Multi-task MSSP model can accurately predict selective Src inhibitors

Multi-task MSSP model can accurately predict selective Src inhibitors | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Multi-task MSSP model can accurately predict selective Src inhibitors Xuecong Tian, Luyang Han, Ying Su, Haiqing Sun, Sizhe Zhang, Chen Chen, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4475200/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Src family kinases (SFKs), non-receptor tyrosine kinases, crucially contribute to invasion, tumor progression, epithelial-mesenchymal transition, angiogenesis, and metastasis. Thus, Src inhibitors offer a promising avenue for cancer therapy. This study introduced a multitask MSSP deep learning model to predict molecule inhibitory activity across multiple Src subtypes. Comparative assessment against four traditional machine learning methods—Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), eXtreme Gradient Boosting (XGBoost)—established the superior performance of the multitask MSSP model. It demonstrated the best comprehensive performance, achieving F1-Score and AUC values of 0.906 and 0.975, respectively. An online web server, "SRC-Predictor," was created to aid the practical application of the multitask MSSP model, predicting compounds' potential inhibitory activity against Src. Finally, compounds ranking in the top twenty based on model predictions were selected for experimental validation. Literature search for these compounds revealed limited research on four of them concerning Src. Molecular docking identified Doramapimod as exhibiting better affinity towards Src compared to reference compounds. It significantly inhibited Lyn kinase activity and influenced the secretion levels of inflammatory factors in LPS-induced macrophages. Experimental validation confirmed that our study provides a novel approach for identifying and screening lead compounds as Src inhibitors. Multitask MSSP deep learning model SRC-Predictor web server Src inhibitors Doramapimod Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 1. Introduction Src family kinases (SFKs), a subset of non-receptor tyrosine kinases, serve as crucial signaling intermediates in metazoans [ 1 ]. They are activated by various growth factors, cytokines, and antigen receptors, thereby regulating crucial biological processes like proliferation, differentiation, apoptosis, migration, and metabolism [ 2 ]. Overexpression or aberrant activation of Src family kinases in both epithelial and non-epithelial cancers significantly contributes to tumorigenesis [ 3 ]. c-Src stands as one of the earliest identified and extensively studied members among the Src family kinases [ 4 ]. Consequently, most inhibitors targeting Src family kinases are designed to selectively target the c-Src subtype. These inhibitors function by binding to the active site of the c-Src subtype, impeding its activity and disrupting cancer-related processes like cell proliferation, migration, and invasion. They competitively interact with the substrate binding site or ATP binding site of the c-Src subtype, thereby impeding the binding of substrates or ATP, ultimately inhibiting the enzyme's catalytic activity [ 5 ]. Additionally, Src family kinase subtypes like Fyn, Yes, Lyn, Lck, and Hck are considered potential therapeutic targets. These subtypes are pivotal in cellular signal transduction and are linked to the onset of diverse diseases, encompassing cancer, immune disorders, and neurological conditions [ 6 ]. While acknowledging the importance of these subtypes in diseases, the development of precise inhibitors remains limited, and research in this domain is in its early stages [ 7 ]. The ongoing development of selective inhibitors targeting different subtypes holds substantial importance in offering more targeted and selective therapeutic strategies. Among the Src family, Lyn, c-Src, and Lck are among the most easily activated tyrosine kinases in various types of cancer cells [ 8 ]. Recent reports highlight the critical role of Lyn kinase in modulating inflammatory signaling, microvascular permeability, neutrophil recruitment, and facilitating hepatic fibrosis [ 9 ]. Additionally, Lyn kinase can induce B-cell leukemia and granulocytic leukemia by phosphorylating the proto-oncogene protein Cbl [ 10 ]. Further analysis of acute myeloid leukemia (AML) cells revealed a significant increase in Lyn kinase activity in 76% of the cells. This elevation led to an increase in the transcription factor STAT5, consequently promoting disease progression [ 11 ]. Notably, inhibiting Lyn kinase activity in AML cell lines significantly reduces cell growth [ 12 ]. Moreover, numerous studies suggest that Lyn's overactivation contributes to the occurrence of various solid tumors, including breast cancer [ 13 ], colon cancer [ 8 ], and prostate cancer [ 8 ]. Therefore, inhibitors targeting the Src family, particularly Lyn, represent a promising strategy for targeting hematologic malignancies and several solid tumor diseases [ 3 ]. Numerous inhibitors targeting the Src family have been developed for the treatment of both hematologic disorders and solid malignancies [ 3 ]. Dasatinib, the first FDA-approved drug with confirmed Src family kinase inhibition, is primarily used to treat hematologic disorders such as chronic myeloid leukemia (CML) and AML [ 14 ]. Bosutinib, known as a Lyn kinase inhibitor, is primarily employed in the treatment of malignant hematologic disorders [ 15 ]. The diversity of kinase inhibitors enables the selection of different drugs based on a patient's resistance profile, thereby extending the duration of treatment [ 16 ]. For instance, Bosutinib, Dasatinib, and Ponatinib effectively overcome BCR-ABL resistance through distinct mechanisms and broader inhibitory capabilities, offering new therapeutic options for patients resistant to Imatinib [ 17 ]. However, only a handful of drugs have gained FDA approval, and achieving selective inhibition of Src kinases without interfering with other related signaling pathways remains a challenge [ 3 ]. Src family kinase inhibitors represent a significant area in cancer therapy with vast research prospects. It's worthwhile to persist in the ongoing development of potential inhibitors for treating diseases. The discovery of kinase inhibitors encompasses high-throughput screening, compound optimization, biological evaluation, and clinical trials, covering multiple stages from chemical database screening to drug approval [ 18 ]. However, traditional experimental assay methods are often expensive, time-consuming, and constrained by limited parameter space [ 19 ]. In recent years, Computer-Aided Drug Design (CADD) has played a pivotal role in the early stages of drug discovery, aligning computational and biological/chemical knowledge [ 20 ]. CADD methods have been widely employed in the discovery and design of Src family kinase inhibitors. In 2019, Koneru et al. employed QSAR modeling and molecular dynamics to redesign the second-generation Src kinase inhibitor RL-45. The newly designed compounds mitigated mutation-associated Src kinase resistance and exhibited enhanced binding to the kinase's active site [ 21 ]. In 2020, Yu et al. employed virtual docking programs to screen 2 million molecules in databases, aiming to identify small molecule binders targeting the pY + 3 site of the Lck kinase SH2 domain within the Src family [ 22 ]. Similarly, in 2020, Zhang et al. discovered potential lead compounds as anti-Src kinase agents by integrating virtual screening of pharmacophores, molecular docking, and dynamics simulations [ 23 ]. These methods significantly aid in drug development and the analysis of large biomedical datasets in cancer therapy. Furthermore, deep learning, a critical branch of artificial intelligence, automates the learning of advanced feature representations from raw data, optimizing the identification of lead compounds and molecular understanding of diseases [ 24 ]. In this study, an interpretable model based on the multitask MSSP deep learning (DL) framework was constructed for virtual screening of active molecules against multiple SRC subtypes. Using the MSSP model, an online platform ( http://ilovemyhome.cc/ ) was created to aid in identifying and adjusting selective SRC inhibitors. Eventually, using the MSSP model, a library of small molecules with the potential to inhibit various Src-related pathways was screened. Four compounds, most likely to act as Lyn kinase inhibitors within the Src family, were further subjected to molecular docking and molecular dynamics simulations to identify lead compounds. Targeting RAW264.7 cells, these lead compounds were explored for their alleviating effects on inflammation. The lead compounds proposed in this study may pave the path toward creating novel precision medicines for inflammation therapy. 2. Materials and Methods 2.1 Materials 2.1.1 Experimental Cells The murine monocyte/macrophage cell line RAW264.7 was provided by our laboratory. 2.1.2 Experimental Drugs Doramapimod (BIRB 796) was purchased from Shanghai YuanYe Biotechnology Co., Ltd. Prior to experimentation, BIRB 796 was dissolved in dimethyl sulfoxide (DMSO) 2.1.3 Experimental Reagents Phosphate-buffered saline (PBS), DMEM cell culture medium, and fetal bovine serum (FBS) were purchased from Gibco (Gaithersburg, Maryland, USA). Fluorescent-labeled antibodies for CD86-PE and CD80-APC were procured from Elabscience (China). Cell Counting Kit-8 (CCK-8) was obtained from Shanghai Beyotime Biotechnology Co., Ltd. Enzyme-linked immunosorbent assay (ELISA) kits for IL-6 and TNF-α were purchased from Elabscience (China). Lipopolysaccharide (LPS) and dimethyl sulfoxide (DMSO) were acquired from Sigma-Aldrich (St. Louis, Missouri, USA). 2.1.4 Experimental Instruments CO2 cell incubator, low temperature and high speed refrigerated centrifuge (Thermo Fisher Scientific, USA); Flow cytometry (BD FACSCalibur, USA); Enzyme labeling instrument (Bio-Rad Company); Low speed centrifuge (Sichuan Shuke Instrument Co., Ltd.). 2.2 Data Collection and Preparation In this study, the dataset containing compounds related to the 8 subtypes of the Src kinase family was sourced from PubChem [ 25 ] and ChEMBL [ 26 ]. Ultimately, a total of 7935 compounds relevant to the 8 subtypes of the Src kinase family, covering 12001 bioassay data, were compiled. The processed dataset was randomly partitioned into a training set (80%), a validation set (10%), and a test set (10%). The training and validation sets were utilized for model development and hyperparameter optimization, while the test set was employed to evaluate the established model's performance. The collected data underwent filtration based on Lipinski's Rule of Five and Veber's Rules: removal of data lacking bioassay values or explicit molecular properties, normalization of biological activity units, calculation of average bioassay values as the final value, exclusion of duplicate molecules and those with a molecular weight exceeding 1000 Da, and standardization of each compound in the dataset.The dataset generated was utilized to train the multitask MSSP model, then compared with baseline models like Random Forest (RF), Support Vector Machine (SVM), k-Nearest Neighbors (KNN), and XGBoost. An overview of the entire process is illustrated in Fig. 1 . 2.3 Multi-task MSSP Deep Learning Framework This experiment established a multitask MSSP model to accurately predict active molecules targeting the Src kinase family. Innovative design was employed to address two distinct feature modalities—molecular fingerprint features and molecular graph structure features—by introducing a novel molecule fusion algorithm. This algorithm mapped different modalities of molecules into shared and molecule-specific representation spaces, utilizing feature similarity to align and merge diverse modalities effectively. The experiment utilized parameter sharing in multitask learning to address shortcomings of traditional machine learning methods in handling correlated subtasks and complex features. To enhance interpretability, the experiment incorporated a graph attention mechanism, quantitatively representing the importance of chemical fragments in predicting molecular properties. The multitask MSSP model was trained on a NVIDIA GeForce RTX 3090 GPU. 2.4 Baseline Machine Learning Algorithms To further validate the effectiveness of the multitask MSSP model in predicting SRC kinase inhibitors, we conducted experiments and compared it with four traditional machine learning algorithms. These four algorithms include Random Forest (RF), Support Vector Machine (SVM), k-Nearest Neighbors (KNN), and eXtreme Gradient Boosting (XGBoost). These algorithms are widely applied classical methods in the fields of chemistry and biology. The RF, SVM, and KNN models were developed using the scikit-learn Python package ( https://github.com/scikit-learn/scikit-learn , version: 1.2.1); the XGBoost model was developed using the XGBoost Python package ( https://github.com/dmlc/xgboost , version: 1.7.4). 2.5 Model Performance Evaluation After establishing the classification model using the training set, the performance was evaluated using the test set. This study employed Accuracy, Recall, Precision, and F1-Score as the model evaluation metrics. The metrics values were computed based on Table 1 using specific formulas (1)-(4). \(Accuracy=\frac{TP+TN}{TP+FP+TN+FN}\) (1) \(Recall=\frac{TP}{TP+FN}\) (2) \(Precision=\frac{TP}{TP+FP}\) (3) \(F1-Score=\frac{2*precision*recall}{precision+recall}\) (4) In addition, we also utilized the area under the receiver operating characteristic curve (AUC) as a comprehensive evaluation metric for the model. Table 1 Confusion matrix Predicted Patients with glioma Normal subjects Patients with glioma TP FN Normal subjects FP TN TN = true negative, TP = true positive, FN = false negative, FP = false positive, TPR = true positive rate, and FPR = false positive rate. 2.6 Molecular Docking By utilizing MSSP to predict the collected dataset related to Src kinases, we selected the top twenty small molecule compounds based on the model's predicted scores. To further validate whether these molecules selected through MSSP screening possess SRC kinase inhibitory activity, a virtual screening experiment was conducted. Following literature review to identify small molecule compounds with less research pertaining to Src kinases, molecular docking simulations were performed using AutoDock Vina software to dock these compounds with the target protein. Lyn kinase was employed as the substrate, and small molecule compounds displaying higher affinity with Lyn kinase were screened and designated as lead compounds. 2.7 Molecular Dynamics Simulations To further investigate the potential lead compounds exhibiting Src subtype Lyn kinase inhibitory activity, we employed molecular dynamics simulations to acquire detailed molecular structure, dynamics, and stability information. Imatinib was selected as the control group, and using the System Builder tool in Maestro software, we constructed the physiological aqueous salt system of the ligand-Lyn kinase protein complex. Employing Maestro's molecular dynamics module, we conducted a 10 nanosecond dynamics simulation of the constructed system. Simulation parameters included a recording interval of 10 picoseconds, utilizing the NPT ensemble type, a temperature of 300K, and a pressure of 1.01325 bar. This experimental design aims to precisely explore the activity and stability of the lead compounds, providing robust support for further drug design endeavors. 2.8 Cell Culture Resuscitated RAW264.7 mouse monocyte/macrophage cell line from liquid nitrogen or ultra-low temperature freezer, seeded in cell culture dishes with DMEM high glucose completed medium containing 10% fetal bovine serum and 1% penicillin-streptomycin, and incubated at 37°C and 5% CO2 in a CO2 incubator. The medium is replaced every 24 hours. Passage can be performed when the cell confluence reaches 70%-80%. Cells in the logarithmic growth phase can be used for subsequent experiments. 2.9 Cell Viability Assay Collect cells in the logarithmic growth phase and count them. Resuspend cells at a concentration of 1 × 105 cells/mL and seed them into a 96-well plate. Set up a normal control group, a positive control group (LPS), and a drug treatment group (LPS + BIRB 796). Each group should have 6 replicates. Treat cells with different concentrations of BIRB 796 (200 nM, 400 nM, 800 nM, 10 µM, 25 µM, 50 µM) and LPS (10 ng/mL) simultaneously. Incubate the cells in a 37°C incubator for 24 h. Add 10 µL of CCK-8 reagent to each well and incubate for an additional 1 h. Measure the absorbance (A) of each well at a wavelength of 450 nm using a Microplate reader. Calculate the cell viability according to the formula. 2.10 Flow Cytometry Collect RAW264.7 cells and seed them in a 24-well plate at a density of 1 × 105 cells per well. Set up a normal control group, a positive control group (LPS), and a drug treatment group (LPS + BIRB 796). Each group should have 3 replicates. Treat the cells with different concentrations of BIRB 796 (25, 50, 100, 200 nM) in combination with LPS. Incubate the cells in a 37°C incubator for 12 hours, then collect the cells. Wash each sample twice with PBS and stain with APC-CD80 and PE-CD80 flow cytometry antibodies for 15 minutes. After washing with PBS, samples were detected by FACSCalibur, and data were analyzed by FlowJo software. 2.11 ELISA Assays Collect RAW264.7 cells and seed them in a 24-well plate at a density of 1 × 105 cells per well. Treat the cells with different concentrations of BIRB 796 (25, 50, 100, 200 nM) in combination with LPS. Incubate the cells in a 37°C incubator for 12 hours. After incubation, collect the cell suspension by centrifuging at 1200 rpm for 7 minutes. Remove the cell supernatant and prepare a standard curve according to the instructions provided in the kit. Use the corresponding ELISA kit to measure the levels of pro-inflammatory cytokines TNF-α and IL-6 secreted by the cells. Follow the instructions provided in the kit for specific procedures. Use an ELISA reader to measure the OD450 values, and calculate the concentrations based on the standard curve. 2.12 Statistical Analysis Data were presented as mean ± SD. Statistical significance was assessed via Prism 8.0 software using one-way analysis of variance (ANOVA) or unpaired t-test. A P value < 0.05 indicated statistical significance. 3. Results and Discussion 3.1 Dataset Analysis According to Table 2 , the proportion of active compounds (inhibitors) across all subtypes within the Src kinase family is above 80%. Notably, the experiment did not employ any data augmentation techniques, as the goal was to maintain the original form of the compound molecules. The dataset's diversity in molecular structures is advantageous for establishing accurate molecular property prediction models [ 27 ], assessed through Scaled Shannon Entropy (SSE) to measure the structural diversity [ 28 – 30 ]. SSE values for all subtypes within the Src family were greater than 0.75, indicating significant structural diversity within the dataset constructed for this study. Additionally, the analysis of molecular weight (MV) and AlogP within the randomly partitioned training, validation, and test sets, as shown in Fig. 2 , demonstrated that the molecules used for modeling cover a wide chemical space. Table 2 SRC family kinase information Kinase subtypes UniProt ID Compouns Active Inactive Blk P51451 884 849 35 Fgr P09769 208 194 14 Fyn P06241 1251 1139 112 Hck P08631 642 604 38 Lck P06239 3230 2907 323 Lyn P07948 715 665 50 c-Src P12931 5086 4222 864 Yes P07947 549 489 60 3.2 Model Evaluation Results In the task of predicting molecular properties, we constructed four basic machine learning models—RF, SVM, KNN, and XGBoost—utilizing PubChem fingerprints. Additionally, we developed a multi-task MSSP model using two different modes—molecular fingerprint features and molecular graph structure features. Upon evaluating the performance of these models, we observed variations across different metrics. The RF model exhibited strong performance, achieving high accuracy, but with relatively lower recall and F1-Score. The SVM model displayed higher precision but had lower recall, resulting in decreased F1-Score and AUC. Results from the KNN model showed a balance between recall and precision, yet slightly underperformed in Precision and AUC. XGBoost demonstrated relatively balanced performance across all evaluation metrics but showed slightly lower recall and F1-Score. Compared to traditional machine learning models, the MSSP model showcased outstanding recall and AUC, reaching 0.912 and 0.975, respectively. This suggests that the MSSP model more effectively identified positive samples and exhibited excellent performance under the ROC curve. The MSSP model addressed the limitations of traditional virtual screening, reducing the occurrence of false-positive results, thereby enhancing the reliability and accuracy in molecular property prediction tasks. Table 3 Model evaluation results Algorithm ACC Recall Precision F1-Score AUC RF 0.965 0.654 0.981 0.785 0.950 SVM 0.960 0.590 1.000 0.742 0.927 KNN 0.962 0.654 0.944 0.773 0.919 XGBoost 0.970 0.718 0.967 0.824 0.947 MSSP 0.935 0.912 0.899 0.906 0.975 3.3 Setting Up and Using the Web Server An online platform named SRC-Predictor ( http://ilovemyhome.cc/ ) has been developed using the multi-task MSSP model to aid experts and researchers in discovering novel selective Src kinase inhibitors. As shown in Fig. 4 , users can input small molecule SMILES sequences or upload a CSV file containing multiple small molecule SMILES sequences. This allows the estimation of inhibitory activity for the interested Src subtype(s) or all subtypes. 3.4 Molecular Docking Results and Final Selected Hits Through molecular docking simulations, the experiment assessed the binding effects between Lyn kinase and candidate compounds (Crizotinib, Doramapimod, Neflamapimod, Tozasertib), alongside the control group Imatinib, as detailed in Supplementary Files S1. Figure 5 illustrates the binding free energies obtained from the molecular docking results. The negative values of the binding free energies indicate stable binding between these drugs and Lyn kinase. The analysis indicates that Neflamapimod exhibited the lowest binding free energy, suggesting its potential as a Lyn kinase inhibitor. Further experiments and research could validate this finding, providing strong leads for potential drug development. The binding mode of BIRB 796 with LYN was predicted based on AutoDock Vina algorithm. The results show that the affinity of BIRB 796 for the docking of LYN molecules is -10 kcal/mol. BIRB 796 can bind to the LYN protein complex with amino acids LYS, GLU, LEU, ASN, SER, and TRP, and it binds to amino acids in the target protein in a variety of ways, including Pi-Cation, Pi-Pi Stacked, Amide-Pi Stacked, pi-alky, Alky1, and Pi-Pi T-shape, There's also the Van der Waals' force and the hydrogen bond. Amino acid residues including GLU and TYR all bind to LYN in the form of hydrogen bonds(Fig. 6a,b). In summary, BIRB 796 can bind to receptor protein tyrosine kinase when inflammation occurs, thus preventing the further transmission of inflammatory signals. Figure 6. Predicted binding mode of BIRB 796 with LYN. The binding mode was predicted based on Autodock vina algorithm and the figure was generated using Discovery Studio software ( http://www.discoverystudio.net/).(A ) Visualization of 2D structures in molecular docking. (B) Visualization of 3D structures in molecular docking. BIRB 796 can bind to theLYN protein complex with amino acids LYS, GLU, LEU, ASN, SER, and TRP. Hydrogen bonds are depicted by purple dashes;Electrostatic Electrostatic forces are depicted by blue dashes and Hydrophobic forces are depicted by red dashes. 3.5 Molecular Dynamics Simulation Ligand Root Mean Square Deviation (RMSD): This indicates the stability of the drug within the binding complex, where a smaller value suggests greater stability. Ligand RMSD values below 5 angstroms (Å) are generally considered indicative of stable binding. Molecular dynamics simulations revealed that, compared to the reference compound, Doramapimod, Neflamapimod, and Tozasertib exhibited superior binding stability to Imatinib, as shown in Fig. 7 (a). Alpha-helix is a crucial secondary structure unit and forms the foundation of a protein's tertiary structure, playing a pivotal role in the protein's overall conformation and function. The α-helix structure possesses a certain rigidity, providing structural support to the protein. All drugs induced alterations in the α-helix conformation of the protein (RMSF), primarily concentrated around residue 50 of the protein. Doramapimod exhibited better results than Imatinib at residues 50, 200, and 400, as depicted in Fig. 7 (b). Figure 7. (a) RMSD plots for the repurposed drugs/LYN and the reference drug (control); (b) RMSF plots depicting the repurposed drugs/LYN and the reference drug. 3.6 Effect of Doramapimod on Cell Viability We investigated the potential cytotoxicity of BIRB 796 on RAW264.7 cells using the CCK-8 assay to determine the optimal safe dosage of the drug. The results showed that, compared to the control group, the cell viability was > 80% when different concentrations of BIRB 796 were added to RAW264.7 cells, indicating that BIRB 796 treatment for 24 hours was non-toxic to macrophages at concentrations below 200 nM (Fig. 8 ). Therefore, the optimal concentration of BIRB 796 for subsequent experiments was determined to be 200 nM. 3.7 BIRB 796 inhibited the expression of RAW264.7 surface molecules induced by LPS M1 macrophages are macrophages that can produce pro-inflammatory cytokines. They express high levels of major histocompatibility complex II (MHC-II) and co-stimulatory molecules CD80 and CD86. Under the stimulation of LPS, M0 type quiescent macrophages can be polarized into M1 macrophages, which is marked by an increase in surface MHC-II, CD80, and CD86, enhancing cell antigen presentation capacity and initiating an immune response. Therefore, we treated RAW264.7 cells with different concentrations of BIRB 796 (25, 50, 100, or 200 nM) in combination with LPS. We then used flow cytometry to measure the fluorescence intensity of surface molecules CD80 and CD86 to investigate the effects of BIRB 796 on LPS-induced CD80 and CD86 expression. The results, as shown in Fig. 9 , demonstrated that treatment with 200 nM BIRB 796 significantly inhibited the expression of CD80 and CD86 induced by LPS compared to the LPS group. This suggests that BIRB 796 can inhibit the differentiation of M0 macrophages into pro-inflammatory M1 macrophages and suppresses the antigen-presenting capacity of the cells. 3.8 BIRB 796 Inhibited the Production of Proinflammatory cytokines in LPS-induced RAW264.7 Macrophages Under the stimulation of LPS, macrophages can become over-activated and differentiate into M1 macrophages, leading to an increase in the secretion of pro-inflammatory cytokines. The results in Fig. 10 showed that compared to the control group, the LPS-treated group significantly upregulated the production of TNF-α and IL-6. Treatment with the drug and LPS simultaneously significantly inhibited the secretion of the inflammatory cytokine TNF-α in a concentration-dependent manner, and also had an inhibitory effect on the secretion of IL-6, indicating that the drug can inhibit the differentiation of macrophages into M1 macrophages and suppress the secretion of pro-inflammatory cytokines induced by LPS, thereby inhibiting the inflammatory response. 4. Conclusion Macrophages, as the first line of innate immune response, engulf external antigens through various receptors, secreting multiple inflammatory factors, presenting antigens to immune cells, and triggering subsequent immune responses. However, in autoimmune diseases, the excessive activation of macrophages due to the engulfment and presentation of self-antigens leads to heightened autoimmune reactions, increased secretion of autoantibodies, pathological changes in multiple tissues, ultimately resulting in organ damage. Lyn kinase, a tyrosine kinase, plays a significant role in macrophage polarization. Lyn kinase promotes the secretion of inflammatory factors and cytotoxicity in M1-type macrophages. The study compiled data on all Src family kinase subtypes and introduced a multi-task MSSP model to forecast molecule inhibitory activity across multiple Src subtypes. By combining molecular docking virtual screening and experimental validation, the lead compound Doramapimod was identified to significantly inhibit Lyn kinase activity. This compound affects the secretion of inflammatory factors in LPS-induced macrophages, suppresses the differentiation of M1-type macrophages, promotes M2-type macrophage differentiation, and elevates anti-inflammatory factor levels. Declarations Ethical Statements: It is not applicable. Funding: Our work is supported by the Major scientific and technological special projects in Xinjiang Uygur Autonomous Region (2022A03016-4). Data availability Data will be provided upon request. Conflicts of interest The authors declare that they have no conflicts of interest. Author Contribution Xuecong Tian: Research design and conceptualization, data analysis, manuscript writing and revision.Luyang Han: Experimental design and implementation, data collection, figure preparation.Ying Su: Literature review, experimental support, data validation.Haiqing Sun: Experimental methods development, data analysis, drafting the manuscript.Sizhe Zhang: Data processing and analysis, results discussion, technical support.Chen Chen: Experimental materials preparation, data collection, research management.Cheng Chen: Data organization, statistical analysis, manuscript revision.Chen-xi Li: Data validation and replication, results discussion, manuscript proofreading.Xiaoyi Lv: Project management, research supervision, final manuscript review.Jinyao Li: Research guidance and oversight, funding acquisition, final manuscript editing and approval. References McClendon CJ, Miller WT (2020) Structure, function, and regulation of the SRMS tyrosine kinase. Int J Mol Sci 21(12):4233 Espada J, Martín-Pérez J (2017) An update on Src family of nonreceptor tyrosine kinases biology. Int Rev cell Mol biology 331:83–122 Martellucci S, Clementi L, Sabetta S, Mattei V, Botta L, Angelucci A (2020) Src family kinases as therapeutic targets in advanced solid tumors: what we have learned so far, Cancers , vol. 12, no. 6, p. 1448 Köhr G, Seeburg PH (1996) Subtype-specific regulation of recombinant NMDA receptor‐channels by protein tyrosine kinases of the src family. J Physiol 492(2):445–452 Hiscox S, Nicholson RI (2008) Src inhibitors in breast cancer therapy. Expert Opin Ther Targets 12(6):757–767 Palacios EH, Weiss A (2004) Function of the Src-family kinases, Lck and Fyn, in T-cell development and activation, Oncogene , vol. 23, no. 48, pp. 7990–8000 Nie L, Ye W-R, Chen S, Chirchiglia D, Wang M (2021) Src family kinases in the central nervous system: their emerging role in pathophysiology of migraine and neuropathic pain. Curr Neuropharmacol 19(5):665–678 Brian BF 4th and, Freedman TS (2021) The Src-family kinase Lyn in immunoreceptor signaling, Endocrinology , vol. 162, no. 10, p. bqab152 de Jesus AA et al (2023) Constitutively active Lyn kinase causes a cutaneous small vessel vasculitis and liver fibrosis syndrome. Nat Commun 14(1):1502 Congleton J, MacDonald R, Yen A (2012) Src inhibitors, PP2 and dasatinib, increase retinoic acid-induced association of Lyn and c-Raf (S259) and enhance MAPK-dependent differentiation of myeloid leukemia cells, Leukemia , vol. 26, no. 6, pp. 1180–1188 Dumas P-Y et al (2017) Hematopoietic niche drives FLT3-ITD acute myeloid leukemia resistance to quizartinib via STAT5-and hypoxia-dependent upregulation of AXL, Haematologica , vol. 104, no. 10, p. 2019 Cao ZX et al (2020) Erlotinib is effective against FLT3-ITD mutant AML and helps to overcome intratumoral heterogeneity via targeting FLT3 and Lyn. FASEB J 34(8):10182–10190 Zhang L et al (2017) Discovery of a small molecule targeting ULK1-modulated cell death of triple negative breast cancer in vitro and in vivo. Chem Sci 8(4):2687–2701 Moy B et al (2014) Bosutinib in combination with the aromatase inhibitor letrozole: A phase II trial in postmenopausal women evaluating first-line endocrine therapy in locally advanced or metastatic hormone receptor‐positive/HER2‐negative breast cancer. Oncologist 19(4):348–349 Daud AI et al (2012) Phase I study of bosutinib, a src/abl tyrosine kinase inhibitor, administered to patients with advanced solid tumors. Clin Cancer Res 18(4):1092–1100 Ward RA, Fawell S, Floc’h N, Flemington V, McKerrecher D, Smith PD (2020) Challenges and opportunities in cancer drug resistance. Chem Rev 121(6):3297–3351 Rossari F, Minutolo F, Orciuolo E (2018) Past, present, and future of Bcr-Abl inhibitors: from chemical development to clinical efficacy. J Hematol Oncol 11(1):1–14 Ferguson FM, Gray NS (2018) Kinase inhibitors: the road ahead. Nat Rev Drug Discovery 17(5):353–377 Cichońska A et al (2021) Crowdsourced mapping of unexplored target space of kinase inhibitors. Nat Commun 12(1):3307 Swain SS, Hussain T (2021) Role of bioinformatics in early drug discovery: an overview and perspective. Comput BioInformatics: Multidisciplinary Appl, pp. 49–67 Koneru JK, Sinha S, Mondal J (2019) In Silico reoptimization of binding affinity and drug-resistance circumvention ability in kinase inhibitors: a case study with RL-45 and Src kinase. J Phys Chem B 123(31):6664–6672 Yu W, Weber DJ, Shapiro P, MacKerell AD (2020) Developing Kinase Inhibitors Using Computer-Aided Drug Design Approaches, Next Generation Kinase Inhibitors: Moving Beyond the ATP Binding/Catalytic Sites , pp. 81–108 Zhang Y, Zhang T-j, Tu S, Zhang Z-h, Meng F-h (2020) Identification of novel Src inhibitors: Pharmacophore-based virtual screening, molecular docking and molecular dynamics simulations, Molecules , vol. 25, no. 18, p. 4094 Zhang L, Tan J, Han D, Zhu H (2017) From machine learning to deep learning: progress in machine intelligence for rational drug discovery. Drug Discovery Today 22(11):1680–1685 Kim S et al (2019) PubChem 2019 update: improved access to chemical data. Nucleic Acids Res, 47, no. D1, pp. D1102-D1109 Gaulton A et al (2012) ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res, 40, no. D1, pp. D1100-D1107 Wang L et al (2016) Chemical fragment-based CDK4/6 inhibitors prediction and web server. RSC Adv 6(21):16972–16981 Yongye AB, Waddell J, Medina-Franco JL (2012) Molecular scaffold analysis of natural products databases in the public domain. Chem Biol Drug Des 80(5):717–724 Gregori-Puigjané E, Mestres J (2006) Shannon entropy descriptors from topological feature distributions. J Chem Inf Model 46(4):1615–1622 Medina-Franco JL, Martínez‐Mayorga K, Bender A, Scior T (2009) Scaffold diversity analysis of compound data sets using an entropy‐based measure, QSAR & Combinatorial Science , vol. 28, no. 11‐12, pp. 1551–1560 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4475200","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":321998071,"identity":"0f8376f3-161e-429f-b160-5c233cfb22d6","order_by":0,"name":"Xuecong Tian","email":"","orcid":"","institution":"Xinjiang University","correspondingAuthor":false,"prefix":"","firstName":"Xuecong","middleName":"","lastName":"Tian","suffix":""},{"id":321998072,"identity":"e9917b10-ab83-4009-ad2b-80274f4c26db","order_by":1,"name":"Luyang Han","email":"","orcid":"","institution":"Xinjiang University","correspondingAuthor":false,"prefix":"","firstName":"Luyang","middleName":"","lastName":"Han","suffix":""},{"id":321998073,"identity":"1dc07d27-a0fd-4b19-a72d-b3c549e225b1","order_by":2,"name":"Ying Su","email":"","orcid":"","institution":"Xinjiang University","correspondingAuthor":false,"prefix":"","firstName":"Ying","middleName":"","lastName":"Su","suffix":""},{"id":321998074,"identity":"a0eb67f1-35ca-48f7-974c-b5318089ed14","order_by":3,"name":"Haiqing Sun","email":"","orcid":"","institution":"Xinjiang University","correspondingAuthor":false,"prefix":"","firstName":"Haiqing","middleName":"","lastName":"Sun","suffix":""},{"id":321998078,"identity":"2e320886-68eb-4b89-92e4-33a44ae14d4b","order_by":4,"name":"Sizhe Zhang","email":"","orcid":"","institution":"Xinjiang University","correspondingAuthor":false,"prefix":"","firstName":"Sizhe","middleName":"","lastName":"Zhang","suffix":""},{"id":321998086,"identity":"741d9d0c-677f-43fe-aadc-cf2a2e5fe11d","order_by":5,"name":"Chen Chen","email":"","orcid":"","institution":"Xinjiang University","correspondingAuthor":false,"prefix":"","firstName":"Chen","middleName":"","lastName":"Chen","suffix":""},{"id":321998089,"identity":"37b53737-326f-4214-9458-855188e45bee","order_by":6,"name":"Cheng Chen","email":"","orcid":"","institution":"Xinjiang University","correspondingAuthor":false,"prefix":"","firstName":"Cheng","middleName":"","lastName":"Chen","suffix":""},{"id":321998094,"identity":"0deb7cfb-1a4e-4371-a257-ce648ac37b2d","order_by":7,"name":"Chen xi Li","email":"","orcid":"","institution":"the First Affiliated Hospital of Xinjiang Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chen","middleName":"xi","lastName":"Li","suffix":""},{"id":321998097,"identity":"730c5c6b-b6e3-4743-b7c0-250f56590749","order_by":8,"name":"Xiaoyi Lv","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzUlEQVRIiWNgGAWjYFCCAyCCjYeNvbHx4QdStMjx8RxuNpYgxS5jOYn0NgEeYpQaHDxjJvGzjS+xTfJhG4MEg52cbgMBLZINZ8wke9vYEtukE9seFDAkG5sdIKCFn+GM2Q3ebWAt7QYSDAcStxHSwgbUcvMvSIvkwTYJHmK0gGy5DbTFmE2CkUgtkg3Hyn/L/mOTY+NJBAayARF+MbhxeLPhmzPHeOTbjz98+KHCTo6gFgYJsIpjMBMIKQcB/gYQWUOM0lEwCkbBKBipAABCmEHN2C0eDAAAAABJRU5ErkJggg==","orcid":"","institution":"Xinjiang University","correspondingAuthor":true,"prefix":"","firstName":"Xiaoyi","middleName":"","lastName":"Lv","suffix":""},{"id":321998098,"identity":"76a1fb4d-5574-4d69-aa7d-55d028052c5a","order_by":9,"name":"Jinyao Li","email":"","orcid":"","institution":"Xinjiang University","correspondingAuthor":false,"prefix":"","firstName":"Jinyao","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2024-05-25 05:23:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4475200/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4475200/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":60435805,"identity":"1f045f82-b255-4d85-9d2d-4fdf9c9d3efa","added_by":"auto","created_at":"2024-07-16 17:26:09","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":253150,"visible":true,"origin":"","legend":"\u003cp\u003eExperimental flow chart\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/6c4b0bea5d3483af368f051a.png"},{"id":60435806,"identity":"c2f1274a-fdb3-48df-bd39-2ff1d4622478","added_by":"auto","created_at":"2024-07-16 17:26:09","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":114907,"visible":true,"origin":"","legend":"\u003cp\u003e(a) Depicts the SSE results. SSE measures the diversity of molecular structures, with values ranging from 0 to 1; a value closer to 1 indicates greater structural diversity. (b) The molecular weight (MV) and AlogP of the randomly partitioned training, validation, and test sets are illustrated.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/37b7a8146f649e35f9d0b36e.png"},{"id":60435810,"identity":"4963c6cd-7d51-4e2d-a4fa-275eae1a9c75","added_by":"auto","created_at":"2024-07-16 17:26:09","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":56345,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves of four baseline machine learning algorithms and multi-task MSSP model.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/ee0d415b330ac815498799e1.png"},{"id":60435807,"identity":"3700f7ce-6f4a-4cb0-9a2d-9a3511645b31","added_by":"auto","created_at":"2024-07-16 17:26:09","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":180381,"visible":true,"origin":"","legend":"\u003cp\u003eSRC-Predictor online platform. The platform utilizes the SpringBoot framework and follows a front-end and back-end separation design principle, capable of handling a certain level of concurrency.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/007d2789f4c848ed136b52ec.png"},{"id":60437412,"identity":"ad6a45dc-be9a-4626-af1a-c75caa9f7e16","added_by":"auto","created_at":"2024-07-16 17:42:09","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":30320,"visible":true,"origin":"","legend":"\u003cp\u003eCandidate compounds (Crizotinib, Doramapimod, Neflamapimod, Tozasertib) and molecular docking results of Imatinib and Lyn in control group.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/d44b848ae870a6045fb29ceb.png"},{"id":60435811,"identity":"451a7b04-40c0-4347-aa19-b0edc048a90f","added_by":"auto","created_at":"2024-07-16 17:26:09","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":195805,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted binding mode of BIRB 796 with LYN. The binding mode was predicted based on Autodock vina algorithm and the figure was generated using Discovery Studio software (http://www.discoverystudio.net/).(A) Visualization of 2D structures in molecular docking. (B) Visualization of 3D structures in molecular docking. BIRB 796 can bind to theLYN protein complex with amino acids LYS, GLU, LEU, ASN, SER, and TRP. Hydrogen bonds are depicted by purple dashes；Electrostatic Electrostatic forces are depicted by blue dashes and Hydrophobic forces are depicted by red dashes.\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/5f388ae037de2054d9cad7cf.png"},{"id":60435813,"identity":"4fc85a80-8956-43af-8fce-47306f3ae348","added_by":"auto","created_at":"2024-07-16 17:26:09","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":284727,"visible":true,"origin":"","legend":"\u003cp\u003e(a) RMSD plots for the repurposed drugs/LYN and the reference drug (control); (b) RMSF plots depicting the repurposed drugs/LYN and the reference drug.\u003c/p\u003e","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/fe4a4d8376d66d1397571210.png"},{"id":60436498,"identity":"a1bdf5db-3ca8-4078-8966-b1460d01ba2b","added_by":"auto","created_at":"2024-07-16 17:34:09","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":14789,"visible":true,"origin":"","legend":"\u003cp\u003eEffect of BIRB 796 on the viability of RAW264.7 cells. Cell viability was assessed after treatment with BIRB 796 (200, 400,800nM or 10，25，50 μM) and stimulated with LPS for 24 h. Data are expressed as mean ± SD from three independent experiments. ###P \u0026lt; 0.001 compared to untreated group. *P \u0026lt; 0.05, ***P \u0026lt; 0.001 compared to LPS group.\u003c/p\u003e","description":"","filename":"floatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/7122587de97e5c4787636b98.png"},{"id":60436496,"identity":"b371dcc2-24aa-401e-b6b2-e404f7c78d74","added_by":"auto","created_at":"2024-07-16 17:34:09","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":83355,"visible":true,"origin":"","legend":"\u003cp\u003eBIRB 796 inhibited the expression of RAW264.7 surface molecules. After RAW164.7 was treated with BIRB 796 (25, 50,100 or 200nM) and and stimulated with LPS for 12 h, the fluorescence intensity of the molecules on the cell surface was detected by flow cytometry. Data are expressed as mean ± SD from three independent experiments. ###P \u0026lt; 0.001 compared to untreated group. *P \u0026lt; 0.05, ***P \u0026lt; 0.001 compared to LPS group.\u003c/p\u003e","description":"","filename":"floatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/da20872875cdfc70ea77f92a.png"},{"id":60435814,"identity":"7255e9bf-0d35-4e34-b573-e8d79aaff4d5","added_by":"auto","created_at":"2024-07-16 17:26:09","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":19957,"visible":true,"origin":"","legend":"\u003cp\u003eBIRB 796 Inhibited the Production of Proinflammatory cytokines in LPS-induced RAW264.7 Macrophages. After RAW164.7 was treated with BIRB 796 (50,100 or 200nM) and and stimulated with LPS for 12 h. The contents of pro-inflammatory cytokines TNF-A and IL-6 secreted by RAW264.7 were detected by ELISA Data are expressed as mean ± SD from three independent experiments. ###P \u0026lt; 0.001 compared to untreated group. *P \u0026lt; 0.05, ***P \u0026lt; 0.001 compared to LPS group.\u003c/p\u003e","description":"","filename":"floatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/6dd3c35749aee69dd1091b37.png"},{"id":87311314,"identity":"8fdb72b1-9656-4667-9f56-10296e46aa18","added_by":"auto","created_at":"2025-07-22 14:54:01","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2180109,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4475200/v1/378662e6-872a-41a5-aa66-ee27ccce972c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Multi-task MSSP model can accurately predict selective Src inhibitors","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eSrc family kinases (SFKs), a subset of non-receptor tyrosine kinases, serve as crucial signaling intermediates in metazoans [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. They are activated by various growth factors, cytokines, and antigen receptors, thereby regulating crucial biological processes like proliferation, differentiation, apoptosis, migration, and metabolism [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Overexpression or aberrant activation of Src family kinases in both epithelial and non-epithelial cancers significantly contributes to tumorigenesis [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. c-Src stands as one of the earliest identified and extensively studied members among the Src family kinases [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Consequently, most inhibitors targeting Src family kinases are designed to selectively target the c-Src subtype. These inhibitors function by binding to the active site of the c-Src subtype, impeding its activity and disrupting cancer-related processes like cell proliferation, migration, and invasion. They competitively interact with the substrate binding site or ATP binding site of the c-Src subtype, thereby impeding the binding of substrates or ATP, ultimately inhibiting the enzyme's catalytic activity [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Additionally, Src family kinase subtypes like Fyn, Yes, Lyn, Lck, and Hck are considered potential therapeutic targets. These subtypes are pivotal in cellular signal transduction and are linked to the onset of diverse diseases, encompassing cancer, immune disorders, and neurological conditions [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. While acknowledging the importance of these subtypes in diseases, the development of precise inhibitors remains limited, and research in this domain is in its early stages [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. The ongoing development of selective inhibitors targeting different subtypes holds substantial importance in offering more targeted and selective therapeutic strategies.\u003c/p\u003e \u003cp\u003eAmong the Src family, Lyn, c-Src, and Lck are among the most easily activated tyrosine kinases in various types of cancer cells [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Recent reports highlight the critical role of Lyn kinase in modulating inflammatory signaling, microvascular permeability, neutrophil recruitment, and facilitating hepatic fibrosis [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Additionally, Lyn kinase can induce B-cell leukemia and granulocytic leukemia by phosphorylating the proto-oncogene protein Cbl [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Further analysis of acute myeloid leukemia (AML) cells revealed a significant increase in Lyn kinase activity in 76% of the cells. This elevation led to an increase in the transcription factor STAT5, consequently promoting disease progression [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Notably, inhibiting Lyn kinase activity in AML cell lines significantly reduces cell growth [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Moreover, numerous studies suggest that Lyn's overactivation contributes to the occurrence of various solid tumors, including breast cancer [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e], colon cancer [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e], and prostate cancer [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Therefore, inhibitors targeting the Src family, particularly Lyn, represent a promising strategy for targeting hematologic malignancies and several solid tumor diseases [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eNumerous inhibitors targeting the Src family have been developed for the treatment of both hematologic disorders and solid malignancies [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Dasatinib, the first FDA-approved drug with confirmed Src family kinase inhibition, is primarily used to treat hematologic disorders such as chronic myeloid leukemia (CML) and AML [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Bosutinib, known as a Lyn kinase inhibitor, is primarily employed in the treatment of malignant hematologic disorders [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. The diversity of kinase inhibitors enables the selection of different drugs based on a patient's resistance profile, thereby extending the duration of treatment [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. For instance, Bosutinib, Dasatinib, and Ponatinib effectively overcome BCR-ABL resistance through distinct mechanisms and broader inhibitory capabilities, offering new therapeutic options for patients resistant to Imatinib [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. However, only a handful of drugs have gained FDA approval, and achieving selective inhibition of Src kinases without interfering with other related signaling pathways remains a challenge [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Src family kinase inhibitors represent a significant area in cancer therapy with vast research prospects. It's worthwhile to persist in the ongoing development of potential inhibitors for treating diseases.\u003c/p\u003e \u003cp\u003eThe discovery of kinase inhibitors encompasses high-throughput screening, compound optimization, biological evaluation, and clinical trials, covering multiple stages from chemical database screening to drug approval [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. However, traditional experimental assay methods are often expensive, time-consuming, and constrained by limited parameter space [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. In recent years, Computer-Aided Drug Design (CADD) has played a pivotal role in the early stages of drug discovery, aligning computational and biological/chemical knowledge [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. CADD methods have been widely employed in the discovery and design of Src family kinase inhibitors. In 2019, Koneru et al. employed QSAR modeling and molecular dynamics to redesign the second-generation Src kinase inhibitor RL-45. The newly designed compounds mitigated mutation-associated Src kinase resistance and exhibited enhanced binding to the kinase's active site [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. In 2020, Yu et al. employed virtual docking programs to screen 2\u0026nbsp;million molecules in databases, aiming to identify small molecule binders targeting the pY\u0026thinsp;+\u0026thinsp;3 site of the Lck kinase SH2 domain within the Src family [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Similarly, in 2020, Zhang et al. discovered potential lead compounds as anti-Src kinase agents by integrating virtual screening of pharmacophores, molecular docking, and dynamics simulations [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. These methods significantly aid in drug development and the analysis of large biomedical datasets in cancer therapy. Furthermore, deep learning, a critical branch of artificial intelligence, automates the learning of advanced feature representations from raw data, optimizing the identification of lead compounds and molecular understanding of diseases [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn this study, an interpretable model based on the multitask MSSP deep learning (DL) framework was constructed for virtual screening of active molecules against multiple SRC subtypes. Using the MSSP model, an online platform (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://ilovemyhome.cc/\u003c/span\u003e\u003cspan address=\"http://ilovemyhome.cc/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was created to aid in identifying and adjusting selective SRC inhibitors. Eventually, using the MSSP model, a library of small molecules with the potential to inhibit various Src-related pathways was screened. Four compounds, most likely to act as Lyn kinase inhibitors within the Src family, were further subjected to molecular docking and molecular dynamics simulations to identify lead compounds. Targeting RAW264.7 cells, these lead compounds were explored for their alleviating effects on inflammation.\u003c/p\u003e \u003cp\u003eThe lead compounds proposed in this study may pave the path toward creating novel precision medicines for inflammation therapy.\u003c/p\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Materials\u003c/h2\u003e \u003cdiv id=\"Sec4\" class=\"Section3\"\u003e \u003ch2\u003e2.1.1 Experimental Cells\u003c/h2\u003e \u003cp\u003eThe murine monocyte/macrophage cell line RAW264.7 was provided by our laboratory.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section3\"\u003e \u003ch2\u003e2.1.2 Experimental Drugs\u003c/h2\u003e \u003cp\u003eDoramapimod (BIRB 796) was purchased from Shanghai YuanYe Biotechnology Co., Ltd. Prior to experimentation, BIRB 796 was dissolved in dimethyl sulfoxide (DMSO)\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.1.3 Experimental Reagents\u003c/h2\u003e \u003cp\u003ePhosphate-buffered saline (PBS), DMEM cell culture medium, and fetal bovine serum (FBS) were purchased from Gibco (Gaithersburg, Maryland, USA). Fluorescent-labeled antibodies for CD86-PE and CD80-APC were procured from Elabscience (China). Cell Counting Kit-8 (CCK-8) was obtained from Shanghai Beyotime Biotechnology Co., Ltd. Enzyme-linked immunosorbent assay (ELISA) kits for IL-6 and TNF-α were purchased from Elabscience (China). Lipopolysaccharide (LPS) and dimethyl sulfoxide (DMSO) were acquired from Sigma-Aldrich (St. Louis, Missouri, USA).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.1.4 Experimental Instruments\u003c/h2\u003e \u003cp\u003eCO2 cell incubator, low temperature and high speed refrigerated centrifuge (Thermo Fisher Scientific, USA); Flow cytometry (BD FACSCalibur, USA); Enzyme labeling instrument (Bio-Rad Company); Low speed centrifuge (Sichuan Shuke Instrument Co., Ltd.).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Data Collection and Preparation\u003c/h2\u003e \u003cp\u003eIn this study, the dataset containing compounds related to the 8 subtypes of the Src kinase family was sourced from PubChem [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] and ChEMBL [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Ultimately, a total of 7935 compounds relevant to the 8 subtypes of the Src kinase family, covering 12001 bioassay data, were compiled. The processed dataset was randomly partitioned into a training set (80%), a validation set (10%), and a test set (10%). The training and validation sets were utilized for model development and hyperparameter optimization, while the test set was employed to evaluate the established model's performance. The collected data underwent filtration based on Lipinski's Rule of Five and Veber's Rules: removal of data lacking bioassay values or explicit molecular properties, normalization of biological activity units, calculation of average bioassay values as the final value, exclusion of duplicate molecules and those with a molecular weight exceeding 1000 Da, and standardization of each compound in the dataset.The dataset generated was utilized to train the multitask MSSP model, then compared with baseline models like Random Forest (RF), Support Vector Machine (SVM), k-Nearest Neighbors (KNN), and XGBoost. An overview of the entire process is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Multi-task MSSP Deep Learning Framework\u003c/h2\u003e \u003cp\u003eThis experiment established a multitask MSSP model to accurately predict active molecules targeting the Src kinase family. Innovative design was employed to address two distinct feature modalities\u0026mdash;molecular fingerprint features and molecular graph structure features\u0026mdash;by introducing a novel molecule fusion algorithm. This algorithm mapped different modalities of molecules into shared and molecule-specific representation spaces, utilizing feature similarity to align and merge diverse modalities effectively. The experiment utilized parameter sharing in multitask learning to address shortcomings of traditional machine learning methods in handling correlated subtasks and complex features. To enhance interpretability, the experiment incorporated a graph attention mechanism, quantitatively representing the importance of chemical fragments in predicting molecular properties. The multitask MSSP model was trained on a NVIDIA GeForce RTX 3090 GPU.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Baseline Machine Learning Algorithms\u003c/h2\u003e \u003cp\u003eTo further validate the effectiveness of the multitask MSSP model in predicting SRC kinase inhibitors, we conducted experiments and compared it with four traditional machine learning algorithms. These four algorithms include Random Forest (RF), Support Vector Machine (SVM), k-Nearest Neighbors (KNN), and eXtreme Gradient Boosting (XGBoost). These algorithms are widely applied classical methods in the fields of chemistry and biology. The RF, SVM, and KNN models were developed using the scikit-learn Python package (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/scikit-learn/scikit-learn\u003c/span\u003e\u003cspan address=\"https://github.com/scikit-learn/scikit-learn\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, version: 1.2.1); the XGBoost model was developed using the XGBoost Python package (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/dmlc/xgboost\u003c/span\u003e\u003cspan address=\"https://github.com/dmlc/xgboost\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e, version: 1.7.4).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Model Performance Evaluation\u003c/h2\u003e \u003cp\u003eAfter establishing the classification model using the training set, the performance was evaluated using the test set. This study employed Accuracy, Recall, Precision, and F1-Score as the model evaluation metrics. The metrics values were computed based on Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e using specific formulas (1)-(4).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"No\" id=\"Taba\" border=\"1\"\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(Accuracy=\\frac{TP+TN}{TP+FP+TN+FN}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(1)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(Recall=\\frac{TP}{TP+FN}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(Precision=\\frac{TP}{TP+FP}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(F1-Score=\\frac{2*precision*recall}{precision+recall}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e(4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn addition, we also utilized the area under the receiver operating characteristic curve (AUC) as a comprehensive evaluation metric for the model.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eConfusion matrix\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePredicted \u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePatients with glioma\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNormal subjects\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePatients with glioma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFN\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNormal subjects\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTN\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTN\u0026thinsp;=\u0026thinsp;true negative, TP\u0026thinsp;=\u0026thinsp;true positive, FN\u0026thinsp;=\u0026thinsp;false negative, FP\u0026thinsp;=\u0026thinsp;false positive, TPR\u0026thinsp;=\u0026thinsp;true positive rate, and FPR\u0026thinsp;=\u0026thinsp;false positive rate.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e2.6 Molecular Docking\u003c/h2\u003e \u003cp\u003eBy utilizing MSSP to predict the collected dataset related to Src kinases, we selected the top twenty small molecule compounds based on the model's predicted scores. To further validate whether these molecules selected through MSSP screening possess SRC kinase inhibitory activity, a virtual screening experiment was conducted. Following literature review to identify small molecule compounds with less research pertaining to Src kinases, molecular docking simulations were performed using AutoDock Vina software to dock these compounds with the target protein. Lyn kinase was employed as the substrate, and small molecule compounds displaying higher affinity with Lyn kinase were screened and designated as lead compounds.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e2.7 Molecular Dynamics Simulations\u003c/h2\u003e \u003cp\u003eTo further investigate the potential lead compounds exhibiting Src subtype Lyn kinase inhibitory activity, we employed molecular dynamics simulations to acquire detailed molecular structure, dynamics, and stability information. Imatinib was selected as the control group, and using the System Builder tool in Maestro software, we constructed the physiological aqueous salt system of the ligand-Lyn kinase protein complex. Employing Maestro's molecular dynamics module, we conducted a 10 nanosecond dynamics simulation of the constructed system. Simulation parameters included a recording interval of 10 picoseconds, utilizing the NPT ensemble type, a temperature of 300K, and a pressure of 1.01325 bar. This experimental design aims to precisely explore the activity and stability of the lead compounds, providing robust support for further drug design endeavors.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e2.8 Cell Culture\u003c/h2\u003e \u003cp\u003eResuscitated RAW264.7 mouse monocyte/macrophage cell line from liquid nitrogen or ultra-low temperature freezer, seeded in cell culture dishes with DMEM high glucose completed medium containing 10% fetal bovine serum and 1% penicillin-streptomycin, and incubated at 37\u0026deg;C and 5% CO2 in a CO2 incubator. The medium is replaced every 24 hours. Passage can be performed when the cell confluence reaches 70%-80%. Cells in the logarithmic growth phase can be used for subsequent experiments.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e2.9 Cell Viability Assay\u003c/h2\u003e \u003cp\u003eCollect cells in the logarithmic growth phase and count them. Resuspend cells at a concentration of 1 \u0026times; 105 cells/mL and seed them into a 96-well plate. Set up a normal control group, a positive control group (LPS), and a drug treatment group (LPS\u0026thinsp;+\u0026thinsp;BIRB 796). Each group should have 6 replicates. Treat cells with different concentrations of BIRB 796 (200 nM, 400 nM, 800 nM, 10 \u0026micro;M, 25 \u0026micro;M, 50 \u0026micro;M) and LPS (10 ng/mL) simultaneously. Incubate the cells in a 37\u0026deg;C incubator for 24 h. Add 10 \u0026micro;L of CCK-8 reagent to each well and incubate for an additional 1 h. Measure the absorbance (A) of each well at a wavelength of 450 nm using a Microplate reader. Calculate the cell viability according to the formula.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e2.10 Flow Cytometry\u003c/h2\u003e \u003cp\u003eCollect RAW264.7 cells and seed them in a 24-well plate at a density of 1 \u0026times; 105 cells per well. Set up a normal control group, a positive control group (LPS), and a drug treatment group (LPS\u0026thinsp;+\u0026thinsp;BIRB 796). Each group should have 3 replicates. Treat the cells with different concentrations of BIRB 796 (25, 50, 100, 200 nM) in combination with LPS. Incubate the cells in a 37\u0026deg;C incubator for 12 hours, then collect the cells. Wash each sample twice with PBS and stain with APC-CD80 and PE-CD80 flow cytometry antibodies for 15 minutes. After washing with PBS, samples were detected by FACSCalibur, and data were analyzed by FlowJo software.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e2.11 ELISA Assays\u003c/h2\u003e \u003cp\u003eCollect RAW264.7 cells and seed them in a 24-well plate at a density of 1 \u0026times; 105 cells per well. Treat the cells with different concentrations of BIRB 796 (25, 50, 100, 200 nM) in combination with LPS. Incubate the cells in a 37\u0026deg;C incubator for 12 hours. After incubation, collect the cell suspension by centrifuging at 1200 rpm for 7 minutes. Remove the cell supernatant and prepare a standard curve according to the instructions provided in the kit. Use the corresponding ELISA kit to measure the levels of pro-inflammatory cytokines TNF-α and IL-6 secreted by the cells. Follow the instructions provided in the kit for specific procedures. Use an ELISA reader to measure the OD450 values, and calculate the concentrations based on the standard curve.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e2.12 Statistical Analysis\u003c/h2\u003e \u003cp\u003eData were presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD. Statistical significance was assessed via Prism 8.0 software using one-way analysis of variance (ANOVA) or unpaired t-test. A P value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 indicated statistical significance.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results and Discussion","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Dataset Analysis\u003c/h2\u003e \u003cp\u003eAccording to Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, the proportion of active compounds (inhibitors) across all subtypes within the Src kinase family is above 80%. Notably, the experiment did not employ any data augmentation techniques, as the goal was to maintain the original form of the compound molecules. The dataset's diversity in molecular structures is advantageous for establishing accurate molecular property prediction models [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], assessed through Scaled Shannon Entropy (SSE) to measure the structural diversity [\u003cspan additionalcitationids=\"CR29\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. SSE values for all subtypes within the Src family were greater than 0.75, indicating significant structural diversity within the dataset constructed for this study. Additionally, the analysis of molecular weight (MV) and AlogP within the randomly partitioned training, validation, and test sets, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, demonstrated that the molecules used for modeling cover a wide chemical space.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSRC family kinase information\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKinase subtypes\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUniProt ID\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCompouns\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eActive\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eInactive\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlk\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eP51451\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e884\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e849\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e35\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFgr\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eP09769\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e208\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e194\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFyn\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eP06241\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1251\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1139\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e112\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHck\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eP08631\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e642\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e604\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e38\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLck\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eP06239\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3230\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2907\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e323\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLyn\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eP07948\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e715\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e665\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ec-Src\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eP12931\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5086\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4222\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e864\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eP07947\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e549\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e489\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e60\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Model Evaluation Results\u003c/h2\u003e \u003cp\u003eIn the task of predicting molecular properties, we constructed four basic machine learning models\u0026mdash;RF, SVM, KNN, and XGBoost\u0026mdash;utilizing PubChem fingerprints. Additionally, we developed a multi-task MSSP model using two different modes\u0026mdash;molecular fingerprint features and molecular graph structure features. Upon evaluating the performance of these models, we observed variations across different metrics. The RF model exhibited strong performance, achieving high accuracy, but with relatively lower recall and F1-Score. The SVM model displayed higher precision but had lower recall, resulting in decreased F1-Score and AUC. Results from the KNN model showed a balance between recall and precision, yet slightly underperformed in Precision and AUC. XGBoost demonstrated relatively balanced performance across all evaluation metrics but showed slightly lower recall and F1-Score.\u003c/p\u003e \u003cp\u003eCompared to traditional machine learning models, the MSSP model showcased outstanding recall and AUC, reaching 0.912 and 0.975, respectively. This suggests that the MSSP model more effectively identified positive samples and exhibited excellent performance under the ROC curve. The MSSP model addressed the limitations of traditional virtual screening, reducing the occurrence of false-positive results, thereby enhancing the reliability and accuracy in molecular property prediction tasks.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel evaluation results\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAlgorithm\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eACC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRecall\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eF1-Score\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.965\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.654\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.981\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.785\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.950\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.960\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.590\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.742\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.927\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKNN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.962\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.654\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.944\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.773\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.919\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eXGBoost\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.970\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.718\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.967\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.824\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.947\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMSSP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.935\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.912\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.899\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.906\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.975\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Setting Up and Using the Web Server\u003c/h2\u003e \u003cp\u003eAn online platform named SRC-Predictor (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://ilovemyhome.cc/\u003c/span\u003e\u003cspan address=\"http://ilovemyhome.cc/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) has been developed using the multi-task MSSP model to aid experts and researchers in discovering novel selective Src kinase inhibitors. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, users can input small molecule SMILES sequences or upload a CSV file containing multiple small molecule SMILES sequences. This allows the estimation of inhibitory activity for the interested Src subtype(s) or all subtypes.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Molecular Docking Results and Final Selected Hits\u003c/h2\u003e \u003cp\u003eThrough molecular docking simulations, the experiment assessed the binding effects between Lyn kinase and candidate compounds (Crizotinib, Doramapimod, Neflamapimod, Tozasertib), alongside the control group Imatinib, as detailed in Supplementary Files S1. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e illustrates the binding free energies obtained from the molecular docking results.\u003c/p\u003e \u003cp\u003eThe negative values of the binding free energies indicate stable binding between these drugs and Lyn kinase. The analysis indicates that Neflamapimod exhibited the lowest binding free energy, suggesting its potential as a Lyn kinase inhibitor. Further experiments and research could validate this finding, providing strong leads for potential drug development.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe binding mode of BIRB 796 with LYN was predicted based on AutoDock Vina algorithm. The results show that the affinity of BIRB 796 for the docking of LYN molecules is -10 kcal/mol. BIRB 796 can bind to the LYN protein complex with amino acids LYS, GLU, LEU, ASN, SER, and TRP, and it binds to amino acids in the target protein in a variety of ways, including Pi-Cation, Pi-Pi Stacked, Amide-Pi Stacked, pi-alky, Alky1, and Pi-Pi T-shape, There's also the Van der Waals' force and the hydrogen bond. Amino acid residues including GLU and TYR all bind to LYN in the form of hydrogen bonds(Fig.\u0026nbsp;6a,b). In summary, BIRB 796 can bind to receptor protein tyrosine kinase when inflammation occurs, thus preventing the further transmission of inflammatory signals.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eFigure 6.\u003c/b\u003e Predicted binding mode of BIRB 796 with LYN. The binding mode was predicted based on Autodock vina algorithm and the figure was generated using Discovery Studio software (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.discoverystudio.net/).(A\u003c/span\u003e\u003cspan address=\"http://www.discoverystudio.net/).(A\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) Visualization of 2D structures in molecular docking. (B) Visualization of 3D structures in molecular docking. BIRB 796 can bind to theLYN protein complex with amino acids LYS, GLU, LEU, ASN, SER, and TRP. Hydrogen bonds are depicted by purple dashes;Electrostatic Electrostatic forces are depicted by blue dashes and Hydrophobic forces are depicted by red dashes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Molecular Dynamics Simulation\u003c/h2\u003e \u003cp\u003eLigand Root Mean Square Deviation (RMSD): This indicates the stability of the drug within the binding complex, where a smaller value suggests greater stability. Ligand RMSD values below 5 angstroms (\u0026Aring;) are generally considered indicative of stable binding. Molecular dynamics simulations revealed that, compared to the reference compound, Doramapimod, Neflamapimod, and Tozasertib exhibited superior binding stability to Imatinib, as shown in Fig.\u0026nbsp;7 (a).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAlpha-helix is a crucial secondary structure unit and forms the foundation of a protein's tertiary structure, playing a pivotal role in the protein's overall conformation and function. The α-helix structure possesses a certain rigidity, providing structural support to the protein. All drugs induced alterations in the α-helix conformation of the protein (RMSF), primarily concentrated around residue 50 of the protein. Doramapimod exhibited better results than Imatinib at residues 50, 200, and 400, as depicted in Fig.\u0026nbsp;7 (b).\u003c/p\u003e \u003cp\u003e \u003cb\u003eFigure 7.\u003c/b\u003e (a) RMSD plots for the repurposed drugs/LYN and the reference drug (control); (b) RMSF plots depicting the repurposed drugs/LYN and the reference drug.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e3.6 Effect of Doramapimod on Cell Viability\u003c/h2\u003e \u003cp\u003eWe investigated the potential cytotoxicity of BIRB 796 on RAW264.7 cells using the CCK-8 assay to determine the optimal safe dosage of the drug. The results showed that, compared to the control group, the cell viability was \u0026gt;\u0026thinsp;80% when different concentrations of BIRB 796 were added to RAW264.7 cells, indicating that BIRB 796 treatment for 24 hours was non-toxic to macrophages at concentrations below 200 nM (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e8\u003c/span\u003e). Therefore, the optimal concentration of BIRB 796 for subsequent experiments was determined to be 200 nM.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003e3.7 BIRB 796 inhibited the expression of RAW264.7 surface molecules induced by LPS\u003c/h2\u003e \u003cp\u003eM1 macrophages are macrophages that can produce pro-inflammatory cytokines. They express high levels of major histocompatibility complex II (MHC-II) and co-stimulatory molecules CD80 and CD86. Under the stimulation of LPS, M0 type quiescent macrophages can be polarized into M1 macrophages, which is marked by an increase in surface MHC-II, CD80, and CD86, enhancing cell antigen presentation capacity and initiating an immune response. Therefore, we treated RAW264.7 cells with different concentrations of BIRB 796 (25, 50, 100, or 200 nM) in combination with LPS. We then used flow cytometry to measure the fluorescence intensity of surface molecules CD80 and CD86 to investigate the effects of BIRB 796 on LPS-induced CD80 and CD86 expression. The results, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e9\u003c/span\u003e, demonstrated that treatment with 200 nM BIRB 796 significantly inhibited the expression of CD80 and CD86 induced by LPS compared to the LPS group. This suggests that BIRB 796 can inhibit the differentiation of M0 macrophages into pro-inflammatory M1 macrophages and suppresses the antigen-presenting capacity of the cells.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e3.8 BIRB 796 Inhibited the Production of Proinflammatory cytokines in LPS-induced RAW264.7 Macrophages\u003c/h2\u003e \u003cp\u003eUnder the stimulation of LPS, macrophages can become over-activated and differentiate into M1 macrophages, leading to an increase in the secretion of pro-inflammatory cytokines. The results in Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e10\u003c/span\u003e showed that compared to the control group, the LPS-treated group significantly upregulated the production of TNF-α and IL-6. Treatment with the drug and LPS simultaneously significantly inhibited the secretion of the inflammatory cytokine TNF-α in a concentration-dependent manner, and also had an inhibitory effect on the secretion of IL-6, indicating that the drug can inhibit the differentiation of macrophages into M1 macrophages and suppress the secretion of pro-inflammatory cytokines induced by LPS, thereby inhibiting the inflammatory response.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Conclusion","content":"\u003cp\u003eMacrophages, as the first line of innate immune response, engulf external antigens through various receptors, secreting multiple inflammatory factors, presenting antigens to immune cells, and triggering subsequent immune responses. However, in autoimmune diseases, the excessive activation of macrophages due to the engulfment and presentation of self-antigens leads to heightened autoimmune reactions, increased secretion of autoantibodies, pathological changes in multiple tissues, ultimately resulting in organ damage.\u003c/p\u003e \u003cp\u003eLyn kinase, a tyrosine kinase, plays a significant role in macrophage polarization. Lyn kinase promotes the secretion of inflammatory factors and cytotoxicity in M1-type macrophages. The study compiled data on all Src family kinase subtypes and introduced a multi-task MSSP model to forecast molecule inhibitory activity across multiple Src subtypes. By combining molecular docking virtual screening and experimental validation, the lead compound Doramapimod was identified to significantly inhibit Lyn kinase activity. This compound affects the secretion of inflammatory factors in LPS-induced macrophages, suppresses the differentiation of M1-type macrophages, promotes M2-type macrophage differentiation, and elevates anti-inflammatory factor levels.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthical Statements:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIt is not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur work is supported by the Major scientific and technological special projects in Xinjiang Uygur Autonomous Region (2022A03016-4).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData will be provided upon request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no conflicts of interest.\u0026nbsp;\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eXuecong Tian: Research design and conceptualization, data analysis, manuscript writing and revision.Luyang Han: Experimental design and implementation, data collection, figure preparation.Ying Su: Literature review, experimental support, data validation.Haiqing Sun: Experimental methods development, data analysis, drafting the manuscript.Sizhe Zhang: Data processing and analysis, results discussion, technical support.Chen Chen: Experimental materials preparation, data collection, research management.Cheng Chen: Data organization, statistical analysis, manuscript revision.Chen-xi Li: Data validation and replication, results discussion, manuscript proofreading.Xiaoyi Lv: Project management, research supervision, final manuscript review.Jinyao Li: Research guidance and oversight, funding acquisition, final manuscript editing and approval.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eMcClendon CJ, Miller WT (2020) Structure, function, and regulation of the SRMS tyrosine kinase. Int J Mol Sci 21(12):4233\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEspada J, Mart\u0026iacute;n-P\u0026eacute;rez J (2017) An update on Src family of nonreceptor tyrosine kinases biology. Int Rev cell Mol biology 331:83\u0026ndash;122\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartellucci S, Clementi L, Sabetta S, Mattei V, Botta L, Angelucci A (2020) Src family kinases as therapeutic targets in advanced solid tumors: what we have learned so far, \u003cem\u003eCancers\u003c/em\u003e, vol. 12, no. 6, p. 1448\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK\u0026ouml;hr G, Seeburg PH (1996) Subtype-specific regulation of recombinant NMDA receptor‐channels by protein tyrosine kinases of the src family. J Physiol 492(2):445\u0026ndash;452\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHiscox S, Nicholson RI (2008) Src inhibitors in breast cancer therapy. Expert Opin Ther Targets 12(6):757\u0026ndash;767\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePalacios EH, Weiss A (2004) Function of the Src-family kinases, Lck and Fyn, in T-cell development and activation, \u003cem\u003eOncogene\u003c/em\u003e, vol. 23, no. 48, pp. 7990\u0026ndash;8000\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNie L, Ye W-R, Chen S, Chirchiglia D, Wang M (2021) Src family kinases in the central nervous system: their emerging role in pathophysiology of migraine and neuropathic pain. Curr Neuropharmacol 19(5):665\u0026ndash;678\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrian BF 4th and, Freedman TS (2021) The Src-family kinase Lyn in immunoreceptor signaling, \u003cem\u003eEndocrinology\u003c/em\u003e, vol. 162, no. 10, p. bqab152\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ede Jesus AA et al (2023) Constitutively active Lyn kinase causes a cutaneous small vessel vasculitis and liver fibrosis syndrome. Nat Commun 14(1):1502\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCongleton J, MacDonald R, Yen A (2012) Src inhibitors, PP2 and dasatinib, increase retinoic acid-induced association of Lyn and c-Raf (S259) and enhance MAPK-dependent differentiation of myeloid leukemia cells, \u003cem\u003eLeukemia\u003c/em\u003e, vol. 26, no. 6, pp. 1180\u0026ndash;1188\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDumas P-Y et al (2017) Hematopoietic niche drives FLT3-ITD acute myeloid leukemia resistance to quizartinib via STAT5-and hypoxia-dependent upregulation of AXL, \u003cem\u003eHaematologica\u003c/em\u003e, vol. 104, no. 10, p. 2019\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCao ZX et al (2020) Erlotinib is effective against FLT3-ITD mutant AML and helps to overcome intratumoral heterogeneity via targeting FLT3 and Lyn. FASEB J 34(8):10182\u0026ndash;10190\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang L et al (2017) Discovery of a small molecule targeting ULK1-modulated cell death of triple negative breast cancer in vitro and in vivo. Chem Sci 8(4):2687\u0026ndash;2701\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoy B et al (2014) Bosutinib in combination with the aromatase inhibitor letrozole: A phase II trial in postmenopausal women evaluating first-line endocrine therapy in locally advanced or metastatic hormone receptor‐positive/HER2‐negative breast cancer. Oncologist 19(4):348\u0026ndash;349\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDaud AI et al (2012) Phase I study of bosutinib, a src/abl tyrosine kinase inhibitor, administered to patients with advanced solid tumors. Clin Cancer Res 18(4):1092\u0026ndash;1100\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWard RA, Fawell S, Floc\u0026rsquo;h N, Flemington V, McKerrecher D, Smith PD (2020) Challenges and opportunities in cancer drug resistance. Chem Rev 121(6):3297\u0026ndash;3351\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRossari F, Minutolo F, Orciuolo E (2018) Past, present, and future of Bcr-Abl inhibitors: from chemical development to clinical efficacy. J Hematol Oncol 11(1):1\u0026ndash;14\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFerguson FM, Gray NS (2018) Kinase inhibitors: the road ahead. Nat Rev Drug Discovery 17(5):353\u0026ndash;377\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCichońska A et al (2021) Crowdsourced mapping of unexplored target space of kinase inhibitors. Nat Commun 12(1):3307\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSwain SS, Hussain T (2021) Role of bioinformatics in early drug discovery: an overview and perspective. Comput BioInformatics: Multidisciplinary Appl, pp. 49\u0026ndash;67\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoneru JK, Sinha S, Mondal J (2019) In Silico reoptimization of binding affinity and drug-resistance circumvention ability in kinase inhibitors: a case study with RL-45 and Src kinase. J Phys Chem B 123(31):6664\u0026ndash;6672\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu W, Weber DJ, Shapiro P, MacKerell AD (2020) Developing Kinase Inhibitors Using Computer-Aided Drug Design Approaches, \u003cem\u003eNext Generation Kinase Inhibitors: Moving Beyond the ATP Binding/Catalytic Sites\u003c/em\u003e, pp. 81\u0026ndash;108\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang Y, Zhang T-j, Tu S, Zhang Z-h, Meng F-h (2020) Identification of novel Src inhibitors: Pharmacophore-based virtual screening, molecular docking and molecular dynamics simulations, \u003cem\u003eMolecules\u003c/em\u003e, vol. 25, no. 18, p. 4094\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang L, Tan J, Han D, Zhu H (2017) From machine learning to deep learning: progress in machine intelligence for rational drug discovery. Drug Discovery Today 22(11):1680\u0026ndash;1685\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim S et al (2019) PubChem 2019 update: improved access to chemical data. Nucleic Acids Res, 47, no. D1, pp. D1102-D1109\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGaulton A et al (2012) ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res, 40, no. D1, pp. D1100-D1107\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang L et al (2016) Chemical fragment-based CDK4/6 inhibitors prediction and web server. RSC Adv 6(21):16972\u0026ndash;16981\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYongye AB, Waddell J, Medina-Franco JL (2012) Molecular scaffold analysis of natural products databases in the public domain. Chem Biol Drug Des 80(5):717\u0026ndash;724\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGregori-Puigjan\u0026eacute; E, Mestres J (2006) Shannon entropy descriptors from topological feature distributions. J Chem Inf Model 46(4):1615\u0026ndash;1622\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMedina-Franco JL, Mart\u0026iacute;nez‐Mayorga K, Bender A, Scior T (2009) Scaffold diversity analysis of compound data sets using an entropy‐based measure, \u003cem\u003eQSAR \u0026amp; Combinatorial Science\u003c/em\u003e, vol. 28, no. 11‐12, pp. 1551\u0026ndash;1560\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Multitask MSSP deep learning model, SRC-Predictor web server, Src inhibitors, Doramapimod","lastPublishedDoi":"10.21203/rs.3.rs-4475200/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4475200/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eSrc family kinases (SFKs), non-receptor tyrosine kinases, crucially contribute to invasion, tumor progression, epithelial-mesenchymal transition, angiogenesis, and metastasis. Thus, Src inhibitors offer a promising avenue for cancer therapy. This study introduced a multitask MSSP deep learning model to predict molecule inhibitory activity across multiple Src subtypes. Comparative assessment against four traditional machine learning methods\u0026mdash;Random Forest (RF), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), eXtreme Gradient Boosting (XGBoost)\u0026mdash;established the superior performance of the multitask MSSP model. It demonstrated the best comprehensive performance, achieving F1-Score and AUC values of 0.906 and 0.975, respectively. An online web server, \"SRC-Predictor,\" was created to aid the practical application of the multitask MSSP model, predicting compounds' potential inhibitory activity against Src. Finally, compounds ranking in the top twenty based on model predictions were selected for experimental validation. Literature search for these compounds revealed limited research on four of them concerning Src. Molecular docking identified Doramapimod as exhibiting better affinity towards Src compared to reference compounds. It significantly inhibited Lyn kinase activity and influenced the secretion levels of inflammatory factors in LPS-induced macrophages. Experimental validation confirmed that our study provides a novel approach for identifying and screening lead compounds as Src inhibitors.\u003c/p\u003e","manuscriptTitle":"Multi-task MSSP model can accurately predict selective Src inhibitors","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-07-16 17:26:04","doi":"10.21203/rs.3.rs-4475200/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"3d815100-40d0-4a74-9e29-34c6d41f8618","owner":[],"postedDate":"July 16th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-07-22T14:53:32+00:00","versionOfRecord":[],"versionCreatedAt":"2024-07-16 17:26:04","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4475200","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4475200","identity":"rs-4475200","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}