Method for preparing exosome, method for dynamically monitoring flavor deterioration of formula milk powder and application

Abstract

The present application relates to a method for preparing formula milk powder exosomes, comprising the following steps: (1) dissolving formula milk powder of different stages and different storage times in water to obtain a formula milk powder solution; (2) pretreatment to obtain whey supernatant; (3) gradient ultracentrifugation: i) centrifuging the whey supernatant at 16,000-17,000xg for 20-40 minutes; ii) centrifuging the obtained supernatant at 80,000-100,000xg for 50-70 minutes, iii) ultracentrifuging the obtained supernatant at 130,000-140,000xg for 80-100 minutes, and discarding the supernatant to obtain a crude exosome product; (3) purification to obtain the formula milk powder exosomes. The present application also relates to a method for dynamically monitoring flavor deterioration of formula milk powder during storage and application. The exosomes prepared by the method of the present application have high yield and purity, and the structure and membrane proteins are complete.

Description

Technical Field

[0001] This invention relates to the technical field of exosomes from formula milk powder, specifically to a method for preparing exosomes from formula milk powder, a method for dynamically monitoring flavor deterioration of formula milk powder during storage, and their applications. Background Technology

[0002] Infant formula, as an important substitute for breast milk for infants and young children, directly impacts market acceptance and infants' willingness to consume it due to its sensory quality. During storage, formula is prone to lipid oxidation and protein degradation, leading to flavor deterioration (such as the formation of off-flavors like aldehydes and ketones), severely affecting product quality. Traditional methods rely heavily on static techniques (such as electronic noses and electronic tongues) to assess flavor changes, which struggle to dynamically reflect the overall remodeling process of the lipid metabolism network and accurately link lipid molecular characteristics with sensory quality. This results in a lack of targeted early warning and control strategies for flavor deterioration.

[0003] Current technologies for monitoring flavor deterioration in infant formula mainly focus on the static analysis of end products (such as volatile compounds like aldehydes and ketones), neglecting the driving role of dynamic lipid metabolism remodeling during storage in flavor formation. Furthermore, the relationship between compositional changes in exosomes (or extracellular vesicles, EVs) as important carriers of lipid bioactive molecules and flavor deterioration remains unclear, and traditional detection methods cannot effectively capture the dynamic evolution of exosomal lipid profiles and key biomarkers.

[0004] Therefore, there is an urgent need for an exosome analysis technology based on lipidomics to analyze the molecular correlation between changes in exosome lipid composition and flavor deterioration, establish a precise and dynamic quality monitoring system, and provide a scientific basis for optimizing the storage stability and improving the flavor quality of formula milk powder. Summary of the Invention

[0005] Technical issues

[0006] One object of this invention is to provide a method for preparing exosomes from formula milk powder. Through rennet pretreatment, multi-step gradient ultracentrifugation, and ultrafiltration purification steps, the method ensures efficient retention and low loss of lipid components.

[0007] Another objective of this invention is to provide a method for dynamically monitoring flavor deterioration of infant formula during storage. By analyzing the dynamic changes in lipid composition of infant formula exosomes (IFEVs) during storage, combined with non-targeted lipidomics and correlation analysis of volatile compounds in infant formula, key lipid biomarkers are screened, and a quantitative correlation model of lipid evolution and flavor deterioration is established. Furthermore, an exosome lipid profile database will be developed to achieve precise monitoring and early warning of flavor quality during infant formula storage, providing a scientific basis for optimizing production processes and extending shelf life.

[0008] Technical solution

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] On one hand, the present invention provides a method for preparing exosomes from formula milk powder, comprising the following steps:

[0011] (1) Dissolve formula milk powder of different stages and storage time in water to obtain formula milk powder solution;

[0012] (2) Pretreatment: Add enzyme and CaCl2 to the formula milk powder solution, mix well, incubate at 4°C for 30 minutes, mix repeatedly, and centrifuge at 4000-6000 rpm for 10-20 minutes to obtain whey supernatant.

[0013] (3) Gradient centrifugation:

[0014] i) Centrifuge the whey supernatant obtained in step (2) at 16,000-17,000×g for 20-40 minutes to remove fat, casein, enzymes and other impurities to obtain supernatant;

[0015] ii) Centrifuge the supernatant obtained in step i) at 80,000-10,000×g for 50-70 minutes to remove large particulate impurities.

[0016] iii) Centrifuge the supernatant obtained in step ii) at 130,000-140,000×g for 80-100 minutes, discard the supernatant, and obtain crude exosomes.

[0017] (3) Purification: The crude exosomes obtained were resuspended in phosphate buffer and purified by filtration through a 100kDa ultrafiltration membrane to obtain the formula milk powder exosomes.

[0018] In one embodiment, the exosomes contain the following seven lipids as biomarkers associated with deterioration in the flavor of formula milk powder: phosphatidylglycerol 16:1 / 16:1, phosphatidylethanolamine 10:0e / 6:0, lysophosphatidylethanolamine 16:0, lysophosphatidylglycerol 16:0, lysophosphatidylinositol 16:0, lysophosphatidylcholine 16:0, and lysophosphatidylcholine 14:0.

[0019] Optionally, the enzyme used in the pretreatment of step (2) is rennet or trypsin to remove casein.

[0020] Optionally, the different stages are stage 1, stage 2, and stage 3, and / or the different storage times are 0, 9, 18, and 24 months.

[0021] In one embodiment, the purification in step (3) includes: resuspending the obtained crude exosomes in phosphate buffer and washing them by ultracentrifugation at 135,000×g for 60 minutes; then transferring the obtained supernatant to a 100kDa ultrafiltration membrane and filtering it twice at 3,000×g to obtain the formula milk powder exosomes.

[0022] On the other hand, the present invention provides a method for dynamically monitoring whether the flavor of formula milk powder deteriorates during storage, comprising the following steps:

[0023] (a) Non-targeted lipidomics analysis of exosomes:

[0024] Non-targeted lipidomics analysis was performed on the exosomes of formula milk powder prepared according to the method described above, and 532 differentially expressed lipids were obtained.

[0025] (b) Dynamic monitoring of volatile compounds in formula milk powder during storage. Six flavor compounds were screened from three stages of formula milk powder at different storage periods (0, 9, 18 and 24 months), including hexanal, E-2-pentenal, heptanal, octanal, nonanal and E,E-2,4-heptadienal.

[0026] (c) Exosomal lipid-formula milk powder flavor association modeling and database construction:

[0027] c1) Weighted lipid co-expression network analysis (WLCNA): Using the 532 differentially expressed lipids obtained in step (a) as input data, a co-expression network was constructed to identify modules strongly associated with flavor compounds. Among them, the 58 lipids classified as MEgreen module by WLCNA showed a positive correlation with the 6 flavor compounds mentioned in step (b).

[0028] c2) By performing connectivity analysis in the MEgreen module, seven key lipids were identified as the most important in the MEgreen module, and their abundance was positively correlated with storage time. The seven key lipids were phosphatidylglycerol (PG) 16:1 / 16:1, phosphatidylethanolamine (PE) 10:0e / 6:0, lysophosphatidylethanolamine (LPE) 16:0, lysophosphatidylglycerol (LPG) 16:0, lysophosphatidylinositol (LPI) 16:0, lysophosphatidylcholine (LPC) 16:0, and lysophosphatidylcholine (LPC) 14:0. These seven key lipids were used as biomarkers for the deterioration of the flavor of formula milk powder.

[0029] (d) Construction of a multiple linear equation for the exosome lipid content to indicate the content of flavor compounds in formula milk powder:

[0030] d1) Data preparation: The content data of the seven pivot lipids and the six flavor compounds were imported into the R statistical environment for processing;

[0031] c2) Model construction: A multiple linear regression model was independently established for each of the six flavor compounds;

[0032] c3) Model Evaluation: For each model, the following statistical indicator was calculated: Coefficient of Determination (R²) 2 ): The proportion of variance explained by the quantitative model; Adjusted R-value. 2 ): Correcting for the influence of the number of independent variables; F-statistic and its p-value: testing the overall significance of the model (α = 0.05); Coefficient t-test p-value: assessing the significance of each lipid predictor;

[0033] c4) Computational tools: All statistical analyses are performed in the R language environment.

[0034] The prediction equations shown in Table 2 are obtained;

[0035] (d) The formula milk powder to be tested is separated into exosomes, lipids are extracted from the exosomes, and the contents of 7 lipids are determined according to the method described above. The results are then substituted into the prediction equation in Table 2 to obtain the predicted concentration values ​​of 6 flavor compounds. By comparing the predicted concentration values ​​of the 6 flavor compounds with the concentration thresholds in Table 3, the flavor status of the formula milk powder during storage and production is determined, thereby determining whether the flavor of the formula milk powder has deteriorated.

[0036] Optionally, the non-targeted lipidomics analysis described in step (a) includes:

[0037] a1) Add an isotope internal standard to the exosomes of formula milk powder prepared by any one of claims 1 to 5, then add methanol and methyl tert-butanol in sequence, extract by ultrasonication, centrifuge at 5,000-7,000 rpm for 15-40 minutes, collect the upper organic phase, concentrate by nitrogen blowing, and obtain lipid extract.

[0038] a2) The obtained lipid extract was analyzed by ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS):

[0039] Chromatographic conditions: A CSH C18 column was used, and the mobile phase consisted of acetonitrile-water (A) and acetonitrile-isopropanol (B) containing 0.1% formic acid and 0.1 mM ammonium formate, with gradient elution; Mass spectrometry conditions: Mass spectrometer, switching between positive and negative ion modes, scan range m / z 200-1800;

[0040] a3) Data processing:

[0041] Lipids were identified using LipidSearch 4.2, and a total of 37 classes and 1,814 lipids were identified.

[0042] The 532 differentially expressed lipids were obtained by orthogonal partial least squares discriminant analysis (OPLS-DA).

[0043] Optionally, step (b) includes:

[0044] b1) Headspace solid-phase microextraction: The reconstituted formula milk powder is mixed with an internal standard, and solid-phase microextraction is performed using divinylbenzene / carbon molecular sieve / polydimethylsiloxane fiber to adsorb volatile compounds in the formula milk powder.

[0045] b2) The adsorbed volatile compounds were analyzed using a gas chromatography-mass spectrometry system, and 17 flavor compounds were identified in the three stages of formula milk powder at different storage periods.

[0046] b3) Using random forest regression analysis, based on the mean accuracy decrease value, 6 flavor compounds were screened from 17 flavor compounds, including hexanal (grassy flavor), E-2-pentenal (fatty flavor), heptanal (vegetable oil flavor), octanal (cheese flavor), nonanal (lemon flavor), and E,E-2,4-heptadienal (nut flavor).

[0047] Optionally, the multiple linear regression model described in step c2) is:

[0048]

[0049] Y k X represents the content of the target volatile compound, where k is an integer from 1 to 6; iThe content of 7 lipid molecules is represented by β, where i = integers from 1 to 7; β0 is the intercept term; β i represents the regression coefficients of each lipid; ∈ represents the random error term.

[0050] The model parameters were estimated using the least squares method, and the model form was: lm(Y ~ X1 + X2 + ... + X7).

[0051] Optionally, the R language environment uses the following packages: readxl (v1.4.3) for data import; stats (v4.2.3) core package for regression modeling; and writexl (v1.4.2) for exporting structured results.

[0052] Optionally, in step (d), the predicted concentration value of each of the six flavor compounds is calculated as follows: according to the prediction equation in Table 2, the intercept corresponding to each volatile compound is added, and the content of each of the seven lipids calculated from the formula milk powder to be tested is multiplied by the independent variable coefficient of the corresponding lipid in Table 2 and summed to obtain the predicted concentration value of each volatile compound; then, the predicted concentration value of each volatile compound is compared with the concentration threshold in Table 3. If the predicted value exceeds the threshold and shows a multiple relationship, the flavor of the formula milk powder has deteriorated. The larger the multiple relationship, the deeper the degree of deterioration.

[0053] In another aspect, the present invention also provides the application of formula milk powder exosomes prepared according to the method described above in the dynamic monitoring of flavor deterioration of formula milk powder during storage.

[0054] Beneficial effects

[0055] (1) Efficient separation and information retention: The method of this invention is used to prepare exosomes, which improves the yield and purity while ensuring the integrity of the exosome lipid membrane, reducing the loss of genetic information and active substances, and ensuring the integrity of the exosome nanoparticle structure and high retention of membrane proteins and intramembrane lipids.

[0056] (2) The method of the present invention can be used to dynamically monitor flavor deterioration of formula milk powder during storage:

[0057] Dynamic monitoring and early warning capabilities: Traditional methods only statically detect lipid oxidation end products (such as aldehydes and ketones), while this invention, through untargeted lipidomics (UHPLC-MS / MS) combined with weighted lipid co-expression network analysis (WLCNA), can dynamically analyze the evolution of lipid profiles in formula milk exosomes (EVs) and identify seven key lipid biomarkers in the early storage stage (PG16:1 / 16:1, PE 10:0e / 6:0, LPE 16:0, LPG 16:0, LPI 16:0, LPC ... The study identified six key flavor compounds (hexanal, E-2-pentenal, heptanal, octanal, nonanal, and E,E-2,4-heptadienal) in three stages of formula milk powder at different storage periods (0, 9, 18, and 24 months) as markers to distinguish formula milk powder at different storage stages. By predicting the content of these six key flavor compounds, the study aims to predict and provide early warning of the degree of flavor deterioration.

[0058] Precise correlation mechanism analysis: Existing technologies cannot quantitatively correlate lipid metabolism and sensory quality. This invention integrates lipidomics and volatile compound data to clarify the driving role of the glycerophospholipid metabolic pathway in flavor deterioration, screens out 7 pivot lipids (PG 16:1 / 16:1, PE 10:0e / 6:0, LPE 16:0, LPG 16:0, LPI 16:0, LPC 16:0, LPC14:0), and establishes a lipid-flavor quantitative prediction model, significantly improving detection specificity and reproducibility.

[0059] Database-driven process optimization: Construct a lipid profile database for formula milk powder, covering the dynamic concentration range of 37 types of lipids (1,814 types) and associated flavor biomarkers, supporting real-time comparison and production process adjustment, and accelerating the entire process of lipid prediction during the shelf life. Attached Figure Description

[0060] Figure 1 A flowchart illustrating the isolation and characterization of exosomes in formula milk powder at different storage stages is shown.

[0061] Figure 2 The particle size distribution of exosomes isolated from formula milk powder stored for different periods is shown. A0, A9, A18, and A24 represent exosomes obtained from stage 1 formula milk powder stored for 0, 9, 18, and 24 months; B0, B9, B18, and B24 represent exosomes obtained from stage 2 formula milk powder stored for 0, 9, 18, and 24 months; and C0, C9, C18, and C24 represent exosomes obtained from stage 3 formula milk powder stored for 0, 9, 18, and 24 months.

[0062] Figure 3Western blot analysis shows the marker proteins of exosomes isolated from formula milk powder with different storage times.

[0063] Figure 4 Transmission electron microscopy (TEM) images of exosomes isolated from formula milk powder stored for different times are shown.

[0064] Figure 5 The results of lipid species identification using LipidSearch 4.2 are shown.

[0065] Figure 6 The results of lipid profile analysis of exosomes are shown.

[0066] Figure 7 The results of differential abundance lipid (DAL) profiling analysis of exosomes isolated from three stages of formula milk powder are shown.

[0067] Figure 8 The table shows all flavor compounds detected in stage 3 infant formula (A) and their classification and content (B); six flavor compounds selected by a random forest model (C); and a flavor wheel constructed based on the sensory characteristics of the six flavor compounds (D).

[0068] Figure 9 The following diagrams are shown: a lipid co-expression network heatmap (A) of weighted lipid co-expression network analysis (WLCNA), the correlation between six identified lipid modules and volatile compounds that cause off-odors (B), the relationship between lipid modules and different storage times (C), and the changes of seven core lipids within the MEgreen module (green module) (D).

[0069] Figure 10 The study presents a quantitative descriptive analysis of eight sensory characteristics of (A) Stage 1, (B) Stage 2, and (C) Stage 3 formula milk powders, as well as (D) the changing trend of consumer acceptance of formula milk powders with the extension of storage time at different stages.

[0070] Figure 11 The oxidative stress status of Raw 264.7 cells after different exosome (EV) interventions is shown. (A) Cell viability; (B) Malondialdehyde (MDA) content; (C) Superoxide dismutase (SOD) activity; and (D) Lactate dehydrogenase (LDH) activity.

[0071] Figure 12 The levels of immune cytokines in Raw 264.7 cells after different EV interventions are shown. (A) Relative expression level of IL-10; (B) Relative expression level of IL-12; and (C) Relative expression level of IL-1β.

[0072] Figure 13 A flowchart is shown for prediction and verification using the multivariate linear prediction equation established in Example 5. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to specific embodiments. The described embodiments are only some, not all, of the embodiments of this invention.

[0074] Current monitoring of flavor deterioration in infant formula mainly focuses on the static analysis of end products (such as volatile compounds like aldehydes and ketones). However, such methods only reflect the end products of lipid oxidation and cannot reveal the dynamic changes in lipid metabolism during storage and their driving mechanisms on flavor.

[0075] Exosomes, as important carriers of bioactive molecules from milk, directly influence membrane stability, signal transduction, and interactions with receptor cells through their lipid composition. Existing separation techniques, such as differential ultracentrifugation and the sucrose gradient method, exhibit discrepancies in efficiency when separating exosomes from dairy products, particularly regarding exosome yield and purity. Furthermore, these two techniques (differential ultracentrifugation and the sucrose gradient method) still present significant challenges when processing infant formula. For instance, protein denaturation and aggregation caused by industrial processing complicate exosome separation and characterization, limiting the comprehensive and accurate quality assessment of exosomes in infant formula.

[0076] Traditional physicochemical detection methods (such as SDS-PAGE and NTA) can characterize the particle size and protein content of exosomes, but they cannot identify specific changes in lipid molecules (such as Cer and Hex2Cer), leading to missed detection of key biomarkers.

[0077] Finally, the dynamic evolution of the lipid profile of exosomes in formula milk powder and its molecular correlation with flavor deterioration are not yet clear in the existing technology. Traditional separation and detection methods cannot effectively capture key biomarkers of exosome lipid metabolism, which limits the precision and accuracy of flavor quality monitoring of formula milk powder during storage.

[0078] Unless otherwise stated, the raw materials and equipment used in this invention are all commonly used in the art, and the methods used in this invention are all conventional methods in the art.

[0079] Unless otherwise stated, parts are by weight, % is by weight, and temperature is in degrees Celsius. The following examples utilize the following raw materials and methods.

[0080] Example 1

[0081] 1-1. Isolation of exosomes from infant formula

[0082] Sample preparation: The infant formula milk powder used was purchased from Yili Dairy Co., Ltd.

[0083] Infant formula milk powder of different stages (Stage 1, Stage 2, and Stage 3) and different storage times (0, 9, 18, and 24 months stored under constant temperature and humidity conditions) were reconstituted according to the manufacturer's instructions and the formula ratio. Specifically, 12.9g of infant formula milk powder was dissolved in 90mL of deionized water (Milli-Q EQ 7000, Millipore, USA) to obtain 100mL of solution. The obtained solution was mixed with 0.05% rennet and 0.3% CaCl2, incubated at 4°C for 30 minutes, and after repeated mixing, centrifuged at 4000rpm for 15 minutes to obtain whey supernatant. Then, gradient ultracentrifugation was performed: the obtained whey supernatant was centrifuged at 16,500×g for 30 minutes at 4°C to remove fat, casein, residual rennet, and other impurities; the obtained supernatant was centrifuged at 90,000×g for 60 minutes at 4°C using a HIMAC CP100NX ultracentrifuge (Hitachi, Tokyo, Japan) to remove large particulate impurities; then, 10 mL aliquots of the supernatant were ultracentrifuged at 135,000×g for 90 minutes at 4°C, and the supernatant was discarded to obtain crude exosomes. Immediate purification: the obtained crude exosomes were resuspended in phosphate-buffered saline (PBS) and washed by ultracentrifugation at 135,000×g for 60 minutes to ensure purity; the obtained supernatant was transferred to a 100 kDa ultrafiltration membrane and filtered twice at 3,000×g to further improve purity, finally obtaining infant formula exosomes (IFEVs), which were stored at 4°C for later use. The exosomes from stage 1, 2, and 3 formula milk powders were grouped into three groups: Group A (A0, A9, A18, and A24 represent exosomes obtained from stage 1 formula milk powder stored for 0, 9, 18, and 24 months), Group B (B0, B9, B18, and B24 represent exosomes obtained from stage 2 formula milk powder stored for 0, 9, 18, and 24 months), and Group C (C0, C9, C18, and C24 represent exosomes obtained from stage 3 formula milk powder stored for 0, 9, 18, and 24 months). The flowchart for exosome isolation and extraction is shown below. Figure 1 As shown.

[0084] Exosome yield: 12.9g of infant formula yields 100mL of infant formula solution, which, after extraction, yields 10mL of exosomes. The exosome concentration is 8.79e+12 particles / mL. Therefore, the exosome yield can be calculated as: 8.79e+12 particles / mL * 10mL / 12.9g = 6.81e+12 particles / g. That is, 1.0g of infant formula yields 6.81 x 10^12 exosome particles.

[0085] 1-2. Characterization of exosomes in formula milk powder

[0086] To confirm that the obtained formula milk powder exosomes are a particulate mixture mainly composed of exosomes, the exosomes prepared in Example 1-1 were characterized and verified according to the International Guidelines for Extracellular Vesicle Research (MISEV Guidelines):

[0087] (1) Particle size, concentration, and purity:

[0088] The particle size distribution and concentration of the prepared EVs were determined using nanoparticle tracking analysis (NTA, Malvern NS300, UK) and the corresponding software NTA 3.4. All EV samples were diluted with 1×PBS buffer and evaluated in triplicate at 22°C. Results are shown in […]. Figure 2 middle.

[0089] like Figure 2 As shown, the exosome particles obtained from stage 1, 2, and 3 formula milk powders (0, 9, 18, and 24 months) mainly fall within the 100-200 nm particle size range, which is a key characteristic of formula milk powder exosomes. Furthermore, based on the numerical variation range of the particle size distribution map, the proportion of larger particles in the exosomes increases with prolonged storage time.

[0090] The protein concentration of exosomes was determined using a bispyridine carboxylic acid protein assay kit (Beyotime Biotechnology Co., Ltd., Shanghai, China) according to the manufacturer's instructions. The purity of the prepared exosomes was expressed as the ratio of particle number to protein concentration. The highest value among the 12 samples was used: the purity of C18 was 3.39175E+12 particles / μg protein. The purity of the 12 samples is shown in Table 1.

[0091] Table 1. Purity of exosomes (unit: particles / μg protein)

[0092]

[0093] (2) Protein biomarker detection: To further characterize the prepared exosomes, Western blot analysis was performed according to the MISEV 2018 guidelines (Théry et al., 2018, Minimal information for studies of extracellular vesicles 2018 (MISEV2018): a position statement of the International Society for Extracellular Vesicles and update of the MISEV2014 guidelines. Journal of Extracellular Vesicles, 7, 1535750) to assess the expression of protein biomarkers for three representative exosome pairs: TSG101, CD9, and CD81, ensuring the integrity of the exosomes. The following antibodies were purchased from Abcam and used for Western blot analysis: anti-TSG101 antibody (Cat. No. ab125011), anti-CD9 antibody (Cat. No. ab236630), and anti-CD81 antibody (Cat. No. ab109201). Results were captured using an automated gel imaging system (SmartGel7500, Sensi Technologies, China) and analyzed using ImageJ software. The results are shown in Figure 3.

[0094] The results showed that all tested exosomes were positive for the aforementioned protein markers, although their expression levels varied. Furthermore, intensity analysis using ImageJ software revealed that the expression levels of these three protein markers in group C exosomes were consistently significantly lower than those in groups A and B.

[0095] (3) Morphological observation:

[0096] The obtained EVs were diluted with PBS to prevent particle aggregation. The bilayer membrane structure and cup-and-disc morphology (approximately 150 nm in diameter) of the exosomes were verified by transmission electron microscopy (TEM). Specifically, exosomes were applied to a copper grid coated with a formvar film and incubated at 4°C for 1 minute. Subsequently, the grid was placed in a droplet of 2% phosphotungstic acid for 1 minute and imaged using a Hitachi H-7650 transmission electron microscope (Hitachi, Tokyo, Japan) at 200 kV. The results are shown in... Figure 4 In the data, AD represents 0, 9, 18, and 24 months of segment 1, and EH represents 0, 9, 18, and 24 months of segment 2.

[0097] like Figure 4As shown, all EVs exhibited a bilayer membrane structure, with round or elliptical shapes, clear boundaries, and a concave center, a typical "cup-and-disc" morphology, with a diameter of approximately 150 nm, largely consistent with NTA results. Furthermore, prolonged storage of formula milk powder led to gradual morphological changes in EVs, transforming them from strictly round to elliptical or even amorphous shapes, with a decrease in the number of observable particles. This observation is consistent with the previously determined total particle number for each exosome type. These morphological changes may be related to alterations in lipid composition and bilayer membrane structure.

[0098] 1-3. Non-targeted lipidomics analysis

[0099] (1) Lipid extraction: An isotopic internal standard composed of 14 isotopes was added to 50 μL of PBS solution of exosome (EV) particles prepared in Example 1-1. Ester, 18:1(d7)MG, 15:0-18:1(d7)DG, 15:0-18:1(d7)-15:0TG, 18:1(d9)SM, Cholesterol (d7); then 500 μL of 70% methanol and 800 μL of 80% methyl tert-butyl ether were added sequentially. After ultrasonic extraction, the mixture was centrifuged at 6,000 rpm for 15 minutes, and the upper organic phase was collected and concentrated by nitrogen blowing to obtain the exosome lipid extract.

[0100] (2) Analysis by ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS, Thermo Fisher Scientific):

[0101] The exosome lipid extract obtained above was redissolved in 200 μL of a 90% isopropanol / acetonitrile mixture, centrifuged at 14,000 × g for 15 min, and then 3 μL of the supernatant was injected into the chromatographic system. Separation was performed using a reversed-phase CSH C18 column (Waters, MA, USA). Chromatographic conditions: CSH C18 column (1.7 μm, 2.1 × 100 mm), mobile phase of acetonitrile-water (6:4, v / v) containing 0.1% formic acid and 0.1 mM ammonium formate (solvent A) and acetonitrile-isopropanol (1:9, v / v) (solvent B), gradient elution (40% solvent B initial, 300 μL / min flow rate). Mass spectrometry data were acquired using a Q-Exactive Plus mass spectrometer in both positive and negative ion modes. Mass spectrometry conditions: Q-Exactive Plus mass spectrometer, positive / negative ion mode switching, scan range m / z 200–1800.

[0102] (3) Lipid data processing: Lipid types were identified using LipidSearch 4.2 (Thermo Fisher, MA, USA), and multivariate statistical analysis (orthogonal partial least squares discriminant analysis (OPLS-DA)) was performed using SIMCA16.1 to screen differentially expressed lipids (DELs, VIP>1) and key metabolic pathways (such as glycerophospholipid metabolism). Results are as follows: Figure 5-7 As shown.

[0103] Lipid types of exosomes

[0104] like Figure 5 As shown, a total of 37 classes and 1,814 lipids were identified in all exosomes. The total lipid content of exosomes from stage 3 formula milk powder was significantly lower than that from exosomes from stage 1 and stage 2 formula milk powder.

[0105] Differentially expressed lipids (DELs) in exosomes

[0106] The lipid profiles of the prepared exosomes were analyzed by combining positive and negative ion modes, and the results are as follows: Figure 6 As shown in the figure, orthogonal partial least squares discriminant analysis (OPLS-DA) revealed significant lipidomical differences among exosomes (EVs) of formula milk powder from different stages, as evidenced by the clear separation in the score plot. Figure 6 A in the model (R). 2 X, R 2 Y and Q 2 The values ​​were 0.852, 0.748, and 0.281, respectively. This indicates that the model has strong explanatory power and moderate predictive ability. The corresponding loading plots show a stronger correlation between lipid molecules and A0, B18, and B24. Figure 6(B in the original text). The model stability was verified through 200 permutation tests, and no overfitting was found. R... 2 The intercept value is 0.507, Q 2 The intercept value is -0.402 ( Figure 6 (C in the text). Receiver operating characteristic (ROC) analysis showed that the area under the curve (AUC) of EVs exceeded 0.99 at different storage times and processing stages. Figure 6 The D in the figure indicates that the OPLS-DA model has high diagnostic potential in lipid classification of EVs. A total of 532 differentially expressed lipids (DELs) were identified, with a projection importance (VIP) > 1, serving as key variables for EV classification. These included 105 sphingolipids (Cer) and 78 triglycerides (TG). Further analysis of sphingolipid chain length and saturation showed significant differences in lipid content among segments containing 30-40 carbon atoms. Most EVs mainly contain lipids with a single double bond, and an increase in the number of double bonds was associated with a decrease in lipid content. Among the identified DELs, phosphatidylglycerol (PG) (18:1 / 18:1) had the highest VIP score, followed by cardiolipin (CL) (18:1 / 16:1 / 18:1 / 16:1) and PG (18:1 / 18:2). Pathway analysis mapped 17 DEL classes to 43 KEGG pathways, including C00195 (sphingolipids), C00422 (triglycerides), C00157 (phosphatidylcholine), C02737 (phosphatidylcholine), and C01194 (phosphatidylinositol). Figure 6 KEGG enrichment analysis identified 24 significantly enriched pathways (p<0.05), with the glycerophospholipid metabolism pathway showing the most enriched lipid types and the smallest p-value. Other enriched pathways included choline metabolism, sphingolipid metabolism, leishmaniasis, and necrotizing cell death in cancer, highlighting the functional relevance of lipid alterations in EVs in dairy products.

[0107] Differential abundance lipid (DAL) profile of exosomes

[0108] Although OPLS-DA identified 532 key differentially expressed lipids (DELs) that affect overall classification, Figure 7 Volcano map in China ( Figure 7 The AC analysis provided a deeper understanding of lipid class variations within and between exosome groups. The distribution of lipid classes differed across treatment stages. Pairwise comparisons within group C showed that the number of differentially abundant lipids (DALs) was greater than in groups A and B. Figure 7 (AC in the text). Notably, the comparison between C18 and C9 identified 62 significantly upregulated DALs and 1,167 significantly downregulated DALs. Figure 7 (B in the middle).

[0109] Venn diagram analysis of DALs stored at different times within the same group revealed 401 shared DALs across the three comparison groups in group C, including 68 Hex2Cer and 67 Hex1Cer species. Figure 7 (D in the middle).

[0110] These findings indicate that the lipid composition changes in stage 3 formula EVs are more pronounced across all three stages, involving a wider range of lipid categories, and highlight the impact of processing and storage on the lipid profile of exosomal tissues.

[0111] Example 2: Dynamic monitoring of volatile compounds in formula milk powder during storage

[0112] Evolution of volatile compounds in stage 3 formula milk powder during storage

[0113] Given that the lipid profile of exosomes from stage 3 formula milk powder changes more significantly with storage time than that of exosomes from stage 1 and stage 2 formula milk powder (as demonstrated in Example 1), in order to reveal the potential mechanism of the dramatic changes in lipid profile of exosomes from stage 3 formula milk powder under different storage times, headspace solid phase microextraction (HS-SPME) combined with gas chromatography-mass spectrometry (GC-MS) was used to analyze volatile compounds in stage 3 formula milk powder under different storage times.

[0114] Technical Procedure: Headspace Solid-Phase Microextraction (HS-SPME): 8 mL of reconstituted stage 3 formula milk powder (the 8 mL mentioned here is prepared using the same method as described in Example 1; that is, after dissolving the stage 3 formula milk powder according to the proportions in Example 1, only 8 mL was used for volatile compound analysis) was transferred to a 20 mL headspace vial. 10 μL of 2-methyl-3-heptanone (150 μg / mL) was added as an internal standard, and the vial was sealed. The sample was then equilibrated at 45°C for 15 minutes to allow the volatile compounds to reach equilibrium in the headspace. Solid-phase microextraction (SPME) was performed using divinylbenzene / carboxymethyl cellulose / polydimethylsiloxane (DVB / CAR / PDMS) fiber (50 / 30 μm, Supelco, Bellefonte, PA, USA) pretreated at 250°C for 1 hour to adsorb the volatile compounds. Volatile compounds were analyzed using a gas chromatography-mass spectrometry (GC-MS) system (Agilent 7890A-5975C, CA, USA) equipped with a DB-WAX column (30 m × 0.25 mm, 0.25 μm). Thermal desorption of volatile compounds was performed in split mode at 250 °C for 5 min. Helium was used as the carrier gas at a constant flow rate of 1 mL / min.

[0115] Data analysis: Qualitative analysis was performed using the NIST spectral library, retention index, and standards, followed by quantitative analysis using the internal standard method.

[0116] Using headspace solid-phase microextraction (HS-SPME) combined with gas chromatography-mass spectrometry (GC-MS), 17 flavor compounds were identified in three stages of infant formula milk powder at different storage times. Figure 8 (A) In general, the concentrations of volatile compounds were higher for longer storage times (C18 and C24) than for shorter times (C0 and C9). Notably, hexanal, E-2-heptenal, 1-pentanol, E-2-hexenal, and 1-penten-3-one accumulated over time. Among the identified volatile compounds, aldehydes were the most abundant class, reaching a peak of 0.80 μg / ml at the C18 stage. Figure 8 (B) Ketones and alcohols showed a brief decrease in the C9 stage followed by a sharp increase, while furans showed a continuous decrease throughout storage. Except for furans, the concentrations of all compound categories in the C24 stage were significantly higher than those in the C9 stage, indicating a significant change in the overall characteristics of volatile compounds during storage.

[0117] Random forest regression analysis was used, based on the average accuracy reduction ( Figure 8 In the study of off-odors, six key volatile biomarkers were identified in stage 3 infant formula at different storage periods (0, 9, 18, and 24 months). These included hexanal (grassy smell), E-2-pentenal (fatty smell), heptanal (vegetable oil smell), octanal (cheese smell), nonanal (lemon smell), and E,E-2,4-heptadienal (nutty smell). These compounds can serve as markers for differentiating infant formula at different storage stages (IF), particularly related to off-odor development. A molecular sensory off-odor wheel was constructed based on their odor descriptors. Figure 8 The D in the text further highlights their contribution to the off-flavor of formula milk powder.

[0118] Example 3: Exosome lipid-formula milk powder flavor association modeling and database construction

[0119] Weighted Lipid Co-expression Network Analysis (WLCNA): Using the 532 differentially expressed lipids (DELs) identified in Examples 1-3 as input data, a co-expression network (soft threshold = 7) was constructed to identify modules strongly associated with flavor compounds. Figure 9 (A) Among them, the MEgreen module (green module) showed a positive correlation with all six flavor compounds (hexanal, E-2-pentenal, heptanal, octanal, nonanal, and 2,4-heptadienal). Figure 9 (B and C in the text). Further, connectivity analysis (threshold 10.0) within the MEgreen module identified seven key lipid molecules as the most important lipid molecules in the MEgreen module, and their abundance was positively correlated with storage time (p<0.05). Figure 9(C and D in the text). The seven key lipids are: phosphatidylglycerol (PG) (16:1 / 16:1), phosphatidylethanolamine (PE) (10:0e / 6:0), lysophosphatidylethanolamine (LPE) (16:0), lysophosphatidylglycerol (LPG) (16:0), lysophosphatidylinositol (LPI) (16:0), lysophosphatidylcholine (LPC) (16:0), and lysophosphatidylcholine (LPC) (14:0). Therefore, the concentrations of these lipids can serve as markers of flavor deterioration in infant formula. They also indicate the storage condition of the infant formula.

[0120] Example 4: Detection of Off-Flavor in Formula Milk Powder Based on Consumer Acceptability and Sensory Evaluation

[0121] To investigate whether the off-odor of formula milk powder caused by changes in exosome lipids can be perceived by the human body, a consumer survey was conducted to evaluate the flavor characteristics of formula milk powder at different storage periods.

[0122] Thirty-two consumers (aged 15–25 years) were randomly selected and their overall acceptance was assessed using a 9-point pleasure scale. Twelve judges (7 women and 5 men, aged 25–35 years) with expertise in sensory evaluation were selected according to ISO 8586-1:1993. Formula milk powder samples were prepared according to the product instructions. A certain amount of milk powder was weighed and added to warm water at 45–50°C according to the mixing ratio specified on the packaging. The samples were then dispensed into 20 ml food-grade PET bottles and incubated at 45°C for 40 minutes before evaluation. To ensure consistency and reproducibility of the evaluation, each judge was trained using an independent sample set. Sensory evaluation consisted of three stages: First, expert sensory evaluators provided descriptive terminology for the infant formula. Second, after discussing different aroma scoring criteria, eight sensory characteristics were selected for analysis. In the third stage, the sensory review panel members evaluated the aroma quality of each sample. To ensure consistent sample temperature, samples were preheated in a 45°C water bath for 30 minutes before each evaluation. For the formal analysis, 20 ml samples of formula milk powder were dispensed into 30 ml disposable tasting cups. All samples were randomly coded with three digits and presented to each rater in triplicate. Sensory characteristics included aftertaste, rustiness, saltiness, fishiness, rancidity, over-steaming, overall flavor, and frankincense. Detailed descriptions and comprehensive definitions of each sensory characteristic are shown in Table S3. The rating system was as follows: 0 indicates no odor, 1–3 indicates weak, 4–5 indicates moderate, 6–8 indicates relatively strong, and 9 indicates very strong. A 1-minute sensory recovery interval was provided between samples, during which coffee beans were offered for olfactory recovery. All participants signed informed consent forms. This study has been approved by the Research Ethics Committee of Beijing Technology and Business University (Approval No.: 85, 2025).

[0123] The data results show ( Figure 10As storage time increased (0, 9, 18, 24 months), the acceptability of formula milk powder continued to decline, indicating that the appearance of off-flavors in formula milk powder had reached the point of human perception. Furthermore, a sensory evaluation expert group assessed eight sensory characteristics (overall flavor, rusty flavor, milky flavor, salty flavor, fishy flavor, oxidized fat flavor, cooked flavor, and aftertaste) of formula milk powder with different storage times using a 9-point scale. The results showed that as storage time increased, the scores of positive sensory characteristics continued to decrease, while the scores of negative sensory characteristics continued to increase. In conclusion, the above results indicate that the flavor differences in formula milk powder caused by exosome lipid oxidation have reached the threshold of human perception.

[0124] Example 5: Verification of intestinal immune and oxidative stress indicators after exosome lipid oxidation caused off-flavor in formula milk powder

[0125] Given that lipid remodeling of the aforementioned EVs can lead to significant changes in the sensory properties, off-flavor formation, and consumer acceptance of formula milk powder during storage, it is necessary to further study its effects on immune cytokines and oxidative stress.

[0126] To confirm that flavor changes caused by lipid oxidation of exosomes from formula milk powder do not lead to safety and functional issues, the intestinal immune and oxidative stress indices were evaluated after intervention with exosomes of the same particle number (prepared in Example 1) in RAW 264.7 cells. When exosomes from formula milk powder prepared in Example 1 (dissolved in PBS) at the same particle concentration (1*10^9 particles / mL) were added to Raw 264.7 cells, no statistically significant differences were observed in cell viability, malondialdehyde (MDA) content, superoxide dismutase (SOD) activity, and lactate dehydrogenase (LDH) activity. Figure 11 This indicates that none of the exosomes affected the safety and functionality of the formula milk powder. qPCR testing also revealed no significant differences in several important immune cytokines, including IL-10, IL-12, and IL-1β. Figure 12 This indicates that lipid changes in EVs do not affect the safety and functionality of formula milk powder.

[0127] Example 6: Construction and application of a multiple linear equation for indicating the content of flavor compounds in formula milk powder using exosome lipid content.

[0128] To visually demonstrate how exosome lipid content indicates the content of flavor compounds in infant formula, a multiple linear equation was constructed using R language (Table 2). This embodiment employs multiple linear regression (MLR) analysis to establish a predictive model for the relationship between lipid content in exosomes and volatile compound content in infant formula. The specific process is as follows:

[0129] a. Data preparation

[0130] The experimental data included exosome samples from formula milk powder stored for four time periods (0, 9, 18, and 24 months), with three technical replicates at each time point. Content data for seven key lipid molecules (LPC(16:0), LPG(16:0), PG(16:1 / 16:1), PE(10:0e / 6:0), LPI(16:0), LPC(14:0), LPE(16:0)) and six volatile flavor compounds (hexanal, E-2-pentenal, heptanal, octanal, nonanal, and 2,4-heptadienal) were collected. Raw data were stored in Excel format and imported into the R statistical environment (v4.2.3) for processing. All data points (n=12) were retained during the analysis; no outlier removal or data transformation was performed. Multicollinearity was assessed using the variance inflation factor (VIF). A VIF > 10 indicated significant multicollinearity (all VIF values ​​in this study were < 5), indicating a reasonable model structure.

[0131] b. Model Building

[0132] Establish independent multiple linear regression models for each volatile compound:

[0133]

[0134] Y k The target volatile compound content (k = 1-6); X i The content of 7 lipid molecules (i = 1-7); β0 is the intercept term; β i represents the regression coefficients of each lipid; ∈ represents the random error term.

[0135] The model parameters were estimated using the Ordinary Least Squares (OLS) method, and the model form was: lm(Y ~ X1 + X2 + ... + X7, data = dataset).

[0136] c. Model Evaluation

[0137] For each model, the following statistical indicator was calculated: coefficient of determination (R²). 2 ): The proportion of variance explained by the quantitative model. Adjusted R-value: 2 ): Corrects for the influence of the number of independent variables. F-statistic and its p-value: Test the overall significance of the model (α = 0.05). Coefficient t-test p-value: Assess the significance of each lipid predictor.

[0138] d. Calculation tools

[0139] All statistical analyses were performed in the R language environment (R Core Team, 2023), mainly using the following packages: readxl (v1.4.3) for data import; stats (v4.2.3) core package for regression modeling; and writexl (v1.4.2) for exporting structured results. The results are shown in Table 2.

[0140] Table 2. Multiple linear equations for predicting the content of volatile compounds in infant formula.

[0141]

[0142]

[0143] The prediction equations shown in Table 2 above indicate that the prediction equation for each volatile compound consists of the content of the seven lipids multiplied by the independent variable coefficients and the intercept. Specifically, the predicted concentration of each volatile compound is calculated as follows: Based on the prediction equations in Table 2, the intercept corresponding to each volatile compound is added, and the content of each of the seven lipids calculated from the tested formula milk powder is multiplied by the independent variable coefficient of the corresponding lipid in Table 2 and summed to obtain the predicted concentration of each volatile compound. The calculation formula is as follows:

[0144]

[0145] Y k Let represent the content of each volatile compound, where k = integers from 1 to 6; β0 is the intercept term; β i The independent variable coefficients for each lipid are shown in Table 2. X i represents the content of each lipid calculated from the formula milk powder to be tested, where i = integers from 1 to 7; e is the residual error term, representing the part not explained by the model. It is usually assumed that e follows a normal distribution with a mean of 0, and is used to capture the difference between the observed and predicted values. It is not a fixed value, but a random residual for each data point, so no specific value is provided in the table.

[0146] Table 2 shows that the lipid content of exosomes and the flavor compound content of formula milk powder exhibit a good linear relationship, which can be directly used in production practice to predict the condition of formula milk powder. 2 A value greater than 0.8 indicates that the model is robust and controllable.

[0147] Application: Microscopic prediction is performed using the prediction equations in Table 2.

[0148] First, in actual production, the exosome lipid content of the milk powder to be tested, obtained by mass spectrometry, rapid detection technology, test strips, etc., can be substituted into the prediction equation in Table 2 to calculate the predicted value of the content of each volatile compound. Then, it is compared with the threshold of volatile compounds specified in the literature (as shown in Table 3). If the predicted value exceeds the threshold and shows a multiple relationship, the flavor of the formula milk powder has deteriorated; the larger the multiple relationship, the more severe the deterioration. The prediction equation in Table 2 can serve as a real-time judgment standard and basis for the storage and production status of formula milk powder.

[0149] Table 3. Summary of concentration thresholds of six flavor compounds in infant formula

[0150]

[0151]

[0152] Secondly, the prediction equation in Table 2 shows a good linear relationship between exosome lipid content and formula milk powder flavor substance content. When the lipid content of two or more products differs by two times or more, it can be intuitively judged that the product with a higher lipid content has a deeper degree of quality deterioration than the other product.

[0153] Finally, the advantages of the prediction equations in Table 2 lie in their simplicity, intuitiveness, and objectivity. Simplicity is reflected in the fact that previous factory practices and laboratory-scale measurements required quantifying the specific content of over two thousand lipids in exosomes, which was not only time-consuming and labor-intensive but also yielded inconsistent results. Using the prediction equations in Table 2, only seven exosomal lipids need to be measured, significantly reducing the waste of human, material, and financial resources and substantially improving production or quality control efficiency. Furthermore, the R-values ​​of all equations are [not specified in the original text]. 2 All values ​​are at least greater than 0.8, indicating that the model is robust, controllable, and repeatable. Its intuitiveness lies in the fact that the outputs of this equation are all numerical, and the results can be obtained simply by comparing them with the thresholds (industry standards) in Table 3. Furthermore, it does not require complex computer simulations; predictions can be achieved using only simple tools such as calculators. Its objectivity lies in the fact that past factory practices and laboratory-scale measurements have relied entirely on manual smelling and visual observation by quality inspectors to determine the condition of formula milk powder, largely conforming to empirical judgment and selection. This judgment and selection is limited by many factors such as age, gender, physiological state, educational background, dietary habits, and taste preferences, often resulting in "multiple results from multiple people." This equation, through screening and modeling, largely solves this problem, remaining unaffected by human judgment.

[0154] Human prediction of flavor deterioration in formula milk powder during storage

[0155] The artificial prediction criteria for flavor deterioration of infant formula during storage are shown in Table 4 below. It has been confirmed that Table 4 is the standard for judging flavor deterioration of infant formula by humans, which comes from subjective judgment, that is, through the five senses to perceive changes in infant formula.

[0156] Table 4. Grading criteria for quantitative descriptive analysis scores of formula milk powder at different stages

[0157]

[0158]

[0159] Compared with the manual prediction method in Table 4, the present invention first uses the prediction equation in Table 2 to calculate the predicted concentration of each flavor compound, and then compares it with the threshold in the monograph shown in Table 3, thereby predicting whether the flavor of formula milk powder will deteriorate. This method has higher precision and accuracy.

[0160] Example 7 Verification

[0161] Reliability verification of the multiple linear prediction equation established in Example 6 in Feihe Xingfeifan formula milk powder

[0162] 1. Verification Objective

[0163] The applicability of the multiple linear prediction equation established in Example 6 to actual production was tested, and its reliability, specificity, and sensitivity were evaluated. Reliability: the degree of deviation between predicted and measured values; Specificity: the ability to distinguish between normal and deteriorated products; Sensitivity: the ability to detect early lipid oxidation signals.

[0164] 2. Validation Materials and Methods

[0165] 2.1 Sample Preparation

[0166] Sample: Feihe Xingfeifan formula milk powder (stage 1, stage 2, stage 3).

[0167] Storage conditions: Store at a constant temperature of 25℃. Sampling time points: 0, 3, 6 / 9, 12, 15, 18, 21, and 24 months.

[0168] Exosome extraction: Exosomes from formula milk powder were separated according to the method in Example 1.

[0169] 2.2 Key indicators are shown in Table 5.

[0170] Table 5

[0171] Test items method Instrument Model 7 types of lipid content UHPLC-MS / MS Thermo Q Exactive Plus 6 flavor compounds HS-SPME-GC-MS Agilent 7890A-5977B

[0172] 2.3 Prediction and Validation Process

[0173] See the flowchart for prediction and verification. Figure 13 .

[0174] 3. Validation Results and Data Analysis

[0175] 3.1 Prediction accuracy (reliability)

[0176] Table 6 shows the stage 2 formula milk powder samples for 3 months, 9 months, 15 months, 21 months, and 24 months.

[0177] Table 6

[0178]

[0179] In addition, the stage 3 formula milk powder stored for 18 months is shown in Table 7.

[0180] Table 7

[0181]

[0182] 3.2 State assessment (specificity and sensitivity)

[0183] (1) Based on the threshold and sensory evaluation in Table 3, the different time points of the stage 2 formula milk powder were verified as shown in Table 8.

[0184] Table 8

[0185]

[0186] Key performance indicators (based on the above 9 samples):

[0187] Sensitivity: 100% (9 out of 9 deteriorated samples were accurately detected)

[0188] Specificity: 100% (correctly identified in 6 out of 6 normal samples)

[0189] (2) Based on the threshold and sensory evaluation in Table 3, the different time points of the three-stage formula milk powder were verified as shown in Table 9.

[0190] Table 9

[0191] Storage time (months) Predicted value of hexanal (μg / mL) Has the threshold been exceeded? Sensory evaluation results Model consistency judgment 0 0.02 no Odorless yes 6 0.08 It is (1.6 × threshold) Slight oily smell yes 18 1.85 Yes (37 × threshold) Strong hyaluronic acid smell yes

[0192] Key performance indicators:

[0193] Sensitivity: 100% (3 / 3 of the deteriorated samples were accurately detected)

[0194] Specificity: 91.7% (11 / 12 normal samples were correctly identified; 1 false positive was due to cross-contamination)

[0195] 4. Verification Conclusion

[0196] Reliability: Predicted values ​​and measured values ​​are in high agreement (average error <7%), R 2A score of >0.75 demonstrates the model's robustness; specificity: the model can eliminate over 95% of interfering factors and accurately distinguish between normal and deteriorated products; sensitivity: an early warning can be triggered when the lipid concentration difference is ≥1.5 times (3-4 weeks earlier than detection by human sensory means).

[0197] Industrial value: This model replaces traditional manual quality inspection, reducing testing costs by 80% and providing early warnings 30 days earlier.

Claims

1. A method for dynamically monitoring whether the flavor of infant formula deteriorates during storage, comprising the following steps: (a) Non-targeted lipidomics analysis of exosomes: Non-targeted lipidomics analysis was performed on the exosomes of the prepared formula milk powder, and 532 differentially expressed lipids were obtained. (b) Dynamic monitoring of volatile compounds during the storage of formula milk powder. Six flavor compounds were screened from the three stages of formula milk powder at different storage periods, including hexanal, E-2-pentenal, heptanal, octanal, nonanal and E,E-2,4-heptadienal. (c) Exosome lipid-formula milk powder flavor association modeling and database construction: c1) Weighted lipid co-expression network analysis (WLCNA): Using the 532 differentially expressed lipids obtained in step (a) as input data, a co-expression network was constructed to identify modules strongly associated with flavor compounds. Among them, the 58 lipids classified as MEgreen modules by WLCNA showed a positive correlation with the 6 flavor compounds mentioned in step (b). c2) Through connectivity analysis in the MEgreen module, seven key lipids were identified as the most important in the MEgreen module, and their abundance was positively correlated with storage time. The seven key lipids were phosphatidylglycerol 16:1 / 16:1, phosphatidylethanolamine 10:0e / 6:0, lysophosphatidylethanolamine 16:0, lysophosphatidylglycerol 16:0, lysophosphatidylinositol 16:0, lysophosphatidylcholine 16:0, and lysophosphatidylcholine 14:

0. These seven key lipids were used as biomarkers for the deterioration of the flavor of formula milk powder. (d) Construction of a multiple linear equation for the exosome lipid content indicator of the flavor compound content in formula milk powder: d1) Data preparation: The content data of the seven pivot lipids and the six flavor compounds were imported into the R statistical environment for processing; d2) Model construction: A multiple linear regression model was independently established for each of the six flavor compounds; d3) Model Evaluation: For each model, the following statistical indicators are calculated: Coefficient of Determination (R²): quantifies the proportion of variance explained by the model; Adjusted Coefficient of Determination (R²) 2 ): Correcting for the influence of the number of independent variables; F-statistic and its p-value: testing the overall significance of the model (α=0.05); Coefficient t-test p-value: assessing the significance of each lipid predictor; d4) Computational tools: All statistical analyses were performed in the R language environment to obtain the prediction equations; (e) The formula milk powder to be tested is separated into exosomes, exosome lipids are extracted, and the contents of 7 lipids are determined according to the method described above. The results are then substituted into the prediction equation of step d4) to obtain the predicted concentration values ​​of 6 flavor compounds. By comparing the predicted concentration values ​​of the 6 flavor compounds with the concentration threshold, the flavor state of the formula milk powder during storage and production is determined, thereby determining whether the flavor of the formula milk powder has deteriorated.

2. The method according to claim 1, characterized in that, The non-targeted lipidomics analysis described in step (a) includes: a1) Add an isotope internal standard to the prepared formula milk powder exosomes, then add methanol and methyl tert-butanol in sequence, extract by ultrasonication, centrifuge at 5,000-7,000 rpm for 15-40 minutes, collect the upper organic phase, concentrate by nitrogen blowing, and obtain exosome lipid extract. a2) The obtained lipid extract was analyzed by ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS / MS): Chromatographic conditions: A CSH C18 column was used, and the mobile phase consisted of acetonitrile-water (A) and acetonitrile-isopropanol (B) containing 0.1% formic acid and 0.1 mM ammonium formate, with gradient elution; Mass spectrometry conditions: Mass spectrometer, switching between positive and negative ion modes, scan range m / z 200-1800; a3) Data processing: Lipids were identified using LipidSearch 4.2, and a total of 37 classes and 1,814 lipids were identified. The 532 differentially expressed lipids were obtained by orthogonal partial least squares discriminant analysis (OPLS-DA).

3. The method according to claim 1, characterized in that, Step (b) includes: b1) Headspace solid-phase microextraction: The reconstituted formula milk powder is mixed with an internal standard, and solid-phase microextraction is performed using divinylbenzene / carbon molecular sieve / polydimethylsiloxane fiber to adsorb volatile compounds in the formula milk powder. b2) The adsorbed volatile compounds were analyzed using a gas chromatography-mass spectrometry system, and 17 flavor compounds were identified in the three stages of formula milk powder at different storage periods. b3) Using random forest regression analysis, based on the mean accuracy decrease value, 6 flavor compounds were screened from 17 flavor compounds, including hexanal, E-2-pentenal, heptanal, octanal, nonanal and E,E-2,4-heptadienal.

4. The method according to claim 1, characterized in that, The multiple linear regression model mentioned in step d2) is: Y k X represents the content of the target volatile compound, where k = an integer from 1 to 6; i The content of 7 lipid molecules is represented by β0, where i = integers from 1 to 7; β0 is the intercept term; β i ϵ represents the regression coefficients of each lipid; ϵ is the random error term. The model parameters are estimated using the least squares method, and the model form is: lm(Y ~ X1 + X2 + ... + X7).

5. The method according to claim 1 or 2, characterized in that, The formula milk powder exosomes are prepared by the following method, which includes the following steps: (1) Dissolve formula milk powder of different stages and storage time in water to obtain formula milk powder solution; (2) Pretreatment: Add enzyme and CaCl2 to the formula milk powder solution, mix well, incubate at 4°C for 30 minutes, mix repeatedly, and centrifuge at 4000-6000 rpm for 10-20 minutes to obtain whey supernatant; (3) Gradient ultracentrifugation: i) Centrifuge the whey supernatant obtained in step (2) at 16,000-17,000 ×g for 20-40 minutes to remove fat, casein, enzymes and other impurities to obtain supernatant; ii) Centrifuge the supernatant obtained in step i) at 80,000-100,000 × g for 50-70 minutes to remove large particulate impurities. iii) Centrifuge the supernatant obtained in step ii) at 130,000-140,000 ×g for 80-100 minutes, discard the supernatant, and obtain the crude exosome product; (3) Purification: The crude exosomes obtained were resuspended in phosphate buffer and purified by filtration through a 100 kDa ultrafiltration membrane to obtain the formula milk powder exosomes.

6. The method according to claim 5, characterized in that, The exosomes contain the following seven lipids as biomarkers associated with deterioration in the flavor of formula milk powder: phosphatidylglycerol 16:1 / 16:1, phosphatidylethanolamine 10:0e / 6:0, lysophosphatidylethanolamine 16:0, lysophosphatidylglycerol 16:0, lysophosphatidylinositol 16:0, lysophosphatidylcholine 16:0, and lysophosphatidylcholine 14:

0.

7. The method according to claim 5, characterized in that, The different stages are stage 1, stage 2 and stage 3, and / or the different storage times are 0, 9, 18 and 24 months.

8. The method according to claim 5, characterized in that, The purification described in step (3) includes: resuspending the obtained crude exosomes in phosphate buffer and washing them by ultracentrifugation at 135,000 × g for 60 minutes; then transferring the supernatant to a 100 kDa ultrafiltration membrane and filtering twice at 3,000 × g to obtain the formula milk powder exosomes.

9. The application of the method according to any one of claims 1 to 8 in the dynamic monitoring of flavor deterioration of formula milk powder during storage.

Images