A method for preparing a small molecule compound for inhibiting the binding of a novel coronavirus to an ACE2 receptor

Abstract

The present application relates to the technical field of biological medicine, and particularly relates to a method for preparing a new coronavirus and ACE2 receptor binding inhibitor drug by using a small molecule compound. Specifically, by using a virtual screening method in computer-aided drug design, including applying an artificial intelligence model, a molecular docking test and a molecular dynamics simulation experiment, 5 candidate inhibitors with good binding affinity and good selectivity to an ACE2 binding site are screened from a public database of small molecule compounds, and pharmacokinetics and toxicology properties of the candidate inhibitors are evaluated, so that a new use of the small molecule compounds for research and development of a new coronavirus infection drug is disclosed.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of biological medicine technology, and particularly relates to a kind of small molecule inhibitor with human angiotensin converting enzyme 2 (ACE2) receptor as action target point. Specifically, the inhibitor is a compound recorded in public database in experimental stage, and the five compounds are found to have new uses of hindering the combination of ACE2 and coronavirus spike glycoprotein by the virtual screening method in computer-aided drug design. Molecular docking test and molecular dynamics simulation experiment verify that it has good binding affinity and binding site selectivity with the above-mentioned action target point, and has great drug potential. BACKGROUND

[0002] The spike glycoprotein of coronavirus invades host cells by combining with ACE2 receptor in human body. Unlike vaccines and macromolecular drugs, small molecule drug development cycle is long, cost is high, and success rate is low, and how to accelerate the research and development process of lead drug is a problem to be solved at present. Scientists have made a lot of efforts to develop small molecule inhibitors hindering the combination of ACE2 and spike glycoprotein, and computer-aided drug design is an efficient, fast and accurate method, in which virtual screening is an important way to quickly explore the potential of new uses of old drugs in the early stage of disease outbreak. In human body, ACE2 is an important regulator in cardiovascular system, and plays a positive role in anti-ventricular remodeling, hypertension, heart failure and other aspects, and is widely distributed in human heart and kidney. Therefore, due to the particularity of ACE2 in function catalysis and as a coronavirus binding receptor, the influence of the drug on the normal physiological function of ACE2 is not considered in the existing drug discovery process targeting ACE2, thereby leading to potential side effects. The present application aims to screen a batch of compounds with good drug properties and small side effects from known database by applying artificial intelligence and traditional computer-aided drug design methods to inhibit the combination of ACE2 and spike glycoprotein, and then used for subsequent experimental determination or clinical application. SUMMARY

[0003] The present application uses artificial intelligence and traditional computer-aided drug design methods to screen a kind of small molecule inhibitors capable of hindering the combination of ACE2 and coronavirus from public database, and endows them with new uses.

[0004] The technical scheme of the present application:

[0005] A kind of small molecule compound in the preparation of coronavirus and ACE2 receptor combination inhibitor drug method, the SMILES and database number of the small molecule compound are as follows:

[0006] (1)[H][C@](CC(=O)NO)(CC1=CC(O)=CC=C1)C(=O)N[C@@]1([H])C2=C(C[C@@]1([H])O)C=CC=C2

[0007] DrugBank ID:DB06837

[0008] (2)OCCN(CC(=O)NO)S(=O)(=O)C1=CC=C(C=C1)C1=CC=CC=C1

[0009] DrugBank ID:DB08029

[0010] (3)CC1(OCC(CO1)OC(=O)NCCNC(=O)OC1COC(C)(OC1)C(O)=O)C(O)=O

[0011] DrugBank ID:DB07579

[0012] (4)[H][C@]12SCC(COC(C)=O)=C(N1C(=O)[C@H]2NC(=O)CSC1=CC=NC=C1)C(O)=O

[0013] DrugBank ID:DB01139

[0014] (5)OC(=O)C1=CC2=C(C=C1)C=C(C=C2)C(F)(F)P(O)(O)=O

[0015] DrugBank ID:DB08397

[0016] Furthermore, the five inhibitors obtained were screened using the following technical method:

[0017] Step 1: Use the Lipinski five-rule screening method to screen small molecule compounds in the DrugBank database to identify drug-like compounds.

[0018] Step 2: Construct a deep learning model to preliminarily determine the binding potential of the drug-like compound obtained in Step 1 to the binding interface of ACE2 and spike glycoprotein.

[0019] Step 3: Using molecular docking methods in computer-aided drug design, calculate the binding affinity of compounds with high binding probability after pre-screening by the deep learning model to the binding interface of ACE2 and spike glycoprotein.

[0020] Step 4: Compare the selectivity of compounds for different binding sites of ACE2 by using the deep learning model in Step 2, the binding site scoring model, and the two molecular docking affinity scoring methods in Step 3.

[0021] Step 5: Using molecular dynamics simulations, examine the dynamic binding stability of compounds with good binding site selectivity from Step 4, analyze their binding mechanisms, evaluate their binding strength, and determine the final candidate inhibitors.

[0022] Step 6: Evaluate the pharmacokinetic (ADME) and toxicological (T) properties (ADMET) of the candidate inhibitors identified in Step 5 to understand their potential effects and risks to humans early in drug discovery.

[0023] Furthermore, step 1 specifically includes:

[0024] Step 1.1: Obtain the 3D structure files of all small molecule compounds from the DrugBank v5.1.8 database.

[0025] Step 1.2: Use the Lipinski five rules to screen for drug-like compounds, namely, relative molecular mass ≤ 500, number of hydrogen bond donors ≤ 5, number of hydrogen bond acceptors ≤ 10, octanol-water partition coefficient -2 ≤ LogP ≤ 5, and number of rotatable bonds in the structure ≤ 10.

[0026] Furthermore, step 2 specifically includes:

[0027] Step 2.1: Construct a convolutional neural network model using artificial intelligence methods. This involves building feature matrices for each atom in the small molecule compound and adjacency matrices between atoms, as well as feature matrices for each amino acid in the protein binding site and adjacency matrices between amino acids. These matrices are used as input to the deep learning model, which outputs the probability of a compound binding to a specific protein binding site. The model is trained using Python code on the DUD-E compound binding affinity decoy database, iterating multiple times until the accuracy requirements are met.

[0028] Step 2.2: Using the constructed deep learning model, quickly screen the large number of drug-like compounds obtained in Step 1.2 to select compounds with binding potential.

[0029] Furthermore, step 3 specifically includes:

[0030] Step 3.1: Using Autodock Vina molecular docking software, dock the compounds with binding potential from Step 2.2 to the binding interface between ACE2 and the spike glycoprotein. First, preprocess the ligand compounds and the protein receptor. Then, set the docking parameters as follows: the docking region is a pocket-shaped region within 25 Å around the geometric center of the binding interface; the exhaustive progress of the docking algorithm is set to 16; each docking generates the top 9 ligand clusters, and the top-ranked conformation is considered the optimal binding state.

[0031] Step 3.2: After docking, Vina and Vinardo binding affinity scores are used to evaluate the binding of each compound to ACE2 and spike glycoprotein to determine whether the two can bind under steady state.

[0032] Step 3.3: Select compounds from Steps 2.2 and 3.2 that rank highly in both binding potential and binding stability (binding affinity) for subsequent screening steps.

[0033] Furthermore, step 4 specifically includes:

[0034] Step 4.1: The binding tendency of the compound obtained in Step 3.3 to the ACE2 and spike glycoprotein binding interfaces was evaluated using a binding site scoring model (from the literature Che XH, Chai SY, Zhang ZZ, et al. Prediction of ligand binding sites using improved blind docking method with a machine learning-based scoring function [J]. Chemical Engineering Science, 2022, 261: 10.).

[0035] Step 4.2: To compare the binding tendency of compounds to different binding sites of ACE2, the deep learning model constructed in Step 2.1, the two docking affinity scoring methods in Step 3.2, and the binding site scoring model in Step 4.1 were reused to comprehensively evaluate the binding tendency of compounds to ACE2 catalytic sites.

[0036] Step 4.3: By comparing the results of the four different scoring models in Step 4.2, select compounds from the compounds selected in Step 3.3 that are easy to bind at the ACE2 and spike glycoprotein binding interface but not easy to bind at the ACE2 catalytic site for subsequent screening steps, so as to minimize the impact of the compounds on the normal catalytic function of ACE2 during the binding process.

[0037] Furthermore, step 5 specifically includes:

[0038] Step 5.1: Long-term molecular dynamics simulations were performed on the compounds with good binding site selectivity selected in Step 4.3. First, based on the docking results in Step 3.1, these compounds with good binding site selectivity were complexed with ACE2 and spike glycoprotein respectively to form new ligand-receptor complexes. Then, their structures were pretreated, cubic periodic boundary conditions were added to the complexes, an SPC explicit water model was added, an appropriate amount of Na ions was added to balance the negative charge of the system, and an OPLS4 force field was selected.

[0039] Step 5.2: Perform molecular dynamics simulations. First, the constructed simulation system was subjected to energy relaxation, followed by a 140 ns molecular dynamics simulation. The first 40 ns used the NVT ensemble, and the last 100 ns used the NPT ensemble, with a time step of 2 fs. One frame of the system architecture was recorded every 100 ps, ​​and the simulation data from the last 1000 frames were used for subsequent result analysis. The simulation temperature was set to 300 K, and the pressure to standard atmospheres. The isothermal method used was the Nose-Hoover Chain method, and the isobaric method used was the Martyna-Tobias-Klein method.

[0040] Step 5.3: Analyze the quality of the simulation results to examine whether the simulation meets the expected settings. Analyze the dynamic changes of ligands and receptors during the simulation to examine the binding stability of the compounds and possible inhibitory mechanisms.

[0041] Step 5.4: Calculate the binding free energy of the ligand and acceptor during the dynamic simulation using the MM-GBSA method to examine the binding strength of the compound.

[0042] Step 5.5: Finally, the compounds selected in Step 4.3 are recombined with the docking conformation of the ACE2 catalytic site and the ACE2 receptor, using the exact same parameters and steps as in Steps 5.1-5.4, to further compare the binding tendency of these compounds to different binding sites of ACE2, thereby verifying the results of our selective screening of binding sites.

[0043] Step 5.6: Determine the final candidate small molecule inhibitors based on the molecular dynamics simulation results.

[0044] Furthermore, step 6 specifically includes:

[0045] Step 6.1: The properties of the candidate inhibitors identified in Step 5.6 in the human body were evaluated using the admetSAR2.0 online website, including their absorption, distribution, metabolism, excretion, and toxicity.

[0046] Step 6.2: Based on the method in Step 6.1, select the most important evaluation indicators among the five properties, compare the advantages and disadvantages of each inhibitor, and give suggestions for future application in the development of new drugs for COVID-19.

[0047] Compared with the prior art, the present invention has the following main innovations and beneficial effects:

[0048] (1) A class of small molecule inhibitors that can block the binding of ACE2 to the spike glycoprotein of SARS-CoV-2 were quickly and accurately screened from public databases using a virtual screening method. Their binding probability, binding stability, inhibition mechanism, binding strength, pharmacokinetic and toxicological properties were also given, which has guiding significance for new drug development.

[0049] (2) The selectivity of small molecule compounds to different binding sites during the discovery of new drugs targeting ACE2 was investigated, and the side effects were minimized as much as possible in the early stages of drug discovery. Attached Figure Description

[0050] Figure 1 This is the overall virtual screening process of the present invention and the structural diagram of the screened small molecule inhibitors;

[0051] Figure 2 For deep learning model architectures used for pre-screening;

[0052] Figure 3 This diagram shows the binding pattern of one of these compounds (DrugBank ID: DB06837) with ACE2 and the spike glycoprotein. In the diagram, A is a 2D binding pattern of the compound with the protein receptor obtained through molecular dynamics simulations; B and C show the changes in a key hydrogen bond interaction between ACE2 and the spike glycoprotein before and after binding to the receptor protein, respectively. ACE2 is represented by a red cartoon model, the receptor-binding domain of the spike glycoprotein is represented by an orange cartoon model, and key amino acid residues are represented by rods. Bond lengths in the diagram are in Å.

[0053] Figure 4 This is a schematic diagram of the binding pattern of one of these compounds (DrugBank ID: DB06837) to the ACE2 catalytic site. In diagram A, we obtain a 2D binding pattern of the compound with the protein receptor through molecular dynamics simulations; in diagram B, we obtain a 3D binding pattern of the compound with the protein receptor through molecular dynamics simulations. ACE2 is represented by a green cartoon model, and amino acid residues 10 Å away from the Zn ion at the ACE2 catalytic center are represented by a white surface model. Detailed Implementation

[0054] The process and results of this invention are explained below with reference to the accompanying drawings and practical application methods. The results obtained in this invention are implemented under the premise of the foregoing technical solution, and detailed implementation methods and specific operating procedures are given. However, the scope of protection of this invention does not include the application of this technical solution in other examples. Although some technical details and features in the above-mentioned technical solution can be modified or equivalently replaced in the screening examples of this invention, they all fall within the spirit and scope of this technical solution for screening small molecule inhibitors with good binding site selectivity that can inhibit the binding of ACE2 to the spike glycoprotein, and should be included within the scope of protection of this application.

[0055] Based on the appendix Figure 1 A virtual screening process was used, with the DrugBank database containing 9137 small molecule compounds. First, 6876 drug-like small molecules meeting the requirements were obtained using the Lipinski five rules. Then, a constructed deep learning model (see attached) was employed. Figure 2 The model predicted the likelihood of 6876 drug-like small molecules binding to the ACE2 and spike glycoprotein interfaces, with 1735 compounds deemed potentially binding. These compounds were then docked to the ACE2 and spike glycoprotein interfaces using molecular docking. Vina and Vinardo scoring methods were used to determine their steady-state binding affinity. Among the 1735 compounds, the deep learning model predictions for the top 300 showed a significant positive correlation with their docking scores. Therefore, 128 compounds from these 300, with Vina scores better than -6.4 kcal / mol and Vinardo scores better than -4.8 kcal / mol, were selected for selective binding site screening. Eleven compounds with good binding site selectivity were selected, as shown in Table 1. In addition, for the convenience of comparison with known active compounds, the binding site selectivity of two compounds, Nilotinib and SSAA09E2, from the literature (Razizadeh M., Nikfar M., Liu Y. Small molecule therapeutics to destabilize the ace2-rbd complex: Amolecular dynamics study [J]. Biophys J, 2021, 120(14): 2793-2804.) is also shown in Table 1.

[0056] Table 1. Results of selective screening of binding sites for different compounds

[0057]

[0058] Next, the molecular dynamics simulation results of the new ligand-receptor complexes formed by these 11 compounds with ACE2 and spike glycoprotein were analyzed. Simulation quality analysis showed that the overall energy of each system remained stable during the simulation, and the temperature and pressure fluctuated around the set values, indicating that the simulation process met expectations. Then, the dynamic changes of the ligand and receptor were analyzed during the 100 ns production simulation after the molecular dynamics simulation. RMSD analysis and visualization showed that 7 of the 11 compounds could stably bind to the receptor protein without leaving their original binding sites during the simulation; therefore, these 7 compounds were used for subsequent analysis. RMSF analysis showed that the binding of each compound could specifically affect the conformational flexibility of key amino acid residues at the ACE2-spike glycoprotein binding interface, thereby interfering with the binding of ACE2 to the spike glycoprotein, but without affecting the overall stability of the protein. (Appendix) Figure 3 Detailed analysis of ligand-acceptor interactions in the system containing ligand compound DB06837 is presented. It can be seen that the binding of this compound has a significant impact on the interaction between ACE2 and the spike glycoprotein. For example, the aromatic ring in the indanate structure of the compound forms a long-term (≥30 ns) π-cation interaction with the LYS417 residue of the spike glycoprotein. This hydrophobic interaction hinders the formation of hydrogen bonds between LYS417 and ACE2's ASP30. When no ligand is bound, a hydrogen bond is formed between the amino hydrogen on the LYS417 side chain and the hydroxyl oxygen on the ASP30 side chain. However, when DB06837 is stably bound, the distance between the two residues increases to 4.7 Å, exceeding the range of a traditional hydrogen bond, thus weakening the binding interaction. Similar situations occur at the TYR505 residue of the spike glycoprotein and the GLU37 residue of ACE2, which also interact with this compound via hydrogen bonds. Furthermore, hydrogen bond interactions between ACE2 ALA387 and the compound, π-cation interactions between LYS26 and the compound, π-π stacking interactions between HIS34 and the compound, and π-π stacking interactions between the spike glycoprotein TYR505 and the compound were also observed. Blocking the interaction between ACE2 and spike glycoprotein residues through compound binding, thereby weakening protein-protein binding ability, is an important pathway for small molecule inhibitors to exert their pharmacological effects.

[0059] In addition to the qualitative analysis described above, the dynamic binding strength between the compounds and the protein receptor was quantitatively characterized by calculating the binding free energy (ΔG). The average ΔG values ​​of different compounds with the receptor during the simulation are listed in Table 2. As can be seen from the table, the average ΔG values ​​of DB07579 and DB08029 are comparable to those of the two active compounds. This demonstrates the effectiveness of our screening method in the discovery of small molecule inhibitors; that is, the screened compounds not only bind stably to the ACE2 and spike glycoprotein binding interface but also exhibit high binding strength. Furthermore, we observed that although the ΔG values ​​of compounds DB08409, DB03313, DB01139, and DB08397 were slightly lower than those of the active compounds, their ΔG fluctuations were smaller during the simulation. This suggests that they have better binding stability than other compounds and possess the potential for further structural modification and optimization.

[0060] Table 2. Average binding free energy at the interface between different compounds and ACE2-spike glycoprotein

[0061]

[0062] Finally, the selectivity of these compounds for two different ACE2 binding sites was compared, and the results were divided into four categories based on their binding to the ACE2 catalytic site:

[0063] (1) Compounds DB06837, DB08029 and DB01139 were found to bind in the narrow “crack” where the ACE2 catalytic site is located, but they did not have a significant effect on the catalytic site centered on Zn ions, and their binding free energy with the catalytic site was worse than their binding interface with ACE2-RBD.

[0064] In order to attach Figure 3 The interactions between compounds are compared, and the binding of compound DB06837 to the ACE2 catalytic site is shown in the appendix. Figure 4 .

[0065] (2) Compounds DB07579 and DB08397 were found to be unable to bind stably near the ACE2 catalytic site.

[0066] (3) Compounds DB08409 and DB03313 were found to form stable coordination with zinc ions in the ACE2 catalytic site during the simulation process, which would inevitably interfere with the binding of the substrate to the ACE2 catalytic site.

[0067] (4) The two active compounds were found to bind in the narrow “crack” where the ACE2 catalytic site is located. Although they did not have a significant effect on the catalytic site centered on Zn ions, their binding free energy with the catalytic site was better than that with the ACE2-spike glycoprotein binding interface. This may lead to a weaker binding tendency with the ACE2-spike glycoprotein binding interface.

[0068] In summary, the first two classes of compounds are considered the most promising small molecule inhibitors in this study. Compounds DB07579 and DB08397 do not bind near the catalytic site of ACE2. DB06837, DB08029, and DB01139 have minimal impact on ACE2 catalytic function, while simultaneously forming stable interactions with the ACE2-RBD binding interface, maintaining high binding strength. Their binding site selectivity is also superior to the two active compounds. Therefore, these five compounds were ultimately identified and their ADMET properties were evaluated. The results are shown in Table 3, with the best-performing inhibitor for each property highlighted in bold. Properties with units in the table represent predictions from the regression model in admetSAR2.0, while those without units represent predictions from the classification model. In these cases, the numerical value indicates the probability of prediction, and the positive or negative sign indicates whether the property is present.

[0069] Table 3. ADMET properties of candidate inhibitors and known active compounds

[0070]

[0071] In summary, a new application of a class of small molecule compounds as inhibitors of ACE2 receptor binding to SARS-CoV-2 has been identified. These compounds also exhibit good ACE2 binding site selectivity, and their ADMET properties have been evaluated. This invention has significant reference value for the development of new drugs for SARS-CoV-2.

Claims

1. A method for preparing a class of small molecule compounds as inhibitors of SARS-CoV-2 binding to the ACE2 receptor, characterized in that, Small molecule compounds were obtained through screening using the following steps: Step 1: Use the Lipinski five-rule screening method to screen small molecule compounds in the DrugBank database to identify drug-like compounds; Step 2: Construct a deep learning model to preliminarily determine the binding potential of the drug-like compound obtained in Step 1 to the binding interface of ACE2 and spike glycoprotein; Step 3: Using molecular docking methods in computer-aided drug design, calculate the binding affinity of compounds with high binding probability after pre-screening by the deep learning model to the binding interface of ACE2 and spike glycoprotein. Step 4: Compare the selectivity of the compound for different binding sites of ACE2 by using the deep learning model in Step 2, the binding site scoring model, and the two molecular docking binding affinity scoring methods in Step 3. Step 5: Using molecular dynamics simulations, examine the dynamic binding stability of compounds with good binding site selectivity from Step 4, analyze their binding mechanisms, evaluate their binding strength, and determine the final candidate inhibitors. Step 3 specifically includes: Step 3.1: Using Autodock Vina molecular docking software, dock the compounds with binding potential from Step 2.2 to the binding interface between ACE2 and spike glycoprotein. First, preprocess the ligand compounds and protein receptors, and then set the docking parameters as follows: the docking region is a pocket-shaped region of 25 Å around the geometric center of the binding interface, the exhaustive progress of the docking algorithm is set to 16, and each docking generates the top 9 ligand clusters, with the first-ranked conformation being considered the best binding state. Step 3.2: After docking, Vina and Vinardo binding affinity scores were used to evaluate the binding of each compound to ACE2 and spike glycoprotein to determine whether the two can bind under steady state. Step 3.3: Select compounds from Steps 2.2 and 3.2 that rank highly in both binding potential and binding stability for subsequent screening steps; Step 4 specifically includes: Step 4.1: Use a binding site scoring model to evaluate the binding tendency of the compound obtained in step 3.3 to the binding interface of ACE2 and spike glycoprotein; Step 4.2: Reuse the deep learning model built in Step 2.1, the two docking affinity scoring methods in Step 3.2, and the binding site scoring model in Step 4.1 to comprehensively evaluate the compound's binding tendency to the ACE2 catalytic site; Step 4.3: By comparing the results of the four different scoring models in Step 4.2, select compounds from the compounds selected in Step 3.3 that are easy to bind at the ACE2 and spike glycoprotein binding interface but not easy to bind at the ACE2 catalytic site for subsequent screening steps.

2. The method for preparing a drug that inhibits the binding of SARS-CoV-2 to the ACE2 receptor using a class of small molecule compounds according to claim 1, characterized in that, Step 1 specifically includes: Step 1.1: Obtain the 3D structure files of all small molecule compounds from the DrugBank v5.1.8 database; Step 1.2: Use the Lipinski five rules to screen for drug-like compounds, namely, relative molecular mass ≤ 500, number of hydrogen bond donors ≤ 5, number of hydrogen bond acceptors ≤ 10, octanol-water partition coefficient -2 ≤ LogP ≤ 5, and number of rotatable bonds in the structure ≤ 10.

3. The method for preparing a drug that inhibits the binding of SARS-CoV-2 to the ACE2 receptor using a class of small molecule compounds according to claim 2, characterized in that, Step 2 specifically includes: Step 2.1: Construct a convolutional neural network model using artificial intelligence methods. Construct the feature matrix of each atom in the small molecule compound and the adjacency matrix between atoms, as well as the feature matrix of each amino acid in the protein binding site and the adjacency matrix between amino acids. Use each matrix as input to the deep learning model. The model outputs the probability of a compound binding to a certain protein binding site. Train the model on the DUD-E compound binding affinity decoy database using Python code. Iterate multiple times until the accuracy requirements are met. Step 2.2: Using the constructed deep learning model, quickly screen the large number of drug-like compounds obtained in Step 1.2 to select compounds with binding potential.

4. The method for preparing a drug that inhibits the binding of SARS-CoV-2 to the ACE2 receptor using a class of small molecule compounds according to claim 3, characterized in that, Step 5 specifically includes: Step 5.1: Perform long-term molecular dynamics simulations on the compounds with good binding site selectivity selected in Step 4.

3. First, based on the docking results in Step 3.1, these compounds with good binding site selectivity are combined with ACE2 and spike glycoprotein to form new ligand-receptor complexes. Then, their structures are pretreated. Next, cubic periodic boundary conditions are added to the complexes, an SPC explicit water model is added, an appropriate amount of Na ions is added to balance the negative charge of the system, and an OPLS4 force field is selected. Step 5.2: Perform molecular dynamics simulation calculations. First, the constructed simulation system was subjected to energy relaxation, and then a 140 ns molecular dynamics simulation was performed. The first 40 ns used the NVT ensemble, and the last 100 ns used the NPT ensemble. The time step was 2 fs, and one frame of the system architecture was recorded every 100 ps. The simulation data of the last 1000 frames were used for subsequent result analysis. The simulation temperature was set to 300 K, and the pressure was standard atmospheric pressure. The isothermal method used was the Nose-Hoover Chain method, and the isobaric method used was the Martyna-Tobias-Klein method. Step 5.3: Analyze the quality of the simulation results, examine whether the simulation meets the expected settings, and analyze the dynamic changes of ligands and receptors during the simulation process to examine the binding stability of the compound and possible inhibitory mechanisms; Step 5.4: Calculate the binding free energy of the ligand and acceptor during the dynamic simulation process using the MM-GBSA method to examine the binding strength of the compound; Step 5.5: Finally, the compounds selected in Step 4.3 are recombined with the docking conformation of the ACE2 catalytic site and the ACE2 receptor, using the exact same parameters and steps as in Steps 5.1-5.4, to further compare the binding tendency of these compounds to different binding sites of ACE2, thereby verifying the results of our selective screening of binding sites. Step 5.6: Determine the final candidate small molecule inhibitors based on the molecular dynamics simulation results.

5. The method for preparing a drug that inhibits the binding of SARS-CoV-2 to the ACE2 receptor using a small molecule compound according to any one of claims 1-4, characterized in that, The small molecule compound described above exhibits good selectivity for two different binding sites of ACE2.

6. The method for preparing a drug that inhibits the binding of SARS-CoV-2 to the ACE2 receptor using a small molecule compound according to any one of claims 1-4, characterized in that, SARS-CoV-2 and ACE2 receptor binding inhibitors are used to treat diseases related to SARS-CoV-2 infection.

7. The method for preparing a drug that inhibits the binding of SARS-CoV-2 to the ACE2 receptor using a small molecule compound according to any one of claims 1-4, characterized in that, The SMILES and database ID of the small molecule compounds are as follows: (1)[H][C@](CC(=O)NO)(CC1=CC(O)=CC=C1)C(=O)N[C@@]1([H])C2=C(C[C@@]1([H])O)C=CC=C2 DrugBank ID:DB06837 (2)OCCN(CC(=O)NO)S(=O)(=O)C1=CC=C(C=C1)C1=CC=CC=C1 DrugBank ID:DB08029 (3)CC1(OCC(CO1)OC(=O)NCCNC(=O)OC1COC(C)(OC1)C(O)=O)C(O)=O DrugBank ID:DB07579 (4)[H][C@]12SCC(COC(C)=O)=C(N1C(=O)[C@H]2NC(=O)CSC1=CC=NC=C1)C(O)=O DrugBank ID:DB01139 (5)OC(=O)C1=CC2=C(C=C1)C=C(C=C2)C(F)(F)P(O)(O)=O DrugBank ID:DB08397。

Images