Screening method and application of CDK12 agonist

Abstract

The invention discloses a screening method and application of a CDK12 agonist, and belongs to the field of medicine and health. The screening method comprises the following steps: acquiring a CDK12 / CyclinK compound model from a database, and optimizing the CDK12 / CyclinK compound model as a target protein for molecular docking; two binding pockets of the CDK12 are selected; screening drug-like compounds from the multi-source compound library, docking the drug-like compounds with two binding pocket networks, and screening the compounds according to pocket binding energy; calculating a valley similarity coefficient between the compounds, carrying out clustering analysis, and respectively screening out clustering compounds for the two binding pockets; identifying a RugBank source drug from the clustering compound, carrying out molecular docking and ADMET property evaluation on the RugBank source drug, and screening out preferential candidate drugs from the two binding pockets respectively; and carrying out molecular dynamics simulation on the preferential candidate drugs, screening out candidate compounds according to docking scores and binding free energy aiming at two binding pockets, carrying out cell experiment verification, and screening out the CDK12 agonist.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of medicine and health, and particularly relates to a screening method and application of CDK12 agonists. BACKGROUND

[0002] CDK12 (cyclin-dependent kinases 12) belongs to the cyclin-dependent kinase family related to transcriptional regulation, is located on human chromosome 17q12, encodes 1490 amino acids, and the protein molecular weight is 164 kDa. The protein structure part mainly includes proline-rich motifs (PRM) and arginine / serine-rich motifs (RS) and a kinase domain with catalytic function. By combining with cyclin K (Cyclin K) to form a complex, the serine residues (Ser2 / Ser5) of the carboxy-terminal domain (CTD) of RNA polymerase II (RNAPII) are phosphorylated. This process directly regulates the transcriptional elongation, and is particularly important for the expression of long genes (>45 kb) and genes rich in exons (such as DNA damage repair genes).

[0003] Loss of function of CDK12 leads to transcriptional elongation defects and genomic instability. Loss of function of CDK12 reduces the RNAPII elongation rate, triggers intron polyadenylation (IPA) events, and produces truncated non-functional transcripts (such as Slc12a1, Fgf1, Cast genes); on the other hand, it leads to down-regulation of DNA damage repair (DDR) genes (such as BRCA1, ATR), causing DNA double-strand breaks. Multiple studies have confirmed that CDK12 is highly expressed in renal tubular epithelial cells (RTECs) and plays a key role in maintaining renal homeostasis. Renal tubule-specific CDK12 knockout mice exhibit polyuria, urine concentration disorders and hydronephrosis, which are related to the down-regulation of Na-K-2Cl co-transporter (NKCC2) caused by transcriptional defects of Slc12a1 gene. Loss of function of CDK12 can exacerbate cisplatin-induced acute kidney injury (AKI) by inhibiting the transcriptional elongation of Fgf1 and Cast genes, resulting in increased DNA damage, increased apoptosis and inhibited proliferation. Overexpression of CDK12 in renal tubules can significantly reduce DNA damage, inhibit apoptosis, and restore the expression of repair genes, protecting renal tubules from cisplatin-induced telomere and genomic DNA damage, and this conclusion is expected to be extrapolated to multiple organs and tissues. In the vascular endothelial system, down-regulation of CDK12 expression triggers endothelial cell lipid metabolism and apoptosis, making heart and kidney damage in hypertensive patients worse; in the muscle system, decreased expression of CDK12 hinders muscle cell differentiation and maturation, causing muscle atrophy and increasing the risk of sarcopenia in patients with chronic kidney disease. Currently, there is no specific agonist of CDK12 reported, therefore, the development of CDK12 agonists is expected to become a new strategy for the repair of multiple organs and tissues in the treatment of acute and chronic kidney injury and its CKD-related cardiovascular and sarcopenia complications, filling the existing treatment gap. SUMMARY

[0004] In view of the deficiencies of the prior art, the purpose of the present application is to provide a screening method and application of CDK12 agonists, which solves the problems in the prior art.

[0005] The purpose of the present application can be achieved by the following technical solutions: The screening method of the CDK12 agonist comprises the following steps: Obtain the CDK12 / CyclinK complex model from the database, and optimize it as the target protein for molecular docking; and select two binding pockets of CDK12; Screen out drug-like compounds from the multi-source compound library; Respectively dock the drug-like compounds with the two binding pockets, and screen the compounds according to the pocket binding energy for the two binding pockets; The inter-compound Tanimoto similarity coefficient is calculated, and clustering analysis is performed, and the clustering compounds are screened for the two binding pockets; The DrugBank-derived drugs are identified from the clustering compounds, and molecular docking and ADMET property evaluation are performed, and the priority candidate drugs are screened for the two binding pockets.

[0006] Molecular dynamics simulation is performed on the priority candidate drugs, and candidate compounds are screened for the two binding pockets according to the docking score and binding free energy; The candidate compounds are verified by cell experiments, and the CDK12 agonist is screened from the candidate compounds.

[0007] Further, the two binding pockets are: ATP binding pocket and allosteric pocket.

[0008] Further, the multi-source compound library includes: ChemDiv, COCONUT and DrugBank.

[0009] Further, the process of screening the drug-like compounds includes: multi-source compound library is optimized for protonation state by OpenBabel, Gypsum-DL generates energy minimized 3D conformation, and drug-like compounds are retained after removing PAINS and warning structures.

[0010] Further, when screening compounds, for the allosteric pocket, PLANET graph neural network model is additionally used to predict pKd value, and 5000 high-affinity compounds with pKd>7.0 are screened.

[0011] Further, the clustering analysis uses K-means clustering, and the number of clustering clusters is set to 100, and according to the clustering analysis result, the compound of the clustering cluster center is selected from each cluster.

[0012] Further, the screened CDK12 agonist is CNP0250284, and its structural formula is: .

[0013] The application of CDK12 agonist in the preparation of acute and chronic kidney injury drugs, the CDK12 agonist is CNP0250284, and its structural formula is: .

[0014] The application of CDK12 agonist in the preparation of multi-organ tissue repair drugs, the CDK12 agonist is CNP0250284, and its structural formula is: .

[0015] A drug includes a CDK12 agonist, and its structural formula is: .

[0016] The beneficial effects of the present application: 1. Based on the fusion innovation of computer-aided drug design and artificial intelligence technology, the present application successfully screens new agonist candidate compounds targeting CDK12 kinase by integrating molecular docking (AutoDockVina), deep learning model (PLANET) and molecular dynamics simulation. Among them, four high-activity compounds targeting the ATP binding pocket of CDK12 and four allosteric regulation candidate compounds targeting the allosteric pocket of CDK12 are first discovered. Molecular docking and 20ns molecular dynamics simulation confirm that they are stably combined with the allosteric site through hydrogen bonding, hydrophobic interaction and π-π stacking, and all candidate compounds meet the Lipinski rule, have no PAINS toxicity risk, and have excellent drugability. After wet experiment verification, it is found that the CNP0250284 compound shows significant CDK12 protein agonist function in multiple cells, which provides a new lead structure for agonizing CDK12 function, treating kidney injury and promoting multi-organ and multi-tissue repair, and the candidate compound has important clinical transformation value.

[0017] 2. The present application simultaneously integrates the repositioning research of DrugBank database and adopts a three-level progressive strategy to develop candidate drugs, so as to quickly lock known drugs with good binding activity and high drugability, thereby significantly reducing the development risk and shortening the research and development cycle.

[0018] 3. The present application obtains the possible binding pockets of CDK12 protein by two ways of literature research and tool calculation prediction, and analyzes and selects. CDK12 is a protein kinase, and its ATP binding pocket is one of the core parts of its function. This pocket is used for binding and catalyzing ATP molecules, so as to transfer the phosphate group to its substrate, which is the core region of CDK12 phosphorylation reaction. The allosteric site is a binding site on the protein different from the orthosteric site. Allosteric modulators can bind to these sites to cause conformational changes in the protein structure, thereby indirectly regulating the activity of the protein and more finely controlling the function of the protein. The potential allosteric site with a higher score obtained by CavityPlus is used as one of the virtual screening candidate pockets.

[0019] 4. The present application integrates ChemDiv (1,648,197 compounds), COCONUT (407,271 natural products) and DrugBank (9,469 approved drugs) databases, and simultaneously considers structural diversity, natural uniqueness and clinical transformation potential, thereby providing a rich and high-quality initial compound pool for subsequent layer-by-layer screening and improving the probability of discovering high-quality candidate drugs.

[0020] 5、The present application ensures the accuracy of the calculation by standardizing the pretreatment including the optimization of the protonation state (pH=7.4) by OpenBabel and the generation of the energy-minimized 3D conformation by Gypsum-DL, and by removing PAINS and alert structures in advance to exclude interferents and toxic molecules, thereby improving the quality and drugability of the candidate compounds from the source.

[0021] 6、The present application uses AutoDockVina to perform grid docking on the dual-pocket, and selects the top 5,000 compounds in terms of the binding energy of each pocket. In view of the characteristics of the allosteric pocket structure, such as flexibility and low conservation, and the limited prediction accuracy of the traditional molecular docking method, the PLANET graph neural network model is additionally used to more accurately evaluate the affinity between the ligand and the allosteric pocket, thereby significantly improving the success rate and reliability of screening for allosteric agonists.

[0022] 7、The present application generates Morgan fingerprints through RDKit, calculates the Kubota similarity coefficient between compounds, adopts K-means clustering, sets the number of clustering clusters to 100, and according to the clustering analysis result, selects the compounds of the clustering cluster center from each cluster to construct a subset of diverse compounds. This effectively avoids repeated screening from a large number of similar structures, while ensuring that the selected candidate compounds have maximum structural diversity, thereby improving the probability of discovering agonists with different mechanisms of action and innovative skeletons.

[0023] 8、The present application adopts a multi-stage virtual screening process to lock CNP0250284 as a CDK12 agonist, and verifies its physiological function through a cell experiment. The compound has been verified to have the function of agonizing CDK12 protein in multiple organ systems, and theoretically, the invention of the agonist can be used for all diseases states with down-regulation of CDK12 function, and is expected to become a new strategy for repairing multiple organs and tissues for treating acute and chronic kidney damage and its CKD-related cardiovascular and sarcopenia complications, thereby filling the existing treatment gap. BRIEF DESCRIPTION OF DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0025] Figure 1 is the flowchart of the agonist screening of the present application; Figure 2 is the protein structure diagram of the CDK12 / CyclinK complex adopted by the present application; Figure 3is a schematic diagram of the ATP binding pocket of the CDK12 protein of the present application; Figure 4 is a schematic diagram of the allosteric pocket of the CDK12 protein of the present application; Figure 5 is a schematic diagram of the CNP0250284 docking with the ATP binding pocket of the present application; Figure 6 is a graph of the experimental results of CNP0250284 applied to 293T cells of the present application; Figure 7 is an immunofluorescence staining graph of CNP0250284 applied to HK2 cells of the present application; Figure 8 is a graph of the experimental results of CNP0250284 applied to c2c12 cells of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0027] As shown in Figure 1 , the screening method of the CDK12 agonist comprises the following steps: S1, obtaining a CDK12 / CyclinK complex model from a database, and optimizing the model as a target protein for molecular docking; and selecting two binding pockets of CDK12; As shown in Figure 2 , obtaining a CDK12 / CyclinK complex model (PDB ID: 8P81, resolution 2.68 Å) from the RCSB protein database (PDB) as a template protein crystal structure. By PDB2PQR tool, the missing hydrogen atoms are supplemented and the charges are calculated, and the atom types are standardized by AutoDockTools, and finally converted into PDBQT format file for molecular docking.

[0028] Through research on existing related literature and tool calculation and prediction, the possible binding pockets of the receptor protein are obtained and analyzed and selected: 1) Related literature research: CDK12 is a protein kinase, and its ATP binding pocket is one of the core parts of its function. This pocket is used to bind and catalyze ATP molecules, thereby transferring phosphate groups to its substrates. This is the core region of CDK12 for phosphorylation reaction, and many small molecule inhibitor designs are aimed at the ATP binding site to block the binding of ATP to CDK12, thereby inhibiting the kinase activity. The structure of the related blocker can be seen in PDB id: 5ACB, 6B3E, 6CKX, 7NXK, 7NXJ, 8P81. This pocket is one of the virtual screening candidate pockets.

[0029] 2) Tool calculation prediction: Allosteric sites are binding sites on proteins that are different from orthosteric sites. Allosteric modulators can bind to these sites, causing conformational changes in the protein structure, thereby indirectly regulating the activity of the protein. This mode of regulation allows for more fine-tuned control of protein function. The potential allosteric site with a high score predicted by CavityPlus is the high-score allosteric site predicted by the CavityPlus online tool developed by the Peking University Kelun Group. This pocket is located on chain A of the protein, and the pocket has a large volume and a relatively deep depth. After visualizing the space accessibility by PyMOL, it is included in the virtual screening.

[0030] The ATP binding pocket of CDK12 protein is shown in Figure 3 , it can be seen that the pocket is located in the core region of the protein kinase domain, with a clear three-dimensional cavity structure. The shape and chemical properties of the pocket enable it to specifically recognize and bind ATP or analogs, and it is the location where ATP molecules bind and exert their phosphorylation function.

[0031] The allosteric pocket of CDK12 protein is shown in Figure 4 , it can be seen that the pocket is located on chain A, with a large volume and a deep depth, and has the structural basis for binding small molecules. Its spatial position is different from that of the ATP binding pocket, and it may have the potential to regulate the kinase activity of CDK12 through allosteric mechanism.

[0032] S2, screen drug-like compounds from multi-source compound libraries; First, the ChemDiv (1,648,197 compounds), COCONUT (407,271 natural products) and DrugBank (9,469 approved drugs) databases were integrated, and 1,479,657 (ChemDiv), 343,713 (COCONUT) and 8,645 (DrugBank) drug-like compounds were retained after optimization of the protonation state (pH=7.4) by OpenBabel, energy minimization of 3D conformations by Gypsum-DL, removal of PAINS and warning structures, and uniform conversion to PDBQT format. Subsequently, a two-strategy virtual screening was carried out.

[0033] S3, network docking of drug-like compounds with two binding pockets, respectively, and screening of compounds according to the binding energy of the two binding pockets; Grid docking (grid size 20x20x20A, exhaustiveness=32) was performed on the two pockets using integrated molecular docking (AutoDockVina), and the top 5,000 compounds were screened according to the binding energy of each pocket. For the allosteric pocket, the PLANET graph neural network model (input pocket 3D structure and 2D chemical structure of drug-like compounds) was additionally used to predict the pKd value, and 5,000 high-affinity compounds with pKd>7.0 were screened, which were cross-validated with the Vina results.

[0034] S4, calculation of the Kubo similarity coefficient between compounds and clustering analysis, and screening of clustered compounds for the two binding pockets, respectively; Morgan fingerprints (radius=2, length=2,048 bits) were generated by RDKit to calculate the Kubo similarity coefficient between compounds, and the formula was as follows: Where a represents the number of features present in molecule A, b represents the number of features present in molecule B, and c represents the number of features common to molecules A and B. The Scikit-Learn toolkit in Python and the calculated Kubo similarity matrix between compounds were used to perform clustering analysis on a certain number of compounds screened by molecular docking. The K-means clustering (K-Means) method was used, and the number of clusters was set to 100. According to the clustering analysis results, the compounds at the cluster centers were selected from each cluster to construct a diverse compound subset.

[0035] S5, focusing on known drugs in the DrugBank database, 38 DrugBank-derived drugs were identified from the clustered compounds, and fine molecular docking and ADME-T property evaluation were performed, and 4 priority candidate drugs were screened for each of the two binding pockets.

[0036] From the clustered compounds, 38 DrugBank-derived drugs were identified, and multi-conformational states were generated by Maestro-LigPrep. High-precision docking was performed using the Schrödinger Glide module (OPLS4 force field, ExtraPrecision mode), and docking scores were calculated. The ADME prediction platform was used to evaluate pharmacokinetic and toxicity indicators (including hERG inhibition risk, hepatotoxicity, Caco-2 permeability, and oral bioavailability). Based on the docking scores and ADME properties, four priority candidate drugs were selected for each binding pocket.

[0037] S6, The priority candidate drugs were subjected to molecular dynamics simulation, and CDK12 agonist candidate molecules for wet experimental verification were selected based on docking scores and binding free energies for the two binding pockets. Molecular dynamics simulation was performed based on GROMACS version 2023.4. In the dynamics simulation, the RESP2(0.5) charge of the small molecule compound was generated by the Multiwfn software, the topology and gro files of GROMACS were generated by the Sobtop software, and the GAFF force field was used for parameterization. The Amberff14SB force field was used to process the protein part, the hydration solvent model was TIP3P, and Na+ / Cl- ions were added to maintain the electrical neutrality of the system.

[0038] Periodic boundary conditions were applied to all three directions of the system. First, the steepest descent method was used for energy minimization, and then the system was pre-equilibrated through NVT and NPT ensembles, respectively. At a stable temperature of 300 K, the Velocity-rescale temperature coupler was used with a time constant of 0.1 ps for 100 ps of NVT equilibration. The Berendsen pressure coupler was used to maintain a pressure of 1 bar with a time constant of 2.0 ps for 100 ps of NPT equilibration. The SPME algorithm was used to handle electrostatic interactions, and a cutoff value of 12 angstroms was used for real-space interactions. The LINCS algorithm was used to constrain all hydrogen atoms involved in the bond. The time step for molecular dynamics simulation was set to 2 fs, and the conformation was recorded every 10 ps. Each complex system was subjected to 20 ns of productive simulation. From the final 2 ns trajectory, 500 frames were extracted for MM / GBSA (molecular mechanics generalized Born surface area) calculation to evaluate the binding free energy of the complex. This calculation was completed by the gmx_MMPBSA tool, and the binding free energy was represented as follows: The binding free energy of the complex can be approximated as the combination of molecular mechanics energy solvation energy and conformational entropy from the internal energy​ ), van der Waals energy (EvdW) ), and electrostatic energy (Eelec). consists of polar solvation term (Epolar) ) and non-polar solvation term (Enonpolar) estimated by generalized Born model, calculated by solvent accessible surface area.

[0039] Binding mode comparative analysis By comparing the candidate compounds with known CDK12 inhibitors (ATP competitive inhibitor SR-4835 and allosteric inhibitor BSJ-01-175), the hydrogen bond network (distance ≤ 3.5 Å, angle ≥ 120°) is analyzed by HBPLUS, the hydrophobic contact (residue distance < 5 Å) is counted by VMD Timeline plug-in, and the binding mode difference between the candidate compounds and the reference inhibitors is systematically compared by combining the MM-GBSA residue energy decomposition heat map, which provides structural biology explanation for the agonist function of the candidate compounds and points out the direction for subsequent residue mutation verification and structure optimization.

[0040] In this embodiment, 4 candidate compounds are screened from the ATP pocket and the allosteric pocket by combining the AutoDockVina docking score, the PLANET model affinity prediction result and the MM / GBSA binding free energy. These molecules have scores significantly higher than the average in deep learning affinity prediction, and the binding energy scores are around -40 kcal / mol. It is preliminarily speculated that these molecules can enhance the activity of CDK12 through conformational regulation mechanism and play the role of agonists. The specific information is as follows: 1) In the screening of the ATP binding pocket, AutoDockVina is used to screen the top 5000 compounds, and 100 representative structures are selected for molecular dynamics simulation by cluster analysis. Finally, molecules such as CNP0123054, CNP0250284, G435-0657 and Y043-6272 perform well in docking score and binding free energy score; the docking scores and binding free energy scores of these candidate compounds are shown in Table 1 as follows: Table 1 Docking scores and binding free energy scores of candidate compounds in ATP binding pocket ​​2) In other pocket screening, AutoDockVina and PLANET independently screened the top 5000 compounds ranked by score, and each performed cluster analysis, and each selected 100 representative structures for molecular dynamics simulation. Combined with docking score and binding free energy, four small molecules J030-0840, L286-0483, L759-0739, SA04-0089 were screened out with the best performance; their docking scores and binding free energy scores are shown in Table 2 below: Table 2 Docking scores and binding free energy scores of candidate compounds for other pockets Among them, the docking situation of CNP0250284 with the ATP binding pocket is as follows Figure 5 which shows the overall binding pose of the compound in the ATP binding pocket, indicating that the compound binds to the CDK12 protein at this position.

[0041] S7, wet experiment (cell experiment) is performed on the candidate compounds to verify and screen CDK12 agonists from the candidate compounds.

[0042] The experimental process is as follows: 1) Cell culture The culture conditions of human embryonic kidney cells 293T are 10% FBS, 1% P / S, DMEM, 37°C, 5% CO2. The culture conditions of human renal tubular epithelial cell line HK-2 are 10% FBS, 1% P / S, DMEM / F12, 37°C, 5% CO2. The culture conditions of mouse myoblast C2C12 are 10% FBS, 1% P / S, DMEM, 37°C, 5% CO2.

[0043] 2) Cell intervention The target cells are seeded in a six-well plate or a confocal culture dish at an appropriate density, and after the cells adhere successfully, the cells are cultured to a density of 70-80%, and the target CDK12 agonist screened out is added to the cell culture medium to intervene in the cells at a concentration of 0 / 10 / 100 / 1000 nmol / L, and the cell lysate is collected after 24 hours of culture to detect the expression of CDK12 protein and downstream factors, and to evaluate whether the drug is effective. Then, according to the previous experimental method, the lysate of cells at different time points 0 / 12 / 24 / 48 hours after drug intervention is collected to evaluate the best effective time of intervention.

[0044] 3) Western blot The cell lysate was configured to extract total protein from the cells. The protein concentration was determined by BCA method, and the protein was denatured by boiling at 75°C for 10 minutes. 40 ug of total protein was separated by gel electrophoresis, transferred to a membrane, blocked with western quick blocking solution for 30 minutes, then ATP6V0A4 primary antibody was added and incubated at 4°C overnight. After washing the membrane with TBST for 10 minutes for 3 times, the corresponding secondary antibody was added and incubated at room temperature for 1 hour. After washing the membrane, electrochemiluminescence exposure and development were carried out.

[0045] 4) Immunofluorescence staining Immunofluorescence analysis was performed on target cells. The slides were incubated with anti-CDK12 antibody (ab246887, Abeam) and secondary antibody (ab150077, Abeam). The cell nucleus was stained with DAPI (Sigma-Aldrich) according to the manufacturer's instructions. Under confocal microscopy, 30 fields were randomly assigned and the average number of CDK12 protein droplets and area of individual cells were counted in a blind manner.

[0046] In this embodiment, the CNP0250284 compound screened was subjected to in vitro biological experiment determination, and the experimental results are shown in Figure 6 to Figure 8 It was found that CNP0250284 had significant agonistic effect on CDK12 protein in multiple cells. Specifically: Figure 6 CNP0250284 was applied to 293T cells, and the left panel represents Western blot, and the right panel represents immunofluorescence staining (green for CDK12 protein; Figure 7 CNP0250284 was applied to HK2 cells, and the immunofluorescence staining (green for CDK12 protein) is shown. In 293t cells ( Figure 6 ) and HK2 cells ( Figure 7 ), the expression of CDK12 protein increased after the application of CNP0250284 and was positively correlated with the drug concentration. The expression of SER2 was the same as the change of CDK12 protein expression.

[0047] Figure 8 CNP0250284 was applied to c2c12 cells, and the effect of the change of CDK12 expression on cells was observed. The left panel of Western blot shows that the expression of CDK12 protein in CNP0250284 group cells increased. The right panel of cell observation shows that the cell morphology of CNP0250284 group is consistent with that of the control group.

[0048] In summary Figure 6- Figure 7, through the structure-based virtual screening strategy, the agonist CNP0250284 targeting CDK12 protein is found, the compound can increase the expression of CDK12 in a concentration-dependent manner, and agonize its activity, and in the future, high-activity and high-selectivity agonists targeting CDK12 protein will be further designed and synthesized for treating acute and chronic kidney injury and promoting multi-organ and multi-tissue repair In the description of the present specification, the description referring to the terms "one embodiment", "an example", "a specific example" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0049] The basic principles, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only illustrative of the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application.

Claims

1. A screening method for CDK12 agonists, characterized in that, Includes the following steps: A CDK12 / CyclinK complex model was obtained from the database, optimized, and used as a target protein for molecular docking; and two binding pockets of CDK12 were selected. Drug-like compounds were screened from a multi-source compound library; The drug-like compounds were network-connected to two binding pockets respectively, and compounds were screened based on the pocket binding energy for each of the two binding pockets. Calculate the similarity coefficient between compounds and perform cluster analysis. For each of the two binding pockets, select the clustered compounds. DrugBank-derived drugs were identified from clustered compounds, and their molecular docking and ADMET properties were evaluated. Priority candidate drugs were screened from each of the two binding pockets. Molecular dynamics simulations were performed on the priority candidate drugs, and candidate compounds were screened based on docking fraction and binding free energy for the two binding pockets. Cellular experiments were conducted to validate the candidate compounds, and CDK12 agonists were screened from the candidate compounds.

2. The screening method for CDK12 agonists according to claim 1, characterized in that, The two binding pockets are the ATP binding pocket and the allosteric pocket.

3. The screening method for CDK12 agonists according to claim 1, characterized in that, The multi-source compound libraries include: ChemDiv, COCONUT, and DrugBank.

4. The screening method for CDK12 agonists according to claim 3, characterized in that, The process of screening the drug-like compounds includes: optimizing the protonation state of a multi-source compound library using OpenBabel, generating energy-minimized 3D conformations using Gypsum-DL, and retaining drug-like compounds after removing PAINS and warning structures.

5. The screening method for CDK12 agonists according to claim 2, characterized in that, During the compound screening, for the allosteric pocket, the PLANET graph neural network model was additionally used to predict the pKd value, and 5000 high-affinity compounds with pKd>7.0 were screened.

6. The screening method for CDK12 agonists according to claim 1, characterized in that, The cluster analysis uses K-means clustering and sets the number of clusters to 100. Based on the cluster analysis results, the compound at the center of each cluster is selected.

7. The screening method for CDK12 agonists according to claim 1, characterized in that, The selected CDK12 agonist is CNP0250284, and its structural formula is as follows: 。 8. The application of CDK12 agonists in the preparation of drugs for acute and chronic kidney injury, wherein the CDK12 agonist is CNP0250284, and its structural formula is: 。 9. Application of CDK12 agonists in the preparation of drugs for multi-organ tissue repair, wherein the CDK12 agonist is CNP0250284, and its structural formula is: 。 10. A drug, characterized in that, Including CDK12 agonists, whose structural formula is: 。

Images