Non-antibiotic small-molecule inhibitor of clostridium difficile toxin B and screening method of non-antibiotic small-molecule inhibitor

Abstract

The invention discloses a non-antibiotic small-molecule inhibitor of clostridium difficile toxin B and a screening method of the non-antibiotic small-molecule inhibitor. According to the method, a plurality of broad-spectrum small-molecule inhibitors targeting all TcdB receptor binding interfaces at the same time are found through virtual screening and a multi-stage calculation strategy, and the inhibition effect of the broad-spectrum small-molecule inhibitors is verified by combining cell experiments; a lead compound and an action mechanism theoretical support are provided for development of non-antibiotic anti-CDI drugs, and a screening strategy and a screening result can also provide a theoretical basis and an experimental basis for development of a next-generation precise anti-CDI therapy.

Description

Technical Field

[0001] The present invention belongs to the technical field of new medical applications and relates to non-antibiotic small molecule inhibitors of Clostridium difficile toxin B and a screening method thereof. Background Art

[0002] Clostridium difficile infection ( Clostridioides difficile Colonic infection (CDI) is a toxin-mediated injury to the colonic epithelium with diverse clinical manifestations ranging from self-limited diarrhea to life-threatening complications such as pseudomembranous colitis, toxic megacolon, intestinal perforation, and systemic inflammatory response.

[0003] Clostridium difficile toxin ( Clostridioides difficile Toxins (CDTs) are important virulence factors for the pathogenesis of Clostridium difficile, mainly including toxin A (TcdA) and toxin B (TcdB), as well as binary toxins expressed in highly virulent strains. Both TcdA and TcdB belong to the large clostridial toxins (LCT) family, which enter intestinal epithelial cells through receptor-mediated endocytosis and glycosylate small guanosine triphosphatase (GTPase) proteins, leading to cytoskeletal dysfunction and cell death. In animal models, TcdA + TcdB - The strains were less pathogenic, while TcdA - TcdB + The strains showed higher pathogenicity, so TcdB dominated the pathogenicity of CDI due to its higher toxicity and wider host response.

[0004] Currently, treatment for C. difficile primarily relies on antibiotics, such as vancomycin and fidaxomicin. However, these drugs further disrupt the balance of intestinal flora, leading to higher recurrence rates (25.3% and 15.4%, respectively) and increased risk of transmission of drug-resistant strains (such as RT027 / NAP1 / ST1 and RT078 / ST11). Therefore, the development of non-antibiotic small molecule inhibitors targeting C. difficile has become an important direction in the research and development of anti-infective drugs. Summary of the Invention

[0005] The purpose of the present invention is to address the deficiencies of the existing technology and provide non-antibiotic small molecule inhibitors of Clostridium difficile toxin B and a screening method thereof. The present invention obtains three high-potential compounds through this method, and verifies the effectiveness of the compounds in inhibiting TcdB toxicity through experiments. The present invention can effectively avoid the risks of intestinal flora destruction caused by antibiotic treatment, and provides a new solution for the clinical treatment of Clostridium difficile infection.

[0006] To achieve the above object, the technical solution adopted by the present invention is: Non-antibiotic small molecule inhibitors of Clostridium difficile toxin B, which are one or more of the compounds Dirlotapide (DB11399), Golvatinib (DB11977), and Laniquidar (DB12799).

[0007] The use of desloratapide in the preparation of drugs for reducing the toxicity of Clostridium difficile toxin B.

[0008] Use of desloratapide in the preparation of drugs for treating Clostridium difficile infection.

[0009] Use of govatinib in the preparation of drugs for reducing the toxicity of Clostridium difficile toxin B.

[0010] Use of govatinib in the preparation of drugs for treating Clostridium difficile infection.

[0011] Application of lanquinda in the preparation of drugs for reducing the toxicity of Clostridium difficile toxin B.

[0012] Application of Laniquida in the preparation of drugs for treating Clostridium difficile infection.

[0013] In addition, the present invention also provides a three-layer filter-type virtual screening method for TcdB inhibitors, comprising: Structural modeling of each TcdB isoform was performed, and the conformation of the receptor binding interface (RBI) was optimized using cryo-electron microscopy data. Perform the first level of filtering: perform molecular docking of the candidate compound with RBI to generate binding energy data, and then process the binding energy data to obtain candidate compounds; Perform the second level of filtering: Consider the toxicity prediction and synthetic feasibility analysis to filter out the candidate compounds that do not meet the preset toxicity and synthetic difficulty requirements, sort the remaining compounds by consensus score, and select the top M compounds; Perform the third level of filtering: Use LeDock and MM / GBSA to optimize the binding conformation, perform more accurate molecular docking and molecular mechanics generalized Born surface area (MM / GBSA) total free energy calculations, screen out the final candidate compounds, and conduct subsequent experimental verification.

[0014] In the above technical solution, further, in the first layer of filtering, the processing of the binding energy data specifically includes: first, inter-group normalization of the binding energy data, and Z-Score normalization of each RBI to reduce the batch effect caused by the intrinsic structural heterogeneity of RBI; then, intra-group probability mapping is performed, and the Z-Score results are normalized by the Sigmoid function, limiting them to the range of [0, 1], thereby achieving comparability across targets; then, pathogenicity weighted summation is performed, and the weights of each subtype of TcdB are calculated according to the pathogenic severity, epidemiological prevalence data and manual subjective importance evaluation of each subtype, thereby achieving weighted summation of the normalized results; finally, consensus scoring is performed, and the weighted summation result is Min-Max normalized, and the data is scaled to [0.01, 0.99] to generate a unified binding energy index, and the compounds are ranked according to this consensus score, and the top N candidate compounds are selected.

[0015] The beneficial effects of the present invention are: Through virtual screening and multi-stage computational strategies, the present invention discovered a broad-spectrum small molecule inhibitor that simultaneously targets each TcdB receptor binding interface, and verified its inhibitory effect in combination with cell experiments, providing lead compounds and theoretical support for the development of non-antibiotic anti-CDI drugs. The screening strategy and screening results can also provide a theoretical basis and experimental foundation for the development of the next generation of precision anti-CDI therapies. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The three-dimensional structures of eight TcdB isoforms (TcdB1-8) constructed using AlphaFold3; Figure 2 This is a diagram showing the results of a Clostridium difficile toxin cell rounding experiment in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The technical solution of the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] The development of existing TcdB small molecule inhibitors has significant deficiencies in terms of broad spectrum (covering multiple TcdB subtypes), toxicity control, and pharmacokinetic properties. Furthermore, systematic virtual screening studies targeting the receptor binding interface (RBI) of different TcdB subtypes are still relatively scarce, and traditional experimental screening has drawbacks such as high cost and long cycle time. This invention integrates deep learning network structure modeling AlphaFold3 and a multi-stage chemical calculation strategy, including Monte Carlo algorithm-driven virtual screening, molecular dynamics simulation stability verification, ADMET toxicity prediction, and SA synthesis feasibility prediction. This method efficiently discovered small molecule lead compounds with high affinity, low toxicity, easy synthesis, and the ability to effectively neutralize multiple TcdB subtypes. Cell experiments have also preliminarily verified their effectiveness in inhibiting the toxicity of Clostridium difficile toxin B. The present invention can not only provide a drug-like small molecule lead compound with high affinity, low toxicity and clinical application value for the clinical treatment of Clostridium difficile infection, which can circumvent the risk of intestinal flora destruction caused by antibiotic treatment, but also provide precise and inexpensive small molecule inhibitors for further research on the molecular mechanism of Clostridium difficile toxins. The screening strategy and screening results can also provide a theoretical basis and experimental foundation for the development of the next generation of precise anti-CDI therapies.

[0019] Identify candidate molecules According to the experimental design, 10,027 FDA-approved small molecule drug compounds were systematically screened and numerically predicted, and the top three lead compounds were obtained through three layers of filters for subsequent molecular dynamics simulation and subsequent experiments. Specifically: First, the SDF structure files of 10,027 candidate lead compounds were obtained from the entire DrugBank database. Open Babel was used to convert them into PDBQT, MOL2 and PDB formats used by various molecular docking and numerical prediction programs while retaining the original three-dimensional molecular conformation.

[0020] Based on the protein sequences of different subtypes of TcdB sequencing results in the NCBI database, the three-dimensional structures of eight common clinical TcdB subtypes (TcdB1-8) were constructed using AlphaFold3. The accuracy of the prediction results was evaluated using pLDDT (predicted Local Distance Difference Test) and pTM value (e.g. Figure 1), the pLDDT values in the TcdB1-8 structure are color-coded into four confidence levels: very high (pLDDT > 90, blue), high (70 < pLDDT ≤ 90, cyan), low (50 < pLDDT ≤ 70, yellow), and very low (pLDDT ≤ 50, orange). At the same time, the pTM values are marked next to the subtype names, with an average value of 0.763. It can be seen that the predicted results are relatively accurate. The root mean square deviation between the predicted TcdB4 structure and the TcdB4 structure measured by cryo-electron microscopy (PDB ID: 7V1N) is 2.71 nm.

[0021] The first layer of filters uses AutoDock Vina based on the fast gradient optimization concept search algorithm and the semi-empirical free energy evaluation function to calculate the binding energy between each RBI and small molecule compounds. Specifically: The molecular docking system was used to perform molecular docking on 16 RBIs of 8 TcdB subtypes constructed by AlphaFold3. Nine optimal conformations were retained for each docking, resulting in a total of 1,443,888 binding energy data (ΔG ADV ). Subsequently, the binding energy data was processed. Since the importance of different subtypes of TcdB in clinical activities is different, and the differences in the structures of RBIs themselves will affect the results of binding energy, a computational framework for multi-target binding energy data was designed to eliminate these influencing factors. First, the binding energy data was normalized between groups. The Z-Score normalization was performed on each RBI to reduce the batch effect caused by the intrinsic structural heterogeneity of RBIs: [[ID=IO]]

[0022] [[ID=I2]]Among them, X i,j represents i the original binding energy of compound j against the σ j represents j the mean and standard deviation of group

[0023] Then, the within-group probability mapping was performed. The results of Z-Score were normalized through the Sigmoid function and limited within the range of [0, 1] to achieve cross-target comparability:

[0024] Subsequently, the pathogenicity weighted summation was performed, and according to the pathogenic severity, epidemiological prevalence data, and subjective importance evaluation of the TcdB1-8 subtypes, the weights of each subtype were calculated to achieve the weighted summation of the normalized results ( W i It should be noted that there are some possible typos in the original text such as "AlphaFold3" which might be "AlphaFold 3" and some inconsistent line breaks and indentations which are maintained as in the original for translation. Also, there are some tags like " ADV " etc. which are kept unchanged as per the requirements. And the text might need further review for more accurate biological and technical understanding in a real - world context., W_ΔG ADV ), evaluate each candidate drug from the perspective of clinical application:

[0025] in w j is the weight of each target.

[0026] Finally, consensus scoring was performed, and the weighted summation results were Min-Max normalized to scale the data to [0.01, 0.99] to generate a unified binding energy index ( NW i , NW_ΔG ADV ) :

[0027] The compounds were ranked according to this consensus score, and the top 100 small molecules were selected to enter the second filter layer.

[0028] The second filter layer uses the small molecule toxicity endpoint prediction program VenomPred 2.0 and the small molecule synthesis feasibility evaluation program SA_Score, which are integrated system algorithms based on random forest, support vector machine, k-nearest neighbor and multi-layer perceptron, to quantitatively analyze and filter the carcinogenicity, mutagenicity, oral toxicity, hepatotoxicity and synthesis feasibility of drug-like small molecule compounds. According to a specific embodiment of the present invention, it can be as follows: The NW_ΔG was analyzed using VenomPred 2.0. ADV The top 100 candidate compounds were subjected to toxicity prediction, and mutagenicity, carcinogenicity, hepatotoxicity, and acute oral toxicity were quantified as values ​​between (0, 1). 0.5 was selected as the threshold to exclude compounds with a single toxicity prediction value greater than 0.6 and an average toxicity prediction value greater than 0.5. The SAscore in RDKit was used to predict the synthetic feasibility score of the top 100 compounds. SAscore is a floating point number between (0.1, 1). The larger the value, the more difficult it is to synthesize. 0.5 was selected as the threshold to retain compounds with SAscore less than 0.5. NW_ΔG was selected after screening. ADV The top 10 molecules with the highest scores are considered as the top 10 molecules and enter the third filter layer.

[0029] The third filter uses LeDock and Schrödinger's MM / GBSA modules, based on simulated annealing and evolutionary optimization algorithms, for more accurate molecular docking and molecular mechanics generalized Born surface area (MM / GBSA) total free energy calculations. The top three lead compounds undergo long-term molecular dynamics simulations and comprehensive ADMET (absorption, distribution, metabolism, excretion, and toxicity) analysis. Finally, the best lead compounds are selected for systematic experiments.

[0030] Through the above method, three high-potential compounds were finally screened out, and their numbers were DB11399, DB11977, and DB12799, corresponding to the compounds Dirlotapide, Golvatinib, and Laniquidar, respectively.

[0031] Cell experiments Clostridium difficile toxin cell rounding assay: Caco-2 human colorectal adenocarcinoma cells were selected for the experiment because they are highly sensitive to TcdB due to the presence of receptors such as FZDs, TFPI and CSPG4, and are widely used as intestinal epithelial injury models.

[0032] DMEM high glucose medium (Gibco) was used, supplemented with 10% fetal bovine serum (FBS, Gibco) and 1% penicillin-streptomycin double antibody solution (Beyotime). The cells were cultured in a constant temperature incubator at 37°C and 5% CO2. When the cells were passaged to 80%, they were digested with 0.25% trypsin-EDTA (Gibco) and divided into bottles at a ratio of 1:3. Recombinant TcdB toxin. TcdB was pre-incubated with PBS buffer containing 2 mM DTT for 30 minutes at 37°C to activate its toxic domain. According to preliminary experiments, the optimal toxicity condition was 10 ng / mL TcdB treatment for 24 hours, inducing 80% cell death. The candidate compound was dissolved in DMSO with a stock concentration of 10 mM. The compound was aliquoted and stored at -80°C. Five concentration gradients of 0.1 μM, 1 μM, 10 μM, 50 μM, and 100 μM were set, with 3 replicates for each concentration. Caco-2 cells were seeded in 96-well plates, 1×10 4 Cells were placed at 400 nm / well. After 24 hours of attachment, different concentrations of inhibitors were added for a 2-hour preincubation period before the experiment was performed. The experiment was divided into four groups: negative control (NS), positive control (TcdB), 10 μM dillotapide group (TcdB + 10 μM dillotapide), and 20 μM dillotapide group (TcdB + 20 μM dillotapide). Cells were incubated overnight at 37°C and observed under a microscope.

[0033] C. difficile toxin B can destroy the cytoskeleton and make cells round. The C. difficile toxin cell rounding test is to evaluate the cytotoxicity of C. difficile toxin by observing the ability of the toxin to make cells round; the results are as follows Figure 2 As shown, it can be seen that the toxicity of Clostridium difficile toxin B to cells can be effectively inhibited by adding desloratapide.

[0034] The embodiments described above are merely some preferred embodiments of the present invention and are not intended to limit the present invention. Persons skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, any technical solution obtained by equivalent substitution or equivalent transformation falls within the scope of protection of the present invention.

Claims

1. Application of desloratapide in the preparation of drugs for reducing the toxicity of Clostridium difficile toxin B.

2. Application of desloratadine in the preparation of drugs for treating Clostridium difficile infection.

3. A three-layer filter-type virtual screening method for TcdB inhibitors, characterized in that: include: Structural modeling of each TcdB isoform was performed, and the conformation of the receptor binding interface (RBI) was optimized using cryo-electron microscopy data. Perform the first level of filtering: perform molecular docking of the candidate compound with RBI to generate binding energy data, and then process the binding energy data to obtain candidate compounds; Perform the second level of filtering: Consider the toxicity prediction and synthetic feasibility analysis to filter out the candidate compounds that do not meet the preset toxicity and synthetic difficulty requirements, sort the remaining compounds by consensus score, and select the top M compounds; Perform the third level of filtering: Use LeDock and MM / GBSA to optimize the binding conformation, perform more accurate molecular docking and molecular mechanics generalized Born surface area (MM / GBSA) total free energy calculations, screen out the final candidate compounds, and conduct subsequent experimental verification.

4. The three-layer filter type virtual screening method according to claim 3, characterized in that: In the first layer of filtering, the binding energy data are processed specifically including: first, inter-group normalization of the binding energy data and Z-Score normalization of each RBI to reduce the batch effect caused by the intrinsic structural heterogeneity of RBI; then, intra-group probability mapping is performed, and the Z-Score results are normalized by the Sigmoid function, limiting them to the range of [0,1] to achieve comparability across targets; then, pathogenicity weighted summation is performed, and the weights of each TcdB subtype are calculated based on the pathogenic severity of each TcdB subtype, epidemiological prevalence data and manual subjective importance evaluation, to achieve weighted summation of the normalized results; finally, consensus scoring is performed, and the weighted summation result is Min-Max normalized, and the data is scaled to [0.01,0.99] to generate a unified binding energy index, and the compounds are ranked according to this consensus score, and the top N candidate compounds are selected.

Images