Method for analyzing the hepatotoxicity difference between proton pump inhibitors and H2 receptor antagonists based on network toxicology

Abstract

The present invention discloses a method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology. The method of the present invention comprises the following steps: mining the action targets of drugs and liver toxicity disease risk genes in a database, constructing a protein interaction network of the drug-induced liver injury disease module DILI, and performing GO biological function and KEGG pathway enrichment analysis on the risk genes; overlapping the drug targets and the DILI disease module and analyzing the drug liver toxicity mechanism; calculating the proximity degree between the drug and the DILI disease module by using a network proximity strategy; and finally performing molecular docking on the drug molecule and the core target. The method of the present invention reveals that H2 receptor antagonists have stronger liver toxicity than proton pump inhibitors, explores the mechanism of potential liver injury of drugs, and provides a reference for the subsequent liver toxicity research of drugs.

Description

Technical Field

[0001] The present invention belongs to the technical field of bioinformatics, and particularly relates to a method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology. Background Art

[0002] Clinically, the treatment of gastric diseases mainly focuses on counteracting gastric acid secretion, aiming to achieve the treatment goal by regulating the gastric acid secretion of patients to a normal level. The drugs used to counteract excessive gastric acid secretion in existing clinical diagnoses and treatments are mainly divided into two categories: receptor antagonists and proton pump inhibitors. The former includes M receptor antagonists, H2 receptor antagonists, and gastrin receptor antagonists, and the latter are proton pump inhibitor rabeprazole drugs. Due to the premise of oral administration and the characteristics of drug metabolism itself, all proton pump inhibitors have undergone significant liver metabolism, while H2 receptor antagonists cannot effectively inhibit gastric acid secretion caused by daytime diet and have poor inhibitory effects on all-day gastric acid secretion. Therefore, to improve the treatment effect, the treatment course must be extended. In relevant animal experimental studies, long-term administration of ranitidine (up to 100 mg / Kg) has produced obvious toxicity to the livers of experimental animals; the long-term use of famotidine in clinical practice can also cause clinical hepatitis. There are many reports on the occurrence of liver toxicity during the clinical use of proton pump inhibitors and H2 receptor antagonists. At the same time, the adverse reaction event reporting system in the FDA database includes a series of disease phenotypes of liver toxicity effects such as hepatitis, jaundice, cholestasis, and abnormal liver function caused by proton pump inhibitors and H2 receptor antagonists in the past few decades of use. Therefore, it is necessary to explore the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists to provide a basis for rational drug use in clinical practice. Summary of the Invention

[0003] To solve the problems in the prior art, the present invention provides a method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology. This method explores the liver toxicity mechanisms and their differences of two types of drugs for treating excessive gastric acid secretion through network medicine analysis and molecular docking technology, demonstrates the toxicological mechanism of drug liver toxicity at the molecular level according to the results of network overlap and gene enrichment, and compares and analyzes the liver toxicity differences between the two types of drugs through network proximity strategies and molecular docking technology, providing new ideas and references for subsequent research on liver toxicity mechanisms and drug adverse reactions.

[0004] The object of the present invention is achieved by the following technical solutions:

[0005] A method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology, comprising the following steps:

[0006] (1). Acquisition and processing of drug targets: Obtain the action targets of proton pump inhibitors and H2 receptor antagonists through online databases, and process the targets in the protein database to match relevant gene symbols.

[0007] (2). Acquisition and processing of liver toxicity risk genes: Use Chemical and Drug Induced LiverInjury, Chemical and Drug Induced Liver Injury, Chronic as keywords to search in the MeSH database. Based on the search results, obtain liver toxicity risk genes from the OMIM database and the GWAS database, and remove genes that failed to map successfully and duplicates.

[0008] (3). Construction of the Drug-Induced Liver Injury (DILI) disease module: Locate the collected liver toxicity risk genes in STRING, observe the distribution of disease risk genes based on network localization, and verify the effectiveness of the DILI disease module through network randomization to obtain the DILI disease module.

[0009] (4). Network proximity between drugs and the DILI disease module: For the disease risk gene set and the drug target set, calculate the shortest path length between drug targets and disease risk genes based on the STRING network using code. Calculate the path length between randomly selected risk genes and drug targets as a reference distance to obtain the relative average shortest distance between drugs and the DILI disease module.

[0010] Use the network proximity strategy to evaluate the closeness between drugs and the DILI disease module, and use code to calculate the proximity between drug targets and DILI disease module risk genes: For a given V (disease gene set) and T (drug target set), d(V,T) represents the shortest path length between v and t in the network nodes, and dc(V,T) represents the average shortest path length between drug targets and disease risk genes in the network. The calculation formula is as follows:

[0011] When the mean μd(V, T) and standard deviation σd(V, T) of the reference distance distribution are used to convert the observed distance into a normalized distance, the calculation formula for the relative average shortest distance Zdc between drugs and the DILI disease module is as follows. Through network proximity calculations, obtain the network proximity results.

[0012]

[0013] (5). Network overlap between drugs and DILI disease modules: Through coding, the drug targets and risk genes are overlapped in the STRING network to calculate the ds value between the drug targets and risk genes, and to analyze the differences in the mechanisms of drug-induced liver toxicity and the network correlation between drugs and disease modules. ds = 0 indicates that the disease risk genes in the overlapping network are exactly the same as the drug targets; ds = 1 indicates that the two are adjacent and the distance is 1; ds = 2 indicates that the shortest distance between the two is 2; ds > 2 indicates that the shortest distance between the two is greater than 2; ds = NA is regarded as no interaction relationship between the two in the PPI. The relevant biological processes and regulatory pathways of the specific targets of the two types of drugs acting in the overlapping network reflect the reasons and differences in the mechanisms of liver toxicity of the two types of drugs at the molecular level and during the occurrence and development of the disease.

[0014] (6). Gene enrichment of DILI disease modules: All risk genes of the DILI disease modules are uploaded to the DAVID bioinformatics database for gene enrichment, and the results obtained are used to analyze the biological processes, molecular functions and related signaling pathways of drug-induced liver injury diseases, and to explore the related mechanisms of drug-induced liver injury.

[0015] (7). Screening the core targets of drugs acting on disease modules: The drug targets are overlapped with the disease risk genes to obtain the candidate core targets of drugs acting on disease modules. The frequencies of genes involved in all signaling pathways in the gene enrichment results are counted, and the frequency rankings corresponding to the candidate core targets are counted. The disease risk genes are uploaded to the STRING database for analysis to obtain the protein-protein interaction network (PPI) and its txt data. The NetworkAnalyzer module in Cytoscape 3.9.0 software is used to analyze the maximum degree of freedom value of the txt data (network nodes). Combining the degree value ranking of the candidate core targets in the PPI network and the gene frequency ranking of the candidate core targets, the core targets of drugs acting on disease modules are screened and determined.

[0016] (8). Molecular docking of drugs and core targets: The corresponding human-derived protein structure of the core target is downloaded from the PDB protein database, and the 2D structure of the drug is downloaded from the PubChem database as a ligand file, and molecular docking is performed using DiscoveryStudio 2019 software.

[0017] Furthermore, the proton pump inhibitors selected in the present invention include: esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole. The H2 receptor antagonists selected include: cimetidine, famotidine, nizatidine, ranitidine, roxatidine. The above-mentioned drugs are first-line drugs that have been on the market earlier in the same drug type, with the best clinical usage and treatment effects, and are representative for research. The structural files of drug molecules are sourced from the DrugBank database, and the action targets of the drugs are identified online through pharmacophore mapping in the PharmMapper database.

[0018] In the present invention, the online drug database in step (1) is selected from Drugbank; the database for drug target mapping is selected from PharmMapper; the protein database is selected from Uniprot; and the target processing is to convert the drug target ID name into UniProt ID, and the drug targets are obtained after removing duplicates.

[0019] In the present invention, the data sources of the liver toxicity risk genes in step (2) are from the OMIM database and the GWAS database, and the standardized terms of liver toxicity diseases are selected based on the MeSH database.

[0020] In the present invention, in step (3), after localizing the liver toxicity risk genes collected in step (2) to STRING, the distribution of disease risk genes is observed according to the network localization, and the disease risk genes are significantly localized; after verifying the effectiveness of the results through network randomization, the constructed DILI disease module is obtained.

[0021] In the present invention, when using the network proximity strategy to evaluate the proximity between the drug and the DILI disease module in step (4), for a given V (disease gene set) and T (drug target set), d(V,T) represents the shortest path length between v and t in the network nodes, and dc(V,T) represents the average shortest path length between the drug targets and the disease risk genes in the network; the proximity between randomly selected targets and risk genes is calculated to obtain the reference distance, and this process is randomized 1000 times. To avoid repeated selection of the same network nodes, when the mean μd(V,T) and standard deviation σd(V,T) of the reference distance distribution are used to convert the observed distance into a normalized distance, the relative average shortest distance Zdc between the drug and the DILI disease module is obtained.

[0022] In the present invention, step (5) is to overlap the drug targets in step (1) with the DILI disease module constructed in step (3) based on the STRING network, and calculate the ds value between the drug targets and the risk genes. ds = 0 indicates that the disease risk genes in the overlapping network are exactly the same as the drug targets; ds = 1 indicates that the two are adjacent and the distance is 1; ds = 2 indicates that the shortest distance between the two is 2; ds > 2 indicates that the shortest distance between the two is greater than 2; ds = NA is regarded as no interaction relationship between the two in the PPI.

[0023] In the present invention, step (6) is to upload the liver toxicity risk genes of the DILI disease module in step (3) to the DAVID biological information database for gene enrichment work, and obtain the results of gene ontology biological processes, molecular functions, cellular components and related biological pathways.

[0024] In the present invention, in step (7), the drug targets are overlapped with the disease risk genes to obtain the candidate core targets of the drug acting on the disease module. The frequencies of the genes involved in all signal pathways in the gene enrichment results are counted, and the frequency rankings corresponding to the candidate core targets are counted. The disease risk genes are uploaded to the STRING database to obtain the target protein-protein interaction network (PPI) and its txt data. The NetworkAnalyzer module in the Cytoscape 3.9.0 software is used to analyze the maximum degree of freedom value of the txt data (network nodes). Combining the degree value ranking of the candidate core targets in the PPI network and the gene frequency ranking of the candidate core targets, the core targets of the drug acting on the disease module are screened and determined. For core target screening, the core targets are screened according to the maximum degree of freedom value ranking in the PPI. The degree of a node is defined as the number of edges connected to the node, representing the topological importance of the network node, which is the most convenient and feasible method for analyzing drug targets. The larger the degree value of a risk gene, the more risk genes it is connected to, so this risk gene is more likely to be related to hepatotoxicity and more likely to be the core target of drug action in the hepatotoxicity disease module. At the same time, the frequency of the candidate core targets appearing in the signal pathway results of the gene enrichment in the disease module is considered. The candidate targets that appear frequently are more closely related to the process of disease occurrence and development, which is an important auxiliary reference for screening the key targets of drug action in the DILI disease module.

[0025] In the present invention, in step (8), the human protein structures of the core targets of proton pump inhibitors and H2 receptor antagonists are downloaded from the Protein Data Bank (PDB): the three-dimensional protein structures determined by X-ray crystallography; the crystal resolution is less than 3 Å; the protein conformations confirmed and supported by the publicly available information are preferred. The 2D structures of the drugs are downloaded from the PubChem database as ligands and saved in the pdbqt format. Molecular docking is performed using the Discovery Studio 2019 software to evaluate the binding affinity between the drug molecules and the core targets.

[0026] Compared with the prior art and related liver toxicity studies, the present invention has the following advantages:

[0027] 1. A method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology and molecular docking according to the present invention successfully establishes a drug-induced liver injury disease module through data mining and the application of network methods. By using overlapping network analysis and gene enrichment, the liver toxicity mechanisms of proton pump inhibitors and H2 receptor antagonists are revealed. Using the network proximity strategy and molecular docking technology, the liver toxicity differences between the two types of drugs are demonstrated computationally, verifying the relevant conclusions of traditional experiments, and providing new ideas and references for the research on liver toxicity mechanisms and the comparative study of drug adverse reactions.

[0028] 2. A method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology and molecular docking according to the present invention, for the first time, uses the overlapping network of drug targets and disease module risk genes to comparatively analyze the different liver toxicity mechanisms of proton pump inhibitors and H2 receptor antagonists, revealing the specific signal pathways, regulatory pathways, and biological processes of the two types of drugs that cause liver toxicity, and providing a valuable reference for subsequent related research.

[0029] 3. A method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology and molecular docking according to the present invention, for the first time, uses the network proximity strategy to reveal the proximity degree between proton pump inhibitors and H2 receptor antagonists and the disease module. The molecular docking verification results also show that H2 receptor antagonists have a stronger binding ability to the core targets of the disease module compared to proton pump inhibitors, reflecting more potential liver side effects, and providing a certain theoretical reference for drug liver toxicity research and the selection of drugs for clinically treating related diseases. Description of the Drawings

[0030] Figure 1 is a flow chart of the steps for studying the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists in the embodiments of the present invention;

[0031] Figure 2It is a diagram showing the differences in drug targets between proton pump inhibitors and H2 receptor antagonists in the embodiments of the present invention;

[0032] Figure 3 It is a network diagram of the DILI disease module in the embodiments of the present invention;

[0033] Figure 4 It is a network proximity result diagram of proton pump inhibitors and H2 receptor antagonists in the embodiments of the present invention;

[0034] Figure 5 It is a diagram of specific targets of H2 receptor antagonists in the embodiments of the present invention;

[0035] Figure 6 It is a diagram of specific targets of proton pump inhibitors in the embodiments of the present invention;

[0036] Figure 7 It is a diagram of the top 10 results of DILI gene enrichment ranking in the embodiments of the present invention. Detailed implementation manners

[0037] The following description of the embodiments is only used to help understand the method and its core idea of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention. The steps of the study on the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists in the embodiments of the present invention are as Figure 1 shown.

[0038] Example 1

[0039] A method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology

[0040] 1. Method steps

[0041] 1.1 Acquisition and processing of drug targets

[0042] Proton pump inhibitors: (esomeprazole, lansoprazole, omeprazole, pantoprazole, rabeprazole); H2 receptor antagonists (cimetidine, famotidine, nizatidine, ranitidine, roxatidine). For 10 drugs, download the structure files of drug molecules (saved in SDF format) by entering the English generic names of the drugs in the DrugBank database. Upload the drug structure files to the Pharmmapper database and online identify the action targets of the drugs through pharmacophore mapping. Standardize the obtained drug targets on the Uniprot website, uniformly convert the target ID type to UniProt ID, and obtain the respective action targets of the drugs after removing duplicates. Use the online analysis tool venny to perform differential analysis on the targets of the two types of drugs.

[0043] 1.2 Obtaining and processing of genes at risk of liver toxicity

[0044] Standardized terms for liver toxicity diseases were determined in the MeSH database. According to the determined standardized terms: Chemical and Drug Induced Liver Injury, Chemical and Drug Induced Liver Injury, Chronic, as keywords, 49 entries were retrieved in the MeSH database. 691 genes at risk of liver toxicity were collected from these 49 entries in the OMIM database; 38 genes at risk of liver toxicity were collected from the GWAS database. After removing genes that failed to be mapped successfully and duplicates, a total of 683 genes at risk of liver toxicity were collected.

[0045] 1.3 Construction of drug-induced liver injury disease modules

[0046] Locate the disease risk genes in 1.2 on the STRING network through coding. Observe the distribution of disease risk genes based on the network location. Verify the effectiveness of the DILI disease module through network randomization, and verify the quality of the constructed DILI disease module using network topology parameters to obtain the network location results and disease module network of the DILI disease module.

[0047] 1.4 Network proximity between drugs and DILI disease modules

[0048] Use the network proximity strategy to evaluate the proximity between drugs and the DILI disease module, and use code to calculate the proximity between drug targets and risk genes in the DILI disease module: For a given V (disease gene set) and T (drug target set), d(V,T) represents the shortest path length between v and t in the network nodes, and dc(V,T) represents the average shortest path length between drug targets and disease risk genes in the network. The calculation formula is as follows:

[0049] Calculate the proximity between randomly selected targets and risk genes to obtain the reference distance. This process is randomized 1000 times.

[0050] To avoid repeated selection of the same network nodes, when the mean μd (V,T) and standard deviation σd (V,T) of the reference distance distribution are used to convert the observed distance into a normalized distance, the calculation formula for the relative average shortest distance Zdc between drugs and the DILI disease module is as follows. Through the calculation of network proximity, the network proximity results are obtained.

[0051]

[0052] 1.5 Network overlap of drugs and the DILI disease module

[0053] To explore the differences in the liver toxicity mechanisms of two types of drugs and the network association between drugs and the disease module, use code to perform network overlap of drug targets and the DILI disease module based on the STRING network, and calculate the ds value of drug targets and risk genes. ds = 0 indicates that the disease risk genes in the overlapping network are exactly the same as the drug targets; ds = 1 indicates that the two are adjacent and the distance is 1; ds = 2 indicates that the shortest distance between the two is 2; ds > 2 indicates that the shortest distance between the two is greater than 2; ds = NA is regarded as no interaction relationship between the two in the PPI. The relevant biological processes and regulatory pathways of the specific targets of the two types of drugs acting in the overlapping network reflect the reasons and differences in the liver toxicity mechanisms of the two types of drugs at the molecular level and during the occurrence and development of the disease.

[0054] 1.6 Gene enrichment of risk genes in the DILI disease module

[0055] Upload all risk genes, core risk genes, and risk genes of the largest connected component in the DILI disease module to the DAVID database for gene enrichment to obtain information such as gene ontology biological processes, molecular functions, cellular components, and related biological pathways, as well as gene enrichment results.

[0056] 1.7 Screening of core targets of drugs acting on the disease module

[0057] Rank the genes that appear in all signaling pathways in the gene enrichment results of the core risk genes. A total of 314 risk genes were counted. Overlap the drug targets of proton pump inhibitors and H2 receptor antagonists with the disease risk genes to obtain 48 candidate core targets for the drugs acting on the disease module, and count the frequency data of these 48 candidate core targets among the 314 risk genes. Upload all the risk genes to the STRING database to analyze the candidate core targets, obtain the target protein-protein interaction network (PPI) and its txt data, and use the NetworkAnalyzer module in the Cytoscape 3.9.0 software to analyze the maximum degree of freedom value of the txt data (network nodes). According to the degree value ranking in the PPI network and the frequency ranking of the candidate core targets, screen the core targets of the drugs acting on the disease module for the next step of molecular docking.

[0058] 1.8 Molecular Docking of Drugs and Core Targets

[0059] Download the human-derived protein structure of the core target of the drug acting on the PDB protein database. The three-dimensional protein structure was determined by X-ray crystallography; the crystal resolution is less than 3 Å; the protein conformation confirmed and supported by public information is preferred. Download the 2D structure of the drug as a ligand from the PubChem database, save it in the pdbqt format, and use the DiscoveryStudio 2019 software for molecular docking to evaluate the binding affinity between the drug molecule and the core target.

[0060] 2. Results

[0061] 2.1 Drug Targets of Proton Pump Inhibitors and H2 Receptor Antagonists

[0062] After data collection and processing, the number of drug targets of proton pump inhibitors and H2 receptor antagonists is shown in Table 1.

[0063]

[0064] A total of 424 drug targets were collected for the 5 drugs of H2 receptor antagonists, and a total of 379 drug targets were collected for the 5 drugs of proton pump inhibitors. Use the online analysis tool venny to perform a differential analysis of the targets of the two types of drugs. The results are shown in Figure 2 .

[0065] 2.2 Obtaining and Processing of Liver Toxicity Risk Genes

[0066] Based on the standardized names of liver toxicity diseases determined by the MeSH database, the keywords "Chemical and Drug Induced Liver Injury" and "Chemical and Drug Induced Liver Injury, Chronic" were selected for retrieval. A total of 691 liver toxicity risk genes were collected from the OMIM database; 38 liver toxicity risk genes were collected from the GWAS database. After removing the genes with unsuccessful mapping and duplicates, a total of 683 liver toxicity risk genes were collected.

[0067] 2.3 DILI disease module

[0068] The 683 risk genes were mapped to the STRING network, and the network analysis tool of Cytoscape 3.9.0 was used to analyze the network topological properties of the disease module. The results showed that 595 risk genes were significantly mapped compared with the randomly selected gene set (Z = 23.68, p < 0.001, 1000 times of random sampling). After verifying the network topological parameters, the quality of the DILI disease module was good. In the PPI network of the DILI disease module, 595 risk genes aggregated to form 2721 edges, and on average, each pair of risk genes had 1.205 edges. 486 risk genes were interconnected and were called core risk genes; the largest connected component of the disease module consisted of 458 risk genes. The constructed DILI disease module network is shown in Figure 3 .

[0069] 2.4 Network proximity analysis of drugs and the DILI disease module

[0070] The relative average neighborhood distance parameter Zdc between 10 drugs and the DILI disease module was calculated using the network proximity strategy. A Zdc value less than 0 indicates that the drug is adjacent to the disease module, and the absolute value is positively correlated with the degree of proximity. The network proximity results are as shown in Figure 4 . The Zdc values of the 10 drugs and the DILI disease module were all less than 0, indicating that the drugs are adjacent to the DILI disease module, showing the potential liver toxicity of the drugs. In the calculation method based on the same disease module and code environment, H2 receptor antagonists are closer to the disease module than proton pump inhibitors, indicating that H2 receptor antagonists are more likely to cause liver toxic side effects than proton pump inhibitors.

[0071] 2.5 Network overlap analysis of drugs and the DILI disease module

[0072] There are 401 risk genes of H2 receptor antagonists with ds ≤ 1 and 399 proton pump inhibitors in the overlapping network results. There are 175 H2 receptor antagonists among the risk genes with ds = 2; 177 proton pump inhibitors. There are 18 risk genes with ds > 2 in both types of drugs in the overlapping network and they are exactly the same.

[0073] According to the differential targets of the two types of drugs in the overlapping network, explore the differences in the mechanisms of liver injury caused by the two types of drugs at the molecular level and in the related processes of disease occurrence and development. The specific targets of H2 receptor antagonist drugs acting on disease modules are shown in Figure 5 . The drug acting on PPARD can reduce the release of apoptosis markers and inflammatory cytokines, thereby improving apoptosis and finally reducing liver injury. The expression of CES1 can reduce neutral lipid accumulation and improve the histological morphology of the liver. The expression level of CES1 is positively correlated with the increase in human lipid storage and blood lipid concentration. Excessive fat increase in the liver may lead to the occurrence of fatty hepatitis and even cirrhosis. The weakening of human CES1 activity has a beneficial effect on liver lipid metabolism. Bone marrow mesenchymal stem cells can produce a high level of BMP7, which can counteract TGFβ1-induced liver cirrhosis in mice. The gene knockout experiment of BMP7 confirmed that bone marrow mesenchymal stem cells relieve liver cirrhosis by producing BMP7 to counteract liver TGFβ1-induced fibrosis. The related drugs of H2 receptor antagonists can act on the BMP7 target, inhibit its antifibrotic function, and induce the body to show liver fibrosis and even more severe liver toxicity phenotypes.

[0074] The expression of the target protein in liver cells can be manifested as playing different functions at different stages of liver diseases and participating in some complex signal transduction processes. Drugs can act on the target and thus have relevant effects on the occurrence and development process of diseases. The specific targets of proton pump inhibitors acting on disease modules are shown in Figure 6 . The expression of ARG1 in liver cancer cells is significantly down-regulated. It plays a key oncogenic role in the progression of primary liver cancer by promoting epithelial-mesenchymal transition; ESR2 and related pathways are related to the severity of primary liver cancer; the lack of ALDH2 will enhance alcohol-induced liver injury. The gene knockout of ALDH2 in animal experiments can further increase the production of reactive oxygen species induced by TGFβ1; in the liver injury model, activated STAT3 in hepatocytes is proven to be an effective anti-inflammatory factor that can protect hepatocytes from injury; it is found that RAC2 is highly expressed in a carbon tetrachloride-induced acute liver injury model in mice, and the knockout of the RAC2 gene significantly reduces the release of these pro-inflammatory substances.

[0075] 2.6 Gene enrichment in the DILI disease module

[0076] In the GO functional analysis results, the top 10 biological process analyses of the enrichment results of biological processes (BP), molecular functions (MF), and cellular components (CC) with P < 0.05 were selected. The results showed that drug-induced liver injury mainly involved biological processes such as inflammatory response, immune response, negative regulation of apoptotic process, positive regulation of transcription from RNA polymerase II promoter, response to lipopolysaccharide, etc.; the cellular components involved were cytosol, plasma membrane, cytoplasm, perinuclear region of cytoplasm, etc.; the molecular functions involved were protein binding, enzyme binding, receptor binding, transcription factor binding, cytokine activity, etc.; looking at the KEGG enrichment analysis results (p < 0.05), the top 10 signals mainly involved Hepatitis C, Hepatitis B, Pathways in cancer, Toll-like receptor signaling pathway. The top 10 results of DILI gene enrichment are shown in Figure 7 。

[0077] 2.7 Core targets of drug action in the disease module

[0078] According to the degree value ranking in the PPI network and the frequency ranking of candidate core targets, the UniProt IDs and protein names of the 7 core targets of drug action in the disease module are shown in Table 2.

[0079]

[0080] 2.8 Molecular docking of drugs and core targets

[0081] Molecular docking was performed using Discovery Studio 2019 software to evaluate the binding affinity between drug molecules and core targets, and the results are shown in Table 3.

[0082]

[0083] The CDOCKER ENERGY value of the binding energy is greater than 0, indicating that the ligand drug molecule can spontaneously bind to the target receptor protein. The larger the value, the closer the binding and the better the affinity. We observed that the binding energy values of the target protein and drug molecules of the same type are relatively close, reflecting the similarity of the drug structures. Most of the values in the table are greater than 0, indicating that the drugs can spontaneously bind to the targets. Whether observed from the overall or individual dimensions of H2 receptor antagonists and proton pump inhibitors, the binding of H2 receptor antagonists to all core targets is far better than that of proton pump inhibitors. The results of molecular docking confirm their stronger binding ability compared to proton pump inhibitors. A better binding situation indicates a stronger interaction between the drug and the core target, which in turn affects the regulation of pathways and biological processes in the disease development process and is more likely to cause liver toxicity during drug use.

[0084] Based on the drug-induced liver injury caused by the long-term use of H2 receptor antagonists and proton pump inhibitors during treatment, drug liver toxicity was established as a disease module. According to the network overlap, gene enrichment, and molecular docking of drug targets and the DILI disease module, a network medicine analysis method was used to explore the relevant mechanisms of drugs and liver toxicity.

[0085] The results of the differential analysis of the overlapping network show that drug targets related to liver injury are involved in many complex signaling pathways, including inflammation-related signaling pathways, liver disease-related signaling pathways, and immune-related signaling pathways. Drugs can act on different targets to affect related biological processes, influence the immune process, cell proliferation and differentiation, the process of inflammatory response, and other signaling pathways that can cause liver injury to play an internal role.

[0086] Through the construction of the PPI network, 7 core targets were screened out, including MAPK1, MAP2K1, RAF1, AKT1, GRB2, MAPK14, and EGFR. RAF1 is involved in the signal pathways of normal cell growth and many carcinogenic transformations. The expression of RAF1 is significantly increased in hepatocellular carcinoma, and its high expression is related to cell proliferation, chemoresistance, and tumor invasiveness. Melatonin is a highly efficient antioxidant produced by the pineal gland, with antioxidant and free radical scavenging effects, and can inhibit liver injury and hepatocyte oxidative stress. Some studies have confirmed that melatonin induces the expression of let7i-3p in hepatocellular carcinoma cells, and further demonstrated that let7i-3p directly inhibits the translation of RAF1 protein and the activation of the carcinogenic pathway downstream of RAF1, thereby inhibiting the growth and metastasis of hepatocellular carcinoma cells and promoting their apoptosis.

[0087] Proton pump inhibitors act on H+ / K+-ATPase to exert acid-suppressing effects. This enzyme is only present on the surface of gastric parietal cells. The H2 receptor is not only present in gastric parietal cells but also in other tissues. Therefore, compared with H2 receptor antagonists, proton pump inhibitors have the advantages of specific action, high selectivity, and fewer side effects. From the perspective of the generation of drug side effects and the molecular-level mechanism of drug action, the results of the network proximity strategy show that H2 receptor antagonists are closer to the DILI disease module than proton pump inhibitors. Molecular docking verification shows that proton pump inhibitors and H2 receptor antagonists have good binding abilities with the core action targets, reflecting the potential liver toxicity characteristics of the drugs. At the same time, H2 receptor antagonists have better binding characteristics with the core targets than proton pump inhibitors, reflecting their stronger potential liver toxicity, which is consistent with the existing conclusions.

[0088] In summary, based on the overlapping network results of drug targets and the DILI disease module and the results of gene enrichment in the disease module, the network medicine analysis method was used to preliminarily explore the liver toxicity-related mechanisms of proton pump inhibitors and H2 receptor antagonists. The results show that the mechanisms of proton pump inhibitors and H2 receptor antagonists causing DILI involve multi-target and multi-pathway regulation. The results of network proximity and molecular docking reflect the relevance between drugs and diseases, and demonstrate at the molecular level by computational means that H2 receptor antagonists have more potential liver toxicity than proton pump inhibitors, providing a reference for the study of liver toxicity and drug adverse reactions.

Claims

1. A method for analyzing the liver toxicity differences between proton pump inhibitors and H2 receptor antagonists based on network toxicology, comprising the following steps: (1). Acquisition and processing of drug targets: Obtain the action targets of proton pump inhibitors and H2 receptor antagonists through an online database, and process the targets in the protein database to match relevant gene symbols; (2). Obtaining and processing of liver toxicity risk genes: Use Chemical and Drug Induced LiverInjury, Chemical and Drug Induced Liver Injury, Chronic as keywords to search in the MeSH database, and obtain liver toxicity risk genes from the OMIM database and GWAS database according to the search results, removing genes that failed to map successfully and duplicates; (3). Constructing a drug-induced liver injury disease module: Locate the collected liver toxicity risk genes in STRING, observe the distribution of disease risk genes according to network localization, and verify the effectiveness of the DILI disease module through network randomization to obtain the DILI disease module; (4). Network proximity between drugs and the DILI disease module: The network proximity strategy was used to evaluate the closeness between drugs and the DILI disease module, and code was used to calculate the proximity between drug targets and risk genes in the DILI disease module. For a given set of disease genes V and a set of drug targets T, d(V,T) represents the shortest path length between v and t in the network nodes, and dc(V,T) represents the average shortest path length between drug targets and disease risk genes in the network. The calculation formula is as follows: ; The calculation formula of the relative average shortest distance Zdc between the drug and the DILI disease module is as follows. Through network proximity calculation, network proximity results are obtained; (5). Network overlap between the drug and the DILI disease module: Use code to perform network overlap of drug targets and risk genes in the STRING network, calculate the ds value between drug targets and risk genes, and analyze the differences in drug liver toxicity mechanisms and the network association between the drug and the disease module; (6). Gene enrichment of the DILI disease module: Upload all risk genes of the DILI disease module to the DAVID bioinformatics database for gene enrichment, and use the obtained results to analyze the biological processes, molecular functions, and related signaling pathways of drug-induced liver injury diseases, and explore the related mechanisms of drug-induced liver injury; (7). Screening the core targets of the drug acting on the disease module: Overlap the drug targets with the disease risk genes to obtain candidate core targets of the drug acting on the disease module, count the frequencies of genes involved in all signaling pathways in the gene enrichment results, and count the frequency rankings corresponding to the candidate core targets; Upload the disease risk genes to the STRING database for analysis to obtain the target protein-protein interaction network PPI and its txt. data, and use the NetworkAnalyzer module in Cytoscape 3.9.0 software to analyze the maximum degree of freedom value of the txt. data; Combine the degree value ranking of the candidate core targets in the PPI network and the gene frequency ranking of the candidate core targets to screen and determine the core targets of the drug acting on the disease module; (8). Molecular docking of the drug and the core target: Download the corresponding human-derived protein structure of the core target from the PDB protein database, download the 2D structure of the drug as a ligand file from the PubChem database, and use DiscoveryStudio 2019 software for molecular docking.

2. The method according to claim 1, wherein: The proton pump inhibitors include esomeprazole, lansoprazole, omeprazole, pantoprazole and rabeprazole; the H2 receptor antagonists include cimetidine, famotidine, nizatidine, ranitidine and roxatidine.

3. The method according to claim 1 or 2, characterized in that: In step (1), the online drug database selected is Drugbank; the database for drug target mapping selected is PharmMapper; the protein database selected is Uniprot; for target processing, the drug target ID names are converted into UniProt IDs, and after removing duplicates, the drug targets are obtained.

4. The method according to claim 1, characterized in that: In step (2), the data sources of liver toxicity risk genes come from the OMIM database and the GWAS database, and the standardized terms of liver toxicity diseases are selected based on the MeSH database.

5. The method according to claim 1, characterized in that: In step (3), after localizing the liver toxicity risk genes collected in step (2) to STRING, the distribution of disease risk genes is observed according to network localization, and the disease risk genes are significantly localized; after verifying the effectiveness of the results through network randomization, the constructed DILI disease module is obtained.

6. The method according to claim 1, characterized in that: When using the network proximity strategy in step (4) to evaluate the proximity between a drug and the DILI disease module, for a given set of disease genes V and a set of drug targets T, d(V,T) represents the shortest path length between v and t in the network nodes, and dc(V,T) represents the average shortest path length between the drug targets and the disease risk genes in the network; the proximity between randomly selected targets and risk genes is calculated to obtain a reference distance. This process is randomized 1000 times. To avoid repeated selection of the same network nodes, when the mean μd(V,T) and standard deviation σd(V,T) of the reference distance distribution are used to convert the observed distance into a normalized distance, the relative average shortest distance Zdc between the drug and the DILI disease module is obtained.

7. The method according to claim 1, characterized in that: In step (5), the drug targets in step (1) and the DILI disease module constructed in step (3) are overlapped based on the STRING network, and the ds value between the drug targets and the risk genes is calculated; ds = 0 indicates that the disease risk genes in the overlapping network are exactly the same as the drug targets; ds = 1 indicates that the two are adjacent and the distance is 1; ds = 2 indicates that the shortest distance between the two is 2; ds > 2 indicates that the shortest distance between the two is greater than 2; ds = NA is regarded as no interaction relationship between the two in the PPI.

8. The method according to claim 1, wherein: In step (6), the liver toxicity risk genes of the DILI disease module in step (3) are uploaded to the DAVID biological information database for gene enrichment work, and the results of gene ontology biological processes, molecular functions, cellular components and related biological pathways are obtained.

9. The method according to claim 1, wherein: Step (8) Download the human protein structures of the core targets of proton pump inhibitors and H2 receptor antagonists in the PDB protein database: three-dimensional protein structures determined by X-ray crystallography; crystal resolution less than 3 Å; protein conformations confirmed and supported by publicly available information are preferred. Download the 2D structure of the drug as a ligand in the PubChem database, save it in the pdbqt format, and perform molecular docking using DiscoveryStudio 2019 software to evaluate the binding affinity between the drug molecule and the core target.

Images