Discovery method of high-selectivity JAK2 activator

Abstract

The invention discloses a discovery method of a high-selectivity JAK2 activator, and belongs to the technical field of medicines, and the method comprises the following steps: S1, obtaining a JAK small molecule data set; s2, characterizing the physicochemical and structural characteristics of JAK micromolecules; s3, dividing a training set and a test set; s4, constructing a virtual screening model based on the ligand, and training and testing the model based on the training and testing set to obtain the virtual screening model based on the ligand; s5, screening out JAK small molecule data with JAK2 activity based on a virtual screening model of the ligand; s6, predicting a potential allosteric site of JAK2, and taking an allosteric pocket with the highest volume as a molecular docking binding site; s7, preparing a ligand; s8, preparing protein crystals; s9, carrying out molecular docking; s10, performing in-vivo and in-vitro biological activity tests to obtain a high-selectivity JAK2 activating agent with targeted molecular docking binding sites and a clear action mechanism; according to the invention, the JAK2 activator with a novel skeleton and good activity and subtype selectivity can be quickly and efficiently found.

Description

Technical Field

[0001] The invention belongs to the field of medical technology, and specifically relates to a method for discovering a highly selective JAK2 activator. Background Art

[0002] Cardiovascular disease refers to a general term for diseases of the heart and its blood vessels. It is currently the leading cause of death in the world, and ischemic heart disease is the main cause of death from cardiovascular disease.

[0003] Currently, the most effective strategy for treating ischemic heart disease clinically is direct percutaneous coronary intervention for reperfusion therapy, in order to minimize the time of myocardial ischemia and reduce the degree of death or damage due to myocardial ischemia and hypoxia. Although reperfusion therapy can restore blood supply to infarcted myocardial tissue in time to reduce the area of ​​infarction, this process can also cause myocardial dysfunction and structural damage, induce myocardial functional metabolic disorders, and cause myocardial damage that is more serious than ischemia itself, leading to worsening of the patient's condition. This tissue damage caused by the restoration of blood flow due to myocardial ischemia is myocardial ischemia-reperfusion injury, which will cause irreversible damage to the structure and function of the myocardium.

[0004] Due to the complex pathogenesis of myocardial ischemia-reperfusion injury, there are currently no targeted drugs on the market to treat this disease. In order to prevent and improve myocardial ischemia-reperfusion injury, it is an urgent task to develop drugs against myocardial ischemia-reperfusion injury based on new mechanisms or new targets.

[0005] Studies have shown that activating the JAK2 / STAT3 signaling pathway can achieve myocardial protection by reducing mitochondrial damage, producing cardioprotective factors, inhibiting pro-apoptotic substances and balancing energy metabolism. The activation of the JAK2 / STAT3 signaling pathway is initiated by the JAK2 protein, which makes JAK2 an important target for combating myocardial ischemia-reperfusion injury.

[0006] However, to date, there are no clinically available drugs targeting JAK2 to combat myocardial ischemia-reperfusion injury, and research reports on JAK2 / STAT3 activators are also limited. In addition, they also have high activity against at least one other kinase, meaning that such compounds are at risk of toxic side effects due to off-target effects. Therefore, in order to develop more highly effective and low-toxic candidate compounds for combating myocardial ischemia-reperfusion injury, discovering JAK2 activators with high kinase selectivity remains a key issue that needs to be addressed.

[0007] In view of this, a method for discovering highly selective JAK2 activators was designed to solve the above problems. Summary of the invention

[0008] To solve the problems raised in the above background technology. The present invention provides a method for discovering a highly selective JAK2 activator, which can quickly and efficiently discover a JAK2 activator with a novel skeleton and good activity and subtype selectivity. It has been verified that the discovered highly selective JAK2 activator, namely CD40, can activate the JAK2 / STAT3 signaling pathway by targeting JAK2, thereby improving the oxidative stress of myocardial ischemia-reperfusion injury and reducing the ability of cell apoptosis.

[0009] To achieve the above object, the present invention provides the following technical solution: a method for discovering a highly selective JAK2 activator, comprising the following steps:

[0010] S1: JAK small molecule data annotated with JAK1, JAK2, JAK3 and TYK2 and bioactivity index values ​​were collected through a large online drug development database, and duplicate, uncertain and non-bioactivity index value JAK small molecule data were removed to obtain a JAK small molecule data set, wherein the JAK small molecule data set includes active JAK small molecule data and inactive JAK small molecule data;

[0011] S2: The molecular descriptors of JAK small molecule data in the JAK small molecule dataset were obtained based on the PaDEL-Descriptor toolkit, and the molecular fingerprints of JAK small molecule data in the JAK small molecule dataset were obtained based on the RDkit. The JAK small molecule data were characterized by molecular descriptors and molecular fingerprints for physical and chemical characteristics and structural characteristics. The molecular descriptors were 13MD, and the molecular fingerprints were CDK, MACCS, ECFP4, ECFP6, and PubChem.

[0012] S3: Based on Scikit-learn, the JAK small molecule dataset is randomly divided into training set and test set in a ratio of 8:2;

[0013] S4: Construct a ligand-based virtual screening model, train and test the constructed ligand-based virtual screening model based on the divided training set and test set to obtain a ligand-based virtual screening model, wherein the ligand-based virtual screening model is a Multitask CapsNet model, including input layer, shared hidden feature layer, PrimaryCaps layer and DigitCaps layer, the input layer receives data input of training set and test set, the data input is a one-dimensional vector, composed of 166 MACCS fingerprints and 13 molecular descriptors, that is, the input layer has 179 neurons, the shared hidden feature layer is composed of two RBMs, the first RBM receives and encodes the original input from the input layer, the second RBM receives and extracts features from the vector from the first RBM based on tensor decomposition for learning, the first and second RBMs contain 256 and 128 neurons respectively, the PrimaryCaps layer receives and converts the extracted features from the second RBM into capsule representation, the DigitCaps layer receives and converts the capsule representation from the PrimaryCaps layer into the predicted JAK small molecule data with JAK2 activity and outputs it, the capsules in the DigitCaps layer interact and combine through a dynamic routing algorithm;

[0014] S5: Ligand-based virtual screening model screens out JAK small molecule data with JAK2 activity;

[0015] S6: Explore the potential allosteric active sites of JAK2 based on AlloFinder, and use the allosteric pocket with the highest volume as the molecular docking binding site;

[0016] S7: LigPrep module based on DS3.5 was used to prepare ligands for the screened JAK small molecule data with JAK2 activity;

[0017] S8: Based on DS3.5 and CHARMM force field conditions, protein crystal preparation was completed, where the protein crystal was a composite crystal of JAK2 protein and ruxolitinib;

[0018] S9: Based on molecular docking binding sites, a sphere was selected for the binding site, which was derived from the JAK small molecule data binding region with JAK2 activity and The residue range within , which is large enough and completely covers the entire binding region of the original ligand of the protein crystal, was used to perform molecular docking of the JAK small molecule data with JAK2 activity and the original ligand of the protein crystal based on CDOCKER, LeDock and AutoDock Vin of DS3.5;

[0019] S10: Using in vivo and in vitro biological activity tests, we obtained a highly selective JAK2 activator that targeted the molecular docking binding site and had a clear mechanism of action.

[0020] Furthermore, in step S1, the determination basis of the active JAK small molecule data and the inactive JAK small molecule data is based on the biological activity index value, and the biological activity index value is EC 50 / K d , when EC 50 / K d Less than 10 μM, which is the data of active JAK small molecules. 50 / K d Greater than 10 μM, indicating inactive JAK small molecule data.

[0021] Furthermore, in step S2, the specific information of the 13MD molecular descriptor is as follows: octanol-water partition coefficient, apparent partition coefficient at pH = 7.4, molecular solubility, molecular mass, number of hydrogen bond donors, number of hydrogen bond acceptors, number of rotatable bonds, number of rings, number of aromatic rings, total number of oxygen atoms and nitrogen atoms, polar surface area, molecular partial polar surface area and molecular surface area.

[0022] Furthermore, in step S4, fine-tuning is performed during the training of the ligand-based virtual screening model. During the fine-tuning process, the loss function proposed by Hinton is used to achieve end-to-end optimization in RBM, and a contribution matrix is ​​generated through a back-propagation algorithm to achieve fine-tuning.

[0023] Furthermore, in step S6, the evaluation indicators predicted by AlloFinder include: allosteric site score, drug-like score, perturbation score and volume.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] The present invention uses a combination of a ligand-based virtual screening model and molecular docking as a virtual screening tool, that is, a combination of a Multitask CapsNet model and molecular docking as a virtual screening tool. This combined method can quickly and efficiently discover JAK2 activators with novel skeletons and good activity and subtype selectivity. It has been verified that the discovered highly selective JAK2 activator, namely CD40, can activate the JAK2 / STAT3 signaling pathway by targeting JAK2, thereby improving the oxidative stress of myocardial ischemia-reperfusion injury and reducing cell apoptosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 The figure is a flow chart of the method for discovering highly selective JAK2 activators of the present invention. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] See also Figure 1 The present invention provides the following technical solution: a method for discovering a highly selective JAK2 activator, comprising the following steps:

[0029] S1: JAK small molecule data annotated with JAK1, JAK2, JAK3 and TYK2 and bioactivity index values ​​were collected through a large online drug development database, and duplicate, uncertain and non-bioactivity index value JAK small molecule data were removed to obtain a JAK small molecule data set, wherein the JAK small molecule data set includes active JAK small molecule data and inactive JAK small molecule data;

[0030] The above-mentioned active JAK small molecule data and inactive JAK small molecule data are judged based on the biological activity index value. The biological activity index value is EC 50 / K d , when EC 50 / K d Less than 10 μM, which is the data of active JAK small molecules. 50 / K d Greater than 10 μM, which is the data of inactive JAK small molecules;

[0031] S2: The molecular descriptors of JAK small molecule data in the JAK small molecule dataset were obtained based on the PaDEL-Descriptor toolkit, and the molecular fingerprints of JAK small molecule data in the JAK small molecule dataset were obtained based on the RDkit. The JAK small molecule data were characterized by molecular descriptors and molecular fingerprints for physical and chemical characteristics and structural characteristics. The molecular descriptors were 13MD, and the molecular fingerprints were CDK, MACCS, ECFP4, ECFP6, and PubChem.

[0032] The detailed information of the above 13MD molecular descriptors are as follows: octanol-water partition coefficient, apparent partition coefficient at pH = 7.4, molecular solubility, molecular mass, number of hydrogen bond donors, number of hydrogen bond acceptors, number of rotatable bonds, number of rings, number of aromatic rings, total number of oxygen atoms and nitrogen atoms, polar surface area, molecular partial polar surface area and molecular surface area;

[0033] S3: Based on Scikit-learn, the JAK small molecule dataset is randomly divided into training set and test set in a ratio of 8:2;

[0034] S4: Construct a ligand-based virtual screening model, train and test the constructed ligand-based virtual screening model based on the divided training set and test set to obtain a ligand-based virtual screening model, wherein the ligand-based virtual screening model is a Multitask CapsNet model, including input layer, shared hidden feature layer, PrimaryCaps layer and DigitCaps layer, the input layer receives data input of training set and test set, the data input is a one-dimensional vector, composed of 166 MACCS fingerprints and 13 molecular descriptors, that is, the input layer has 179 neurons, the shared hidden feature layer is composed of two RBMs, the first RBM receives and encodes the original input from the input layer, the second RBM receives and extracts features from the vector from the first RBM based on tensor decomposition for learning, the first and second RBMs contain 256 and 128 neurons respectively, the PrimaryCaps layer receives and converts the extracted features from the second RBM into capsule representation, the DigitCaps layer receives and converts the capsule representation from the PrimaryCaps layer into the predicted JAK small molecule data with JAK2 activity and outputs it, the capsules in the DigitCaps layer interact and combine through a dynamic routing algorithm;

[0035] The above-mentioned ligand-based virtual screening model will be fine-tuned during the training process. During the fine-tuning process, the loss function proposed by Hinton is used to achieve end-to-end optimization in RBM, and the contribution matrix is ​​generated through the back-propagation algorithm to achieve fine-tuning;

[0036] S5: Ligand-based virtual screening model screens out JAK small molecule data with JAK2 activity;

[0037] S6: Explore the potential allosteric active sites of JAK2 based on AlloFinder, and use the allosteric pocket with the highest volume as the molecular docking binding site;

[0038] The evaluation indicators predicted by AlloFinder mentioned above include: allosteric site score, drug-like score, perturbation score and volume;

[0039] The results of AlloFinder prediction above showed that there may be four potential allosteric binding sites in JAK2 protein, with pocket volumes of 2215, 1961, 1894, 1413 and The allosteric pocket with the highest volume was selected as the best allosteric binding site for molecular docking;

[0040] S7: LigPrep module based on DS3.5 was used to prepare ligands for the screened JAK small molecule data with JAK2 activity;

[0041] S8: Based on DS3.5 and CHARMM force field conditions, protein crystal preparation was completed, where the protein crystal was a composite crystal of JAK2 protein and ruxolitinib;

[0042] S9: Based on molecular docking binding sites, a sphere was selected for the binding site, which was derived from the JAK small molecule data binding region with JAK2 activity and The residue range within , which is large enough and completely covers the entire binding region of the original ligand of the protein crystal, was used to perform molecular docking of the JAK small molecule data with JAK2 activity and the original ligand of the protein crystal based on CDOCKER, LeDock and AutoDock Vin of DS3.5;

[0043] S10: Using in vivo and in vitro biological activity tests, we obtained a highly selective JAK2 activator that targeted the molecular docking binding site and had a clear mechanism of action.

[0044] ADME / T prediction:

[0045] ADMETlab 2.0 open source program was used to evaluate various ADME / T properties of the obtained highly selective JAK2 activators;

[0046] PAINS Filtering:

[0047] Use the PAINS screening tool to reduce the occurrence of false positives;

[0048] The PAINS filter of RDKit was used to filter the highly selective JAK2 activators obtained by virtual screening;

[0049] Drugability assessment:

[0050] The Lipinski rule of the obtained highly selective JAK2 activators was evaluated using the SwissADME web server;

[0051] The RDkit toolkit is used to calculate the quantitative estimate of drugability (QED) of small molecules.

[0052] After ADME / T prediction, PAINS filtering, and drugability evaluation, 49 JAK small molecules were retained based on their structures;

[0053] Enzyme level activity test:

[0054] Based on DSF technology, the interaction between ligand and protein is evaluated by measuring the thermal stability parameters of the protein:

[0055] The protein, small molecules and SYPRO orange were mixed in a buffer solution, and the system was placed in a Bio-Rad fluorescent quantitative PCR instrument. The system was heated from 25°C to 95°C at a rate of 1°C per minute, and the signal excited by the FRET filter was detected.

[0056] With temperature as the independent variable and fluorescence signal as the dependent variable, a temperature-fluorescence curve was prepared. The Boltzmann fitting curve in GraphPad Prism6.0 analysis software was used to obtain Tm. With DMSO as the control, the following was calculated:

[0057] △Tm=Tm(small molecule)-Tm(DMSO);

[0058] The melting temperature of the protein was detected by DSF technique;

[0059] After testing with DSF technology, it was found that the highly selective JAK2 activator has JAK2 activity;

[0060] ADP-Glo ​​Kinase Assay Kit-Based JAK2 Agonist Activity and Isoform Selectivity Model:

[0061] Prepare a mixed solution of ATP and Kinase substrate and a kinase solution;

[0062] Initiate kinase reactions;

[0063] Add mixed solution, kinase solution, and 1% DMSO to the positive control wells, and 1% DMSO to the negative control wells;

[0064] To stop the kinase reaction and consume the remaining ATP, add ADP-Glo TM Reagents;

[0065] Add kinase detection reagent to convert ADP to ATP, and use coupled luciferase / luciferin reaction to detect newly synthesized ATP;

[0066] Detect the luminescent signal with an ELISA reader to confirm the activation / antagonism effect of the JAK2 active compound;

[0067] The EC values ​​of activating compounds were obtained using the Caliper MSA method. 50 ;

[0068] This model was used to test the activity of JAK small molecules against JAK1, JAK3 and TYK2 at various concentrations to explore subtype selectivity. It was found that the JAK small molecule, CD40, showed the best activity and subtype selectivity against JAK2.

[0069] Myocardial cell activity test:

[0070] (1) Establishment of cardiomyocyte model

[0071] H9C2 cardiomyocytes were cultured in DMEM high-glucose medium containing 10% fetal bovine serum and 1% penicillin / streptomycin in a 37°C constant-temperature incubator containing 5% CO2. The cell status was observed every other day, the culture medium was replaced, and the cells were passaged regularly.

[0072] According to the experimental design, cells were plated at an appropriate density and randomly divided into 4 groups after the cells adhered to the wall:

[0073] (A) Control group (Control), H9C2 cells were routinely cultured for 24 h;

[0074] (B) OGD / R group, simulating myocardial injury in vivo, H9C2 cells were cultured in glucose-deficient and serum-deficient medium, exposed to hypoxic conditions of 1% O2, 5% CO2, and 94% N2 and cultured in a 37°C constant temperature incubator for 8 h. After hypoxia, the cell culture medium was replaced with normal culture medium and cultured in a constant temperature incubator for 16 h.

[0075] (C) Lead compound + OGD / R group, cells were pretreated with different concentrations of lead compounds in hypoxia and low glucose medium under OGD / R for 6 h;

[0076] (D) Lead compound + AG490 + OGD / R group, cells were treated with 20 μM lead compound and 5 μM AG490 in hypoxia and low glucose medium under OGD / R for 6 h;

[0077] (2) Cardiomyocyte apoptosis test

[0078] Cell Counting Kit-8 was used to detect the viability of H9C2 cardiomyocytes in each experimental group. H9C2 cells were seeded in a 96-well plate at a density of 3500 cells / well. After specific treatment, 10 μL of CCK-8 solution was added to each well and incubated in an incubator for 4 hours. The optical density D value of each well was measured at 450 nm using an ELISA reader to calculate the cell viability. The calculation formula of cell viability is as follows:

[0079] Relative cell survival rate = (D value of experimental group - D value of blank group) / (D value of control group - D value of blank group) × 100%;

[0080] After treatment, the culture medium was collected and transferred to a new 96-well plate, the LDH reaction mixture was added to each well, and the plate was incubated at room temperature for 30 min. The absorbance of each well was measured at 440 nm using a microplate reader to quantitatively assess cell death;

[0081] (3) Flow cytometry

[0082] H9C2 cells were seeded in 6-well plates at a density of 2.5×105 cells / well. After treatment, the cells were washed with cold PBS and resuspended in annexin binding buffer. Subsequently, the cells were placed in 5μL Annexin V-FITC and 5μL 7-AAD staining solution in the dark and incubated at room temperature for 10 minutes. 400μL 1× Annexin V-FITC binding buffer was added and the FACS Calibur flow cytometer was used to complete the detection within 1 hour to determine the cell apoptosis.

[0083] (4) Active oxygen test

[0084] The cells were incubated with 10 μM probe DCFH-DA for 25 min, then washed twice with PBS and imaged using a fluorescence microscope, and then the fluorescence intensity was checked using a fluorescence microplate reader to detect the generation of reactive oxygen species;

[0085] (5) TUNEL staining

[0086] H9C2 cardiomyocytes were cultured in 24-well plates for 24 h. After treatment, the cells were fixed with 4% paraformaldehyde for 15 min, and then permeabilized with 0.5% Triton-X 100 for 20 min at room temperature. The permeabilized samples were immersed in PBS and rinsed three times, each time for 5 min. The treated samples were placed in a humidified box to keep the samples moist. Then 50 μL of TUNEL buffer was added to each sample, and it was incubated in a humidified box at 37°C for 1 h, washed with PBS three times, and then counterstained with DAPI and incubated in the dark for 5 min. TUNEL images were collected using a Carl Zeiss microscope and ZEN software.

[0087] The results showed that CD40 could inhibit OGD / R-induced oxidative stress and apoptosis in H9C2 cardiomyocytes.

[0088] Myocardial ischemia-reperfusion injury in vivo rat activity test:

[0089] Male SPF SD rats aged 6-8 weeks and weighing 200-250g were randomly divided into 3 groups: drug group (50mg / kg), sham operation group and surgery group. All drugs in the drug group were dissolved in dimethyl sulfoxide solution. The sham operation group and surgery group were perfused with equal volume of normal saline for 7 consecutive days. The myocardial ischemia-reperfusion injury model was established 30min after the last drug administration.

[0090] Five batches of rats were used in the experiment. The first batch of rats was used to confirm the establishment of the myocardial ischemia-reperfusion injury model. The second batch of rats was subjected to TTC staining (n=6) to detect the myocardial infarction area. The third batch of rats was subjected to ELISA detection (n=9) to evaluate the content of CK-MB and cTnI in the heart tissue of each group. The fourth batch of rats was subjected to HE staining (n=6). The fifth batch of rats was used to determine the expression of p-JAK2, JAK2, p-STAT3, STAT3, active Capsase3, Bcl-2 and Bax proteins in myocardial tissue.

[0091] All rats were housed under the following conditions:

[0092] (A) Temperature 20-26°C;

[0093] (B) Relative humidity 40-70%;

[0094] (C) 12:12-h light-dark cycle;

[0095] (D) Individually ventilated cage;

[0096] (1) Construction of myocardial ischemia-reperfusion injury model

[0097] Rats were anesthetized with 2% isoflurane, fixed in supine position after successful anesthesia, tracheotomy connected to Jiangwan I type micro-artificial ventilator, positive pressure ventilation, tidal volume 6-7ml / kg, respiratory rate 60 times / min, inhalation: exhalation ratio 1:2, left thoracotomy, pericardium cut open, heart exposed, 6-0 silk thread passed between the root of the left atrial appendage and the pulmonary artery cone, left anterior descending branch of coronary artery temporarily ligated, rats were subjected to 30 minutes of ischemia, reperfusion for 24 hours, 30 minutes later, the ligature was released for 2 hours, myocardial ischemia reperfusion injury model was established, and the successful ligation of coronary artery was confirmed by the color change of myocardial tissue;

[0098] (2) TTC staining

[0099] After reperfusion was established, the heart was removed and washed with 0.9% saline to remove blood. The rat myocardial tissue was taken and cut into slices from the apex to the base of the heart with a thickness of about 1 mm. The tissue slices were immersed in 1% TTC solution at 37°C for 30 min and then fixed with 4% polysorbate for 2 h. After staining, the slices were photographed and analyzed and calculated by Image-pro Plus v6.0 software.

[0100] (3) HE staining

[0101] After the heart tissue is dehydrated and paraffin-embedded, the trimmed wax block is placed in a paraffin slicer and sliced ​​to a thickness of 3-4 μm. The sliced ​​tissue slices are placed in a 40°C water bath for spreading, and an anti-slip glass slide is inserted into the water surface at an angle to pick up the slices, so that the slices are attached to the appropriate position of the slide, and baked in a 60°C oven for 3 hours. After the slices are taken out, subsequent staining is performed at room temperature, and HE staining is performed according to the instructions. The morphological changes are observed under an optical microscope and photographed;

[0102] (4) Western blotting

[0103] Total protein was extracted according to the instructions of the protein extraction kit, and the protein concentration was determined by the bicinchoninic acid protein assay. 40 μg of total protein was separated under the following electrophoresis conditions: 75 V for 30 min on the stacking gel (6%) and 180 V for 1 h on the separation gel (12%). The gel was carefully peeled off the glass plate and transferred to the PVDF membrane. The PVDF membrane was immersed in 5% skim milk and placed on a shaker for 2 h. The PVDF membrane was washed 3 times with TBST, and then placed in the primary antibody dilution solution and incubated overnight in a 4°C refrigerator shaker. The PVDF membrane was washed 5 times with TBST, and an appropriate amount of secondary antibody was added and incubated on a shaker at room temperature for 2 h. The PVDF membrane was washed again with TBST 5 times, each time for 5 min. The protein bands were detected using an enhanced chemiluminescence detection kit, and the results were observed and analyzed using Image-Pro Plu software.

[0104] The results showed that after CD40 treatment, the apoptosis rate of myocardial cells in rats with myocardial ischemia-reperfusion injury was significantly reduced, and myocardial function indicators were significantly improved.

[0105] Target validation:

[0106] (1) Target stabilization technology based on drug affinity reaction

[0107] H9C2 cells were collected, lysed with RIPA lysis buffer on ice for 15 min, and centrifuged at 12000 rpm / min for 15 min to obtain the supernatant. The protein concentration was determined by Bradford reagent. Before drug treatment, the protein was divided into two parts. The samples were treated with CD40 and DMSO at room temperature for 6 hours. The two parts were then divided into three parts. Different volumes of protease E (0.05 mg / mL) were added according to the mass ratio of protease E to protein mass (1:500, 1:1000 and 1:1500), and incubated at 40°C for 10 min. 5× loading buffer was added and boiled for 10 min. All parts of each sample were used for Western blot analysis. Another detection method was roughly the same as above, except that each sample was controlled with the same enzyme concentration, incubated at different drug concentrations, and its sample expression was detected;

[0108] (2) Using SPR technology to confirm the binding ability of compounds to JAK2

[0109] Based on SPR technology, biomembrane interferometry was used to track the interaction between JAK2 activators and proteins in real time. The proteins were bound to the surface of CM5 chips through amino coupling. Small molecules of different concentrations were injected into Biacore and passed through the chip surface at a flow rate of 30 μL / min, which was the binding phase. The chip was then rinsed with running buffer at the same flow rate, which was the dissociation phase. The KD value of the small molecule to be tested was calculated using BIA evaluation software, and the affinity binding constant was detected using SPR technology to confirm the binding ability of the highly selective JAK2 activator with JAK2.

[0110] (3) Molecular dynamics simulation of the dynamic binding ability and key binding sites of compounds with JAK2

[0111] In order to explore the dynamic binding ability of highly selective JAK2 activators with JAK2 protein and the key amino acids located at potential allosteric sites, the GROMACS molecular dynamics simulation program was first used to perform molecular dynamics simulation on the complex obtained by molecular docking. The electrostatic potential of the small molecule was calculated and analyzed before the simulation, and the appropriate electric field was selected and the system was energy optimized to achieve the best effect. Sampling was performed in the last 2ns of the dynamics simulation, and the binding free energy was calculated using the MM-PBSA method to obtain the dynamic binding ability of the highly selective JAK2 activator with JAK2 protein. The MM / GBSA method was used to decompose the binding free energy to confirm the key amino acid binding sites of the highly selective JAK2 activator with JAK2 protein.

[0112] The results showed that CD40 could interact with JAK2 kinase.

[0113] In summary, the method of the present application can be used to discover highly selective JAK2 activators.

[0114] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for discovering a highly selective JAK2 activator, characterized in that: The following steps are involved: S1: JAK small molecule data annotated with JAK1, JAK2, JAK3 and TYK2 and bioactivity index values ​​were collected through a large online drug development database, and duplicate, uncertain and non-bioactivity index value JAK small molecule data were removed to obtain a JAK small molecule data set, wherein the JAK small molecule data set includes active JAK small molecule data and inactive JAK small molecule data; S2: The molecular descriptors of JAK small molecule data in the JAK small molecule dataset were obtained based on the PaDEL-Descriptor toolkit, and the molecular fingerprints of JAK small molecule data in the JAK small molecule dataset were obtained based on the RDkit. The JAK small molecule data were characterized by molecular descriptors and molecular fingerprints for physical and chemical characteristics and structural characteristics. The molecular descriptors were 13MD, and the molecular fingerprints were CDK, MACCS, ECFP4, ECFP6, and PubChem. S3: Based on Scikit-learn, the JAK small molecule dataset is randomly divided into training set and test set in a ratio of 8:2; S4: Construct a ligand-based virtual screening model, train and test the constructed ligand-based virtual screening model based on the divided training set and test set to obtain a ligand-based virtual screening model, wherein the ligand-based virtual screening model is a Multitask CapsNet model, including input layer, shared hidden feature layer, PrimaryCaps layer and DigitCaps layer, the input layer receives data input of training set and test set, the data input is a one-dimensional vector, composed of 166 MACCS fingerprints and 13 molecular descriptors, that is, the input layer has 179 neurons, the shared hidden feature layer is composed of two RBMs, the first RBM receives and encodes the original input from the input layer, the second RBM receives and extracts features from the vector from the first RBM based on tensor decomposition for learning, the first and second RBMs contain 256 and 128 neurons respectively, the PrimaryCaps layer receives and converts the extracted features from the second RBM into capsule representation, the DigitCaps layer receives and converts the capsule representation from the PrimaryCaps layer into the predicted JAK small molecule data with JAK2 activity and outputs it, the capsules in the DigitCaps layer interact and combine through a dynamic routing algorithm; S5: Ligand-based virtual screening model screens out JAK small molecule data with JAK2 activity; S6: Explore the potential allosteric active sites of JAK2 based on AlloFinder, and use the allosteric pocket with the highest volume as the molecular docking binding site; S7: LigPrep module based on DS3.5 was used to prepare ligands for the screened JAK small molecule data with JAK2 activity; S8: Based on DS3.5 and CHARMM force field conditions, protein crystal preparation was completed, where the protein crystal was a composite crystal of JAK2 protein and ruxolitinib; S9: Based on molecular docking binding sites, a sphere was selected for the binding site, which was derived from the JAK small molecule data binding region with JAK2 activity and The residue range within , which is large enough and completely covers the entire binding region of the original ligand of the protein crystal, was used to perform molecular docking of the JAK small molecule data with JAK2 activity and the original ligand of the protein crystal based on CDOCKER, LeDock and AutoDock Vin of DS3.5; S10: Using in vivo and in vitro biological activity tests, we obtained a highly selective JAK2 activator that targeted the molecular docking binding site and had a clear mechanism of action.

2. The method for discovering a highly selective JAK2 activator according to claim 1, characterized in that: In step S1, the determination basis of active JAK small molecule data and inactive JAK small molecule data is based on the biological activity index value, and the biological activity index value is EC 50 / K d , when EC 50 / K d Less than 10 μM, which is the data of active JAK small molecules. 50 / K d Greater than 10 μM, indicating inactive JAK small molecule data.

3. The method for discovering a highly selective JAK2 activator according to claim 1, characterized in that: In step S2, the specific information of the 13MD molecular descriptor is as follows: octanol-water partition coefficient, apparent partition coefficient at pH = 7.4, molecular solubility, molecular mass, number of hydrogen bond donors, number of hydrogen bond acceptors, number of rotatable bonds, number of rings, number of aromatic rings, total number of oxygen atoms and nitrogen atoms, polar surface area, molecular partial polar surface area and molecular surface area.

4. The method for discovering a highly selective JAK2 activator according to claim 1, characterized in that: In step S4, fine-tuning is performed during the training of the ligand-based virtual screening model. During the fine-tuning process, the loss function proposed by Hinton is used to achieve end-to-end optimization in RBM, and a contribution matrix is ​​generated through a back-propagation algorithm to achieve fine-tuning.

5. The method for discovering a highly selective JAK2 activator according to claim 1, characterized in that: In step S6, the evaluation indicators predicted by AlloFinder include: allosteric site score, drug-like score, perturbation score and volume.

Images