Acoustic Cavitation-Induced Unfolding and Solubilization of Velvet Antler Protein for Antioxidant Peptide Release: Substrate Modification Kinetics, Quantum Chemistry, and Keap1/Nrf2-Associated Cellular Responses

Xuanying Xin; Haowen Wang; Sungkwon Park; Changguo Yan; Qiang Li; Rui Du; Xiangzi Li

doi:10.1016/j.ultsonch.2026.107920

. 2026 Jun 8;131:107920. doi: 10.1016/j.ultsonch.2026.107920

Acoustic Cavitation-Induced Unfolding and Solubilization of Velvet Antler Protein for Antioxidant Peptide Release: Substrate Modification Kinetics, Quantum Chemistry, and Keap1/Nrf2-Associated Cellular Responses^☆

Xuanying Xin ^a,^b,¹, Haowen Wang ^a,^b,¹, Sungkwon Park ^c,¹, Changguo Yan ^a,^b, Qiang Li ^a,^b, Rui Du ^a,^b,^⁎, Xiangzi Li ^a,^b,^⁎

PMCID: PMC13276608 PMID: 42276016

Graphical abstract

Keywords: Acoustic cavitation, Velvet antler protein, Protein unfolding, Antioxidant peptides, Keap1-Nrf2 pathway

Abstract

The targeted extraction of bioactive peptides from structurally robust biological matrices, such as velvet antler protein (VAP), is severely hindered by high steric hindrance and profound conformational stability. This study established a multiscale analytical framework to investigate ultrasound-assisted VAP modification and its association with antioxidant peptide release and Keap1/Nrf2-related cellular responses. High-intensity ultrasound pretreatment (optimized at 450 W, 30 min, 30 mL/g) effectively deconstructed the dense interfacial architecture of VAP, sub-micronizing the particle size to ∼175 nm and promoting an ultrasound-associated conformational transition from α −helices to random coils. Thermodynamic deconstruction revealed that acoustic shear forces significantly attenuated the denaturation enthalpy (ΔH) and elevated surface hydrophobicity (H₀), alleviating steric constraints and driving an increase in targeted peptide yield from 44.82% to 68.32%. By integrating peptidomics, machine learning, and density functional theory (DFT), five representative antioxidant lead candidates were prioritized and subsequently synthesized for validation. DFT analyses suggested that these selected sequences may possess favorable electron-donating properties, potentially associated with aromatic residues, adjacent hydrophobic residues, and Pro-related conformational restriction. Furthermore, molecular docking suggested that these peptides may interact with the Keap1 Kelch domain, while in vitro cellular assays showed that the selected peptides restored endogenous antioxidant enzymes (SOD, CAT, GSH-Px) in H₂O₂-challenged RAW264.7 macrophages. These findings provide physicochemical and cellular evidence supporting ultrasound-assisted VAP modification and antioxidant peptide discovery, offering a laboratory-scale, mechanistically guided strategy for discovering functional peptides from structurally dense biological resources. Nevertheless, further pilot-scale validation is required before industrial translation.

1. Introduction

With advances in enzymatic hydrolysis, physical pretreatment, peptidomics, and computational screening, substantial progress has been made in the preparation and preliminary discovery of food-derived bioactive peptides. Therefore, the current challenge is no longer limited to peptide extraction itself, but increasingly lies in the targeted release, identification, classification, and mechanistic validation of specific active peptides from complex hydrolysates. This issue is particularly evident for structurally robust biological matrices, in which compact protein architecture can limit protease accessibility and obscure the relationship between substrate modification, peptide release, and biological function. Velvet antler (Cervus elaphus), a unique mammalian organ characterized by its capacity for complete annual regeneration, exhibits antioxidant and anti-aging activities and contains more than 50% protein by dry weight, making it a promising reservoir of bioactive peptides [1], [2]. However, native velvet antler protein (VAP) possesses a compact and highly ordered structure, which reduces aqueous solubility and restricts the exposure of proteolytic cleavage sites. As a result, potentially active peptide sequences may remain embedded within the protein matrix and cannot be efficiently released or accurately identified by conventional hydrolysis alone [3]. Under conventional enzymatic hydrolysis conditions, exogenous proteases struggle to surmount the high thermodynamic barrier required for substrate unfolding. Consequently, numerous potentially active antioxidant sequences, especially hydrophobic and aromatic residue-rich segments, may remain embedded within the compact protein core, limiting both their enzymatic release and subsequent identification. Thus, a key scientific question is how physical substrate modification can improve peptide release while enabling the accurate screening and mechanistic classification of antioxidant peptides from the resulting complex peptide mixture. Therefore, physical pretreatment that loosens the compact protein structure may improve protease accessibility and promote the release of bioactive peptides [4].

Among a plethora of non-thermal physical pretreatment technologies, sonochemical processing techniques—notably high-intensity low-frequency ultrasound (HILFU)—have garnered significant attention owing to their exceptional capacity for substrate modification. The fundamental mechanism is rooted in sound-field-induced acoustic cavitation, characterized by the violent asymmetric collapse of transient microbubbles. Acoustic cavitation generates localized shear forces, micro-jets, and shock waves that can modify protein structure [5]. Extant literature indicates that these vigorous mechanical stresses can propagate along the protein backbone, effectively disrupting the non-covalent interactions stabilizing the α-helix. This process induces protein dissociation and facilitates the exposure of buried hydrophobic residues, thereby markedly enhancing the degree of hydrolysis (DH) and the apparent solubilization of the substrate. Nevertheless, the current body of research is constrained by notable methodological and theoretical lacunae [6], [7]. Most studies remain limited to descriptive accounts of macro-scale improvements in hydrolysis or the total antioxidant capacity of crude extracts, often regarding cavitation-induced modification as a purely phenomenological process. However, how ultrasound alters protein unfolding and thermodynamic stability remains insufficiently understood. Furthermore, the sub-microscopic mechanisms by which such physical modifications govern the specific release of targeted antioxidant sequences remain poorly elucidated.

Specific peptides liberated via ultrasonic-mediated physical modification are primarily intended to target and modulate cellular oxidative stress networks. Within the defense architectures safeguarding against both endogenous and exogenous oxidative damage, the Keap1-Nrf2 signaling axis has been established as a pivotal regulatory hub [8]. Some food-derived antioxidant peptides may interact with the Kelch domain of Keap1 and thereby influence Keap1/Nrf2-associated antioxidant responses. This interaction disrupts the Keap1-Nrf2 protein–protein interaction via steric hindrance, subsequently facilitating Nrf2 nuclear translocation and the up-regulation of downstream Phase II detoxifying enzymes [9], [10]. Despite the identification of these molecular targets, the accurate screening, prioritization, and experimental validation of active peptides from complex ultrasonic hydrolysates remain technically challenging. In particular, LC-MS/MS can identify a large number of peptide sequences, but these sequences should be regarded as putative candidates until selected peptides are synthesized or purified and experimentally validated [11]. Similarly, the electronic basis of these peptides as chemical entities—specifically their capacity to quench free radicals via single-electron transfer (SET) or hydrogen atom transfer (HAT) mechanisms—remains obscure at the molecular orbital level.

Therefore, this study aimed to link ultrasound-induced VAP structural modification with antioxidant peptide release and cellular mechanism validation. First, ultrasound pretreatment was optimized, and changes in particle size, morphology, secondary structure, thermodynamic stability, and surface hydrophobicity were characterized. Second, LC-MS/MS-based peptidomics and machine learning were used to screen candidate peptides. Third, DFT calculations and molecular docking were applied to evaluate their electronic antioxidant features and Keap1-binding potential. Finally, five lead peptides were synthesized and validated in H₂O₂-challenged RAW264.7 macrophages. Overall, this work establishes a laboratory-scale mechanistic framework for ultrasound-assisted VAP modification and antioxidant peptide discovery. However, the feasibility of pilot-scale or industrial-scale implementation, including acoustic field uniformity, energy efficiency, reactor configuration, and continuous processing performance, requires further dedicated investigation.

2. Materials and methods

2.1. Materials and main reagents

Velvet antler powder was purchased from a local market, pulverized, and passed through an 80-mesh sieve prior to use. Alcalase (enzyme activity >200,000 U/g), 1,1-diphenyl-2-picrylhydrazyl (DPPH), 2,2′-azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), 8-anilino-1-naphthalenesulfonic acid (ANS), 5,5′-dithiobis (2-nitrobenzoic acid) (DTNB), ferrozine, and Coomassie Brilliant Blue G-250 were obtained from Solarbio (Beijing, China). Fmoc-amino acid derivatives, resins, and trifluoroacetic acid (TFA) for solid-phase peptide synthesis (SPPS) were procured from Shanghai GL Biochem Co., Ltd. (Shanghai, China). HPLC-grade formic acid (FA) and acetonitrile (ACN) were purchased from Merck KGaA (Darmstadt, Germany). Fetal Bovine Serum, FBS (BaiDi Biotechnology Co., Ltd. (BDBIO)) (Shanghai, China). CCK-8 assay, BCA protein assay, RIPA lysis buffer, and nuclear and cytoplasmic extraction kits were supplied by Beyotime Biotechnology (Shanghai, China). Commercial kits for the determination of superoxide dismutase (SOD), glutathione peroxidase (GSH-Px), and malondialdehyde (MDA) were purchased from Nanjing Jiancheng Bioengineering Institute (Nanjing, China). Primary antibodies (anti-Nrf2, anti-Keap1, anti-HO-1, and anti-β-actin) and horseradish peroxidase (HRP)-conjugated secondary antibodies were obtained from Abcam (Cambridge, MA, USA). All other chemical reagents were of analytical grade, and ultrapure water (18.2 MΩ·cm) was prepared using a Milli-Q system (Millipore, Bedford, MA, USA).

2.2. Ultrasound pretreatment and determination of protein solubility

The velvet antler substrate was suspended in ultrapure water at prescribed liquid-to-solid (L/S) ratios. Ultrasound pretreatment was conducted using an ultrasonic processor (SCIENTZ-IID, Ningbo Scientz Biotechnology Co., Ltd., Ningbo, China) equipped with a 6-mm diameter titanium alloy probe operating at 20 kHz (maximum output: 900 W). The probe was immersed 1.5–2.0 cm below the liquid surface and centrally positioned to ensure homogeneous acoustic cavitation and minimize energy dissipation. Sonication was operated in pulsed mode (2 s on/2 s off) within an ice-water bath. The thermal protection threshold was set at 55 °C to preclude irreversible thermal denaturation of the proteins. To establish the parameter ranges for the Box-Behnken design (BBD), single-factor experiments were conducted using protein solubility (%) as the primary response variable. The effects of ultrasound power (150, 300, 450, 600, and 750 W), duration (10, 20, 30, 40, and 50 min), and L/S ratio (10, 20, 30, 40, and 50 mL/g) were evaluated independently by varying one parameter while maintaining the others at their respective central levels (450 W, 30 min, and 30 mL/g). The resulting suspensions were centrifuged at 8000×g for 15 min at 4 °C, and the supernatants were harvested. The soluble protein content was determined via the Bradford assay. Solubility was quantified as the ratio of the mass of soluble protein in the supernatant to the total initial protein mass of the substrate [12].

2.3. Optimization of ultrasound pretreatment via response surface methodology (RSM)

Building upon the preliminary findings of the single-factor experiments, a BBD was implemented to optimize the ultrasound-assisted extraction (UAE) parameters. A three-factor, three-level experimental matrix was constructed using Design-Expert 12.0 software (Stat-Ease, Inc., Minneapolis, MN, USA). Ultrasound power (X₁), duration (X₂), and L/S ratio (X₃) were designated as independent variables, while protein solubility (Y) served as the response variable. The randomized experimental design comprised 17 runs, incorporating five replicate center points to assess pure error and ensure model reproducibility. The experimental data were fitted to the following second-order polynomial regression model:

Y = β_{0} + \sum_{i = 1}^{3} β_{i} X_{i} + \sum_{i = 1}^{3} β_{i i} X_{i}^{2} + \sum_{i = 1}^{2} \sum_{j = i + 1}^{3} β_{i j} X_{i} X_{j}

Note: Where Y denotes the predicted response (solubility); β₀ is the intercept; β_i, β_ii, and β_ij represent the regression coefficients for the linear, quadratic, and interaction effects, respectively; and X_i and X_j signify the coded levels of the independent variables.

2.4. Characterization of micro-morphology and dispersion properties

The micro-morphological features of the freeze-dried control (CK) and ultrasound-pretreated (US) samples were examined via Scanning Electron Microscopy (SEM). Briefly, the samples were mounted onto aluminum stubs using double-sided conductive adhesive tape and subsequently sputter-coated with a thin layer of gold under vacuum to enhance conductivity. Surface topography was observed using a field-emission scanning electron microscope (SU8010, Hitachi High-Tech Corp., Tokyo, Japan) operating at an accelerating voltage of 5.0 kV. For dispersion characteristics, the samples were diluted to a concentration of 1.0 mg/mL and equilibrated at 25 °C for 120 s. The average hydrodynamic diameter and ζ −potential were determined using a dynamic light scattering (DLS) instrument (Zetasizer Nano ZS, Malvern Panalytical, UK). All analyses were performed in triplicate, and the results are expressed as the mean ± standard deviation.

2.5. Analysis of molecular conformation and thermodynamic properties

CD Spectroscopy: The secondary structure of the VAP samples was analyzed using a CD spectrometer (J-1500, JASCO, Japan). Samples were diluted to 0.1 mg/mL in phosphate buffer and scanned in the far-ultraviolet (far-UV) region (190–250 nm). A quartz cuvette with a 0.1-cm path length was utilized, with a scanning speed of 100 nm/min and a bandwidth of 1.0 nm. After baseline correction by subtracting the solvent spectrum, the raw ellipticity (mdeg) was recorded and used for comparative CD spectral analysis. Mean residue ellipticity was not calculated because the purpose of this experiment was to compare relative secondary structural changes between CK and US samples rather than to determine absolute secondary structure contents. Since all samples were standardized to the same protein concentration (0.1 mg/mL) and measured using the same 0.1-cm path length cuvette under identical instrumental conditions, comparison of raw ellipticity profiles was considered valid for evaluating ultrasound-associated conformational changes.

Differential Scanning Calorimetry (DSC): The thermal stability and thermodynamic transitions of the proteins were evaluated using a DSC system. Approximately 3.0 mg of freeze-dried sample was weighed into an aluminum pan and hermetically sealed, with an empty pan used as a reference. The samples were heated from 20 to 120 °C at a constant rate of 10 °C/min under a continuous nitrogen purge (50 mL/min). The peak denaturation temperature (Tp) and the enthalpy of denaturation (ΔH) were calculated using the manufacturer's software [13].

Determination of Surface Hydrophobicity (H₀): Surface hydrophobicity was determined using 8-anilino-1-naphthalenesulfonic acid (ANS) as an extrinsic fluorescent probe. Protein solutions were serially diluted to concentrations of 0.05, 0.1, 0.2, 0.3, and 0.4 mg/mL. A 4.0-mL aliquot of each sample was mixed with 20 μL of ANS (8 mM in 0.01 M phosphate buffer, pH 7.0) and incubated in the dark at room temperature for 15 min. The fluorescence intensity was measured using a fluorescence spectrophotometer with excitation and emission wavelengths set at 390 nm and 470 nm, respectively. The initial slope of the linear regression of fluorescence intensity against protein concentration was calculated and designated as the surface hydrophobicity index (H₀)[14].

2.6. Preparation of velvet antler peptides and determination of yield

The substrate dispersions resulting from the optimized ultrasound treatment (US) and the control group (CK) were adjusted to pH 8.0. Alcalase was then introduced at an enzyme-to-substrate (E/S) ratio of 2% (w/w, based on the total protein content). Enzymatic hydrolysis was conducted at 50 °C for 4 h in a thermostatic water bath shaker (150 rpm). Subsequently, the reaction was terminated by heating the mixture in a boiling water bath for 10 min to ensure complete enzyme inactivation. After cooling to room temperature, the hydrolysates were centrifuged at 8000×g for 15 min at 4 °C. The resulting supernatants were harvested and lyophilized (−80 °C, 48 h) to obtain the final peptide powders, designated as US-peptides and CK-peptides, respectively. The peptide yield (%) was defined and calculated as the ratio of the mass of the lyophilized peptide powder to the total initial protein mass of the velvet antler in the enzymatic hydrolysis system, multiplied by 100%.

2.7. Determination of multi-dimensional antioxidant activity in vitro

The lyophilized peptide powders were reconstituted into a series of aqueous solutions with concentrations ranging from 1.0 to 5.0 mg/mL. Except for the reducing power, the radical and ion scavenging activities were calculated according to the following generalized equation:

Scavenging rate (\%) = (1 - \frac{A_{s} - A_{b}}{A_{c}}) \times 100

where As represents the absorbance of the reaction system containing the sample; Ab is the background absorbance of the sample (with deionized water replacing the chromogenic reagent); and Ac denotes the absorbance of the blank control (with deionized water replacing the sample).

For the DPPH assay, a 2.0-mL aliquot of the sample was mixed with 2.0 mL of DPPH-ethanol solution (0.1 mM) and incubated in the dark for 30 min, after which the absorbance was measured at 517 nm. For the ABTS assay, 0.5 mL of the sample was reacted with 4.0 mL of ABTS working solution for 6 min in the dark, and the absorbance was recorded at 734 nm [15], [16].

Reducing power was determined via the potassium ferricyanide method. Briefly, a 1.0-mL sample was mixed with equal volumes of phosphate buffer (0.2 M, pH 6.6) and 1% (w/v) potassium ferricyanide, followed by incubation at 50 °C for 20 min. The reaction was terminated by adding 10% (w/v) trichloroacetic acid. After centrifugation, 1.0 mL of the supernatant was reacted with ferric chloride, and the absorbance was measured at 700 nm.

The scavenging activities against highly toxic reactive oxygen species (ROS), including hydroxyl radicals (⋅ OH) and superoxide anions (O₂⁻), were evaluated using the salicylic acid-ferrous sulfate method (510 nm) and the pyrogallol autoxidation assay (320 nm), respectively. Furthermore, the ferrous ion (Fe²⁺) chelating activity was quantified via the ferrozine-complexation assay at 562 nm.

2.8. LC-MS/MS analysis of velvet antler peptides

2.8.1. Peptide desalting and pretreatment

The reconstituted peptides were desalted via C18 solid-phase extraction (SPE) using Sep-Pak cartridges (Waters Corp., Milford, MA, USA). The resulting eluates were concentrated using a vacuum centrifugal evaporator, re-dissolved in LC-MS grade ultrapure water containing 0.1% (v/v) FA, and quantified via a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) prior to storage at −80 °C for subsequent analysis.

2.8.2. Mass spectrometry identification

Peptide separation and identification were performed using an EASY-nLC 1200 system coupled online to a Q Exactive HF-X high-resolution mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). The samples were initially loaded onto a C18 trapping column (Acclaim PepMap 100, 100 μm × 2 cm) for enrichment and subsequently separated on a C18 analytical column (75 μm × 25 cm, 2 μm). Mobile phase A consisted of 0.1% (v/v) FA in ultrapure water, whereas mobile phase B comprised 80% (v/v) acetonitrile (ACN) containing 0.1% (v/v) FA. A 60-min linear gradient was implemented from 2% to 100% mobile phase B at a constant flow rate of 300 nL/min. The mass spectrometer was operated in data-dependent acquisition (DDA) mode, with a full-scan resolution of 70,000 across an m/z range of 100–1500. The top 20 most abundant precursor ions were selected for higher-energy collisional dissociation (HCD) fragmentation at a normalized collision energy (NCE) of 28%, with an MS/MS resolution of 15,000.

2.8.3. Sequence identification and bioinformatics analysis

Raw data files were imported into PEAKS Studio 11 software (Bioinformatics Solutions Inc., Waterloo, ON, Canada) and searched against the UniProtKB Cervidae database. To account for the non-specific nature of enzymatic hydrolysis, the enzyme specificity was set to “None” to allow non-specific peptide identification. Mass tolerances for precursor and fragment ions were set at 20 ppm and 0.02 Da, respectively. No fixed modification was specified. Oxidation (M), N-terminal acetylation, and hydroxylation of proline (P) and lysine (K)—to account for the collagenous nature of velvet antler—were set as variable modifications. The global false discovery rate (FDR) was strictly maintained at ≤1% at the peptide-spectrum match (PSM) level. Key physicochemical properties, including theoretical molecular weight (MW), isoelectric point (pI), hydrophobicity scores, and terminal residue distribution frequencies, were calculated using Biopython and custom Python scripts.

2.9. Machine learning-based high-throughput screening of bioactive peptides

To circumvent the inherent limitations and labor-intensive nature of conventional empirical screening, the 7,269 high-confidence peptide sequences identified via LC-MS/MS were subjected to high-throughput in silico screening using the PeptideRanker server. This predictive platform, utilizing an N-to-1 neural network algorithm, evaluates peptides based on their sequence-specific features and physicochemical properties, providing bioactivity probability scores ranging from 0 to 1. The top 50 peptides with the highest predicted scores were prioritized as candidate sequences for further computational analysis. These sequences were subsequently subjected to molecular docking and quantum chemical analyses to evaluate their potential interaction mechanisms and antioxidant-related electronic properties. The PeptideRanker output was used only as an initial prioritization step and does not represent direct experimental evidence of bioactivity.

2.10. Quantum chemical calculations in a physiological aqueous microenvironment

Geometry optimizations of all molecular structures were performed without symmetry constraints at the B3LYP/6-311G(d) level of theory using the Gaussian 16 software package, incorporating Grimme’s D3 dispersion correction with Becke–Johnson damping (D3BJ). To accurately simulate the physiological aqueous microenvironment, implicit solvation effects were accounted for using the polarizable continuum model (PCM), with water designated as the dielectric medium. Subsequent wavefunction analyses were conducted using Multiwfn 3.8 (dev) to extract key descriptors within the conceptual density functional theory (CDFT) framework, including frontier molecular orbitals (FMOs). These computational descriptors provided high-resolution mechanistic insights into the local reactivity of the peptides and their susceptibility to free radical attack. Finally, three-dimensional molecular electrostatic potential (ESP) surfaces and interaction interfaces were rendered for visualization using VMD software [17], [18].

2.11. Molecular docking and molecular dynamics simulations

The high-resolution crystal structure of the Keap1 Kelch domain (PDB ID: 2FLU) was retrieved from the RCSB Protein Data Bank. Structural refinement, including the removal of crystallographic water molecules, co-crystallized ligands, and heteroatoms, was performed using PyMOL. The predominant solution-phase conformations of the peptides, pre-optimized via DFT in an aqueous environment, were employed as ligands. Following energy minimization of both the receptor and ligands using MOE 2019 software (Chemical Computing Group, Montreal, QC, Canada), the active binding pocket of Keap1 was precisely defined. Flexible molecular docking was executed via the MOE Dock module, and the binding affinity was quantified by the lowest binding free energy (kcal/mol). Finally, the optimal docking poses and their underlying non-covalent interaction networks were visualized and analyzed in three dimensions using PyMOL [19]. To further evaluate the predicted dynamic stability of the Keap1–peptide complexes under physiological-like conditions, all-atom molecular dynamics (MD) simulations were performed according to a previously reported method with minor modifications [20]. The initial structures of the Keap1–peptide complexes obtained from molecular docking were used as the starting conformations for MD simulations. All simulations were conducted using the AMBER 24 software package. The Amber14SB force field was used to parameterize Keap1 and the peptides, and the TIP3P water model was used to describe water molecules. Each complex was placed in a truncated octahedral box filled with TIP3P water molecules, with a minimum distance of 10.0 Å between the complex and the box boundary. Na⁺ and Cl⁻ ions were added to neutralize the system and maintain an ionic strength of 0.15 M. Before production simulation, each system underwent a multi-step equilibration procedure, including 10,000 steps of energy minimization, gradual heating for 100 ps under the NVT ensemble, and 1 ns of NPT equilibration with positional restraints. Subsequently, a 100 ns unrestrained production simulation was performed under the NPT ensemble at 300 K and 1 bar. The Langevin thermostat was used for temperature control, and the Berendsen barostat was used to maintain pressure. The SHAKE algorithm was applied to constrain bonds involving hydrogen atoms, allowing a time step of 2 fs. Trajectory analyses, including root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (Rg), solvent-accessible surface area (SASA), and hydrogen bond number, were performed using the CPPTRAJ module. The binding free energy of each Keap1–peptide complex was calculated using the MM-PBSA method implemented in the MMPBSA.py script. For binding free energy calculations, 200 representative conformations were extracted from the equilibrated trajectory segment of 80–100 ns. The binding free energy values were expressed as mean ± SD to evaluate the predicted binding stability and account for statistical fluctuations during the simulation.

2.12. SPPS of lead peptides

Five core lead peptides (WFL, WGPF, WIF, IFW, and GFPWP) were chemically synthesized via standard Fmoc-based SPPS. Successive cycles of amino acid coupling and deprotection were performed in the C-to-N direction on the solid-phase resin, with coupling efficiency monitored in real-time via the ninhydrin (Kaiser) test. Following synthesis, the peptides were cleaved from the resin using a TFA-based cleavage cocktail. The resulting crude peptides were precipitated in ice-cold diethyl ether, and their molecular weights were verified by mass spectrometry. Finally, the peptides were purified via preparative reverse-phase high-performance liquid chromatography (RP-HPLC). All synthesized peptides used for subsequent cell-based assays attained a chromatographic purity of ≥95%. Therefore, only five representative lead candidates were chemically synthesized and used for subsequent in vitro validation, whereas the remaining predicted sequences were considered putative candidates requiring further validation.

2.13. Cell culture and in vitro assays

2.13.1. Cell culture and cytotoxicity assessment

RAW264.7 macrophages were maintained in high-glucose Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) penicillin–streptomycin, and incubated at 37 °C in a humidified atmosphere containing 5% CO₂. To evaluate the potential cytotoxicity of the peptides, a Cell Counting Kit-8 (CCK-8) assay was employed. Cells in the logarithmic growth phase were seeded into 96-well plates and allowed to adhere for 24 h. Subsequently, the cells were exposed to varying concentrations of peptides (0–400 μM) for another 24 h. Following incubation with the CCK-8 reagent according to the manufacturer’s instructions, the absorbance at 450 nm was recorded using a microplate reader, and the cell viability was quantified.

2.13.2. Determination of intracellular antioxidant indices and gene regulation

Cells were seeded into 6-well plates and pretreated with peptides at non-toxic concentrations for 24 h, followed by a challenge with 500 μM H₂O₂ for 4 h to induce oxidative stress. Following cell lysis, the activities of endogenous superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GSH-Px), along with the levels of malondialdehyde (MDA), were quantified in strict accordance with the manufacturer's instructions. Simultaneously, total RNA was extracted and reverse-transcribed into cDNA. The relative mRNA expression levels of pivotal antioxidant genes (Nrf2 and HO-1) were determined via quantitative real-time PCR (qRT-PCR) with 40 amplification cycles. The data were normalized to the internal reference gene β-actin and analyzed using the 2^−ΔΔ^Ct method.

2.13.3. Western blot analysis

Total cellular protein fractions were isolated and quantified using a BCA protein assay kit. Equivalent amounts of protein were subjected to SDS-PAGE and electrotransferred onto polyvinylidene fluoride (PVDF) membranes. The membranes were blocked with 5% (w/v) non-fat dry milk and sequentially incubated with specific primary antibodies against Nrf2, Keap1, HO-1, phosphorylated NF-κB p65 (p-p65), total NF-κB p65 (p65), phosphorylated IκBα (p-IκBα), total IκBα, and β-actin overnight at 4 °C. After washing, the membranes were incubated with HRP-conjugated secondary antibodies for 2 h at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) detection system and imaged. The relative protein expression levels within the Keap1/Nrf2 and NF-κB signaling pathways were quantified by densitometry using ImageJ software. HO-1, Keap1, Nrf2, p65, and IκBα were normalized to β-actin, whereas phosphorylated proteins were normalized to their corresponding total proteins, including p-p65/p65 and p-IκBα/IκBα.

2.14. In silico ADME prediction of lead peptides

To further evaluate the potential gastrointestinal behavior and translational relevance of the five lead peptides, key ADME-related parameters of WFL, WIF, IFW, WGPF, and GFPWP were predicted using the ADMETlab 3.0 platform (https://admetlab3.scbdd.com). The canonical SMILES or molecular structures of the five synthesized peptides were submitted to the ADMET Evaluation module. Six parameters related to gastrointestinal behavior and metabolic persistence were selected, including Caco-2 permeability, human intestinal absorption (HIA), aqueous solubility (LogS), distribution coefficient at physiological pH (LogD7.4), plasma half-life, and liver microsomal clearance. These parameters were used to provide preliminary information regarding aqueous stability, intestinal absorption potential, membrane permeability, and metabolic persistence.

2.15. Statistical analysis

All data are expressed as the mean ± standard deviation (SD) and were obtained from at least three independent biological replicates. Statistical analyses were performed using SPSS version 27.0 (IBM Corp., Armonk, NY, USA). Significant differences between experimental groups were evaluated using one-way analysis of variance (ANOVA) followed by Duncan’s multiple range test. A value of P < 0.05 was considered to be statistically significant. Graphical representations were generated using GraphPad Prism software (version 9.0, GraphPad Software, San Diego, CA, USA).

3. Results

3.1. Influence of ultrasonic process optimization on the solubility of velvet antler protein

3.1.1. Influence of operational variables on substrate dissolution kinetics

The spatiotemporal distribution of acoustic energy within the liquid matrix is a pivotal physical determinant of protein disaggregation efficiency. As illustrated in Fig. 1A, under fixed sonication duration and liquid-to-solid (L/S) ratio, the solubility of the velvet antler extract exhibited a non-linear response to increasing ultrasonic power, characterized by an initial increment followed by a subsequent decline. Solubility reached its maximum value at an ultrasonic power of 450 W. The solubilization effect during this phase is primarily ascribed to the micro-jets and mechanical shear stresses induced by transient acoustic cavitation. These physical forces effectively disrupted the compact architecture at the substrate-particle interface, facilitating the diffusion of soluble components into the aqueous phase. Concurrently, moderate acoustic energy input severed specific intramolecular hydrogen bonds stabilizing the macromolecular conformation, triggering partial unfolding and increasing the available hydration sites for solvent interaction. However, an attenuation in solubility was observed when the ultrasonic power surpassed 450 W. This suggests that the excessive acoustic energy and intense shear stress caused an overexposure of internal hydrophobic residues. Within the strictly controlled ice-water bath environment, this severe disruption of the hydrophilic-hydrophobic balance triggered macromolecular re-aggregation driven by profound hydrophobic interactions, ultimately leading to the secondary precipitation of previously soluble fractions.

Similarly (Fig. 1B), the sonication duration determines the cumulative acoustic dose delivered to the substrate. Solubility trended upward within the 10–30 min interval, reaching a maximum at 30 min, indicating that the initial acoustic field effectively loosened the three-dimensional protein network and enhanced solid–liquid mass transfer efficiency. However, a regression in solubility occurred when the duration was extended beyond 40 min. This phenomenon may be primarily attributed to the exposure of reactive sites on disaggregated macromolecules under prolonged intense shear, which subsequently re-crosslinked via disulfide bonds or strong hydrophobic interactions into insoluble polymers. The L/S ratio (Fig. 1C) exerts a dual regulatory effect on the concentration gradient and the acoustic field distribution within the extraction system. Increasing the L/S ratio from 10 to 30 mL/g significantly enhanced solubility, as the larger solvent volume boosted the driving force of the concentration gradient and diminished fluid viscosity, thereby optimizing the efficiency of cavitation bubble formation and implosion. A further increase to 50 mL/g resulted in a plateau or slight reduction in solubility, primarily due to the dilution of acoustic energy density per unit volume, which attenuated the effective shear forces acting on the solid-phase interface. Previous studies have shown that moderate ultrasound can improve protein solubility or dispersion by cavitation-induced disaggregation, partial unfolding, particle size reduction, and exposure of water-accessible groups, whereas excessive ultrasound may induce hydrophobic association and secondary aggregation. Similar effects have been reported in walnut protein isolate, and soy protein isolate [21], [22]. Consistent with these studies, VAP solubility in the present work showed a non-linear response to ultrasound power, duration, and L/S ratio. However, unlike most plant or dairy proteins, VAP is a collagen-rich and structurally dense animal-derived matrix; therefore, the solubility improvement observed here was further connected with interfacial disruption, sub-micronization, conformational destabilization, and subsequent peptide yield enhancement.

3.1.2. Construction of the response surface model and analysis of Multi-variable coupling effects

To quantitatively elucidate the interactive effects among the operational variables, a second-order polynomial response surface model was constructed using a Box–Behnken design (BBD). The experimental design matrix and observed solubility values are provided in Supplementary Table S2, and the ANOVA results of the quadratic model are summarized in Supplementary Table S3. The model was highly significant (F = 140.53, P < 0.0001), whereas the lack of fit was not significant (F = 3.62, P = 0.1228), indicating that the model adequately described the relationship between ultrasound parameters and VAP solubility. The high coefficient of determination (R² = 0.9945) and adjusted R² (0.9874) further confirmed the good agreement between the predicted and experimental values.

Examination of the two-dimensional contour plots (Fig. 1D–F) and three-dimensional response surface plots (Fig. 1G–I) indicated that the gradient along the ultrasonic power axis was the most pronounced, suggesting that ultrasonic power was the predominant factor affecting VAP solubilization under the tested conditions. The contour profiles further indicated interaction effects between ultrasonic power and duration, as well as between ultrasonic power and L/S ratio. This pattern indicates significant synergistic and constraining coupling effects between these variable pairs. These findings confirm that efficient protein unfolding and solubilization are not solely dependent on high energy input, but rather require the precise synchronization of acoustic energy density, cumulative exposure time, and fluid hydrodynamic properties to mitigate secondary aggregation caused by localized overheating. By solving the second-order regression equation and calibrating for practical laboratory conditions, the optimal extraction parameters were determined to be an ultrasonic power of 450 W, a duration of 30 min, and an L/S ratio of 30 mL/g. This response surface result is consistent with recent experimental studies on ultrasound-assisted protein modification, in which moderate ultrasound treatment rather than maximum energy input produced the most favorable response. Adal [23] reported that high-intensity ultrasound increased the solubility of faba bean protein isolate, with the maximum solubility observed at an appropriate sonication time, indicating a time-dependent response to ultrasound treatment. Wang et al. [24] also showed that appropriate ultrasound power promoted the depolymerization of oxidized soybean protein aggregates and improved their functional properties, whereas high-power ultrasound induced a tendency toward reaggregation. Consistent with these findings, VAP showed a non-linear response to ultrasound conditions in the present study. Excessive ultrasound power or prolonged treatment reduced solubility, probably due to the overexposure of hydrophobic regions and subsequent secondary aggregation. However, unlike previous studies that mainly focused on plant protein isolates and their functional properties, the present study used VAP solubility as the primary response variable and further linked the optimized ultrasound conditions with interfacial disruption, particle sub-micronization, conformational destabilization, surface hydrophobicity, and subsequent peptide yield.

3.2. Deconstruction of the solid-liquid interface and sub-micronization of particle size distribution

The pronounced surge in macro-solubility is governed by the fundamental remodeling of the substrate's micro-physical architecture. Scanning electron microscopy (SEM; Fig. 1N) revealed that the untreated control (CK) exhibited a continuous, dense, and coalesced block-like morphology. Following optimized ultrasonic intervention (US), the intense mechanical erosion caused by cavitation micro-jets and shockwaves induced the dissociation of these aggregates into irregular micro-particles characterized by porous features and heightened surface roughness. DLS (Fig. 1L) further quantified this disaggregation kinetics; while the CK group exhibited an extremely broad particle size distribution (centered at ∼613 nm), the US group showed a significant leftward shift and convergence, with the primary peak migrating to ∼175 nm. This drastic reduction in spatial scale, coupled with the formation of a porous matrix, exponentially increased the effective interfacial contact area, providing the geometric prerequisite for the profundity of solvent penetration.

3.3. Irreversible conformational unfolding and secondary structure evolution

Acoustic cavitation-induced mechanical stress drives the profound evolution of protein spatial conformation. CD analysis (Fig. 1J) was performed using raw ellipticity profiles recorded under identical protein concentration and path length conditions. The CK group exhibited characteristic negative bands at approximately 208 nm and 222 nm, indicative of an α-helical structural component. Following sonication, the magnitude of the negative peaks in the US group significantly attenuated, suggesting the partial disruption of the intramolecular hydrogen bonding network that stabilizes the rigid secondary structure. Furthermore, the two spectra intersected to form a clear isodichroic point near 203 nm, with an intensified negative absorption trend at 198 nm in the US group. This spectroscopic pattern suggests that ultrasound pretreatment was associated with a conformational transition of VAP from a more ordered α-helical structure toward a more flexible random-coil-rich state. However, this CD result should be interpreted as spectroscopic evidence of structural modification rather than direct mechanistic proof of the specific cavitation forces responsible for the transition.

3.4. Thermodynamic destabilization and exposure of the hydrophobic microenvironment

The extension of the molecular conformation directly triggered thermodynamic destabilization within the system. DSC thermograms (Fig. 1K) revealed that, compared to the sharp endothermic denaturation peak of the CK group, the US group exhibited a marked downward shift in the denaturation temperature (Tp) and a significant contraction in the enthalpy of denaturation (ΔH). This reduction in the thermal stability barrier provides thermodynamic evidence for the loose structural reorganization forced by the physical field. The disruption of cohesive energy further drove the exposure of residues, originally buried within the folded core, to the aqueous environment. Surface hydrophobicity (H₀) analysis (Fig. 1M) showed that the linear slope for the US group was approximately 2.3-fold higher than that of the CK group. The pronounced exposure of the hydrophobic microenvironment not only enhanced the surface hydration potential—the molecular driver of macro-solubility—but also eliminated the critical steric constraints for the targeted liberation of specific antioxidant peptide sequences sequestered within the hydrophobic core during subsequent proteolysis.

3.5. Influence of ultrasonic substrate modification on targeted enzymatic peptide yield

The spatial conformation of macromolecular substrates significantly modulates the thermodynamic progression and mass transfer efficiency during exogenous proteolysis. As illustrated in Fig. 2A, under identical enzymatic kinetics, the peptide yield of the untreated control (CK) group was 44.82% ± 0.35%, whereas the ultrasonically modified substrate (US group) exhibited a marked elevation in peptide yield, reaching 68.32% ± 0.52% (P < 0.05). This substantial enhancement suggests that the acoustic cavitation-induced structural remodeling successfully reduced the steric hindrance of the native protein, thereby facilitating greater accessibility for the Alcalase to target specific cleavage sites.

Fig. 2 — Influence of ultrasonic substrate modification on peptide yield and the multi-dimensional in vitro antioxidant network. (A) Comparison of enzymatic peptide yields between the control and ultrasound-pretreated groups. (B) DPPH radical scavenging activity. (C) ABTS radical scavenging activity. (D) Reducing power at 700 nm. (E) Hydroxyl radical scavenging activity. (F) Superoxide anion radical scavenging activity. (G) Fe²⁺ chelating capacity. CK, peptides prepared from untreated VAP; US, peptides prepared from ultrasound-pretreated VAP. Data are presented as mean ± SD (n = 3). Statistical differences were analyzed between CK and US at the same concentration. *P < 0.05 and **P < 0.01.

3.6. Influence of ultrasound pretreatment on the in vitro antioxidant activities of peptides

3.6.1. Analysis of DPPH/ABTS radical scavenging activity and reducing power

DPPH and ABTS radical scavenging activities, alongside reducing power, serve as established indicators for evaluating the electron- or hydrogen-donating capacities of antioxidants. As shown in Fig. 2B–D, the antioxidant activities of both CK and US peptides exhibited a clear dose-dependent pattern. Within the concentration range of 1.0–5.0 mg/mL, the US peptides showed higher DPPH and ABTS radical scavenging activities and reducing power than the CK peptides at the same concentration. Specifically, the DPPH scavenging rate of the US peptides increased from 37.77 ± 1.01% at 1.0 mg/mL to 95.54 ± 0.75% at 5.0 mg/mL, whereas that of the CK peptides increased from 23.00 ± 0.63% to 70.60 ± 1.79%. Similarly, at 5.0 mg/mL, the ABTS scavenging rate of the US group reached 98.67 ± 0.95%. In addition, the reducing power of the US group was 0.810 ± 0.015 at 700 nm, which was markedly higher than that of the CK group (0.486 ± 0.041). These results indicate that ultrasound pretreatment enhanced both radical-scavenging capacity and electron-donating ability of the resulting VAP-derived peptides. Ultrasound-induced conformational unfolding facilitated the exposure of numerous internal hydrophobic amino acids (e.g., leucine and valine) and sulfur-containing residues. Subsequent proteolysis effectively liberated low-molecular-weight fragments enriched with these active groups, augmenting the concentration of free sulfhydryl and phenolic hydroxyl donors, thereby neutralizing and stabilizing free radicals more efficiently.

3.6.2. Analysis of hydroxyl/superoxide radical scavenging and ferrous ion chelating capacity

Hydroxyl radicals (⋅OH) and superoxide anions (O₂⁻) are the primary reactive oxygen species (ROS) responsible for inducing oxidative damage in biological macromolecules. As illustrated in Fig. 2E and F, the scavenging activities of the US peptides toward ⋅OH and O₂⁻ were significantly higher than those of the CK group. At a concentration of 5.0 mg/mL, the ⋅OH scavenging rate of the US group reached 90.96% ± 1.45%. This improvement is ascribed to the facilitated generation of low-molecular-weight bioactive peptides, whose reduced steric dimensions allow for greater proximity to and more effective quenching of target radicals. Furthermore, transition metal ions such as Fe²⁺ can catalyze the generation of ⋅OH via the Fenton reaction. The results in Fig. 2G demonstrate that the US peptides exhibited a more robust Fe²⁺ chelating capacity, likely due to the increased exposure of amino and carboxyl groups capable of forming stable complexes with metal ions.

3.7. Peptidomic characterization and physicochemical properties analysis of VAP-derived peptides

3.7.1. Total ion chromatogram (TIC) analysis

To characterize the peptide composition of the ultrasound-assisted enzymatic hydrolysate, LC-MS/MS analysis was performed. As shown in Fig. 3A, the total ion chromatogram (TIC) reflects the chromatographic complexity and ion abundance distribution of the VAP-derived peptide mixture. Within the 0–65 min elution window, multiple ion peaks were detected, with relatively abundant signals mainly distributed between 10 and 50 min, indicating that the hydrolysate contained a diverse peptide population. It should be noted that the TIC profile itself does not directly prove ultrasound-induced protein unfolding or substrate destabilization. Rather, it provides peptide compositional information that should be interpreted together with the structural and functional evidence obtained from SEM, DLS, CD, DSC, surface hydrophobicity, and peptide yield analyses. Therefore, the peptide profile observed here may reflect the combined effects of ultrasound-assisted substrate modification, Alcalase hydrolysis specificity, and the intrinsic collagen-rich characteristics of VAP.

3.7.2. Sequence identification, length distribution, and molecular weight characteristics

The molecular weight (MW) distribution and sequence length of peptides are critical determinants of their steric hindrance, transmembrane transport capacity, and bioavailability. High-throughput peptidomic analysis revealed that a total of 7,269 unique peptide sequences were successfully identified and characterized in this study. As illustrated in Fig. 3B, the molecular weights of VAP-derived peptides exhibited a distinct low-range distribution profile, with the vast majority of components concentrated below 1 kDa, substantiating the exceptional efficiency of the synergistic degradation process. The statistical analysis of sequence length (Fig. 3C) further substantiated these micro-structural characteristics. The identified peptides were predominantly short-chain oligopeptides; specifically, sequences containing 2–10 amino acid residues constituted the overwhelming majority, totaling 6,712 sequences (92.3% of the total library). Within this distribution, tetrapeptides and pentapeptides exhibited the highest abundance, reaching frequencies of 2,285 and 1,861, respectively. This short-peptide-dominant profile suggests a pronounced synergistic effect between acoustic cavitation-induced structural loosening and enzymatic proteolysis. Low-molecular-weight oligopeptides are generally considered to possess favorable structural features for potential intestinal transport and biological utilization. However, whether the identified VAP-derived peptides can remain stable during gastrointestinal digestion and be absorbed through PepT1 or other transporter-mediated pathways was not directly verified in this study and therefore requires further validation.

3.7.3. Amino acid composition analysis

Amino acid composition serves as the fundamental structural basis governing the specific spatial conformation and biological activities of peptides. As illustrated in Fig. 3D, the VAP peptides are characterized by an exceptionally high content of hydrophobic amino acids (HAAs), with the top five residues identified as leucine (Leu, 14.65%), proline (Pro, 11.38%), isoleucine (Ile, 10.38%), valine (Val, 7.89%), and phenylalanine (Phe, 6.58%). The high proportion of hydrophobic and aromatic residues may contribute to antioxidant potential by improving interactions with radical species or hydrophobic microenvironments. Nevertheless, cellular uptake and membrane transport were not directly measured in this study; therefore, the relationship between amino acid composition and cellular internalization should be regarded as a possible explanation rather than direct evidence. Furthermore, the abundance of acidic amino acids (Asp and Glu, totaling 9.67%) was found to be slightly higher than that of basic amino acids (Lys, Arg, and His, totaling 7.48%). Notably, the high detection frequencies of Gly (7.83%) and Pro reflect the collagen-rich nature of the velvet antler matrix, from which numerous active fragments with typical collagenous motifs (e.g., Gly-X-Y) were efficiently liberated under the synergistic effects of ultrasound and enzymes.

3.7.4. Characteristics of isoelectric point (pI) and hydrophobicity distribution

The physicochemical properties of peptides are inextricably linked to their primary sequences. As shown in Fig. 3E, the isoelectric points (pI) of the identified peptides exhibited distinct multi-modal distributions in the kernel density plots, with the predominant peak concentrated between pI 5.0 and 6.0 and a secondary peak near pI 4.0. This predominantly acidic pI distribution is consistent with the high proportion of acidic amino acids observed, enabling the peptides to carry a net negative charge under physiological pH conditions, which is conducive to maintaining superior solubility and colloidal stability. Regarding the hydrophobicity distribution (Fig. 3F), the scores followed a quasi-normal distribution ranging from −20 to 20, with the core distribution peak noticeably shifted toward the positive region (0–10). This balanced amphiphilicity is a critical structural prerequisite for bioactive peptides to exert their physiological functions as functional additives.

3.7.5. N-terminal and C-terminal amino acid distribution and structure-activity relationships

The terminal residues of peptides not only manifest the cleavage specificity of proteases but also serve as pivotal determinants governing their metabolic stability and biological activities. The terminal analysis of the 7,269 identified peptides (Fig. 3G–H) exhibited a distinct pattern, with the C-terminus being highly enriched in non-polar aliphatic (55.15%) and aromatic (18.86%) residues. Notably, Leu (25.31%), Ile (16.29%), and Phe (13.32%) were the most prevalent residues at this position. According to structure–activity relationship (SAR) studies, the presence of hydrophobic or aromatic residues at the C-terminus is essential for both ACE-inhibitory and antioxidant activities, as these residues can effectively compete for the active sites of enzymes via hydrophobic interactions or efficiently quench free radicals. Similarly, the N-terminus was predominantly composed of non-polar residues (45.95%), with Thr (12.85%) and Leu (11.37%) being the most abundant. The enrichment of hydrophobic residues at peptide termini may influence peptide–membrane interactions and enzymatic susceptibility. However, gastrointestinal stability, epithelial transport, and bioavailability were not directly measured in this study; therefore, these structure–activity relationships should be regarded as possible explanations requiring further validation.

3.8. Machine learning-based high-throughput virtual screening and sequence characterization

Leveraging the PeptideRanker deep neural network model, this study conducted a comprehensive prediction of the general bioactivity potential of the 7,269 peptide sequences. This broad-spectrum screening was subsequently combined with quantum chemical calculations for further antioxidant-related prioritization. The screening results revealed that the top 50 candidate peptides exhibited predicted bioactivity scores exceeding 0.90 (Fig. 4A), indicating that they were suitable for further computational evaluation and experimental selection. However, PeptideRanker provides a general bioactivity prediction based on sequence-derived features and does not directly confirm antioxidant activity. Therefore, these high-scoring peptides should be regarded as putative candidates rather than experimentally validated antioxidant peptides.

Fig. 4 — High-throughput screening and electronic property characterization of antioxidant lead peptides from VAP based on PeptideRanker and DFT. (A) Distribution of predicted general bioactivity scores for the 7,269 identified VAP sequences calculated via the PeptideRanker deep neural network model (the top 50 lead peptides are highlighted). (B) Optimized geometric configurations of the core lead peptides obtained under the PCM implicit solvation model. Note: Due to the folded conformation of the peptides, the three-dimensional effects are presented in two-dimensional images, which may create the illusion of overlapping structures.

Structural characterization indicated that these high-scoring sequences were primarily composed of 2–5 amino acid residues, which closely matched the molecular weight distribution concentrated below 1 kDa observed in the peptidomic analysis. Sequence profiling further revealed that these potential bioactive peptides were enriched in aromatic residues with conjugated π-systems (Trp and Phe) and bulky hydrophobic side chains (Leu and Ile). This profile may be associated with the combined effects of ultrasound-assisted substrate modification, Alcalase hydrolysis specificity, and the intrinsic amino acid composition of VAP. Nevertheless, the PeptideRanker output may be influenced by training dataset bias and limited generalizability to specific functional subclasses, such as antioxidant peptides. Therefore, the PeptideRanker results were used only as an initial prioritization step, and these sequences were further evaluated by DFT analysis, molecular docking, molecular dynamics simulation, chemical synthesis, and in vitro assays rather than being regarded as experimentally confirmed active peptides at this stage.

3.9. DFT-based characterization of electronic properties and antioxidant-related reactivity of lead peptides

The redox activity of peptides is governed not only by their amino acid composition but, more fundamentally, by their three-dimensional electronic density topological features within a physiological solvent milieu. In the present study, DFT calculations were employed to perform full geometry optimizations (Fig. 4B) and Frontier Molecular Orbital (FMO) analyses on the 50 pre-screened lead peptides. These calculations were conducted using the PCM implicit solvation model with water as the dielectric medium. This electronic-scale characterization allows for a deeper understanding of the peptides' reactivity and their potential to interact with free radicals through electron or hydrogen atom transfer mechanisms.

3.9.1. Screening of lead peptides and electronic level advantages

To evaluate the chemical reactivity and stability of the lead peptides at the electronic level, the maximum Highest Occupied Molecular Orbital energy (E_HOMO) and the minimum energy gap (Δ_E) were employed as the primary criteria (Fig. S1,S2, Fig. 5), leading to the identification of sequences with exceptional electron-donating propensity, notably WFL, WP, WIF, IFW, WGPF and GFPWP. These peptides exhibited pronounced electronic advantages, characterized by relatively high E_HOMO values (e.g., −5.6765 eV for WFL and −5.6823 eV for WIF) and narrow Δ_E ranges (5.0–5.1 eV), indicating lower IP. In plain terms, a higher E_HOMO value means that the peptide is more willing to donate electrons, whereas a smaller energy gap means that electron transfer can occur more easily. Therefore, these peptides may trigger SET with a relatively low energy barrier, thereby facilitating free radical quenching.

Fig. 5 — Frontier molecular orbital (FMO) energy level profiles and electron density topological characteristics of velvet antler lead peptides. (A) Donut chart illustrating the energy gap (ΔE) distribution of 50 lead peptides. (B) Topological visualization of Highest Occupied Molecular Orbital (HOMO) and Lowest Unoccupied Molecular Orbital (LUMO) electron density for 10 core active peptides, along with their corresponding ΔE values.

3.9.2. DFT-based interpretation of structure–orbital relationships in lead peptides

By integrating the topological characteristics of the FMO electron density distribution, this study identified two distinct structural mechanisms: Aliphatic alkyl induction and hydrophobic shielding networks (WFL, WIF, and IFW): In these sequences, the HOMO electron density was primarily localized on the indole ring of Trp or the phenyl ring of Phe. The bulky non-polar alkyl side chains of Ile and Leu exerted a critical synergistic effect. On one hand, the strong electron-donating inductive effect (+I effect) of the alkyl groups enhanced the electron density of the aromatic rings, significantly elevating the ground-state E_HOMO levels. On the other hand, the bulky hydrophobic side chains may create a relatively low-polarity microenvironment around the predicted electron-donating center. This DFT-based interpretation suggests that such local hydrophobic surroundings could contribute to the stabilization of radical intermediates and may partly explain the favorable radical-quenching potential of these peptides. Possible Pro-related conformational restriction in WGPF, WP, and GFPWP: As representative collagenous motifs of velvet antler, the core feature of WP and GFPWP is the incorporation of rigid Pro residues. The unique pyrrolidine ring restricts the rotational degrees of freedom of the peptide backbone, achieving “conformational locking.” In this stable configuration, the reactive indole ring of Trp is precisely anchored at spatially accessible sites. This geometric stability minimizes intramolecular thermal vibrations during SET, thereby reducing the reorganization energy of the system and providing an optimal physical foundation for rapid electronic transitions.

3.10. Molecular docking and molecular dynamics-based prediction of peptide–Keap1 interactions

Within the cellular redox regulatory network, the antioxidant effects of peptides may be related not only to intrinsic electron-donating capacity, but also to their potential interaction with regulatory proteins such as Keap1. In this study, molecular docking was used as a predictive tool to explore the possible interaction modes between selected peptides and the Keap1 Kelch domain.

3.10.1. Evaluation of binding affinity and thermodynamic stability

The core sequences previously prioritized via peptidomics and quantum chemical screening (e.g., WFL, WGPF, WIF, IFW, and GFPWP) demonstrated superior binding performance, with binding energies consistently lower than −7.0 kcal/mol. Notably, WGPF exhibited the most favorable predicted binding affinity among all candidates, with a binding energy of −10.29 kcal/mol. In molecular docking, a more negative binding energy generally indicates a more stable predicted peptide–protein interaction. Thus, this result suggests that WGPF may bind more stably within the Keap1 Kelch domain than the other candidate peptides, providing a possible structural basis for its Keap1/Nrf2-related cellular antioxidant effect.

3.10.2. Visualization of predicted binding Hotspots and interaction patterns

Based on residue contribution analysis (Figs. 6B and S3–S5), the docking results provided a predicted interaction map between the selected peptides and the Keap1 Kelch domain. The representative peptides were predicted to interact with several residues within the Keap1 cavity, including Val285, Val142, Val283, Val48, Val287, Ile238, Ala45, and Ala189. These interactions were mainly associated with hydrophobic contacts and hydrogen bonds, suggesting that both non-polar side chains and peptide backbone groups may contribute to the predicted peptide–Keap1 binding stability. For Pro-containing peptides such as WGPF and GFPWP, the rigid pyrrolidine ring of Pro may restrict backbone flexibility and help maintain a relatively favorable binding orientation within the receptor cavity. However, these interaction patterns should be interpreted as docking-based predictions rather than experimentally verified binding mechanisms.

Fig. 6 — Molecular docking simulations and residue interaction network analysis of the lead peptides with the Keap1 protein. (A) Circular heatmap illustrating the binding free energies of 50 lead peptides with the Keap1 Kelch domain (PDB ID: 2FLU). (B) Three-dimensional (3D) binding poses and two-dimensional (2D) residue interaction diagrams of representative core peptides with Keap1.

3.10.3. Dynamic stability analysis of Keap1–peptide complexes by molecular dynamics simulation

To further evaluate the predicted dynamic stability of the Keap1–peptide complexes beyond static docking poses, 100 ns molecular dynamics simulations were performed for Keap1 complexed with WFL, WIF, IFW, WGPF, and GFPWP. The stability of each complex was assessed using RMSD, radius of gyration (Rg), solvent-accessible surface area (SASA), residue fluctuation (RMSF), hydrogen bond number, and estimated binding free energy. As shown in Fig. S7, all Keap1–peptide complexes underwent different degrees of conformational adjustment during the early simulation stage and subsequently reached relatively stable states. The RMSD profiles suggested that the complexes did not undergo continuous structural drift during the 100 ns simulations. The Rg and SASA curves remained generally stable, indicating that no obvious structural collapse, expansion, or abnormal change in solvent exposure occurred after peptide binding. RMSF analysis showed that most Keap1 residues maintained relatively low fluctuation amplitudes, suggesting that peptide binding did not markedly perturb the overall structural stability of Keap1. In addition, the hydrogen bond profiles and negative estimated binding free energies further supported the predicted stability of the Keap1–peptide complexes. Among the five peptides, WGPF and IFW showed relatively stable dynamic profiles, consistent with their favorable docking results. Nevertheless, these MD results should be interpreted as computational evidence supporting the predicted stability of peptide–Keap1 interactions rather than as direct experimental proof of binding. Direct biophysical assays, such as SPR, ITC, cellular thermal shift assays, or competitive Keap1–Nrf2 interaction assays, are still required to confirm the actual binding affinity and inhibitory capacity of these peptides.

3.11. Experimental validation of antioxidant efficacy and analysis of molecular mechanisms

3.11.1. Evaluation of in vitro chemical antioxidant activity

Based on dual criteria integrating frontier molecular orbital (FMO) energy levels and receptor binding affinities, while ensuring high sequence novelty, five core peptides (WFL, WIF, IFW, WGPF, and GFPWP) were prioritized for the experimental validation of their antioxidant effects. To ensure the structural authenticity of the identified sequences, the presence of the candidate peptides within the VAP hydrolysate was rigorously verified through high-resolution MS/MS fragmentation analysis (Fig. S6). To evaluate the direct radical-neutralizing capacity of the identified lead peptides, their DPPH and ABTS scavenging activities were determined across a concentration range of 0–250 μM (Fig. 7A and B). The experimental results revealed that all five peptides exhibited significant antioxidant activities in a dose-dependent manner. Among them, WFL—which possessed the narrowest energy gap (ΔE)—demonstrated the most superior in vitro chemical antioxidant potential; at a concentration of 250 μM, its scavenging rates for DPPH and ABTS reached 98.06% ± 0.33% and 99.15% ± 0.25%, respectively. These findings confirm the high predictive accuracy of the combined machine learning and quantum chemical screening approach.

3.11.2. Cell viability and biosafety assessment

The CCK-8 assay was employed to assess the impact of the lead peptides across a concentration gradient (25–400 μM) on cell viability (Fig. 7C). Our findings indicated that cell viability for all groups remained within the range of 98.59% ± 1.48% to 104.87% ± 1.05% across the entire concentration range tested. Statistical analysis revealed no significant deviations between the peptide-treated groups and the blank control (P > 0.05), thereby substantiating the excellent biosafety profiles of these five novel peptides. Notably, a slight pro-proliferative trend (approximately 104.5% viability) was observed in all groups at 100 μM; consequently, a conservative concentration of 100 μM was selected for subsequent cytoprotection experiments.

3.11.3. Intracellular antioxidant indicators and mitigation of lipid damage

To further evaluate the bioactivity of the five lead peptides within a complex intracellular microenvironment, the activities of key endogenous antioxidant enzymes (SOD, CAT, and GSH-Px) and the levels of the lipid peroxidation product (MDA) were quantified (Fig. 7D, E, F, and I). The intracellular antioxidant enzyme system constitutes the primary line of defense against oxidative stress. As depicted in Fig. 7D–F, H2O2-induced oxidative stress severely compromised the cellular antioxidant equilibrium, resulting in a marked decline in SOD, CAT, and GSH-Px activities in the Model group compared to the Blank group, falling to 31.81 ± 4.04 U/mg, 8.32 ± 0.49 U/mg, and 51.63 ± 0.61 U/mg, respectively (P < 0.05). Following treatment with 100 μM of the five lead peptides, a significant recovery in enzymatic activities was observed across all intervention groups. Notably, WGPF exhibited the most potent protective efficacy, restoring SOD activity to 77.13 ± 1.51 U/mg (a recovery rate of 88.2%); simultaneously, its CAT (21.67 ± 0.49 U/mg) and GSH-Px (110.15 ± 2.24 U/mg) activities reached peak levels across all tested groups. The GFPWP, WFL, IFW, and WIF groups also demonstrated substantial restorative effects, with enzymatic activities exhibiting a stepwise increase relative to the Model group. These findings suggest that all five lead peptides significantly enhance the cellular capacity to scavenge superoxide anions and hydrogen peroxide by bolstering endogenous antioxidant enzyme activities.

The concentration of malondialdehyde (MDA), a prominent lipid peroxidation product, serves as a critical biomarker for assessing the degree of oxidative damage to cellular membranes. As illustrated in Fig. 7I, H2O2 exposure triggered a dramatic surge in MDA levels in the Model group, increasing from 2.15 ± 0.18 to 12.45 ± 0.47 nmol/mg, indicating a severe loss of membrane integrity. Following peptide intervention, the lipid peroxidation process was effectively mitigated. WGPF demonstrated the most superior membrane-protective capacity, markedly reducing MDA levels to 3.74 ± 0.09 nmol/mg. MDA concentrations in the GFPWP, WFL, IFW, and WIF groups were similarly reduced to 4.53 ± 0.09, 5.79 ± 0.06, 6.57 ± 0.17, and 7.13 ± 0.09 nmol/mg, respectively. These results demonstrate that the five peptides effectively maintain cellular membrane stability by suppressing lipid peroxidation. Notably, although WFL performed best in cell-free chemical radical scavenging assays, WGPF exhibited superior overall efficacy in the cytoprotection experiments. This observation is consistent with the molecular docking results showing the favorable predicted binding affinity of WGPF toward Keap1, suggesting that the cellular protection of WGPF may not be explained solely by direct chemical radical scavenging, but may also be associated with Keap1/Nrf2-related endogenous antioxidant regulation.

3.11.4. Transcriptional regulation of the Keap1/Nrf2 signaling pathway

The relative mRNA expression levels of canonical components within the Keap1/Nrf2 signaling pathway were quantified via qPCR (Fig. 7G and H). Compared with the Blank group, the Model group exhibited a marked down-regulation in the mRNA expression of Nrf2 (Fig. 7H) and heme oxygenase-1 (HO-1) (Fig. 7G), which declined to 0.42 ± 0.03 and 0.35 ± 0.04, respectively (P < 0.05). Intervention with all five lead peptides significantly up-regulated the transcriptional levels of these target genes. Notably, the WGPF-treated group demonstrated the most robust restorative effect, with Nrf2 mRNA levels recovering to 0.95 ± 0.06 (approximating basal levels) and HO-1 expression reaching 0.88 ± 0.08. These findings suggest that the five lead peptides may enhance Keap1/Nrf2-associated antioxidant response element (ARE) through transcriptional modulation, thereby bolstering cellular resistance to oxidative stress.

3.11.5. Regulatory mechanism of key protein expression in the Keap1/Nrf2 pathway

To further elucidate the underlying mechanism at the protein level, the expression of key signaling components was quantitatively analyzed using Western blotting (Fig. 7J–M). The results demonstrated that H₂O₂ exposure led to an aberrant elevation in Keap1 expression (relative level of 1.68 ± 0.12). Following peptide intervention, Keap1 levels were significantly down-regulated, with WGPF reducing it to 1.12 ± 0.04 (Fig. 7L), corroborating the high binding affinity (−10.29 kcal/mol) for the Keap1 pocket predicted by molecular docking. Concomitant with the suppression of Keap1 expression, the protein levels of Nrf2 (Fig. 7M) and HO-1 (Fig. 7K) increased significantly. Specifically, WGPF treatment restored Nrf2 and HO-1 levels to 0.60 ± 0.05 and 0.52 ± 0.04, respectively. Statistical analysis revealed that the cytoprotective efficacy of WGPF was significantly superior to that of WFL, despite the latter possessing the highest chemical activity. These results, together with molecular docking analysis, suggest that the five lead peptides may enhance Keap1/Nrf2-associated antioxidant regulation. In addition, the newly added NF-κB-related protein results further indicate that peptide treatment may attenuate ROS-induced inflammatory signaling activation, thereby providing broader mechanistic support for their cellular antioxidant effects.

3.11.6. Effects of lead peptides on NF-κB signaling-related protein expression

To further investigate whether the antioxidant effects of the five lead peptides involved additional redox-sensitive signaling pathways, NF-κB pathway-related proteins were analyzed by Western blotting. As shown in Fig. 7J, H₂O₂ stimulation markedly increased the phosphorylation levels of NF-κB p65 and IκBα in RAW264.7 cells. Quantitative analysis showed that the p-p65/p65 ratio was significantly elevated in the Model group compared with the Blank group (Fig. 7N), indicating activation of NF-κB signaling under oxidative stress. Similarly, the p-IκBα/IκBα ratio was also markedly increased after H₂O₂ exposure (Fig. 7O), suggesting enhanced IκBα phosphorylation, which is associated with NF-κB pathway activation. Treatment with WFL, WIF, IFW, WGPF, and GFPWP significantly reduced the p-p65/p65 and p-IκBα/IκBα ratios compared with the Model group. These results indicate that the five lead peptides attenuated H₂O₂-induced NF-κB activation in RAW264.7 cells. Together with the Keap1/Nrf2-related results, these findings suggest that the antioxidant protection of VAP-derived lead peptides may involve coordinated regulation of Keap1/Nrf2-associated antioxidant defense and NF-κB-related oxidative inflammatory signaling.

3.12. In silico ADME prediction of gastrointestinal behavior and metabolic persistence of lead peptides

To further address the potential gastrointestinal behavior and translational relevance of the five lead peptides, key ADME-related parameters associated with aqueous stability, intestinal absorption, membrane permeability, and metabolic persistence were predicted using ADMETlab 3.0. Although these computational parameters cannot directly determine gastrointestinal digestion stability, they provide useful preliminary evidence for evaluating whether the peptides possess favorable physicochemical characteristics for maintaining solubility, interacting with the intestinal environment, and supporting subsequent absorption.

Overall, WFL, WIF, IFW, WGPF, and GFPWP exhibited moderate predicted aqueous solubility, with LogS values ranging from −3.329 to −2.532 (Table S1). Their LogD7.4 values ranged from 1.011 to 2.052, indicating a balanced hydrophilic–lipophilic profile under physiological pH conditions. These results suggest that the five short peptides are unlikely to be extremely insoluble or excessively hydrophobic in aqueous biological environments, which may be beneficial for maintaining their dissolved state in gastrointestinal fluids and for subsequent interaction with intestinal epithelial interfaces.

Regarding intestinal absorption-related properties, the predicted Caco-2 permeability values ranged from −6.163 to −5.630 log cm/s, indicating that the five peptides may possess a certain degree of epithelial permeability, although passive transcellular diffusion is expected to be limited. The HIA values differed among the five peptides, suggesting sequence-dependent variation in intestinal absorption potential. Considering their short-chain peptide structures and acceptable physicochemical profiles, these results support the possibility that intestinal uptake may not rely solely on passive diffusion, and that transporter-assisted absorption, including possible PepT1-mediated transport, may contribute to the absorption of some peptide sequences. However, this mechanism was not directly demonstrated by the current prediction and requires further validation using Caco-2 transport assays and PepT1 inhibition experiments.

In terms of metabolic persistence, the predicted plasma half-lives of the five peptides ranged from 0.582 to 0.986 h, suggesting that these short peptides may retain a certain systemic residence time after absorption, while also undergoing relatively rapid turnover, which is consistent with the metabolic characteristics of short bioactive peptides. Liver microsomal clearance values were generally low for most peptides, indicating that the peptide group may possess a certain degree of predicted hepatic microsomal stability, although sequence-dependent differences were observed. Taken together, these ADME prediction results suggest that WFL, WIF, IFW, WGPF, and GFPWP possess favorable preliminary physicochemical and metabolic features related to gastrointestinal behavior, intestinal absorption potential, and metabolic stability.

4. Discussion

The liberation of specific bioactive peptides from structurally dense, macroscopic biological matrices often encounters limited efficiency when relying solely on direct exogenous proteolysis. This limitation arises because the substrate, in its native, tightly folded state, possesses extraordinary conformational stability. Exogenous enzyme molecules, driven only by stochastic thermal motion, are generally incapable of independently surmounting the substantial thermodynamic energy barrier required for the complete unfolding of the substrate. Consequently, critical proteolytic cleavage sites sequestered within the hydrophobic core remain shielded by formidable steric hindrance [25], [26]. The central premise of this study lies in the deployment of a high-intensity, low-frequency acoustic cavitation field as a physical “unfolding” strategy prior to enzymatic hydrolysis, which profoundly modulates the mass transfer kinetics at the solid–liquid interface. Experimental data indicated that under optimal ultrasonic intervention (450 W), the particle size of substrate aggregates markedly shifted to a centered distribution at 175.2 nm; this physical fragmentation provided an exponentially increased effective surface area for subsequent enzymatic attack. This mechanism closely aligns with the findings of NIU et al. [27] regarding ultrasonically modified soy protein, who noted that high-intensity ultrasound not only significantly reduces protein particle size through mechanical shear forces but also triggers a conformational transition from a compact to an extended state; for instance, the β-sheet content of U-7SN was augmented by 20.8%. This conformational deconstruction effectively disrupted the internal hydrophobic barriers of native globular proteins, thereby liberating a greater abundance of proteolytic cleavage sites. Furthermore, ultrasound-induced moderate proteolysis—as evidenced by the increase in low-molecular-weight peptides in SDS-PAGE—further attenuated system viscosity and potentiated surface charge (indicated by an elevated ζ −potential), which macroscopically manifested as a significant enhancement in proteolytic susceptibility. In summary, the 450 W acoustic cavitation field, through the dual mechanisms of physical comminution and structural unfolding, established an optimal substrate topography for subsequent high-efficiency enzymatic reactions.

A more fundamental mechanism lies in the remodeling of the protein's intramolecular cohesive energy by the acoustic cavitation field. Synergistic analysis via DSC and CDrevealed that the transient shear stress and micro-jets generated by acoustic cavitation effectively disrupted the intramolecular hydrogen bonding network that stabilizes the α-helical conformation of VAP. This observation is highly congruent with the findings of MENG et al. [28]regarding whey protein isolate (WPI), which demonstrated a similar structural reorganization characterized by the transformation of α-helices and β-sheets into random coils, accompanied by a marked increase in surface hydrophobicity. Specifically, the reduced particle size and enhanced surface hydrophobicity observed in WPI are essentially macroscopic manifestations of the attenuated intramolecular cohesive constraints. In the VAP system, this energy remodeling suggests that the mechanical energy provided by ultrasound precisely overcame the non-covalent interactions that maintain the native compact conformation, thereby inducing partial unfolding of the molecule. This conformational state not only exposed previously buried hydrophobic groups and active sites but also conferred enhanced interfacial affinity and conformational flexibility of the VAP substrate. Consequently, by drawing parallels with the established modification patterns of WPI, this study confirms that the acoustic cavitation field is not a stochastic physical disruption; rather, it remodels the intramolecular potential energy landscape, facilitating a conformational transition of VAP from a low-energy native state to a more accessible state with heightened processing activity and antioxidant potential.

Concurrently, a downward contraction in both the denaturation temperature (T_p) and enthalpy of denaturation (ΔH) was observed. This indicates that the mechanical shear forces effectively “compensated” for a portion of the activation energy required for substrate unfolding, transitioning the system into a sensitized, destabilized state and facilitating the exogenous proteases in surmounting the initial thermodynamic barrier.

These findings are strongly corroborated by recent research on bovine bone collagen by QI et al. [29], whose energetic calculations explicitly demonstrated that ultrasonic pretreatment attenuated the ΔH by 27.69% and reduced the activation energy (E_a) of the proteolytic process by 25.73%. This downward shift in thermodynamic parameters essentially reflects the deconstruction of the compact protein architecture by the physical field, mitigating the energy hurdles that otherwise impede enzymatic hydrolysis. This “structural pre-activation” mechanism not only augmented the substrate-enzyme affinity but also macroscopically drove the efficient liberation of low-molecular-weight bioactive peptides, thereby significantly enhancing the biological activity of the products. This alteration of the energy landscape, coupled with the elevation of surface hydrophobicity (H₀), promoted the exposure of sequestered internal hydrophobic residues to the aqueous phase. The enhancement of H₀ in VAP observed in the present study aligns closely with the trends reported by Malik et al. [30], who confirmed that ultrasound-induced molecular unfolding forces the exposure of hydrophobic micro-domains previously buried within the core—regions that are typically enriched with aromatic and hydrophobic residues, which serve as primary cleavage sites. Combined with the observed decrease in Tp, it can be inferred that the rise in H₀ signifies both the dissociation of intramolecular cohesive energy and the elimination of steric hindrance at the enzyme-binding interface. While significantly reducing the apparent Michaelis constant (K_m), this process cleared the spatial obstacles for Alcalase-mediated specific cleavage of sequestered antioxidant motifs, ultimately propelling the peptide yield from 44.82% in the untreated control to 68.32%.

Previous studies on peptide antioxidant activity have often focused on amino acid composition, whereas the electronic basis of radical scavenging remains less clearly described. In this study, DFT analysis was used to provide an electronic-level explanation for the antioxidant potential of the selected VAP-derived peptides. The lead peptides, particularly Trp- and Phe-containing sequences such as WFL and WIF, showed relatively high HOMO energies and narrow energy gaps. These features suggest favorable electron-donating capacity and lower energy barriers for SET. The DFT results further suggest that hydrophobic residues may contribute to this electronic advantage. In WFL, WIF, and IFW, adjacent Leu or Ile residues may enhance the local electron-rich environment through an inductive effect. This interpretation is consistent with the DFT-based observation of Zheng et al. [31], who reported that electron-donating groups can reduce ionization potential and facilitate electron transfer. In the present peptide system, these alkyl side chains may increase electron density around adjacent aromatic residues, thereby contributing to the higher HOMO levels. In addition, hydrophobic side chains may help stabilize aromatic electron-donating centers and radical intermediates by creating a relatively low-polarity local microenvironment. Pro-containing sequences such as WGPF and GFPWP may possess another structural advantage. The pyrrolidine ring of Pro restricts backbone rotation and may help maintain a relatively constrained peptide conformation. Such conformational restriction could reduce unnecessary structural fluctuation and keep the Trp-containing reactive region spatially accessible. This may support efficient radical interaction and contribute to the antioxidant activity of Pro-containing peptides. Although DFT and molecular docking provided useful computational clues for interpreting the antioxidant potential and Keap1-binding tendency of the selected peptides, these results should be regarded as supportive predictions rather than direct biological evidence. The HOMO/LUMO-derived descriptors can explain the theoretical electron-donating tendency of peptides, but the actual radical-quenching pathway requires further confirmation by reaction kinetic analysis or specific SET/HAT-related assays. Similarly, docking-derived binding energy reflects a predicted peptide–Keap1 interaction and cannot replace direct binding validation. Future studies should therefore include SPR, ITC, cellular thermal shift assay, or competitive Keap1–Nrf2 binding assays to verify the actual binding affinity and inhibitory capacity of these peptides.

Within the cellular antioxidant defense system, the Keap1/Nrf2 signaling axis represents a key regulatory pathway for interpreting the potential mechanism of VAP-derived lead peptides. In this study, molecular docking was used as a predictive structural tool to explore the possible interaction between the core peptide WGPF and the Keap1 Kelch domain. The docking results suggested that WGPF could be accommodated within the Keap1 binding cavity through a combination of hydrogen bonding and hydrophobic interactions. In particular, residues such as Arg380, Arg415, and Asn382, which have been reported to participate in Keap1–Nrf2 recognition, were located near the predicted peptide-binding region. Meanwhile, the aromatic residues Trp and Phe in WGPF were predicted to interact with the hydrophobic pocket surrounding residues such as Val285, thereby contributing to favorable spatial complementarity. This binding pattern suggests that VAP-derived lead peptides may occupy or partially shield the Keap1 binding cavity, which could influence Keap1/Nrf2-associated antioxidant regulation. This interpretation is consistent with the study of Yin et al. [32], which reported that short peptides such as SPW and QKW may exert antioxidant effects by competitively occupying key Keap1 sites and interfering with Keap1–Nrf2 protein–protein interactions. In addition, the present docking results also indicated that Pro-containing sequences, such as WGPF and GFPWP, may possess conformational advantages. The rigid pyrrolidine ring of Pro can restrict backbone flexibility, potentially reducing conformational fluctuation and helping the peptide maintain a relatively stable orientation within the binding pocket. This “conformational locking” effect may partly explain why WGPF showed favorable predicted binding with Keap1. More importantly, the docking results were supported by the cellular antioxidant experiments. Although WFL showed strong radical-scavenging activity in cell-free DPPH and ABTS assays, WGPF exhibited more pronounced protection in H₂O₂-challenged RAW264.7 cells, as reflected by the restoration of SOD, CAT, and GSH-Px activities, reduction of MDA accumulation, downregulation of Keap1, and upregulation of Nrf2 and HO-1. These findings suggest that the cellular antioxidant effect of WGPF may not be explained solely by direct radical quenching, but may also be associated with Keap1/Nrf2-related endogenous antioxidant regulation. Similar integration of molecular docking with experimental validation has been reported by Chu et al. [33], who used docking analysis to support the Keap1/Nrf2-mediated antioxidant mechanism of ultrasound-assisted enzymatic loach protein hydrolysates. In addition, Zhang et al. [34] linked docking-predicted hydrogen bonding with experimentally observed digestive behavior of pork during heating, further supporting the value of interpreting docking results together with experimental evidence. Nevertheless, molecular docking remains predictive in nature, and direct binding assays, such as SPR, ITC, or competitive Keap1–Nrf2 interaction assays, are still required in future studies to confirm the actual peptide–Keap1 interaction and competitive inhibitory capacity.

The cytoprotective efficacy and signaling responses of VAP-derived lead peptides in living cells further support their biological regulatory potential, which transcends mere chemical radical scavenging. Traditional paradigms have largely ascribed the antioxidant mechanisms of natural peptides to direct, unidirectional radical quenching. However, in the H₂O₂ −induced RAW264.7 cell injury model, peptide intervention not only effectively mitigated the dramatic elevation of the lipid peroxidation product malondialdehyde (MDA) but also significantly reconstituted the activities of endogenous antioxidant enzymes, including SOD, CAT, and GSH-Px. This enzymatic restorative capacity parallels recent findings on representative bio-regulatory peptides derived from sea cucumber and bovine collagen [[35], [36], [37]]. Moreover, the transcriptional and protein-level results suggest that VAP-derived lead peptides may attenuate Keap1-associated repression of Nrf2. This dual blockade, impacting both the stability and the absolute expression levels of the repressor protein, represents a more profound level of therapeutic intervention compared to conventional peptides that merely facilitate Nrf2 dissociation. Although the newly added ADME prediction provides preliminary information on the potential gastrointestinal behavior and metabolic persistence of the five lead peptides, these results remain computational and should not be interpreted as direct evidence of gastrointestinal stability, intestinal absorption, systemic exposure, or in vivo efficacy. The predicted LogS, LogD7.4, Caco-2 permeability, HIA, plasma half-life, and liver microsomal clearance values suggest that the peptides may differ in solubility, amphiphilicity, absorption potential, and metabolic persistence; however, these properties require experimental verification. Future studies should combine simulated gastrointestinal digestion, Caco-2 monolayer transport, PepT1 inhibition assays, LC-MS/MS-based metabolic tracing, pharmacokinetic analysis, and animal models of oxidative injury to determine the biological relevance and in vivo efficacy of these peptides. In addition, the ultrasound pretreatment parameters were optimized under laboratory-scale conditions using a probe-type ultrasonic system. Therefore, the optimized conditions of 450 W, 30 min, and 30 mL/g should be regarded as a laboratory-scale basis rather than a directly transferable industrial protocol. During scale-up, factors such as acoustic field distribution, cavitation intensity, specific energy input, reactor geometry, heat transfer, processing throughput, and batch-to-batch uniformity may substantially affect process performance. Further pilot-scale validation using continuous-flow or multi-transducer ultrasonic systems is required to evaluate the technical feasibility, energy efficiency, reproducibility, peptide yield, and antioxidant activity of ultrasound-assisted VAP pretreatment under larger-scale processing conditions.

5. Conclusion

In summary, ultrasound-assisted pretreatment improved VAP solubility and enhanced peptide release during enzymatic hydrolysis. Integrated peptidomics, machine learning, DFT, molecular docking, and molecular dynamics simulations enabled the prioritization of five representative antioxidant peptide candidates: WFL, WIF, IFW, WGPF, and GFPWP. In vitro validation showed that these peptides exerted antioxidant protective effects in H₂O₂-challenged RAW264.7 cells, accompanied by regulation of Keap1/Nrf2 and NF-κB-associated responses. Overall, this study provides a laboratory-scale strategy for discovering antioxidant peptides from structurally dense biological resources.

6. Data statement

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

CRediT authorship contribution statement

Xuanying Xin: Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization. Haowen Wang: Methodology, Formal analysis, Data curation, Conceptualization. Sungkwon Park: Writing – original draft, Methodology, Data curation, Conceptualization. Changguo Yan: Resources, Methodology, Formal analysis. Qiang Li: Methodology, Formal analysis. Rui Du: Writing – review & editing, Visualization, Supervision, Resources, Project administration, Methodology, Conceptualization. Xiangzi Li: Writing – review & editing, Validation, Project administration, Methodology.

Funding

This study was supported by the Natural Science Foundation of Jilin Province (Grant No. YDZJ202101ZYTS105), the National Natural Science Foundation of China (grant number: 32060767) and the Research Fund of Engineering Research Center of North-East Cold Region Beef Cattle Science & Technology Innovation, Ministry of Education, and the “111” Project (D20034), China.

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.

Footnotes

^☆

This article is part of a special issue entitled: ‘Sono-processing for enhanced food quality’ published in Ultrasonics Sonochemistry.

^{Appendix A}

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ultsonch.2026.107920.

Contributor Information

Rui Du, Email: duruijlau@163.com.

Xiangzi Li, Email: lxz@ybu.edu.cn.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Supplementary Data 1

Includes additional data, figures, and tables supporting the experimental and computational analyses of VAP-derived antioxidant peptides.

mmc1.docx^{(9.3MB, docx)}

References

1.Chen S., Li Y., Yang Y., Zhao S., Shi H., Yang C., Zhang A. Comparison of the composition, immunological activity and anti-fatigue effects of different parts in sika deer antler. Front. Pharmacol. 2024;15 doi: 10.3389/fphar.2024.1468237. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Liu L., Jiao Y., Yang M., Long G., Hu W. Network pharmacology, molecular docking and molecular dynamics to explore the potential immunomodulatory mechanisms of deer antler. Int. J. Mol. Sci. 2023;24 doi: 10.3390/ijms241210370. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Jiang W., Chang K., Sainan Z.C., Zhang T. Effects of heat treatment on structural and functional properties of velvet antler polypeptides. J. Food Process. Preserv. 2021 doi: 10.1111/jfpp.15490. [DOI] [Google Scholar]
4.Herrera-Lavados C., Tabilo-Munizaga G., Carvajal-Mena N., Jara-Quijada E., Martínez-Oyanedel J., Pérez-Won M. Obtaining bioactive peptides by enhancing enzymatic hydrolysis of salmon by-product proteins through pulsed electric fields (PEF) Food Res. Int. 2025;208 doi: 10.1016/j.foodres.2025.116103. [DOI] [PubMed] [Google Scholar]
5.Morton J., Khavari M., Priyadarshi A., Kaur A., Grobert N., Mi J., Tzanakis I. Dual frequency ultrasonic cavitation in various liquids: high-speed imaging and acoustic pressure measurements. Phys. Fluids. 2023;10(1063/5) [Google Scholar]
6.Flores-Jiménez N.T., Ulloa J.A., Urías-Silvas J.E., Ramírez-Ramírez J.C., Bautista-Rosales P.U., Gutiérrez-Leyva R. Influence of high-intensity ultrasound on physicochemical and functional properties of a guamuchil Pithecellobium dulce (Roxb.) seed protein isolate. Ultrason. Sonochem. 2022;84 doi: 10.1016/j.ultsonch.2022.105976. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Pacheco A., Pacheco F., Cunha J.S., Nalon G.A., Gusmão J.V.F., Santos F., De Castro Leite Júnior B.R. Physicochemical properties and in vitro antioxidant activity characterization of protein hydrolysates obtained from pumpkin seeds using conventional and ultrasound-assisted enzymatic hydrolysis. Foods. 2025;14 doi: 10.3390/foods14050782. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Xin X.Y., Ruan C.H., Liu Y.H., Jin H.N., Park S.K., Hur S.J., Choi S.H. Identification of novel antioxidant and anti-inflammatory peptides from bovine hemoglobin by computer simulation of enzymolysis, molecular docking and molecular dynamics. Curr. Res. Food Sci. 2024;9 doi: 10.1016/j.crfs.2024.100931. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Lin H., Zhao J., Xie Y., Tang J., Wang Q., Zhao J., Liu P. Identification and molecular mechanisms of novel antioxidant peptides from fermented broad bean paste: a combined in silico and in vitro study. Food Chem. 2024;450 doi: 10.1016/j.foodchem.2024.139297. [DOI] [PubMed] [Google Scholar]
10.Liu W., Liu R., Qin Q., Wang H., Zhang X., Meng G. Molecular docking and molecular dynamics simulation of wheat gluten-derived antioxidant peptides acting through the Keap1-Nrf2 pathway. J. Sci. Food Agric. 2024 doi: 10.1002/jsfa.13647. [DOI] [PubMed] [Google Scholar]
11.Aiello G., Cropotova J., Kvangarsnes K., d’Adduzio L., Fanzaga M., Bollati C., Lammi C. Ultrasonicated Atlantic herring side streams as source of multifunctional bioactive and bioavailable peptides. npj Sci. Food. 2025;9 doi: 10.1038/s41538-025-00388-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.López-Mártir K.U., Armando Ulloa J., Urías-Silvas J.E., Rosas-Ulloa P., Ramírez-Ramírez J.C., Resendiz-Vazquez J.A. Modification of the physicochemical, functional, biochemical and structural properties of a soursop seed (Annona muricata L.) protein isolate treated with high-intensity ultrasound. Ultrason. Sonochem. 2024;105 doi: 10.1016/j.ultsonch.2024.106870. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Nemergut M., Štulajterová M., Škrabana R., Hovan A., Ambro L.U., Strunga A., Sedlák E. Concentration-dependent dimerization of staphylokinase variants with engineered surface charges. Prot. Sci. 2026;35:3. doi: 10.1002/pro.70494. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Fan S.-H., Wang W.-Q., Zhou Y.-W., Gao X.-J., Zhang Q., Zhang M.-H. Research on the interaction mechanism and structural changes in human serum albumin with hispidin using spectroscopy and molecular docking. Molecules. 2024;29 doi: 10.3390/molecules29030655. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Ahmed E.H., Endalew S.A. In vitro antioxidant and free-radical scavenging activities of polar leaf extracts of Vernonia amygdalina. BMC Complementary Med. Ther. 2023;23 doi: 10.1186/s12906-023-03923-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Wang H., Huang G., Zhang X. Analysis and properties of polysaccharides extracted from Brassica oleracea L. var. capitata L. by hot water extraction/ultrasonic-synergistic enzymatic method. Ultrason. Sonochem. 2025;114 doi: 10.1016/j.ultsonch.2025.107244. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Gu Z., Bo M., Gao Z., Ma C., Ma P. Regioisomerism effects on the thermal decomposition mechanism of fused triazole-based high-nitrogen compounds: a DFT study. J. Mol. Model. 2025;31(8):204. doi: 10.1007/s00894-025-06427-3. [DOI] [PubMed] [Google Scholar]
18.Karunarathna B., Wanniarachchi J.D., Prashantha M.A.B., Govender K.K. Enhancing styrene monomer recovery from polystyrene pyrolysis: insights from density functional theory. J. Mol. Model. 2023;29(8):255. doi: 10.1007/s00894-023-05661-x. [DOI] [PubMed] [Google Scholar]
19.Vidal-Limon A., Aguilar-Toalá J.E., Liceaga A.M. Integration of molecular docking analysis and molecular dynamics simulations for studying food proteins and bioactive peptides. J. Agric. Food Chem. 2022;70(4):934–943. doi: 10.1021/acs.jafc.1c06110. [DOI] [PubMed] [Google Scholar]
20.Xin X.Y., Park S., Wang H.W., Hur S., Li X.Z., Choi S. Quantum-chemistry-guided identification of a dual-functional cod-derived peptide targeting oxidative stress and inflammation. Bioorg. Chem. 2026;178 doi: 10.1016/j.bioorg.2026.109963. [DOI] [PubMed] [Google Scholar]
21.Chen L., Chen J., Ren J., Zhao M. Effects of ultrasound pretreatment on the enzymatic hydrolysis of soy protein isolates and on the emulsifying properties of hydrolysates. J. Agric. Food Chem. 2011;59(6):2600–2609. doi: 10.1021/jf103771x. [DOI] [PubMed] [Google Scholar]
22.Zhao F., Zhai X., Liu X., Lian M., Liang G., Cui J., Wang W. Effects of high-intensity ultrasound pretreatment on structure, properties, and enzymolysis of walnut protein isolate. Molecules. 2022;27(1) doi: 10.3390/molecules27010208. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Adal E. Modification of faba bean protein isolate by high-intensity ultrasound treatment: screening of physicochemical, thermal, and structural properties. J. Food Meas. Charact. 2024;18(3):2438–2449. doi: 10.1007/s11694-024-02379-z. [DOI] [Google Scholar]
24.Wang Y., Li B., Guo Y., Liu C., Liu J., Tan B., Jiang L. Effects of ultrasound on the structural and emulsifying properties and interfacial properties of oxidized soybean protein aggregates. Ultrason. Sonochem. 2022;87 doi: 10.1016/j.ultsonch.2022.106046. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Vate N.K., Sharifi E., Coppola A., Montuori E., Undeland I., de Pascale D., Abdollahi M. Impact of pretreatment degree and enzyme type on the production of radical scavenging and antiproliferative peptides from starfish. Mar. Drugs. 2026;24(3) doi: 10.3390/md24030120. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Zhang Y., Olsen K., Grossi A., Otte J. Effect of pretreatment on enzymatic hydrolysis of bovine collagen and formation of ACE-inhibitory peptides. Food Chem. 2013;141(3):2343–2354. doi: 10.1016/j.foodchem.2013.05.058. [DOI] [PubMed] [Google Scholar]
27.Niu X., An D., Li L. Effects of ultrasonic pretreatment on the fibrillation of soy protein isolate, β‑conglycinin, and glycinin: Kinetics, structure, and functional properties. Ultrason. Sonochem. 2026;107836 doi: 10.1016/j.ultsonch.2026.107836. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Meng Y., Liang Z., Zhang C., Hao S., Han H., Du P., Liu L. Ultrasonic modification of whey protein isolate: Implications for the structural and functional properties. LWT. 2021;152 doi: 10.1016/j.lwt.2021.112272. [DOI] [Google Scholar]
29.Qi L., Mao L., Qin X., Guo Y., Zhang C. Ultrasonic pretreatment assisted enzymolysis for preparation of low molecular weight osteogenic collagen peptides: Kinetics, thermodynamics, and osteogenic activity. Ultrason. Sonochem. 2025;120 doi: 10.1016/j.ultsonch.2025.107525. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Malik M.A., Sharma H.K., Saini C.S. High intensity ultrasound treatment of protein isolate extracted from dephenolized sunflower meal: effect on physicochemical and functional properties. Ultrason. Sonochem. 2017;39:511–519. doi: 10.1016/j.ultsonch.2017.05.026. [DOI] [PubMed] [Google Scholar]
31.Zheng Y.Z., Deng G., Guo R., Chen D.F., Fu Z.M. DFT studies on the antioxidant activity of naringenin and its derivatives: effects of the substituents at C3. Int. J. Mol. Sci. 2019;20(6) doi: 10.3390/ijms20061450. [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Yin J.Y., Han Y.N., Liu M.Q., Piao Z.H., Zhang X., Xue Y.T., Zhang Y.H. Structure-guided discovery of antioxidant peptides bounded to the Keap1 receptor as hunter for potential dietary antioxidants. Food Chem. 2022;373(Pt A) doi: 10.1016/j.foodchem.2021.130999. [DOI] [PubMed] [Google Scholar]
33.Chu Z., Qu G., Yang F., Hu Z., Luo F., Lin Q. Ultrasound-assisted enzymatic hydrolysis of loach (Misgurnus anguillicaudatus) protein for antioxidant peptides: process optimization, structural characterization, and Keap1/Nrf2-mediated antioxidant mechanism. Ultrason. Sonochem. 2025;123 doi: 10.1016/j.ultsonch.2025.107665. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Zhang M., Han Y., Liu H., Chen B., Li Q., Li C. Microstructure and digestive behaviors of inner, middle, and outer layers of pork during heating. Food Chem. 2024;458 doi: 10.1016/j.foodchem.2024.140263. [DOI] [PubMed] [Google Scholar]
35.Abdo A.A.A., Hou Y., Hassan F.A., Al-Sheraji S.H., Aleryani H., Alanazi A., Sang Y. Antioxidant potential and protective effect of modified sea cucumber peptides against H(2)O(2)-induced oxidative damage in vitro HepG2 cells and in vivo zebrafish model. Int. J. Biol. Macromol. 2024;266(Pt 2) doi: 10.1016/j.ijbiomac.2024.131090. [DOI] [PubMed] [Google Scholar]
36.Wang J., Liu J., John A., Jiang Y., Zhu H., Yang B., Wen L. Structure identification of walnut peptides and evaluation of cellular antioxidant activity. Food Chem. 2022;388 doi: 10.1016/j.foodchem.2022.132943. [DOI] [PubMed] [Google Scholar]
37.Xin X., Park S., Zhang W., Yu J., Hur S.J., Li X., Choi S.H. Integrative discovery of novel antioxidant peptides from bovine hide via peptidomics-guided molecular simulation, quantum chemistry, and experimental validation. Food Res. Int. 2026;226 doi: 10.1016/j.foodres.2025.118196. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials