Elucidating the interaction between pyrazine flavor compounds and coffee proteins: Insights from multiscale structural and molecular dynamics simulations Authors: Xinxin Yu, Zhang Su, Guanghua Xia, Dachuan Zhang, Hehe Li, Wenjiang Dong Published: 2026 ABSTRACT AbstractFor roasted coffee, its distinctive aroma is imparted by pyrazine compounds. Understanding how they interact with coffee proteins (CP) is crucial for controlling coffee flavor retention. This study demonstrated that the type and number of alkyl groups determine the binding strength between pyrazines and CP. At 2.00 and 10.00 mg/L of the pyrazine compounds, the binding affinity of CP for the pyrazines followed the order: 2,3,5-trimethylpyrazine (TRP) > 2-ethyl-3,5-dimethylpyrazine (EDP) > 2-ethyl-3-methylpyrazine (EMP) > 2-ethylpyrazine (EP). Multi-spectroscopic analysis indicated that pyrazines quenched the fluorescence of CP through both dynamic and static quenching processes, primarily via binding near Tyr and Trp residues. Hydrophobic interactions primarily governed the binding of TRP and EDP to CP, whereas van der Waals forces and hydrogen bonding were dominant for EP and EMP. Pyrazine binding induced the aggregation of CP, with TRP exhibiting the most pronounced effect on secondary structure, including a decrease in α-helix content and an increase in β-sheet content indicative of partial protein unfolding. Molecular simulation confirmed that TRP exhibited the smallest highest occupied molecular orbital-lowest unoccupied molecular orbital (HOMO–LUMO) energy gap (5.338505 eV) with CP, indicating a greater propensity for interaction. The study revealed the relationship between pyrazine structure and its binding capacity to CP, providing a theoretical basis for flavor regulation during coffee processing and storage.Keywords: Plant-based proteins, Pyrazine compounds, Food flavors, Multispectral techniques, Molecular simulations, Interactions Introduction Coffee is one of the world's three major beverage crops and is among the most widely consumed (Zhai et al., 2024). In recent years, coffee consumption in China has increased markedly, propelling the country into the ranks of the world's top ten coffee-consuming nations (Xu et al., 2026, Xu et al., 2026). Coffee is cherished by consumers for its distinctive and complex flavor profiles. The volatile flavor compounds that are generated during roasting engage in highly intricate interactions, and their final composition and aroma are further modulated by multiple stages of coffee processing and storage (Lu et al., 2026; Cao et al., 2023). In food systems, the interaction of flavor compounds with proteins is crucial for their retention or release and determines food sensory quality and consumer acceptance (Ke et al., 2026; Sharma et al., 2025). These interactions depend on flavor compound contents and both protein and flavor compound structures (Guo et al., 2024; Jia et al., 2024), which can facilitate the formation of covalent and non-covalent interactions between them (Wei et al., 2024). Among these, non-covalent interactions are reversible and include van der Waals forces, hydrophobic interactions, hydrogen bonds, and ionic bonds (Li et al., 2025). They are often exploited to mitigate flavor loss during processing and to modulate flavor release during consumption (Wei et al., 2024). Interactions of major proteins with aroma compounds in products such as cow's milk (Ma et al., 2019; Kou et al., 2025), meat products (Sun et al., 2023), and plant-based meat (Guo et al., 2022) have been extensively reported in recent years. However, for plant proteins and plant-based beverages, no studies have yet elucidated the retention and release of characteristic flavor compounds in coffee beverages. Coffee protein (CP) is one of the major endogenous components in coffee; although it undergoes Maillard reactions and thermal degradation with reducing sugars during roasting, about 12% of CP remains in roasted coffee beans (Lee et al., 2024). The molecular weight of each monomer is 55 kDa, comprising a 33 kDa (α) subunit together with a 24 kDa (β) subunit linked by disulfide bonds (Ribeiro et al., 2021). These monomers further associate into trimeric and hexameric forms through hydrophobic interactions (Souza Almeida, Furlan Goncalves Dias, Kawazoe Sato, & Leite Nobrega De Moura Bell, 2021). Although coffee protein is belongs to the 11S storage protein family, its three-dimensional structure has not yet been elucidated; protein prediction tools offer a potential means to resolve its three-dimensional structure. This also provides a structural foundation for investigating its interactions with key flavor compounds in coffee. The volatile compounds primarily responsible for coffee aroma include pyrazines, aldehydes, furans, ketones, sulfur-containing compounds, and phenolic compounds, generated mainly via Maillard reactions, caramelization, Strecker degradation, and pyrolysis throughout roasting (Wang et al., 2025). Among these, sulfur-containing compounds and pyrazines constitute the second largest class of volatile compounds (25–39%) in roasted coffee (Angeloni et al., 2021; Wang et al., 2025). They are formed from α-amino acids via complex Strecker degradation and are recognized for imparting nutty, cocoa-like, and roasty notes to coffee (de Melo Pereira et al., 2019). Alkyl-pyrazines such as 2-ethylpyrazine and 2-ethyl-3,5-dimethylpyrazine are identified as key aroma contributors in coffee brew (Tong et al., 2023). The four carbon atoms in the heterocyclic ring can be substituted with diverse functional groups (e.g., alkyl, acyl), and varying numbers of alkyl groups lead to structural variations in pyrazine (Ma et al., 2019). Generally, alkyl-pyrazine analogs possess similar chemical properties; however, it remains unclear how differences in alkyl group type and position on the pyrazine ring affect their interaction with CP and adsorption strength. Although the interactions between pyrazines and other proteins, such as myofibrillar proteins (Shen et al., 2019; Yu et al., 2023) and pea proteins (Guo et al., 2024), have been investigated, CP differs from these proteins in its structural organization. Myofibrillar proteins possess a filamentous structure stabilized by hydrogen bonds and ionic interactions, while CP forms trimeric and hexameric assemblies stabilized primarily by hydrophobic interactions and disulfide bonds. Therefore, the binding behavior of pyrazines with CP may differ from that observed in previously studied protein systems, and a dedicated investigation is necessary. An in-depth analysis of the interactions between coffee proteins and representative volatile compounds provides new insights into the mechanisms governing the retention or release of key flavor compounds in coffee products, which is crucial for maintaining and enhancing flavor quality during coffee processing and storage. This study selected four typical pyrazine compounds present in coffee, each with a different alkyl substitution pattern: [2-ethylpyrazine (EP), 2-ethyl-3,5-dimethylpyrazine (EDP), 2-ethyl-3-methylpyrazine (EMP), 2,3,5-trimethylpyrazine (TRP)]. The capacity of CP to adsorb pyrazine compounds of different concentrations and types was evaluated by headspace solid-phase microextraction coupled with gas chromatography-mass spectrometry (HS-SPME-GC/MS). Multispectroscopy and scanning electron microscopy (SEM) were employed to investigate how various pyrazine compounds affected CP conformation. Additionally, the interaction mechanisms of pyrazines with CP were investigated from thermodynamic, molecular dynamics (MD), and quantum chemical perspectives. This study overcomes the limitations of previous approaches, which relied solely on flavoromics and sensory evaluation to assess coffee flavor quality, delving into the impact of pyrazine flavor compounds on coffee protein structure as well as the adsorption and release mechanisms of characteristic flavor compounds. The findings promise to address the issue of flavor degradation in coffee during environmental exposure and storage, providing a theoretical foundation for enhancing coffee flavor quality. Materials and methods Materials and reagents Robusta green coffee (C. canephora) beans were obtained from Baisha Yunshiling Coffee CO., Ltd. (Hainan, China), with fresh cherries harvested in February 2024. Petroleum ether (boiling point, 60–90 °C), ethylene glycol (C2H6O2), anhydrous ethanol, hydrochloric acid, and sodium hydroxide were purchased from Xilong Scientific Co., Ltd. (Guangzhou, China). Choline chloride (ChCl) and 2-ethyl-3,5-dimethylpyrazine (97%) were obtained from Shanghai Yuanye Biotechnology Co., Ltd. (Shanghai, China). 2-ethyl-3-methylpyrazine (98%), 2-ethylpyrazine (99%), and 2,3,5-trimethylpyrazine (99%) were all supplied by Shanghai Macklin Biochemical Technology Co., Ltd. The remaining chemicals used were of analytical grade. Preparation of coffee protein Green coffee beans were ground using a coffee grinder (Kenia, Mahlkonig, Germany) and passed through a 40-mesh sieve. The resulting powder (100 g) was mixed with 300 mL of petroleum ether and extracted in a constant-temperature water bath for 1 h. After filtration, the residue was collected and dried in a forced-air drying oven (DHG-9055A, Shanghai Yiheng Scientific Instruments Co., Ltd.) at 40 °C to obtain defatted green coffee powder. Coffee protein (CP) was extracted according to the method described by Chen et al. (2021) with slight modifications. Briefly, the defatted coffee powder was added into a natural deep eutectic solvent (ChCl/ethylene glycol = 1:3, mol/mol) at a 1:5 (w/v) solid-to-solvent ratio. Thereafter, the mixture was stirred for 2 h at 600 rpm in a 50 °C water bath (model HH-6, Shanghai Sheyan Instrument Co., Ltd.), followed by centrifugation for 10 min at 5000 rpm. Supernatants were collected, while protein was precipitated with ice-cold ethanol and stored at 4 °C overnight. Subsequently, the precipitated protein was recovered by centrifugation (5000 rpm, 10 min) and repeatedly washed with ice-cold ethanol to remove residual natural deep eutectic solvent. The pellet was then redissolved in ice-cold ethanol and transferred into a dialysis bag. A 48 h dialysis process was conducted with ultrapure water at 4 °C, with the water changed at 4 h intervals, to further eliminate low molecular weight solvent residues. After dialysis, the resulting solution underwent another centrifugation step (5000 rpm, 15 min) to collect the final precipitate, and CP was obtained after freeze-drying. The protein content of the prepared CP was analyzed using a Dumas nitrogen analyzer (NDA701, VELP, Milan, Italy) and was determined to be 81.93 ± 1.66%. Preparation of coffee protein-pyrazine complexes solution A standard CP solution was prepared by dissolving CP in 0.01 mol/L PBS (pH 7.4) to a concentration of 5.00 mg/mL. The mixture was stirred for 1 h using a magnetic stirrer and then centrifuged for 10 min (5000 rpm) to obtain supernatants for subsequent use. Subsequently, 1.0 mg/mL stock solutions of four pyrazine compounds were prepared using 0.01 mol/L phosphate buffer (pH 7.4) containing 10% methanol as the solvent. The prepared stock solutions were stored in sealed brown bottles to prevent degradation. Then, 2.00 mL of the CP solution (5.00 mg/mL) was transferred into a 20.00 mL vial, followed by the addition of 100.00 μL or 500.00 μL of the respective pyrazine stock solutions (1.00 mg/mL). The system was diluted with PBS to 5.00 mL and sealed with an aluminum cap. As a result, the final concentration of CP was 2.00 mg/L, and pyrazine concentrations were 2.00 and 10.00 mg/L, respectively. The sample vials were then incubated at 30 °C for 16 h until equilibrium was reached. Solutions containing only the solvent and pyrazine compounds (without CP) served as controls. The mixtures containing CP with EP, EMP, TRP, and EDP were designated as PEP, PEMP, PTRP, and PEDP, respectively. Binding ability assessment The GC/MS system (7890B-5977B, Agilent Technologies, Santa Clara, CA, USA) was employed to evaluate the binding ability of CP to four pyrazine compounds. Prior to injection, the SPME fiber was conditioned in the GC inlet at 250 °C for 30 min. During 30 min of sample incubation at 40 °C, volatile compounds were adsorbed onto the SPME fiber. Afterward, the fiber was inserted into the GC inlet for automatic injection in splitless mode. Initially, the oven temperature was held at 40 °C for 3 min, then increased to 250 °C at 15 °C/min, and maintained for 5 min. The ion source, auxiliary heater, and quadrupole temperatures were set at 230 °C, 250 °C, and 150 °C, respectively. Other parameters were set according to Zhang et al. (2022). The binding ability of the four pyrazine compounds was calculated using the following equation (1). Surface hydrophobicity measurement To measure surface hydrophobicity (H0), the ANS probe method was employed (Yu et al., 2023). Specifically, each 3.60 mL aliquot of CP solutions at graded concentrations (0.40, 0.60, 0.80, 1.00, and 1.20 mg/mL) was supplemented with 0.40 mL of each pyrazine solution (0.02 or 0.10 mg/mL), resulting in final pyrazine concentrations of 2.00 and 10.00 mg/L, respectively. After mixing, 4.00 mL of each mixture was combined with 8.00 mM ANS (40.00 μL) and incubated at 25 °C for 15 min in the dark. A fluorescence spectrophotometer (F-4600, Hitachi Co., Ltd., Japan) was used to measure fluorescence intensities at excitation and emission wavelengths of 390 and 470 nm, respectively, with a slit width of 5 nm. H0 was calculated from the initial slope of the fluorescence intensity versus CP concentration plot. Multi-spectroscopic analysis The standard CP solution (5.0 mg/mL, 0.5 mL) was mixed with 0, 100, 200, 300, 400, 500 μL pyrazine solutions (1.00 mg/mL). PBS was added to dilute the mixtures to 5.0 mL, yielding a CP concentration of 0.50 mg/mL and pyrazine concentrations of 0, 2.0, 4.0, 6.0, 8.0, and 10.0 mg/L, respectively. Fluorescence spectroscopy A fluorescence spectrophotometer (RF-4600, Hitachi, Japan) was used to record fluorescence spectra at excitation wavelengths of 280 nm and emission wavelengths ranging from 295 to 500 nm, with a slit width of 2.5 nm and a scan rate of 1200 nm/min (Li et al., 2023). The complex solutions were incubated for 2 h in a constant-temperature water bath at 298, 303, and 310 K. The fluorescence intensity of each sample was recorded immediately after incubation, and background correction was performed by subtracting the signal from control solutions containing the same concentration of pyrazine compounds but without CP. To eliminate potential fluorescence artifacts caused by the inner-filter effect, all fluorescence intensities were corrected using the method reported by Pang et al. (2025). The correction was performed using the following equation (2): where Fcor and Fobs denote the corrected and measured fluorescence intensities, respectively, and Aex and Aem are the absorbances of the sample at the excitation wavelength (280 nm) and the emission wavelength (335 nm), respectively. These absorbance values were recorded directly on the fluorescence spectrophotometer prior to each fluorescence measurement. Pyrazine compounds (quenchers) can alter the intrinsic fluorophore microenvironment (aromatic amino acid residues) of CP via various quenching pathways, including static quenching, dynamic quenching, or a combined static-dynamic mechanism (Xu et al., 2026). In this study, the Stern-Volmer equation (3) was applied to characterize the fluorescence quenching mechanism of CP by the four pyrazine compounds (Jia et al., 2022). where F0 and F are the fluorescence intensities of CP in the absence and presence of quencher, respectively; [Q] is the quencher concentration; Kq and Ksv represent the quenching rate constant, Stern-Volmer quenching constant, respectively. The τ0 is the average lifetime of molecules in the absence of quencher (about 10−8 s). The binding constant (Ka) and the number of binding sites (n) were determined using the double-logarithm equation (4): Calculations of ΔH (enthalpy change) and ΔS (entropy change), and ΔG (free energy change) during CP and pyrazine compounds interactions at 298 K–303 K were performed with the Van't Hoff equations (5) and (6) (Wang et al., 2026). where R is the gas constant (approximately 8.314 J mol−1 K−1), T is the absolute temperature, Ka is the binding constant at the corresponding temperature. Synchronous fluorescence spectra For pyrazine–CP complexes, synchronous fluorescence spectra were recorded using fixed wavelength intervals (Δλ = 15 nm and Δλ = 60 nm) at 298 K. Each spectrum was obtained at an excitation wavelength of 280 nm, with emission scanned from 265 to 330 nm (for Δλ = 15 nm) and from 220 to 330 nm (for Δλ = 60 nm). A slit width of 2.5 nm and a scan rate of 1200 nm/min were maintained throughout the measurements (Li et al., 2023). Three-dimensional fluorescence spectra measurements A fluorescence spectrophotometer (RF-4600, Hitachi, Japan) was used to record three-dimensional fluorescence spectra of CP–pyrazine complexes at 25 °C. The complexes contained CP at 0.50 mg/mL and pyrazine at 6.0 mg/L. The excitation and emission wavelength ranges were 200–450 nm and 200–500 nm, respectively. Data were collected with a 5 nm sampling interval for both excitation and emission, a slit width of 2.5 nm, and a scan rate of 1200 nm/min (Guo et al., 2024). Fourier transform infrared spectroscopy Sample preparation was the same as described in Section 2.6.3. Fourier transform infrared (FTIR) spectra of CP, PEP, PEMP, PTRP, and PEDP were recorded at 25 °C using an FTIR spectrometer (Nicolet 6700, Thermo Fisher Scientific, USA) over the range of 4000-500 cm−1. The complex solutions were freeze-dried, mixed with potassium bromide (KBr) powder at a 1:100 mass ratio, ground thoroughly, and pressed into transparent pellets, with a pure KBr pellet used as the background reference. For noise reduction, spectra were collected at a resolution of 4 cm−1 with 32 scans. The spectral data were analyzed using OMNIC software (Feng et al., 2026). Circular dichroism spectra The composition and content of the secondary structure of CP were determined and analyzed by CD (Bio-Logic MOS-500, Bio-Logic, France). The preparation of CP and pyrazine solutions was as described in Section 2.6.3. The samples were mixed thoroughly, and only circular dichroism (CD) spectra obtained at 303 K were considered. Spectral recording was performed over the range of 190–250 nm. The scan rate and spectral resolution were 20 nm/min and 0.1 nm, respectively. The secondary structure content was analyzed using the online CD analysis platform: http://dichroweb.cryst.bbk.ac.uk (Zhang et al., 2024). SDS-PAGE analysis SDS-PAGE was conducted to analyze the molecular weights of the samples following the method described by Zhang et al. (2025) with slight modifications. Gels consisting of a 5% stacking gel and a 10% separating gel were prepared. Samples were mixed with 5× reducing loading buffer at a 4:1 (v/v) ratio, followed by heating at 95 °C for 5 min and cooling to room temperature. Subsequently, each sample (5 μL) and a protein molecular weight marker (10 μL, 6.5–270 kDa) were loaded into gel lanes and electrophoresed at 80 V for 10 min. The voltage was then increased to 110 V after the dye front entered the separating gel and maintained until the dye reached the bottom of the gel. After staining with Coomassie Brilliant Blue for 2 h, the gel was destained overnight in a decolorizing solution until clear bands were visible. Images were captured using a Tanon-2500 gel documentation system (Tanon, Shanghai, China). Scanning electron microscopy The surface microstructure of freeze-dried CP and complex samples was examined using SEM (S-3400N, Hitachi, Tokyo, Japan). Prior to observation, samples were mounted on conductive adhesive and sputter-coated with a thin gold layer. The microstructure was observed at an accelerating voltage of 10 kV and a magnification of 8000×. Molecular docking The amino acid sequence of the 11S storage protein (UniProt ID: P93079), the primary coffee protein, was retrieved from UniProt. Its three-dimensional structure was modeled using AlphaFold2 (Jumper et al., 2021) and subsequently energy-minimized. The structures of the four pyrazine compounds were obtained from the PubChem database with the following CID numbers: EP (CID: 26331), EMP (CID: 27457), TRP (CID: 26808), and EDP (CID: 26334). Each small molecule was then geometry-optimized using the MMFF94s force field to eliminate unreasonable bond lengths and angles, and all rotatable bonds were set as flexible during docking. AutoDock Vina 1.2 was used for semi-flexible molecular docking (Feng et al., 2026). Due to the absence of prior information regarding the binding site, a blind docking strategy was adopted. A grid box was constructed to be sufficiently large to encompass the entire 11S storage protein structure, with a grid spacing of 0.375 Å. The docking calculations employed the default scoring function of AutoDock Vina. One hundred independent docking runs were performed, and the resulting conformations were ranked by binding energy. The conformation with the lowest binding energy (kcal/mol) was selected as the most stable complex. Visualization and analysis were performed using PyMOL 3.0 (The PyMOL Molecular Graphics System, Version 3.0, Schrödinger, LLC, USA). Molecular dynamics simulation Geometries of the four pyrazine compounds were optimized, and their vibrational frequencies were calculated using the r2SCAN-3c method within the ORCA program. Single-point energy calculations were then performed at the B3LYP-GD3/def2-TZVP level of theory, using the auxiliary def2/J basis set (Sazlı and Karataş, 2025). RESP atomic charges were derived through wavefunction analysis using the Multiwfn program, and GAFF force-field parameters were assigned for each flavor compound. Electrostatic potential (ESP) surfaces and frontier molecular orbitals (highest occupied molecular orbital, HOMO; lowest unoccupied molecular orbital, LUMO) were also calculated to evaluate the electronic stability and chemical reactivity of pyrazines. The GROMACS software package (version 2021.6) was used to perform MD simulations to analyze interactions and binding stability between pyrazine compounds and CP (Zhang et al., 2025). For all systems, the Amber14SB force field in combination with an explicit TIP3P water model was applied. Each complex was placed in a cubic box of TIP3P water molecules, and the total system charge was neutralized by adding counterions. These systems were energy-minimized in two stages: an initial minimization using the steepest-descent algorithm (force tolerance 1000.0 kJ mol−1·nm−1), followed by a refined minimization using the conjugate-gradient method (force tolerance 100.0 kJ mol−1·nm−1). Thereafter, the system was equilibrated in two phases: an NVT simulation (1 ns) at 303.15 K under periodic boundary conditions, followed by an NPT simulation (1 ns) at 1 bar. After equilibration, a 100 ns production MD run was conducted under NVT conditions at 303.15 K, maintained by the V-rescale thermostat. Hydrogen bonds were constrained using the LINCS algorithm. Long-range electrostatic interactions were treated using the Particle-Mesh-Ewald (PME) method with a 1.0 threshold. The stability of complexes formed between CP and pyrazine compounds was further evaluated by analyzing root-mean-square deviation (RMSD), root-mean-square fluctuation (RMSF), radius of gyration (Rg), and solvent-accessible surface area (SASA) (Ren et al., 2023). Statistical analysis All experiments were conducted with three independent replicates, with data expressed as mean ± standard deviation. Statistical significance (P < 0.05) was assessed using one-way analysis of variance (ANOVA) ollowed by Duncan's multiple comparison test. OriginPro 2021 software (OriginLab, Massachusetts, USA) was used for figure processing and plotting. Results and discussion Binding ability between pyrazine compounds and coffee proteins The binding capacities of the four pyrazine compounds with CP are shown in Fig. 1A. Except for TRP, CP showed no significant change in adsorption capacity for the pyrazines when their concentration increased from 2.00 to 10.00 mg/L. In contrast, a previous study reported that pyrazine flavor compounds exhibited a gradual increase in binding capacity to pea protein as pyrazine concentration increased from 0.08 to 0.40 mM (Guo et al., 2024). However, when TRP concentration increased from 2.00 mg/L to 10.00 mg/L, its binding capacity with CP increased significantly from 11.09% to 16.83%. This suggests that the reactivity and positional arrangement of flavor compounds are crucial for protein-flavor interactions (Fu et al., 2026; Jia et al., 2024). This discrepancy may be attributed to differences in the molecular structure and hydrophobicity of the pyrazine compounds. TRP, with its three uniformly distributed methyl groups, possesses higher molecular reactivity than EP, EMP, and EDP. At 2.00 mg/L, the initial binding sites accessible to all pyrazines may already be largely occupied. However, TRP, owing to its higher reactivity, may further induce conformational changes in CP that expose additional binding sites, thereby enabling continued adsorption at higher concentrations. Specifically, regardless of the concentration of the flavor compound, PTRP exhibited the highest binding affinity, followed by PEDP, PEMP, and PEP. In particular, CP showed increasing adsorption capacity for pyrazines with increasing numbers of methyl groups, indicating that alkyl chain length and degree of substitution critically influence binding strength (Li et al., 2023). This binding affinity order indicates that pyrazines with a higher degree of alkyl substitution exhibit stronger binding to CP, which is generally consistent with the trend observed for pea protein (Guo et al., 2024), despite differences in the specific alkyl groups. The binding ability (A) and surface hydrophobicity decrease rate (B) of CP with four pyrazine compounds at concentrations of 2.0 mg/L and 10.0 mg/L. Note: The different letters indicate significant differences (P < 0.05). Surface hydrophobicity analysis H0 is an important indicator of changes in protein conformation and can be used to infer the structural effects of pyrazine compounds on the hydrophobic regions of CP (Jia et al., 2024). For comparison, the H0 of CP in the absence of pyrazines was defined as 100%. According to the results, at 2.0 mg/L pyrazine, CP showed a significant decrease in H0 (P < 0.05) upon pyrazine addition. Moreover, H0 values for PEP, PEMP, PTRP, and PEDP decreased to 93.77%, 91.98%, 89.93%, and 89.36%, respectively (Fig. 1B). This indicates that pyrazines can bind to the hydrophobic groups of CP, consistent with previous findings that flavor compounds interact with hydrophobic patches of myofibrillar and pea proteins via hydrophobic interactions, leading to reduced H0 (Guo et al., 2024; Wei et al., 2024). Sun et al. (2025) also reported that increased steric hindrance from substituents enhances the hydrophobic character of the pyrazine ring, facilitating its insertion into protein hydrophobic cavities. At this concentration, the four pyrazines exerted differential effects on H0 of CP, with the order PEDP > PTRP > PEMP > PEP, suggesting that the type and number of substituents on the pyrazine ring are critical determinants of binding. Interestingly, when pyrazine concentration increased to 10.0 mg/L, a significant increase in H0 was observed for PEP and PEMP (P < 0.05). This may be associated with the structural unfolding of CP induced by higher pyrazine concentrations, which can expose buried hydrophobic amino acid residues (e.g., Tyr, Trp) to the molecular surface, thereby increasing H0 (Li et al., 2025; Liu et al., 2025, Liu et al., 2025). In contrast, the addition of EDP and TRP resulted in a significant decrease in H0 (P < 0.05). Owing to its longer alkyl chain and higher degree of substitution, EDP possesses the highest LogP value (1.5, sourced from PubChem), and its pronounced hydrophobic character facilitates its embedding into the hydrophobic regions of protein (Zhang et al., 2022). Consequently, PEDP exhibited the greatest reduction in H0 (97.82%). For TRP, the three uniformly distributed methyl groups enhance its ability to interact with the hydrophobic domains of CP through hydrophobic interactions (Guo et al., 2024), also leading to a significant decrease in H0 (P < 0.05). The divergent H0 trends observed at different pyrazine concentrations suggest that the binding behavior is influenced not only by alkyl chain length and substitution pattern but also by solvent effects in solution. These phenomena likely contribute to the differences in binding mechanisms and interaction forces between CP and the various pyrazine compounds, which were subsequently investigated in detail through fluorescence spectroscopy and molecular simulations. Fluorescence quenching analysis and thermodynamic parameters Aromatic amino acids in proteins, including tryptophan, phenylalanine, and tyrosine, exhibit intrinsic fluorescence. When these residues interact with small molecules, conformational changes may lead to variations in fluorescence intensity (Fu et al., 2026). The fluorescence spectra of CP complexed with the four pyrazine compounds are shown in Fig. 2A–D. CP exhibited a characteristic emission peak around 320 nm. As the pyrazine concentration in the system increased, the maximum fluorescence intensity gradually decreased, indicating that all four pyrazines induced fluorescence quenching of CP. A simultaneous red shift of the emission maximum was observed, suggesting that interaction with pyrazines exposed the aromatic amino acid residues of CP to a more polar solvent environment (Yu et al., 2023). This further supports that CP binds pyrazines primarily through hydrophobic regions associated with Trp and Tyr residues (Zhang et al., 2022). The TRP and EDP consistently exhibited stronger quenching ability than the other two pyrazines, indicating that quenching efficiency is related to the number and type of alkyl substituents. The order of quenching ability was TRP > EDP > EMP > EP. Similarly, as reported by Guo et al. (2024), an increase in alkyl groups on the pyrazine ring resulted in more pronounced conformational changes in pea protein, accompanied by enhanced energy transfer from protein fluorophores in the excited state to pyrazine molecules. The effect of temperature on quenching was most pronounced for TRP, whereas the decrease in fluorescence intensity caused by the other three pyrazines showed less dependence on temperature. Similar fluorescence quenching behavior has also been observed in studies of collagen-flavor compound interactions by Xu et al. (2026). Fluorescence quenching spectra of PEP (A), PEMP (B), PTRP (C) and PEDP (D) (a: 298 K, b: 303 K, c: 310 K). Stern-Volmer plots (E) and logarithmic plots (F) of PEP (a), PEMP (b), PTRP (c) and PEDP (d). To investigate the mechanism underlying quenching, binding mode, and interaction of the pyrazine compounds with CP, the temperature dependence of fluorescence quenching was examined to distinguish between dynamic and static quenching. The Stern–Volmer constant (KSV) reflects quenching efficiency of the pyrazine compounds toward CP fluorescence; a higher KSV indicates greater sensitivity of the fluorophore to the quencher (Gan et al., 2025). In dynamic quenching, KSV typically increases with temperature due to enhanced diffusion and collision frequency. In contrast, KSV often decreases with increasing temperature for static quenching, as higher temperatures can destabilize the ground-state complex (Zhang et al., 2025). Interaction parameters of CP with the four pyrazines at different temperatures were obtained from fitted curves (Fig. 2E and F), as summarized in Table 1. For all systems, the KSV values exhibited a non-monotonic trend with temperature: they increased from 298 to 303 K and then decreased at 310 K, with the maximum observed at 303 K. This non-monotonic behavior suggests that both static and dynamic quenching mechanisms contribute to the overall quenching. Although time-resolved fluorescence was not performed, the Kq values overwhelmingly exceeding the diffusion-controlled limit (2 × 1010 M−1 s−1), unambiguously confirms static quenching as the dominant mechanism. This approach has been widely adopted in protein-flavor interaction studies (Wang et al., 2022; Liu et al., 2025). It is important to note that KSV reflects the collisional/quenching efficiency, whereas Ka is an equilibrium constant quantifying the thermodynamic binding affinity (Pang et al., 2025). Despite this conceptual difference, both parameters followed a consistent order at 303–310 K, indicating that pyrazines with higher quenching efficiency also exhibit stronger binding affinity to CP. At the same temperature, KSV and Kq values were higher for PTRP and PEDP than for PEP and PEMP, indicating that the number of alkyl groups on the pyrazine ring influences the quenching pathway. The binding stoichiometry (n) was about 1 for all complexes, suggesting a single primary binding site on CP for each pyrazine. The binding constant (Ka) reflects binding affinity, and values greater than 103 M−1 indicate strong interactions. The decrease in Ka with increasing temperature suggests that higher temperatures destabilize the complexes (Sun et al., 2023). Ka values for PTRP were significantly higher (P < 0.05) than those for other pyrazine-CP complexes at 303–310 K, indicating that pyrazine compounds with a higher number of methyl substituents exhibit stronger binding affinity, which is consistent with the GC-MS results presented in Section 3.1. Table 1The thermodynamic parameters of coffee proteins and four pyrazine compounds.Table 1CompoundsT(K)Ksv (103 M−1)Kq (1011 M−1s−1)Ka (102 M−1)nΔH (kJ·mol−1)ΔS (J·mol−1K−1)ΔG (kJ·mol−1)PEP2985.25 ± 0.15bD5.25 ± 0.05bD77.11 ± 3.80aD1.05 ± 0.13−23.02−2.85−22.183035.68 ± 0.23aD5.68 ± 0.03aD66.15 ± 1.58bD1.02 ± 0.05−22.163104.65 ± 0.23cD4.65 ± 0.03cD17.32 ± 1.86cD0.89 ± 0.06−19.22PEMP29811.36 ± 0.04bC11.36 ± 0.04bC2799.63 ± 17.8aA1.34 ± 0.06−152.94−408.95−31.0730311.85 ± 0.13aC11.85 ± 0.03aC1010.88 ± 11.5bC1.24 ± 0.01−29.0331011.48 ± 0.16bC11.48 ± 0.06bC362.91 ± 15.1cB1.13 ± 0.04−27.06PTRP29821.50 ± 0.10cA21.50 ± 0.10cA1077.95 ± 12.6bB1.18 ± 0.0279.03361.54−28.7130322.11 ± 0.15aA22.11 ± 0.15aA1824.74 ± 14.5aA1.24 ± 0.04−30.5231021.87 ± 0.11bA21.87 ± 0.11bA521.31 ± 15.1cA1.10 ± 0.04−27.99PEDP29820.91 ± 0.12bB20.91 ± 0.12bB756.48 ± 14.1bC1.14 ± 0.03123.49507.79−27.8330321.38 ± 0.24aB21.38 ± 0.24aB1721.87 ± 20.4aB1.24 ± 0.07−30.3731019.20 ± 0.13cB19.20 ± 0.13cB150.42 ± 20.0cC0.98 ± 0.07−24.79Note: The ΔH and ΔS values in the table are thermodynamic parameters for pyrazine compounds and coffee proteins in the 298–303 K range, calculated using the Van't Hoff equation. Different lowercase letters indicate statistically significant differences in data among the same pyrazine compound at different temperatures (P < 0.05), while different uppercase letters indicate statistically significant differences in data among different pyrazine compounds at the same temperature (P < 0.05). The thermodynamic parameters of coffee proteins and four pyrazine compounds. Note: The ΔH and ΔS values in the table are thermodynamic parameters for pyrazine compounds and coffee proteins in the 298–303 K range, calculated using the Van't Hoff equation. Different lowercase letters indicate statistically significant differences in data among the same pyrazine compound at different temperatures (P < 0.05), while different uppercase letters indicate statistically significant differences in data among different pyrazine compounds at the same temperature (P < 0.05). The main interaction forces between CP and pyrazine compounds were determined by evaluating thermodynamic parameters using the Van't Hoff equation. Typically, changes in Gibbs free energy (ΔG), enthalpy (ΔH), and entropy (ΔS) determine the spontaneity of binding and the nature of the interaction (Sun et al., 2023). It is generally accepted that ΔH > 0 and ΔS > 0 indicate hydrophobic interactions; ΔH < 0 and ΔS < 0 suggest van der Waals forces and hydrogen bonding; whereas ΔH < 0 and ΔS > 0 indicate electrostatic interactions (Tong et al., 2023). It should be noted that the Ka values for PTRP and PEDP do not strictly follow a Van't Hoff relationship across the entire 298–310 K range; therefore, ΔH and ΔS were calculated only for the 298–303 K range. Previous studies have similarly employed a two-point calculation for protein-flavor compound interactions (Wang et al., 2022; Liu et al., 2025). This non-linear Van't Hoff behavior is likely due to changes in protein conformation or solvent reorganization at higher temperatures, which alters the balance of binding forces. PEP and PEMP exhibited negative ΔH and ΔS values in 298–310 K intervals, implying that van der Waals forces and hydrogen bonding dominate these complexes. Conversely, PTRP and PEDP showed positive ΔH and ΔS values in the 298–303 K interval, indicating the dominance of hydrophobic interactions. As shown in Table 1, all pyrazine–CP complexes exhibited negative ΔG values, confirming spontaneous adsorption. These results further demonstrate that differences in the number of alkyl substituents in pyrazine compounds lead to variations in the CP–pyrazine interaction mode (Tong et al., 2023). Notably, this alkyl-dependent shift in dominant interaction forces has not been observed in myofibrillar protein-pyrazine systems, where electrostatic interactions and van der Waals forces remain predominant regardless of pyrazine structure (Shen et al., 2019; Yu et al., 2023). This alkyl-dependent shift in dominant forces also contrasts with that reported for pea protein (Guo et al., 2024), where hydrophobic interactions dominated for MP (Methylpyrazine) while van der Waals/hydrogen bonding dominated for DP (Dimethylpyrazine) and TRP. This difference may reflect the distinct hydrophobic cavity architecture of CP compared to pea 11S globulin. Synchronous fluorescence spectra Synchronous fluorescence spectroscopy has been widely applied to investigate changes in the fluorophore microenvironment during protein-ligand interactions (Tong et al., 2023). For tyrosine (Tyr) and tryptophan (Trp) residues in proteins, their spectral characteristics can be monitored by setting emission and excitation wavelength intervals (Δλ) at 15 and 60 nm, respectively (Jia et al., 2022). As shown in Fig. 3, CP exhibited decreased fluorescence intensity with increasing pyrazine concentration. A higher maximum intensity was observed at Δλ = 60 nm, indicating that the intrinsic fluorescence of CP mainly originates from Trp residues. A red shift of the maximum emission peak was observed at Δλ = 15 nm, suggesting that Tyr residues in CP became exposed to a more hydrophilic environment (Kou et al., 2025). Interestingly, at Δλ = 60 nm, the maximum emission peak of CP exhibited a red shift in the presence of EP and EMP, but a blue shift in the presence of EDP and TRP. This indicates that binding with EP or EMP increased the polarity (decreased hydrophobicity) around Trp residues, whereas binding with EDP or TRP rendered the microenvironment of Trp residues more hydrophobic (Zhang et al., 2025). Overall, EP, EMP, TRP, and EDP likely bind near both Trp and Tyr residues of CP, and CP undergoes distinct conformational changes upon adsorption of different pyrazines. These changes are directly related to the number of alkyl substituents on the pyrazine ring. Synchronous fluorescence spectra of different concentrations of EP (a), EMP (b), TRP (c), and EDP (d) with CP at Δλ = 15 nm (A) and Δλ = 60 nm (B). Three-dimensional fluorescence spectra measurements Three-dimensional fluorescence spectroscopy was conducted to investigate conformational changes of CP induced by the adsorption of different pyrazines. Fig. 4 presents the three-dimensional (3D) fluorescence contours and corresponding spectral features of the CP–pyrazine systems. Peak 1 represents fluorescence information from amino acid residues, primarily Trp and Tyr, whereas Peak 2 reflects fluorescence characteristics associated with the π→π∗ transition of the C=O group in the polypeptide backbone (Chen et al., 2026). Upon addition of EP (λex/λem = 285.0/325.0 nm, FI = 678.2), EMP (285.0/320.0 nm, 573.6), EDP (285.0/320.0 nm, 436.0), and TRP (285.0/320.0 nm, 399.5) to CP (280.0/315.0 nm, 901.0), the intensity of Peak 1 decreased markedly. The order of peak intensity was PEP > PEMP > PEDP > PTRP, indicating different interaction strengths between CP and the pyrazines. Previous studies have also shown that pyrazines with more alkyl substituents interact more strongly with pea protein and exhibit greater quenching effects on protein fluorescence (Guo et al., 2024). The present results are consistent with these findings, demonstrating that such interactions alter the polarity of the microenvironment surrounding amino acid residues, supporting the fluorescence and synchronous fluorescence results (Sections 3.3, 3.4). As indicated by Peak 2, the addition of EP (λex/λem = 235.0/320.0 nm, FI = 183.7), EMP (235.0/315.0 nm, 189.5), EDP (235.0/325.0 nm, 201.7) and TRP (235.0/315.0 nm, 207.9) to CP alone (235.0/320.0 nm, FI = 190.2) altered both fluorescence intensity and emission maxima. Notably, TRP exhibited the highest fluorescence intensity. Therefore, binding with flavor compounds caused alterations in the secondary structure of the protein (Ren et al., 2023). 3D fluorescence spectra of CP (A), PEP (B), PEMP (C), PTRP (D), and PEDP (E). Note: The different letters in picture (D) indicate significant differences (p < 0.05). SDS-PAGE analysis Fig. 5A shows SDS-PAGE profiles of CP after binding with different pyrazine compounds. The major bands of CP appeared near 33 kDa (acidic globulin subunit α) and 24 kDa (basic globulin subunit β), corresponding to the subunits of the 11S globulin linked by disulfide bonds (Campos et al., 2022). Compared with CP alone, PEP, PEMP, PTRP, and PEDP exhibited similar band patterns. The addition of pyrazine molecules resulted in slight shifts or changes in intensity in the β-subunit region, indicating interactions without altering molecular weight. Moreover, no new bands were observed, suggesting that CP forms complexes with pyrazines via non-covalent interactions (Bi et al., 2022). Similarly, Chen et al. (2026) reported that interactions between myofibrillar protein and thyme phenolics are predominantly non-covalent in nature. SDS-PAGE (A), FTIR spectra (B), CD spectra (C) and secondary structure content (D) of CP and CP-pyrazine complexes. Fourier transform infrared spectroscopy FTIR spectroscopy was conducted to investigate changes in the chemical structure of CP after binding with the four pyrazine compounds. All samples exhibited similar FTIR absorption profiles (Fig. 5B), indicating the presence of analogous functional groups in the pyrazine–CP systems. However, the characteristic peaks of CP shifted to varying extents upon the addition of pyrazines. A noticeable red shift was observed in the broad band around 3400 cm−1 (from 3452 cm−1 to 3416 cm−1), attributed to O–H and N–H stretching vibrations and intermolecular hydrogen bonding, suggesting hydrogen-bond interactions in the pyrazine–CP complexes (Jia et al., 2024). The C–H stretching bands of CP (near 2960 cm−1) also exhibited a red-shift (from 2967 cm−1 to 2961 cm−1) after pyrazine addition, indicating the involvement of hydrophobic interactions in complex formation, consistent with previous studies on pea protein-flavor compound interactions (aldehydes, alcohols, pyrazines, and furans) (Fu et al., 2026). The amide I band (1600–1700 cm−1), mainly associated with C=O stretching vibrations, provides information on protein secondary structure (Jia et al., 2022). Changes in this region were observed for PEP, PEMP, PTRP, and PEDP, indicating alterations in protein secondary structure, which will be further discussed in combination with circular dichroism (CD) results. Circular dichroism spectra To further investigate changes in secondary structure after adsorption of EP, EMP, TRP, and EDP, CD spectroscopy was employed. Fig. 5C and D displays the CD spectra and quantified secondary-structure contents. For native CP, the secondary-structure composition was 30.67% α-helix, 19.35% β-sheet, 21.00% β-turn, and 28.99% random coil. Upon addition of pyrazines, the α-helix content decreased by 1.87%, 2.78%, 3.45%, and 2.68% for PEP, PEMP, PTRP, and PEDP, respectively, while the β-sheet content increased by 0.09%, 2.73%, 1.04%, and 3.24%. No significant change was observed in β-turn content (p > 0.05). The decrease in α-helix content suggests interactions between pyrazine molecules and amino acid residues of the polypeptide chain, which disrupt hydrogen-bond networks and promote partial unfolding, thereby converting the protein from a rigid to a more flexible conformation (Li et al., 2025). The increase in β-sheet content indicates aggregation and exposure of linear hydrophobic groups, likely associated with intra- and intermolecular hydrogen bond formation (Tong et al., 2023). Consistently, as reported by Guo et al. (2024), the introduction of hydrophobic pyrazine molecules can enhance protein hydrophobicity, thereby promoting β-sheet formation. Microstructure by SEM The effects of pyrazine compounds on the microstructure of CP were examined using SEM. As shown in Fig. 6, in the absence of pyrazines, CP exhibited a smooth surface and was uniformly dispersed with only minor aggregation. Upon the addition of different pyrazines, pronounced aggregation was observed. This is likely because pyrazines induced partial unfolding of CP, exposing hydrophobic groups and reducing surface hydrophobicity, thereby promoting aggregation (Ma et al., 2019). Furthermore, unfolding of protein structure enhanced intermolecular interactions, leading to cross-linking and formation of larger aggregates (Zhang et al., 2024). The extent of aggregation varied depending on the alkyl substituents of the pyrazines, reflecting their differential effects on CP surface hydrophobicity and conformation. These microstructural changes further demonstrate that pyrazine compounds alter the conformation of CP. Consistently, Ma et al. (2019) reported similar findings and noted that binding initiates at the protein surface and gradually penetrates into the interior. Surface microstructure of CP (A), PEP (B), PEMP (C), PTRP (D) and PEDP (E) at 8000× magnification. Molecular docking Molecular docking, as a computational modeling approach, is used to predict and validate ligand-protein binding sites and interaction forces, providing insights into interaction characteristics (Sharma et al., 2025). Fig. 7 presents the three-dimensional and two-dimensional structures, binding sites, and binding energies of the optimal complexes formed between CP and EP, EMP, TRP, and EDP, respectively. The docking energy of −5.4, −5.2, −5.0, and −4.6 kcal/mol for PEDP, PTRP, PEMP, and PEP, respectively, indicated that EDP and TRP exhibit higher binding affinity and form more stable complexes with CP than EMP and EP. Visualization of docking results shows that Ser63 of the coffee 11S storage protein forms hydrogen bonds with the nitrogen atoms of EP and EMP, while π–σ and π–π stacking interactions occur with Phe49 and Phe60, and van der Waals interactions are observed between Glu74, Gly215, and Gly214, and hydrophobic contacts with Phe51. These findings indicate that hydrogen bonding, van der Waals forces, and hydrophobic interactions collectively contribute to the binding of EP and EMP to the coffee 11S storage protein. As shown in Fig. 7, TRP forms a hydrogen bond with Ser63, engages in π–σ and π–π stacking with Phe49, Phe60, and Gly215, participates in van der Waals interactions with Glu74, Glu61, Asn64, Lys217, Gly216, Glu218, and Gly214, and exhibits hydrophobic contact with Phe60. For EDP, a hydrogen bond is formed with Glu218, π–σ and π–π stacking interactions occur with Phe49, van der Waals interactions involve Gly216, Asn64, Gln201, Ser63, Gly215, and Gly214, and hydrophobic interactions involve a broader set of residues (Phe49, Phe51, Phe60, and Lys217). This observation is consistent with the surface hydrophobicity analysis presented in Section 3.2. The 2D and 3D plots of PEP, PEMP, PTRP and PEDP complexes best docking model obtained from molecular docking. In summary, hydrogen bonds, hydrophobic interactions, and van der Waals forces are the primary contributors to pyrazine adsorption by CP, consistent with the thermodynamic parameters listed in Table 1. Similarly, Guo et al. (2024) reported that hydrophobic, van der Waals, and hydrogen-bonding interactions dominate the binding of pea protein with pyrazine flavor compounds. According to Bi et al. (2022), adsorption of alcohols and aldehydes by pea protein is attributed to hydrogen bonding and hydrophobic interactions. The molecular docking results further confirm that the number of hydrophobic alkyl groups on the pyrazine ring is a key factor influencing the adsorption and release of flavor compounds. The docking binding energy (−4.6 to −5.4 kcal/mol) is a scoring function value used for ranking relative binding affinities, rather than a direct quantitative measure of the thermodynamic Gibbs free energy. The scoring function is typically calculated in a simplified environment that does not fully account for solvent effects, entropy contributions, and protein conformational dynamics (Zhang et al., 2024). Similar discrepancies between computational and experimental binding energies have been observed in other molecular interaction studies. For instance, Gan et al. (2024) and Liu et al. (2025) utilized the docking scores served as a qualitative indicator rather than a quantitative replacement for thermodynamic parameters. Therefore, in this study, the docking results were used solely to predict binding sites and interaction forces. The trend observed in the docking energy rankings was similar to that of the Gibbs free energy rankings obtained from thermodynamic calculations, confirming the reliability of the computational predictions in this comparative analysis. It should be noted that the binding sites predicted by molecular docking in this study were not validated by site-directed mutagenesis or competitive binding experiments, which would provide more definitive evidence. Future studies employing such experimental approaches are warranted to confirm the predicted interaction sites. MD simulation The reaction mechanisms of CP with various pyrazine compounds were further investigated using electrostatic potential (ESP) and frontier molecular orbital analyses (Hu et al., 2026). ESP mapping provides a visual representation of electrostatic distribution on the molecular surface. All pyrazine molecules exhibited a distinct region of positive electrostatic potential (blue) around the nitrogen atoms. This indicates relatively low electron density in this region, characterizing it as an electrophilic site capable of electrostatic attraction with negatively charged amino acid residues (e.g., glutamic acid and aspartic acid side chains) in proteins (Sepahvandian and Zabardasti, 2026). Frontier molecular orbitals, namely HOMO and LUMO, determine chemical reactivity. A smaller HOMO–LUMO energy gap is associated with greater chemical reactivity and polarizability. As shown in Fig. 8A, PTRP exhibited the smallest energy gap (5.338505 eV), whereas the gaps for PEP, PEMP, and PEDP were 5.411525 eV, 5.430953 eV, and 5.454066 eV, respectively. Therefore, PTRP possesses the highest chemical reactivity and a greater tendency to interact with CP, which explains the stronger adsorption of TRP observed in previous sections. Furthermore, the energy gap, representing the energy required for electron excitation from HOMO to LUMO, reflects both molecular reactivity and stability (Igbokwe et al., 2024). A larger gap indicates greater electronic stability and a lower likelihood of electron transfer. The ESP and HOMO-LUMO isotope visualization (A), secondary structure (B), RMSD (C), RMSF (D), SASA (E) and Rg (F) of PEP, PEMP, PTRP and PEDP complexes. To further investigate the structural effects of pyrazine flavor compounds on coffee protein (CP), alterations in secondary structure were analyzed during MD simulation trajectories. The secondary structures of CP in the presence of different pyrazines exhibited slight and fluctuating changes, with most structural elements remaining unchanged (Fig. 8B). Notable variations were localized around residues 250–310: the PEP, PTRP, and PEDP systems showed changes mainly in 3-helix and α-helix content, whereas the PEMP system exhibited changes primarily in turn structures. Fig. 5C and D also indicate that pyridine compounds induce changes in the α-helix, β-sheet, and β-turns of CP, particularly near residue 100, where the β-sheet structure of PEDP exhibits distinct alterations. This indicates that different pyrazine compounds exert distinct effects on the secondary structure of CP. These findings are consistent with previous reports, such as Zhang et al. (2024), who observed significant secondary-structure alterations in 11S soy protein following piperine treatment, highlighting the influence of small molecules on protein structural stability. RMSD was used to evaluate structural deviation from the initial conformation upon ligand binding (Wang et al., 2022). Lower RMSD values generally correspond to greater structural stability and smaller conformational changes (Zhang et al., 2025). As shown in Fig. 8C, RMSD values for CP, PEP, PEMP, PTRP, and PEDP reached equilibrium after about 40 ns of simulation. The average RMSD of PTRP (0.862 Å) was lower than that of CP (1.041 Å), PEP (1.149 Å), PEMP (1.407 Å), and PEDP (1.172 Å), indicating that PTRP exhibits higher structural stability. The root-mean-square fluctuation (RMSF) profiles of the CP–pyrazine complexes were highly similar (Fig. 8D), suggesting limited fluctuations at residue-level binding sites and only minor conformational adjustments across most amino acid residues. This consistency reflects the stability of the simulated complexes (Feng et al., 2026). The compactness of the CP–pyrazine complexes was further evaluated by combining analyses of SASA (Fig. 8E) and Rg (Fig. 8F). A reduction in SASA indicates that the ligand becomes buried within a complementary protein pocket, thereby minimizing solvent exposure at the interface. Concurrently, a lower Rg value reflects increased overall structural compactness of the protein-ligand system (Chen et al., 2026). Throughout the simulations, all complexes exhibited lower mean SASA values (PEP: 257.2, PEMP: 255.1, PTRP: 257.4, PEDP: 257.5 nm2) than native CP (266.8 nm2), confirming effective burial of the pyrazines. The extent of reduction varied, with PEMP showing the largest decrease (11.7 nm2). Similarly, the Rg values of the complexes (PEP: 2.626, PEMP: 2.617, PTRP: 2.539, PEDP: 2.561 nm) remained consistently below that of CP (2.654 nm), with PTRP forming the most compact complex. The combined SASA and Rg results suggest distinct binding modes. While PEMP induced the most pronounced local burial at the interface (largest ΔSASA), PTRP achieved the highest global structural compactness (lowest Rg). This indicates that TRP, with multiple methyl groups, promotes a tightly integrated complex with improved shape complementarity and hydrophobic contacts, resulting in enhanced overall stability, consistent with its lowest RMSD (Li et al., 2023; Xu et al., 2026). Conversely, binding of EMP, despite effective interfacial burial, may lead to slightly less optimal overall packing, as reflected by its higher RMSD. Conclusion The present study investigated the mechanisms underlying the interactions of CP with alkylpyrazines, demonstrating that the number and position of alkyl groups on the pyrazine ring determine binding affinity and interaction mode. Pyrazines quenched CP fluorescence through combined static and dynamic mechanisms, with quenching efficiency increasing with alkyl substitution. Thermodynamic analysis indicated that hydrophobic interactions dominated for TRP and EDP, whereas van der Waals forces and hydrogen bonding were predominant for EP and EMP. Molecular docking confirmed these interactions as key stabilizing forces. Binding resulted in a consistent decrease in α-helix content and an increase in β-sheet content, indicating partial unfolding of CP. Molecular dynamics simulations confirmed stable complex formation, with TRP forming the most compact and stable complex, consistent with its lowest RMSD and radius of gyration. Quantum chemical calculations provided an electronic-level explanation: TRP exhibited the smallest HOMO–LUMO energy gap, corresponding to its highest reactivity and strongest binding. Although this study used endogenous fluorescence measurements to identify differences in the quenching mechanisms of pyrazine compounds with varying substituents on CP, it did not employ time-resolved fluorescence spectroscopy to confirm the type of quenching involved. Furthermore, it remains unclear whether structural differences in CP arising from different extraction methods could affect its binding affinity toward pyrazine flavor compounds. Therefore, future investigations into the interactions between the coffee matrix and flavor compounds should integrate multiple analytical techniques to cross-validate the results and comprehensively account for how compositional variations in coffee proteins influence their flavor-binding capacity. This study has established a preliminary structure-activity relationship between pyrazine alkylation and binding to coffee proteins, thereby providing a theoretical basis for regulating the retention and release of characteristic flavors in coffee products. CRediT authorship contribution statement Xinxin Yu: Formal analysis, Visualization, Writing – original draft, Writing – review & editing. Zhang Su: Investigation, Methodology, Writing – original draft. Guanghua Xia: Resources, Supervision. Dachuan Zhang: Writing – original draft, Writing – review & editing. Hehe Li: Project administration, Methodology, Data curation. Wenjiang Dong: Validation, Conceptualization, Funding acquisition. 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. Acknowledgments This research was supported by the Hainan Provincial Natural Science Foundation of China (326ZD1016, 325QN458), the Cross-Innovation Open Project of Food Flavor and Health, Beijing Technology & Business University (FFHCI-2025044) , Central Public-interest Scientific Institution Basal Research Fund (1630012026101;1630012025119) and Hainan Provincial Key Research and Development Program of China (ZDYF2026XDNY140). References