Out-of-the-box deep learning prediction of quantum-mechanical partial charges by graph representation and transfer learning.

Accurate prediction of atomic partial charges with high-level quantum mechanics (QM) methods suffers from high computational cost. Numerous feature-engineered machine learning (ML)-based predictors with favorable computability and reliability have been developed as alternatives. However, extensive expertise effort was needed for feature engineering of atom chemical environment, which may consequently introduce domain bias. In this study, SuperAtomicCharge, a data-driven deep graph learning framework, was proposed to predict three important types of partial charges (i.e. RESP, DDEC4 and DDEC78) derived from high-level QM calculations based on the structures of molecules. SuperAtomicCharge was designed to simultaneously exploit the 2D and 3D structural information of molecules, which was proved to be an effective way to improve the prediction accuracy of the model. Moreover, a simple transfer learning strategy and a multitask learning strategy based on self-supervised descriptors were also employed to further improve the prediction accuracy of the proposed model. Compared with the latest baselines, including one GNN-based predictor and two ML-based predictors, SuperAtomicCharge showed better performance on all the three external test sets and had better usability and portability. Furthermore, the QM partial charges of new molecules predicted by SuperAtomicCharge can be efficiently used in drug design applications such as structure-based virtual screening, where the predicted RESP and DDEC4 charges of new molecules showed more robust scoring and screening power than the commonly used partial charges. Finally, two tools including an online server (http://cadd.zju.edu.cn/deepchargepredictor) and the source code command lines (https://github.com/zjujdj/SuperAtomicCharge) were developed for the easy access of the SuperAtomicCharge services.

Catalyzing a Cure: Discovery and development of LRRK2 inhibitors for the treatment of Parkinson's disease.

Parkinson's disease (PD) is an age-related second most common progressive neurodegenerative disorder that affects millions of people worldwide. Despite decades of research, no effective disease modifying therapeutics have reached clinics for treatment/management of PD. Leucine-rich repeat kinase 2 (LRRK2) which controls membrane trafficking and lysosomal function and its variant LRRK2-G2019S are involved in the development of both familial and sporadic PD. LRRK2, is therefore considered as a legitimate target for the development of therapeutics against PD. During the last decade, efforts have been made to develop effective, safe and selective LRRK2 inhibitors and also our understanding about LRRK2 has progressed. However, there is an urge to learn from the previously designed and reported LRRK2 inhibitors in order to effectively approach designing of new LRRK2 inhibitors. In this review, we have aimed to cover the pre-clinical studies undertaken to develop small molecule LRRK2 inhibitors by screening the patents and other available literature in the last decade. We have highlighted LRRK2 as targets in the progress of PD and subsequently covered detailed design, synthesis and development of diverse scaffolds as LRRK2 inhibitors. Moreover, LRRK2 inhibitors under clinical development has also been discussed. LRRK2 inhibitors seem to be potential targets for future therapeutic interventions in the treatment and management of PD and this review can act as a cynosure for guiding discovery, design, and development of selective and non-toxic LRRK2 inhibitors. Although, there might be challenges in developing effective LRRK2 inhibitors, the opportunity to successfully develop novel therapeutics targeting LRRK2 against PD has never been greater.

Discovering a novel dual specificity tyrosine-phosphorylation-regulated kinase 1A (DYRK1A) inhibitor and its impact on tau phosphorylation and amyloid-ß formation.

Dual-specificity tyrosine-regulated kinase 1 A (DYRK1A) is crucial in neurogenesis, synaptogenesis, and neuronal functions. Its dysregulation is linked to neurodegenerative disorders like Down syndrome and Alzheimer's disease. Although the development of DYRK1A inhibitors has significantly advanced in recent years, the selectivity of these drugs remains a critical challenge, potentially impeding further progress. In this study, we utilised structure-based virtual screening (SBVS) from NCI library to discover novel DYRK1A inhibitors. The top-ranked compounds were then validated through enzymatic assays to assess their efficacy towards DYRK1A. Among them, NSC361563 emerged as a potent and selective DYRK1A inhibitor. It was shown to decrease tau phosphorylation at multiple sites, thereby enhancing tubulin stability. Moreover, NSC361563 diminished the formation of amyloid ß and offered neuroprotective benefits against amyloid ß-induced toxicity. Our research highlights the critical role of selective DYRK1A inhibitors in treating neurodegenerative diseases and presents a promising starting point for the development of targeted therapies.

Unveiling Novel ERCC1-XPF Complex Inhibitors: Bridging the Gap from In Silico Exploration to Experimental Design.

Modifications in DNA repair pathways are recognized as prognostic markers and potential therapeutic targets in various cancers, including non-small cell lung cancer (NSCLC). Overexpression of ERCC1 correlates with poorer prognosis and response to platinum-based chemotherapy. As a result, there is a pressing need to discover new inhibitors of the ERCC1-XPF complex that can potentiate the efficacy of cisplatin in NSCLC. In this study, we developed a structure-based virtual screening strategy targeting the inhibition of ERCC1 and XPF interaction. Analysis of crystal structures and a library of small molecules known to act against the complex highlighted the pivotal role of Phe293 (ERCC1) in maintaining complex stability. This residue was chosen as the primary binding site for virtual screening. Using an optimized docking protocol, we screened compounds from various databases, ultimately identifying more than one hundred potential inhibitors. Their capability to amplify cisplatin-induced cytotoxicity was assessed in NSCLC H1299 cells, which exhibited the highest ERCC1 expression of all the cell lines tested. Of these, 22 compounds emerged as promising enhancers of cisplatin efficacy. Our results underscore the value of pinpointing crucial molecular characteristics in the pursuit of novel modulators of the ERCC1-XPF interaction, which could be combined with cisplatin to treat NSCLC more effectively.

Unveiling Novel Urease Inhibitors for Helicobacter pylori: A Multi-Methodological Approach from Virtual Screening and ADME to Molecular Dynamics Simulations.

Helicobacter pylori (Hp) infections pose a global health challenge demanding innovative therapeutic strategies by which to eradicate them. Urease, a key Hp virulence factor hydrolyzes urea, facilitating bacterial survival in the acidic gastric environment. In this study, a multi-methodological approach combining pharmacophore- and structure-based virtual screening, molecular dynamics simulations, and MM-GBSA calculations was employed to identify novel inhibitors for Hp urease (HpU). A refined dataset of 8,271,505 small molecules from the ZINC15 database underwent pharmacokinetic and physicochemical filtering, resulting in 16% of compounds for pharmacophore-based virtual screening. Molecular docking simulations were performed in successive stages, utilizing HTVS, SP, and XP algorithms. Subsequent energetic re-scoring with MM-GBSA identified promising candidates interacting with distinct urease variants. Lys219, a residue critical for urea catalysis at the urease binding site, can manifest in two forms, neutral (LYN) or carbamylated (KCX). Notably, the evaluated molecules demonstrated different interaction and energetic patterns in both protein variants. Further evaluation through ADMET predictions highlighted compounds with favorable pharmacological profiles, leading to the identification of 15 candidates. Molecular dynamics simulations revealed comparable structural stability to the control DJM, with candidates 5, 8 and 12 (CA5, CA8, and CA12, respectively) exhibiting the lowest binding free energies. These inhibitors suggest a chelating capacity that is crucial for urease inhibition. The analysis underscores the potential of CA5, CA8, and CA12 as novel HpU inhibitors. Finally, we compare our candidates with the chemical space of urease inhibitors finding physicochemical similarities with potent agents such as thiourea.

Structure-Based Virtual Screening for Methyltransferase Inhibitors of SARS-CoV-2 nsp14 and nsp16.

The ongoing COVID-19 pandemic still threatens human health around the world. The methyltransferases (MTases) of SARS-CoV-2, specifically nsp14 and nsp16, play crucial roles in the methylation of the N7 and 2'-O positions of viral RNA, making them promising targets for the development of antiviral drugs. In this work, we performed structure-based virtual screening for nsp14 and nsp16 using the screening workflow (HTVS, SP, XP) of Schrödinger 2019 software, and we carried out biochemical assays and molecular dynamics simulation for the identification of potential MTase inhibitors. For nsp14, we screened 239,000 molecules, leading to the identification of three hits A1-A3 showing N7-MTase inhibition rates greater than 60% under a concentration of 50 µM. For the SAM binding and nsp10-16 interface sites of nsp16, the screening of 210,000 and 237,000 molecules, respectively, from ZINC15 led to the discovery of three hit compounds B1-B3 exhibiting more than 45% of 2'-O-MTase inhibition under 50 µM. These six compounds with moderate MTase inhibitory activities could be used as novel candidates for the further development of anti-SARS-CoV-2 drugs.

DeepAtomicCharge: a new graph convolutional network-based architecture for accurate prediction of atomic charges.

Atomic charges play a very important role in drug-target recognition. However, computation of atomic charges with high-level quantum mechanics (QM) calculations is very time-consuming. A number of machine learning (ML)-based atomic charge prediction methods have been proposed to speed up the calculation of high-accuracy atomic charges in recent years. However, most of them used a set of predefined molecular properties, such as molecular fingerprints, for model construction, which is knowledge-dependent and may lead to biased predictions due to the representation preference of different molecular properties used for training. To solve the problem, we present a new architecture based on graph convolutional network (GCN) and develop a high-accuracy atomic charge prediction model named DeepAtomicCharge. The new GCN architecture is designed with only the atomic properties and the connection information between the atoms in molecules and can dynamically learn and convert molecules into appropriate atomic features without any prior knowledge of the molecules. Using the designed GCN architecture, substantial improvement is achieved for the prediction accuracy of atomic charges. The average root-mean-square error (RMSE) of DeepAtomicCharge is 0.0121 e, which is obviously more accurate than that (0.0180 e) reported by the previous benchmark study on the same two external test sets. Moreover, the new GCN architecture needs much lower storage space compared with other methods, and the predicted DDEC atomic charges can be efficiently used in large-scale structure-based drug design, thus opening a new avenue for high-performance atomic charge prediction and application.

Machine learning on ligand-residue interaction profiles to significantly improve binding affinity prediction.

Structure-based virtual screenings (SBVSs) play an important role in drug discovery projects. However, it is still a challenge to accurately predict the binding affinity of an arbitrary molecule binds to a drug target and prioritize top ligands from an SBVS. In this study, we developed a novel method, using ligand-residue interaction profiles (IPs) to construct machine learning (ML)-based prediction models, to significantly improve the screening performance in SBVSs. Such a kind of the prediction model is called an IP scoring function (IP-SF). We systematically investigated how to improve the performance of IP-SFs from many perspectives, including the sampling methods before interaction energy calculation and different ML algorithms. Using six drug targets with each having hundreds of known ligands, we conducted a critical evaluation on the developed IP-SFs. The IP-SFs employing a gradient boosting decision tree (GBDT) algorithm in conjunction with the MIN + GB simulation protocol achieved the best overall performance. Its scoring power, ranking power and screening power significantly outperformed the Glide SF. First, compared with Glide, the average values of mean absolute error and root mean square error of GBDT/MIN + GB decreased about 38 and 36%, respectively. Second, the mean values of squared correlation coefficient and predictive index increased about 225 and 73%, respectively. Third, more encouragingly, the average value of the areas under the curve of receiver operating characteristic for six targets by GBDT, 0.87, is significantly better than that by Glide, which is only 0.71. Thus, we expected IP-SFs to have broad and promising applications in SBVSs.

Virtual screening and biological evaluation to identify pharmaceuticals potentially causing hypertension and hypokalemia by inhibiting steroid 11ß-hydroxylase.

Several drugs were found after their market approval to unexpectedly inhibit adrenal 11ß-hydroxylase (CYP11B1)-dependent cortisol synthesis. Known side-effects of CYP11B1 inhibition include hypertension and hypokalemia, due to a feedback activation of adrenal steroidogenesis, leading to supraphysiological concentrations of 11-deoxycortisol and 11-deoxycorticosterone that can activate the mineralocorticoid receptor. This results in potassium excretion and sodium and water retention, ultimately causing hypertension. With the risk known but usually not addressed in preclinical evaluation, this study aimed to identify drugs and drug candidates inhibiting CYP11B1. Two conceptually different virtual screening methods were combined, a pharmacophore based and an induced fit docking approach. Cell-free and cell-based CYP11B1 activity measurements revealed several inhibitors with IC50 values in the nanomolar range. Inhibitors include retinoic acid metabolism blocking agents (RAMBAs), azole antifungals, α2-adrenoceptor ligands, and a farnesyltransferase inhibitor. The active compounds share a nitrogen atom embedded in an aromatic ring system. Structure activity analysis identified the free electron pair of the nitrogen atom as a prerequisite for the drug-enzyme interaction, with its pKa value as an indicator of inhibitory potency. Another important parameter is drug lipophilicity, exemplified by etomidate. Changing its ethyl ester moiety to a more hydrophilic carboxylic acid group dramatically decreased the inhibitory potential, most likely due to less efficient cellular uptake. The presented work successfully combined different in silico and in vitro methods to identify several previously unknown CYP11B1 inhibitors. This workflow facilitates the identification of compounds that inhibit CYP11B1 and therefore pose a risk for inducing hypertension and hypokalemia.

Discovery of novel methionine adenosyltransferase 2A (MAT2A) allosteric inhibitors by structure-based virtual screening.

Methionine adenosyltransferase 2A (MAT2A) has been indicated as a drug target for oncology indications. Clinical trials with MAT2A inhibitors are currently on-going. Here, a structure-based virtual screening campaign was performed on the commercially available chemical space which yielded two novel MAT2A-inhibitor chemical series. The binding modes of the compounds were confirmed with X-ray crystallography. Both series have acceptable physicochemical properties and show nanomolar activity in the biochemical MAT2A inhibition assay and single-digit micromolar activity in the proliferation assay (MTAP -/- cell line). The identified compounds and the relating structural data could be helpful in related drug discovery projects.

Discovery of VEGFR inhibitors through virtual screening and energy assessment.

Vascular endothelial growth factor receptor-2 (VEGFR-2) is crucial in promoting tumor angiogenesis and cancer metastasis. Thus, inhibition of VEGFR-2 has appeared as a good tactic for cancer treatment. To find out novel VEGFR-2 inhibitors, first, the PDB structure of VEGFR-2, 6GQO, was selected based on atomic nonlocal environment assessment (ANOLEA) and PROCHECK assessment. 6GQO was then further used for structure-based virtual screening (SBVS) of different molecular databases, including US-FDA approved drugs, US-FDA withdrawn drugs, may bridge, MDPI, and Specs databases using Glide. Based on SBVS, receptor fit, drug-like filters, and absorption, distribution, metabolism, excretion, and toxicity (ADMET) analysis of 427877 compounds, the best 22 hits were selected. From the 22 hits, hit 5 complex with 6GQO was put through molecular mechanics/generalized born surface area (MM/GBSA) study and hERG binding. The MM/GBSA study revealed that hit 5 possesses lesser binding free energy with more inferior stability in the receptor pocket than the reference compound. The VEGFR-2 inhibition assay of hit 5 disclosed an IC50 of 165.23 nM against VEGFR-2, which can be possibly enhanced through structural modifications.

Discovery of novel TLR4/MD-2 inhibitors: Receptor structure-based virtual screening studies and anti-inflammatory evaluation.

In this study, a receptor structure-based virtual screening strategy was constructed using a computer-aided drug design. First, the compounds were filtered based on the Lipinski pentad and adsorption, distribution, metabolism, excretion, and toxicity profiles. Then, receptor structure-based pharmacophore models were constructed and screened. Finally, the in vitro toxicity and anti-inflammatory activities of hit compounds were initially evaluated to investigate their in vitro anti-inflammatory effects and mechanisms of action. The results revealed that hit 94 had the best anti-inflammatory activity and low toxicity while inhibiting the activation of Toll-like receptor (TLR) 4/myeloid differentiation factor 2 (MD2)-associated signaling pathways of nuclear factor-κB and mitogen-activated protein kinase. In vivo adjuvant arthritis results also revealed that hit 94 ameliorated foot swelling to a greater extent in rats compared with the positive control drug indomethacin. These results suggest that hit 94 can be used as a potential TLR/MD2 inhibitor for inflammatory diseases.

Virtual screening for potential discoidin domain receptor 1 (DDR1) inhibitors based on structural assessment.

Discoidin domain receptor 1 (DDR1) (EC Number 2.7.10.1) has recently been considered as a promising therapeutic target for idiopathic pulmonary fibrosis (IPF). However, none of the currently discovered DDR1 inhibitors have been included in clinical studies due to low target specificity or druggability limitations, necessitating various approaches to develop novel DDR1 inhibitors. In this study, to assure target specificity, a docking assessment of the DDR1 crystal structures was undertaken to find the well-differentiated crystal structure, and 4CKR was identified among many crystal structures. Then, using the best pharmacophore model and molecular docking, virtual screening of the ChEMBL database was done, and five potential molecules were identified as promising inhibitors of DDR1. Subsequently, all hit compound complex systems were validated using molecular dynamics simulations and MM/PBSA methods to assess the stability of the system after ligand binding to DDR1. Based on molecular dynamics simulations and hydrogen-bonding occupancy analysis, the DDR1-Cpd2, DDR1-Cpd17, and DDR1-Cpd18 complex systems exhibited superior stability compared to the DDR1-Cpd1 and DDR-Cpd33 complex systems. Meanwhile, when targeting DDR1, the descending order of the five hit molecules' binding free energies was Cpd17 (- 145.820 kJ/mol) > Cpd2 (- 131.818 kJ/mol) > Cpd18 (- 130.692 kJ/mol) > Cpd33 (- 129.175 kJ/mol) > Cpd1 (- 126.103 kJ/mol). Among them, Cpd2, Cpd17, and Cpd18 showed improved binding characteristics, indicating that they may be potential DDR1 inhibitors. In this research, we developed a high-hit rate, effective screening method that serves as a theoretical guide for finding DDR1 inhibitors for the development of IPF therapeutics.

Discovery and biological evaluation of novel CARM1/HDAC2 dual-targeting inhibitors with anti-prostate cancer agents.

Prostate cancer (PCa) is a clinically heterogeneous disease with a progressively increasing incidence. Concurrent inhibition of coactivator-associated arginine methyltransferase 1 (CARM1) and histone deacetylase 2 (HDAC2) could potentially be a novel strategy against PCa. Herein, we identified seven compounds simultaneously targeting CARM1 and HDAC2 through structure-based virtual screening. These compounds possessed potent inhibitory activities at the nanomolar level in vitro. Among them, CH-1 was the most active inhibitor which exhibited excellent and balanced inhibitory effects against both CARM1 (IC50 = 3.71 ± 0.11 nM) and HDAC2 (IC50 = 4.07 ± 0.25 nM). MD simulations presented that CH-1 could stably bind the active pockets of CARM1 and HDAC2. Notably, CH-1 exhibited strong anti-proliferative activity against multiple prostate-related tumour cells (IC50 < 1 µM). In vivo, assessment indicated that CH-1 significantly inhibited tumour growth in a DU145 xenograft model. Collectively, CH-1 could be a promising drug candidate for PCa treatment.

In silico identification and biological evaluation of a selective MAP4K4 inhibitor against pancreatic cancer.

Inhibiting a specific target in cancer cells and reducing unwanted side effects has become a promising strategy in pancreatic cancer treatment. MAP4K4 is associated with pancreatic cancer development and correlates with poor clinical outcomes. By phosphorylating MKK4, proteins associated with cell apoptosis and survival are translated. Therefore, inhibiting MAP4K4 activity in pancreatic tumours is a new therapeutic strategy. Herein, we performed a structure-based virtual screening to identify MAP4K4 inhibitors and discovered the compound F389-0746 with a potent inhibition (IC50 120.7 nM). The results of kinase profiling revealed that F389-0746 was highly selective to MAP4K4 and less likely to cause side effects. Results of in vitro experiments showed that F389-0746 significantly suppressed cancer cell growth and viability. Results of in vivo experiments showed that F389-0746 displayed comparable tumour growth inhibition with the group treated with gemcitabine. These findings suggest that F389-0746 has promising potential to be further developed as a novel pancreatic cancer treatment.

Identification of Four New Chemical Series of Small Drug-Like Natural Products as Potential Neuropilin-1 Inhibitors by Structure-Based Virtual Screening: Pharmacophore-Based Molecular Docking and Dynamics Simulation.

Neuropilin-1 (NRP-1), a surface transmembrane glycoprotein, is one of the most important co-receptors of VEGF-A165 (vascular endothelial growth factor) responsible for pathological angiogenesis. In general, NRP-1 overexpression in cancer correlates with poor prognosis and more tumor aggressiveness. NRP-1 role in cancer has been mainly explained by mediating VEGF-A165-induced effects on tumor angiogenesis. NRP-1 was recently identified as a co-receptor and an independent gateway for SARS-CoV-2 through binding subunit S2 of Spike protein in the same way as VEGF-A165. Thus, NRP-1 is of particular value as a target for cancer therapy and other angiogenesis-dependent diseases as well as for SARS-CoV-2 antiviral intervention. Herein, The Super Natural II, the largest available database of natural products (∼0.33 M), pre-filtered with drug-likeness criteria (absorption, distribution, metabolism and excretion/toxicity), was screened against NRP-1. NRP-1/VEGF-A165 interaction is one of protein-protein interfaces (PPIs) known to be challenging when approached in-silico. Thus, a PPI-suited multi-step virtual screening protocol, incorporating a derived pharmacophore with molecular docking and followed by MD (molecular dynamics) simulation, was designed. Two stages of pharmacophorically constrained molecular docking (standard and extra precisions), a mixed Torsional/Low-mode conformational search and MM-GBSA ΔG binding affinities calculation, resulted in the selection of 100 hits. These 100 hits were subjected to 20 ns MD simulation, that was extended to 100 ns for top hits (20) and followed by post-dynamics analysis (atomic ligand-protein contacts, RMSD, RMSF, MM-GBSA ΔG, Rg, SASA and H-bonds). Post-MD analysis showed that 19 small drug-like nonpeptide natural molecules, grouped in four chemical scaffolds (purine, thiazole, tetrahydropyrimidine and dihydroxyphenyl), well verified the derived pharmacophore and formed stable and compact complexes with NRP-1. The discovered molecules are promising and can serve as a base for further development of new NRP-1 inhibitors.

Keras/TensorFlow in Drug Design for Immunity Disorders.

Homeostasis of the host immune system is regulated by white blood cells with a variety of cell surface receptors for cytokines. Chemotactic cytokines (chemokines) activate their receptors to evoke the chemotaxis of immune cells in homeostatic migrations or inflammatory conditions towards inflamed tissue or pathogens. Dysregulation of the immune system leading to disorders such as allergies, autoimmune diseases, or cancer requires efficient, fast-acting drugs to minimize the long-term effects of chronic inflammation. Here, we performed structure-based virtual screening (SBVS) assisted by the Keras/TensorFlow neural network (NN) to find novel compound scaffolds acting on three chemokine receptors: CCR2, CCR3, and one CXC receptor, CXCR3. Keras/TensorFlow NN was used here not as a typically used binary classifier but as an efficient multi-class classifier that can discard not only inactive compounds but also low- or medium-activity compounds. Several compounds proposed by SBVS and NN were tested in 100 ns all-atom molecular dynamics simulations to confirm their binding affinity. To improve the basic binding affinity of the compounds, new chemical modifications were proposed. The modified compounds were compared with known antagonists of these three chemokine receptors. Known CXCR3 compounds were among the top predicted compounds; thus, the benefits of using Keras/TensorFlow in drug discovery have been shown in addition to structure-based approaches. Furthermore, we showed that Keras/TensorFlow NN can accurately predict the receptor subtype selectivity of compounds, for which SBVS often fails. We cross-tested chemokine receptor datasets retrieved from ChEMBL and curated datasets for cannabinoid receptors. The NN model trained on the cannabinoid receptor datasets retrieved from ChEMBL was the most accurate in the receptor subtype selectivity prediction. Among NN models trained on the chemokine receptor datasets, the CXCR3 model showed the highest accuracy in differentiating the receptor subtype for a given compound dataset.

Virtual Screening and ADMET Prediction to Uncover the Potency of Flavonoids from Genus Erythrina as Antibacterial Agent through Inhibition of Bacterial ATPase DNA Gyrase B.

The emergence of antimicrobial resistance due to the widespread and inappropriate use of antibiotics has now become the global health challenge. Flavonoids have long been reported to be a potent antimicrobial agent against a wide range of pathogenic microorganisms in vitro. Therefore, new antibiotics development based on flavonoid structures could be a potential strategy to fight against antibiotic-resistant infections. This research aims to screen the potency of flavonoids of the genus Erythrina as an inhibitor of bacterial ATPase DNA gyrase B. From the 378 flavonoids being screened, 49 flavonoids show potential as an inhibitor of ATPase DNA gyrase B due to their lower binding affinity compared to the inhibitor and ATP. Further screening for their toxicity, we identified 6 flavonoids from these 49 flavonoids, which are predicted to have low toxicity. Among these flavonoids, erystagallin B (334) is predicted to have the best pharmacokinetic properties, and therefore, could be further developed as new antibacterial agent.

A Natural Compound Containing a Disaccharide Structure of Glucose and Rhamnose Identified as Potential N-Glycanase 1 (NGLY1) Inhibitors.

N-glycanase 1 (NGLY1) is an essential enzyme involved in the deglycosylation of misfolded glycoproteins through the endoplasmic reticulum (ER)-associated degradation (ERAD) pathway, which could hydrolyze N-glycan from N-glycoprotein or N-glycopeptide in the cytosol. Recent studies indicated that NGLY1 inhibition is a potential novel drug target for antiviral therapy. In this study, structure-based virtual analysis was applied to screen candidate NGLY1 inhibitors from 2960 natural compounds. Three natural compounds, Poliumoside, Soyasaponin Bb, and Saikosaponin B2 showed significantly inhibitory activity of NGLY1, isolated from traditional heat-clearing and detoxifying Chinese herbs. Furthermore, the core structural motif of the three NGLY1 inhibitors was a disaccharide structure with glucose and rhamnose, which might exert its action by binding to important active sites of NGLY1, such as Lys238 and Trp244. In traditional Chinese medicine, many compounds containing this disaccharide structure probably targeted NGLY1. This study unveiled the leading compound of NGLY1 inhibitors with its core structure, which could guide future drug development.

In Silico Screening of Natural Flavonoids against 3-Chymotrypsin-like Protease of SARS-CoV-2 Using Machine Learning and Molecular Modeling.

The "Long-COVID syndrome" has posed significant challenges due to a lack of validated therapeutic options. We developed a novel multi-step virtual screening strategy to reliably identify inhibitors against 3-chymotrypsin-like protease of SARS-CoV-2 from abundant flavonoids, which represents a promising source of antiviral and immune-boosting nutrients. We identified 57 interacting residues as contributors to the protein-ligand binding pocket. Their energy interaction profiles constituted the input features for Machine Learning (ML) models. The consensus of 25 classifiers trained using various ML algorithms attained 93.9% accuracy and a 6.4% false-positive-rate. The consensus of 10 regression models for binding energy prediction also achieved a low root-mean-square error of 1.18 kcal/mol. We screened out 120 flavonoid hits first and retained 50 drug-like hits after predefined ADMET filtering to ensure bioavailability and safety profiles. Furthermore, molecular dynamics simulations prioritized nine bioactive flavonoids as promising anti-SARS-CoV-2 agents exhibiting both high structural stability (root-mean-square deviation < 5 Å for 218 ns) and low MM/PBSA binding free energy (<-6 kcal/mol). Among them, KB-2 (PubChem-CID, 14630497) and 9-O-Methylglyceofuran (PubChem-CID, 44257401) displayed excellent binding affinity and desirable pharmacokinetic capabilities. These compounds have great potential to serve as oral nutraceuticals with therapeutic and prophylactic properties as care strategies for patients with long-COVID syndrome.