis, 1 mL of emulsion was dual‐stained with FITC and Nile Red (10 μL each) and left in the dark for 10 min. Images were acquired at excitation wavelengths of 488 nm and 543 nm to characterize the particle size and morphology (CLSM, FV3000; Olympus, Tokyo, Japan).

Salt ions, pH, and temperature stability

Aliquots (3 mL) of the emulsion were subjected to the following treatments: the pH was adjusted to 2.0, 3.0, 4.0, 5.0, 6.0, 7.0 and 8.0 using 0.1 mol L⁻¹ hydrochloric acid/sodium hydroxide (HCl/NaOH), then left to stand for 20 min; the final concentration of salt ions was adjusted to 50, 100, 200 and 400 mmol L⁻¹ with sodium chloride (NaCl), then left to stand for 20 min; the emulsion was placed in test tubes and heated in a water bath at 60, 70, 80, 90 and 100 °C for 20 min (with the emulsion at room temperature as the control). Particle size was measured before and after each treatment to assess stability.

Oxidative stability of emulsion

Aliquots (10 mL) of the emulsion were stored in sealed glass vials for 2 to 10 days at 37 °C and 50 °C.

Determination of peroxide value (POV)

An emulsion sample of 0.3 mL was mixed with 1.5 mL of isooctane/isopropanol (3:1, v/v), vigorously vortexed, and centrifuged at 10 528 × g for 5 min. Then, 200 μL of the upper phase was collected and mixed with 2.8 mL of methanol/butanol (2:1, v/v). Subsequently, 15 μL of ferrous solution (0.132 mol L⁻¹ barium chloride (BaCl₂) + 0.144 mol L⁻¹ iron(II) sulfate (FeSO₄)) and 15 μL of 3.94 mol L⁻¹ ammonium thiocyanate (NH₄SCN) were added.

Determination of thiobarbituric acid reactive substances (TBARS)

An emulsion sample of 1 mL was thoroughly vortexed with 2 mL of thiobarbituric acid (TBA) solution, heated in boiling water for 15 min, then rapidly cooled and centrifuged (1685 × g, 10 min). The absorbance at 532 nm was recorded on a UV‐1800 spectrophotometer. A standard curve was prepared using 1,1,3,3‐tetraethoxypropane for quantification.

Statistical analysis

All experiments were performed in triplicate, and results are expressed as mean ± standard deviation. One‐way analysis of variance (ANOVA) was performed using IBM SPSS Statistics 24. Duncan's multiple range test was used to determine significant differences between means, with a significance level of P < 0.05. Figures and tables were generated using Excel 2019 and Origin 2024.

RESULTS AND DISCUSSION

Preparation and characterization of WPI‐I‐PC

Polyphenol content

The polyphenol content of the complexes was measured to evaluate the effect of PC binding (Fig. 1(A)). Under alkaline conditions (pH 9, 10, and 11), covalent complexes were formed between WPI and PC, whereas under neutral conditions (pH 7), non‐covalent complexes were predominantly obtained. ⁹ All covalent complexes exhibited significantly higher polyphenol content than non‐covalent complexes (P < 0.05), confirming enhanced covalent binding between PC and WPI/WPI‐I. This is attributed to alkaline‐induced polyphenol oxidation generating semiquinone radicals that covalently bind to protein amino acid side chains. ²⁵ Vasava et al. ¹⁰ also reported superior polyphenol incorporation in covalent complexes. Covalent complexes achieved the highest PC binding at pH 10 due to greater polyphenol oxidation relative to pH 9, while binding decreased at pH 11 from excessive alkalinity‐induced protein denaturation. ²⁶ Additionally, WPI bound less PC than WPI‐I in both covalent and non‐covalent complex systems, as inulin and PC bound to distinct WPI sites without competition, ¹⁵ and WPI structural modification by inulin conjugation exposed additional PC‐binding sites. ²⁷ WPI‐I‐PC (4 h) exhibited a more complete reaction than WPI‐I‐PC (2 h), and WPI‐I‐PC (c) had comparable PC binding to WPI‐I‐PC (4 h), indicating more pronounced structural changes and exposed active sites under these two conditions.

Figure 1.

Polyphenol content (A), sulfhydryl content (B), free amino group content (C), particle size (D), and surface hydrophobicity (E) of WPI‐PC and WPI‐I‐PC complexes at different pH. Different lowercase letters denote that the same conjugate is significantly different at different pH values (P < 0.05); different capital letters denote significant differences between different conjugates at the same pH (P < 0.05). (Same in Fig. 4).

Sulfhydryl content

Protein sulfhydryl content was determined to characterize microstructural changes (Fig. 1(B)), with covalent complexes showing markedly lower sulfhydryl content than non‐covalent complexes (P < 0.05). Urea was used to eliminate disulfide bond interference during sulfhydryl group quantification. Non‐covalent interactions involve reversible hydrogen bonding between polyphenolic hydroxyls and proteins, whereas alkaline covalent binding induces polyphenol hydroxyl oxidation to quinones that condense with protein sulfhydryl groups to form conjugated structures, reducing free sulfhydryl levels. ²⁸ Comparable effects occurred in other protein–polyphenol systems, such as casein–quercetin complexes ²⁹ and anthocyanin–soy protein isolate complexes. ³⁰ Among the non‐covalent complexes, WPI‐I‐PC (c) exhibited the highest sulfhydryl content, likely because ultrasonic cavitation unfolds proteins and dissociates aggregates, increasing hydrophobic group exposure. ³⁰ All covalent complexes showed minimal sulfhydryl group levels at pH 10, confirming the highest PC binding under this condition. Overall, WPI‐I‐PC had higher sulfhydryl content than WPI‐PC, presumably due to the wet‐heating treatment applied during the formation of WPI‐I conjugated via the Maillard reaction. The elevated temperatures promoted protein unfolding and exposure of functional groups, thereby facilitating subsequent binding with PC. ²⁷

Free amino group content

Free amino group content further indicated the extent of polyphenol binding (Fig. 1(C)). Non‐covalent complexes retained significantly more free amino groups than covalent complexes (P < 0.05), because the covalent binding of polyphenols forms stable C–N/C–S bonds with amino acid residues, ¹³ while non‐covalent binding is reversible and preserves most amino groups. Among the non‐covalent complexes, WPI‐I‐PC (2 h), WPI‐I‐PC (4 h), and WPI‐I‐PC (c) showed a sequential decline in free amino groups relative to WPI‐PC. This decline results from the Maillard reaction between WPI and inulin, which consumed free amino groups. As the reaction time was extended, the extent of the Maillard reaction increased, leading to a further reduction in free amino group content. ³¹ WPI‐I‐PC (c) showed the lowest free amino group content due to ultrasound‐induced protein structural modification, which enhanced amino group accessibility and subsequent PC binding. ⁵ Among the covalent complexes, the amino content of WPI‐I‐PC (4 h) and WPI‐I‐PC (c) was higher at pH 9 than at pH 10 and 11 (P < 0.05), indicating incomplete reaction at pH 9. Across all pH conditions, WPI‐I‐PC complexes exhibited lower free amino group content than WPI‐PC, due to partial consumption during WPI‐I conjugation.

Particle size

Non‐covalent complexes exhibited significantly larger particle sizes than covalent complexes (P < 0.05; Fig. 1(D)), likely due to reduced protein solubility and aggregation induced by polyphenols at neutral pH. Under alkaline conditions, however, covalent bond formation between hydrophilic polyphenol groups and proteins enhanced solubility, thereby reducing particle size. ²⁵ Guo et al. ³² also reported that the whey protein–quercetin–junction gel ternary complex formed in alkaline conditions displayed markedly smaller particle sizes than those formed under neutral conditions. Among the non‐covalent complexes, the particle size followed the order: WPI‐I‐PC (4 h) > WPI‐I‐PC (c) > WPI‐PC ≈ WPI‐I‐PC (2 h). In covalent complexes, WPI‐PC and WPI‐I‐PC (2 h) displayed a slight size increase with rising pH, attributable to the crosslinking and oxidation of polyphenols in covalent protein–polyphenol complexes promoting particle size growth and fostering a stable network structure. ¹³ Conversely, WPI‐I‐PC (c) exhibited smaller particle sizes, as ultrasonic cavitation mechanically disrupts aggregates and promotes uniform dispersion. ³³

Surface hydrophobicity

Here, H ₀ was measured using ANS fluorescence, ³⁴ with native WPI displaying a value of 1356.00 ± 1.64. As shown in Fig. 1(E), PC binding significantly reduced the H ₀ of all samples (P < 0.05), as hydrophilic hydroxyl groups of PC shielded hydrophobic protein regions and limited protein unfolding. ³³ Covalent complexes exhibited markedly reduced H ₀ relative to non‐covalent complexes (P < 0.05), consistent with greater masking of hydrophobic sites by covalently bound PC. ³⁵ In non‐covalent complexes, hydrophobic interactions between aromatic rings of PC and protein residues also reduced ANS accessibility. ²⁹ Similarly, the H ₀ of WPI covalently conjugated with EGCG/quercetin/apigenin/naringenin was lower than that of unmodified WPI. ³⁶ The minimum H ₀ occurred at pH 10, reflecting a tighter binding between WPI and PC, while a slight increase at pH 11 likely resulted from alkaline‐induced protein unfolding. WPI‐I‐PC (c) exhibited higher H ₀ than WPI‐PC, presumably due to ultrasound‐driven exposure of hydrophobic groups. ³⁰ In contrast, WPI‐I‐PC (2 h) and WPI‐I‐PC (4 h) complexes demonstrated lower H ₀ than WPI‐PC at pH 9 and pH 10 but higher values at pH 11, possibly linked to pH‐dependent changes in inulin solubility. ³⁷

FTIR analysis

The FTIR spectra obtained at different pH values are shown in Figure 2. The primary FTIR bands of WPI are centered at 3400–3300 cm⁻¹ (amide A band, O–H and N–H stretching vibrations), 1700–1600 cm⁻¹ (amide I band, C=O stretching vibrations), and 1550–1450 cm⁻¹ (amide II band, N–H bending and C–N stretching). ³⁸ After the addition of PC, the amide A band (except for the WPI‐I‐PC (c)) and amide I band of the complexes exhibited blue shifts and red shifts to varying degrees, respectively, suggesting the participation of C=O, N–H, and C–N groups within the peptide backbone, further confirming the modification of the secondary conformation of WPI‐I‐PC. ³⁹ Relative to non‐covalent complexes, covalent WPI‐I‐PC complexes underwent blue shifts at 3400–3300 cm⁻¹ and 1100–1000 cm⁻¹ regions, and a red shift in the amide I band, thereby confirming covalent bond formation between PC and WPI‐I.

Figure 2.

FTIR spectra of WPI‐PC and WPI‐I‐PC complexes.

For secondary structure analysis, the amide I region was fitted using Fourier self‐deconvolution and Gaussian functions, with the resulting curves presented in Supporting Information, Fig. S1(A)–(D). Protein secondary structure was assigned based on its individual positions. ³⁶ The secondary structure composition of the complexes is shown in Supporting Information, Table S1. PC binding reduced the proportions of α‐helix, β‐sheet, and random coil structures, while increasing the β‐turn content. Among protein secondary structures, the α‐helix and β‐sheet represent non‐flexible internal structures of the polypeptide chain and are associated with conformational stability, whereas β‐turn and random coil reflect flexible segments. ⁴⁰ Accordingly, it could be concluded that the conformation of WPI became looser after binding with PC. The alterations were more pronounced in covalent than in non‐covalent complexes, underscoring the stronger influence of covalent conjugation on protein secondary structure.

Thermal stability

The thermal stability of the complexes was assessed by DSC (Fig. 3). Compared with the T _p values of WPI (101.2 °C), WPI‐I (2 h) (111.9 °C), WPI‐I (4 h) (115.1 °C), and WPI‐I (c) (108.4 °C) from our previous studies, ⁷ the T _p of the complexes after reacting with PC increased: for WPI‐PC to 105.4 °C, for WPI‐I‐PC (2 h) to 114.4 °C, for WPI‐I‐PC (4 h) to 116.8 °C, and for WPI‐I‐PC (c) to 116.6 °C, indicating an improvement in thermal stability. ⁴¹ Notably, the covalent complexes demonstrated a higher T _p relative to the non‐covalent complexes, implying that covalent bonding more effectively enhanced thermal stability. The thermal stability of covalent complexes increased with rising pH, suggesting tighter binding under alkaline conditions. Compared with WPI‐PC, the WPI‐I‐PC complexes demonstrated superior thermal stability, likely due to steric hindrance from the conjugated polysaccharide, which restricted protein unfolding. ²⁵ Similarly, EGCG could elevate the T _p of β‐lactoglobulin‐exopolysaccharide. ³⁹ Among all samples, WPI‐I‐PC (c) achieved the highest T _p (peaking at pH 11), indicating the best thermal stability. As shown in Table S2, the covalent complexes exhibited higher ΔH values compared to the non‐covalent complexes, with ΔH increasing at higher pH, implying a greater energy requirement for their unfolding.

Figure 3.

DSC curves of WPI‐PC and WPI‐I‐PC complexes at different pH.

Emulsifying properties

The EAI and ES of the complexes are shown in Fig. 4(A),(B). Consistent with previous studies from our laboratory, the incorporation of PC significantly improved both the EAI and ES of the complexes relative to WPI and WPI‐I. ⁷ Similarly, Meng and Li ⁸ found that the binding of WPI to gallic acid (GA)/CA/EGCG could improve the emulsifying capacity of WPI. Covalent complexes demonstrated superior EAI to non‐covalent complexes, as the covalent linkage between PC and WPI promoted hydrophobic group exposure, thereby enhancing surface activity and emulsification capacity.29, 33 Among non‐covalent complexes, WPI‐I‐PC (4 h) showed the highest EAI. In covalent complexes, the EAI of samples slightly decreased at pH 11, possibly due to alkaline‐induced protein denaturation. Overall, ternary complexes (WPI‐I‐PC) had superior EAI to binary complexes (WPI‐PC), with the highest value observed for WPI‐I‐PC (4 h).

Figure 4.

The EAI (A), ES (B), ABTS radical scavenging activity (C), and DPPH radical scavenging activity (D) of WPI‐PC and WPI‐I‐PC complexes at different pH.

However, while non‐covalent binding enhanced ES, covalent binding reduced it slightly. This may be due to differing effects on interfacial charge: non‐covalent interactions can increase electrostatic repulsion between droplets, whereas covalent conjugation may weaken it. ⁴⁰ In the covalent complexes, ES peaked at pH 11 and decreased at pH 9 and 10, showing an opposite trend to that of EAI. Overall, WPI‐I‐PC (c) achieved the highest ES, attributed to ultrasound‐assisted reduction in particle size and formation of a denser, more uniform emulsion microstructure. ³³

Antioxidant capacity

As shown in Fig. 4(C),(D), covalent complexes demonstrated significantly greater ABTS and DPPH free radical scavenging capacities compared to non‐covalent complexes, due to the more stable binding and greater number of accessible sites provided by covalent linkages. ⁴⁰ Vasava et al. ¹⁰ also demonstrated that the covalent conjugation of WPI with caffeic acid/ferulic acid enhanced antioxidant capacity relative to non‐covalent binding. While ABTS free radical scavenging did not differ significantly with pH (P > 0.05), DPPH free radical scavenging at pH 9 and 10 was higher than at pH 11 (P < 0.05). WPI‐I‐PC (2 h), WPI‐I‐PC (4 h), and WPI‐I‐PC (c) complexes all displayed the highest ABTS and DPPH free radical scavenging activities at pH 10, corresponding to their highest grafting degree and thus the most available phenolic hydroxyl groups. Furthermore, WPI‐I‐PC exhibited stronger antioxidant activity than WPI‐PC. This enhancement could be explained by both the antioxidant melanoidins and other compounds generated during the WPI‐I Maillard reaction and the higher bound PC content in the ternary structure. ⁴² Similarly, the myofibrillar protein–dextran–EGCG ternary complex exhibited superior antioxidant capacity to the myofibrillar protein–dextran binary complex. ¹⁴ This result confirms the crucial role of phenolic hydroxyl groups in enhancing antioxidant activity.

Application of WPI‐I‐PC covalent complex in Pickering emulsion

Particle size and FI of Pickering emulsion

Most samples, with the exception of WPI‐PC (c) and WPI‐I‐PC (c), formed markedly larger particles than the WPI‐stabilized emulsion, as shown in Fig. 5(A) (P < 0.05). The addition of inulin led to a thicker interfacial layer and consequently larger particle sizes, ⁴³ whereas ultrasonication counteracted this effect by disrupting aggregates. ¹⁹

Figure 5.

Particle size (A), flocculation index (B), and interfacial protein adsorption content (C) of emulsions stabilized by complexes; apparent viscosity of emulsions varies with shear rate (D); variation of G' (E) and G" (F) with oscillation frequency of the emulsion. Different letters denote significant differences between groups (P < 0.05).

The FI reflected the extent of droplet aggregation (Fig. 5(B)). The WPI‐stabilized emulsion exhibited the highest FI (32.00%), while systems stabilized with WPI‐I and WPI‐I‐PC complexes showed significantly reduced FI values (P < 0.05), confirming that covalent modification with inulin and/or PC effectively enhanced the stability of the emulsion. The complexes could maintain emulsion stability and inhibit aggregation by providing steric repulsion and electrostatic interactions. ⁴⁴ Notably, the WPI‐I‐PC (4 h) and WPI‐I‐PC (c) complexes exhibited the lowest FI (7.92% and 8.24%, respectively), confirming that ternary complexes form more stable emulsions than binary systems.

Interfacial protein adsorption content of Pickering emulsion

The interfacial protein content reflects the extent of protein adsorption at the oil–water interface. Figure 5(C) shows that both binary and ternary complex‐stabilized emulsions displayed markedly higher interfacial protein adsorption than the WPI‐stabilized system (76.80%) (P < 0.05), with WPI‐I‐PC (4 h) (84.92%) and WPI‐I‐PC (c) (84.37%) reaching the highest levels. The enhanced emulsifying activity likely resulted from improved complex solubility via covalent conjugation between WPI and inulin, and from PC‐induced hydrophobic group exposure, thereby enhancing the affinity of the complexes for the oil phase and accelerating interfacial adsorption. ⁴⁵ Thus, the ternary complexes improved emulsion stability by promoting interfacial protein adsorption.

Rheological property of Pickering emulsion

All emulsions exhibited shear‐thinning behavior, with apparent viscosity decreasing at higher shear rates (Fig. 5(D)). In the initial stage, the emulsions displayed elevated viscosity due to disordered, dense droplet packing. With increasing shear, structural deformation and droplet disruption occurred, resulting in progressive viscosity reduction. Ultimately, the disruptive effect of shear exceeded the inter‐droplet interactions. ⁴⁶ The emulsions stabilized by WPI‐I‐PC ternary complexes had higher apparent viscosity than those stabilized by WPI‐I binary complexes, which in turn exceeded the apparent viscosity of WPI‐stabilized emulsions. This increase is attributed to the denser droplet network formed by the polysaccharide and polyphenol components. Under a shear rate of 100 s⁻¹, the apparent viscosities of all emulsions were similar. However, emulsions stabilized with ternary complexes maintained greater viscosity than binary or WPI‐stabilized systems, thus possessing superior physical stability against shear and oscillation. This reflects stronger internal structural interactions and greater flow resistance in the WPI‐I‐PC ternary complexes. ⁴⁷ This is attributed to enhanced structural homogeneity through hydrogen bonding induced by the polyphenol. ⁴⁸

The viscoelastic properties were assessed through frequency sweep tests. As shown in Fig. 5(E),(F), both G′ and G′′ showed progressively enhanced values at higher frequencies. Across the 0.1–1 rad s⁻¹ range, the G′′ value slightly exceeded the G′ value, while G′ dominated between 1 and 100 rad s⁻¹, demonstrating predominantly elastic behavior in all emulsion systems. ⁴⁹ Ternary complexes exhibited a higher G′ than both binary complexes and WPI, likely due to the covalent grafting of WPI with inulin and PC, which strengthened interfacial and emulsifying properties, and promoted tighter droplet packing and crosslinking, thereby strengthening the emulsion network structure. ⁵⁰ This covalent conjugation facilitates the formation of an elastic‐dominant network during homogenization, thereby improving long‐term storage stability. ⁴⁷

CLSM analysis

All emulsions exhibited an oil‐in‐water (O/W) structure, containing evenly distributed oil droplets within the continuous protein‐based aqueous phase (Fig. 6). The relatively homogeneous droplet size distribution indicated good system stability. ⁵¹ In the overlaid dual‐color fluorescence images of FITC (green) and Nile Red (red), a yellow ring layer covering the periphery of oil droplets was observed, confirming the interfacial adsorption of WPI, WPI‐I, and WPI‐I‐PC at the oil–water interface, which hindered droplet coalescence and enhanced stability. ⁴⁵ Notably, WPI‐I (c) and WPI‐I‐PC (c)‐stabilized emulsions displayed slightly smaller droplet sizes than other groups, potentially related to the structural refinement induced by ultrasonic cavitation.

Figure 6.

CLSM images of emulsion stabilized complexes. I represents the protein phase, II represents the oil phase, III represents the mixed phase.

Stability of Pickering emulsion

Particle size was applied to investigate the pH stability of emulsions (Fig. 7(A),(B)). For binary complex‐stabilized emulsions, size increased markedly near the IP of WPI (pH 4–5) due to reduced electrostatic repulsion and protein solubility. ⁵² The addition of inulin reduced the particle size at pH 4–5, as covalent conjugation enhanced complex solubility. ²⁵ Ternary complex‐stabilized emulsions showed maximal particle size at pH 2, as excessively acidic conditions caused denaturation of the protein–polyphenol complexes, triggering droplet aggregation and phase separation. Within the pH range of 5–8, the emulsions maintained a stable particle size (P > 0.05), demonstrating that PC addition effectively prevented flocculation near the IP of the protein.

Figure 7.

Particle size changes in response to environmental stresses: pH (A: binary, B: ternary), high temperature (C: binary, D: ternary), and salt ion concentration (E: binary, F: ternary). Different letters denote significant differences between groups of the same conjugates (P < 0.05) (Same in Fig. 8).

Thermal stability was assessed by monitoring changes in particle size after heat treatment. As shown in Fig. 7(C),(D), all emulsions exhibited increased particle size following high‐temperature treatment in a water bath at 70–100 °C for 20 min, with a more significant change observed at higher temperatures. This is attributed to heat‐induced protein structural changes and internal group exposure, enhancing intermolecular hydrophobic interactions and consequently inducing emulsion flocculation and precipitation.46, 53 Among these, binary complex‐stabilized emulsions showed smaller particle size changes than WPI‐stabilized emulsions, as the steric hindrance effect between polysaccharide chains and protein peptide chains prevented the stretching of WPI peptide chains and improved thermal stability. ⁵⁴ Notably, the thermal stability of emulsions stabilized by ternary complexes was more prominent compared to the binary complexes, owing to their tighter structure. These findings aligned with DSC results, where ternary complexes had the highest denaturation temperature, confirming their optimal thermal stability.

The addition of salt is commonly used to strengthen food flavor during production. The stability of emulsions under varying NaCl concentrations was assessed via particle size (Fig. 7(E),(F)). As presented in Fig. 7(E), binary complex‐stabilized emulsions exhibited gradual particle size growth with rising NaCl concentration. This can be explained by the electrostatic shielding effect of salt ions through diminished inter‐droplet repulsion. ⁵⁵ Furthermore, salt ions altered the strength of hydrophobic interactions of proteins at the oil–water interface, further inducing droplet flocculation. Notably, WPI‐I (4 h)‐stabilized emulsions showed improved stability and reduced particle size variation compared to WPI‐stabilized systems. Ternary complex‐stabilized emulsions exhibited superior salt tolerance, with minimal size change or even reduction at low NaCl levels (Fig. 7(F)). However, at 400 mmol L⁻¹ NaCl, the particle size increased sharply (P < 0.05), likely from disruption of the hydration layer. ⁵⁶ Different WPI‐I‐PC complex‐stabilized emulsions exhibited largely consistent particle size variation trends. Overall, both binary and ternary complexes improved the salt ion stability of emulsions, with the ternary complexes demonstrating superior performance.

Oxidative stability of Pickering emulsion

Peroxide value (POV) was measured to assess primary lipid oxidation in the emulsions (Fig. 8(A),(B)), while thiobarbituric acid reactive substances (TBARS) were analyzed for secondary oxidation products (Fig. 8(C),(D)). ⁵⁷ The results demonstrated that during the accelerated oxidation tests at 37 °C and 50 °C, the POV of all emulsions increased significantly with prolonged storage time. The POV of emulsions stored at 50 °C was higher than that at 37 °C, consistent with the accelerated lipid oxidation at elevated temperatures. WPI‐stabilized emulsions showed the highest POV, while binary and ternary complex‐stabilized emulsions exhibited significantly lower values. This reduction is attributed to the formation of a denser interfacial film that limits oxygen access,58, 59 and improved emulsion stability that reduces oil‐droplet exposure. Moreover, the inherent antioxidative capacity of WPI‐I Maillard reaction products contributed to diminished lipid oxidation in binary complex‐stabilized emulsions. Among these, emulsions stabilized by ternary complexes exhibited superior oxidation stability, with WPI‐I‐PC (4 h) and WPI‐I‐PC (c) showing the lowest POV. This is presumably because the covalent conjugation of PC with WPI and inulin strengthened the overall antioxidant capacity of the complexes, effectively inhibiting the progression of oxidation. Similarly, the TBARS trend paralleled the POV trend. After 2 days of storage, all emulsions exhibited comparable oxidation levels (P > 0.05), but differences emerged after 4 days. Consistent with the POV results, emulsions stabilized by binary and ternary complexes had lower oxidation degrees compared to those stabilized by WPI. Overall, the ability of PC to significantly enhance antioxidant activity and inhibit lipid oxidation was demonstrated by the superior antioxidant effects of ternary complexes over binary complexes.

Figure 8.

Changes in POV (A, B) and TBARS (C, D) in emulsions over a 10‐day storage period at 37 °C and 50 °C.

CONCLUSION

In this study, WPI‐I‐PC ternary non‐covalent complexes and covalent complexes were prepared by an ultrasound‐assisted wet‐heating method based on WPI‐I binary complexes. The physicochemical properties, functional characteristics, and Pickering emulsion stabilization performance of the complexes prepared under different pH values, heating durations, and with or without ultrasound assistance were systematically investigated. The covalent binding between PC and WPI‐I was evidenced by the increased polyphenol content and decreased sulfhydryl content, free amino group content, particle size, and H ₀, with the most significant effects observed at pH 10. The WPI‐I‐PC covalent complexes exhibited superior thermal stability, emulsifying activity, and antioxidant capacity compared with non‐covalent complexes and binary systems, highlighting the critical role of PC in enhancing antioxidant capacity. In the Pickering emulsion system, the ternary complexes acted as superior stabilizers, demonstrating higher interfacial protein adsorption, lower FI, and increased apparent viscosity, along with typical shear‐thinning behavior and an elastic gel structure (G' > G″). These emulsions maintained stability under different pH, thermal, and salt conditions, and exhibited low levels of lipid oxidation during storage. Among them, WPI‐I‐PC (4 h) and WPI‐I‐PC (c) demonstrated the best physicochemical stability, suggesting that the ultrasound‐assisted wet‐heating method is a highly efficient preparation technique. Collectively, this study provides new insights into the design of protein–polysaccharide–polyphenol ternary complexes and establishes an efficient ultrasound‐assisted strategy for fabricating high‐performance Pickering emulsion stabilizers, thereby contributing to the development of functional food ingredients. Future research could focus on emulsion systems loaded with lipophilic active ingredients to systematically evaluate their encapsulation efficiency, stability, and gastrointestinal digestion properties, thereby expanding their application prospects in functional foods.

AUTHOR CONTRIBUTIONS

Jia‐Xin Lu: writing – original draft, investigation, data curation, formal analysis. Yu‐Qing Qin: writing – original draft, methodology, data curation, software. Jing‐Nan Ren: resources, investigation. Qi An: writing – review and editing, formal analysis. Li‐Fen Ma: conceptualization. Xi‐Wen Hu: methodology. Gang Fan: writing – review and editing, funding acquisition, project administration. Si‐Yi Pan: writing – review and editing, funding acquisition.

Supporting information

Figure S1. Curve‐fitted of the amide I region of WPI‐PC (A), WPI‐I‐PC (2 h) (B), WPI‐I‐PC (4 h) (C), and WPI‐I‐PC (c) (D) complexes at different pH.

Table S1. The secondary structure content (%) of WPI‐PC and WPI‐I‐PC complexes at different pH.

Table S2. The thermal property parameters of WPI‐PC and WPI‐I‐PC complexes at different pH.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.