If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
School of Food and Biological Engineering, Shaanxi University of Science and Technology, Xi'an 710021, ChinaShaanxi Research Institute of Agricultural Products Processing Technology, Xi'an 710021, China
Brown goat milk products have gained popularity for their unique taste and flavor. The emergence of chain-reversal phenomenon makes the design and development of goat milk products gradually tend to a consumer-oriented model. However, the precise mechanism of how browning and fermentation process causes characteristics is not clear. In an effort to understand how the treatments potentially lead to certain metabolite profile changes in goat milk, comprehensive, quantitative metabolomics and peptidomics analysis of goat milk samples after browning and fermentation were undertaken. An intelligent hybrid z-score standardization-principal components algorithm-multimodal denoizing autoencoder was used for feature fusion and hidden layer fusion in high-dimensional variable space. The fermentation process significantly improved the flavor of brown goat yogurt through the tricarboxylic acid-urea-glycolysis composite pathway. Bitter peptides HPFLEWAR, PPGLPDKY, and PPPPPKK have strong interactions with both putative dipeptidyl peptidase IV and angiotensin-converting enzyme, proving that brown goat yogurt can be considered as effective provider of potential putative dipeptidyl peptidase IV and angiotensin-converting enzyme inhibitors. The level of health-promoting bioactive components and sensory contributed to consumer selection. The proposed multimodal data integrative analysis platform was applicable to explain the effect of the dynamic changes of metabolites and peptides on consumer preferences.
). Food quality perception was affected by sensory, health, convenience, and process-related variables according to the Total Food Quality Model. Among them, sensory and health-related consumption purposes were considered as critical factors, which could override the influences of the default thinking mode and determine the choice of yogurt (
Chemical composition, such as peptides, free fatty acids (FA), and free AA, was closely related to the physical properties and sensory quality of goat milk and its products (
Fatty acid profiles of milk from Holstein cows, Jersey cows, buffalos, yaks, humans, goats, camels, and donkeys based on gas chromatography-mass spectrometry.
). As the heat inactivation of the lipase reduces the lipolysis, the concentration change of UFA in UHT milk is significantly less than that in raw milk. A total of 1,317 peptides were identified in brown goat milk and yogurt, of which 297 peptides significantly increased after fermentation (
Novel insight into the transformation of peptides and potential benefits in brown fermented goat milk by mesoporous magnetic dispersive solid phase extraction-based peptidomics.
). Moreover, thermal processing of the protein-sugar system in goat milk could enhance the mouthfulness, umami, and continuity attributes of peptides, as well as improve the antimicrobial and antioxidant properties of peptides and proteins. Fermentation process of goat milk results in the generation of a series of different metabolites and hydrolysis of goat milk proteins into oligopeptides at the action of enzymes released by lactic acid bacteria (
). Functionality is enhanced further by the release of bioactive peptides. Compounds crosslinked by metabolites, AA, peptides, and proteins from glycosylation and fermentation products bestow desirable senses and nutrition to goat milk products.
To understand the complex interactions that occurred in biological systems, the advantage of using various analytical strategies to obtain comprehensive coverage of components has been well-understood. Several studies have focused on the peptidome in yogurt with multiple-species starters (
). Metabolomics is the golden standard to identify and quantify metabolites from goat milk products driven by various biological statuses and functions, which has been used for elucidating metabolic differences of yogurt from different starter cultures (
Metabolic footprint analysis of volatile metabolites by gas chromatography-ion mobility spectrometry to discriminate between different fermentation temperatures during Streptococcus thermophilus milk fermentation.
Metabolic footprint analysis of volatile metabolites to discriminate between different key time points in the fermentation and storage of starter cultures and probiotic Lactobacillus casei Zhang milk.
). Biomimetic systems, mainly the electronic nose and tongue, have the potential to reproduce the human sensory system mechanism, against fatigue or stress of human (
). These strategies have been moved from focusing on individual components (e.g., metabolite) to encompassing the entire metabolome, as well as even assessing the complementary omics measurements (e.g., peptidomics, proteomics, lipidomics) in parallel (
A combined metabolomics and peptidomics approach to discriminate anomalous rind inclusion levels in Parmigiano Reggiano PDO grated hard cheese from different ripening stages.
). The data obtained from the same sample but acquired in different modalities were expected to display some level of correspondence. However, it was still a major challenge to directly correlate the comprehensive components information for functional analysis.
The general framework to integrate spatial-omics and biomimetic data focused on starting to map denoizing and normalized signals to pathways of interest and subsequently combining the mapped data with flavor characteristics, molecular nutrition, and other protein expression data (
). To obtain pathways of interest authenticates from denoized and normalized spatial-omics data, segmentation was available for training classification models, with input as compounds intensity and output as pathways and functional features such as bitter, sour, and tricarboxylic acid cycle (
). For downstream integration analysis, multimodal data types were loaded into the data object structure and compared between biological conditions. Integration of multimodal data and meta-analysis was improving the accuracy of multimodal data via imputation analysis, which uses information from similar (adjacent) compounds determined in a reduced-dimension potential space to adjust intensity values (
This study aimed to investigate the effect of browning and fermentation on the release of metabolites and peptides. Additionally, combining the spatial-omics and biomimetic data connects metabolite and peptide profiles with flavor, as structural similarities correlated with compounds contents levels. The specific idea of data acquisition was shown in Figure 1. Through spectra dimensionality reduction and standardization, multimodal data from multiple samples were combined and compared with the fermentation and protein glycosylation process. Further, the effect on flavor and nutrition evolution was evaluated, as those factors drive the consumption of goat yogurt.
Figure 1A schematic diagram summarizing the workflow for metabolite and peptides profiling.
No animal subjects were used in this work, so ethical approval for the use of animals was thus not required.
Sample Collection and Fermentation Protocol
Whole goat milk was collected from a bulk tank containing goat milk from the Saanen dairy goat between March and July 2021 in Shaanxi Province, China. The milk samples were pasteurized at 91°C for 10 min and homogenized, packaged, refrigerated at 4°C, and transported to the laboratory immediately (
). Glucose was added at 8 g/100 mL to the whole goat milk system and placed in a glass bottle, incubated in an autoclave at 45°C for 30 min, and cooled to room temperature. Then, the mixture was heated at 95°C and maintained for 180 min to prepare the brown goat milk samples. Afterward, the brown goat milk was cooled to 42°C. Lactobacillus bulgaricus and Streptococcus thermophilus (1:1; Angel Yeast Co., Ltd.) as yogurt starter culture was inoculated into the brown goat milk, and the mixture was anaerobically fermented at 42 ± 1°C until the pH reached 4.5. The total fermentation time was 10 h. After fermentation, the brown goat yogurts were stored at 4°C to interrupt the fermentation process. Each sample was processed in sextuplicate and held at −80°C before analysis.
Sample Preparation
Regarding the extraction step of the untargeted metabolomics workflow, the goat milk, brown goat milk and brown goat yogurt samples were centrifuged at 15,000 × g at 4°C for 30 min and the top fat layer was discarded. For each sample, 1 mL of supernatant was mixed with 7 mL of precooled acetonitrile by vortexing for 5 min and then centrifuged at 15,000 × g at 4°C for 20 min to precipitate macromolecular aggregates. Supernatant was concentrated by rotary evaporation before reconstituting in 500 μL of 40% acetonitrile and filtered through 0.22 μm microporous membranes in amber vials until ultra-high-pressure liquid chromatography-quadrupole Orbitrap mass spectrometer (UHPLC-Q Orbitrap MS) analysis.
Extraction step of peptidomics analysis involved goat milk samples with triplicate replications. Specifically, goat milk, brown goat milk, and brown goat yogurt samples were centrifuged at 15,000 × g at 4°C for 30 min to separate the top fat layer and precipitate the cells. The supernatant was transferred to a fresh tube, added with an equal 20% (vol/vol) trichloroacetic acid solution, and left on ice for 10 min to precipitate proteins. After centrifugation at 15,000 × g at 4°C for 20 min, the supernatant was purified and desalted by C18 Sep-Pak cartridge (Waters Corporation). The C18 resin was activated with 60% acetonitrile containing 0.1% formic acid and equilibrated with 0.1% formic acid solution. Goat milk samples were loaded on C18 resin, and washed 3 times with 0.1% formic acid to remove salts and carbohydrates. Afterward, the elution of peptides from C18 resin was accomplished using 60% acetonitrile in 0.1% formic acid. The eluted peptides were dried in SpeedVac Concentrator (Bionoon) and reconstituted in 500 μL of 60% acetonitrile acidified with 0.1% formic acid for Orbitrap analysis. Quality control (QC) samples of metabolomics and peptidomics were prepared by mixing equal portions of goat milk, brown goat milk, and brown goat yogurt samples to ensure good repeatability and stability of the instrument system.
Liquid Chromatography Tandem Q Orbitrap MS Analysis
The liquid chromatography-MS platform for metabolite and peptide analysis was UltiMate 3000-UHPLC system tandem quadrupole Orbitrap mass spectrometer (Thermo Fisher Scientific) equipped with a heated electrospray ionization source. The chromatographic separation was achieved through a thermostated (35°C) Hypersil GOLD C18 column (100 mm × 2.1 mm, 1.7 μm; Thermo Fisher Scientific) with mobile phase A (0.1% formic acid in ultrapure water) and mobile phase B (0.1% formic acid in acetonitrile) at a flow rate of 0.3 mL/min. The elution gradient of 5 μL metabolites was as follows: 0.1 min, 2% B; 3 min, 20% B; 9 min, 98% B; 14 min, 98% B; 14.1 min, 2% B; 15 min, 2% B. The elution gradient of 5 μL of metabolites was as follows: 0 to 1 min, 2% B; 1 to 3 min, 2 to 20% B; 3 to 9 min, 98% B; 9 to 14 min, 98% B; 14 to 14.1 min, 98 to 2% B; 14.1 to 15 min, 2% B. The separation of 10-μL peptide mixtures had the following elution gradients: 0 to 9 min, 2% B; 9 to 70 min, 2 to 35% B; 70 to 75 min, 35 to 98% B; 75 to 95 min, 98% B; 95 to 95.1 min, 98 to 2% B; and 95.1 to 100 min, 2% B.
Mass spectra of eluted metabolites and peptides were collected in data dependent acquisition mode with one precursor scan followed by 15 candidate ions scans per cycle. For metabolomics, the source parameters were as follows: 40 arb of sheath gas flow, 8 arb of auxiliary gas flow, 320°C of heated capillary, 350°C of the ion source, 55 V of S-lens RF, and 3.5 or −3.0 kV of spray voltage for positive and negative ion modes, respectively. Full scan mass spectra were acquired at mass ranges of m/z 100 to 1,500 with a resolution of 70,000 full width at half maximum, and MS/MS fragmentation spectra were triggered and acquired depending on preselected criteria with a resolution of 35,000 full width at half maximum using higher-energy collision-induced dissociation with collision energy at 17.5, 35, and 52.5 eV. For peptidomics, the full mass scan was acquired at mass scans of m/z 100–2000 with the automatic gain control (AGC) target of 3 × 106 and accumulation time of 250 ms. In the MS/MS mode, accumulation time of 120 ms, isolation width of m/z 4, and an AGC target of 2 × 105 were employed. The ionization source parameters, resolution and collision energy were identical to the metabolomics workflow.
Biomimetic Technology Analysis
Chromatic parameters were reflected in Chroma-Meter (CR-300 colorimeter, Konica Minolta). Using the International Commission on Illumination L*, a*, and b* (CIELAB) color measurement system with D65 illuminant and measuring angle of 10°, samples were tested in transparent quartz cuvettes at ambient temperature. L* represented the brightness of goat milk samples, a* and b* represented the chromaticity coordinates, in which a* indicates the red-green components, and b* indicates the yellow-blue components. Measurements were replicated 6 times with 3 biological duplicates. Mean values were used. The statistical models used to determine the total color difference (ΔE) were the following (
The taste of goat milk, brown goat milk, and brown goat yogurt were determined using SA402B electronic tongue (Intelligent Sensor Technology Co., Ltd.). Seven chemical sensors were used for different taste characteristics, including sourness, saltiness, umami, richness, astringency, sweetness, and bitterness. Sensors were immersed in the internal solution (3.3 mmol/L potassium chloride and saturated silver chloride) and reference solution (30 mmol/L potassium chloride and 0.3 mmol/L tartaric acid) for 24 h. Accurately weighed 100 g of samples, then diluted to 200 g with ultrapure water and centrifuged at 10,000 × g for 15 min at 4°C. For taste evaluation, poured 25 mL of supernatant into beakers, and the sensor array was soaked in sample resolution for 120 s before measuring to provide reliable analysis. Before each measurement, the sensor array was rinsed until stable readings were obtained. Each sample was measured 6 times, and the mean values were used for subsequent analysis.
Sensory Evaluation
Goat milk and yogurt samples were coded with random 3-digit numbers and displayed on the small glass pots. The panel consisted of 40 assessors (20 males and 20 females) with an average age of 25 to 45 yr. The sensorial valorization of raw goat milk, brown goat milk, and brown goat yogurt was conducted by a hedonic test based on a 9-point hedonic scale (1 = disliked, 5 = moderately liked, and 9 = extremely liked). Thereby, attributes such as appearance, texture, acid, sweetness, bitterness, caramelization, and fattiness were evaluated for the samples. In addition, all the necessary environments and the same conditions of light and temperature were prepared for the assessors, and samples were randomly distributed to assessors, along with deionized water and unsalted crackers to cleanse their palates. Each milk sample was evaluated 6 times.
Data Preprocessing
Raw data set of untargeted metabolomics was imported to the Compound Discoverer software (version 3.2, Thermo Fisher Scientific) without any conversion to complete background drift correction, peak extraction with the signal-to-noise ratio threshold of 3, retention times alignment with a mass tolerance of 5 ppm and m-score threshold of 10, compound grouping, and elemental compositions prediction using mzCloud (https://www.mzcloud.org/) and ChemSpider (http://www.chemspider.com/). The adduct ions [M+H]+, [M+Na]+, [M+NH4]+, and [M+K]+ were selected for the positive ionization mode, and [M-H]−, [M-2H]2−, [M+Na-2H]−, and [M+HCOO]− for the negative ionization mode. For the identified compounds, a similarity search with fragmentation data was performed using mzCloud and normalization was performed with constant median (
Metabolomic profile of milk fermented with Streptococcus thermophilus cocultured with Bifidobacterium animalis ssp. lactis, Lactiplantibacillus plantarum, or both during storage.
). Raw data set of peptidomics was processed using MaxQuant software (Max Planck Institute of Biochemistry, Martinsried, Germany) with Andromeda as the peptide search engine. Bos taurus and Capra hircus database from UniProtKB database (https://www.uniprot.org) were used to retrieve sequences. No enzyme and missed cleavage were defined. Glucosylation, methylglyoxal derived hydroimidazolone, pyrraline, carboxymethylation, protein N-terminal acetylation, oxidation, and dimethylation were defined as variable modifications and cysteine carbamidomethylation as fixed modifications. Results were filtered using 0.01 false discovery rates cut-off at the protein level. Other settings were the same as regular search. Sequence information was given in their confidence and the identified peptides with high confidence were used for downstream protein identification analysis.
Peptide sequences were submitted to the bioactive peptide databases, including EROP-Moscow, BIOPEP, and MBPDB for search and novelty check. The isoelectric point (pI) was evaluated by Expasy-pI/Mw tool, the water solubility was predicted by Innovagen online software (http://www.innovagen.com/proteomics-tools) and the Ames mutagenicity and developmental toxicity potential were determined by ToxinPred (http://crdd.osdd.net/raghava//toxinpred/).
Obtained untargeted metabolomics and peptidomics data matrices were uploaded into MetaboAnalyst 5.0 (https://www.metaboanalyst.ca) to complete data preprocessing and statistical analysis, including missing value replacement, log-transformation, and Pareto scaling. Pattern recognition analysis, including hierarchical clustering, principal component analysis (PCA) and partial least square discriminant analysis (PLS-DA) was performed to forecast the group-based clustering of metabolites and peptides. Differential metabolites and peptides were defined as follows: P < 0.05 in one-way ANOVA test, fold-change (FC) 2 or <0.5, and variable important in projection (VIP) value >1.0 (
Characterisation of key odorants causing honey aroma in Feng-flavour Baijiu during the 17-year ageing process by multivariate analysis combined with foodomics.
). Furthermore, the R2Y and Q2 of the permutation test were used to assess the goodness-of-fit and predictive ability of the PLS-DA models.
Integrative Analysis of Multimodal Spatial Data
A general framework (Figure 2A) of integrating complementary multimodal numbers was proposed. Differential abundant peptides and metabolites were fused to provide systems biological interpretation for further holistic analysis. Principal components analysis was employed for feature selection and dimensionality reduction. It should be pointed out that before using principal components analysis, it was necessary to standardization to prevent over capturing some features with too large values. The Z-score standardization method was adopted.
[2]
where
represents the component's eigenvalue of i sample after standardization, xi represents the initial eigenvector, and n represents the number of samples. The contribution rate of the feature vector needed to be calculated to determine the number of features m at the final dimensionality reduction (
). Value of m was obtained by the following formula.
[3]
where x(i) is the initial spatial coordinate of sample i,
is the spatial coordinate after mapping to the low-dimensional space, and m is the dimension of the initial data after dimension reduction. Multimodal denoizing autoencoder (MDA) added noise to the input data and trained to reconstruct the original noisy data as output layer (Figure 2B). This approach led to increased generalization and robustness in the learning model. Based on the MDA, feature fusion and hidden layer fusion strategies (Figure 2C) for fusion of the spatial-omics and biomimetic signals were used to improve the recognition performance of classifiers for sensory characteristics (
Figure 2Integrative analysis of multimodal spatial data. (A) General framework of integrating complementary multimodal data. (B) Schematic structure of a denoizing autoencoder. (C) Schematic structure of feature fusion and hidden layer fusion.
In the feature fusion strategy, the spatial-omics feature was assumed to be xspa, biomimetic feature was assumed to be xbio, and then these 2 features were merged to form a new feature xspa+bio.
[4]
Noise-added input xspa+bio' was mapped to
through the qD function.
[5]
Noise-added input
was mapped to the hidden layer through the
activation function.
[6]
Original noise-free input data zspa+bio was reconstructed through the
hidden function.
[7]
Among them, the parameters θ and θ' were trained by the unsupervised back-propagation algorithm to minimize the average reconstruction error LH(Xspa+bio,zspa+bio).
In the hidden layer fusion strategy, the spatial-omics feature and biomimetic feature were processed separately to form 2 new hidden layer features, which were then merged.
Noise was added into spatial-omics feature xspa, and then mapped to
through the qD function.
[8]
Noise-added input
was mapped to the hidden layer through the
activation function.
[9]
Original noise-free input data zspa was reconstructed through the
hidden function.
[10]
Among them, the parameters θ and θ' were trained by the unsupervised back-propagation algorithm to minimize the average reconstruction error LH(Xspa,zspa).
Noise was added into biomimetic feature xbio, and then mapped to
through the qD function.
[11]
Noise-added input
was mapped to the hidden layer through the
activation function.
[12]
Original noise-free input data zbio was reconstructed through the
hidden function.
[13]
Among them, the parameters θ and θ' were trained by the unsupervised back-propagation algorithm to minimize the average reconstruction error LH(Xbio,zbio).
Merged, the yspa and ybio formed the
.
[14]
For the feature fusion strategy, the features from spatial-omics data of each sample and biomimetic feature were directly linked using the BestFirst approach as the input for the MDA.
Statistical Analysis
The SPSS statistics (version 22.0, SPSS-IBM) software was used for variance analyses and Duncan's multiple range test, and P < 0.05 was regarded as significant. The experimental results were expressed as the mean values ± standard deviation.
RESULTS AND DISCUSSION
General Effect of Browning Treatment on Raw Goat Milk
Caramelization, as the main nonenzymatic reaction, significantly contributes to the color change in goat milk. The browning color of brown goat milk and brown goat yogurt is one of the particular characteristics that attract consumers. L* value, which refers to the white and blackness of the sample, was significantly decreased after Maillard reaction, and a significantly lower L* value (90.87) was observed in the brown goat milk compared with the raw goat milk (Figure 3A). The decrease of L* value demonstrated the decrease of lightness and darker color after the Maillard reaction, which may be due to the darker colored compounds formation (
). The darker colored compounds formed after browning process were mainly due to the aldol and aldehyde-amine condensations of fission products, reductones, and Strecker degradation products in the final stage of Maillard reaction. In addition, a* and b* values were significantly increased in brown goat milk. The simultaneous increase of a* and b* values indicate that the Maillard reaction made the brown goat milk lose blueness and greenness, meanwhile showing more yellowish and reddish color. There are 2 main phenomena during browning treatment: (1) introduction of the glycation groups at the protein surface, including glucosylation, methylglyoxal derived hydroimidazolone, and pyrraline modifications, which increased the solubility; and (2) glucose/lactose-mediated proteins cross-linking during browning treatment reduced the solubility, browning of the goat milk system was good indication of this phenomenon.
Figure 3Biomimetic technology analysis and univariate statistical analysis, chemometric analysis of untargeted metabolomics. (A) Chromatic parameters. a–c Means in the different bars with different letters were significantly different (P < 0.05). L* represents the lightness, a* indicates the red-green components, and b* indicates the yellow-blue components. (B) Radar chart of E-tongue data. (C) Categories of identified metabolites and the distribution of substances in all samples. (D) Heatmap of the abundant of individual metabolites subclasses. (E) Principal component (PC) analysis score map for raw goat milk samples. (F) Partial least square discriminant analysis (PLS-DA) score plot of raw goat milk, brown goat milk, and brown goat yogurt samples. (G) Permutation test results of the PLS-DA model. (H) Volcano plot of goat milk samples comparing brown goat milk to goat milk samples. The x-axis represents the log2 fold-change (FC) value, and y-axis corresponds to the −log10(P). “Sig” represents metabolites with significant change, and “Unsig” indicates metabolites with no significant change. (I) Volcano plot of goat milk samples comparing brown goat yogurt to brown goat milk samples, with red and blue color indicating statistical significance. (J) The top 25 matched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. (K) The significant correlation network for matched pathways.
The average intensity values of the 7 taste attributes determined by electronic tongue were visually exhibited in a radar chart (Figure 3B), including sourness, saltiness, umami, richness, astringency, bitterness, and sweetness. The 3 vivid taste attribute indicators were bitterness, sweetness, and umami, followed by sourness and saltiness. Results of the sensory evaluation of goat milk and products based on the attributes of appearance, texture, acid, sweetness, bitterness, caramelization, and fattiness are presented in Supplemental Figure S2 (https://data.mendeley.com/datasets/jfr6pzfgch;
). As expected, brown goat milk groups showed significantly higher bitterness, caramelization, and sweetness scores compared with raw goat milk (P < 0.05). These results suggested that the results of sensory evaluation yielded very similar to those of the taste attributes determination using electronic tongue. Furthermore, browning produced the caramelized flavor, which was the most significant evaluation attribute compared with goat milk, and enhanced the consumer preference. Previous studies have proved that the caramelized flavor of brown milk and yogurt was favored by consumers, particularly in China, Japan, Europe, Southeast Asia, and other regions (
General Effect of Fermentation Treatment on Brown Goat Milk
The color of brown goat milk and brown goat yogurt changed with the same pattern, but with different offsets (Figure 3A). The increase of L* value after fermentation may be due to the higher degree of light scattering and micelle aggregation phenomenon. The a* and b* values in brown goat yogurt were significantly higher than those in brown goat milk. When the samples were compared individually with raw goat milk, brown goat milk (ΔE = 14.16) had a smaller color change than the brown goat yogurt (ΔE = 16.66), and these ΔE offsets were attributed to the fermentation.
Following the results of electronic tongue and sensory evaluation, it was found that brown goat yogurt was significantly higher in sourness than brown goat milk (P < 0.05), which might be related to the hydrolysis of milk protein by starter bacteria to AA and organic acid during fermentation. Following fermentation, the bitterness and sweetness intensity scores were significantly lower than that of brown goat milk (P < 0.05), while the astringency was significantly higher than brown goat milk (P < 0.05). These might be due to the more prominent sourness caused by the fermentation process, which shields the astringency and bitterness to some extent. In general, sweetener inhibited bitterness receptor TAS2R31 and TAS2R43 responses in a concentration-dependent manner (
). In addition, the fatty flavor in brown goat yogurt resulted in the upregulation of “Ca2+-signaling,” the common mechanism between the bitter, fatty, and sweet perceptions. Following activation of fatty receptors and inhibition of bitterness receptors, store-operated calcium channels opened, triggering inhibition of TRPM5 channels (
), which in brown goat yogurt suppressed the consumers perception of bitterness. The lower texture score of brown goat milk compared with brown goat yogurt may be due to the whey proteins aggregation caused by the free–SH group interacting with the S–S bond of cysteine-containing proteins, such as α-LA, β-LG, and κ-CN via –SH/S–S interchange reactions (
). The improvement of the flavor and texture of brown goat milk was obviously by the fermentation process, due to significant metabolic changes caused by starter bacteria.
Quantification of Metabolites and Peptides
The significant difference peptides and lipids were quantified by UHPLC-Q Orbitrap MS using internal standard (IS). Briefly, LGLHQKPGV was chosen as the IS for peptides quantified and SPLASH mixed lipid standard (Alabaster) for lipids (
). The IS were added to goat milk and product samples to complete quantification after extraction. Peptides prepared by solid-phase synthesis approach were obtained from GenScript Ltd. Reconstituted samples (200 μL) with 8 different concentrations, including 0.5 mg/L peptide IS and 20 μL lipid IS, were injected to construct the calibration curve (
Novel insight into the transformation of peptides and potential benefits in brown fermented goat milk by mesoporous magnetic dispersive solid phase extraction-based peptidomics.
). The significant differences in metabolites were quantified by the external standards, which were gained from Sigma-Aldrich, Dr. Ehrenstorfer GmbH, and LGC Standards with 98% purity.
To verify the suitability of the established method for routine monitoring, the linearity, limit of detection (LOD), limit of quantitation (LOQ), precision, and overall recovery were assessed in accordance with the US-FDA bio-analytical method validation guidance. The 68 standards exhibited satisfactory linearity over the tested concentration range with the correlation coefficients (r) greater than 0.99 (Supplemental Table S1; https://data.mendeley.com/datasets/jfr6pzfgch;
The LOQ of the instrument was determined as 3 times the LOD. Both LOD and LOQ of UHPLC-Q Orbitrap MS were in the range of 0.001 to 4.290 and 0.002 to 12.871 mg/L (Supplemental Table S1). The intraday and interday relative standard deviation for all standards was lower than 15%, demonstrating that the established method was accurate and reproducible the determination of metabolites, peptides, and lipid in goat milk and products.
Metabolite Modality for the Evolution of Flavor
Overall, 782 metabolites were annotated in positive and negative ionization modes for each group of comparison (goat milk, brown goat milk, and brown goat yogurt samples) to improve the metabolome coverage. Identified metabolites spanned 17 chemical categories, of which the categories as high numbers of species include organic heterocyclic compounds, lipid metabolites, acids, amines, AA, and their derivatives, peptides, and esters (Figure 3C), which may indicate that they cast essential roles of the anabolism of other metabolites. Next, a heatmap combined with the hierarchical cluster analysis was constructed to give a good overview of the smallest and largest intensities based on the chemical pattern and cluster the 3 sample groups according to the similarities of their chemical profile. Heatmap displayed the data with the samples on the horizontal axis and the metabolites categories on the vertical axis. The generated heatmap formed Cluster I and Cluster II (Figure 3D). Cluster I aggregated the brown goat milk and yogurt samples. In contrast, Cluster II groups goat milk. Goat milk samples were characterized by higher levels of amines, sulfur compounds, nucleosides, esters, acids, and peptides. When the browning treatment and fermentation process were carried out, highlighted the increase in the relative contribution of amino acids and their derivatives, vitamins, lipid metabolites, phenols, ketones and alcohols was observed, when compared with brown goat milk and brown goat yogurt samples. Regarding the chemical profile of goat milk submitted to treatment, it is clearly demonstrated that browning and fermentation treatment helps to protect the flavor and amora of this goat milk mainly by developing ketones, alcohols, phenols, and aldehydes. The ketone-containing compounds would be generated through the Maillard reaction as final Maillard reaction products. The content of ketone-containing compounds accounted for a higher proportion than the rest in brown goat milk and brown goat yogurt, which positively contributed to the perception of fatty and creamy flavors (
). 4-Chloro-2-methylthio-6-(trifluoromethyl)pyrimidine and decyl sulfate were increased significantly after browning (Supplemental Table S2; https://data.mendeley.com/datasets/jfr6pzfgch;
), and then reacted with the carbonyl intermediates of Maillard reaction, which contributed to the unique flavor of brown goat milk. Furthermore, the higher levels of flavonoids and pyridine as organic heterocyclic compound in brown goat milk could contribute to its unique color and flavor.
Unsupervised and supervised multivariate statistical analyses were used to visualize the compositional variability among the goat milk, brown goat milk and brown goat yogurt samples. The underlying objective was to investigate whether the glucose-based browning treatment left distinctive fingerprint in raw goat milk and, if s)o, whether the fermentation process amplified or masked it. The above 785 metabolites converged in an unsupervised multivariate model of PCA. As indicated in Figure 3E, there were clear differences among the goat milk, brown goat milk and brown goat yogurt samples in terms of expressed metabolites, but there was no separation between samples in each group. Notably, a greater difference was found in goat milk compared with the other milk samples, confirming that the changes in metabolomic characteristics were related to the browning and fermentation treatment. The QC samples were tightly clustered, suggesting good reproducibility, reliability, and stability of the instrumental system and ensuring the quality of downstream analyses. Supervised PLS-DA mode was further employed to eliminate the irrelevant factors that influence on metabolic data matrix and enhanced the comprehension of different metabolic patterns (Figure 3F). The intergroup differences in PLS-DA were magnified and the intragroup differences were shrunken in comparison to that in PCA, which might due to filtering out information unrelated to the classification information (
). The R2X, R2Y, Q2, and permutation of 1,000 tests of the PLS-DA model indicated the model was robust and had low risk of overfitting (Figure 3G). The metabolites with VIP value >1.0 in PLS-DA analysis were used for further selecting the significantly differential metabolites by univariate analysis.
In univariate analysis, 254 and 263 metabolites showed significant upregulation (P < 0.05) at brown goat milk compared with those in goat milk and in brown goat yogurt compared with the brown goat milk, whereas, 263 and 241 metabolites showed significant downregulation (Figure 3H and 3I). In addition, 213 metabolites complied with VIP >1, FC value >2, P < 0.05 were considered to be significantly differentially expressed. Despite showing lower fold-change, several metabolites [such as citramalic acid, LysoPC (18:3), and acetylcholin] also consisted in the submission list as they were shown as significantly altered compounds with great contribution to the PLS-DA analysis, according to the high VIPvalue. Pathway enrichment analysis was carried in MetaboAnalyst 5.0 platform, based on the hypergeometric test, integrating high-quality Bos taurus KEGG database as the backend knowledge bases, to preliminarily determine the metabolic pathways most closely attribute to the differentially expressed metabolites. According to the meta-analysis, 75 statistically pathways were identified covering a broad range of metabolic pathways, in which the top 25 were shown in Figure 3J. Significance of these pathways was determined by P-value and enrichment ratio (i.e., the metabolic pathway with a darker color and larger bubbles was the most significant, including the tricarboxylic acid cycle, alanine metabolism, gluconeogenesis, nucleotide sugars metabolism, aspartate metabolism, phospholipid biosynthesis, as well as the urea cycle). However, some of these pathways may be irrelevant to canonical metabolic pathways. To further visually reveal the synergistic-regulatory connection among various pathways, a correlation network with the pathway pairs determined (Figure 3K) was constructed. The network contained 39 nodes, with 93 edges and 50 connected pathways. Examining Figure 3K, it is apparent that tricarboxylic acid cycle, glutamate metabolism, alanine metabolism and urea cycle had significant correlations with other pathways, and they were carbon and nitrogen metabolism pathways, which indicated that microbial strains could decompose organic substances for the synthesis of small molecular metabolites in the goat milk system during browning and fermentation treatment.
Peptide Modality for the Evolution of Nutrition
In the fermentation process, Lactobacillus bulgaricus and Streptococcus thermophilus as yogurt starter culture sustain the bacterial growth by hydrolyzing goat milk proteins to AA and peptides as external nitrogen sources. Therefore, in this study, a streamlined analytical pipeline for peptidomics was presented. The liquid chromatography-high resolution mass spectrometry analysis of peptidomics was completed in 100 min, and typical peaks corresponding to high abundance peptides were presented in total ion chromatogram. The generated data set was integrated with publicly available databases. A total of 196 peptides with different charges were identified, and the major part (92.3%) had molecular masses below 2,600 Da (Figure 4A). These peptides were mainly released by native enzymes in milk derived from αS1-CN, αS2-CN, β-CN, phospholipase, and acyl-coenzyme A oxidase, which agreed with the other report (
). Moreover, glucose or lactose in goat milk system can format the glucose- or lactose- and heat-induced protein modifications in the heating. The identification of glycosylation modifications of goat milk protein was outlined in Supplemental Figure S3 (https://data.mendeley.com/datasets/jfr6pzfgch;
). After assessment by PCA (Figure 4B), the peptides in the brown goat milk exhibited more difference from the other groups (goat milk and brown goat yogurt), whereas the QC group suggested good reproducibility, reliability, and stability of the instrumental system.
Figure 4Univariate statistical analysis and chemometric analysis of peptidomics. (A) Distribution of molecular weight of total peptides. MV = molecular mass. (B) Principal component analysis (PCA) score map for raw goat milk samples (n = 6, blue), brown goat milk samples (n = 6, red), brown goat yogurt samples (n = 6, green), and QC sample (n = 6, purple). QC = quality control. (C) Numerical distribution of bioactive peptides. ACE = angiotensin-converting enzyme; DPP IV = putative dipeptidyl peptidase IV; DPP III = dipeptidyl peptidase III; UMPS = ubiquitin-mediated proteolysis. (D) Numerical distribution of sensory peptides. (E) Schematic diagram describing metabolite alterations involved in lipid metabolism in raw goat milk (R), brown goat milk (B), and brown goat yogurt (Y) samples. Intensities of significantly different metabolites in the heatmap were expressed as relative levels compared with raw goat milk. LysoPC: lysophosphatidylcholine; PC: phosphatidylcholine; PE: phosphatidylethanolamine; PS: phosphatidylserine.
Previous reviews reported that bioactive peptides, consisting of 2 to 20 AA residues, were easily absorbed in the intestine and facilitated the enhancement of biological activity during digestion (
). Molecular weight of most identified peptides was lower than 3 kDa, indicating that these peptides might possess higher bioactivities and bioavailability.
), which was linked to digital data sets (BIOPEP). Figure 4C showed the numerical distribution of putative dipeptidyl peptidase IV (DPP IV) inhibitor, angiotensin-converting enzyme (ACE) inhibitor, dipeptidyl peptidase III (DPP III) inhibitor, antioxidative, stimulating, α-glucosidase inhibitor, renin inhibitor, regulating, antithrombotic, antiamnestic, hypotensive, activating ubiquitin-mediated proteolysis (UMPS), anti-inflammatory, CaMPDE inhibitor, neuropeptide, and bacterial permease ligand, with various peptides having more than one hypothetical function. High number of these bioactivities was due to the accompanying derivation of similar peptides from the same sequence portion of the parental protein. Specific peptide sequences may rigger various sensory perceptions (e.g., bitter, umami, sour, salty, sweet, astringent, or kokumi). The distribution of sensory peptides among the identified peptides was listed in Figure 4D, 42.96% functional domain were correlated with bitterness. After the water solubilities prediction as well as the Ames mutagenicity and developmental toxicity potential prediction, 21 peptides with biological activity, good water solubility, and nontoxicity were selected (Table 1). Further, drawing upon the results of previous literatures and this sensory profile data, the bitter peptide profiles of goat milk, brown goat milk and brown goat yogurt samples were predicted by hydrophobicity (expressed as Q-value) and the identified peptide sequences were compared with the sensory peptides contained in the BIOPEP database. Among the 21 screened peptides, 7 peptides (HPFLEWAR, PAGLPDKY, PPGLPDKY, PPPPPKK, AFLKLFR, AKCMFFK, and AMKPWTQPK) were considered to be bitter due to the Q-value above 1,400 cal/mol (Table 1). The contents of these peptides in brown goat milk (7.24 mg/L) were significantly higher than in raw goat milk (0.60 mg/L) and brown goat yogurt (1.98 mg/L), which may have been related to the higher heating temperature in browning process or greater protease hydrolysis during fermentation. Browning apparently induced site-specific breakdown of the protein backbone by a radical-mediated oxidative fragmentation that occurs on the protein backbone, where the radicals as intermediates during the Maillard reaction were formed (
). However, to confirm that these 7 peptides contribute to the perception of bitterness in the mouth, assessment of their concentration relative to the perception threshold, as well as investigation of the effect on bitter receptors, were necessary.
Table 1The contents of goat milk protein-derived peptides and prediction of bioactivity, solubility, and toxicity
It has been proved that the peptides containing Pro residues and their location have a decisive effect on DPP IV inhibitory activities. Peptides with the Pro residue in the second position of the N terminus were generally considered to be potential strong DPP IV inhibitors due to their substrate-like structure (
). Among these released fragments from goat milk, brown goat milk, and brown goat yogurt samples, 6 sequences with the PeptideRanker scores of >0.5 and AA residues of <10 (i.e., GPFPILV, PPPPPKK, HPFLEWAR, HPWVCQR, PPGLPDKY, and APHRKMHR) have the typical Pro residue at the second position and displayed in Table 2. A total of 3, 4, 4, and 4 peptides, respectively, were qualified to be DPP IV inhibitors. To determine the potential interactions between goat milk protein-derived bioactive peptides with DPP IV enzyme, Pepsite2 prediction software was used. Table 2 displayed the PeptideRanker scores, Pepsite 2 (P-value), and potential binding residues of the identified 6 sequences in goat milk, brown goat milk and brown goat yogurt samples with DPP IV, where the HPWVCQR (0.69 mg/L) presented the lowest P-value (0.000357). The identified 6 peptides were capable of binding to at least 4 key residues of DPP IV. In particular, APHRKMHR (0.13 mg/L) from brown goat milk could bind up to 16 important residues and remain the only peptides that can bind to the Arg123, Glu203, Glu204, Asn711, and His741 of the catalytic triad, demonstrating the potential to inhibit DPP IV activity. GPFPILV (caprine β-CN-derived heptapeptide) shared in the 3 samples could bind to 4 important residues from DPP IV, and its minimal inhibitory concentration for Escherichia coli DPC6053 was 163.7 ± 1.33 μm (
Molecular basis of the anti-diabetic properties of camel milk through profiling of its bioactive peptides on dipeptidyl peptidase IV (DPP-IV) and insulin receptor activity.
). It was further found that PPPPPKK and HPFLEWAR from brown goat yogurt samples could bind to 8 and 6 important active site residues of DPP IV. The brown goat yogurt sample has the greatest number of peptides with DPP IV inhibitor activities. Therefore, brown goat yogurt can be considered as an effective provider of potential DPP IV inhibitors.
Table 2Identified bioactive peptides in goat milk, brown goat milk, and brown goat yogurt samples and their potential binding sites with dipeptidyl peptidase IV (DPP IV) enzymes
The ACE-inhibitory peptide has been proved to be the most common bioactive peptides in fermented milk products. Milk-derived ACE-inhibitory peptides have gradually received attention and were considered to be safer and gentler alternatives for blood pressure management (
). Low molecular weight peptides exhibited high ACE-inhibitory activity. Moreover, the ACE activity of peptide was controlled by AA sequence. Among the peptides from goat milk, brown goat milk and brown goat yogurt samples, 14 sequences with the PeptideRanker scores of >0.5 and amino acids residues of <10 were displayed in Table 3, which meet the characteristics of typical proline, tryptophan and lysine residues at the C-terminus, and aromatic residue (His, Phe, Pro, Tyr, or Trp) or hydrophobic residue (Ile, Leu, or Val) at the N-terminal position (
An exploration of angiotensin-converting enzyme (ACE) inhibitory peptides derived from gastrointestinal protease hydrolysate of milk using a modified bioassay-guided fractionation approach coupled with in silico analysis.
). The active site of typical ACE consisted of 3 pockets which were P1 (Ala354, Glu384, and Tyr523), P2 (Gln281, His353, His513, Lys511, and Tyr520), and P3 (Glu162;
). Peptides AKCMFFK, AMKPWTQPK, FGARCPQPR, HFAPWCK, PAGLPDKY, PPGLPDKY, PPPPPKK, DCGEDWFR, and VGINYWLAHK were able to bind to 8 important active residues of ACE and exert the competitive inhibitory effect of ACE substrate by spatially blocking the active sites. VGINYWLAHK were expressed in the 3 goat milk samples, also identified as antioxidative peptides, with 0.20, 0.24, 0.18 for goat milk, brown goat milk, and yogurt, respectively. Similar to DPP IV inhibitors, brown goat yogurt samples contained the largest number of ACE-inhibitory peptides, which may be related to the strong proteolysis of Lactobacillus bulgaricus and Streptococcus thermophilus during the fermentation. These findings showed that peptides from brown goat yogurt samples have ACE-inhibitory activities. The results provided better knowledge of the peptide profiles with biological properties, which help to increase consumers understanding about the nutritional value and potential health benefits of brown goat yogurt.
Table 3Identified bioactive peptides in goat milk, brown goat milk, and brown goat yogurt samples and their potential binding sites with angiotensin-converting enzyme (ACE) inhibitors
Brown goat milk and brown goat yogurt were both highly diverse and dynamic at the metabolites and peptides level, in many cases with functional characteristics as they pertain to the sensory. To construct the comprehensive profile of brown and fermented samples as well as identify the relationships between peptides and metabolites, multimodal analysis integrating peptidomics and untargeted metabolomics data from the same biological samples was performed.
For the feature fusion strategy, the features from spatial-omics data of each sample and biomimetic feature were directly linked using the BestFirst approach as the input for the MDA. Mean accuracy and standard deviation of the linear support vector machines approach results with the feature fusion were listed in Supplemental Table S4 (https://data.mendeley.com/datasets/jfr6pzfgch;
). In the case of both the brown goat milk and brown goat yogurt samples, after the fusion of biomimetic data (ΔE, sourness, saltiness, umami, richness, astringency, bitterness, sweetness, and all data), the highest classification accuracy of 83.25% was achieved in the brown goat milk sample using the linear support vector machines approach, with a standard deviation of 1.31%, whereas in the case of the brown goat yogurt sample, the highest accuracy of 79.33%, with a standard deviation of 1.37%. In the hidden layer fusion strategy, in both the brown goat milk and brown goat yogurt samples, the highest classification accuracy of 85.68% was achieved in the brown goat milk sample, with a standard deviation of 1.45%, whereas in the case of the brown goat yogurt sample, the highest accuracy of 82.65%, with a standard deviation of 1.22%. The results showed that in these 2 blocks, the classification performance was significantly improved after the fusion of spatial-omics hidden layer and biomimetic hidden layer. Compared with feature fusion strategy, hidden layer fusion strategy achieved more significant improvements in classification. Integrated peptidomics and untargeted metabolomics data served as an easy-to-implement approach to explore multimodal data solutions and improves robustness of spatial-omics solutions. Raising awareness of common spatial-omics fusion analyses pitfalls will contribute to avoid the overinterpreting of results and miss valuable insights when performing spatial-omics analyses on biological data.
Biochemical Landscape of Sensory and Nutrition in Goat Milk Products
Balance of sensory and nutrition is a major challenge for the goat milk industry (
). Goat milk was prone to physiological and nutritional changes upon thermal processing and fermentation. Previous studies have demonstrated that the highest amount of FA found in goat milk were oleic acid (C18:1c), palmitic acid (C16:0), myristic acid (C14:0), decanoic acid (C10:0) and stearic acid (C18:0), visualized in Supplemental Figure S1B. The content of these compounds was largely affected by processing (
Fatty acid profiles of milk from Holstein cows, Jersey cows, buffalos, yaks, humans, goats, camels, and donkeys based on gas chromatography-mass spectrometry.
). As the heat inactivation of lipase reduces the lipolysis, the concentration alter of UFA in brown goat milk were less than that in raw goat milk, including eicosapentaenoic acid and arachidonic acid. Lipid metabolites were essential sources of flavor compounds in complex goat milk substrates. Glycerol 3-phosphate was formed from glucose via glycolysis and esterified with fatty acyl-CoA to drive lipogenesis, directly linked the triglyceride biosynthesis. Content of glycerol 3-phosphatein brown goat milk (102.048 mg/L) was significantly higher than that in brown goat yogurt (79.871 mg/L). Hydrolysis of triacylglycerols caused “goaty” aroma, and the branched-chain fatty acids of triacylglycerols were hydrolyzed by microbial lipase, led to the increase in the content of the free fatty acids such asbutyric acid, octanoic acid, and hexanoic acid, as well as off-odor (
). Phosphatidylethanolamine (O-16:0), ceramide 3, LysoPE (16:0), LysoPC (18:3), and phosphocholine (PC) were specifically upregulated after the fermentation process (Figure 4E). Phosphatidylethanolamine, LysoPE, PC, and LysoPC were polar lipids in goat milk, participated in the cell signaling and formation of the neural system. They acted as emulsifiers to ensure the stability of the water/oil emulsion system of goat milk, and also determined the lipase activity on lipid droplets in the digestive tract (
Development of a high-throughput method for the comprehensive lipid analysis in milk using ultra-high performance supercritical fluid chromatography combined with quadrupole time-of-flight mass spectrometry.
). Polar lipids rich in UFA had high susceptibility to oxidation and produced peculiar smells, whereas brown goat yogurt had lower levels of LysoPC (18:3; 0.031 mg/L). Accumulation of these lipid metabolites can considerably affect flavor.
Sugars were first decomposed into glucose in fermentation, which can synthesize glucose-1-phosphate under the action of glucokinase and phosphofructokinase-1. Glucose-1-phosphate synthesized the d-galactose with the significantly increased level after fermentation under the action of UTP-glucose-1-phosphate uridylyltransferase, UDP-glucose 4-epimerase, and galactokinase. l-Phenylalanine was the important substrate to produce phenylacetaldehyde, 2-coumaric acid, and phenylpyruvic acid. Identified compounds in Maillard reaction derived from xylose-phenylalanine exhibited flower and fruit aroma (
). Goat milk was characterized by high content of threonine (6.01 g/100 g protein) and low content of cysteine (0.27 g/100 g protein). Significant increases in the phenylalanine contents follow the Maillard reaction and fermentation processes. Pyruvate was the important metabolic hub substrate to produce Ile, Leu, and Val through isoleucine, leucine, and valine and synthesize acetyl-CoA with the action of pyruvate dehydrogenase and participate in the tricarboxylic acid cycle. These AA created a bitter taste and were important in controlling the flavor and nutritional value of brown goat milk. In addition, citrate, oxaloacetate, oxalosuccinate, and cis-aconitate were the crucial intermediates involved in the tricarboxylic acid cycle, glutamate metabolism and alanine metabolism. Fermentation tended to decrease the levels of these 4 compounds in brown goat yogurt samples (Figure 5), implying that fermentation affected the energy metabolism in goat milk.
Figure 5Schematic diagram describing metabolite alterations involved in the urea cycle, tricarboxylic acid cycle, and nucleotide sugar metabolism in raw goat milk (R), brown goat milk (B), and brown goat yogurt (Y) samples. Intensities of significantly different metabolites in the heatmap are expressed as relative levels compared with raw goat milk. GDP: guanosine diphosphate; GTP: guanosine triphosphate.
). Fermented goat milk generally tended to be sourer than brown goat milk, at least partly due to the conversation of citrate. It should be noted that fermentation significantly altered the biosynthesis of N2-hexanoyl-l-lysine, trans-aconitic acid, and arginine. It can be found from Figure 5 that aspartic acid converted lysine and N2-hexanoyl-l-lysine in the process of fermentation. Increased l-phenylalanine degradation gave rise to phenyl acetic acid, phenyl lactic acid, and benzoic acid, showing a much stronger correlation with bitterness (
). Content of l-carnitine increased after the fermentation; l-carnitine is an AA metabolite that can accelerate the conversion of fatty acids into energy, facilitating weight loss and is indispensable for fatty acid and glucose metabolism and hormonal regulation. These findings suggested fermentation enhanced tricarboxylic acid cycle and glycolysis in goat milk, affecting the flavor and taste.
Due to the low odor threshold, aldehydes, especially with significant difference trans-p-coumaraldehyde (P < 0.05), have overwhelming impact on the overall aroma. Studies have shown that trans-p-coumaraldehyde can inhibit xanthine oxidase, which could generate reactive oxygen species through a cascade reaction, causing a series of harmful reactions, including endothelial dysfunction, apoptosis, cardiac mechanical energy inflammation, sclerosis, uncoupling, aging, and cancer (
). This study found that trans-p-coumaraldehyde was screened out among the goat milk, brown goat milk, and brown goat yogurt, and trans-p-coumaraldehyde had the highest levels in the brown goat yogurt. The increase in trans-p-coumaraldehyde content of brown goat yogurt could be associated with the higher degradation of residual lipids under the action of starter bacteria. Moreover, the most significant organic heterocyclic compounds contents were enriched in brown goat milk, including 2-(2-chlorophenyl)-1-cyclohexyl-6-oxopiperidine-3-carboxylic acid and 2,2-dimethyl-N-[1-(methylsulfonyl)-2,3-dihydro-1H-indol-5-yl]propenamide. The 2 organic heterocyclic compounds vanished after fermentation process. Taurochenodeoxycholic acid, as a bile acid, may be derived from the heterocyclic intermediate in final stage of Maillard reaction, which has been proved to reduce the severity of colitis (
). Taurochenodeoxycholic acid was raised by the fermentation. Taurine metabolism was also enriched in the brown milk yogurt metabolomes compared with yogurt. Ketones and alcohols usually possessed high odor thresholds and do not affect the flavor in the Maillard reaction system. Ketones were generated through lipid degradation, Maillard reaction, and the interaction between these 2 reactions. Ketones possessed relatively high dour threshold and thus showed low impact on the flavor of goat milk. Significant increase in the contents of the significantly different ketones of brown goat yogurt was (4E)-1,7-bis(4-hydroxyphenyl)-4-hepten-3-one. (4E)-1,7-bis(4-hydroxyphenyl)-4-hepten-3-one as the top metabolite could reduce pain in neuropathic pain-associated behaviors (
Dietary supplementation of gingerols- and shogaols-enriched ginger root extract attenuate pain-associated behaviors while modulating gut microbiota and metabolites in rats with spinal nerve ligation.
). Alcohols arise mainly from the oxidation and degradation of lipids. l-Iditol, bis(4-ethylbenzylidene)sorbitol, d-(-)-mannitol, 1,4:3,6-dianhydrohexitol, and 4,5-difluoro-1,2-benzenediol were identified as significant alcohols, in which l-iditol and d-(-)-mannitol were promoted to accumulate after fermentation. These indicated that brown goat yogurt could improve harmful reactions and products nutritional quality.
Bitter Evolution by the Browning and Fermentation Treatment
Bitter but nutritious foods were usually avoided as bitterness was ordinarily a warning of potential toxicity (
Lotus seed green embryo extract and a purified glycosyloxyflavone constituent, narcissoside, activate TRPV1 channels in dorsal root ganglion sensory neurons.
J. Agric. Food Chem.2022; 70 (35343690): 3969-3978
). Main AA of the identified peptides were Trp, Tyr, Phe, Pro, Leu, and Ala, which produced strong bitterness. Contents of 24 significant short peptides were generally enriched in brown goat milk, in which 9 peptides were significantly increased after fermentation, whereas l-valyl-l-prolyl-N5-(diaminomethylene)-l-ornithine significantly decreased. Most significantly increased peptides in the brown goat milk compared with raw goat milk included Leu-Leu, Leu-Tyr, Val-Leu, Phe-Tyr, Leu-Leu-Tyr, Tyr-Leu, Met-Phe, Leu-Gly-Leu, Ala-Tyr, and Lys-Leu, were considered as bitter peptides. This illustrated that Maillard reaction products after browning displayed more bitterness in the brown goat milk. Further, these peptides were not found in the brown goat yogurt sample, which demonstrated that fermentation process could weaken the bitterness produced by Maillard reaction. Most significantly increased peptides in the brown goat yogurt compared with brown goat milk included Phe-Leu, Asp-Val-Lys, and Leu-Phe, which increased by 28.1-, 24.8-, and 20.0-fold. Phe-Leu and 5-O-Pro-Trp have been reported as bitter peptide, Leu-Phe as sour peptide and Ile-Pro-Ile as umami, suggesting that brown goat yogurt showed more sourness and umami than brown goat milk. AFLKLFR peptide with high Q-value (1,871 cal/mol) coexisted in the goat milk and brown goat yogurt samples, and had high peak intensity in brown goat yogurt sample. This supported that during browning and fermentation process, peptide bonds were opened and AFLKLFR embedded in the protein structure was gradually exposed and released, producing bitter with taste cells. HPFLEWAR, PPGLPDKY, PPPPPKK were particularly noteworthy because they had the potent interaction with both DPP IV and ACE. These 7 peptides were highly likely to activate T2R40 or TA2R7 receptors (probability: 86.65–73.09%) using BitterX software (Supplemental Table S5; https://data.mendeley.com/datasets/jfr6pzfgch;
). These findings illustrated that the fermentation process was believed to be beneficial as it improved the flavor and organoleptic properties of brown goat milk. Rich variety of amino acids and their derivatives as well as peptides in brown goat yogurt not only served as taste contributors and essential nutrients but also health-promoting bioactive components.
CONCLUSIONS
Overall, the metabolites and peptides from goat milk, brown goat milk, and brown goat yogurt samples in the current study were primarily considered in the context of the potential biological activity and sensory characteristics. Biomimetic technology and sensory evaluation analysis demonstrated the improvement of the flavor, color and texture of brown goat milk was obviously by the fermentation process as the end metabolites produced by starter bacteria formed a balanced flavor through the shield of astringency and bitterness by the more prominent sourness. Spatial-omics spotted that the significantly different compounds, such as polar lipids, bitter amino acids and peptides, were changed with the fermentation process through tricarboxylic acid cycle-urea cycle-glycolysis composite pathway. The spatial-omics-biomimetic data synchronization acquisition network ensured that the 2 fusion methods improved the recognition accuracy of the consumer's purchase desire due to the flavor characteristics and molecular nutrition, thus demonstrating the complementarity of the modalities. The proposed multimodal spatial data integrative analysis platform endowed high robustness and meaningful interpretation with multimodal fusion data.
ACKNOWLEDGMENTS
This work was supported by the National Natural Science Committee Foundation of China (Beijing; Grant No. 32272401). The authors have not stated any conflicts of interest.
REFERENCES
Ali Redha A.
Valizadenia H.
Siddiqui S.A.
Maqsood S.
A state-of-art review on camel milk proteins as an emerging source of bioactive peptides with diverse nutraceutical properties.
Molecular basis of the anti-diabetic properties of camel milk through profiling of its bioactive peptides on dipeptidyl peptidase IV (DPP-IV) and insulin receptor activity.
Metabolic footprint analysis of volatile metabolites by gas chromatography-ion mobility spectrometry to discriminate between different fermentation temperatures during Streptococcus thermophilus milk fermentation.
Lotus seed green embryo extract and a purified glycosyloxyflavone constituent, narcissoside, activate TRPV1 channels in dorsal root ganglion sensory neurons.
J. Agric. Food Chem.2022; 70 (35343690): 3969-3978
Novel insight into the transformation of peptides and potential benefits in brown fermented goat milk by mesoporous magnetic dispersive solid phase extraction-based peptidomics.
Characterisation of key odorants causing honey aroma in Feng-flavour Baijiu during the 17-year ageing process by multivariate analysis combined with foodomics.
Metabolomic profile of milk fermented with Streptococcus thermophilus cocultured with Bifidobacterium animalis ssp. lactis, Lactiplantibacillus plantarum, or both during storage.
An exploration of angiotensin-converting enzyme (ACE) inhibitory peptides derived from gastrointestinal protease hydrolysate of milk using a modified bioassay-guided fractionation approach coupled with in silico analysis.
A combined metabolomics and peptidomics approach to discriminate anomalous rind inclusion levels in Parmigiano Reggiano PDO grated hard cheese from different ripening stages.
Dietary supplementation of gingerols- and shogaols-enriched ginger root extract attenuate pain-associated behaviors while modulating gut microbiota and metabolites in rats with spinal nerve ligation.
Metabolic footprint analysis of volatile metabolites to discriminate between different key time points in the fermentation and storage of starter cultures and probiotic Lactobacillus casei Zhang milk.
Development of a high-throughput method for the comprehensive lipid analysis in milk using ultra-high performance supercritical fluid chromatography combined with quadrupole time-of-flight mass spectrometry.
Fatty acid profiles of milk from Holstein cows, Jersey cows, buffalos, yaks, humans, goats, camels, and donkeys based on gas chromatography-mass spectrometry.
00761-5&id=gr1.jpg){kind=link}
00761-5&id=gr2.jpg){kind=link}
00761-5&id=gr3.jpg){kind=link}
00761-5&id=gr4.jpg){kind=link}
00761-5&id=gr5.jpg){kind=link}