Novel dual-taste umami and salty peptides from goat whey: Mechanism and interaction with taste receptors

Abstract

Goat whey was analyzed using machine learning and peptidomics to identify umami and salty peptides. Of the 42 candidates tested, four peptides (AD-7, GA-8, PL-9 and GQ-10) exhibited both umami and salty characteristics with low taste thresholds (0.20–0.25 mmol/L). Molecular docking revealed that these peptides bind to the T1R1/T1R3 umami receptor via hydrogen bonds involving key residues such as ALA176, ASN150, LYS155, GLN222, Asp219 and LYS5, with acidic residues enhancing umami perception. For saltiness, they bind to TMC4 through electrostatic and hydrogen bonding networks and to TRPV1 through hydrophobic interactions involving Leu, Phe, and Met. Molecular dynamics simulations confirmed stable complex formation with all three receptors. Non-targeted metabolomics identified 199 metabolites, including betaine, contributing to savoury flavor. These findings establish goat whey as a sustainable source of natural taste-active compounds and provide a scientific basis for its utilisation in flavourings, thus reducing waste and the environmental impact of cheese production.

Highlights

• •The discovery of dual-tasteumami-salty peptides from goat whey.
• •The key umami mechanism is binding to T1R1/T1R3 via ALA176, ASN150, LYS155, GLN222, Asp219 and LYS5
• •Peptides bind TMC4 via electrostatic interactions and H-bonds; Leu, Phe, and Met bind TRPV1 to mediate salty taste.
• •It provides a molecular foundation for natural flavor enhancers and salt replacers.

1. Introduction

In recent years, the food industry has seen a growing demand for natural salty and umami flavor compounds (Ramesh, Venkatappa, & Bhat, 2025). Goat whey is a key by-product of cheese production with a global annual output of 180–190 million tons (Kurnick, Michellim, Yada, Leite Junior, & Tribst, 2024; Pires, Marnotes, Rubio, Garcia, & Pereira, 2021). It contains 3.5–3.8% high protein and 4.6% lactose and is also rich in minerals such as calcium and phosphorus (Hejtmánková et al., 2012). At present, most of these high-nutrient components are downgraded for use as feed or fertilizer (Uribarrena et al., 2024). Goat whey is rich in A2-type β-casein, lactoferrin, and immunoglobulin (Jung et al., 2017; Soloshenko, Lych, Voloshyna, & Shkotova, 2020). It offers unique advantages for releasing peptides with functional and flavor properties. However, current research on whey peptides focuses on bitter and bioactive peptides. For instance, Sebald et al. employed QQQ-MS and SWATH-MS techniques to identify bitter peptides in cheese (Sebald, Dunkel, & Hofmann, 2020). In contrast, Estévez et al. employed hydrolysis to produce whey peptides with ACE and antioxidant activities. These peptides were added to yoghurt and their acceptability and stability tested (Estévez, Fucinos, Rodríguez-Sanz, & Rúa, 2022). At present, whey protein-derived bioactive peptides have been explored. But the potential to release peptides with both functional and flavor properties remains largely unexplored.

Saltiness is traditionally provided by inorganic salts like sodium chloride, and umami by glutamate. Recent research points out peptides are linked amino acids that can make umami, sweet, sour, bitter, or salty tastes (Song, Zhang, et al., 2023; Xie et al., 2024). The significant correlation between the salty and umami characteristics of these peptides (Zhao et al., 2024), rooted in their synergistic perception (Song et al., 2024), makes them ideal for enhancing saltiness perception in low-sodium foods (Xie et al., 2024). Consequently, the pursuit of dual-taste peptides exhibiting concurrent salty and umami tastes has become a major research drive (Chang et al., 2023; Zhang et al., 2022). Studies have shown that peptides from various sources are being successfully utilized. For example, yeast extract peptides is used to reduce sodium in savoury products like low-sodium soy sauce and soups (Zheng et al., 2021). The peptide QPGDY, which is isolated from peanut protein, enhances saltiness. Others, such as those found in Jinhua ham and clams, contribute to the umami flavor. (Diepeveen, Moerdijk-Poortvliet, & van der Leij, 2022; Ji et al., 2025; Wang et al., 2025). Moreover, chicken-derived umami peptides have been demonstrated to possess saltiness-enhancing properties alongside ACE inhibitory activity, highlighting the potential of dual-taste peptides in food applications (Zhang et al., 2025). However, most findings are based on animal proteins and plant proteins, while umami peptides from goat whey are unexplored.

Molecular docking has emerged as a powerful tool for investigating ligand-receptor interactions and elucidating the molecular mechanisms underlying taste perception (Feng et al., 2024). This approach has been widely used to characterize the potential interactions of peptides with the umami receptor T1R1/T1R3 and to salt taste receptors, including TMC4 and TRPV1. For example, umami active peptides derived from fermented wheat germ have been shown to interact with the Venus flytrap domain of T1R1/T1R3, providing molecular insights into their umami perception (Jin et al., 2025). Salt taste perception involves multiple physiological pathways, including an amiloride-sensitive pathway associated with the highly selective epithelial sodium channel (ENaC) and amiloride-insensitive mechanisms involving other ion channels and salt-sensing pathways. Among these, TRPV1 has been implicated in amiloride-insensitive salt sensing, whereas TMC4 functions as a voltage-dependent chloride channel involved in chloride-associated salt taste perception (Le et al., 2022). In humans, salt perception is predominantly mediated by amiloride-insensitive mechanisms, positioning these pathways as primary targets for effective sodium reduction strategies. Recent studies have identified several salt-enhancing peptides, such as DP, DR, KSWDDFFTR, WGFGDD, DWELMP, and ILAPPER, which augment saltiness perception via specific interactions with the TMC4 and TRPV1 receptor (Askar et al., 2025; Li et al., 2024; Wang et al., 2025; Yu et al., 2025). However, no conformational changes of the protein and ligand were observed during the molecular docking process. Therefore, molecular dynamics simulations are required to obtain atomic trajectories on spatial and temporal scales, providing detailed information on protein conformational changes and fluctuations (Wang et al., 2025). In contrast, the structure activity relationships governing umami-salt dual-taste peptides derived from whey proteins remain inadequately characterized. Therefore, integrating molecular docking with molecular dynamics simulations may provide deeper insights into the potential interactions of these peptides with the umami receptor T1R1/T1R3 and the salt taste-related ion channels TMC4 and TRPV1, thereby helping to elucidate the molecular basis underlying their dual-taste properties.

This study focused on salty-umami goat whey-derived peptides, using peptidomics to identify their sequences and clarify their structural basis. The electronic tongue was used to evaluate their taste intensity and synergistic effects, and molecular docking to analyze the potential binding of the peptides with T1R1/T1R3 and TRPV1 taste receptors. This integrated approach provides theoretical insights into the flavor presentation and potential receptor interaction mechanisms of salty-umami peptides from goat whey, supporting their future application in low-sodium food products.

2. Materials and methods

2.1. Materials and chemicals

Goat whey was purchased from a local market in Jinning District, Kunming City, Yunnan Province, China. All samples were collected under standard storage conditions and stored at 4 °C prior to analysis. Methanol, acetonitrile, and isopropanol (chromatography grade) were purchased from CNW Technologies (Shanghai, China). Sugar, monosodium glutamate, NaCl, isoleucine, and citric acid solution were of food grade. All other chemicals and solvents used for analysis were of analytical grade.

2.2. Sample preparation

Following the method of Su et al. (2012) with slight modifications, goat whey was placed in 10 kDa and 3 kDa ultrafiltration tubes and fractionated using an HC-3018R centrifuge (Anhui Zhongke Zhongjia, China). Centrifugation conditions were as follows: speed, 4000 rpm; time, 10 min; and temperature, 4 °C. Three fractions were obtained: U1 (> 10 kDa), U2 (3–10 kDa), and U3 (< 3 kDa). All three fractions were collected and subjected to sensory analysis. The fraction with the strongest umami taste was selected for further analysis.

2.3. UPLC-MS analysis with metabolite analysis

The sample was thawed and subjected to metabolite extraction using 400 μL of pre-cooled methanol and acetonitrile (1:1, v/v), containing isotope internal standards. Following vortexing, sonication, incubation at −40 °C and centrifugation, the resulting supernatant was collected for UPLC-MS/MS analysis. Chromatography was performed on a 2.1 × 50 mm, 1.7 μm ACQUITY UPLC BEH Amide column using a Vanquish UHPLC system. The mobile phases were (A) 25 mmol/L ammonium acetate and 25 mmol/L ammonia in water and (B) acetonitrile. The column temperature was 4 °C and the injection volume was 2 μL. Mass spectrometry was conducted using an Orbitrap Exploris 120 instrument (Thermo Fisher Scientific), applying a spray voltage of 3.8 kV in positive mode and 3.4 kV in negative mode. Full MS and MS/MS resolutions were set to 60,000 and 15,000 respectively, with a stepped NCE of 20/30/40. Raw data were converted to mzXML format using ProteoWizard and processed with a custom R package for metabolite identification against the Biotree DB (V3.0) database.

2.4. Identification of peptides by nano-LC-MS/MS

200 μg of total peptides were separated and analyzed using a nano-UPLC (Evosep One) coupled to a timsTOF Pro2 mass spectrometer (Bruker, Germany) with a nano-electrospray ion source. Separation was performed on a reversed phase column (PePSep C18, 1.9 μm, 150 μm × 15 cm, Bruker, Germany). Mobile phases consisted of water containing 0.1% formic acid (phase A) and acetonitrile containing 0.1% formic acid (phase B). Data acquisition was carried out in data-dependent acquisition (DDA) PASEF mode. The full MS (MS1) scan covered an m/z range of 100–1700 with a resolution of 60,000 (Orbitrap). For MS/MS analysis, the top 4 precursor ions per cycle were selected for fragmentation with an isolation window of 2.0 m/z. The MS2 resolution was set to 15,000 (Orbitrap). The automatic gain control (AGC) target was 200% (normalized), and the maximum injection time was 50 ms. Dynamic exclusion was enabled to avoid repeated selection of previously fragmented ions. During PASEF MS/MS scanning, the collision energy was increased linearly with ion mobility, from 20 eV (1/K₀ = 0.6 Vs/cm²) to 59 eV (1/K₀ = 1.6 Vs/cm²). Only precursor ions with charge states of 2+ to 6+ were considered for fragmentation.

Raw MS files were processed using SpectroMine software (4.2.230428.52329) and the built-in Pulsar search engine. MS spectra lists were searched against their species-level UniProt FASTA databases (uniprot-Capra_hircus_9925_2023_08.fasta), Carbamidomethyl [C] as a fixed modification, Oxidation (M) and Acetyl (Protein N-term) as variable modifications. No-Enzyme (Unspecific) was used as enzyme. The false discovery rate (FDR) was set to 0.01 for both PSM and peptide levels. Peptide identification was performed with an initial precursor mass deviation of up to 20 ppm and a fragment mass deviation of 20 ppm. All the other parameters were reserved as default.

2.5. Virtual screening of peptides

Following the method of Xu et al. (2024), goat whey was screened for potential taste-active peptides. Screening was performed using multiple predictive models, including Peptide Ranker (http://distilldeep.ucd.ie/PeptideRanker/, PeptideRanker value exceeding 0.5), UMPred-FRL (http://pmlabstack.pythonanywhere.com/UMPredFRL), Umami_YYDS (https://github.com/SynchronyML/Umami_YYDS/blob/main/README.md), and TastePeptides-Meta (http://www.tastepeptides-meta.com/Umami_IP). These models were used for joint prediction, and further evaluation was conducted to assess the relationship between peptide characteristics and umami flavor (Niu et al., 2024). Toxicity is a critical safety concern in peptide development. Good water solubility, stability, and non-allergenicity are essential for normal metabolism (Gu et al., 2024). Computational methods were used to predict peptide toxicity, sensitivity, solubility, and stability. Specifically, the following tools were employed: ToxinPred (https://webs.iiitd.edu.in/raghava/toxinpred/), AllerTOP v.2.0 (http://www.ddg-pharmfac.net/AllerTOP), Innovagen tool (http://www.innovagen.com/), and Expasy-ProtParam (https://web.expasy.org/protparam/).

2.6. Synthetic peptides

Six taste peptides, including GAKDMWRA (GA-8), SGKDPNHF (SG-8), PLFKGMTRP (PL-9), ADKYFHA (AD-7), GQGAKDMWRA (GQ-10), and AGQGAKDMWR (AG-10), were synthesized. Peptide synthesis was carried out by Anhui Guoping Pharmaceutical Co., Ltd. (Hefei, China) using Fmoc solid-phase peptide synthesis (FMOC-SPPS). The purity and molecular weight of the peptides were determined by high-performance liquid chromatography (HPLC) and mass spectrometry (MS). Each peptide had a purity of 95%.

2.7. Sensory evaluation and electronic tongue

2.7.1. Panel training

To systematically evaluate the gustatory contributions of goat whey peptides, we adopted the sensory descriptive analysis protocol established by Yang et al. (2025). At Yunnan Agricultural University, sensory experiments do not require approval from the Human Body Ethics Committee. Nevertheless, appropriate protocols were strictly followed to protect participants' rights and privacy. Informed consent was signed by all participants, who acknowledged the potential risks of the experiment. Participation was voluntary, and all participants had the right to withdraw at any time. The sensory panel consisted of 10 healthy volunteers (5 males and 5 females, aged 18–25 years) with no history of smoking or olfactory impairment. Panelists underwent training to accurately identify the five basic tastes: sour, sweet, bitter, salty, and umami. Included sweet sucrose solution (10 mg/mL), umami monosodium glutamate solution (3.5 mg/mL), salty NaCl solution (3.5 mg/mL), bitter isoleucine solution (2.5 mg/mL), and sour citric acid solution (0.8 mg/mL). The concentration of the standard solution was determined to be 5 points, with a maximum score of 10 points and a minimum score of 0 points. After training, panelists were required to accurately describe the five solutions. Sensory evaluations were carried out in the sensory analysis laboratory. The laboratory was maintained at 25 ± 2 °C, and the average of three repeated tests was recorded as the evaluation result. Samples were dissolved in ultrapure water, transferred to plastic containers, and assigned random numerical codes to ensure evaluation objectivity. Panelists rinsed their mouths with 50–60 mL ultrapure water at least twice between samples to minimize sensory fatigue and carry-over effects.

2.7.2. Sensory evaluation of different fractions

Sensory evaluation of goat whey fractions was carried out by dissolving U1, U2, and U3 with water to a final concentration of 5 mg/mL. The supernatant obtained after centrifugation (8000 rpm, 15 min, 4 °C) was used for sensory analysis. Each fraction was evaluated three times.

2.7.3. Descriptive sensory analysis and umami thresholds of synthetic peptides

Each peptide was dissolved in water at 1 mg/mL. Taste dilution analysis (TDA) was performed by gradually diluting the solution until it could no longer be distinguished from the blank control. The highest dilution at which panelists could accurately identify the taste was defined as the recognition threshold. Panelists evaluated the sensory properties of peptide solutions, including sour, sweet, bitter, salty, and umami.

2.7.4. Electronic tongue

The taste profiles of synthetic peptides were evaluated using an SA402B electronic tongue (Intelligent Sensor Technology, Inc., Japan). Each peptide was dissolved in ultrapure water at a concentration of 0.5 mg/mL. Following a standardized cleaning and equilibration protocol using the provided solutions, measurements were conducted with five flavor sensors (C00, AE1, CA0, CT0, AAE). Each sample was analyzed in four cycles, with the first cycle discarded and the mean of the subsequent three cycles used for data reporting.

2.7.5. Taste-enhancing effect analysis

To clearly evaluate the intrinsic taste and its enhancing effects, we performed QDA and TI analyses. Following the method of Liu, Xu, Li, Shen, and Liu (2025) and Chen et al. (2025). To evaluate the saltiness and umami-enhancing effects of the peptides, quantitative descriptive analysis (QDA) was employed. MSG solutions at 3.00, 5.00, and 7.00 mg/mL were designated as umami benchmarks, with intensities of 5, 10, and 15 points, respectively. Four peptides (1.00 mg/mL) were dissolved in a 3 mg/mL MSG solution for the QDA test. For saltiness assessment, each peptide was dissolved at a concentration of 1 g/L in a 0.5% NaCl solution for the QDA. Reference NaCl solutions of 0.5%, 1.0%, and 1.5% were prepared, corresponding to saltiness intensity scores of 5, 10, and 15, respectively. Assessors sampled 5 mL for 5 s before expectorating, and the taste-enhancing effect was quanti­fied based on perceived intensity. Using TI under the same conditions, the dynamic interactions of four taste-enhancing peptides (GQ-10, GA-8, AD-7, and PL-9) with both MSG and NaCl were further explored. Taste intensity was recorded at 5, 10, 20, 30, 45,60, 75, 90, 120, 150, and 180 s, continuing until umami sensation dis­appeared. A 20-min break was enforced between tastings to minimize fatigue, with water rinsing allowed. Each sample was tested in a ran­domized order and replicated three times.

2.8. Molecular docking of the screened peptides with taste receptors

Amino acid sequences for umami receptors T1R1 (Q7RTX1) and T1R3 (Q7RTXO) were retrieved from UniProtKB. Homology modeling was performed using Swiss-Model with PDB ID 1ewk as template (T1R1: 34.34% identity; T1R3: 33.55% identity). Models were aligned in PyMOL 2.5.41 (DeLano, 2002), transmembrane domains removed, and the T1R1/T1R3 heterodimer was generated. Structures were optimized using Protein Preparation Wizard (Maestro 13.0): hydrogen atoms added, disulfide bonds created, missing residues restored via Prime, hydrogen bonds optimized, and protonation states calculated at pH 7.4 (OPLS4 force field). Global energy minimization removed clashes. Quality was assessed via Ramachandran plot.

The crystal structure of the TMC4 protein used for docking was obtained from the AlphaFold database, and the 3D structure of the peptide was constructed using PyMol 2.5.2. In this study, molecular docking was performed using AutoDock Vina 1.2.3. Prior to docking, the receptor protein was processed using PyMol 2.5.2, which involved removing water molecules, salt ions, and small molecules. Next, set up the docking box so that it encloses the entire protein. In addition, use ADFRsuite 1.0 to convert all preprocessed small molecules and receptor proteins into the PDBQT format required for docking with AutoDock Vina 1.2.3. When integrating, set the global search depth to 32 and keep the other parameters at their default settings. We consider the docking conformation with the highest score to be the binding conformation, and finally use PyMol 2.5.2 to visualize and analyze the docking results.

TRPV1 crystal structure (PDB ID 5IS0) was downloaded from PDB. Peptide structures were generated in PyMOL 2.5.4 and minimized (MMFF94 force field). Water, ions, and ligands were removed from the receptor. The binding pocket was defined by setting a docking grid to encompass residues within 15 Å of the receptor's active site. Both ligands and receptor were converted to PDBQT format using ADFRsuite 1.0 (Meng et al., 2024). AutoDock Vina 1.2.3(Trott et al., 2010). was used for docking with global search exhaustiveness set to 32. The highest scored docking pose was designated as the binding conformation and visualized in PyMOL 2.5.4 and LigPlot+(Wallace, Laskowski, & Thornton, 1995).

2.9. MDS of the screened peptides with taste receptors

All-atom MD simulations were conducted using AMBER 24 software with the ff14SB force field. Systems were prepared using the LEaP module with truncated octahedral TIP3P water boxes (10 Å cutoff) and charge-neutralizing Na⁺/Cl⁻ ions. Energy minimization comprised 2500 steps of steepest descent and 2500 steps of conjugate gradient minimization. Heating (0–298.15 K, 200 ps, constant volume), NVT equilibration (500 ps), and NPT equilibration (500 ps) preceded 100 ns production runs under periodic boundary conditions. Nonbonded interactions were truncated at 10 Å, with PME for long-range electrostatics. SHAKE constraints were applied to hydrogen bonds. Langevin dynamics (γ = 2 ps⁻¹) maintained temperature at 298.15 K. Pressure was set to 1 atm with 2 fs integration steps. Trajectories were saved every 20 ps.

2.10. Statistical analysis

All experiments were repeated three times. Results are expressed as mean ± standard deviation (n = 3). Data were analyzed using one-way ANOVA with IBM SPSS Statistics version 20.0. The statistical significance level was set at P < 0.05.

3. Results and discussion

3.1. Sensory characteristics of ultrafiltration fractions

In this study, three fractions (>10 kDa, 10–3 kDa, and < 3 kDa) were obtained by ultrafiltration and analyzed by sensory evaluation. As shown in Fig. 1A, U3 had the strongest umami taste (5.4 points), followed by U2 (3.5 points), while U1 had the weakest intensity. For salty taste, U3 (2.9 points) was significantly higher than U1 (1.5 points) (P < 0.05). In summary, among the three fractions, U3 had the strongest umami characteristics, while U1 had the weakest. These results are consistent with Cui, Li, Wu, and Hu (2023), who reported that low-molecular-weight compounds are key contributors to umami flavor. In addition, U3 had significantly stronger umami and saltiness compared to U2 (P < 0.05). Therefore, U3 was selected for further analysis.

Fig. 1.

Sensory evaluation and molecular characterization of goat whey ultrafiltration fractions and peptides. (A) Sensory radar of UF fractions (Three fractions were obtained: U1 (> 10 kDa), U2 (10–3 kDa), and U3 (< 3 kDa)). (B) Non-volatile metabolites in U3. (C) LC-MS sequencing of goat whey peptides. (D) Statistical results of relative molecular masses of peptide segments in goat whey. (E) Peptide-length distribution. (F) N/C-terminal residue composition.

3.2. UPLC-MS analysis with metabolite analysis

This study used LC-MS/MS-based non-targeted metabolomics to profile umami and salty compounds in goat whey. Preliminary sensory evaluation indicated that the <3 kDa ultrafiltration fraction of goat whey exhibited the most pronounced umami characteristics. This fraction was selected for in-depth analysis. A total of 173 metabolites were identified (Fig. 1B). Organic acids and their derivatives (34.1%), amino acids and their derivatives (28.9%), and sugars and their derivatives (12.7%) were the most abundant classes. The high proportion of organic acids is attributable to acid hydrolysis of lactose and proteins in whey, which generates new carboxyl-containing compounds (León-López, Pérez-Marroquín, Estrada-Fernández, & Campos-Lozada, 2022), as well as to the conversion of lactose into lactic acid by indigenous lactic acid bacteria (Frankowski, Miracle, & Drake, 2014). As detailed in (Table S1), Key organic acids in goat whey, including citric acid (0.0003 ± 0), malic acid (0.0173 ± 0.0025) and lactate(0.4278 ± 0.0343), reportedly synergize with inorganic ions and umami and sweet amino acids to enhance overall umami perception. (Jin et al., 2025; Liu et al., 2024). Amino acids and their derivatives such as proline (0.0005 ± 0.0001), and theanine (0.0002 ± 0), interact with taste receptors through specific functional groups and are pivotal to umami formation (Huang, Duan, Wang, Xiao, & Zhang, 2019; Kurihara, 2009; Rotzoll, Dunkel, & Hofmann, 2005; Wang et al., 2022). Beyond these amino acids, various dipeptides in non-volatile metabolites have also been reported as key taste modulators. For example, Leu-Leu and Ile-Leu significantly enhance umami thickness and continuity through stable binding to the calcium-ensing receptor (Wu, Lin, Cui, & Li, 2025). Arg-Ala exhibits a clear salt-enhancing effect by increasing the response frequency of taste bud cells to NaCl (Xu et al., 2017). Although nucleotides and analogues (10.4%), alkaloids (4.6%), vitamins and coenzymes (3.5%), lipids (4.6%) and phenolic acids (1.2%) were present at lower levels, they contributed distinct flavors, For example, betaine(0.1876 ± 0.0032), a major alkaloid, imparts the characteristic taste of seafood (Tu, Wu, Wang, & Shi, 2020), while lipids may prolong umami persistence (Gao et al., 2021). Collectively, these findings demonstrate that goat whey is a rich source of umami- and salty-tasting compounds, underscoring its potential for high value food applications.

3.3. Identification of peptides in goat whey

Statistical analysis of peptide molecular masses in goat whey is shown in (Fig. 1). As bioactive small molecules, umami peptides contribute various taste characteristics and significantly enhance umami and salty taste perception (Fan et al., 2025). Mass spectrometric profiling of goat whey showed 130 proteins and 2383 peptides (Fig. 1C). Molecular mass profiling revealed pronounced enrichment of low-mass peptides: 63.11% of sequences were < 1.5 kDa (Fig. 1D). Chain length analysis indicated that 44.40% (1058 peptides) contained 7–10 residues, 48.80% (1163 peptides) contained 11–15 residues, and only 6.8% exceeded 15 residues (Fig. 1E). These results highlight the dominance of short peptides (≤10 residues) as precursors for umami active and salt-active sequences. Sequence feature mapping showed that hydrophobic residues (Leu, Val, Ala) were highly enriched at the N-terminus, while Lys dominated the C-terminus (Fig. 1F). Overall, 493 peptides contained cationic (Lys, Leu, Thr, Arg) at the C-terminus, and 65 peptides contained N-terminal Glu, conferring a zwitterionic characteristic that may enhance electrostatic interactions with the T1R1/T1R3 receptor. Notably, 36.2% of all peptides contained Lys or Leu at the terminus. Such a motif has been reported to enhance both salty perception by activating G-protein-coupled signaling (Yang et al., 2024). Collectively, these ≤10-residue, low-molecular-mass peptides constitute a rich source for the discovery of dual-taste umami-salty peptides in goat whey.

3.4. Virtual screening of potential dual-taste umami and salty peptides

Based on the above analysis, a comprehensive screening strategy combining short peptides (≤10 amino acids, molecular weight < 1500 Da) with multiple umami peptide prediction databases was employed for integrated identification (Fig. 2). The PeptideRanker score was used as an indicator to predict the biological activity of peptides. Peptides with scores >0.5 are generally considered potential bioactive peptides. Based on this scores (PeptideRanker >0.5), 242 peptides with fewer than 10 amino acids were initially identified. To enhance screening reliability and reduce the bias of a single model, multiple umami peptide prediction models were used. Confidence in potential umami peptide identification was improved by intersecting results across models (Niu et al., 2024). Therefore, integrated prediction models (UMPred-FRL, Umami-YYDS, and Taste Peptides-Meta) were applied to screen the 242 peptides, yielding a total of 42 potential umami peptides. Non-allergenicity, safety, non-toxicity, good water solubility, and stability are prerequisites for the practical application of food derived peptides (Li et al., 2024). These indicators were used to evaluate the 42 potential umami peptides. Finally, six peptides were identified from 42 peptides through this strict multi-step screening process. These peptides included GAKDMWRA (GA-8, Gly-Ala-Lys-Asp-Met-Trp-Arg-Ala), SGKDPNHF (SG-8, Ser-Gly-Lys-Asp-Pro-Asn-His-Phe), PLFKGMTRP (PL-9, Pro-Leu-Phe-Lys-Gly-Met-Thr-Arg-Pro), ADKYFHA (AD-7, Ala-Asp-Lys-Tyr-Phe-His-Ala), GQGAKDMWRA (GQ-10, Gly-Gln-Gly-Ala-Lys-Asp-Met-Trp-Arg-Ala), and AGQGAKDMWR (AG-10, Ala-Gly-Gln-Gly-Ala-Lys-Asp-Met-Trp-Arg). As shown in Table S2, six peptides had PeptideRanker, UMPred-FRL, and Umami_YYDS scores exceeding 0.5. They were high water solubility, non-toxic, and non-allergenic; thus, they qualify as prime candidates for subsequent synthesis and validation experiments.

Fig. 2.

Virtual Screening of Peptides Flowchart.

3.5. Sensory evaluations

3.5.1. Taste characteristics of dual-taste umami and salty peptides

To verify the sources of umami and salty tastes in goat whey, we conducted sensory evaluations on selected peptides. Six peptides, including PL-9, AD-7, GA-8, SG-8, GQ-10, and AG-10. Previous studies have shown that most flavor-enhancing peptides exhibit multiple flavor attributes, resulting in complex taste profiles (Yang et al., 2024). Consistent with these findings, our sensory analysis revealed that all six synthetic peptides possessed multiple flavor attributes; their sensory scores are detailed in Fig. 3A, Compared to the standard solution (5 points), AD-7 has the highest umami intensity (4.2 points), followed by PL-9 (3.8 points), GQ-10 (2.9 points), GA-8 and AG-10 (both 2.8 points), and SG-8 (2.2 points). The sweetness intensity of GA-8 (2.5 points) was significantly higher than the other five peptides (P < 0.05). In terms of sweetness intensity, peptides PL-9 (1.9 points) and AD-7 (1.8 points) scored marginally higher than the other three, but there was no statistical difference (P > 0.05). AD-7 (3.0 points) and PL-9 (2.1 points) were significantly saltier than the others (P < 0.05). Notably, the molecular weights of all six peptides were lower than 1500 Da, supporting the view that short-chain peptides make significant contributions to umami taste (Xiang et al., 2025). The taste threshold of the six peptides was determined using threshold discrimination analysis (TDA). The detection thresholds for the six peptides ranged from 0.20 to 0.53 mmol/L, which were comparable to the previously reported thresholds for yeast protein hydrolysates (0.13–0.50 mmol/L) (Ya et al., 2024) and for salty peptides derived from porcine bone collagen (0.10–0.32 mmol/L) (Chen et al., 2025). Notably, GA-8, PL-9, AD-7, and GQ-10 exhibited relatively lower detection thresholds (0.2–0.25), indicating enhanced taste perception sensitivity of these four peptides. In summary, the four peptides GA-8, PL-9, AD-7, and GQ-10 exhibited strong umami and salty taste characteristics.

Fig. 3.

Sensory analysis (A) Sensory evaluation of synthetic peptides. (B) Electronic tongue analysis. (C) Umami-enhancing characteristics of four umami peptides. (D) The time-intensity curve of four umami-enhancing peptides. (E) The time-intensity curve of four Salty-enhancing peptides.

3.5.2. Determination of four dual-taste umami and salty peptides electronic tongue

To validate the sensory analysis, the taste profiles of four synthetic peptides (GA-8, PL-9, AD-7 and GQ-10), which have low thresholds and pronounced umami-salty characteristics, were assessed using an electronic tongue (Table S3). All tested peptides exhibited umami taste, as illustrated in Fig. 3B, AD-7 demonstrated the highest umami intensity (4.02 points), followed by GQ-10 (3.40 points) and PL-9 (3.31 points); GA-8 demonstrated the weakest intensity (2.39 points). The superior umami intensity of AD-7 is likely attributable to its lower molecular weight, which may facilitate receptor binding, consistent with previous findings (Dang, Gao, Ma, & Wu, 2015). Notably, PL-9 exhibited a distinct umami taste despite its lack of typical umami amino acids, suggesting that its spatial conformation and specific interactions with taste receptors may determine the perceived intensity (Pan et al., 2025).

For salty taste, AD-7 (6.35 points) had the highest score, followed by GQ-10 (5.00 points), and GA-8 (−0.4 points; lowest). The strong salty taste of AD-7 may be due to its C-terminal aspartic acid which has been reported as a key feature of salty peptides (Gao et al., 2024). The saltiness of GQ-10 may be due to the synergistic effects, including bitterness inhibition by N-terminal glycine and the saltiness potential of its acidic amino acids. PL-9 and GQ-10 showed strong bitterness in the electronic tongue results. However, sensory evaluation indicated that all four peptides had a slight bitter taste. This discrepancy may be due to partial inhibition of bitterness during sensory evaluation, a phenomenon undetectable by the electronic tongue. As a result, the latter had higher bitterness intensity. Such differences between electronic tongue and sensory evaluation results are consistent with previous reports (Feng et al., 2024). The pronounced bitterness of PL-9 detected by the electronic tongue can be attributed to its structural characteristics. PL-9 comprises 66.7% hydrophobic residues, including an N-terminal Pro-Leu-Phe cluster and a C-terminal Pro residue, which facilitate strong interactions with the hydrophobic sensitive sensors of the electronic tongue (Xiang, Xia, Fang, & Zhong, 2024). Conversely, the lower perceived bitterness during sensory evaluation may be explained by the conformational adaptation of hydrophobic bitter residues, which appears to enhance saltiness perception while simultaneously reducing bitter receptor activation (Liu et al., 2020). This effect may be further compounded by cross modal masking from the concurrent presence of umami and salty stimuli in the evaluation matrix. Because the electronic tongue cannot fully evaluate taste interactions and sensory evaluation is susceptible to external factors and individual variability, combining both methods provides a more robust scientific approach to taste analysis (Ismail, Hwang, & Joo, 2020).

3.5.3. Taste-enhancing effects

3.5.3.1. QDA analysis

To evaluate the umami and saltiness-enhancing effects of the four peptides (1 mg/mL) in MSG (3 mg/mL) and 5% NaCl solutions, respectively. All peptides significantly enhanced MSG solution's umami intensity (Fig. 3C). AD-7 showed the strongest umami enhancement, followed by GQ-10, PL-9, and GA-8. Notably, the peptides also enhanced NaCl's saltiness, but with different ranking: GQ-10 exhibited the strongest saltiness enhancement, followed by AD-7, GA-8, and PL-9, indicating selective taste receptor recognition among the four peptides.

3.5.3.2. TI analysis

Fig. 3D, shows the temporal intensity (TI) profiles of umami for MSG control (3 mg/mL) and peptide-MSG complexes. The peptide supplemented solutions exhibited rapid ascent to peak intensity followed by gradual decline, enhancing umami perception. As shown in Table S4, the maximum intensity (I_MAX) values were: control 5.67, AD-7 11.75, PL-9 9.25, GA-8 8.42, and GQ-10 10.08. The peptides also extended umami duration from 75 s (control) to 100 s (PL-9, GA-8) and 110 s (AD-7, GQ-10). Fig. 3E, presents the TI profiles of saltiness for 5% NaCl control and peptide-NaCl complexes. Similar kinetic patterns were observed, with I_MAX values of: control 5.75, AD-7 10.08, PL-9 8.17, GA-8 7.75, and GQ-10 11.75. The peptides significantly extended saltiness duration from 30 s (control) to 45 s (GA-8), 60 s (AD-7, PL-9), and 75 s (GQ-10), demonstrating potent flavor-enhancing properties.

3.6. Molecular docking

3.6.1. Molecular docking of peptides with T1R1/T1R3 taste receptors

The umami receptor T1R1/T1R3 is a heterodimer with a Venus flytrap (VFT) binding domain, to which umami compounds bind to elicit umami taste (Belloir et al., 2017). Low docking energy indicates a stable conformation and a higher possibility of binding (Feng et al., 2024). Therefore, T1R1/T1R3 was used as the receptor for docking to further explore the umami mechanisms of selected peptides. Since the crystal structure of the T1R1/T1R3 complex has not been reported, the complex was generated using homology modeling, protein superposition, and structural optimization (Dong, Wan, Huang, Xu, & Lei, 2023). Ramachandran plot analysis (Fig. 4A2) showed that most amino acids (> 95%) were in reasonable regions, and only 5 amino acids were in the disallowed region. The data showed that the model had good quality and was suitable for subsequent molecular docking.

As shown in Table 1, summarizes the docking energies of the four peptides with T1R1/T1R3. Lower values indicate stronger binding. The binding strength could be ranked as follows: AD-7 (−8.779 kcal/mol) < GAK-8 (−8.671 kcal/mol) < PL-9 (−8.302 kcal/mol) < GQ-10 (−7.858 kcal/mol). It is worth noting that AD-7 had the lowest docking energy (−8.779 kcal/mol), which indicates that it has a stronger binding affinity with T1R1/T1R3 and a higher umami potential, consistent with the electronic tongue results. (Fig. 4B2-E2) presents a 2D docking map, the binding of peptides to the T1R1/T1R3 complex involved carbon‑hydrogen bonds, conventional hydrogen bonds, salt bridges, hydrophobic interactions (Pi-Pi, T-shaped, Pi-alkyl), and electrostatic interactions. The proportions of interaction types were as follows: hydrogen bonds: AD-7 (75.86%), GA-8 (64.74%), PL-9 (80.77%), and GQ-10 69.44%); hydrophobic forces: AD-7 (22.2%), GA-8 (25.81%), PL-9 (18.52%), and GQ-10 (5.71%); and electrostatic forces: AD-7 (6.9%), GA-8 (6.45%), and GQ-10 (11.43%). In keeping with other research, the peptide binds to the receptor mainly through hydrogen bond (Yu et al., 2025; Zhao et al., 2024). This is consistent with previous studies, showing that Ser, Asp, and Leu are key residues for the binding to T1R1/T1R3 (Feng et al., 2024). For GA-8, residues Lys155, Asn150, Thr179, Leu173, and Ala176 had the highest frequency. In contrast to AD-7, GA-8 contained fewer acidic amino acids. Asp219, most frequently observed in AD-7, is an acidic residue and has been reported to contribute to umami intensity (Shi et al., 2024)，Residues ASN150, LYS155, GLN222, and LYS5 were most frequent in PL-9. In GQ-10, residues TYR182, ASN150, LYS155, GLU172, and LYS5 were most frequent. For easier analysis, we identified residues that appeared more than three times as key binding sites. These residues are shown in Fig. 4F. Among these sites, ALA176, ASN150, LYS155, GLN222, and LYS5 showed the most frequent interactions. This indicates that these residues are critical for peptide binding.

Table 1.

Binding energies of synthetic peptides with the receptor.

Ligand_name	Target_name	Docking_score(kcal/mol)
AD-7	T1R1-T1R3	−8.779
GA-8	T1R1-T1R3	−8.671
PL-9	T1R1-T1R3	−8.302
GQ-10	T1R1-T1R3	−7.858
AD-7	TMC4	−6.844
GA-8	TMC4	−6.755
GQ-10	TMC4	−7.681
PL-9	TMC4	−6.863
AD-7	TRPV1	−6.416
GA-8	TRPV1	−6.566
PL-9	TRPV1	−6.023
GQ-10	TRPV1	−6.850

Fig. 4.

The molecular docking map of umami peptides and umami receptors T1R1/T1R3: The optimized T1R1/T1R3 model (A1) and Raman map (A2). The molecular docking of the four peptides with T1R1/T1R3 of the 3D diagram (B1-E1) and 2D diagram (B2-E2). (F) The count of interaction forces at individual docking sites for the for Umami and Salty peptides with T1R1/T1R3.

In summary, among the tested peptides, AD-7 demonstrated the strongest binding affinity to T1R1/T1R3, with the lowest docking energy (−8.779 kcal/mol) and the highest proportion of hydrogen bonds (75.86%). Multiple residues, including ALA176, ASN150, LYS155, GLN222, and LYS5, were involved in receptor binding. Asp219 was identified as a key acidic amino acid, as its higher frequency in AD-7 correlated with the lowest docking energy and the highest umami intensity, suggesting that this residue plays a critical role in stabilizing the peptide receptor complex and enhancing umami perception through synergistic hydrogen bonding and salt-bridge interactions. These findings support the rational design of novel umami peptides.

3.6.2. Molecular docking of peptides with TMC4 and TRPV1 taste receptors

The TMC4 protein adopts a typical α-helical transmembrane bundle conformation (Fig. 5A1). Ramachandran plot reveals two low-energy states, indicating two relatively stable conformational states during simulation (Fig. 5A2). As shown in Table 1, the binding energies follow the order: GQ-10 (−7.681 kcal/mol) < PL-9 (−6.863) < AD-7 (−6.844) < GA-8 (−6.755), which is consistent with the salt-enhancing capability observed in QDA experiments. Analysis of binding forces (Fig. 5B1-E2) shows that hydrogen bonding is the primary driving force, accounting for 60%–78% of interaction. Consistent with the findings of Askar et al. (2025). Electrostatic and hydrophobic interactions serve as supplementary forces, contributing 0%–17% and 14%–27%, respectively. Notably, AD-7 lacks hydrophobic interactions, which may explain its weaker binding affinity. Molecular docking details (Fig. 5F) reveal that GQ-10 binds to TMC4 through residues LYS389, GLN383, GLN377, GLU384, ASN221, VAL374, and SER214. Previous studies have identified LYS and GLU as key residues for TMC4 binding (Xie et al., 2023). The involvement of LYS389 was not observed in the other three peptides, suggesting that this specific interaction may be the key factor contributing to the superior salt-enhancing property of GQ-10.

Fig. 5.

The molecular docking map of peptides and salt receptors TMC4: The optimized TMC4 model (A1) and Raman map (A2). The molecular docking of the four peptides with TMC4 of the 3D diagram (B1-E1) and 2D diagram (B2-E2). (F) The count of interaction forces at individual docking sites for the for Umami and Salty peptides with TMC4.

TRPV1 acts as a multimodal integrator, where low sodium activates or sensitizes this pathway to enhance saltiness perception (Li, Sun, et al., 2024). The binding energies of the four peptides with TRPV1 are −6.85 (GQ-10), −6.566 (GA-8), −6.416 (AD-7), and − 6.032 (PL-9) kcal/mol (Table 1). No significant correlation was observed between docking energy and peptide length (Zhang et al., 2024). Analysis of binding forces reveals that hydrogen bonds and hydrophobic interactions are the primary mechanisms for TRPV1 binding: GQ-10 and GA-8 rely more on hydrogen bonds (55.55% and 50%, respectively), while PL-9 and AD-7 rely more on hydrophobic interactions (57.89% and 52.17%, respectively); electrostatic forces play only a minor role (Li, Liu, et al., 2024). As illustrated in Fig. S1B—F, the prevalence of bitter amino acids (Leu, Phe, Met) at the binding sites (e.g., Leu664, Met547, Leu577) suggests that these residues contribute to the salty taste via hydrophobic interactions, stabilizing the complex by minimizing water contact between the peptide and receptor (Huang et al., 2022).

In summary, hydrogen bonding dominates the peptide and TMC4 interaction, with GQ-10's unique binding to LYS389 explaining its highest salt-enhancing activity. For TRPV1, hydrophobic interactions involving bitter amino acids are the main stabilizing forces. These findings deepen the understanding of salt taste mechanisms and provide a theoretical foundation for developing low-sodium seasonings targeting both TMC4 and TRPV1.

3.7. MDS analysis

3.7.1. MDS analysis of T1R1/T1R3

Molecular docking only considers the binding of flexible ligands to rigid receptors, which may impose certain limitations on its assessment of ligand-receptor stability. (Feng et al., 2024). Therefore, to further validate the binding affinity and stability of the peptide-protein interactions, a 100 ns MDS was performed to explore the binding stability. The RMSD measures the displacement of atoms from their initial conformations over time and serves as an indicator of system stability, where lower RMSD values correspond to greater stability. (Song, Zhuang et al., 2023). AD-7 forms the most stable complex with the receptor, with the lowest ligand RMSD (∼0.15–0.20 nm) and minimal protein perturbation (∼0.4–0.5 nm), indicating that their conformations remain stable throughout the simulation (Liang et al., 2026). In comparison, GQ-10 exhibits intermediate stability, characterized by ligand RMSD values ranging from 0.20 to 0.25 nm. PL-9 displays moderate fluctuations with a ligand RMSD between 0.25 and 0.35 nm. However, its impact on the overall protein stability remains limited, with deviations maintained at approximately 0.4–0.5 nm. GA-8 exhibited the highest ligand RMSD (∼0.4 nm) and induced the greatest overall protein conformational adjustment (∼0.7–0.9 nm), suggesting a lack of specific anchoring interactions and an ongoing conformational search within the binding pocket (Fig. 6A-B).

Fig. 6.

Results of molecular dynamics simulations of peptides with the T1R1/T1R3 receptor. (A). Ligand RMSD in MD. (B) Complex RMSD in MD. (C) Protein RMSF in MD. (D) Complex Rg in MD. (E) Number of H-bonds. (F) Complex SASA in MD.

RMSF analysis was employed to evaluate the flexibility of amino acid residues throughout the simulation, revealing distinct mobility patterns across different structural regions of the protein (Zhang et al., 2022). As illustrated in Fig. 6C, the RMSF distribution across all systems exhibits similar overall patterns, with elevated fluctuations predominantly localized in the loop regions and terminal domains, while the core structural domains maintain relatively low flexibility. The GA-8 system displays higher peak values across multiple residue segments (>0.5–1.0 nm), indicating that this peptide enhances local flexibility, which potentially corresponds to induced-fit effects occurring around the binding pocket. In contrast, the AD-7 and PL-9 systems exhibit lower overall RMSF values, suggesting that these peptides contribute to maintaining the rigidity and stable conformation of the binding region. This may be attributed to the hydrogen bonding interactions between T1R1/T1R3 and the bifunctional peptides (Chen et al., 2025).

Rg was employed to characterize the structural compactness and stability of the system. A larger Rg value signifies that the system underwent significant expansion during the dynamic simulation, whereas a smaller Rg indicates that the system remained compact and stable throughout the trajectory (Liang et al., 2026). Similar to the RMSF results, the evolution of Rg over time reflects the overall structural integrity of the protein. The Rg for GA-8 increased continuously (from approximately 3.0 to 3.2 nm), suggesting that the protein conformation tended to become more loose. While the Rg of GQ-10 showed a slight increase, the magnitude was relatively limited. In contrast, the Rg values for AD-7 and PL-9 remained largely stable at approximately 2.95–3.0 nm, indicating that these complexes maintained a more compact folded state, which is typically associated with more stable binding interactions (Fig. 6D).

All four peptide systems exhibited stable hydrogen bond networks throughout the simulation, albeit with notable differences. AD-7 and PL-9 maintained consistently high hydrogen bond counts (∼8–14) throughout the entire molecular dynamics trajectory, which was favorable for enhancing binding stability and consistent with the RMSF analysis. GQ-10 demonstrated slightly lower hydrogen bond numbers but still maintained continuous hydrogen bonding interactions. In contrast, GA-8 exhibited lower and more fluctuating hydrogen bond counts, suggesting that the polar interactions were less persistent, which represents one of the key factors contributing to its inferior overall stability (Zou et al., 2024) (Fig. 6E).

SASA reflects the degree of protein surface exposure and the stability of the hydrophobic core. The GA-8 system exhibited the highest SASA (∼440–460 nm²), indicating greater exposure of hydrophobic surfaces, which was consistent with the increased Rg and elevated RMSD observed in this peptide. In contrast, AD-7 and PL-9 showed lower and more stable SASA values (∼410–430 nm²), suggesting more compact protein structures and better protection of the hydrophobic core. Hydrogen bonds played a crucial role in ligand-receptor interactions, stabilizing the complex structures formed by these peptides (Zhang et al., 2024). GQ-10 displayed intermediate SASA values and structural stability between the two groups (Fig. 6F).

The binding free energy is a fundamental tool for analyzing changes in ligand binding patterns by measuring the thermodynamic properties of the ligand. A negative value of ΔG indicates that the system is stable, while a positive value indicates instability. The lower the binding energy, the stronger the interaction, and the more stable the bond (Gao et al., 2024). As shown in Table S5, T1R1-T1R3/AD-7，T1R1-T1R3/GA-8，T1R1-T1R3/GQ-10，T1R1-T1R3/PL-9 are −71.19 ± 3.19，-37.05 ± 0.30，-58.04 ± 0.87，-73.00 ± 0.80. T1R1-T1R3/AD-7 and T1R1-T1R3/PL-9 exhibit high binding energies, followed by T1R1-T1R3/GQ-10, whose value is less than −50.0 kcal/mol, indicating that they have strong binding affinities for the receptor.

Integrated analysis of MDS reveals that AD-7 forms the most stable T1R1/T1R3 complex with sustained hydrogen bonding and compact folding, while PL-9 and GQ-10 exhibit moderate stability and GA-8 shows the weakest binding. These findings demonstrate that stable peptide-receptor interactions are critical determinants of taste potency.

3.7.2. MDS analysis of TMC4 and TRPV1

As shown in Fig. 7A (Ligand RMSD), the TMC4/AD-7 system exhibited relatively high ligand RMSD values (approximately 0.20–0.30 nm), with a noticeable increase and fluctuations after 20 ns, indicating considerable rearrangement within the binding site. In contrast, the TMC4/GA-8, TMC4/GQ-10, and TMC4/PL-9 systems showed lower ligand RMSD values (approximately 0.12–0.20 nm) with smaller fluctuations, suggesting more stable binding modes. Among these, GA-8 and PL-9 displayed the lowest mean RMSD (approximately 0.15 nm), indicating superior binding stability. Fig. 7B (Complex RMSD) illustrates overall complex stability. The TMC4/AD-7 complex exhibited significantly higher RMSD (approximately 0.45–0.55 nm), which increased rapidly within the first 10 ns and then plateaued, indicating poor stability. By contrast, the other three systems maintained RMSD values within approximately 0.28–0.35 nm, with PL-9 showing the lowest and most steady RMSD (mean ∼ 0.28 nm), suggesting the least perturbation to the protein structure. RMSF analysis (Fig. 7C) showed that all systems had low RMSF (<0.2 nm) in most regions, indicating a stable backbone. The TMC4/AD-7 system displayed the highest peak (>0.6 nm) in the 150–180 residue region, implying higher local flexibility that may compromise binding. GA-8 and PL-9 exhibited generally lower RMSF values, indicating stronger stabilization. The radius of gyration (Rg) (Fig. 7D) revealed that all systems maintained Rg within 2.55–2.62 nm. TMC4/AD-7 showed higher Rg (2.60–2.62 nm) with larger fluctuations, reflecting a looser structure, while PL-9 had the lowest Rg (2.55–2.57 nm) with minimal fluctuations, demonstrating the most compact structure. Hydrogen bond analysis (Fig. 7E) showed that GQ-10 possessed the highest average number of hydrogen bonds (approximately 10), indicating an extensive network that enhances binding stability. GA-8 ranked second (∼9 bonds), while AD-7 exhibited fewer bonds (∼8 bonds) with larger fluctuations. SASA analysis (Fig. 7F) indicated that AD-7 had higher and more variable SASA, while GQ-10 and PL-9 showed lower and more stable SASA values, consistent with their compact conformations. Binding free energy calculations (Table S5) gave ΔG values of −45.74 ± 3.60 (AD-7), −26.50 ± 2.97 (GA-8), −50.23 ± 4.05 (GQ-10), and − 42.50 ± 4.26 (PL-9) kcal/mol, with energy decomposition revealing electrostatic and van der Waals interactions as the main contributors.

Fig. 7.

Results of molecular dynamics simulations of peptides with the TMC4 receptor. (A) Ligand RMSD in MD. (B) Complex RMSD in MD. (C) Protein RMSF in MD. (D) Complex Rg in MD. (E) Number of H-bonds. (F) Complex SASA in MD.

As shown in Fig. S2A—B, ligand RMSD values for all four peptides in the TRPV1 binding pocket increased rapidly within the first ∼10 ns and then reached a stable plateau, consistent with previous reports (Chen et al., 2023; Chen et al., 2025). Among the peptides, GQ-10 exhibited the lowest RMSD (∼0.20–0.30 nm) with the smallest fluctuations, suggesting the most stable conformation within the binding site. GA-8 showed intermediate RMSD (∼0.25–0.30 nm), while AD-7 and PL-9 displayed higher values (∼0.40–0.55 nm), with PL-9 showing a slight late-stage elevation, indicating conformational rearrangement. Complex RMSD values converged to ∼0.5–0.65 nm, with GQ-10 showing slightly higher complex RMSD, suggesting more pronounced overall protein conformational changes. RMSF analysis (Fig. S2C) revealed that GA-8 and GQ-10 exhibited elevated peaks (∼1.0–1.5 nm) at several residue segments, indicating enhanced local flexibility, whereas AD-7 and PL-9 showed lower RMSF profiles, contributing to rigidity in the binding region. Rg values (Fig. S2D) initially increased slightly then fluctuated around ∼4.0 nm, indicating preserved overall folding. GA-8 and GQ-10 had higher mean Rg values, reflecting a more relaxed conformation, while AD-7 and PL-9 displayed lower Rg with minimal fluctuations, suggesting a more compact global structure. Hydrogen bond analysis (Fig. S2E) showed that GA-8 formed the highest number of hydrogen bonds (approximately 6–12) with high persistence, indicating a dominant role of polar interactions. AD-7 and GQ-10 maintained intermediate levels, while PL-9 exhibited the lowest hydrogen bond count (∼2–5), suggesting a greater reliance on hydrophobic effects. SASA analysis (Fig. S2F) showed that GA-8 had the highest SASA (∼440–460 nm²), indicating greater hydrophobic surface exposure, while AD-7 and PL-9 had lower and more stable SASA (∼410–430 nm²), implying a more compact hydrophobic core. Binding free energies (Table S5) for TRPV1 were − 42.65 ± 0.54 (AD-7), −69.85 ± 5.18 (GA-8), −42.98 ± 0.48 (GQ-10), and − 18.43 ± 3.01 (PL-9) kcal/mol, with electrostatic and van der Waals interactions again being the main contributors.

Integrated analysis of the two receptors reveals that all four peptides bind stably to both TMC4 and TRPV1, but with distinct preferences. On TMC4, PL-9 and GA-8 exhibited the most compact and stable complexes (lowest RMSD, Rg, and SASA), while GQ-10 enhanced stability through the most abundant hydrogen bond network. This may be the reason why GQ-10 exhibits higher saltiness enhancement perception. On TRPV1, GQ-10 showed the most stable ligand conformation, whereas AD-7 and PL-9 induced minimal perturbation to the overall protein structure and maintained a more compact receptor conformation. Notably, binding free energies on TMC4 (range − 26.50 to −50.23 kcal/mol) were generally more negative than on TRPV1 (range − 18.43 to −69.85 kcal/mol), although GA-8 showed an exceptionally strong affinity for TRPV1 (−69.85 kcal/mol). Collectively, these bioactive peptides stably bind to the active pockets of both salt taste receptors through finely tuned intermolecular interactions (electrostatic, van der Waals, and hydrogen bonding), which is of great significance for enhancing saltiness perception. The results also suggest that TMC4 may serve as a more consistently favorable target for most of the tested peptides, supporting its relevance in low-sodium food development.

4. Conclusion

This study identified four novel dual-taste peptides (GA-8, PL-9, AD-7, and GQ-10) from goat whey, with AD-7 exhibiting the most pronounced umami intensity and GQ-10 demonstrating the strongest saltiness enhancement. Molecular docking revealed that these peptides form stable complexes with the T1R1/T1R3 umami receptor through hydrogen bond networks, identifying ALA176, ASN150, LYS155, GLN222, Asp219 and LYS5 as a critical interaction site. A positive correlation was established between acidic amino acid content and umami intensity. For saltiness perception, molecular docking and dynamics simulations showed that the peptides bind to TMC4 via electrostatic and hydrogen bonding interactions (with GQ-10 distinctly engaging Lys389) and to TRPV1 through hydrophobic interactions involving Leu, Phe, and Met. Molecular dynamics simulations further confirmed that all four peptides maintain stable binding conformations with T1R1/T1R3, TMC4, and TRPV1 throughout 100 ns trajectories, as evidenced by low RMSD, compact Rg, and favorable binding free energies.

However, this study has several limitations. The molecular mechanisms were primarily elucidated through in silico docking and dynamics simulations, which require experimental validation. Additionally, the relative contributions of ENaC, TMC4, and TRPV1 to the observed saltiness enhancement remain to be delineated. Future research should therefore focus on validating these findings using cellular assays to clarify pathway-specific contributions. Addressing these aspects will further establish the potential of whey-derived peptides as effective salt replacers.

CRediT authorship contribution statement

Jiarui Yang: Writing – review & editing, Investigation, Data curation. Xue Yang: Software, Methodology, Investigation. Xiaoyan Yu: Writing – review & editing, Methodology, Formal analysis, Data curation. Qinying Li: Writing – review & editing. Qingyuan Liu: Writing – review & editing. Hong Li: Writing – review & editing. Wentao Zheng: Project administration, Funding acquisition. Yanan Shi: Writing – review & editing, Supervision, Project administration, Funding acquisition.

Ethics declaration

Informed consent and patient details

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

Studies in human

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. Ethics committee approval was not required under relevant laws and institutional guidelines. Informed consent and patient details Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed. Studies in Human This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was conducted in accordance with the guidelines of the Ethics Committee of Yunnan Agricultural University. The Ethics Committee of Yunnan Agricultural University granted an exemption. The authors provided the following explanation: Ethical review and approval were waived for this study by the Ethics Committee of Yunnan Agricultural University, as the study involved the sensory evaluation of goat whey peptides, which are food-derived and safe for human consumption. The evaluation followed standard protocols, and informed consent was obtained from all participants. The study did not involve clinical trials, medical interventions, or the collection of human biological materials; therefore, no additional legal, regulatory, or institutional approvals were required.

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.