Application of metabolite in sow vaginal secretion in early pregnancy diagnosis

Abstract

The invention provides an application of metabolites in sow vaginal secretion in early pregnancy diagnosis, and finds that three metabolic molecules can be used as markers for the early pregnancy diagnosis of sows for the first time on the basis of non-targeted metabonomics and machine learning conjoint analysis. According to the present invention, the metabolic markers are respectively INDOLEPROPIONIC ACID (indolepropionic acid), cis-Amonitate (cis-Amonitate) and 4-p-Coumaroylquinic acid (4-p-coumaroylquinic acid), the ROC curve AUC values of the metabolic markers are respectively 0.88, 0.92 and 0.77, and the metabolic markers have excellent diagnosis performance during the combination diagnosis. The pregnancy diagnosis is carried out by adopting the vaginal secretions of the sows 18 days after hybridization, the effective pregnancy diagnosis time is advanced by 5-8 days on the basis of a wave diagnosis method, and the method has important significance for reducing cost and increasing ultrasonic efficiency in live pig feeding.

Description

Technical Field

[0001] This invention belongs to the field of biotechnology diagnostics, specifically relating to the application of metabolites in sow vaginal secretions in early pregnancy diagnosis. Background Technology

[0002] In the pig farming industry, increasing the average annual number of weaned piglets (PSY) produced by breeding sows is a crucial goal for farmers. However, in actual production, various reasons lead to breeding failures in sows, resulting in prolonged non-pregnant periods and inevitably causing economic losses to pig farms. Therefore, timely diagnosis of sow pregnancy status and re-breeding non-pregnant sows are essential for increasing PSY and mitigating economic losses for pig farms. Currently, the most commonly used methods for pregnancy diagnosis in pig farms include ultrasound detection, boar estrus testing, platelet counting, radioimmunoassay, rectal examination, and biological hormone diagnosis. However, these methods suffer from high operational difficulty, high costs, and significant stress on sows.

[0003] In recent years, research on vaginal secretions has gradually increased, and they are often used to diagnose reproductive tract diseases in female mammals. However, research on the correlation between metabolites in vaginal secretions and whether animals are pregnant is relatively lacking. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide the application of metabolites in sow vaginal secretions in early pregnancy diagnosis.

[0005] This invention provides metabolic markers in vaginal secretions for early pregnancy diagnosis in sows, including at least one of NDOLEPROPIONIC ACID, cis-Aconitate, and 4-p-Coumaroylquinic acid.

[0006] Preferably, the present invention also provides a metabolic biomarker, including a combination of two or more of the above three metabolic biomarkers.

[0007] The present invention also provides the application of any one or more of the aforementioned metabolic markers in the early pregnancy diagnosis of sows.

[0008] Furthermore, the aforementioned application involves detecting the expression levels of the aforementioned metabolic markers in the vaginal secretions of sows and using this information to determine the pregnancy status of the sows. Those skilled in the art will understand that conventional methods known in the art can be used to determine the levels of metabolites, including but not limited to enzyme-linked immunosorbent assay (ELISA), colloidal gold assay, and mass spectrometry.

[0009] The metabolic biomarkers provided by this invention can also be used in the development of early pregnancy diagnostic products for sows, including but not limited to the development of ELISA test kits, colloidal gold test strips, and other products for early pregnancy diagnosis in sows.

[0010] Preferably, the early pregnancy diagnosis is performed 18 days after mating of the sow.

[0011] Preferably, large crossbred sows are used as the testing subjects.

[0012] This invention, based on combined metabolomics and machine learning analysis of sow vaginal secretions, is the first to discover that the aforementioned three metabolic molecules can serve as biomarkers for early pregnancy diagnosis in sows and their applications. All three metabolic biomarkers exhibit good diagnostic performance, especially when two or more are combined for diagnosis, resulting in even better diagnostic performance. This invention provides a new and effective detection method for early pregnancy diagnosis in sows. In particular, using vaginal secretions from sows 18 days post-mating for pregnancy diagnosis not only avoids stress on pregnant sows but also advances the effective pregnancy diagnosis time by 5-8 days compared to conventional methods. This helps farms seize the optimal time for rebreeding non-pregnant sows, reducing non-productive days, further lowering feeding costs, and increasing annual sow productivity. Furthermore, based on the metabolic biomarkers provided by this invention, various early pregnancy diagnostic products for sows can be developed. Attached Figure Description

[0013] Figure 1 Correlation analysis diagram for quality control samples; Figure 2 Cluster heatmaps for pregnant, non-pregnant, and QC samples; Figure 3 PCA analysis chromatograms for pregnant, non-pregnant, and quality control samples; Figure 4 PLS-DA analysis chromatograms for pregnant, non-pregnant, and quality control samples; Figure 5 Volcano plot of differential metabolites; Figure 6 A graph showing the parameters of the random forest model; Figure 7 Metabolic molecules with an MDI value greater than 0 in the random forest model; Figure 8 A graph of the support vector machine model parameters; Figure 9 Metabolic molecules that have an impact weight exceeding 0.3% on the decision boundary in a support vector machine model; Figure 10 Venn diagrams showing the significant differences in expression levels of metabolic molecules, the influence weights of random forests, and the influence weights of support vector machines for metabolic molecules with high levels of expression. Figure 11 Here are the structural diagrams of the three target molecules; Figure 12 ROC curves for the three target molecules; Figure 13 The expression levels of the three target molecules differed in different groups of samples; Figure 14 ROC curves for the diagnostic efficiency of the three target molecules in independent samples; Figure 15 ROC curves for the diagnostic efficiency of four combinations of three target molecules; Figure 16 The ROC curves show the diagnostic efficiency of four combinations of three target molecules in independent samples. Detailed Implementation

[0014] The embodiments of the present invention will be described in detail below with reference to specific examples. These embodiments are for illustrative purposes only and should not be construed as limiting the scope of the invention. Rather, they should be understood as a more detailed description of certain aspects, characteristics, and implementations of the invention. Unless otherwise specified, the instruments, reagents, and materials used in the following embodiments are commercially available.

[0015] Example 1: Preparation of vaginal secretion samples 1. Treatment of experimental animals All experimental samples used in this study were provided by Xinjiang Tiankang Livestock Technology Co., Ltd., and consisted of large white-grained crossbred gilts aged 225-250 days, weighing 135-150 kg, in good condition, and fed and housed under identical conditions. All sows were in good health, free from genetic diseases or deformities, and the pen temperature was maintained between 20-25°C and humidity between 50-55%. An all-in, all-out production strategy was employed, and a complete feed was provided. Breeding was performed in batches after estrus synchronization, and pregnancy was confirmed by ultrasound at 23-25 ​​days of age.

[0016] 2. Sample collection Eighteen days after batch mating, 15 sows were randomly selected. A sterile, long-handled medical cotton swab was inserted 10-20 cm into the vagina of each sow and removed after 30 minutes. At the same time, vaginal secretions were randomly collected from 10 unmated or failed-mating gilts from the same batch. All collected samples were frozen in liquid nitrogen and then transferred to a -80°C freezer for later storage.

[0017] 3. Sample pretreatment Samples were thawed at 4°C, sonicated on ice for 30 min, and transferred to new EP tubes for lyophilization. 200 μL of pre-cooled water and 800 μL of pre-cooled methanol were added, vortexed, sonicated on ice for 20 min, and incubated at -20°C for 1 h to precipitate proteins. The precipitated proteins were centrifuged at 16000g at 4°C for 20 min, and the supernatant was collected. The precipitated proteins were air-dried, and 200 μL of SDT was added for BCA quantification to determine protein concentration. Based on the protein content in the samples, the number of cells in different samples was estimated. For each sample, a supernatant containing an equal volume of cells was collected and evaporated to dryness using a high-speed vacuum centrifuge. For mass spectrometry analysis, 35 μL of pre-cooled methanol-water (1:1, v / v) solution was added to reconstitute the supernatant, and the sample was centrifuged at 20000g at 4°C for 20 min. An appropriate amount of supernatant was injected for analysis.

[0018] 4. Preparation of Quality Control Samples (hereinafter referred to as QC samples) All samples were taken in equal amounts and mixed uniformly to obtain QC samples, which were used to balance the chromatography-mass spectrometry system and determine the instrument status, and to evaluate the system stability throughout the experiment.

[0019] Example 2: Screening for Differential Metabolites 1. Non-targeted metabolomics detection using liquid chromatography-mass spectrometry (LC-MS) technology. Throughout the analysis, the sample was placed in an autosampler at 4°C. The sample was analyzed using a Shimadzu Nexera x2 ultra-high performance liquid chromatography (UHPLC) system with an ACQUITY UPLC® HSS T3 (2.1×100 mm, 1.8 µm) column (Waters, Milford, MA, USA). The injection volume was 4 μL, the column temperature was 40°C, and the flow rate was 300 µL / min. The mobile phases were A: 0.1% formic acid aqueous solution and B: 0.1% formic acid acetonitrile solution. The gradient elution program was as follows: 0–1 min, B maintained at 2%; 1–5 min, B linearly changed from 2% to 48%; 5–7 min, B linearly changed from 48% to 80%; 7–11 min, B linearly changed from 80% to 100%; 11–13 min, B maintained at 100%; 13–13.01 min, B linearly changed from 100% to 2%; 13.01–15 min, B maintained at 2%.

[0020] Subsequently, each sample was detected using electrospray ionization (ESI) in both positive (+) and negative (-) modes. Mass spectrometry analysis was performed using a TripleTOF® 6600 mass spectrometer (ABSCIEX) in positive and negative ion information-dependent bioassay (IDA) mode. The detection procedures were completed by Shanghai Baipu Biotechnology Co., Ltd.

[0021] Figure 1 This is a correlation analysis chart of the quality control samples. The three scatter plots in the lower left corner represent the expression levels of each metabolic molecule in different samples; the histogram on the diagonal shows the data distribution in each sample; and the three graphs in the upper right corner show the correlation coefficients between corresponding samples. As can be seen from the chart, the expression levels of each metabolic molecule in different samples are close to a linear relationship, the expression data of each metabolic molecule in different samples are close to a normal distribution, and the correlation coefficients between different samples are all greater than 0.9, which is at a high level. This indicates that the measurement data have good consistency and confirms the reliability of the mass spectrometry data.

[0022] 2. Differential expression analysis to screen differentially expressed metabolites After standardizing the expression levels of ion peaks in mass spectrometry (z-score), different models were used for analysis. First, dimensionality reduction methods were employed to describe the data characteristics of two types of samples: pregnant (represented by P in the chart) and non-pregnant (represented by NP in the chart). The dimensionality reduction analysis methods used were mainly principal component analysis (PCA) and partial least squares analysis (PLS-DA), and volcano plots were used to visualize the P-values ​​and fold changes.

[0023] Depend on Figure 2 It can be seen that there are significant differences in the expression of metabolic molecules between pregnant samples (P1~P15) and non-pregnant samples (NP1~NP10), while the expression levels of each molecule in QC samples (QC1~QC3) are highly consistent, indicating that QC samples are very stable and confirming the reliability of the experiment.

[0024] By performing PCA dimensionality reduction analysis on the samples, such as Figure 3 As shown in the figure, QC samples cluster in the PCA plot, indicating that the detection instrument is stable and reliable during sequencing, and the detection results are highly reliable. The figure also shows a general separation trend between pregnant and non-pregnant samples. Partial least squares discriminant analysis, as shown... Figure 4 As shown, there is a very significant separation trend between pregnant and non-pregnant samples, indicating that there are significant differences in the metabolic characteristics of vaginal secretions between pregnant and non-pregnant sows; while the quality control samples are concentrated in the figure, indicating that the experimental results are highly reliable.

[0025] This invention detected 126,695 ion peaks in sow vaginal secretion samples using chromatography-mass spectrometry (LC-MS). After noise reduction and peak extraction, 3,245 metabolic molecules in the sow vaginal secretions were identified, such as... Figure 5As shown in the differential metabolite volcano plot, 510 metabolites were screened that showed significant differences in expression levels between pregnant and non-pregnant sows, with a fold change (FC) value greater than 1.5 or less than 0.6667. Among them, 323 metabolites were significantly upregulated (red dots in the figure with log2(FC) > 0.58) and 187 metabolites were significantly downregulated (blue dots in the figure with log2(FC) < -0.58).

[0026] 3. Machine learning models for screening differentially expressed metabolites After obtaining the mass spectrometry data, the training and test sets were randomly split in a 7:3 ratio. The Random Forest (RF) and Support Vector Machine (SVM) models were then trained using the training set. After training, the accuracy of the models was tested using the test set. Finally, by combining p-value scores, full-joint (FC) values, RF feature weights, and SVM feature weights, metabolic molecules that showed significant differences between pregnant and non-pregnant samples and had high weights in the models were selected. This invention uses Python to build the machine learning model and employs the Anaconda environment and Spyder for compilation.

[0027] Random Forest Model: like Figure 6 As shown, the area under the receiver operating characteristic (ROC) curves (AUC) of the model on both the training and test sets is 1.00. In the confusion matrix, the model fits the training set samples perfectly, but a small number of pregnant samples in the test set are misclassified. In the prediction probability box plot, the model has higher prediction reliability for pregnant samples and weaker prediction reliability for non-pregnant samples. The learning curve shows that the model's training set score remains at a high level. When the number of training samples is small, the model's cross-validation score is low. When the number of training samples reaches more than 15, the cross-validation score increases and approaches the training set score.

[0028] As shown in Table 1, the random forest model performed well on the training set, with a precision, recall, F1 score, and accuracy of 1.00, indicating that the model could fit the training set data well. However, on the test set, the model performance was slightly worse, especially its predictive performance for non-pregnant samples was relatively weak.

[0029] Table 1 Performance evaluation results of the random forest model

[0030] In random forest models, Mean Decrease Impurity (MDI) is used to evaluate the importance of model features. The main principle is to evaluate the average decrease in Gini Imporyance (GI) caused by a feature across all trees. This is calculated by first calculating the GI values ​​of the parent nodes before and after the split in a single node when the feature is introduced. m ) and left and right child nodes (GI) l GI r The difference in Gini coefficient (VIM) Gini,j,m Then, the reduction in impurity of the feature across all split nodes in a single tree is summed (VIM). Gini,i,j Finally, the average total importance of the feature across all trees in the model (VIM) is calculated. Gini,j The formula is as follows:

[0031] Set the hyperparameter n_estimators to 100, and leave the other hyperparameters at their default values. For example... Figure 7 As shown, there are 123 metabolic molecules with an MDI value greater than 0 in the random forest model.

[0032] Support Vector Machine Model: like Figure 8 As shown, the AUC values ​​of the ROC curves for both the training and test sets of the model are 1.00. In the confusion matrix, the model can perfectly fit the samples, and there are no misclassified samples in either the training or test sets. The prediction probability box plot shows that the model has relatively strong predictive reliability for pregnant samples, while its predictive reliability for non-pregnant samples may be weaker. The learning curve shows that the model's training set score remains at a high level, while the model's cross-validation score is low when the number of training samples is small, and gradually increases when the number of training samples reaches a certain level.

[0033] As shown in Table 2, the support vector machine model performs well in diagnosing sow pregnancy based on vaginal secretions. It demonstrates good predictive accuracy and reliability in both the training and test sets and can effectively distinguish between pregnant and non-pregnant sow samples based on metabolomics detection results.

[0034] Table 2 Performance evaluation results of the support vector machine model

[0035] In the linear kernel support vector machine model, the model decision function is: f(x) = w Tx+b, where w is the weight vector, x is the feature vector, T denotes transpose (i.e., the weight vector is transposed to facilitate calculation with the feature vector), and b is the bias term. The coefficients of the weight vector directly reflect the degree of influence of each feature on the classification task decision; therefore, the importance of features in the model can be measured by the absolute value of the weight coefficients. In this invention, the support vector machine model regularization parameter (C) is 1, the kernel function is a linear kernel function, and all other hyperparameters are default. Metabolic molecules with a weight exceeding 0.3% affecting the decision boundary are selected. Figure 9 As shown, there are 127 metabolic molecules in the support vector machine model that have an influence weight of more than 0.3% on the decision boundary.

[0036] Example 3: Determining Metabolic Markers and Their Diagnostic Performance Metabolic molecule Venn diagrams obtained from three screening methods—significant difference in expression levels, random forest, and support vector machine—( Figure 10 Three common metabolic molecules were identified in the screening process: INDOLEPROPIONIC ACID (…). Figure 8 and Figure 9 The Chinese code is POS2295), cis-Aconitate ( Figure 8 and Figure 9 (coded as NEG2539) and 4-p-Coumaroylquinic acid ( Figure 8 and Figure 9 The Chinese code is NEG13945.

[0037] The mass spectrometry characteristics of the three metabolic molecules are shown in Table 3, and their molecular structures are shown in the figure. Figure 11 As shown.

[0038] Table 3. Mass spectrometry detection characteristics of three metabolic molecules

[0039] The expression differences of these metabolic molecules were calculated and ROC curves were plotted. The AUC value, optimal threshold, sensitivity, specificity, and Youden index of each ROC curve were then calculated. In this experiment, the expression level of metabolic molecules is a relative expression level, represented by the area of ​​the metabolic molecule detected by liquid chromatography-mass spectrometry.

[0040] like Figure 12 As shown in the ROC curves, all three metabolic molecules exhibited high AUC values: INDOLEPROPIONIC ACID (0.88), cis-Aconitate (0.92), and 4-p-Coumaroylquinic acid (0.77). This indicates that these three metabolic molecules are feasible as metabolic markers for diagnosing pregnancy in sows from vaginal secretions.

[0041] like Figure 13 As shown in Table 4, the expression levels of INDOLEPROPIONIC ACID were increased in pregnant samples compared to non-pregnant samples, while the expression levels of cis-Aconitate and 4-p-Coumaroylquinic acid decreased, and the differences were statistically significant (P<0.05). Combined with Table 4, this indicates that the three target metabolites have different properties in diagnosing sow pregnancy. INDOLEPROPIONIC ACID is expressed at higher levels in the vaginal secretions of pregnant sows than in non-pregnant sows, and it has high sensitivity and specificity, as well as a relatively high Youden index. It is an ideal metabolic biomarker for pregnancy diagnosis. The method of judgment is as follows: if the expression level of INDOLEPROPIONIC ACID detected is higher than its optimal threshold (i.e., the diagnostic threshold), pregnancy is diagnosed; if it is lower than the optimal threshold, non-pregnancy is diagnosed. The expression levels of cis-Aconitate and 4-p-Coumaroylquinic acid in the vaginal secretions of non-pregnant sows were higher than those in pregnant sows. Cis-Aconitate exhibited ideal sensitivity and specificity, along with a high Youden index, indicating its high accuracy in identifying pregnant sows. Conversely, 4-p-Coumaroylquinic acid showed higher sensitivity but lower specificity, suggesting its reliability in screening non-pregnant sows. The diagnostic method was as follows: when the expression level of cis-Aconitate or 4-p-Coumaroylquinic acid was above its respective optimal threshold (i.e., diagnostic threshold), the sow was diagnosed as non-pregnant; when it was below the optimal threshold, the sow was diagnosed as pregnant.

[0042] Table 4. Diagnostic efficiency of ROC curve analysis for three types of metabolic molecules.

[0043] Example 4: Validation of the diagnostic performance of the screened metabolic biomarkers in independent samples. Combined with Table 5 and Figure 14The diagnostic efficiency of the three types of metabolic molecules was verified using independent samples. In the independent samples, INDOLEPROPIONIC ACID was expressed at a higher level in vaginal secretions from pregnant sows than in non-pregnant sows, and its specificity and sensitivity both reached 1.00, indicating that this metabolic molecule has a good ability to identify both pregnant and non-pregnant sows. The expression levels of cis-Aconitate and 4-p-Coumaroylquinic acid were lower in vaginal secretion samples from pregnant sows compared to non-pregnant sows. Cis-Aconitate had high specificity, indicating that it has a stronger diagnostic ability for pregnant sows, while 4-p-Coumaroylquinic acid had high sensitivity, indicating that it has a stronger ability to identify non-pregnant sows.

[0044] Table 5 Diagnostic efficiency of three types of metabolites in independent samples

[0045] Example 5: Diagnostic performance of metabolic biomarker combinations Through Table 6 and Figure 15 As shown in Table 6, the diagnostic methods for sow pregnancy using either pairwise or combined combinations of the three metabolic molecules exhibit high sensitivity and specificity, with AUC values ​​of ROC curves all above 0.88, indicating reliable screening for different sows. The combination of INDOLEPROPIONIC ACID + cis-Aconitate and the three-molecule combination showed the highest sensitivity, indicating better reliability compared to other combinations. Furthermore, the overall performance of these two combinations was consistent; therefore, the INDOLEPROPIONIC ACID + cis-Aconitate combination should be chosen as the preferred method for diagnosing sow pregnancy.

[0046] Table 6. Diagnostic efficiency of ROC curve analysis for different combinations of three types of metabolic molecules.

[0047] Example 6: Diagnostic performance of metabolic biomarker combinations in independent samples Combined with Table 7 and Figure 16The results show that different combinations of metabolic molecules exhibit high sensitivity in independent samples, indicating higher reliability for screening pregnant sows. Among the different combinations, the INDOLEPROPIONIC ACID + cis-Aconitate combination and the combination of the three molecules showed the same sensitivity, specificity, Youden index, and ROC curve AUC value. Both combinations accurately identified pregnant and non-pregnant samples in independent samples, further confirming the value of this combination as a diagnostic method for early pregnancy in sows.

[0048] Table 7. Diagnostic efficiency of independent sample tests for different combinations of three types of metabolic molecules.

[0049] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A metabolic marker in vaginal secretions for early pregnancy diagnosis in sows, characterized in that, The metabolic markers include at least one of NDOLEPROPIONIC ACID, cis-Aconitate, and 4-p-Coumaroylquinic acid.

2. The metabolic biomarker as described in claim 1, characterized in that, The metabolic markers include two or more of INDOLEPROPIONIC ACID, cis-Aconitate, and 4-p-Coumaroylquinic acid.

3. The application of a metabolic marker in sow vaginal secretions in early pregnancy diagnosis, characterized in that, The metabolic biomarker is the metabolic biomarker as described in claim 1 or 2.

4. The application as described in claim 3, characterized in that, The expression levels of the aforementioned metabolic markers in the vaginal secretions of sows were detected and used to determine the pregnancy status of sows.

5. The application of the metabolic markers as described in claim 1 or 2 in the development of diagnostic products for early pregnancy in sows.

6. The metabolic biomarker as described in claim 1 or 2, or the application as described in claim 3 or 4, characterized in that, The early pregnancy diagnosis is performed 18 days after mating of the sow.

7. The metabolic biomarker as described in claim 1 or 2, or the application as described in claim 3 or 4, characterized in that, The sow in question is a Large White crossbred sow.

Images