Study on the inhibitory effects and mechanisms of four polyphenolic compounds from perilla extract on advanced glycation end products (AGEs)

Abstract

This study investigates the inhibitory effects and underlying mechanisms of major polyphenolic compounds in perilla extract on the formation of advanced glycation end products using a high-temperature in vitro glycation model. Ultra-performance liquid chromatography coupled with mass spectrometry identified quercetin, luteolin, rosmarinic acid, sinapic acid, and other representative bioactive constituents in PE. These polyphenolic compounds significantly inhibited protein-bound Nε-(carboxymethyl)lysine (41.3 % - 61.1 %), mainly through free radical scavenging, metal ion chelation, and suppression of protein aggregation, thereby mitigating protein oxidation and modulating protein conformation. Further mechanistic analysis revealed that these polyphenols regulated protein surface hydrophobicity, altered protein particle size, and influenced intermolecular interactions, thereby impacting the glycation process. Overall, this study elucidates the molecular mechanisms by which polyphenolic compounds inhibit AGEs formation under high-temperature conditions and provides a theoretical basis for the development of effective AGEs inhibition strategies in processed meat products.

Graphical abstract

Highlights

• •Polyphenolic compounds inhibit AGEs formation with an inhibition rate of up to 60 %.
• •Polyphenolic compounds exhibit significant anti-protein aggregation activity.
• •Hydrophobic and molecular interactions play a crucial role in inhibiting the glycation process.

1. Introduction

Advanced glycation end products (AGEs) are harmful compounds generated through a series of non-enzymatic reactions between the carbonyl groups of reducing sugars and the free amino groups of proteins, initially forming unstable Schiff bases. These Schiff bases subsequently undergo Amadori rearrangement to produce Amadori products, which further degrade into reactive dicarbonyl compounds, such as glyoxal (GO) and methylglyoxal (MGO). These intermediates ultimately react with amino groups to form AGEs (Han et al., 2025). Protein-rich foods such as soybeans, fish, and meat (especially processed foods like braised duck, red-cooked pork, and fried chicken), are prone to AGE formation during high-temperature cooking processes like frying, baking, and roasting. During these processes, proteins react with sugars, thereby diminishing food quality. More importantly, long-term consumption of foods with high AGE content can lead to their accumulation in the human body, increasing the risk of chronic diseases such as diabetes, Alzheimer's disease, and cardiovascular disorders (Twarda-Clapa et al., 2022). Therefore, inhibiting the formation of AGEs in food has become a crucial strategy for mitigating related health risks and preventing associated diseases.

In recent years, natural plant-derived substances have garnered considerable attention in food processing, particularly for their roles in inhibiting the formation of hazardous compounds such as AGEs and heterocyclic amines. For example, vegetable extracts (e.g., celery, carrot, and yam extracts) have been shown to effectively inhibit AGE and heterocyclic amines formation in grilled mackerel, primarily through their free radical scavenging properties (Zhang et al., 2023). Similarly, Xu et al. (2024) demonstrated that the addition of citrus peel extract to grilled pork patties significantly suppressed lipid and protein oxidation by scavenging free radicals, thereby reducing the generation of heterocyclic amines and AGEs. Perilla, a widely used medicinal and edible plant, is rich in various bioactive constituents, including polyphenols, flavonoids, terpenoids, and aromatic compounds (Yu et al., 2017). Compared to other plant extracts, perilla extract (PE) exhibits multiple beneficial properties, including anti-allergic, anti-inflammatory, antioxidant, anticancer, antibacterial, and antidepressant effects, making it a promising candidate for further investigation (Zeng et al., 2024). These characteristics suggest that PE may serve as an effective inhibitor of harmful compound formation during the heating of meat products.

Although numerous studies have demonstrated the potential of plant-derived compounds in reducing AGE formation during food processing, two major challenges remain: identifying the specific inhibitory components and elucidating their mechanisms of action. First, due to the complexity and variability of plant metabolites, a comprehensive understanding of the chemical composition of these extracts is essential prior to their application in food systems (Zhang et al., 2023). Advanced analytical techniques, such as ultra-performance liquid chromatography-mass spectrometry (UPLC-MS), nuclear magnetic resonance, and high-resolution mass spectrometry, are commonly employed for the identification and quantification of bioactive compounds. For instance, Bakr et al. (2019) successfully employed high-resolution high-performance liquid chromatography-mass spectrometry to identify alkaloid constituents in Erythrina × neillii leaf extract. Second, the complexity of thermal food processing introduces changes in protein properties, making it difficult to directly assess AGE formation in real food systems. To address this, in vitro glycation models have been widely adopted to investigate the inhibitory potential of plant-derived compounds under controlled conditions. For example, Zhu et al. (2021) demonstrated that catechins could inhibit protein aggregation in an in vitro glycation model by binding to lysine and arginine residues and scavenging free radicals.

Accordingly, the objective of this study was to investigate the inhibitory effects of PE on AGE formation in duck meat products and to elucidate the underlying mechanisms involved. The composition of PE was first characterized using UPLC-MS to identify its major bioactive components. Subsequently, the inhibitory effects of four representative polyphenolic compounds in PE on AGE formation were systematically evaluated by quantifying Nε-(carboxymethyl)lysine (CML), Nε-(carboxyethyl)lysine (CEL), α-dicarbonyl compounds, Amadori products, glucose substitution degree, and protein solubility. Furthermore, the inhibition mechanisms were explored using Fourier-transform infrared spectroscopy (FTIR), particle size analysis, zeta potential measurements, antioxidant activity assays, and molecular interaction analyses. This research provides a theoretical foundation and technical guidance for the development of nutritious, healthy, and flavorful braised meat products with reduced levels of harmful compounds.

2. Materials and methods

2.1. Materials

Duck breast meat was purchased from Huaying Food Co., Ltd. (Henan, China). Perilla extract was obtained from Daqin Plant Essence Company (Shanxi, China). Quercetin, luteolin, sinapic acid, potassium chloride, magnesium chloride, trichloroacetic acid, glucose, nitroblue tetrazolium, urea, Tris(hydroxymethyl)aminomethane (Tris), Ethylenediaminetetraacetic acid, β-mercaptoethanol (β-Me), ammonium persulfate, glycine, 5,5′-dithiobis-(2-nitrobenzoic acid), ferrous chloride, ferrozine, and aminoguanidine hydrochloride (AGH) were purchased from Shanghai Macklin Biochemical Technology Co., Ltd. (Shanghai, China). Rosmarinic acid (RosA), 4-phenylbutyric acid, stachydrine, Triton X-100, o-phenylenediamine, phenol, acetic acid, GO, and MGO were obtained from Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). 8-Anilino-1-naphthalenesulfonic acid (ANS) was purchased from Shanghai yuanye Bio-Technology Co., Ltd. (Shanghai, China). Sodium borohydride, chloroform, and sodium borate buffer solution were purchased from Beijing bailingwei Biotechnology Co., Ltd. (Beijing, China). C18 MCX solid-phase extraction cartridges were obtained from BKMAN Corporation (Hunan, China). Nε-carboxyethyllysine (CEL) and CML enzyme-linked immunosorbent assay (ELISA) kits were acquired from Shanghai Enzyme-linked Biotechnology Co., Ltd. (Shanghai, China).

2.2. Component analysis of PE using UPLC-MS

A 50 mg PE sample was added to 1200 μL of pre-cooled (at −20 °C) 70 % methanol-water internal standard extraction solution. The preparation of the internal standard solution involved dissolving 1 mg of the standard substance in 70 % methanol-water to create a 1000 μg/mL stock solution, which was further diluted with 70 % methanol to prepare a 250 μg/mL internal standard solution. The prepared sample solution was vortexed, centrifuged, and the supernatant was collected. It was then filtered through a 0.22 μm membrane and stored in an injection vial for UPLC-MS analysis according to the method of Bakr et al. (2019).

2.3. Establishment of the in vitro glycation model

Myofibrillar proteins were extracted from duck breast meat following the method described by Zhu et al. (2021) and subsequently dissolved in 20 mM PBS. The protein concentration was adjusted to 10 mg/mL using a biuret reagent kit (Solarbio Science & Technology Co., Ltd., Beijing, China).

The preparation of the protein-glucose glycation model was conducted based on the method of Abdallah et al. (2016), with slight modifications. Briefly, the 10 mg/mL protein solution was mixed with a 40 mg/mL glucose solution at a 1:1 (v/v) ratio. PE PE), azelaic acid, 4-phenylbutyric acid, stachydrine, quinine, quercetin, luteolin, RosA, and sinapic acid were then added to the mixture to achieve a final concentration of 0.5 mg/mL for each additive. The resulting mixture was homogenized thoroughly and subjected to heating at 200 °C for 270 s. After heating, the samples were cooled to room temperature to establish the in vitro glycation model. AGH at the same concentration was used as a positive control, while a parallel sample without any additives served as the blank control.

2.4. Determination of AGEs, intermediate products, and reactants

2.4.1. Measurement of free and bound (covalently binds to the ε-amino group of the lysine residue of the protein) CML and CEL

The quantification of free CML and CEL was primarily based on the method described by Niu et al. (2017) with slight modifications. Briefly, 1 mL of the glycation reaction mixture was combined with 10 mL of pre-cooled 5 % trichloroacetic acid in a 50 mL centrifuge tube. The mixture was vortexed for 10 min and centrifuged at 8000 ×g for 5 min to precipitate proteins. Subsequently, 5 mL of the resulting supernatant was transferred to a pre-activated MCX solid-phase extraction cartridge for affinity adsorption, enrichment, and purification. The concentrations of free and protein-bound CML and CEL in different groups were determined using commercial CML and CEL ELISA kits.

The measurement of bound CML and CEL was conducted following the method of Sun et al. (2021) with modifications. Specifically, 0.4 mL of the sample solution was accurately pipetted into a test tube, followed by the addition of 4 mL of sodium borate buffer (0.2 M, pH 9.2) and 0.8 mL of sodium borohydride (2 M). The mixture was incubated at 4 °C for 12 h. After the reaction, 3 mL of concentrated hydrochloric acid was added to acidify the mixture, followed by hydrolysis at 110 °C for 24 h. The hydrolysate was then diluted to a final volume of 8 mL, and 3 mL of the solution was loaded onto a pre-activated MCX extraction column for further enrichment and purification. The concentrations of CML and CEL were subsequently quantified using CML and CEL ELISA kits.

2.4.2. Determination of α-dicarbonyl compounds

The measurement of α-dicarbonyl compounds in the glycation model was performed according to the method described by Chu et al. (2023) using an Agilent 1260 High-performance liquid chromatography (HPLC) system (Agilent Technologies, USA). Briefly, 1 mL of the glycation reaction sample was mixed with 200 μL of o-phenylenediamine solution (5 mg/mL) and derivatized at 60 °C for 3 h. After filtration through a 0.22 μm organic phase membrane, the reaction solution was subjected to HPLC analysis. HPLC conditions were as follows: column temperature, 25 °C; injection volume, 20 μL; detection wavelength, 313 nm (DAD detector). The mobile phase consisted of acetonitrile (A) and 0.15 % acetic acid (B), with a flow rate of 0.8 mL/min. The gradient elution program was: 0–10 min, 80 %–40 % A; 10–12 min, 40 %–48 % A; 12–13 min, 48 %–60 % A; 13–15.5 min, 60 %–80 % A; and 15.5–20.5 min, 80 %–8 % A. A standard curve was constructed using GO and MGO standards at concentrations of 0–0.03 %, and the relative contents of GO and MGO in the samples were calculated based on their corresponding peak areas.

2.4.3. Determination of Amadori products

The measurement of Amadori products was conducted according to the method described by Zhao et al. (2022a). Briefly, 100 μL of the glycation reaction sample was mixed with 100 μL of a 0.25 mM nitroblue tetrazolium solution in 0.1 M sodium carbonate buffer. The mixture was incubated at 27 °C for 30 min, and the absorbance of the reaction solution at 530 nm was measured using a microplate reader (Varioskan Lux, Thermo Fisher Scientific, USA). The content of Amadori products in the sample was calculated using the following formula:

$$\text{Amadori products}\left(\text{nmol}/\mathrm{L}\right)=\mathrm{A}530\times 12.64$$ (1)

where: A530: Absorbance of the reaction solution at 530 nm; 12.64: Conversion factor (M⁻¹ × cm⁻¹).

2.4.4. Determination of glucose substitution degree and protein solubility

The glucose substitution degree was determined using the sulfuric acid-phenol method (Zeng et al., 2022). Briefly, 4 mL of the glycation reaction sample was mixed with 10 mL of a 15 % trichloroacetic acid, vortexed for 20 min, and centrifuged at 9500 ×g and 4 °C for 10 min. The supernatant was discarded, and the precipitate was washed once with 15 % trichloroacetic acid to remove unreacted reducing sugars. The precipitate was then dissolved in 20 mM PBS buffer, and 1 mL of the solution was transferred to a 10 mL centrifuge tube. Subsequently, 0.5 mL of 6 % phenol solution and 2.5 mL of concentrated sulfuric acid were added. The mixture was left to stand at room temperature for 20 min, after which the absorbance was measured at 490 nm. A standard curve was constructed using glucose solutions ranging from 0 to 0.1 mg/mL, following the same procedure. The standard curve equation was determined as y = 8.52× + 0.30 (R² = 0.99) (Supplementary material 3 A). The glucose substitution degree in the samples was calculated by substituting the absorbance values into the equation.

Protein solubility was measured according to the method described by Hu et al. (2022) with slight modifications. The protein concentration of the reaction solution was adjusted to 2.5 mg/mL and incubated at 4 °C for 1 h. After centrifugation at 6000 ×g for 10 min, the supernatant was collected, and the protein concentration was determined using a biuret reagent kit (Solarbio Science & Technology Co., Ltd., Beijing, China). Protein solubility was calculated using the following formula:

$$\text{Protein solubility}\left(\%\right)=\text{Supernatant protein concentration}/2.5\times 100\%$$ (2)

2.4.5. Scanning Electron microscopy (SEM)

The glycation reaction samples, freeze-dried for 48 h, were mounted on a sample holder and coated with gold using an ion sputter coater. The microstructure of the samples was observed under a scanning electron microscope (S-3400, Hitachi, Japan) at a magnification of 3500× and an accelerating voltage of 15 kV.

2.4.6. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE)

The glycation reaction sample (40 μL) was combined with 10 μL of 5× protein loading buffer (with or without β-Me), vigorously vortexed, and heat-denatured at 95 °C for 10 min in a metal bath. After cooling to room temperature, the mixture was centrifuged at 4000 ×g for 5 min, and the supernatant was collected. An 8 μL sample was loaded onto a 12 % separating gel with a 5 % stacking gel. Electrophoresis was initially performed at 80 V for 30 min, followed by an increase to 120 V for 90 min. The gel was stained with Coomassie Brilliant Blue staining solution and destained with destaining solution. The electrophoresis results were imaged and analyzed using a BIO-RAD imaging system (Bio-Rad Laboratories, Inc., USA).

2.4.7. Determination of free thiol and carbonyl content

The free thiol content was measured according to the method described by Lv et al. (2022) with slight modifications. Briefly, 1 mL of the glycation reaction sample was mixed with 4.5 mL of Tris-glycine buffer (0.086 M Tris, 0.09 M glycine, 4 M ethylenediaminetetraacetic acid, pH 8.0) and 0.5 mL of Ellman's reagent solution (4 mg of 5,5′-dithiobis-(2-nitrobenzoic acid) dissolved in 1 mL of Tris-glycine buffer). The mixture was vortexed thoroughly and incubated at 25 °C for 30 min. The absorbance was measured at 412 nm, with 1 mL of PBS buffer used as the control. The free thiol content was calculated using the following formula:

$$\text{Free thiol content}\left(\text{nmol}/\mathrm{mg}\right)=\frac{{\mathrm{A}}_{412}}{13600\times \text{Protein concentration}}\times \text{Dilution factor}\times {10}^6$$ (3)

The carbonyl content in the glycation reaction samples was determined using a protein carbonyl content assay kit (Solarbio Science & Technology Co., Ltd., Beijing, China). The results were expressed as μmol/mg of protein.

2.5. Investigation of inhibition mechanisms

2.5.1. Zeta potential and particle size

The protein concentration in the glycation reaction samples was adjusted to 1 mg/mL using a biuret reagent kit. Subsequently, 2 mL of the sample was analyzed using a Zetasizer Lab particle size analyzer (Malvern Panalytical Ltd., UK) to measure the particle size and Zeta potential.

2.5.2. Surface hydrophobicity

The surface hydrophobicity (H₀) of proteins in the glycation reaction system was determined using the ANS fluorescence probe method according to Xue et al. (2018), which is based on the principle that the fluorescence intensity of ANS increases upon binding to exposed hydrophobic regions of proteins via hydrophobic interactions. The protein concentration of the glycation reaction samples was adjusted to 0.01, 0.05, 0.1, 0.5, and 1 mg/mL. Subsequently, 4 mL of each sample was transferred to a 10 mL brown centrifuge tube, and 20 μL of 5 mM ANS solution was added. After thorough vortexing, the fluorescence spectra were measured using an F-4700 fluorescence spectrophotometer (Hitachi High-Technologies Corporation, Japan) with an excitation wavelength of 390 nm and an emission range of 400–600 nm The fluorescence intensity at excitation /emission wavelength = 390/470 nm was plotted against the protein concentration in the samples, and the initial slope of the curve was determined as the H₀.

2.5.3. FTIR

A mixture of 1 mg of freeze-dried glycation reaction sample and 100 mg of potassium bromide was ground thoroughly and pressed into a transparent pellet using a pellet press. The pellet, required to be transparent and free of cracks, was placed in a sample holder for analysis using a Nicolet iS20 Fourier transform infrared (Thermo Fisher Scientific, USA). Spectra were recorded in the full wavenumber range (4000–400 cm⁻¹), with potassium bromide serving as the background. Chemical bond changes were analyzed based on the obtained spectra. The infrared spectra were processed and deconvoluted using OMNIC software and PeakFit 4.12. The protein secondary structure was further analyzed in the amide I region (1600–1700 cm⁻¹), following the method described by Wang & Xie et al. (2019).

2.5.4. Determination of antioxidant capacity and iron ion chelation ability

The total antioxidant capacity of the glycation reaction system was measured using a total antioxidant capacity assay kit (Nanjing Jiancheng Bioengineering Institute, China). Briefly, the DPPH working solution was mixed with the glycation reaction sample and allowed to stand for 30 min. The absorbance was then measured at 405 nm. A standard curve was constructed using the same procedure with standard solutions. The total antioxidant capacity of the samples was calculated based on the absorbance at 405 nm and the standard curve.

The iron ion chelation ability was determined according to the method described by Zhao et al. (2022a). Briefly, 1 mL of the glycation reaction sample was mixed sequentially with 4 mL of deionized water, 0.1 mL of 2 mM ferrous chloride solution, and 0.2 mL of 5 mM ferrozine solution. The mixture was vortexed thoroughly and incubated at room temperature for 10 min. The absorbance of the solution was measured at 562 nm, and the iron ion chelation ability of the glycation reaction samples was indirectly reflected by the absorbance value.

2.5.5. Protein-protein interactions

Protein-protein interactions were assessed following the method described by Jin et al. (2023) with slight modifications. Briefly, 0.5 mL of the glycation reaction sample was mixed separately with five solutions: 0.05 M sodium chloride solution (S1), 0.6 M sodium chloride solution (S2), 0.6 M sodium chloride solution with 1.5 M urea (S3), 0.6 M sodium chloride solution with 8 M urea (S4), 0.6 M sodium chloride solution with 8 M urea with 0.5 M β-Me (S5). Each mixture was vortexed for 1 min, incubated at 4 °C for 1 h, followed by centrifugation at 10000 ×g and 4 °C for 10 min. Protein concentrations in the supernatants (A1 to A5) were measured. The differences in protein concentrations (A2–A1, A3–A2, A4–A3, and A5–A4) were used to evaluate the strengths of ionic bonds, hydrogen bonds, hydrophobic interactions, and disulfide bonds, respectively, in the glycation reaction samples.

2.5.6. Interactions between polyphenolic compounds and proteins

The interactions between polyphenolic compounds and proteins were investigated according to the method described by Hu et al. (2022) with slight modifications. Briefly, 0.5 mL of quercetin, luteolin, RosA, or sinapic acid was added to a 1 mg/mL protein solution in triplicate, achieving final concentrations of 0.2, 0.4, 0.6, 0.8, and 1.0 mM. The mixtures were incubated at 5 °C, 15 °C, and 25 °C for 30 min. The fluorescence intensity of the proteins was scanned using a microplate reader at Ex/Em of 280/310–500 nm. The protein fluorescence quenching mechanism was determined using the Stern-Volmer equation. The Stern-Volmer quenching constant (Ksv), quenching rate constant (Kq), binding constant (Ka), and number of binding sites (n) were calculated using the double logarithmic equation. Additionally, thermodynamic parameters were characterized using the Van't Hoff equation and thermodynamic equations, and the driving forces for the binding between duck myofibrillar protein and polyphenolic compounds were determined.

$$\frac{{\mathrm{F}}_0}{\mathrm{F}}=1+{\mathrm{K}}_{\mathrm{q}}{\mathrm{\tau }}_0\left[\mathrm{Q}\right]=1+{\mathrm{K}}_{\mathrm{sv}}\left[\mathrm{Q}\right]$$ (4)

$$\log \left(\frac{{\mathrm{F}}_0-\mathrm{F}}{\mathrm{F}}\right)=\log {\mathrm{K}}_{\mathrm{a}}+\text{nlog}\left[\mathrm{Q}\right]$$ (5)

$$\ln {\mathrm{K}}_{\mathrm{a}}=-\frac{\mathrm{\Delta H}}{\mathrm{RT}}+\mathrm{C}$$ (6)

$$\mathrm{\Delta G}=-\text{RTln}{\mathrm{K}}_{\mathrm{a}}$$ (7)

$$\mathrm{\Delta G}=\mathrm{\Delta H}-\mathrm{T\Delta S}$$ (8)

where, F₀ and F represent the fluorescence intensities in the absence and presence of the quencher Q, respectively; Q is the concentration of the quencher; τ₀ (10⁻⁸ s) is the average fluorescence lifetime of the protein; R is the gas constant (8.314 J/mol/K); T is the absolute temperature; ΔH and ΔS are the changes in enthalpy and entropy, respectively, during the binding process.

2.6. Molecular docking

Protein-polyphenol interactions were validated through molecular docking following the methodology of Zhang et al. (2024). The Myosin-9 protein sequence was retrieved from the UniProt database (ID: U3IIB9). As no experimentally resolved structure was available, the 3D structure was predicted using AlphaFold3 and subsequently energy-minimized with Rosetta Relax. Structural preparation was performed using PyMOL (Schrödinger), including water removal and hydrogen addition. The 3D structures of four polyphenolic compounds (quercetin, luteolin, RosA, and sinapic acid) were obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/). Blind docking was conducted using the CB-DOCK2 web server (https://cadd.labshare.cn/cb-dock2/), which employs: Artificial neural network-based cavity detection; Rigid docking with AutoDock Vina. The top-ranked complexes were selected based on binding affinity (ΔG) for further analysis. Protein-ligand interactions were analyzed using the PLIP web server (https://plip-tool.biotec.tu-dresden.de/), and 3D representations were generated with PyMOL.

2.7. Statistical analysis

All experiments were conducted with a minimum of 3 repetitions independently -1for data analysis. Findings are expressed as the mean ± standard deviation. ANOVA and Tukey comparison test were performed with a significant level of 0.05 for the significance analysis of data via the statistical SPSS software (SPSS 17, Inc., Chicago, IL, USA). Graphs were generated using Origin 8.0.

3. Results and discussion

3.1. PE composition analysis

UPLC-MS analysis of PE was conducted in both positive and negative ionization modes, with the corresponding total ion chromatograms shown in Supplementary material 1 A and 1B. Mass spectrometric analysis revealed that PE contains multiple classes of compounds, including 168 alkaloids, 336 organic acids, 118 flavonoids, 151 phenolic acids, along with various amino acids, nucleotides, and their derivatives. The proportions of these compounds are shown in Fig. 1A, while Fig. 1B displays the molecular structures of four representative polyphenolic compounds (quercetin, lignan, RosA and sinapic acid). Previous studies have demonstrated that substances with antioxidant activity and metal ion chelation capabilities can reduce protein conformational damage under high-temperature conditions, prevent protein cross-linking, and slow the formation of AGEs (Zhao et al., 2022). Although PE contains a variety of beneficial components, whether these components can inhibit glycation reactions remains to be clarified. Based on the UPLC-MS results, eight compounds from four major categories with relatively high proportions were selected to investigate their effects on glycation: phenolic acids (RosA and sinapic acid), flavonoids (quercetin and luteolin), organic acids (azelaic acid and 4-phenylbutyric acid), and alkaloids (stachydrine and quinine).

Fig. 1.

Four polyphenol structural formulaes of PE. (A) Main components of PE and their relative percentages.(B) Molecular structures of the four polyphenol compounds.

Numerous studies have shown that these compounds contribute to antioxidant activity, antibacterial effects, inhibition of harmful substances, and improvement of product quality. For example, He et al. (2022) demonstrated that the addition of 0.25 % RosA to fried meatballs reduced the total content of heterocyclic amines by 59 %. However, it remains unclear whether these eight compounds in PE can influence glycation reactions and inhibit AGEs production.

3.2. Effects of main components in PE on glycation reactions

3.2.1. Inhibitory effects of main components in PE on CML and CEL

CML and CEL are among the most prominent AGEs, and their levels are widely used as key indicators to reflect variations in overall AGEs content (Han et al., 2025). Based on the evaluation of the effects of four polyphenolic compounds at low (0.05 mg/mL), medium (0.5 mg/mL), and high (5 mg/mL) concentrations on CML and CEL formation in the glycation model (Supplementary Material 2), the medium concentration (0.5 mg/mL) exhibited the most pronounced inhibitory effect on both AGEs (P < 0.05). Therefore, 0.5 mg/mL was selected as the optimal concentration for subsequent mechanistic investigations. Fig. 2A and B present the effects of different compounds on the levels of both bound and free forms of CML and CEL in the in vitro glycation model. Compared to the control group, all tested compounds demonstrated varying degrees of inhibition. Specifically, quinine, quercetin, luteolin, RosA, and sinapic acid exhibited the most significant inhibitory effects on bound CML, with inhibition rates of 46.0 %, 41.3 %, 53.8 %, 60.5 %, and 61.1 %, respectively. These inhibition rates were notably higher than those reported for caffeic acid and dihydrocaffeic acid in the study by Zhang et al. (2024). In terms of CEL, quercetin, luteolin, RosA, and sinapic acid also displayed superior inhibitory effects compared to the AGH group, with inhibition rates of 9.8 %, 10.8 %, 11.8 %, and 12.7 %, respectively. While AGH inhibits AGEs by trapping reactive intermediates, these polyphenolic compounds exhibit multiple advantageous mechanisms. As demonstrated in this study (Fig. 2C), they not only effectively capture key intermediate dicarbonyl compounds (GO and MGO) but also exhibit potent intrinsic antioxidant activity (Fig. 4F), thereby protecting protein structural integrity. This multimodal action - combining carbonyl scavenging, antioxidant protection, and structural stabilization - highlights the superior inhibitory potential of polyphenolic compounds compared to single-mechanism agents like AGH. Although no significant inhibition was observed for free CML, its concentration was considerably lower than that of bound CML, suggesting that bound CML may serve as a more relevant indicator of inhibitory efficacy. Furthermore, considering the limited safety profile of quinine for food applications, quercetin, luteolin, RosA, and sinapic acid were selected for subsequent mechanistic studies on AGEs inhibition (Sun et al., 2021).

Fig. 2.

Effects of different compounds of PE treatments on AGEs formation and related factors in the glycation model. (A) Bound and free CML contents. (B) Bound and free CEL contents. (C) Amadori product content and GO levels. (D) Glucose substitution degree. (E) Protein solubility. Data are expressed as mean ± standard deviation. Different letters indicate statistically significant differences (P < 0.05).

Fig. 4.

Effects of four polyphenol compounds on protein structure, oxidative properties, and molecular interactions in the glycation model. (A) Free thiol group content. (B) Carbonyl content. (C) H₀. (D) Chemical bonds. (E) Secondary structure. (F) Total antioxidant capacity. (G) Metal ion chelation ability. (H) Interaction forces between MPs. Data are expressed as mean ± standard deviation. Different letters indicate statistically significant differences (P < 0.05).

3.2.2. Inhibitory effects of four polyphenolic compounds on α-dicarbonyl compounds and Amadori products

The standard curves for GO and MGO were constructed by plotting chromatographic peak areas against standard concentrations, yielding the equations y = 330,955× – 263 (R² = 0.98) for GO and y = 682,079× - 395 (R² = 0.99) for MGO (Supplementary material 3B, 3C). These calibration curves were subsequently used to quantify the contents of GO and MGO in the glycation model. Experimental results indicated that MGO levels were below the detection limit of the HPLC method and were therefore not detected. As shown in Fig. 2C, compared with the control group, both the GO peak area and Amadori product content were significantly reduced in the quercetin and luteolin treatment groups (P < 0.05). These findings confirm that quercetin and luteolin effectively inhibit the formation of Amadori products and scavenge intermediate dicarbonyl compounds such as GO, thereby attenuating AGEs accumulation. This inhibitory effect may be attributed to the strong free radical scavenging capacities of quercetin and luteolin during glycation (Wu & Yen, 2005). Additionally, luteolin has been reported to stabilize protein structures and prevent protein cross-linking, particularly in bovine serum albumin, further contributing to its antiglycation activity (Qin et al., 2021). Accordingly, luteolin exhibited the strongest inhibitory effect on both Amadori products and GO. In contrast, while RosA and sinapic acid significantly reduced Amadori product content, they demonstrated weaker abilities to trap dicarbonyl compounds, suggesting that their inhibitory effects primarily occur at the early stages of glycation. This distinction may be related to structural differences between flavonoids and phenolic acids, particularly regarding the number and position of phenolic hydroxyl groups (Zuo et al., 2018). Moreover, phenolic acids are more susceptible to thermal degradation, and their inhibitory effects on glycation are strongly concentration-dependent, with decreased concentrations leading to reduced efficacy (Zhao et al., 2022). Collectively, these results indicate that quercetin and luteolin primarily inhibit AGEs formation by scavenging dicarbonyl intermediates, whereas RosA and sinapic acid exert their effects by reducing Amadori product formation during the early phases of glycation.

3.2.3. Effects of four polyphenolic compounds on glucose substitution degree and protein solubility

The extent of glucose participation in the glycation reaction was assessed by measuring the glucose substitution degree, as illustrated in Fig. 2D. Compared with the control group, all treatment groups exhibited a reduction in glucose substitution degree, with the inhibitory efficacy ranking as follows: RosA > quercetin > sinapic acid > luteolin. These results indicate that all four polyphenolic compounds effectively reduced the involvement of reducing sugars in the glycation process, thereby limiting the reaction between reducing sugars and free amino groups and consequently inhibiting glycation progression. This inhibitory effect is likely attributable to the strong antioxidant capacities of these polyphenolic compounds, which enable the scavenging of reactive oxygen species generated during glycation, thereby preventing the accumulation of early-stage glycation products (e.g., Amadori products) and intermediates (e.g., GO) (Wu & Yen, 2005). Additionally, these compounds may suppress the autoxidation of reducing sugars under aerobic conditions, further limiting AGEs formation (Twarda-Clapa et al., 2022). These findings are consistent with the subsequent antioxidant activity results, further confirming the close relationship between antioxidant capacity and AGEs inhibition.

During glycation, the tertiary structure of proteins undergoes conformational changes under heating conditions, which in turn affects protein solubility. Elevated temperatures induce protein unfolding, exposing hydrophilic groups and increasing the likelihood of glycation with reducing sugars. This facilitates the formation of insoluble protein aggregates, leading to a decline in protein solubility. As shown in Fig. 2E, quercetin and luteolin treatments resulted in decreased protein solubility. This suggests that while these flavonoids may not significantly interfere with the direct interaction between proteins and reducing sugars, they likely exert inhibitory effects by interacting with protein molecules, scavenging dicarbonyl compounds, and reducing the availability of free amino groups for glycation reactions. These results are consistent with the findings of Cömert and Gökmen (2019). In contrast, RosA and sinapic acid significantly improved protein solubility, thereby alleviating the adverse effects of thermal processing on protein properties. Several mechanisms may account for this observation. First, RosA and sinapic acid possess strong free radical scavenging and metal ion chelation activities, which reduce protein oxidation, inhibit the formation of large protein aggregates, and prevent the reaction between lysine residues and reactive α-dicarbonyl compounds, thereby preserving protein solubility. Second, as depicted in Fig. 1B, RosA and sinapic acid contain multiple phenolic hydroxyl groups that can interact with proteins via hydrogen bonding and hydrophobic interactions (Fig. 4C and Table 2), masking exposed hydrophobic regions and enhancing solubility. Third, RosA exhibits particularly strong electrostatic stabilization due to its high negative charge density (Fig. 4C). This charge repulsion may further inhibit protein aggregation, complementing the protective effects of antioxidant activity and molecular interactions. However, further studies are required to validate these hypotheses and fully elucidate the underlying mechanisms responsible for the inhibitory effects of RosA and sinapic acid on protein aggregation and glycation.

Table 2.

Kq, Ka, and thermodynamic parameters (ΔH, ΔS, and ΔG) for the interactions between different polyphenolic compounds and duck meat proteins. Data are expressed as mean ± standard deviation.

Compounds	T (K)	Kq (1010L/mol.s)	Ka (L/mol)	∆G (kJ/mol)	∆H (kJ/mol)	∆S (J/K)
Quercetin	278	6.55 ± 0.17	3.39	−2.82	2.50	19.14 ± 0.03
	288	6.31 ± 0.07	3.50	−3.00
	298	8.60 ± 0.75	3.65	−3.21
luteolin	278	19.18 ± 3.83	9.98	−5.32	50.14	198.40 ± 1.61
	288	16.43 ± 0.08	14.13	−6.34
	298	23.96 ± 1.42	43.22	−9.33
RosA	278	31.14 ± 0.20	10,817.45	−21.47	48.06	251.34 ± 1.83
	288	33.63 ± 0.51	35,283.00	−25.07
	298	42.91 ± 3.30	43,197.88	−26.44
Sinapic Acid	278	15.80 ± 0.21	1426.26	−16.79	−26.09	−33.52 ± 0.10
	288	16.00 ± 0.50	939.94	−16.39
	298	20.60 ± 1.52	669.11	−16.12

3.2.4. SEM analysis

To further validate the inhibitory effects of the selected compounds on AGEs formation, the microstructural characteristics of the glycation products were examined using SEM. As shown in Fig. 3A, the SEM images of the in vitro glycation model revealed that the control group exhibited a microstructure characterized by numerous fine and densely distributed protein aggregates. In contrast, the treatment groups displayed a marked reduction in these fine granular structures, suggesting that quercetin, luteolin, RosA, and sinapic acid effectively inhibited protein aggregation. These microstructural observations, combined with the results for CML and CEL levels, provide additional evidence supporting the role of these polyphenolic compounds in mitigating AGEs formation under high-temperature conditions. The observed reduction in protein aggregation is consistent with the findings reported by Zhu et al. (2021), further confirming the inhibitory potential of these compounds against glycation-induced protein aggregation.

Fig. 3.

Effects of four polyphenol compounds on the basic characteristics of the glycation model. (A) SEM microstructure observation. (B) SDS-PAGE analysis under conditions with and without β-Me.

3.2.5. SDS-PAGE analysis

Fig. 3B presents the SDS-PAGE results under reducing (with β-Me) and non-reducing (without β-Me) conditions. The electrophoretic profiles were primarily composed of two major proteins—myosin heavy chain (MHC) and actin—along with several lower-intensity bands corresponding to troponin T, troponin I, troponin C, myosin light chain 1 (MLC-1), and myosin light chain 2 (MLC-2), consistent with the findings of Deng et al. (2021). Under reducing conditions, protein bands appeared more intense and well-defined, indicating that β-Me disrupted disulfide bonds, weakened intermolecular interactions, and promoted protein unfolding and degradation. Conversely, under non-reducing conditions, the proteins largely maintained their intact tertiary structures, and stronger intermolecular interactions resulted in fainter bands, underscoring the key role of disulfide bonds in both intramolecular and intermolecular protein cross-linking.

In general, narrowing, fading, or blurring of higher molecular weight protein bands, along with the emergence of lower molecular weight bands, is indicative of protein degradation. Under reducing conditions, a notable fading of the actin band was observed in the control group, likely attributable to protein degradation or reduced aggregation caused by glycation. This phenomenon also suggests that the molecular weight of AGEs may exceed that of native proteins (Li et al., 2025), or that glycation disrupts protein aggregation (Zhang et al., 2024). In the four treatment groups, the intensity of the troponin T band significantly decreased, while bands in the 11–25 kDa range became more pronounced. The reduction in high molecular weight proteins, coupled with the increased abundance of low molecular weight species, indicates enhanced protein degradation and prevention of intermolecular cross-linking. Collectively, these results demonstrate that quercetin, luteolin, RosA, and sinapic acid effectively inhibit the formation of high molecular weight protein aggregates, mitigate glycation-induced cross-linking, and thereby play a critical role in reducing AGEs accumulation (Sheng et al., 2016).

3.2.6. Analysis of changes in free thiol and carbonyl content

As shown in Fig. 4A, all treatment groups exhibited a significant increase in free thiol content compared with the control group (P < 0.05). Previous studies have demonstrated a positive correlation between free thiol content and protein oxidative stability, with higher thiol levels reflecting better preservation of protein structural integrity (Li et al., 2025). Therefore, the four compounds investigated in this study - quercetin, luteolin, RosA, and sinapic acid - may enhance protein oxidative stability by maintaining protein structural integrity, reducing the disruption of internal protein structure caused by environmental changes, thereby minimizing thiol exposure. These findings further suggest that these polyphenolic compounds possess potential protective effects against protein oxidative damage.

The analysis of protein carbonyl content revealed that the quercetin, luteolin, and RosA treatment groups significantly reduced carbonyl levels. Protein carbonylation is widely recognized as a key marker of oxidative protein modification, and its content is positively correlated with the extent of protein oxidation (Deng et al., 2021). These results are consistent with the free thiol content findings, further confirming that the selected polyphenols protect protein conformation and modulate the glycation process by mitigating protein carbonylation. Additionally, previous studies have demonstrated that elevated carbonyl content primarily arises from the conversion of amino acid residues to carbonyl derivatives or the cleavage of oxidized peptides, both of which are closely associated with the accumulation of dicarbonyl compounds (Yang et al., 2019). This correlation indirectly substantiates the capacity of these polyphenolic compounds to scavenge dicarbonyl intermediates and prevent glycation-induced oxidative damage.

3.3. Inhibition mechanism of AGEs by four polyphenolic compounds

3.3.1. Zeta potential and particle size analysis

It is generally accepted that protein particle size increases following glycation modification, primarily due to the attachment of sugars to the protein surface or cross-linking between protein molecules during the glycation process. These modifications increase the volume and mass of protein molecules, thereby resulting in larger particle sizes (Awasthi et al., 2019). As shown in Table 1, the effects of the four polyphenolic compounds on protein particle size in the glycation model were evaluated. Compared with the control group, significant reductions in particle size were observed in the luteolin, RosA, and sinapic acid treatment groups (P < 0.05), suggesting that these three polyphenolic compounds effectively inhibited protein cross-linking and contributed to their anti-AGEs activity. In contrast, a slight increase in particle size was detected in the quercetin treatment group. This increase may be attributed to the formation of copolymers resulting from the interaction between quercetin and reactive dicarbonyl compounds, which subsequently altered protein particle size. Additionally, quercetin may form aggregates with myofibrillar proteins, thereby occupying potential glycation sites and reducing interactions with reducing sugars, while simultaneously increasing the apparent particle size. This hypothesis warrants further validation through detailed molecular interaction studies.

Table 1.

Effects of different polyphenol treatments on the average particle size and Zeta potential of proteins in the glycation model. Data are expressed as mean ± standard deviation. Different letters indicate statistically significant differences (P < 0.05).

Groups	Average particle size (μm)	Zeta potential (mV)
Control	3308.33 ± 137.63ab	−15.24 ± 0.83a
Quercetin	3729.67 ± 247.56a	−16.88 ± 1.04ab
luteolin	2916.67 ± 255.38bc	−16.30 ± 0.37ab
RosA	2348.67 ± 387.02c	−17.05 ± 0.12b
Sinapic Acid	2656.67 ± 150.14c	−16.69 ± 0.21ab

Changes in Zeta potential reflect variations in protein surface charge and are indicative of protein structural stability (Zhu et al., 2021). Table 1 also presents the effects of the four polyphenolic compounds on the Zeta potential of proteins within the glycation model. Compared with the control, all four compounds led to increased Zeta potential values, suggesting enhanced surface charge stability. This increase in absolute Zeta potential may be attributable to the loss of positively charged lysine residues during glycation (Zhang et al., 2024). Accordingly, these polyphenolic compounds likely stabilize protein conformations by promoting electrostatic repulsion between protein molecules, thereby improving the overall stability of the glycation system and contributing to their anti-glycation effects. These findings are consistent with the structural stabilization effects observed in the SDS-PAGE analysis.

3.3.2. H₀ analysis

Glycation and polyphenol binding can modulate protein surface hydrophobicity, primarily through mechanisms associated with the exposure of hydrophobic amino acid residues and the degree of protein aggregation (Xue et al., 2018). As shown in Fig. 4C, the addition of four polyphenolic compounds—quercetin, luteolin, RosA, and sinapic acid—led to a significant reduction in protein surface hydrophobicity (P < 0.05). This observation suggests that these polyphenolic compounds may competitively occupy the ANS binding sites on the protein surface, thereby reducing ANS fluorescence. This finding is consistent with previous research by Zhang et al. (2024), who reported that caffeic acid and dihydrocaffeic acid similarly reduced protein hydrophobicity by competing with ANS for binding sites. Furthermore, the present study indicates that these polyphenolic compounds can stabilize protein structures by protecting hydrophobic regions, which are typically embedded within the protein's non-polar core, from exposure during heat-induced unfolding. By preventing the exposure of hydrophobic domains, these compounds effectively inhibit increases in protein hydrophobicity and reduce oxidative aggregation, thereby suppressing AGEs formation (Lv et al., 2022). This mechanism is further supported by the findings of Li et al. (2024), who reported that the addition of tea polyphenols to glycation systems decreased both surface hydrophobicity and β-sheet content, thereby limiting the accumulation of Maillard reaction intermediates and subsequent AGEs formation. These results underscore the critical relationship between protein surface hydrophobicity and AGEs accumulation. Notably, the reduction in H₀ was more pronounced in the flavonoid groups (quercetin and luteolin) than in the phenolic acid groups (RosA and sinapic acid) (P < 0.05). This difference may be attributed to the stronger metal ion-chelating abilities of flavonoids (Fig. 4G), which can further stabilize protein conformation. Additionally, the number and distribution of polar groups in polyphenolic compounds are closely associated with their influence on protein surface hydrophobicity (Ke & Li, 2024).

3.3.3. Protein conformation analysis

As illustrated the Fig. 4D, the FTIR spectra of the in vitro glycation model supplemented with different polyphenolic compounds exhibited similar overall profiles, indicating that no new covalent bonds were formed between the polyphenolic compounds and proteins. Within the 3200–3600 cm⁻¹ range, a prominent absorption peak corresponding to the amide A band was observed, primarily attributed to O—H stretching vibrations (Yin et al., 2025). Notably, quercetin, luteolin, and RosA induced a slight red shift accompanied by increased peak intensity in the amide A band, suggesting the formation of enhanced hydrogen bonding interactions. This enhancement may strengthen intra- or intermolecular hydrogen bonding within protein structures, thereby contributing to improved protein conformational stability, consistent with the observed reductions in protein carbonyl content.

The FTIR spectra also included characteristic absorption bands of proteins: the amide I band (1600–1700 cm⁻¹), amide II band (1500–1600 cm⁻¹), and amide III band (1200–1350 cm⁻¹). Among them, the amide I band provides the most reliable information on protein secondary structure, enabling the quantification of α-helices (1650–1660 cm⁻¹), β-sheets (1600–1640 cm⁻¹), β-turns (1660–1700 cm⁻¹), and random coils (1640–1650 cm⁻¹) (Wang & Xie, 2019). Variations in these secondary structure elements reflect conformational changes, protein stability, and related functional properties. Generally, α-helical structures represent stable conformations, while significant disruptions can result in their transformation into β-sheets, β-turns, or random coils, thereby reducing protein stability and potentially causing partial or complete denaturation (Lv et al., 2022). As shown in the Fig. 4E, the control group exhibited the lowest proportion of α-helices and higher proportions of β-sheets and β-turns, indicating significant conversion of α-helical structures to β-sheet structures under thermal treatment conditions, consistent with significant protein denaturation. In contrast, treatment groups showed varying degrees of increased α-helix proportions, suggesting that polyphenolic compounds can inhibit the conversion of α-helices to β-sheets. By maintaining the stability of α-helical structures and preventing their unfolding, these compounds reduce the exposure of internal amino groups that would otherwise participate in glycation reactions (Lv et al., 2022). This structural stabilization not only preserves protein conformational integrity but also limits the involvement of exposed residues in subsequent glycation. Furthermore, an increase in β-turn content is often associated with enhanced intermolecular interactions, potentially promoting protein aggregate formation (Deng et al., 2021). Therefore, these experimental results further confirm that polyphenolic compounds can exert anti-glycation effects by stabilizing protein structures and inhibiting protein aggregate formation. These findings are consistent with the results from particle size analysis and SDS-PAGE, providing comprehensive evidence that polyphenolic compounds exert their anti-glycation effects through the dual mechanisms of conformational stabilization and aggregation inhibition.

3.3.4. Total antioxidant capacity and metal ion chelation ability

The Fig. 4F and G illustrate the total antioxidant capacity and metal ion chelation ability of the control group and treatment groups. The total antioxidant capacity followed this order: RosA > quercetin > sinapic acid > luteolin, while metal ion chelation ability ranked as: RosA > sinapic acid > quercetin > luteolin. Notably, RosA exhibited superior performance in both aspects, highlighting its key role in mitigating protein oxidative damage and inhibiting glycation. These results demonstrate that the anti-glycation effects of these polyphenolic compounds mainly operate through two mechanisms: scavenging free radicals and chelating metal ions. By sequestering redox-active metal ions, they prevent glucose autoxidation under aerobic conditions, which involves electron transfer, coordination with glucose hydroxyl groups, and oxidative cleavage of C—C bonds, thereby accelerating glycation (Hayase et al., 1996). Additionally, these compounds increased protein thiol content and reduced carbonyl levels (Fig. 4A and B), further protecting proteins from oxidative damage and aggregation. This dual protective effect ultimately inhibits glycation reactions and reduces the accumulation of CML and CEL (Han et al., 2025). The superior antioxidant and metal-chelating abilities of RosA suggest its potential as an effective inhibitor of protein glycation and oxidative damage in food systems.

3.3.5. Interaction forces between proteins

The Fig. 4H illustrates the chemical interactions between proteins in different treatment groups, including ionic bonds, hydrogen bonds, hydrophobic interactions, and disulfide bonds. Compared to the control group, the treatment groups exhibited weakened disulfide bonds and enhanced ionic interactions, consistent with the FTIR analysis results (Fig. 4D). Notably, the luteolin treatment group showed the weakest hydrophobic interactions, corresponding to its reduced surface hydrophobicity. This suggests that luteolin may inhibit protein aggregation by reducing diminishing hydrophobic interactions. During the initial stages of thermal treatment, weaker interactions such as ionic bonds are more susceptible to disruption (Luo et al., 2024). However, all treatment groups exhibited significantly stronger ionic bonds than the control, indicating that polyphenolic compounds contribute to preserving intermolecular interactions and stabilizing protein structure. These findings are supported by the observed changes in free thiol and carbonyl contents (Fig. 4A and B), further confirming the protective role of polyphenolic compounds against protein oxidative damage and glycation. Interestingly, the RosA treatment group exhibited significantly reduced hydrogen bonding, likely due to competition between RosA and proteins for hydrogen bond formation, thereby decreasing protein-protein hydrogen bonding. During glycation and thermal treatment, protein unfolding exposes internal hydrophobic groups and thiols, promoting hydrophobic interactions and disulfide bond formation (Hu et al., 2023). However, all treatment groups displayed significantly lower disulfide bond content compared to the control, consistent with the SDS-PAGE results (Fig. 3B). These findings suggest that polyphenolic compounds interfere with protein cross-linking, protect protein structure, and prevent AGEs formation mediated by protein aggregation (Han et al., 2025).

3.3.6. Interaction forces between proteins and polyphenolic compounds

Fluorescence quenching mechanisms are generally classified into static and dynamic quenching. Based on the thermodynamic parameters of protein–polyphenol interactions presented in the table, the quenching rate constants (kq) for all four experimental groups exceed the maximum fluorescence dynamic quenching constant (2.00 × 10¹⁰ L/mol/s), indicating that protein fluorescence loss is attributed to static quenching (Hu et al., 2022). By analyzing the enthalpy change (ΔH) and entropy change (ΔS), the interaction modes between proteins and polyphenolic compounds can be categorized into four types: (i) hydrophobic interactions (ΔH > 0; ΔS > 0), (ii) electrostatic and hydrophobic interactions (ΔH > 0; ΔS < 0), (iii) electrostatic interactions (ΔH < 0; ΔS > 0), and (iv) van der Waals forces and hydrogen bonding interactions (ΔH < 0; ΔS < 0) (Hu et al., 2022). According to the data in the table and Supplementary material 4A-D, quercetin, luteolin, and RosA primarily interact with proteins via hydrophobic interactions, whereas sinapic acid mainly forms van der Waals and hydrogen bonding interactions with proteins (Hu et al., 2023). From the structural perspective (Fig. 1B), quercetin, luteolin, and RosA contain multiple hydrophobic groups (e.g., benzene rings), which enable them to interact strongly with the hydrophobic regions of proteins, thereby enhancing binding affinity. In contrast, sinapic acid has lower hydrophobicity and higher polarity, leading to its predominant interaction via van der Waals forces and hydrogen bonding. The surface hydrophobicity of the simulated system aligns with the strength of hydrophobic interactions as indicated by ΔH and ΔS values, further confirming the close correlation between surface hydrophobicity and polyphenol–protein binding affinity. When evaluating binding strength based on the binding constant (Ka), the ranking of polyphenol–protein interactions follow the order: RosA > sinapic acid > luteolin > quercetin. This trend corresponds to the inhibitory effects of the four polyphenolic compounds on CML and CEL formation, consistent with previous findings that glycation influences polyphenol–protein interactions (Tang et al., 2017). Polyphenolic compounds exert anti-glycation effects by binding to proteins and occupying key glycation sites, thereby preventing glycation reactions (Zhang et al., 2024). Measuring the binding affinity between polyphenolic compounds and proteins provides insights into the stability of polyphenol occupation at glycation sites, which in turn helps explain their glycation inhibition capacity.

3.3.7. Molecular docking analysis

Molecular docking simulations were performed to elucidate the molecular interactions between the target protein and four polyphenolic compounds (quercetin, luteolin, RosA, and sinapic acid) in relation to their anti-glycation properties. The Fig. 5 revealed that all tested polyphenols predominantly formed non-covalent interactions with the protein, as supported by thermodynamic data. Specifically, quercetin and luteolin demonstrated strong hydrophobic interactions, particularly with leucine residues, which likely contribute to protein stabilization. In contrast, RosA and sinapic acid exhibited a mixed interaction profile involving both hydrophobic forces and hydrogen bonding. Notably, sinapic acid was found to interact with Lys-186 through hydrophobic contacts while simultaneously forming hydrogen bonds, effectively blocking the AGEs binding site and thereby inhibiting glycation (Zhang et al., 2024). The calculated binding energies of −8.4 kcal/mol for quercetin, −8.8 kcal/mol for luteolin, and − 8.6 kcal/mol for RosA further corroborated their strong binding affinities. These findings suggest that while quercetin, luteolin and RosA primarily maintain protein structural stability through direct interactions, sinapic acid exerts its anti-glycation effect mainly by competitively occupying the glycation site through dual hydrophobic and hydrogen-bonding interactions with Lys-186.

Fig. 5.

Molecular docking analysis of four polyphenolic compounds with the target protein. (A, E) Three-dimensional (3D) and two-dimensional (2D) interaction diagrams of quercetin bound to the protein. (B, F) 3D and 2D interaction diagrams of luteolin with the protein. (C, G) 3D and 2D interaction diagrams of RosA in complex with the protein. (D, H) 3D and 2D interaction diagrams of sinapic acid docked into the protein active site.

4. Conclusion

The major constituents of PE were identified using LC-MS/MS, leading to the selection of eight predominant compounds—azelaic acid, 4-phenylbutyric acid, salvianolic acid B, quinine, quercetin, luteolin, RosA, and sinapic acid—based on their potential bioactivity and high concentrations. Among them, quercetin, luteolin, RosA, and sinapic acid exhibited significant inhibitory effects on bound CML, with inhibition rates of 41.3 %, 53.8 %, 60.5 %, and 61.1 %, respectively. Given their pronounced anti-glycation activity, these four compounds, classified as flavonoids (quercetin and luteolin) and phenolic acids (RosA and sinapic acid), were selected for further investigation of their inhibitory mechanisms against AGEs formation during high-temperature food processing. Our findings demonstrate that quercetin, luteolin, RosA, and sinapic acid exhibit strong free radical scavenging activity and metal ion chelation capacity. These properties contribute to protein stabilization, preventing structural damage and inhibiting protein cross-linking. These compounds also modulate protein-protein interactions, influence conformational changes, and enhance structural stability. Additionally, they alter protein surface properties, engage in non-covalent interactions, and effectively occupy glycation sites, reducing the availability of reactive glycation loci. Consequently, these compounds suppress the formation of Amadori products and capture dicarbonyl intermediates such as GO and MGO, thereby mitigating AGEs accumulation.

CRediT authorship contribution statement

Xue Han: Writing – original draft, Methodology, Formal analysis, Conceptualization. Xue Sun: Writing – review & editing, Investigation, Data curation. Zihang Shi: Writing – review & editing, Methodology. Xiankang Fan: Writing – original draft, Formal analysis. Yangyang Hu: Formal analysis, Data curation. Chen Chen: Methodology. Qiang Xia: Writing – review & editing. Yangying Sun: Investigation. Daodong Pan: Project administration, Funding acquisition, Conceptualization.

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.