Method for evaluating toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology

Abstract

The invention relates to the technical field of toxicity evaluation of bisphenol compounds, and particularly discloses a method for evaluating toxicity of bisphenol compounds based on a virtual screening and pharmacophore modeling technology, which utilizes the virtual screening and pharmacophore modeling technology to construct a comprehensive calculation framework of binding estrogen receptor pharmacophore modeling and molecular docking. And systematically evaluating the endocrine disruption effect of the bisphenol compound. And an effective evaluation method is provided for supervision hazard identification of the bisphenol compounds.

Description

Technical Field

[0001] This invention relates to the field of toxicity assessment technology for bisphenol compounds, and more specifically, to a method for assessing the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling techniques. Background Technology

[0002] Bisphenols are a diverse class of synthetic chemicals, primarily used as monomers in the production of polycarbonate plastics and epoxy resins, thus establishing their position as one of the most widely used industrial chemicals in modern manufacturing. Statistics show that global production of bisphenol A (BPA) alone was approximately 10 million tons in 2022, with a global market value reaching US$23.61 billion in 2023 and projected to reach US$34.25 billion by 2030. Bisphenol A exhibits diverse industrial applications, with polycarbonate manufacturing accounting for 65-70% of total production, making it the primary use. Epoxy and vinyl ester resin formulations constitute a secondary application category, accounting for 25-30% of consumption. The remaining 5% of BPA production is used for specialized applications in high-performance polymers and as a functional additive in PVC formulations, thermal paper coatings, and other industrial materials.

[0003] Health hazards associated with exposure to bisphenol A (BPA) compounds have become a major public health concern, encompassing a wide range of adverse effects across multiple physiological systems. Specifically, scientific evidence has linked BPA exposure to learning difficulties, increased aggression, ADHD, reduced fertility, and an increased risk of substance dependence, with particular susceptibility observed in infants, children, and pregnant women. Furthermore, BPA exposure is associated with an increased risk of hormone-related cancers, including ovarian, breast, prostate, and colon cancer, and also induces resistance to various chemotherapy drugs such as doxorubicin, cisplatin, and vinblastine.

[0004] Bisphenolic compounds exert endocrine-disrupting effects primarily through their interaction with nuclear hormone receptors (especially estrogen receptors α and β). Existing computational methods (such as the traditional QSAR model) have limitations in assessing structurally diverse bisphenolic compounds, making it difficult to rapidly and systematically evaluate the toxicity of a large number of compounds.

[0005] Therefore, it is necessary to design a method for assessing the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling techniques to address the problem that existing technologies cannot rapidly and systematically assess the toxicity of a large number of compounds. Summary of the Invention

[0006] In view of this, the present invention proposes a method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology, in order to solve the problem that the existing technology cannot quickly and systematically evaluate the toxicity of a large number of compounds.

[0007] This invention proposes a method for assessing the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling techniques, comprising the following steps:

[0008] Multiple pharmacophore models were generated based on the three-dimensional crystal structures of human estrogen receptors α and β.

[0009] The generated pharmacophore models were validated using receiver operating characteristic (ROC) curves, and the effective pharmacophore model was selected.

[0010] Construct a library of bisphenol compounds containing different bridging groups, substitution modes, and functional group modifications;

[0011] The validated effective pharmacophore model was used as a molecular filter to virtually screen the bisphenol compound library. Based on the quality of the alignment between the compound and the pharmacophore, compounds with the ability to bind to multiple pharmacophore models were screened as lead compounds.

[0012] Molecular docking experiments were performed on the screened lead compounds to evaluate the interaction mode and binding ability of the lead compounds with the human estrogen receptor;

[0013] The selected lead compounds were subjected to ADMET property analysis and toxicity prediction.

[0014] Furthermore, when generating multiple pharmacophore models based on the three-dimensional crystal structures of human estrogen receptors α and β, the following are included:

[0015] Human estrogen receptor α and β proteins were prepared based on the three-dimensional crystal structures of the human estrogen receptor α and β, and the co-crystal ligands that recognize human estrogen receptor α and β proteins were prepared. Within a radius range of amino acid residues, ligand binding sites are defined based on the identified amino acid residues, and multiple pharmacophore models are generated based on the defined ligand binding sites and the three-dimensional crystal structures of the human estrogen receptors α and β.

[0016] Furthermore, when validating the generated pharmacophore models using receiver operating characteristic (ROC) curves, pharmacophore models with an AUC value ≥ 0.8 are identified as effective pharmacophore models.

[0017] Furthermore, when constructing a bisphenol compound library containing different bridging groups, substitution modes, and functional group modifications, the following steps are taken:

[0018] Bisphenol compounds were retrieved from the PubChem database and then manually screened to remove duplicate entries;

[0019] The screened compounds were standardized using the molecular preparation protocol implemented in Discovery Studio 2021;

[0020] Three-dimensional conformations were generated from the screened compounds using the FAST conformation search algorithm.

[0021] Furthermore, the standardization includes the neutralization of charged substances, tautomer standardization, and stereochemical partitioning where applicable.

[0022] Furthermore, during the virtual screening, when the scale-free fit threshold between the compound and the pharmacophore model is ≥2, the compound is determined to have binding ability to this pharmacophore model.

[0023] Furthermore, during the molecular docking experiment, multiple different binding postures are generated for each lead compound. The binding postures are scored using the CDOCKER interaction energy, and the highest-scoring posture for each lead compound is selected for interaction analysis and visualization.

[0024] Furthermore, the CDOCKER interaction energy includes van der Waals interactions, electrostatic contributions, and internal ligand strain energy.

[0025] Furthermore, when performing ADMET property analysis on the screened lead compounds, the pharmacokinetic parameters of the screened lead compounds were analyzed, including water solubility, blood-brain barrier penetration potential, and human intestinal absorption probability.

[0026] Furthermore, the toxicity prediction specifically involves: conducting toxicological assessments on the screened lead compounds, including hepatotoxicity potential, carcinogenicity predictions for multiple species and sex combinations, acute oral toxicity and chronic toxicity, and skin irritation potential.

[0027] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0028] 1. This invention generates a pharmacophore model that accurately characterizes receptor-ligand binding requirements by analyzing key interaction features in the estrogen receptor structure. The validation process ensures that the model has good discriminative ability, improving the reliability of subsequent screening.

[0029] 2. This invention utilizes a validated pharmacophore model as a molecular filter to rapidly and efficiently identify compounds with potential endocrine-disrupting effects from a large number of bisphenol compounds. This overcomes the limitations of traditional experimental methods in large-scale screening.

[0030] 3. This invention provides a detailed structural basis for compound-receptor interactions by simulating the binding process of compounds and receptors, which helps to understand the mechanisms of endocrine disruption. Attached Figure Description

[0031] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0032] Figure 1 A flowchart illustrating a method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling techniques, provided in an embodiment of the present invention;

[0033] Figure 2 The three-dimensional structures of estrogen receptors α and β and their corresponding ligands provided in the embodiments of the present invention;

[0034] Figure 3 This is a graph showing the ROC curve analysis of the 20 pharmacophore models in Example 1 of the present invention;

[0035] Figure 4 This is a diagram showing the distribution of bisphenol compounds bound to different pharmacophore models in Example 1 of the present invention;

[0036] Figure 5 This is a topological binding diagram of phenolic alcohol and triphenolic alcohol with estrogen receptor pharmacophores in Example 1 of the present invention;

[0037] Figure 6 This is a schematic diagram of the molecular docking structure of phenolic alcohol and triphenolic alcohol with estrogen receptors α and β in Example 1 of the present invention. Detailed Implementation

[0038] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0039] like Figure 1 As shown in some embodiments of this application, a method for assessing the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling techniques includes the following steps:

[0040] Multiple pharmacophore models were generated based on the three-dimensional crystal structures of human estrogen receptors α and β.

[0041] The generated pharmacophore models were validated using receiver operating characteristic (ROC) curves, and the effective pharmacophore model was selected.

[0042] Construct a library of bisphenol compounds containing different bridging groups, substitution modes, and functional group modifications;

[0043] The validated effective pharmacophore model was used as a molecular filter to virtually screen the bisphenol compound library. Based on the quality of the alignment between the compound and the pharmacophore, compounds with the ability to bind to multiple pharmacophore models were screened as lead compounds.

[0044] Molecular docking experiments were performed on the screened lead compounds to evaluate the interaction mode and binding ability of the lead compounds with the human estrogen receptor;

[0045] The selected lead compounds were subjected to ADMET property analysis and toxicity prediction.

[0046] Specifically, the three-dimensional crystal structures of human estrogen receptors α and β were obtained by retrieving from a protein database (PDB). Specifically, the ERα structure complexed with the natural ligand was obtained using PDB ID: 1JSO, and the ERβ structure was retrieved using PDB ID: 2SFZ. Both structures were based on their high resolution. The selection is based on the presence of eutectic ligands within the ligand binding domain.

[0047] Understandably, this invention generates a pharmacophore model that accurately characterizes receptor-ligand binding requirements by analyzing key interaction features in the estrogen receptor structure. The validation process ensures that the model has good discriminative ability, improving the reliability of subsequent screening.

[0048] Understandably, this invention utilizes a validated pharmacophore model as a molecular filter to rapidly and efficiently identify compounds with potential endocrine-disrupting effects from a large number of bisphenol compounds. This overcomes the limitations of traditional experimental methods in large-scale screening.

[0049] Understandably, this invention provides a detailed structural basis for compound-receptor interactions by simulating the binding process of compounds to receptors, which helps to understand the mechanisms of endocrine disruption.

[0050] In some embodiments of this application, when generating multiple pharmacophore models based on the three-dimensional crystal structures of human estrogen receptors α and β, the following steps are included:

[0051] Human estrogen receptor α and β proteins were prepared based on the three-dimensional crystal structures of the human estrogen receptor α and β, and the co-crystal ligands that recognize human estrogen receptor α and β proteins were prepared. Within a radius range of amino acid residues, ligand binding sites are defined based on the identified amino acid residues, and multiple pharmacophore models are generated based on the defined ligand binding sites and the three-dimensional crystal structures of the human estrogen receptors α and β.

[0052] Specifically, the retrieved protein structures were fully prepared using Discovery Studio 2021, including water molecule removal, hydrogen atom addition, and side chain conformation optimization. Subsequently, co-crystal ligands were identified. The ligand binding site is defined by all amino acid residues within the radius, ensuring that all potential interaction sites related to ligand binding are captured during subsequent pharmacophore formation.

[0053] Specifically, the interaction pharmacophore generation module implemented in Discovery Studio 2021 was used to generate structure-based pharmacophore models. For each estrogen receptor subtype, ten different pharmacophore models were systematically generated based on ligand-receptor interaction patterns observed in the crystal structure. The pharmacophore generation algorithm analyzed various intermolecular interactions, including hydrogen bonds (donor and acceptor), hydrophobic contacts, aromatic stacking interactions, and electrostatic interactions between cocrystal ligands and their respective binding sites.

[0054] The generated pharmacophore models are characterized by specific features representing the basic binding requirements: hydrogen bond donor (D), hydrogen bond acceptor (A), aromatic ring (R), hydrophobic region (H), and repulsion volume (S). Each pharmacophore model is assigned a unique identifier (ERα from ESR1_01 to ESR1_10; ERβ from ESR2_01 to ESR2_10).

[0055] In some embodiments of this application, when the generated pharmacophore models are validated using receiver operating characteristic (ROC) curves, pharmacophore models with an AUC value ≥ 0.8 are identified as effective pharmacophore models.

[0056] Specifically, the predictive performance of the generated pharmacophore model was rigorously evaluated using established validation protocols with the Enhanced Useful False Positive Database (DUDE). A balanced dataset containing 150 known active estrogen receptor ligands and 150 carefully selected false positive molecules was constructed for validation.

[0057] Validation was performed using receiver operating characteristic (ROC) curve analysis, with the area under the curve (AUC) as the primary performance metric. Additional statistical parameters, including sensitivity, specificity, and Matthews correlation coefficient, were calculated to provide a comprehensive assessment of model performance. Pharmacophore models with an AUC value ≥ 0.8 were considered effective and retained for subsequent virtual screening applications.

[0058] In some embodiments of this application, the construction of a bisphenol compound library including different bridging groups, substitution modes, and functional group modifications includes:

[0059] Bisphenol compounds were retrieved from the PubChem database and then manually screened to remove duplicate entries;

[0060] The screened compounds were standardized using the molecular preparation protocol implemented in Discovery Studio 2021;

[0061] Three-dimensional conformations were generated from the screened compounds using the FAST conformation search algorithm.

[0062] In some embodiments of this application, the standardization includes neutralization of charged substances, tautomer standardization, and stereochemical partitioning where applicable.

[0063] Specifically, a comprehensive library of bisphenol compounds was systematically compiled from the PubChem database using a structure- and keyword-based search strategy. The search protocol used "bisphenol" as the primary query, followed by manual screening to ensure the inclusion of structurally relevant compounds and to remove duplicate entries. The final dataset in the bisphenol compound library contains 427 unique bisphenol compounds, covering a diverse range of structural variations, including different bridging groups, substitution patterns, and functional group modifications.

[0064] All retrieved compounds were normalized using molecular preparation protocols implemented in Discovery Studio 2021, including neutralization of charged substances, tautomer normalization, and stereochemical allocation where applicable. Three-dimensional conformations were generated using the FAST conformational search algorithm, with a maximum of 200 conformations generated for each compound to ensure adequate sampling of the conformational space.

[0065] In some embodiments of this application, when performing the virtual screening, if the scale-free fit threshold of the compound and the pharmacophore model is ≥2, it is determined that the compound has the ability to bind to the pharmacophore model.

[0066] Specifically, the validated effective pharmacophore models were used as molecular filters to virtually screen a library of bisphenol compounds. The screening process was implemented using the ligand profiling module in Discovery Studio 2021, employing a systematic approach to identify compounds that met the spatial and chemical requirements defined for each pharmacophore model.

[0067] The screening protocol utilizes specific fit criteria to evaluate the quality of compound-pharmacophore alignment. A compound is considered a potential hit compound if a scale-free fit threshold of ≥2 is applied, indicating that the compound binds to the pharmacophore model. The fit value quantifies the alignment between the compound's 3D structure and pharmacological features, considering spatial overlap and feature matching. Compounds exhibiting binding ability to multiple pharmacophore models are prioritized as lead compounds for detailed molecular docking analysis.

[0068] In some embodiments of this application, when performing the molecular docking experiment, multiple different binding postures are generated for each lead compound, the binding postures are scored using the CDOCKER interaction energy, and the highest-scoring posture for each lead compound is selected for interaction analysis and visualization.

[0069] In some embodiments of this application, the CDOCKER interaction energy includes van der Waals interaction, electrostatic contribution, and internal ligand strain energy.

[0070] Specifically, the CDOCKER algorithm implemented in Discovery Studio 2021 was used for molecular docking experiments. This algorithm employs a CHARMm-based molecular dynamics method for ligand placement and scoring. The ligand binding sites for ERα and ERβ were defined using co-crystal ligand coordinates, and the binding site spheres were extended to... The radius is adjusted to accommodate different compound sizes and binding modes.

[0071] For each lead compound, ten distinct binding postures were generated through systematic conformational sampling and energy optimization. The docking protocol incorporates full ligand flexibility while treating the protein as rigid, consistent with high-resolution crystal structure data. Binding postures were assessed using the CDOCKER interaction energy, which combines van der Waals interactions, electrostatic contributions, and internal ligand strain energies. The highest-scoring posture for each compound was selected for detailed interaction analysis and visualization.

[0072] In some embodiments of this application, when performing ADMET property profiling on the screened lead compounds, the pharmacokinetic parameters of the screened lead compounds are analyzed, including water solubility, blood-brain barrier penetration potential, and human intestinal absorption probability.

[0073] In some embodiments of this application, the toxicity prediction specifically involves: performing a toxicological assessment on the screened lead compounds, the assessment including hepatotoxicity potential, carcinogenicity prediction for multiple species and sex combinations, acute oral toxicity and chronic toxicity, and skin irritation potential.

[0074] Specifically, the ADMET descriptor module in Discovery Studio 2021 was used to perform a comprehensive absorption, distribution, metabolism, excretion, and toxicity (ADMET) profiling of all compounds identified through virtual screening. The analysis covered key pharmacokinetic parameters, including water solubility, blood-brain barrier penetration potential, and probability of absorption into the human intestinal tract.

[0075] Toxicological assessments were performed using the Toxicity Prediction Computer-Aided Technology (TOPKAT) module, which provides predictive models for various toxicity endpoints. Specific assessments included hepatotoxicity potential, carcinogenicity prediction across multiple species and sex combinations, acute oral toxicity and chronic toxicity, and skin irritation potential.

[0076] Example 1

[0077] S1. The three-dimensional crystal structures of human estrogen receptor α (ERα) and β (ERβ) were retrieved from the Protein Database (PDB). Using Discovery Studio 2021, human estrogen receptor α (ERα) and β (ERβ) were prepared by removing water molecules, adding hydrogen atoms, and optimizing the side chain conformation. Co-crystal ligands were then identified. The ligand binding site is defined by all amino acid residues within the radius range.

[0078] S2. Based on the defined ligand binding sites and the three-dimensional crystal structures of human estrogen receptors α (ERα) and β (ERβ), using the interaction pharmacophore generation function implemented in Discovery Studio 2021, for each estrogen receptor subtype, 10 different pharmacophore models are systematically generated based on the ligand-receptor interaction patterns observed in the crystal structures, resulting in a total of 20 pharmacophore models. The generated pharmacophores mainly consist of hydrogen bond donors (D), aromatic rings (R), hydrophobic interactions (H), and repulsive volumes (S). Each pharmacophore model is assigned a unique identifier (ERα: ESR1_01 to ESR1_10; ERβ: ESR2_01 to ESR2_10). The three-dimensional crystal structures of human estrogen receptors α (ERα) and β (ERβ) and the three-dimensional structures of their corresponding ligands are as follows: Figure 2 As shown.

[0079] S3. The pharmacophore model was validated using 150 active ligands and 150 false positive molecules retrieved from the DUDE database. The corresponding receiver operating characteristic (ROC) curves of the pharmacophore model were generated. The pharmacophore models with an AUC value ≥ 0.8 were selected and identified as effective pharmacophore models.

[0080] The ROC curve analysis results are as follows: Figure 3As shown, the AUC values ​​ranged from 0.507 to 0.883. Using an AUC threshold of 0.8, eight valid pharmacophore models were identified for ERα and seven for ERβ. The highest-performing models demonstrated excellent discriminative ability, with ESR1_01 achieving an AUC of 0.883 (sensitivity: 0.807, specificity: 0.893) and ESR2_01 achieving an AUC of 0.842 (sensitivity: 0.807, specificity: 0.753); these validated pharmacophore models served as robust filters for subsequent virtual screening of bisphenol compounds.

[0081] S4. Bisphenol compounds were retrieved from the PubChem database and then manually screened to remove duplicate entries, constructing a bisphenol compound library containing 427 bisphenol compounds. Validated effective pharmacophore models were used as molecular filters to perform virtual screening of the bisphenol compound library. The screening process was implemented through the ligand profiling module in Discovery Studio 2021. During the virtual screening, when the scale-free fit threshold between a compound and the pharmacophore model was ≥2, the compound was determined to have binding ability to this pharmacophore model. Compounds with binding ability to multiple pharmacophore models were screened out as lead compounds.

[0082] The screening results are as follows: Using a scale-free fitting value threshold >2, 187 compounds capable of binding to at least one pharmacophore model were successfully identified, generating a total of 237 binding conformations. After applying the fitting value ≥2 criterion, 90 compounds were ultimately retained. The results and interaction patterns of the compounds with the pharmacophores during the screening are as follows: Figure 4-5 As shown, among the screened compounds, 42 exhibited the ability to bind simultaneously with three or more pharmacophore models, indicating strong estrogen receptor interaction potential. Two compounds, phenolol (CID: 3292100) and triphenolol (CID: 75304), exhibited the highest binding affinity, each forming favorable interactions with six different pharmacophore models, demonstrating their significant potential as endocrine disruptors. Figure 4 This demonstrates the number of compounds that successfully bind to each validated pharmacophore model (ESR1_01-ESR1_09 and ESR2_01-ESR2_09), showcasing the relative selectivity and binding ability of each pharmacophore model to the bisphenol compound library. Figure 5 The binding conformations and interaction patterns of phenolic alcohols (CID:3292100) and triphenolic alcohols (CID:75304) with multiple pharmacophore models are shown, and key pharmacological features involved in molecular recognition are highlighted, including hydrogen bond donors (D), aromatic rings (R), hydrophobic regions (H), and repulsive volumes (S).

[0083] S5. The CDOCKER algorithm was used to study the molecular docking of phenolic alcohol and triphenol with two estrogen receptor subtypes. Ten different binding postures were generated for each compound for the two receptors, and the highest-scoring conformation was selected for detailed interaction analysis: for ERα, the binding energies of phenolic alcohol and triphenol were 7.102 kcal / mol and 3.624 kcal / mol, respectively, while for ERβ, the corresponding values ​​were 22.006 kcal / mol and 19.992 kcal / mol, respectively.

[0084] like Figure 5 As shown, interaction analysis reveals that the two compounds bind primarily through hydrophobic interactions with key amino acid residues. For ERα binding, key residues include LEU346, ALA350, TRP383, LEU384, LEU387, LEU525, and CYS530, with MET421 additionally involved through van der Waals interactions. Notably, the triphenol forms an additional hydrogen bond with HIS524, potentially contributing to its enhanced binding stability compared to phenolic alcohols. For ERβ interactions, both compounds exhibit extensive hydrophobic contacts with multiple residues, including MET295, LEU298, LEU301, ALA302, MET336, LEU339, MET340, LEU343, PHE356, ILE373, ILE376, and LEU476. Hydrogen-bonded interactions were observed between phenolic alcohols and THR299, GLU305, and ARG346, while triphenolic alcohols formed hydrogen bonds with THR299 and GLU305. The higher binding energy observed with ERβ indicates a stronger affinity for both compounds compared to ERα.

[0085] S6. A comprehensive ADMET analysis and toxicological evaluation were performed on all 90 identified bisphenol compounds, and the results are as follows: Figure 6As shown, solubility assessment indicated that most compounds exhibited moderate to high solubility levels, with 78 compounds (86.7%) falling into category 2–3 on the solubility scale. However, absorption analysis revealed that 58 compounds (64.4%) exhibited poor absorption properties, potentially limiting their bioavailability. Toxicological predictions presented significant safety concerns, with 77 compounds (85.6%) predicted to be hepatotoxic, indicating potential liver injury upon exposure. Carcinogenicity assessments showed species- and sex-specific variations, with mouse models showing a balanced distribution between carcinogenic and non-carcinogenic predictions, while rat models, particularly males, exhibited a higher proportion of non-carcinogenic classifications. Acute toxicity parameters showed considerable variation, with oral LD50 values ​​ranging from 0.281 to 31.245 g / kg body weight, and chronic LOAEL values ​​spanning from 0.002 to 1.082 g / kg body weight. Most compounds (82.2%) were predicted not to cause skin irritation, while 16 compounds (17.8%) were associated with mild irritation potential.

[0086] Figure 6 Part A presents the DMET property distribution results: distribution of key pharmacokinetic parameters, including solubility levels (0-5), blood-brain barrier (BBB) ​​penetration levels (0-4), and absorption levels (0-3) in the compound library; Part B presents (B) toxicity prediction results, a summary of binary toxicity predictions, showing hepatotoxicity potential, plasma protein binding (PPB) tendency, and skin irritation classification. Stacked bar charts represent positive (red) and negative (green) predictions for hepatotoxicity and PPB binding; Part C presents carcinogenicity analysis: species- and sex-specific carcinogenicity predictions, comparing mouse and rat models in male and female populations. Based on the FDA TOPKAT model, results are categorized as non-carcinogenic (green), single carcinogenic (yellow), and multiple carcinogenic (red); Part D presents dose-response analysis: acute oral toxicity (LD50) in rats. 50 Distribution of LD50 and oral (LD50) values 50 (g / kg body weight) and inhalation (LC) 50 mg / m 3 Correlation analysis between toxicity endpoints ( / h). Scatter plots show the relationships between different toxicity measurements in the compound series.

[0087] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for assessing the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling techniques, characterized in that, Includes the following steps: Multiple pharmacophore models were generated based on the three-dimensional crystal structures of human estrogen receptors α and β. The generated pharmacophore models were validated using receiver operating characteristic (ROC) curves, and the effective pharmacophore model was selected. Construct a library of bisphenol compounds containing different bridging groups, substitution modes, and functional group modifications; The validated effective pharmacophore model was used as a molecular filter to virtually screen the bisphenol compound library. Based on the quality of the alignment between the compound and the pharmacophore, compounds with the ability to bind to multiple pharmacophore models were screened as lead compounds. Molecular docking experiments were performed on the screened lead compounds to evaluate the interaction mode and binding ability of the lead compounds with the human estrogen receptor; The selected lead compounds were subjected to ADMET property analysis and toxicity prediction.

2. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 1, characterized in that, When generating multiple pharmacophore models based on the three-dimensional crystal structures of human estrogen receptors α and β, including: Human estrogen receptor α and β proteins were prepared based on the three-dimensional crystal structures of the human estrogen receptor α and β, and co-crystal ligands of the human estrogen receptor α and β proteins were identified. Within a radius range of amino acid residues, ligand binding sites are defined based on the identified amino acid residues, and multiple pharmacophore models are generated based on the defined ligand binding sites and the three-dimensional crystal structures of the human estrogen receptors α and β.

3. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 2, characterized in that, When validating the generated pharmacophore models using receiver operating characteristic (ROC) curves, pharmacophore models with an AUC value ≥ 0.8 are considered valid pharmacophore models.

4. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 3, characterized in that, When constructing a bisphenol compound library containing different bridging groups, substitution modes, and functional group modifications, the following steps are taken: Bisphenol compounds were retrieved from the PubChem database and then manually screened to remove duplicate entries; The screened compounds were standardized using the molecular preparation protocol implemented in Discovery Studio 2021; Three-dimensional conformations were generated from the screened compounds using the FAST conformation search algorithm.

5. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 4, characterized in that, The standardization includes the neutralization of charged substances, tautomer standardization, and stereochemical partitioning where applicable.

6. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 5, characterized in that, When performing the virtual screening, if the scale-free fit value threshold between the compound and the pharmacophore model is ≥2, the compound is determined to have binding ability to the pharmacophore model.

7. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 6, characterized in that, During the molecular docking experiment, multiple different binding postures were generated for each lead compound. The binding postures were scored using the CDOCKER interaction energy, and the highest-scoring posture for each lead compound was selected for interaction analysis and visualization.

8. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 7, characterized in that, The CDOCKER interaction energy includes van der Waals interactions, electrostatic contributions, and internal ligand strain energy.

9. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 8, characterized in that, When performing ADMET property analysis on the screened lead compounds, the pharmacokinetic parameters of the screened lead compounds are analyzed, including water solubility, blood-brain barrier penetration potential, and human intestinal absorption probability.

10. The method for evaluating the toxicity of bisphenol compounds based on virtual screening and pharmacophore modeling technology according to claim 9, characterized in that, The toxicity prediction specifically involves conducting toxicological assessments on the selected lead compounds, including hepatotoxicity potential, carcinogenicity predictions for multiple species and sex combinations, acute oral toxicity and chronic toxicity, and skin irritation potential.

Images