Endogenous markers of exposure to bisphenols and methods for assessing exposure toxicity effects

Abstract

The present application discloses endogenous markers of exposure to bisphenols and methods for assessing exposure toxic effects. The present invention couples non-targeted and targeted metabolomics to systematically characterize trace and abundant serum metabolites, thereby improving the coverage, confirmation and sensitivity of metabolite marker identification, and is conducive to screening out bisphenol exposure metabolite markers with strong specificity and high sensitivity. The present invention discovered the high specificity and sensitivity of the histidine / kynurenine ratio in assessing the toxic effects of exposure to bisphenols, and constructed a classification model that can qualitatively predict exposure to bisphenols based on a support vector machine machine learning algorithm, providing a reliable model tool for rapid discrimination of toxic effects of exposure to bisphenols.

Description

Technical Field

[0001] The present invention belongs to the field of environmental toxicology and risk assessment of chemicals, and specifically relates to endogenous markers of exposure to bisphenols and a method for evaluating exposure toxicity effects. Background Art

[0002] Bisphenol A (BPA) is a plastic additive widely used in plastics, food packaging, thermal paper and rubber. Numerous zoological and human epidemiological studies have shown that BPA has obvious endocrine disrupting effects, and long-term exposure to environmental doses of BPA can significantly increase the risk of neurodegenerative diseases, behavioral problems, reproductive disorders and cancer. Therefore, in recent years, the use and addition of BPA in products has been gradually restricted worldwide to avoid the health hazards of BPA exposure. Therefore, BPA has been gradually replaced by other substitutes (BPs) to produce "BPA-free products".

[0003] BPA is a typical metabolic disruptor, which is associated with disorders of energy, lipid, amino acid and nucleic acid metabolism, as well as metabolic diseases such as obesity and type 2 diabetes. BPs (bisphenol AF (BPAF), bisphenol B (BPB) and bisphenol AP (BPAP)) are very similar to BPA in chemical structure and may also cause significant metabolic disruption. Identifying specific metabolic biomarkers associated with exposure and effects helps to reveal the common toxicity mechanisms of environmental pollutants. However, there are currently few studies related to the identification of BPs metabolic biomarkers, which mainly focus on cell studies and lack mammalian data. Metabolomics has the advantage of high-throughput identification of metabolites in biological samples, and has gradually become an important technical means for screening the targets of environmental pollutants and revealing their modes of action in environmental toxicology research. However, current research mainly focuses on single non-targeted metabolomics methods, which mainly detect metabolites with high abundance and are difficult to capture trace metabolites. Targeted metabolomics can achieve sensitive and specific detection of trace metabolites due to its high selectivity. Combined untargeted and targeted metabolomics analysis can improve the coverage and sensitivity of metabolite identification and provide a promising strategy for comprehensive characterization of trace and high-abundance metabolite markers.

[0004] At present, most BPs are put on the market for use without systematic toxicity assessment. The market lacks supervision over these new BPs, and lacks orderly testing and regulatory means. The concentration levels of some BPs detected in mammalian biological samples are comparable to BPA, and they show toxicity comparable to or greater than that of BPA, which has aroused widespread concern. Faced with the rapid growth of BPs components, traditional toxicological tests are difficult to meet the needs of BPs exposure health risk assessment due to problems such as long time consumption, high testing costs, and difficulty in circumventing animal ethics. There is an urgent need to develop effective rapid toxicity assessment tools. It is a feasible method to construct a BPs exposure toxicity prediction tool by identifying the common molecular characteristics induced by BPs exposure. Summary of the invention

[0005] Traditional toxicology tests cannot meet the rapidly growing demand for BPs exposure health risk assessment due to the problems of long time consumption, high testing costs, and difficulty in avoiding animal ethics. Therefore, the purpose of the present invention is to provide endogenous markers of bisphenol exposure and a method for evaluating the toxic effects of exposure, so as to achieve rapid identification of the toxic effects of exposure to bisphenol A substitutes. This method does not require large-scale animal experiments and toxicology index detection, and can be quickly evaluated by detecting the content of biomarkers.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] A method for screening endogenous markers of bisphenol exposure comprises the following steps:

[0008] (1) Animal exposure and serum collection: Experimental animals were gavaged with bisphenols at an exposure (dosing) concentration of 50-400 mg / kg / day for more than 15 days. After the exposure, serum was obtained from the experimental animals, purified, and analyzed.

[0009] The experimental animals include rodents or non-rodents;

[0010] The rodents are preferably rats and mice;

[0011] The non-rodent animal is preferably a rabbit, dog, monkey, or others;

[0012] The bisphenol substances include bisphenol A and its substitutes;

[0013] The bisphenol A substitutes include bisphenol AF (BPAF), bisphenol B (BPB) and bisphenol AP (BPAP);

[0014] The method of obtaining serum from experimental animals is that after the exposure, the experimental animals are fasted for more than 12 hours, anesthetized, and subjected to abdominal incision to collect blood from the abdominal aorta, which is then centrifuged to obtain serum;

[0015] (2) Non-targeted metabolomics analysis: The samples were analyzed and measured using ultra-high performance liquid chromatography coupled to a quadrupole / orbitrap high-resolution mass spectrometer, and several fragment ion spectra were collected. A non-targeted metabolomics screening and analysis process was established based on the software, including peak detection, peak alignment, peak extraction, peak integration, compound identification and annotation, missing value interpolation, quality control correction, background subtraction, and peak area normalization, and the collected fragment ion spectra were analyzed. The screened compounds were identified and annotated using the database.

[0016] The software is preferably Compound Discoverer 3.3 SP2;

[0017] The databases include ChemSpider, mzVault, Mass Lists and mzCloud;

[0018] (3) Targeted metabolomics analysis: A high-throughput detection method that can simultaneously detect multiple metabolites is constructed using ultra-high performance liquid chromatography tandem quadrupole / orbitrap high-resolution mass spectrometry. The chromatographic peaks of each compound are extracted based on the precise molecular mass of the parent ion, and quantitative analysis is performed using the isotope internal standard method.

[0019] Among them, the linear range of amino acid compounds was 125~4000 μg / L, the linear range of carbohydrates and energy metabolites was 62.5~2000 μg / L, and the linear range of nucleotides was 54.75~1750 μg / L;

[0020] (4) Analysis of the effects of bisphenol exposure on serum metabolites: The metabolites identified by non-targeted metabolomics analysis and targeted metabolomics analysis were combined for analysis to evaluate the metabolic interference effects caused by BPs exposure and to identify the differential metabolites between different exposure groups and the control group; the differential metabolites that appeared in all exposure groups were considered endogenous markers of bisphenol exposure;

[0021] The combined analysis preferably uses a supervised multivariate statistical analysis method (such as partial least squares discriminant analysis (PLS-DA)), and evaluates whether the model is overfitting through a permutation test, and screens differential metabolites based on one-way analysis of variance.

[0022] The present invention claims the use of histidine and / or kynurenine in evaluating the toxic effects of exposure to bisphenols.

[0023] The present invention claims the use of the histidine / kynurenine ratio in evaluating the toxic effects of exposure to bisphenols.

[0024] A method for evaluating the toxic effects of exposure to bisphenols, comprising the following steps:

[0025] S1. Obtain the histidine / kynurenine ratio data of each exposure group and divide it into training set data and test set data;

[0026] The data were subjected to natural logarithm transformation and z-score standardization to prevent the non-normal distribution of the original data from affecting the prediction accuracy of the model or making the model unable to converge; 70% of the data were set as a training set and 30% of the data were set as a test set;

[0027] S2, using the "histidine / kynurenine ratio" in the training set data as input, and using the "whether the person is exposed to bisphenols" in the training set data as output, to train the preset neural network model;

[0028] S3, inputting the "histidine / kynurenine ratio" in the test set data into the trained preset neural network model, and predicting the classification result of "whether the samples are exposed to bisphenols"; the classification result of "whether the samples are exposed to bisphenols" includes: the number correctly predicted to be exposed to bisphenols, the number incorrectly predicted to be exposed to bisphenols, the number incorrectly predicted to be not exposed to bisphenols, and the number correctly predicted to be not exposed to bisphenols;

[0029] S4. Calculate the assessment accuracy rate according to the classification result of “whether or not exposed to bisphenols”;

[0030] S5. The evaluation accuracy is represented by a ROC curve.

[0031] Compared with the prior art, the present invention has the following advantages and effects:

[0032] 1. The present invention couples non-targeted and targeted metabolomics to systematically characterize trace and abundant serum metabolites, thereby improving the coverage, confirmation and sensitivity of metabolite marker identification, and is conducive to screening out metabolite markers of bisphenol exposure with strong specificity and high sensitivity.

[0033] 2. The present invention identifies the common molecular features induced by different bisphenols at different exposure doses, clarifies the potential targets of exposure toxic effects caused by bisphenols, and can provide a highly universal biological indicator for the toxicity evaluation of bisphenols.

[0034] 3. The present invention discovered the high specificity and sensitivity of the histidine / kynurenine ratio in assessing the toxic effects of exposure to bisphenols, and constructed a classification model that can qualitatively predict exposure to bisphenols based on a support vector machine machine learning algorithm, providing a reliable model tool for rapid identification of the toxic effects of exposure to bisphenols. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1This is the chromatographic separation diagram of 283 metabolites in the standard solution.

[0036] Figure 2 Metabolites identified in nontargeted metabolomics analysis in positive and negative ion scanning modes.

[0037] Figure 3 The differences in metabolic profiles between the control group and the different BPs exposure groups were shown based on partial least squares discriminant analysis.

[0038] Figure 4 yes Figure 3 Permutation test results for partial least squares discriminant analysis.

[0039] Figure 5 The metabolites that changed significantly after exposure to different doses of BPs were analyzed based on the volcano plot.

[0040] Figure 6 is the number of metabolites that cause interference with different doses of BPs exposure.

[0041] Figure 7 It is based on KEGG pathway enrichment analysis to analyze the metabolic pathways that are significantly disturbed after exposure to different BPs.

[0042] Figure 8 The chemical classification of serum metabolites affected by exposure to the three BPs includes: OADs, carboxylic acids and their derivatives; LLLLMs, lipids and lipid-like molecules; OOCs, oxygenated organic compounds; NNAs, nucleosides, nucleotides and their analogs; OHCs, organic heterocyclic compounds; CADs, carboxylic acids and their derivatives; OOs, organic oxygen compounds; SSDs, steroids and steroid derivatives; BSDs, benzene and its substituted derivatives; PMNs, pyrimidine nucleosides; and IDDs, imidazopyridines.

[0043] Fig. 9 The impact of metabolic pathways caused by the exposure of three BPs was analyzed based on KEGG pathway enrichment.

[0044] Fig.10 Metabolic flux analysis of the top three metabolic pathways affected by BPs exposure.

[0045] Fig.11 is a Venn diagram showing the association of the affected metabolite sets after exposure to different doses of BPs.

[0046] Fig.12 is the fold change distribution of the common metabolic characteristics exposed to high concentrations of BPs.

[0047] Fig.13 It is the changes of histidine and kynurenine and their ratio in different BPs exposure groups.

[0048] Fig.14 The control group was distinguished from the BPs-exposed group based on histidine and kynurenine concentration levels.

[0049] Fig.15 The support vector machine model was used to evaluate the predictive performance of histidine, kynurenine and their ratio on BPs exposure toxicity. DETAILED DESCRIPTION

[0050] The present invention is further described in detail below in conjunction with embodiments and drawings, but the embodiments of the present invention are not limited thereto.

[0051] Example 1

[0052] A method for screening endogenous markers of bisphenol exposure comprises the following steps:

[0053] (1) Animal exposure: 60 female and 60 male SD rats weighing 180-220 g were purchased from Zhuhai Baishitong Biotechnology Co., Ltd., China. Given that dietary intake is the main route of human exposure to BPs, this study used oral administration for exposure, with an exposure interval of 15 days. There were three exposure groups and one control group for each BP, with 5 females and 5 males in each group. The exposure concentration was determined based on the exposure concentration range (26-259 mg / kg / day) used by the National Toxicology Program of the U.S. Department of Health and Human Services to evaluate the developmental and reproductive toxicity of subchronic bisphenol AF exposure. The low, medium, and high exposure groups were exposed to 50, 200, and 400 mg / kg / day of BPs, respectively, while the control group was only given an equal amount of corn oil. After the exposure, the rats were fasted for 12 h, anesthetized with isoflurane, positioned dorsally, and subjected to ventral incision to collect blood from the abdominal aorta. Two milliliters of blood were collected using lithium heparin anticoagulant tubes and centrifuged at 3500 rpm for 5 minutes to obtain serum.

[0054] (2) Serum sample purification: Take 100 μL serum supernatant and add 300 μL methanol precooled at -80°C. Vortex the sample for 30 seconds, place it at -80°C for 60 min, and centrifuge it at 14,000 rpm for 15 minutes. Take 200 μL supernatant, add 50 μL isotope internal standard and 50 μL vitamin C (to prevent metabolite oxidation), and vacuum concentrate to dryness. Then, add 200 μL of 1% formic acid-10% methanol aqueous solution to reconstitute, let it stand for 15 minutes, vortex it for 30 seconds, centrifuge it at 14,000 rpm for 15 minutes, and take 120 μL of supernatant for analysis.

[0055] (3) Non-targeted metabolomics analysis: Ultra-high performance liquid chromatography tandem quadrupole / orbitrap high-resolution mass spectrometer (Orbitrap Exploris 240) was used for analysis and determination. The specific conditions were as follows: the chromatographic separation column was a Hypersil GOLDAQ C18 column (150 × 2.1 mm, 1.9 um, Thermo Fisher Scientific, Germany), the organic phase (B) was methanol, the aqueous phase (B) was deionized water (negative ion mode) or 0.1% formic acid (positive ion mode), the column temperature was 35 °C, and the injection volume was 2 μL. The gradient elution program was as follows: 0-2 min 2% B, flow rate 0.25 mL / min; 2-8 min 2-98% B, flow rate 0.25 mL / min; 8.01-11 min 98% B, flow rate 0.40 mL / min; 11.01-14 min 2% B, flow rate 0.25 mL / min. The mass spectrometer was detected in electrospray ionization (ESI) mode, with spray voltages of 3.5 and –2.5 kV in positive and negative ion modes, respectively. The ESI source parameters were as follows: sheath gas, auxiliary gas, and sweep gas were set to 45, 8, and 1 Arb, respectively. The ion transfer tube and vaporizer temperatures were 320 and 350 °C, respectively. The data acquisition process included full mass spectrometry scan, dynamic exclusion, target mass exclusion, peak detection, and fragment ion scan. In full scan mode, the resolution was 120,000 and the scan range was 70–1000 m / z. The process blank sample was analyzed in full scan mode, and the parent ions with the top 100 signal responses were screened out to establish an ion exclusion list. Fragment ion scan: isolation window: 2 m / z; collision energy type: standardized; HCD collision energy: 20%, 40%, 60%, 80%; resolution: 15,000; scan range mode: automatic; AGC target: standard; number of fragment ion spectra collected: 5.

[0056] (4) Non-targeted metabolomics screening analysis: Based on Compound Discoverer 3.3 SP2 software, a non-targeted metabolomics screening analysis process was established, including peak detection, peak alignment, peak extraction, peak integration, compound identification and annotation, missing value interpolation, quality control correction, background subtraction, and peak area normalization. The fragment ion spectra collected in step (3) were analyzed. The screened compounds were identified and annotated using databases such as ChemSpider, mzVault, Mass Lists, and mzCloud. The quality assurance and control (QC) measures during the non-targeted metabolomics analysis are as follows: ① Open process samples were prepared to eliminate interference caused by material and reagent contamination; ② 10 µL of each serum sample was mixed to prepare a QC sample, which was interspersed with the sample for measurement to monitor the variability during the instrument analysis process.

[0057] (5) Targeted metabolomics analysis: Purchase single standards to prepare mixed standard solutions and generate standard curves. Use ultra-high performance liquid chromatography tandem quadrupole / orbitrap high-resolution mass spectrometry to construct a high-throughput detection method that can simultaneously detect 283 metabolites ( Figure 1 ), including 103 amino acids, 83 carbohydrates, 72 nucleotides and 25 energy metabolism metabolites. The chromatographic conditions and mass spectrometry methods were the same as those for non-targeted screening. The chromatographic peaks of each compound were extracted based on the accurate molecular mass of the parent ion, and the isotope internal standard method was used for quantitative analysis. Among them, the linear range of amino acid compounds was 125~4000 μg / L, the linear range of carbohydrates and energy metabolites was 62.5~2000 μg / L, and the linear range of nucleotides was 54.75~1750 μg / L.

[0058] (6) Analysis of the effect of BPs exposure on serum metabolites: Based on the non-targeted metabolomics screening workflow, 434 metabolites were identified in the positive ion mode and 498 metabolites were identified in the negative ion mode ( Figure 2 ). In addition, a targeted quantitative method was used to complete the quantification of 94 metabolites. Subsequently, all identified metabolites were combined for analysis to evaluate the metabolic interference effects caused by BPs exposure and to identify differential metabolites. Partial least squares discriminant analysis (PLS-DA) showed that there were significant differences in the metabolic profiles between the control group and the different BPs exposure groups ( Figure 3 The intercepts of goodness of fit (R2) and goodness of prediction (Q2) indicate that the PLS-DA model is reliable and does not overfit ( Figure 4 ).

[0059] One-way analysis of variance and Dunnett's multiple comparison test were used to evaluate the inter-group differences in serum metabolites after BPs exposure. The volcano plot was used to analyze the metabolites that changed significantly after exposure to different doses of BPs. Metabolites that met the change fold change >1.2 and p value <0.05 were considered to be significantly changed. The results showed that the number of disturbed metabolites was positively correlated with the exposure dose of BPs. Compared with the control group, low, medium, and high concentrations of BPAF exposure can lead to significant changes in the concentrations of 258 (138 downregulated and 120 upregulated), 303 (170 downregulated and 133 upregulated), and 405 (192 downregulated and 213 upregulated) metabolites, respectively ( Figure 5 ). In addition, compared with the control group, low, medium and high concentrations of BPAP exposure can cause significant changes in the concentrations of 15, 94 and 189 metabolites, respectively, and low, medium and high concentrations of BPB exposure can cause significant changes in the concentrations of 137, 174 and 277 metabolites, respectively. Different concentrations of BPAF, BPB and BPAP exposure caused significant changes in 554, 391 and 236 serum metabolites, respectively ( Figure 6). The number of metabolites affected by low-concentration BPAF exposure was significantly higher than that caused by low-concentration BPB and BPAP. Then, metabolite pathway enrichment analysis was performed using the online tool MetaboAnalyst 6.0 to study the metabolic pathways that were significantly disturbed after exposure to different BPs. The results are as follows Figure 7 As shown, the red dotted line is the threshold of significant change. BPAF, BPB and BPAP exposure significantly interfered with 10, 10 and 4 metabolic pathways, respectively. In summary, BPs exposure interfered with serum metabolites in a dose-dependent manner, and the degree of metabolic interference effect was BPAF>BPB>BPAP.

[0060] Subsequently, we integrated the metabolites affected by these three BPs exposure to explore the potential pathways of BPs exposure. The results showed that the disturbed metabolites mainly included carboxylic acids and their derivatives, fatty acid acyl groups, organic oxygen compounds, and steroids and steroid derivatives ( Figure 8 ). The results of KEGG pathway enrichment analysis showed that the most significantly affected metabolic pathway was arginine-proline metabolism, followed by alanine-aspartate-glutamate metabolism and valine-leucine-isoleucine biosynthesis ( Fig. 9 ). Metabolic flux analysis showed that these three metabolic pathways are interrelated ( Fig.10 After BPs exposure, the levels of creatine, arginine, 4-aminobutyric acid, γ-aminobutyric acid, proline, and pyruvate were significantly downregulated, while the levels of succinate, citrate, glutamate, ornithine, threonine, L-isoleucine, N-acetyl-L-aspartate, and L-valine were significantly upregulated.

[0061] (7) Identification of metabolic characteristics of BPs exposure: We further identified the potential metabolic characteristics of BPs exposure. The Venn diagram showed that there were 0, 12, and 32 metabolites co-interfered by the three BPs in the low, medium, and high concentration exposure groups, respectively ( Fig.11 Among them, 18 of the 32 metabolites in the high-concentration exposure group were successfully annotated (Table 1). These metabolites mainly included glycerophospholipids, carboxylic acids and their derivatives, pyrimidine nucleosides, and organic oxygen compounds, with a change factor between 0.033 and 0.793 and 1.21 and 2.91 ( Fig.12 ). It is worth noting that histidine and kynurenine were the only two metabolic markers that were significantly changed in all BPs exposure groups. After BPs exposure, serum histidine levels were significantly upregulated, while kynurenine levels were significantly downregulated ( Fig.13). Among them, the changes of histidine and kynurenine in the BPAF exposure group were the most obvious. Compared with the control group (4024 μg / L), the mean concentrations of histidine increased to 6580, 8607, and 10795 μg / L after low, medium, and high concentrations of BPAF exposure, respectively, showing an obvious linear growth relationship. Compared with the control group (11.3), the mean levels of histidine and kynurenine ratios increased to 38.5, 55.0, and 88.2 after low, medium, and high concentrations of BPAF exposure, respectively, also showing an obvious linear growth relationship. However, this linear growth trend was greatly weakened in the BPB and BPAP exposure groups.

[0062] Table 1: List of common metabolic features of high-concentration BPs exposure

[0063]

[0064] Subsequently, we attempted to use histidine and kynurenine levels to distinguish the control group and BPs-exposed group samples. The two-dimensional scatter plot showed that the control group and the BPs-exposed group had good separation, among which the separation between the BPAF-exposed group and the control group samples was the most obvious ( Fig.14 ). Therefore, histidine and kynurenine levels can effectively indicate the toxicity of BPs exposure, which suggests that they can be used as biomarkers to evaluate the toxic effects of exposure to BPA substitutes.

[0065] Example 2

[0066] A method for evaluating the toxic effects of exposure to bisphenols, comprising the following steps:

[0067] Support vector machine (SVM) is a supervised learning algorithm. Its basic idea is to find an optimal decision hyperplane so that the distance between the two types of data closest to the hyperplane on both sides is maximized. Based on the SVM machine learning algorithm, "histidine, kynurenine and histidine / kynurenine ratio" are used as input variables, and the binary variable "whether exposed to BPs" is used as the output variable to construct a rapid prediction model for exposure toxicity of bisphenol A substitutes. The specific steps are as follows: ① Import the original data and perform natural logarithm transformation and z-score standardization to avoid the non-normal distribution of the original data affecting the prediction accuracy of the model or making the model unable to converge; ② Set 70% of the data as the training set and 30% of the data as the test set; ③ Train the SVM in the training set to determine the number of support vectors; ④ Put the trained SVM into the test set and classify the data in the test set; ⑤ Output the prediction results in the form of confusion matrix and calculate the accuracy; ⑥ Characterization and selection of model prediction performance: Select the ROC curve to characterize the prediction effect of the model.

[0068] The ROC curve uses the false positive rate as the horizontal axis and the true positive rate as the vertical axis. According to the performance of the test data in the SVM classifier, a set of point pairs is obtained. As the test data changes, its false positive rate and true positive rate will also change, so a series of point pairs can be obtained. Connecting these point pairs will get the ROC curve, which can well characterize the prediction effect of the classifier.

[0069] In order to compare the prediction effects of different biomarkers, the area under the curve (AUC) and accuracy (ACC) were used as evaluation criteria. Generally speaking, an AUC range of 0.5 to 1 indicates that the prediction results of the classifier have reference value, and the higher the value, the better the prediction effect of the classifier.

[0070] The results showed that when kynurenine was used to construct the classification prediction model, the AUC and ACC values ​​were 0.833 and 0.801, respectively, and its prediction accuracy was slightly higher than that of the classification prediction model constructed using histidine (AUC: 0.867, ACC: 0.786) ( Fig.15 ). When the histidine / kynurenine ratio is used, its accuracy in predicting BPs exposure toxicity is better than that of histidine or kynurenine alone. Therefore, compared with histidine and kynurenine, the histidine / kynurenine ratio is a better biomarker for rapid assessment of the toxic effects of exposure to BPA substitutes.

[0071] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A method for screening endogenous markers of exposure to bisphenols, characterized in that The following steps are involved: (1) Animal exposure and serum collection: Experimental animals were gavaged with bisphenols at an exposure concentration of 50-400 mg / kg / day for more than 15 days. After the exposure, serum was obtained from the experimental animals, purified, and analyzed. (2) Non-targeted metabolomics analysis: The samples were analyzed and measured using ultra-high performance liquid chromatography coupled to a quadrupole / orbitrap high-resolution mass spectrometer, and several fragment ion spectra were collected. A non-targeted metabolomics screening and analysis process was established based on the software Compound Discoverer 3.3 SP2, including peak detection, peak alignment, peak extraction, peak integration, compound identification and annotation, missing value interpolation, quality control correction, background subtraction, and peak area normalization. The collected fragment ion spectra were analyzed. The screened compounds were identified and annotated using the database. (3) Targeted metabolomics analysis: A high-throughput detection method that can simultaneously detect multiple metabolites is constructed using ultra-high performance liquid chromatography tandem quadrupole / orbitrap high-resolution mass spectrometry. The chromatographic peaks of each compound are extracted based on the precise molecular mass of the parent ion, and quantitative analysis is performed using the isotope internal standard method. (4) Analysis of the effects of bisphenol exposure on serum metabolites: The metabolites identified by non-targeted metabolomics analysis and targeted metabolomics analysis were combined for analysis to evaluate the metabolic interference effects caused by BPs exposure and to identify the differential metabolites between different exposure groups and the control group; the differential metabolites that appeared in all exposure groups were considered endogenous markers of bisphenol exposure; The combined analysis described in step (4) is a supervised multivariate statistical analysis method, and the overfitting of the model is evaluated by permutation test, and the screening of differential metabolites is performed based on one-way analysis of variance.

2. The method according to claim 1, characterized in that: The bisphenol substances described in step (1) include bisphenol A and its substitutes.

3. The method according to claim 2, characterized in that: The bisphenol A substitutes include bisphenol AF, bisphenol B and bisphenol AP.

4. The method according to claim 1, characterized in that: The step (1) of obtaining serum from experimental animals is that after the exposure, the experimental animals are fasted for more than 12 hours, anesthetized, and subjected to abdominal incision to collect blood from the abdominal aorta, which is then centrifuged to obtain serum.

5. The method according to claim 1, characterized in that: The experimental animals described in step (1) include rodents or non-rodents.

6. The method according to claim 5, characterized in that: The rodents include rats and mice.

Images