Optimization of enzymatic hydrolysis for generating potential multifunctional peptides from goat milk: Bioactivity evaluation, structural characterization and molecular interaction analysis

Abstract

Food-derived bioactive peptides have attracted increasing interest for functional food development. In this study, goat milk proteins were enzymatically hydrolyzed to generate peptides with inhibitory activities against α-glucosidase (α-G), dipeptidyl peptidase-IV (DPP-IV), and xanthine oxidase (XO). Protease screening and response surface methodology were applied to optimize the composite hydrolysis process. The optimized goat milk hydrolysate exhibited inhibitory activities of 73.05 ± 1.38%, 88.63 ± 2.76%, and 80.69 ± 1.65% against α-G, DPP-IV, and XO, respectively. The active peptide fraction was further purified and analyzed by LC–MS/MS, identifying 24 peptide sequences. Bioinformatic screening highlighted LRF and FLPYPYY as potential multifunctional enzyme-inhibitory peptides. Molecular docking, 100 ns molecular dynamics simulations, and MM/PBSA analysis revealed favorable peptide–enzyme interactions and structural stability. This study provides an integrated enzymatic, analytical, and computational strategy for the discovery of goat milk-derived potential multifunctional bioactive peptides.

Highlights

• •Multifunctional peptides with hypoglycemic and uric acid-lowering activities were obtained from goat milk.
• •Papain and neutral protease synergistically enhanced peptide bioactivity.
• •Optimized hydrolysis improved inhibition of α-G, DPP-IV and XO.
• •LRF and FLPYPYY showed potential as key bioactive peptides.

1. Introduction

Diabetes and hyperuricemia are two major chronic metabolic disorders with rapidly increasing prevalence worldwide, posing serious challenges to global public health (Sun et al., 2025). Recent epidemiological estimates indicate that the number of adults living with diabetes worldwide was around 589 million in 2024, and this figure is expected to rise substantially to approximately 852.5 million by 2050 (Duncan et al., 2026). Meanwhile, the prevalence of hyperuricemia and gout has also risen markedly worldwide, and the number of gout cases is expected to reach 95.8 million globally by 2050 (Cross et al., 2024). Current pharmacological treatments, including acarbose, sitagliptin and allopurinol, are widely used for the management of these metabolic disorders. Nevertheless, the long-term use of these drugs may be associated with adverse effects, such as gastrointestinal discomfort caused by α-glucosidase (α-G) inhibitors and rare but severe hypersensitivity reactions associated with allopurinol (Stamp & Chapman, 2020; Van de Laar, 2008). Consequently, increasing attention has been paid to the development of safe and effective natural functional ingredients for preventing and managing metabolic diseases.

Bioactive peptides derived from food proteins are increasingly recognized as effective agents for nutritional intervention, owing to their high bioavailability, low toxicity and diverse regulatory functions (Chalamaiah et al., 2018). Existing studies indicate that hypoglycemic peptides modulate glucose metabolism primarily through the inhibition of key enzymes associated with carbohydrate digestion and incretin degradation, such as α-G and dipeptidyl peptidase-IV (DPP-IV) (Deng et al., 2020; Okechukwu et al., 2020). Hypouricemic peptides mainly exert their effects by suppressing uric acid formation via inhibition of xanthine oxidase (XO), a key enzyme involved in purine metabolism (Hou et al., 2022). A growing number of studies have demonstrated that peptides generated from food proteins can inhibit key metabolic enzymes, indicating their promise as functional food components for the dietary regulation of metabolic disorders.

Milk proteins are widely recognized as valuable substrates for generating bioactive peptides, owing to their well-balanced amino acid profiles, high nutritional quality and favorable digestibility (Urrutia-Baca et al., 2025). Among various dairy sources, goat milk has gained increasing interest due to its enhanced digestibility, reduced allergenicity and improved bioavailability relative to bovine milk (Wang et al., 2025). In particular, goat milk proteins, predominantly composed of caseins and whey proteins, serve as important precursors that can be enzymatically hydrolyzed to yield a diverse range of bioactive peptides (Du et al., 2022). However, current goat milk products remain mainly limited to traditional dairy powders, resulting in relatively low added value and considerable product homogeneity (Sepe & Argüello, 2019). More importantly, studies on multifunctional peptides derived from goat milk with simultaneous enzyme-inhibitory activities related to glucose and uric acid metabolism are still limited, and their structural characteristics and interaction mechanisms remain insufficiently understood.

In recent years, the integration of peptide identification with in silico approaches has provided an efficient strategy for screening potential bioactive peptides and elucidating their molecular interactions. Molecular docking can characterize possible binding modes, binding affinities, and key non-covalent interactions between peptides and target enzymes, thereby providing structural information for candidate prioritization. Molecular dynamics (MD) simulations further complement docking analysis by evaluating the dynamic stability, conformational flexibility, compactness, solvent exposure, and persistence of intermolecular interactions within peptide–enzyme complexes. Such combined computational approaches have been increasingly applied to food-derived enzyme-inhibitory peptides, including α-G-, DPP-IV-, and XO-inhibitory peptides, to facilitate candidate screening and provide insights into their potential molecular interactions (Coban et al., 2021; Wang et al., 2026; Yang et al., 2025; Zhao et al., 2021). Similar integration of experimental evaluation with molecular docking and simulations has also been employed to obtain complementary structural interpretations of biomolecular interactions (Farzandi et al., 2026; Feizi-Dehnayebi et al., 2026). Previous studies on goat milk-derived peptides have mainly focused on either hypoglycemic or hypouricemic activity, whereas their potential effects on both glucose and uric acid metabolism within the same hydrolysate remain less explored. Therefore, this study evaluated the inhibitory activities of goat milk hydrolysates against α-G, DPP-IV, and XO and optimized the enzymatic hydrolysis conditions. Bioactive peptide fractions were subsequently identified by LC–MS/MS, bioactivity prediction, and molecular docking to screen potential multifunctional peptide candidates. Molecular dynamics simulations were further conducted on the selected peptide–enzyme complexes to evaluate their dynamic stability and conformational behavior and to complement the interaction information obtained from molecular docking. The novelty of this study lies in integrating the concurrent evaluation of glucose- and uric acid-related enzyme targets with peptide identification, bioactivity screening, and structure-based analyses to prioritize potential multifunctional peptides from goat milk. The overall experimental and computational workflow is summarized in Fig. 1. This integrated approach provides further information on the potential of goat milk proteins as a source of peptides associated with both enzyme-inhibitory activities related to glucose and uric acid metabolism.

Fig. 1.

Schematic workflow from goat milk proteins to potential multifunctional peptides.

2. Materials and methods

2.1. Materials

Whole goat milk powder was provided by Xi'an Baiyue Goat Milk Group Co., Ltd. (Xi'an, China). Neutral protease (NZ), compound protease (NFH), flavor protease (NFW), alkaline protease (NS), bromelain (PB), neutral protease (PZ), papain (XM), high-activity neutral protease (XZG), and neutral protease (XZ) were purchased from Novozymes (China) Investment Co., Ltd. (Beijing, China). Pepsin was obtained from Kehao Biotechnology Co., Ltd. (Guangzhou, China). Trypsin, α-glucosidase (α-G), p-nitrophenyl-α-D-glucopyranoside, acarbose, sitagliptin, allopurinol, dipeptidyl peptidase-IV (DPP-IV), N-glycyl-prolyl-p-nitroanilide hydrochloride (Gly-Pro-pNA), xanthine oxidase (XO), and xanthine was purchased from Shanghai Yuanye Bio-Technology Co., Ltd. (Shanghai, China), and all remaining reagents were of analytical grade.

2.2. Preparation of reconstituted goat milk

According to the method reported by Akshit et al. (2024) with slight modifications, reconstituted goat milk was prepared by dissolving 12.5 g of whole goat milk powder in deionized water, adjusting the final volume to 100 mL with deionized water, and stirring with a magnetic stirrer until homogeneous. Before enzyme addition, the reconstituted milk was brought to the hydrolysis temperature specified for the corresponding protease.

2.3. Preparation of goat milk hydrolysates

Nine commercially available food-grade proteases were individually added to the reconstituted goat milk. The hydrolysis temperature for each commercial protease was selected within the operating range recommended by the respective manufacturer. Specifically, neutral protease (NZ), compound protease (NFH), flavor protease (NFW), alkaline protease (NS), and bromelain (PB) were used at 55 °C; papain (XM) at 48 °C; and high-activity neutral protease (XZG), neutral protease (XZ), and neutral protease (PZ) at 45 °C. The enzyme dosage was defined based on the protein content of the reconstituted goat milk substrate and expressed as U g⁻¹ protein. For combined protease treatments, the relative proportions of XM and XZ were calculated based on enzyme activity units (U: U). Enzymatic hydrolysis was performed for 5 h, after which enzyme activity was quenched by boiling the mixture for 10 min. The samples were cooled, centrifuged (8000 rpm, 10 min), and the supernatant was retained for subsequent analysis.

During hydrolysis, the pH was maintained by continuous addition of 0.1 mol L⁻¹ NaOH, and the volume consumed was recorded. The degree of hydrolysis (DH) was subsequently determined using the pH-stat method (Wu et al., 2016) based on the following equation:

$$\mathrm{DH}=\mathrm{B}\times {\mathrm{M}}_{\mathrm{b}}\times \frac{1}{\mathrm{\alpha }}\times \frac{1}{{\mathrm{M}}_{\mathrm{p}}}\times \frac{1}{{\mathrm{H}}_{\mathrm{tot}}}\times 100$$

where B represents the volume of NaOH (mL), M_b denotes the concentration of NaOH (mol L⁻¹), α refers to the degree of dissociation of α-amino groups, M_p corresponds to the protein mass (g), and H_tot is the total peptide bond content (mmol g⁻¹), with a value of 8.35 mmol g⁻¹ for goat milk.

2.4. Enzyme inhibitory activity assays

The α-G inhibitory activity was determined in 96-well microplates according to a previously reported method with slight modifications (Chen et al., 2014). Four groups were prepared: blank control, blank, experimental control and experimental group, with a total volume of 250 μL in each well. In the experimental group, each well was loaded with 25 μL of sample mixed with 25 μL of 2.5 mmol L⁻¹ PNPG solution. After incubation at 37 °C for 5 min, 50 μL of α-G solution (0.2 U mL⁻¹) was subsequently introduced. In the corresponding control groups, the sample solution and/or enzyme solution were replaced with 0.1 mol L⁻¹ PBS (pH 6.8) as appropriate. The reaction mixture was incubated at 37 °C for 25 min and subsequently terminated by adding 100 μL of 0.2 mol L⁻¹ Na₂CO₃. The absorbance was then measured at 405 nm using a microplate reader (Hangzhou Youmi Instrument Co., Ltd., Zhejiang, China), with acarbose used as the positive control.

A modified protocol derived from a reported method was applied to determine DPP-IV inhibitory activity (Li-Chan et al., 2012). The assay was conducted in 96-well microplates with four groups: blank control, blank, experimental control and experimental group, with a total volume of 200 μL per well. In the experimental group, 25 μL of sample was combined with 25 μL of 1.59 mmol L⁻¹ Gly-Pro-pNA substrate, after which 50 μL of DPP-IV solution (0.01 U mL⁻¹) was added. For the control groups, the sample and/or enzyme solutions were replaced with 0.1 mol L⁻¹ Tris–HCl buffer (pH 8.0). Incubation was carried out at 37 °C for 60 min, after which the reaction was quenched using 100 μL of 1 mol L⁻¹ CH₃COONa solution (pH 4.0). Measurements of absorbance at 405 nm were performed, including sitagliptin as the standard positive control.

XO inhibitory activity was determined based on a modified method reported previously (Li et al., 2018). The assay was conducted in 96-well microplates and included four groups: blank control, blank, experimental control and experimental group, with a final reaction volume of 200 μL in each well. In the experimental group, 50 μL of sample was mixed with 50 μL of XO solution (0.2 U mL⁻¹). For the corresponding control groups, the sample and/or enzyme solutions were substituted with 0.1 mol L⁻¹ PBS buffer (pH 7.4) as required. After addition of 100 μL of 1 mmol L⁻¹ xanthine substrate, the mixture was incubated at 37 °C for 30 min. Absorbance was measured at 295 nm using allopurinol as the positive control.

The inhibition rate was calculated as:

$$\text{inhibitionrate}/\%=\left[1-\frac{\left({\mathrm{A}}_1-{\mathrm{A}}_2\right)}{\left({\mathrm{A}}_3-{\mathrm{A}}_4\right)}\right]\times 100$$

where A₁ represents the absorbance of the reaction system containing both sample and enzyme; A₂ corresponds to the absorbance of the sample control; A₃ denotes the absorbance of the blank system without sample; and A₄ refers to the absorbance of the blank control.

2.5. Screening of composite proteases

Based on the hydrolysis results of single proteases, enzymes with relatively high enzyme-inhibitory activities related to glucose and uric acid metabolism were selected for combination experiments. Thermal stability and simulated gastrointestinal digestion assays were used to select the optimal composite enzyme, and the ratio of each enzyme in the mixture was further optimized.

2.6. Thermal stability analysis

Goat milk hydrolysates with inactivated enzymes were exposed to heat treatment at 63 °C for 30 min, 90 °C for 20 min and 116 °C for 20 min following a previously reported method (Zhao et al., 2020). After cooling to room temperature, the samples were centrifuged at 8000 rpm for 10 min, and the supernatant obtained was collected for subsequent analysis.

2.7. Simulated gastrointestinal digestion

Simulated gastrointestinal digestion was carried out based on a previously reported protocol with minor modifications (Liu et al., 2020). The enzyme-inactivated hydrolysates were first adjusted to pH 1.5 using 0.1 mol L⁻¹ HCl and incubated with 2% pepsin at 37 °C for 2 h. The gastric phase was terminated by heating at 98 °C for 10 min, after which the pH was readjusted to 7.5. Subsequently, 2% trypsin was added and the samples were further incubated at 37 °C for 2 h. Following trypsin inactivation at 98 °C for 10 min, the mixture was centrifuged, and the supernatant was collected for subsequent analysis.

2.8. Optimization of enzymatic hydrolysis conditions

Using the optimal composite enzyme ratio, full-fat goat milk was selected as the substrate. The influences of enzyme dosage (3000–7000 U g⁻¹), hydrolysis time (1–5 h), and hydrolysis temperature (40–60 °C) on DH and enzyme inhibitory activities were systematically examined. On the basis of the single-factor experiments, response surface methodology with a Box–Behnken design was employed to optimize the hydrolysis conditions, with α-G inhibition (Y₁) and XO inhibition (Y₂) defined as the response variables.

2.9. Peptide separation and identification

After centrifugation, the hydrolysates were subjected to ultrafiltration to obtain fractions with molecular weights below 10 kDa, followed by desalting prior to mass spectrometric analysis (Hu et al., 2025). Peptide separation was performed on a Zorbax 300SB-C18 column using 0.1% formic acid in water (A) and 0.1% formic acid in 84% acetonitrile (B) as the mobile phases. The elution gradient was programmed as 4–50% B over 0–50 min, increased to 100% B from 50 to 54 min, and maintained at 100% B until 60 min. MS analysis was conducted on a Q Exactive HF-X system (Thermo Fisher) operating in positive ion mode, and the acquired data were processed using MaxQuant 1.5.5.1.

2.10. Bioactive peptide screening

PeptideRanker (http://distilldeep.ucd.ie/PeptideRanker/) was used to evaluate the bioactivity of identified peptides, with scores above 0.5 indicating potential bioactivity. BIOPEP-UWM (https://biochemia.uwm.edu.pl/biopep-uwm/) was applied to evaluate the potential bioactivities of the selected peptides.

2.11. Molecular docking analysis

Peptides with bioactivity scores greater than 0.5 and potential α-G, DPP-IV, or XO inhibitory activity were docked to the corresponding target enzymes. The crystal structures of α-G (PDB: 2QMJ), DPP-IV (PDB: 2BGR), and XO (PDB: 1N5X) were obtained from the Protein Data Bank (http://www.rcsb.org/pdb). Prior to docking, the receptor structures were prepared by removing water molecules, adding hydrogen atoms and charges, and converting them to PDBQT format. Molecular docking was performed using AutoDock Vina 1.1.2 (The Scripps Research Institute, La Jolla, CA, USA). The docking search space was defined according to the corresponding target binding region, and the exact grid centers and box dimensions are described in the Results and Discussion section. The Vina parameters were set to exhaustiveness = 8, num_modes = 9, and energy_range = 3 kcal mol⁻¹. For each enzyme–peptide pair, the top-ranked pose with the lowest predicted binding energy was selected for subsequent interaction analysis and molecular dynamics simulation. To validate the docking protocol, structurally resolved co-crystallized ligands were extracted and re-docked into their original binding sites using the same docking parameters. The top-ranked redocked pose was compared with the crystallographic conformation, and the heavy-atom root-mean-square deviation (RMSD) was calculated in Discovery Studio based on corresponding non‑hydrogen atoms. The RMSD value between the redocked and crystallographic ligand conformations was calculated according to the following equation (Carugo & Pongor, 2001):

$$\mathit{RMSD}=\sqrt{\frac{1}{N}\sum _{i=1}^N{‖r_i^{\mathit{redock}}-r_i^{\mathit{crystal}}‖}^2}$$

where (N) is the number of corresponding heavy atoms, and ($r_i^{\mathit{redock}}$) and ($r_i^{\mathit{crystal}}$) represent the coordinates of the corresponding atoms in the redocked and crystallographic ligand conformations, respectively.

2.12. Molecular dynamics simulations

Molecular dynamics (MD) simulations of the selected enzyme–peptide complexes were performed using GROMACS 2022.2 software (Abraham et al., 2015). The top-ranked docking conformations with the lowest predicted binding energies were used as the initial structures for the MD simulations. The receptor topology files were generated using the pdb2gmx tool with the Amber99SB-ILDN force field (Lindorff-Larsen et al., 2010). The topology parameters of the peptides were generated using Sobtop 1.0 (dev3.1) with the GAFF2 force field, and RESP charges were assigned to the peptide atoms (Wang et al., 2004).

Each enzyme–peptide complex was placed in a cubic simulation box, with a minimum distance of 1.0 nm between the protein atoms and the box boundary. The system was solvated using the TIP3P water model. Na⁺ and Cl⁻ ions were added using gmx genion to neutralize the system and establish a NaCl concentration of 0.15 M. Long-range electrostatic interactions were calculated using the particle mesh Ewald method, while the cutoff distances for van der Waals and short-range electrostatic interactions were both set to 1.2 nm. Covalent bonds were constrained using the LINCS algorithm, and the integration time step was set to 2 fs.

Before the production simulation, energy minimization was sequentially performed using the steepest-descent and conjugate-gradient algorithms to eliminate unfavorable steric contacts and local high-energy conformations. The minimized systems were subsequently equilibrated for 1 ns under the canonical ensemble (NVT), followed by 1 ns under the isothermal–isobaric ensemble (NPT). Positional restraints were applied to both the receptor proteins and peptides during the equilibration stages. The temperature was maintained at 300 K using the V-rescale thermostat with a coupling time constant of 1 ps. During NPT equilibration, the pressure was maintained at 1 bar using the Parrinello–Rahman barostat with a coupling time constant of 1 ps. After equilibration, the positional restraints were removed, and unrestrained production MD simulations were performed for 100 ns for all systems.

The simulation trajectories were analyzed using the tools implemented in GROMACS 2022.2. RMSD profiles were evaluated for the receptor, peptide ligand, and receptor–peptide complex to characterize their time-dependent conformational behavior during the 100 ns simulations. Backbone RMSD, residue root-mean-square fluctuation (RMSF), radius of gyration (Rg), and solvent-accessible surface area (SASA) were calculated using gmx rms, gmx rmsf, gmx gyrate, and gmx sasa, respectively. Two-dimensional and three-dimensional free-energy landscapes (FELs) were constructed using gmx sham to characterize preferential conformational states sampled during the simulations. In the FEL plots, the two coordinates labeled PC1 and PC2 correspond to RMSD and Rg, respectively. MM/PBSA binding free energies, energy-component decomposition, and per-residue free-energy contributions were calculated using gmx_MMPBSA version 1.6.4 with the MMPBSA.py v.16.0 kernel. For each complex, the 90–100 ns segment of the production trajectory was selected, and 100 conformations were extracted at equal intervals for energy-statistical analysis.

2.13. Statistical analysis

Unless otherwise stated, all experiments were conducted using three independently prepared hydrolysate batches under identical experimental conditions, which were treated as independent experimental replicates (n = 3). Each assay was performed in technical triplicate for each independently prepared hydrolysate. The technical replicate measurements were averaged to obtain one value for each independent hydrolysis experiment, and statistical analyses were performed using the three independent experimental values. Data are presented as the mean ± standard deviation (SD).

Response-surface model fitting and analysis of variance were performed using Design-Expert 12 (Stat-Ease, Inc., Minneapolis, MN, USA). Other statistical analyses were performed using IBM SPSS Statistics 22.0 (IBM Corp., Armonk, NY, USA). Differences among treatment groups were analyzed by one-way analysis of variance followed by Tukey's honestly significant difference test. Differences were considered statistically significant at p < 0.05. Different lowercase letters (a–e) indicate significant differences among treatments, whereas the same lowercase letter indicates no significant difference. Figures were prepared using Origin 2026 (OriginLab Corp., Northampton, MA, USA).

3. Results and discussion

3.1. Effects of single proteases on hydrolysis and enzyme inhibition

3.1.1. Degree of hydrolysis (DH)

The degree of hydrolysis (DH) serves as an important parameter for evaluating protein hydrolysis, representing the extent of peptide bond cleavage and affecting both the yield and functional properties of the resulting products (Qiao et al., 2020). The hydrolysis curves of goat milk hydrolysates (GMH) are shown in Fig. 2A. After 5 h of hydrolysis, the DH of goat milk increased progressively with time. Different proteases exhibited varying hydrolysis efficiencies toward goat milk, with DH values ranging from 0 to 10.7%. The DH obtained with different proteases followed the order: NS > NFH > XZ > XZG > PZ > XM > NZ > PB > NFW.

Fig. 2.

Degree of hydrolysis (DH) of GMH obtained using different proteases during 5 h hydrolysis(A); α-G inhibitory activity(B); DPP-IV inhibitory activity(C); XO inhibitory activity(D); effects of protease combinations on enzyme inhibitory activities(E); changes in enzyme inhibitory activities after simulated gastrointestinal digestion(F); effects of thermal treatment on enzyme inhibitory activities(G).

The differences in DH among proteases may be attributed to variations in their cleavage sites on the substrate, as well as differences in the molecular weight and amino acid composition of the peptides produced during hydrolysis (Tu et al., 2018). Meanwhile, the increase in DH over time may result from the unfolding or denaturation of native proteins, which increases the accessibility of active sites and enhances enzymatic reactivity, leading to the cleavage of more peptide bonds (Ambigaipalan & Shahidi, 2015).

GMH with alkaline protease (NS) exhibited the highest DH, which may be attributed to the fact that alkaline protease is an endopeptidase with multiple catalytic sites capable of efficiently cleaving peptide bonds in proteins (Lapsongphon & Yongsawatdigul, 2013). In contrast, flavor protease exhibited relatively low DH due to the limited availability of cleavage sites (Merz et al., 2015). Similar findings were reported by Du et al. (2022), who evaluated the effects of seven proteases on whey protein hydrolysis and observed that alkaline protease yielded the highest DH. Likewise, Liu et al. (2022) demonstrated that alkaline protease achieved the highest DH when hydrolyzing mung bean proteins using five different proteases.

3.1.2. Enzyme inhibitory activities of goat milk hydrolysates

Protease type markedly influenced the enzyme-inhibitory activities related to glucose and uric acid metabolism of GMH. As shown in Fig. 2B–D, the inhibitory activities of the hydrolysates differed after 5 h of hydrolysis depending on the enzyme used. α-G inhibition ranged from 15.75% to 53.37%, with XM showing the highest activity, followed by XZ, NZ, PB and NFH. DPP-IV inhibition ranged from 58.90% to 97.08%, with XM exhibiting the strongest effect, followed by PB, XZ, PZ and NFW. In contrast, XO inhibition ranged from 35.25% to 69.50%, with XZ showing the highest activity, followed by XZG, NZ, NFH and PZ.

The differences in enzyme-inhibitory activities related to glucose and uric acid metabolism among goat milk hydrolysates obtained using different proteases may be attributed to variations in cleavage sites during enzymatic hydrolysis. Different proteases generate peptides with distinct amino acid compositions and molecular structures, which consequently influence their bioactivities. Du et al. (2023) reported that hydrolysis of goat milk proteins with papain and bromelain produced peptides with relatively high DPP-IV inhibitory activity.

Based on the inhibitory activity results and statistical significance analysis, XZG (A), PB (B), XM (C), and XZ (D) were selected for subsequent combination experiments under hydrolysis conditions of 50 °C for 5 h.

3.2. Selection of composite proteases

3.2.1. Effects of protease combinations on hydrolysis and bioactivity

Different proteases exhibit distinct cleavage sites on protein substrates, resulting in hydrolysates with different peptide compositions and biological activities. Consequently, peptides produced by single-enzyme hydrolysis often fail to achieve optimal enzyme-inhibitory activities related to glucose and uric acid metabolism simultaneously; therefore, protease combinations were applied to enhance bioactivity. Based on the results of single-enzyme hydrolysis, XZG (A), PB (B), XM (C), and XZ (D) were selected and used for enzymatic hydrolysis of goat milk at 50 °C for 5 h. For combined protease treatments, the proportions of individual enzymes were calculated based on enzyme activity units (U: U). The optimal composite enzyme combinations were subsequently determined based on the resulting enzyme-inhibitory activities.

As shown in Fig. 2E, after hydrolysis with composite proteases, the DPP-IV inhibitory activities of goat milk hydrolysates were all above 80%. Among the tested combinations, hydrolysates obtained using XZG + XM + XZ (ACD) and XM + XZ (CD) exhibited relatively stronger inhibitory activities (p < 0.05). The α-G inhibition rates were 59.45% ± 1.23% and 53.12% ± 3.18%, respectively; the DPP-IV inhibition rates were 84.78% ± 2.94% and 90.56% ± 3.04%, respectively; and the XO inhibition rates were 73.24% ± 0.66% and 68.59% ± 0.43%, respectively. Based on the inhibitory activity results and statistical significance analysis, the ACD and CD combinations were selected for subsequent experiments on the preparation of goat milk hydrolysates using composite enzymatic hydrolysis.

3.2.2. Stability of goat milk hydrolysates during simulated gastrointestinal digestion

To investigate changes in enzyme-inhibitory activities related to glucose and uric acid metabolism during simulated gastrointestinal digestion, GMH obtained from the composite enzymatic hydrolysis were subjected to in vitro digestion stability tests. Following simulated gastrointestinal digestion, the inhibitory activities of hydrolysates prepared using the ACD and CD enzyme combinations remained relatively stable (Fig. 2F). The α-G inhibition rates were 67.43% ± 1.25% and 66.16% ± 3.59%, respectively, with no significant difference (p > 0.05). Similarly, the DPP-IV inhibition rates were 85.15% ± 4.48% and 89.42% ± 1.19%, respectively, also showing no significant difference (p > 0.05). In contrast, the XO inhibition rates were 80.53% ± 0.13% and 77.03% ± 0.53%, respectively, with a significant difference observed between the two groups (p < 0.05).

During simulated gastrointestinal digestion, the inhibitory activities of the hydrolysates increased at the initial stage and subsequently decreased after exposure to gastric and intestinal conditions, while remaining significantly higher than those of the untreated control (p < 0.05). This trend may be attributed to the sequential hydrolysis processes mediated by pepsin and trypsin. During gastric digestion, the release of smaller peptides may expose additional active sites, resulting in enhanced enzyme-inhibitory activities. However, further hydrolysis during the intestinal phase may degrade previously generated active peptides or alter their sequence characteristics, leading to partial changes in activity. Similar trends have been reported for casein hydrolysates. Mudgil et al. (2021) demonstrated that simulated gastrointestinal digestion influenced the α-G inhibitory activity of camel and bovine casein hydrolysates and contributed to the generation of potential antidiabetic peptides. Carrera-Alvarado et al. (2024) reported increased DPP-IV inhibitory activity after digestion, likely due to the release of smaller peptide fragments with stronger inhibitory effects (Zhang et al., 2024). Likewise, Qi, Chen, Guan, Sun, Wang, and Ma (2022) found that whey protein isolates released more potent XO-inhibitory peptides after simulated gastrointestinal digestion.

3.2.3. Stability of goat milk hydrolysates under thermal treatment

As shown in Fig. 2G, temperature had different effects on the inhibitory activities of GMH depending on the enzyme combinations. For α-G, the inhibition activity of hydrolysates from the ACD group decreased after heat treatment, whereas that of the CD group increased (p < 0.05). The AD group showed a slight increase after high-temperature treatment, although no significant difference was observed at 90 °C (p > 0.05), while the AC group exhibited increased activity at 63 °C followed by a decrease at higher temperatures. For DPP-IV, the inhibition activity of the ACD group slightly decreased with increasing temperature (80–90%), whereas that of the CD group gradually increased (87–92%), indicating that heat treatment had a limited overall effect. Similar results were reported by Bunsroem et al. (2022), although Pu et al. (2024) observed a reduction in activity after heating. In contrast, high temperature significantly enhanced XO inhibition, with all enzyme combinations reaching maximum activity at 116 °C, and the CD group showing the highest inhibition rate (84.58% ± 0.83%, p < 0.05).

Heating at different temperatures may induce conformational changes in bioactive peptides, such as denaturation or aggregation, thereby enhancing or reducing their bioactivity. This may result from the disruption or modification of specific amino acids, as well as enhanced molecular interactions that affect the biological activity of food components (Ashaolu et al., 2023; Sun et al., 2022). Moreover, heat treatment may induce Maillard reactions between peptides and reducing sugars present in goat milk (Rivero-Pino et al., 2020a). The differences in inhibitory activity after heat treatment may be attributed to the distinct thermal stability of peptide fragments generated by different proteolytic hydrolysis processes, which ultimately influences the overall activity of the hydrolysates. Based on the significance analysis results, the XM + XZ (CD) combination was selected for subsequent goat milk hydrolysis experiments.

3.3. Optimization of enzymatic hydrolysis

3.3.1. Effects of protease ratio on bioactivity

Different proteases generate distinct peptide profiles during protein hydrolysis, and enzyme dosage can also influence the yield of bioactive peptides. Based on the stability evaluation described above, papain (XM, C) and neutral protease (XZ, D) were selected for further investigation. Goat milk hydrolysates were generated at 50 °C for 5 h, after which the optimal proportion of the two enzymes in the composite system was assessed. The XM:XZ ratios were defined on the basis of enzyme activity units (U: U). As illustrated in Fig. 3A, the XM: XZ activity ratio of 1:2 resulted in the strongest inhibitory activities against DPP-IV and α-G, with values of 97.19% ± 0.55% and 58.13% ± 0.49%, respectively (p < 0.05). In contrast, the maximum XO inhibition (75.61% ± 1.59%) was observed at an XM: XZ activity ratio of 1:3 (p < 0.05). Considering the overall inhibitory effects against α-G, DPP-IV and XO, the XM: XZ activity ratio of 1:2 was selected for subsequent process optimization.

Fig. 3.

Effects of XM:XZ ratio on enzyme inhibitory activities of GMH (A); effects of hydrolysis time on enzyme inhibitory activities (B); effects of enzyme dosage on enzyme inhibitory activities (C); effects of hydrolysis temperature on enzyme inhibitory activities (D); standard curves of positive control drugs (E).

3.3.2. Effects of hydrolysis conditions

(a) Hydrolysis time

As illustrated in Fig. 3B, the enzyme-inhibitory activities related to glucose and uric acid metabolism of goat milk hydrolysates increased during the initial stage of hydrolysis and subsequently declined with prolonged time. The highest inhibitory effects against XO and α-G were obtained at 4 h (p < 0.05), whereas DPP-IV inhibition reached its maximum at 3 h (p < 0.05). However, no significant difference was detected between 3 h and 4 h (p > 0.05). Considering these results, 4 h was selected as the optimal hydrolysis time. The decrease in inhibitory activity after prolonged hydrolysis may be attributed to the reduction in available proteolytic cleavage sites, leading to further degradation of previously generated bioactive peptides and disruption of protein structures (Hartman et al., 2026). In addition, excessive hydrolysis time may result in enzyme inactivation, thereby reducing peptide bioactivity (Felix et al., 2020).

(b) Enzyme dosage

With increasing enzyme dosage, the enzyme-inhibitory activities related to glucose and uric acid metabolism of goat milk hydrolysates first increased and then declined (Fig. 3C). The highest inhibitory activities against α-G, DPP-IV and XO were observed at an enzyme dosage of 4000 U g⁻¹ (p < 0.05). Therefore, 4000 U g⁻¹ was selected as the optimal enzyme dosage. Enzyme dosage affects the generation of bioactive peptides through its impact on enzyme–substrate interactions and the extent of hydrolysis. At low enzyme concentrations, substrates tend to accumulate around enzyme active sites, limiting effective interactions. Increasing enzyme concentration enhances substrate dispersion and hydrolysis efficiency, thereby promoting the release of hypoglycemic and hypouricemic peptides. However, excessive enzyme-to-substrate (E/S) ratios may lead to over-hydrolysis, resulting in further degradation of bioactive peptides into inactive fragments and consequently reducing their biological activity (Hao et al., 2024).

(c) Hydrolysis temperature

The enzyme-inhibitory activities related to glucose and uric acid metabolism of goat milk hydrolysates increased initially and then declined with rising hydrolysis temperature (Fig. 3D). The strongest inhibitory effects on α-G and XO were obtained at 55 °C (p < 0.05), whereas peak DPP-IV inhibition was observed at 50 °C (p < 0.05). No significant difference was detected between 50 °C and 55 °C (p > 0.05). Accordingly, 55 °C was selected as the optimal hydrolysis temperature. Temperature plays a crucial role in enzymatic hydrolysis by affecting enzyme stability and reaction kinetics. At relatively low temperatures, enzyme activity is limited, resulting in reduced production of bioactive peptides. Within the optimal temperature range, enzymes maintain structural stability and enhanced molecular activity, leading to higher reaction rates and improved bioactivity. However, excessively high temperatures may cause denaturation of both enzymes and substrate proteins, resulting in loss of catalytic activity and reduced peptide bioactivity (Eğilmez & Haspolat, 2024).

3.3.3. Response surface optimization

Based on the results of the single-factor experiments, α-G inhibition and XO inhibition were selected as response variables. DPP-IV inhibition was monitored throughout the single-factor experiments but was not selected as a response variable because it remained consistently high within a comparatively narrow range, thereby providing limited discrimination among the hydrolysis conditions. In contrast, α-G and XO inhibition showed greater condition-dependent variation and were therefore selected as the more discriminating response variables for process optimization. DPP-IV inhibition was retained as an independent validation endpoint under the optimized conditions. Response surface methodology was subsequently applied to optimize the preparation conditions of hypoglycemic and hypouricemic peptides from goat milk within the defined parameter ranges. A Box–Behnken design (BBD) was further employed to evaluate the interactions among the three hydrolysis variables. In the BBD, A represents the total composite enzyme dosage (U g⁻¹ protein; XM: XZ = 1:2, based on enzyme activity units), B represents the hydrolysis temperature (°C), and C represents the hydrolysis time (h), while Y₁ and Y₂ represent α-G inhibition (%) and XO inhibition (%), respectively. The experimental design and corresponding results are presented in Table 1.

Table 1.

Experimental design and results of Box-Behnken method to optimize the process of goat milk hydrolysis by multi-enzymes.

Run	A (U g−1)	B (°C)	C (h)	Y1 (%)	Y2 (%)
1	3000 (−1)	50	4	71.2809	78.029
2	3000	55	5	67.4115	73.1015
3	5000 (1)	60	4	68.8567	58.0598
4	4000 (0)	55	4	64.9873	65.3537
5	5000	50	4	58.787	59.7455
6	4000	50	5	63.4955	79.4229
7	4000	60	3	60.2788	62.4361
8	4000	55	4	66.8521	59.3241
9	4000	50	3	68.5304	57.4114
10	4000	55	4	64.4279	63.2142
11	5000	55	3	63.7752	53.6186
12	3000	60	4	60.3254	62.8251
13	3000	55	3	70.908	69.4707
14	5000	55	5	70.0688	64.6729
15	4000	60	5	67.2717	59.6806

Note. A, composite enzyme dosage (XM: XZ = 1:2 on an activity-unit basis, U g⁻¹ protein); B, hydrolysis temperature (°C); C, hydrolysis time (h); Y₁, α-G inhibition (%); Y₂, XO inhibition (%).

Regression analysis of the experimental data was performed using Design-Expert software, and quadratic polynomial models were established to characterize the relationships between the independent factors and the corresponding responses, as shown below:

$$Y_{\mathit{1}}=65.42–1.05A-0.6702B+0.5944C+5.26\mathit{AB}+2.45\mathit{AC}+3.01\mathit{BC}+1.27A^2–1.88B^2+1.35C^2$$

$$Y_{\mathit{2}}=62.63–5.92A-3.95B+4.24C+3.38\mathit{AB}+1.86\mathit{AC}-6.19\mathit{BC}+1.26A^2+0.778B^2+1.33C^2$$

According to the absolute values of the linear regression coefficients, the relative influence of the factors on α-G inhibition activity followed the order: enzyme dosage (A) > hydrolysis temperature (B) > hydrolysis time (C), whereas that on XO inhibition activity followed the order: enzyme dosage (A) > hydrolysis time (C) > hydrolysis temperature (B). The ANOVA results are summarized in Table 2. The coefficients of determination (R²) for Y₁ and Y₂ were 0.9773 and 0.9723, respectively, indicating that the proposed models explained 97.73% and 97.23% of the variability in α-G and XO inhibitory activities. The adjusted R² values were 0.9365 and 0.9224, respectively, showing good agreement with the R² values and confirming the reliability of the models.

Table 2.

ANOVA for the response surface models of α-G and XO inhibitory activities of goat milk hydrolysates.

Source	α-G inhibition					XO inhibition
	SS	df	MS	F-value	p-value	SS	df	MS	F-value	p-value
Model	213.66	9	23.74	23.96	0.0014**	774.48	9	86.05	19.49	0.0022**
A	8.90	1	8.90	8.98	0.0302*	280.01	1	280.01	63.41	0.0005***
B	3.59	1	3.59	3.63	0.1152	124.88	1	124.88	28.28	0.0031**
C	2.83	1	2.83	2.85	0.1520	144.00	1	144.00	32.61	0.0023**
AB	110.52	1	110.52	111.54	0.0001***	45.68	1	45.68	10.35	0.0236*
AC	23.96	1	23.96	24.18	0.0044**	13.78	1	13.78	3.12	0.1376
BC	36.17	1	36.17	36.50	0.0018**	153.35	1	153.35	34.73	0.0020**
A2	5.94	1	5.94	6.00	0.0580	5.83	1	5.83	1.32	0.3027
B2	13.03	1	13.03	13.15	0.0151*	2.24	1	2.24	0.5061	0.5086
C2	6.73	1	6.73	6.79	0.0479*	6.52	1	6.52	1.48	0.2785
Residual	4.95	5	0.9908			22.08	5	4.42
Lack of fit	1.73	3	0.5772	0.3583	0.7933	3.39	3	1.13	0.1210	0.9398
Pure error	3.22	2	1.61			18.69	2	9.34
Total	218.61	14				796.56	14

Note. A, composite enzyme dosage (U g⁻¹ protein); B, hydrolysis temperature (°C); C, hydrolysis time (h); AB, interaction between enzyme dosage and temperature; AC, interaction between enzyme dosage and hydrolysis time; BC, interaction between temperature and hydrolysis time; A², B², and C², quadratic effects. SS, sum of squares; df, degrees of freedom; MS, mean square. * p < 0.05, ** p < 0.01, *** p < 0.001.

Both models showed high statistical significance (p < 0.01), whereas the lack-of-fit was not significant (p > 0.05), demonstrating that the models were suitable for predicting the responses within the studied ranges. For α-G inhibition, the linear factor A, interaction terms AB, AC and BC, as well as the quadratic components B² and C², exhibited significant effects (p < 0.05). By comparison, in the case of XO inhibition, significant contributions (p < 0.05) were mainly associated with the linear terms A, B and C and the interaction terms AB and BC. Three-dimensional response surface and contour plots were generated from the regression models to visualize the interactions among A, B and C (Fig. 4). The analysis revealed that lower enzyme dosage and hydrolysis temperature, together with extended hydrolysis time, promoted the generation of peptides with enhanced α-G and XO inhibitory activities. These results suggest that synergistic interactions among hydrolysis parameters play a critical role in determining the yield and bioactivity of peptides derived from goat milk proteins.

Fig. 4.

Response surface and contour plots showing the interactive effects of enzyme dosage and hydrolysis temperature on α-G inhibitory activity (A); enzyme dosage and hydrolysis time on α-G inhibitory activity (B); hydrolysis temperature and hydrolysis time on α-G inhibitory activity (C); enzyme dosage and hydrolysis temperature on XO inhibitory activity (D); enzyme dosage and hydrolysis time on XO inhibitory activity (E); hydrolysis temperature and hydrolysis time on XO inhibitory activity (F).

3.3.4. Validation of the optimized hydrolysis conditions

The optimal conditions for producing hypoglycemic and XO inhibitory peptides from goat milk by composite enzymatic hydrolysis were determined using response surface methodology. The optimized parameters included an enzyme dosage of 3000 U g⁻¹ with an XM:XZ ratio of 1:2, a hydrolysis temperature of 50 °C, and a hydrolysis time of 4.109 h. For practical operation, the hydrolysis time was adjusted to 4.1 h. Under these optimized conditions, the predicted inhibition rates of α-G and XO were 71.28% and 78.86%, respectively.

Validation experiments were conducted under the optimized conditions. The results showed that the α-G inhibition rate was 73.05% ± 1.38%, the DPP-IV inhibition rate was 88.63% ± 2.76%, and the XO inhibition rate was 80.69% ± 1.65%. No significant difference was found between the experimental and predicted values, confirming the reliability and validity of the hydrolysis parameters optimized by RSM.

Du et al. (2022) reported that the DPP-IV inhibitory activity of goat milk casein hydrolysates increased with protein concentration. At 2.5 mg mL⁻¹, papain hydrolysates achieved an inhibition rate of 71.0% ± 0.98%. In contrast, the hydrolysates obtained in this study exhibited higher inhibitory activity, which may be attributed to differences in protease selection, substrate characteristics and concentration, as well as hydrolysis conditions, all of which can affect the composition and yield of bioactive peptides.

3.3.5. Positive-control drugs as internal assay references

The standard curves of the positive control drugs acarbose, sitagliptin, and allopurinol are shown in Fig. 3E. The linear regression equation for acarbose was y = 0.7093× + 24.806 (R² = 0.9868). The linear regression equation for sitagliptin was y = 21.534× + 26.889 (R² = 0.9815), while that for allopurinol was y = 0.4126× + 53.983 (R² = 0.9653).

Following optimization of the enzymatic hydrolysis process by response surface methodology, the inhibition percentages of the goat milk hydrolysate corresponded, according to the respective regression curves, to those produced by 68.02 μg mL⁻¹ acarbose for α-G, 2.87 μg mL⁻¹ sitagliptin for DPP-IV, and 64.73 μg mL⁻¹ allopurinol for XO under the specific in vitro assay conditions.

The positive-control drugs were included to confirm the reliability and responsiveness of the enzyme inhibition assays. The inhibitory activities of goat milk hydrolysates showed comparable trends with the reference standards under the established assay conditions, indicating the potential of the hydrolysates as sources of enzyme-inhibitory peptides. However, considering that the hydrolysates are complex mixtures containing diverse peptide components, the comparison with pure reference compounds was only used as an internal assay reference rather than a direct evaluation of inhibitory potency.

3.4. Peptide identification and mechanistic analysis

3.4.1. Bioactivity of preparative chromatography fractions

Hypoglycemic and uric acid–lowering peptides generally possess relatively low molecular weights. Therefore, the supernatant of goat milk obtained after enzymatic hydrolysis and centrifugation was ultrafiltered using a 10 kDa membrane to obtain the <10 kDa fraction, designated as EG. The EG fraction was further separated by preparative chromatography, and the chromatogram is shown in Fig. 5A.

Fig. 5.

Preparative chromatogram of fraction EG; α-G, DPP-IV and XO inhibitory activities of fractions EG1, EG2, EG3 and EG4 (A); MS/MS spectrum of LRF and FLPYPYY (B).

The resulting fractions (EG1, EG2, EG3 and EG4) were collected, freeze-dried, and reconstituted to 2 mg mL⁻¹, followed by evaluation of their inhibitory activities against α-G, DPP-IV, and XO (Fig. 5A). Fraction EG3 exhibited the strongest inhibitory activities among the four fractions (p < 0.05), with α-G, DPP-IV and XO inhibition rates of 42.23% ± 1.67%, 57.13% ± 4.77% and 53.72% ± 3.16%, respectively. Therefore, the EG3 fraction was selected for subsequent LC–MS/MS analysis and peptide identification.

3.4.2. LC–MS/MS identification of peptides

The EG3 fraction was analyzed using a Q Exactive HF-X mass spectrometer. Peptide sequences were identified through database searching with MaxQuant 1.5.5.1, followed by bioactivity assessment and protein annotation. A total of 24 peptides, with molecular weights ranging from 400 to 1400 Da, were detected, mainly derived from β-casein, αS1-casein and αS2-casein. The identified sequences and their corresponding precursor proteins are listed in Table S1, while the MS/MS spectra are provided in Supplementary Fig. S1.

The identified peptides were first evaluated using PeptideRanker, and peptides with scores above 0.5 were retained. Their potential enzyme-inhibitory activities were subsequently evaluated using BIOPEP-UWM. Among them, six peptides, namely LRF, FLPYPYY, LVYPF, LQYPYQGPIVL, VMFPPQ and VVVPPF, showed potential α-G inhibitory activity. Ten peptides, including FLSW, LVYPF, FLPYPYY, HLPLP, GPFP, LQYPYQGPIVL, IQAF, LQWQ, VVVPPF and VMFPPQ, exhibited potential DPP-IV inhibitory activity. In addition, LRF showed potential XO inhibitory activity. Notably, LRF was predicted to possess dual inhibitory potential against α-G and XO, whereas FLPYPYY was predicted to target both α-G and DPP-IV. Moreover, LVYPF, LQYPYQGPIVL, VMFPPQ and VVVPPF were predicted to possess both α-G and DPP-IV inhibitory activities.

Previous research has indicated that the bioactivity of hypoglycemic and hypouricemic peptides is closely related to their amino acid composition and structural characteristics. Peptides with α-G inhibitory activity are often characterized by the presence of residues such as Ser, Thr, Lys, Tyr or Arg at the N-terminus, Pro within the peptide sequence, and Met, Arg or Ala at the C-terminus (Ibrahim et al., 2018; Berraquero-García et al., 2023; Liu et al., 2021). In addition, peptides containing Pro, Leu and Ala often exhibit strong α-G inhibitory activity, mainly through hydrophobic interactions (Li et al., 2023; Ren et al., 2016; Wang et al., 2019). Several peptides identified in the EG3 fraction, including LRF, LVYPF, VVVPPF, VMFPPQ and FLPYPYY, conform to these structural characteristics.

DPP-IV inhibitory peptides are commonly characterized by the presence of Pro residues, particularly at the second position from the N-terminus, often in combination with Leu, Val or Phe (Harnedy-Rothwell et al., 2020; Rivero-Pino et al., 2020b). In addition, these peptides tend to exhibit a high proportion of hydrophobic amino acids, such as Ala, Gly, Leu, Val and Phe, together with hydrophilic residues including Gln, Ser, Lys and Arg (Nongonierma et al., 2017; Jia et al., 2020; Nongonierma & FitzGerald, 2019). Short peptides containing Pro at the C-terminal position have also been reported to display strong DPP-IV inhibitory activity (Nongonierma & FitzGerald, 2013). Notably, several peptides identified in the present study, including FLSW, LVYPF, LQWQ, IQAF, GPFP and HLPLP, exhibit these structural features.

Previous studies suggest that XO inhibitory activity is closely associated with the presence of aromatic amino acids, particularly Trp and Phe. Trp-containing peptides may exert inhibitory effects on XO due to their structural similarity to allopurinol (Nongonierma & FitzGerald, 2012; Nongonierma et al., 2013). In addition, peptides enriched in Phe have been reported to display stronger XO inhibition (He et al., 2019), while Trp-rich peptides derived from walnut protein have also shown pronounced inhibitory activity (Li et al., 2019). Several peptides identified in the EG3 fraction, including FLPYPYY, IQAF, LQWQ, FLSW and LRF, are consistent with these structural characteristics. Notably, both LRF and FLPYPYY contain key structural features associated with enzyme-inhibitory activities related to glucose and uric acid metabolism, including aromatic and hydrophobic residues, suggesting their potential as multifunctional bioactive peptides. The MS/MS spectra corresponding to LRF and FLPYPYY are presented in Fig. 5B. Based on the above analysis, LRF and FLPYPYY were selected for subsequent molecular docking analysis.

3.4.3. Molecular docking analysis

Molecular docking was performed to investigate the binding modes of LRF and FLPYPYY with their corresponding target enzymes. The reliability of the docking protocol was evaluated by redocking the co-crystallized ligands into the original binding sites. The detailed docking parameters and the superposition of crystallographic and redocked ligand conformations are provided in Supplementary Table S2, Table S3 and Fig. S2. The RMSD values of the redocked ligands for α-G (PDB: 2QMJ), DPP-IV (PDB: 2BGR), and XO (PDB: 1N5X) were 1.9817, 1.6828, and 0.8953 Å, respectively. These values were below 2.0 Å, confirming the reliability of the docking protocol for subsequent peptide–enzyme interaction analysis (Xiong et al., 2026).

The binding energies of FLPYPYY with α-G and DPP-IV were −9.7 and −8.8 kcal mol⁻¹, respectively, while LRF showed binding energies of −7.5 and −7.2 kcal mol⁻¹ toward α-G and XO, respectively. The negative values indicated favorable interactions between the peptides and target enzymes. Similar binding characteristics have been reported for food-derived enzyme-inhibitory peptides, in which hydrogen bonding, hydrophobic interactions, and van der Waals forces contribute to peptide recognition and stabilization within enzyme-binding regions (Du et al., 2023).

As shown in Fig. 6A, FLPYPYY interacted with α-G through hydrogen bonds with Tyr733, Arg647, Glu767, and Gly766, together with a Pi–Alkyl interaction involving Lys765. Hydrogen bonding has been frequently observed in α-G inhibitory peptide–enzyme complexes and contributes to ligand stabilization within active regions (Velarde-Salcedo et al., 2013). The LRF–α-G complex (Fig. 6B) was stabilized by hydrogen bonds involving Asp649, Arg653, Glu788, and Gly766, accompanied by Pi–Pi and Pi–Alkyl interactions with Tyr636 and Pro676. These interactions were consistent with previous reports that aromatic residues and hydrophobic contacts participate in peptide recognition by α-G (Yang et al., 2014).

Fig. 6.

Molecular docking interactions of FLPYPYY with α-G (A); molecular docking interactions of LRF with α-G (B); molecular docking interactions of FLPYPYY with DPP-IV (C); molecular docking interactions of LRF with XO (D); root mean square deviation (RMSD) profiles of the four enzyme–peptide complexes and the corresponding free enzymes during 100 ns molecular dynamics simulations (E); radius of gyration (Rg) profiles of the four enzyme–peptide complexes (F); solvent-accessible surface area (SASA) profiles of the four enzyme–peptide complexes (G); and root mean square fluctuation (RMSF) profiles of the four enzyme–peptide complexes (H).

For DPP-IV, FLPYPYY formed hydrogen bonds with Tyr48 and Arg560, while Gly628 and Lys554 contributed through van der Waals interactions. In addition, π–π interactions were observed between FLPYPYY and Trp629, Trp627, and Tyr547 (Fig. 6C). Aromatic and hydrophobic residues have been reported to play important roles in DPP-IV inhibitory peptide binding, particularly through interactions with the enzyme active region (Du et al., 2023). In the LRF–XO complex (Fig. 6D), hydrogen bonds were formed with Arg37, Gly820, Gly35, and Tyr817, together with Pi–Alkyl and Pi–Sigma interactions involving Val591 and Arg37. The presence of aromatic residues in LRF, especially Phe, agrees with previous observations that Phe-, Tyr-, and Trp-containing peptides tend to exhibit stronger XO inhibitory potential due to their involvement in enzyme–peptide interactions (He et al., 2019).

Overall, the docking results revealed that LRF and FLPYPYY could establish multiple non-covalent interactions with α-G, DPP-IV, and XO. Although the docking results cannot directly represent the inhibitory potency of individual peptides in the absence of purified peptide validation, they provide structural information supporting the potential contribution of LRF and FLPYPYY to the enzyme-inhibitory activities observed in goat milk hydrolysates.

3.4.4. Molecular dynamics simulations

Molecular dynamics (MD) simulations were performed to evaluate the structural stability and dynamic behavior of the four docking-derived enzyme–peptide complexes over 100 ns. As shown in Fig. 6E, the α-G–FLPYPYY and α-G–LRF complexes reached relatively stable states after approximately 20 ns, with RMSD values fluctuating around 0.16 and 0.10 nm, respectively, generally lower than those of free α-G. The DPP-IV–FLPYPYY complex required approximately 70 ns to reach equilibrium and subsequently fluctuated around 0.18 nm, whereas the XO–LRF complex stabilized after approximately 20 ns with an RMSD of approximately 0.17 nm. Similar reductions in RMSD after peptide binding have been reported for α-G-, DPP-IV-, and XO-inhibitory peptide systems (Chen et al., 2025; Li et al., 2026; Qi et al., 2022; Wang et al., 2026). Overall, the stable RMSD profiles indicated that peptide binding did not induce progressive structural deviation and generally restricted global conformational fluctuations.

The Rg and SASA profiles of the four complexes showed only limited fluctuations throughout the simulations (Fig. 6F–G), indicating maintenance of overall structural compactness and solvent exposure. Consistently, the RMSF values of most residues in the α-G–FLPYPYY, α-G–LRF, and DPP-IV–FLPYPYY complexes remained below 0.3 nm, whereas most residues in the XO–LRF complex remained below 0.5 nm (Fig. 6H). Although several local regions exhibited greater flexibility, the combined RMSD, Rg, SASA, and RMSF results suggest that all four complexes maintained relatively stable global structures during the simulated period.

Free-energy landscape (FEL) analysis further characterized the preferred conformational states of the complexes (Fig. 7). Using RMSD and Rg as reaction coordinates, the α-G–FLPYPYY, α-G–LRF, DPP-IV–FLPYPYY, and XO–LRF complexes showed concentrated low-free-energy regions centered at approximately 0.14/2.81, 0.12/2.825, 0.17/2.72, and 0.16/3.14 nm, respectively. The presence of well-defined low-energy basins suggests preferential sampling of relatively restricted and energetically favorable conformational states, consistent with previous RMSD–Rg-based FEL analyses of peptide–enzyme complexes (Wang et al., 2026).

Fig. 7.

Two-dimensional (2D) and three-dimensional (3D) free-energy landscapes of the α-G–FLPYPYY complex (A), α-G–LRF complex (B), DPP-IV–FLPYPYY complex (C), and XO–LRF complex (D) during molecular dynamics simulations. PC1 and PC2 represent RMSD and Rg, respectively, and the color gradient represents the relative free-energy distribution.

MM/PBSA analysis was subsequently performed using 100 conformations extracted from the 90–100 ns trajectory interval. The calculated binding free energies of the α-G–FLPYPYY, α-G–LRF, DPP-IV–FLPYPYY, and XO–LRF complexes were −4.16, −4.74, −23.79, and −21.46 kcal mol⁻¹, respectively (Fig. 8A). The negative values indicate energetically favorable peptide–enzyme association. Energy decomposition showed that van der Waals and electrostatic interactions generally favored complex formation, whereas polar solvation opposed binding, consistent with previous studies of α-G- and XO-inhibitory peptides (Li et al., 2025; Wang et al., 2026). Because MM/PBSA values depend on the receptor–ligand system and computational protocol, these absolute energies should not be interpreted as a direct quantitative ranking of inhibitory potency across different enzymes.

Fig. 8.

MM/PBSA binding free-energy analysis of the α-G–FLPYPYY, α-G–LRF, DPP-IV–FLPYPYY, and XO–LRF complexes. Energy-component decomposition of the peptide–enzyme binding free energy (A); per-residue binding free-energy decomposition showing the energetic contributions of key residues in the corresponding enzyme–peptide complexes (B).

Per-residue free-energy decomposition identified several residues with substantial energetic contributions to peptide association (Fig. 8B). Major contributors included Phe735, Gly766, Tyr636, and Pro736 in α-G–FLPYPYY; Tyr733, Pro676, Gly766, Leu640, Arg653, Arg643, Tyr636, Thr639, and Lys765 in α-G–LRF; Arg560, Val558, Ser458, Tyr456, and Ser473 in DPP-IV–FLPYPYY; and Ile153, Pro95, Phe165, Asn163, Met157, Val164, Phe99, and Val162 in XO–LRF. These residues may contribute to maintaining the corresponding enzyme–peptide associations.

Collectively, the MD, FEL, and MM/PBSA results provide complementary structural and energetic evidence that FLPYPYY and LRF can form dynamically stable associations with their respective enzyme targets. Nevertheless, these computational analyses cannot independently establish enzyme inhibition or a definitive inhibitory mechanism. Further validation using synthesized peptides, concentration–response assays, IC₅₀ determination, and enzyme-kinetic or structural experiments is still required.

4. Conclusions

Goat milk proteins were enzymatically hydrolyzed to obtain hydrolysates with in vitro inhibitory activities against α-glucosidase, DPP-IV, and xanthine oxidase. A papain–neutral protease system was selected and optimized using response surface methodology, producing a hydrolysate with enhanced enzyme-inhibitory activities. Ultrafiltration and preparative chromatography enabled enrichment of the active peptide fraction, from which candidate peptides were identified by LC–MS/MS. LRF and FLPYPYY were screened as potential multifunctional enzyme-inhibitory peptides through bioinformatic analysis and molecular docking. Molecular dynamics simulations and MM/PBSA calculations further supported the stability and favorable interactions of the peptide–enzyme complexes. These findings indicate that goat milk hydrolysates represent a promising source of potential multifunctional peptides. However, the individual bioactivities, bioavailability, safety, and physiological effects of the identified peptides require further experimental validation before functional-food applications can be established.

CRediT authorship contribution statement

Xinru Yu: Writing – original draft, Methodology. Guowei Shu: Writing – review & editing, Supervision, Funding acquisition, Conceptualization. Wenjing Hu: Formal analysis. Ge Zhang: Project administration, Funding acquisition. Pian Xie: Methodology, Funding acquisition. Kai Song: Resources, Project administration, Investigation, Funding acquisition. Guoliang Li: Supervision, Investigation, Funding acquisition, Data curation.

Declaration of competing interest

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

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.