Learning with uncertainty to accelerate the discovery of histone lysine-specific demethylase 1A (KDM1A/LSD1) inhibitors.

Machine learning including modern deep learning models has been extensively used in drug design and screening. However, reliable prediction of molecular properties is still challenging when exploring out-of-domain regimes, even for deep neural networks. Therefore, it is important to understand the uncertainty of model predictions, especially when the predictions are used to guide further experiments. In this study, we explored the utility and effectiveness of evidential uncertainty in compound screening. The evidential Graphormer model was proposed for uncertainty-guided discovery of KDM1A/LSD1 inhibitors. The benchmarking results illustrated that (i) Graphormer exhibited comparative predictive power to state-of-the-art models, and (ii) evidential regression enabled well-ranked uncertainty estimates and calibrated predictions. Subsequently, we leveraged time-splitting on the curated KDM1A/LSD1 dataset to simulate out-of-distribution predictions. The retrospective virtual screening showed that the evidential uncertainties helped reduce false positives among the top-acquired compounds and thus enabled higher experimental validation rates. The trained model was then used to virtually screen an independent in-house compound set. The top 50 compounds ranked by two different ranking strategies were experimentally validated, respectively. In general, our study highlighted the importance to understand the uncertainty in prediction, which can be recognized as an interpretable dimension to model predictions.

A chemoinformatic analysis on natural glycosides with respect to biological origin and structural class.

Covering: up to 202216.19% of reported natural products (NPs) in the Dictionary of Natural Products (DNP) are glycosides. As one of the most important NPs' structural modifications, glycosylation can change the NPs' polarity, making the aglycones more amphipathic. However, until now, little is known about the general distribution profile of the natural glycosides in different biological sources or structural types. The reason, structural or species preferences of the natural glycosylation remain unclear. In this highlight, chemoinformatic methods were employed to analyze the natural glycosides from DNP, the most comprehensively annotated NP database. We found that the glycosylation ratios of NPs from plants, bacteria, animals and fungi decrease successively, which are 24.99%, 20.84%, 8.40% and 4.48%, respectively. Echinoderm-derived NPs (56.11%) are the most frequently glycosylated, while those produced by molluscs (1.55%), vertebrates (2.19%) and Rhodophyta (3.00%) are the opposite. Among the diverse structural types, a large proportion of steroids (45.19%), tannins (44.78%) and flavonoids (39.21%) are glycosides, yet aminoacids and peptides (5.16%), alkaloids (5.66%) are comparatively less glycosylated. Even within the same biological source or structural type, their glycosylation rates fluctuate drastically between sub- or cross-categories. The substitute patterns of flavonoid and terpenoid glycosides and the most frequently glycosylated scaffolds were identified. NPs with different glycosylation levels occupy different chemical spaces of physicochemical property and scaffold. These findings could help us to interpret the preference of NPs' glycosylation and investigate how NP glycosylation could aid NP-based drug discovery.

Binding Mechanism of CD47 with SIRPα Variants and Its Antibody: Elucidated by Molecular Dynamics Simulations.

The intricate complex system of the differentiation 47 (CD47) and the signal-regulatory protein alpha (SIRPα) cluster is a crucial target for cancer immunotherapy. Although the conformational state of the CD47-SIRPα complex has been revealed through crystallographic studies, further characterization is needed to fully understand the binding mechanism and to identify the hot spot residues involved. In this study, molecular dynamics (MD) simulations were carried out for the complexes of CD47 with two SIRPα variants (SIRPαv1, SIRPαv2) and the commercially available anti-CD47 monoclonal antibody (B6H12.2). The calculated binding free energy of CD47-B6H12.2 is lower than that of CD47-SIRPαv1 and CD47-SIRPαv2 in all the three simulations, indicating that CD47-B6H12.2 has a higher binding affinity than the other two complexes. Moreover, the dynamical cross-correlation matrix reveals that the CD47 protein shows more correlated motions when it binds to B6H12.2. Significant effects were observed in the energy and structural analyses of the residues (Glu35, Tyr37, Leu101, Thr102, Arg103) in the C strand and FG region of CD47 when it binds to the SIRPα variants. The critical residues (Leu30, Val33, Gln52, Lys53, Thr67, Arg69, Arg95, and Lys96) were identified in SIRPαv1 and SIRPαv2, which surround the distinctive groove regions formed by the B2C, C'D, DE, and FG loops. Moreover, the crucial groove structures of the SIRPα variants shape into obvious druggable sites. The C'D loops on the binding interfaces undergo notable dynamical changes throughout the simulation. For B6H12.2, the residues Tyr32LC, His92LC, Arg96LC, Tyr32HC, Thr52HC, Ser53HC, Ala101HC, and Gly102HC in its initial half of the light and heavy chains exhibit obvious energetic and structural impacts upon binding with CD47. The elucidation of the binding mechanism of SIRPαv1, SIRPαv2, and B6H12.2 with CD47 could provide novel perspectives for the development of inhibitors targeting CD47-SIRPα.

EnMCB: an R/bioconductor package for predicting disease progression based on methylation correlated blocks using ensemble models.

MOTIVATION: Based on the concept that contiguous cytosine-phosphorothioate-guanine (CpG) sites in the same DNA strand may be modified by a methyltransferase or demethylase together, current study found that the combination of multiple CpGs into a single block may promote cancer diagnosis. However, there is no R package available for building models based on methylation correlated blocks. RESULTS: Here, we present a package named stacked ensemble of machine learning models for methylation correlated blocks (EnMCB) to build signatures based on DNA methylation correlated blocks for survival prediction. The Cox regression, support vector regression, mboost and elastic-net model were combined in the ensemble model. Methylation profiles from The Cancer Genome Atlas were used as real datasets. The package automatically partitions the genome into blocks of tightly co-methylated CpG sites, termed methylation correlated blocks. After partitioning and modeling, the diagnostic capacities for predicting patients' survivals are given. AVAILABILITY AND IMPLEMENTATION: EnMCB is freely available for download at GitHub (https://github.com/whirlsyu/EnMCB/) and Bioconductor (http://bioconductor.org/packages/release/bioc/html/EnMCB.html). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Accelerating the identification of subtype selective inhibitors via Three-Dimensional Biologically Relevant Spectrum (BRS-3D): The monoamine oxidase subtypes as a case study.

Subtype-selective drugs are of great therapeutic importance as they are expected to be more effective and with less side-effects. However, discovery of subtype selective inhibitors was hampered by the high similarity of the binding sites within subfamilies. In this study, we further evaluated the applicability of "Three-Dimensional Biologically Relevant Spectrum (BRS-3D)" for the identification of subtype-selective inhibitors. A case study was performed on monoamine oxidase, which has two subtypes related to distinct diseases. The inhibitory activity against MAO-A/B of 347 compounds experimentally tested in this research was reported. Compound M124 (5H-thiazolo[3,2-a]pyrimidin-5-one) with IC50 less than 100 nM (SI = 23) was selected as a probe to investigate the structure selectivity relationship. Similarity search led to the identification of compound M229 and M249 with IC50 values of 7.4 nM, 4 nM and acceptable selectivity index over MAO-A (M229 SI > 1351, M249 SI > 2500). The molecular basis for subtype selectivity was explored through docking study and attention based DNN model. Additionally, in silico ADME properties were characterized. Accordingly, it is found that BRS-3D is a robust method for subtype selectivity in the early stage of drug discovery and the compounds reported here can be promising leads for further experimental analysis.

Novel C7-Substituted Coumarins as Selective Monoamine Oxidase Inhibitors: Discovery, Synthesis and Theoretical Simulation.

There is a continued need to develop new selective human monoamine oxidase (hMAO) inhibitors that could be beneficial for the treatment of neurological diseases. However, hMAOs are closely related with high sequence identity and structural similarity, which hinders the development of selective MAO inhibitors. "Three-Dimensional Biologically Relevant Spectrum (BRS-3D)" method developed by our group has demonstrated its effectiveness in subtype selectivity studies of receptor and enzyme ligands. Here, we report a series of novel C7-substituted coumarins, either synthesized or commercially purchased, which were identified as selective hMAO inhibitors. Most of the compounds demonstrated strong activities with IC50 values (half-inhibitory concentration) ranging from sub-micromolar to nanomolar. Compounds, FR1 and SP1, were identified as the most selective hMAO-A inhibitors, with IC50 values of 1.5 nM (selectivity index (SI) < -2.82) and 19 nM (SI < -2.42), respectively. FR4 and FR5 showed the most potent hMAO-B inhibitory activity, with IC50 of 18 nM and 15 nM (SI > 2.74 and SI > 2.82). Docking calculations and molecular dynamic simulations were performed to elucidate the selectivity preference and SAR profiles.

Integrating Multifaceted Information to Predict Mycobacterium tuberculosis-Human Protein-Protein Interactions.

Tuberculosis (TB) is one of the biggest infectious disease killers caused by Mycobacterium tuberculosis (MTB). Studying the protein-protein interactions (PPIs) between MTB and human can deepen our understanding of the pathogenesis of TB and offer new clues to the treatment against MTB infection, but the experimentally validated interactions are especially scarce in this regard. Herein we proposed an integrated framework that combined template-, domain-domain interaction-, and machine learning-based methods to predict MTB-human PPIs. As a result, we established a network composed of 13 758 PPIs including 451 MTB proteins and 3167 human proteins ( http://liulab.hzau.edu.cn/MTB/ ). Compared to known human targets of various pathogens, our predicted human targets show a similar tendency in terms of the network topological properties and enrichment in important functional genes. Additionally, these human targets largely have longer sequence lengths, more protein domains, more disordered residues, lower evolutionary rates, and older protein ages. Functional analysis demonstrates that these proteins show strong preferences toward the phosphorylation, kinase activity, and signaling transduction processes and the disease and immune related pathways. Dissecting the cross-talk among top-ranked pathways suggests that the cancer pathway may serve as a bridge in MTB infection. Triplet analysis illustrates that the paired targets interacting with the same partner are adjacent to each other in the intraspecies network and tend to share similar expression patterns. Finally, we identified 36 potential anti-MTB human targets by integrating known drug target information and molecular properties of proteins.

Cheminformatic Insight into the Differences between Terrestrial and Marine Originated Natural Products.

This is a new golden age for drug discovery based on natural products derived from both marine and terrestrial sources. Herein, a straightforward but important question is "what are the major structural differences between marine natural products (MNPs) and terrestrial natural products (TNPs)?" To answer this question, we analyzed the important physicochemical properties, structural features, and drug-likeness of the two types of natural products and discussed their differences from the perspective of evolution. In general, MNPs have lower solubility and are often larger than TNPs. On average, particularly from the perspective of unique fragments and scaffolds, MNPs usually possess more long chains and large rings, especially 8- to 10-membered rings. MNPs also have more nitrogen atoms and halogens, notably bromines, and fewer oxygen atoms, suggesting that MNPs may be synthesized by more diverse biosynthetic pathways than TNPs. Analysis of the frequently occurring Murcko frameworks in MNPs and TNPS also reveals a striking difference between MNPs and TNPs. The scaffolds of the former tend to be longer and often contain ester bonds connected to 10-membered rings, while the scaffolds of the latter tend to be shorter and often bear more stable ring systems and bond types. Besides, the prediction from the naïve Bayesian drug-likeness classification model suggests that most compounds in MNPs and TNPs are drug-like, although MNPs are slightly more drug-like than TNPs. We believe that MNPs and TNPs with novel drug-like scaffolds have great potential to be drug leads or drug candidates in drug discovery campaigns.

Theoretical Calculation and Experimental Verification Demonstrated the Impossibility of Finding Haptens Identifying Triphenylmethane Dyes and Their Leuco Metabolites Simultaneously.

Detection of triphenylmethane dyes (TDs), especially the widely used malachite green (MG) and crystal violet (CV), plays an important role in safety control of aquatic products. There are two chromatic forms of TDs: oxidized or reduced. Usually, only one form can be detected by reported ELISA antibodies. In this article, molecular shape superimposing and quantum mechanics calculation were employed to elucidate the differences between MG, CV, and their reduced chromatic forms (leucomalachite green, LMG and leucocrystal violet, LCV). A potential hapten was rationally designed and synthesized. Polyclonal antibodies were raised through immunizing New Zealand white rabbits and BALB/C mice. We tested the cross-reactivity ratios between the hapten and TDs. The cross-reactivity ratios were correlated with the difference in surface electrostatic potential. The determination coefficients (r²) of the correlations are 0.901 and 0.813 for the rabbit and mouse antibody, respectively. According to this linear model, the significant difference in the atomic charge seemed to make it impossible to find a hapten that can produce antibodies with good cross-reactivities with both reduced and oxidized TDs.

Structural Transformation Detection Contributes to Screening of Behaviorally Active Compounds: Dynamic Binding Process Analysis of DhelOBP21 from Dastarcus helophoroides.

In light of reverse chemical ecology, the fluorescence competitive binding assays of functional odorant binding proteins (OBPs) is a recent advanced approach for screening behaviorally active compounds of insects. Previous research on Dastareus helophoroides identified a minus-C OBP, DhelOBP21, which preferably binds to several ligands. In this study, only (+)-ß-pinene proved attractive to unmated adult beetles. To obtain a more in-depth explanation of the lack of behavioral activity of other ligands we selected compounds with high (camphor) and low (ß-caryophyllene) binding affinities. The structural transformation of OBPs was investigated using well-established approaches for studying binding processes, such as fluorescent quenching assays, circular dichroism, and molecular dynamics. The dynamic binding process revealed that the flexibility of DhelOBP21 seems conducive to binding specific ligands, as opposed to broad substrate binding. The compound (+)-ß-pinene and DhelOBP21 formed a stable complex through a secondary structural transformation of DhelOBP21, in which its amino-terminus transformed from random coil to an α-helix to cover the binding pocket. On the other hand, camphor could not efficiently induce a stable structural transformation, and its high binding affinities were due to strong hydrogen-bonding, compromising the structure of the protein. The other compound, ß-caryophyllene, only collided with DhelOBP21 and could not be positioned in the binding pocket. Studying structural transformation of these proteins through examining the dynamic binding process rather than using approaches that just measure binding affinities such as fluorescence competitive binding assays can provide a more efficient and reliable approach for screening behaviorally active compounds.

Three-Dimensional Biologically Relevant Spectrum (BRS-3D): Shape Similarity Profile Based on PDB Ligands as Molecular Descriptors.

The crystallized ligands in the Protein Data Bank (PDB) can be treated as the inverse shapes of the active sites of corresponding proteins. Therefore, the shape similarity between a molecule and PDB ligands indicated the possibility of the molecule to bind with the targets. In this paper, we proposed a shape similarity profile that can be used as a molecular descriptor for ligand-based virtual screening. First, through three-dimensional (3D) structural clustering, 300 diverse ligands were extracted from the druggable protein-ligand database, sc-PDB. Then, each of the molecules under scrutiny was flexibly superimposed onto the 300 ligands. Superimpositions were scored by shape overlap and property similarity, producing a 300 dimensional similarity array termed the "Three-Dimensional Biologically Relevant Spectrum (BRS-3D)". Finally, quantitative or discriminant models were developed with the 300 dimensional descriptor using machine learning methods (support vector machine). The effectiveness of this approach was evaluated using 42 benchmark data sets from the G protein-coupled receptor (GPCR) ligand library and the GPCR decoy database (GLL/GDD). We compared the performance of BRS-3D with other 2D and 3D state-of-the-art molecular descriptors. The results showed that models built with BRS-3D performed best for most GLL/GDD data sets. We also applied BRS-3D in histone deacetylase 1 inhibitors screening and GPCR subtype selectivity prediction. The advantages and disadvantages of this approach are discussed.

Anti-inflammatory terpenes from flowers of Inula japonica.

Five new terpenes (1-5) and ten known compounds (6-15) were isolated from Inula japonica, and their structures were identified by spectroscopic analysis. Compounds 3 and 14 showed positive inhibitory effects on nitric oxide production. Furthermore, compound 14 suppressed both leukotriene C4 synthesis and degranulation in c-kit ligand-induced bone marrow-derived mast cells.

[Growth and photosynthetic characteristics of Ardisia corymbifera var. tuberifera in bionic wild cultivation].

OBJECTIVE: To explore the growth dynamic and photosynthetic characteristics of Ardisia corymbifera var. tuberifera in bionic wild cultivation. METHODS: The test of bionic wild cultivation was conducted with two-year-old seedlings as well as three-year-old cuttings of Ardisia corymbifera var. tuberifera, and their growth dynamic and photosynthetic characteristics were studied. RESULTS: Both seedlings and cuttings of Ardisia corymbifera var. tuberifera maintained good growth in imitation of the wild, the height were increased by 9.2 cm and 12 cm, base diameter were increased by 2.48 mm and 2.39 mm,and crown width were increased by 10.6 cm and 17 cm, respectively. Two-year-old seedlings branched only once a year, and three-year-old cuttings branched thrice a year. There were no significant differences in all photosynthetic parameters of the two kinds of seedlings, and the range of growth light intensity of Ardisia corymbifera var. tuberifera was wider. Owing to the weaker light quantum efficiency under lower light and relatively lower light photosynthetic capacity under higher light, its competition ability was lower in the field. CONCLUSION: The increases of height and crown width were greater in cuttings than those in seedlings, and the transplanting survival rate of the former was larger.

Conformational plasticity of SpyCas9 induced by AcrIIA4 and AcrIIA2: Insights from molecular dynamics simulation.

CRISPR-Cas9 systems constitute bacterial adaptive immune systems that protect against phage infections. Bacteriophages encode anti-CRISPR proteins (Acrs) that mitigate the bacterial immune response. However, the structural basis for their inhibitory actions from a molecular perspective remains elusive. In this study, through microsecond atomistic molecular dynamics simulations, we demonstrated the remarkable flexibility of Streptococcus pyogenes Cas9 (SpyCas9) and its conformational adaptability during interactions with AcrIIA4 and AcrIIA2. Specifically, we demonstrated that the binding of AcrIIA4 and AcrIIA2 to SpyCas9 induces a conformational rearrangement that causes spatial separation between the nuclease and cleavage sites, thus making the endonuclease inactive. This separation disrupts the transmission of signals between the protospacer adjacent motif recognition and nuclease domains, thereby impeding the efficient processing of double-stranded DNA. The simulation also reveals that AcrIIA4 and AcrIIA2 cause different structural variations of SpyCas9. Our research illuminates the precise mechanisms underlying the suppression of SpyCas9 by AcrIIA4 and AcrIIA2, thus presenting new possibilities for controlling genome editing with higher accuracy.

A terpenoids database with the chemical content as a novel agronomic trait.

Natural products play a pivotal role in drug discovery, and the richness of natural products, albeit significantly influenced by various environmental factors, is predominantly determined by intrinsic genetics of a series of enzymatic reactions and produced as secondary metabolites of organisms. Heretofore, few natural product-related databases take the chemical content into consideration as a prominent property. To gain unique insights into the quantitative diversity of natural products, we have developed the first TerPenoids database embedded with Content information (TPCN) with features such as compound browsing, structural search, scaffold analysis, similarity analysis and data download. This database can be accessed through a web-based computational toolkit available at http://www.tpcn.pro/. By conducting meticulous manual searches and analyzing over 10 000 reference papers, the TPCN database has successfully integrated 6383 terpenoids obtained from 1254 distinct plant species. The database encompasses exhaustive details including isolation parts, comprehensive molecule structures, chemical abstracts service registry number (CAS number) and 7508 content descriptions. The TPCN database accentuates both the qualitative and quantitative dimensions as invaluable phenotypic characteristics of natural products that have undergone genetic evolution. By acting as an indispensable criterion, the TPCN database facilitates the discovery of drug alternatives with high content and the selection of high-yield medicinal plant species or phylogenetic alternatives, thereby fostering sustainable, cost-effective and environmentally friendly drug discovery in pharmaceutical farming. Database URL: http://www.tpcn.pro/.

The impact of oxygen on metabolic evolution: a chemoinformatic investigation.

The appearance of planetary oxygen likely transformed the chemical and biochemical makeup of life and probably triggered episodes of organismal diversification. Here we use chemoinformatic methods to explore the impact of the rise of oxygen on metabolic evolution. We undertake a comprehensive comparative analysis of structures, chemical properties and chemical reactions of anaerobic and aerobic metabolites. The results indicate that aerobic metabolism has expanded the structural and chemical space of metabolites considerably, including the appearance of 130 novel molecular scaffolds. The molecular functions of these metabolites are mainly associated with derived aspects of cellular life, such as signal transfer, defense against biotic factors, and protection of organisms from oxidation. Moreover, aerobic metabolites are more hydrophobic and rigid than anaerobic compounds, suggesting they are better fit to modulate membrane functions and to serve as transmembrane signaling factors. Since higher organisms depend largely on sophisticated membrane-enabled functions and intercellular signaling systems, the metabolic developments brought about by oxygen benefit the diversity of cellular makeup and the complexity of cellular organization as well. These findings enhance our understanding of the molecular link between oxygen and evolution. They also show the significance of chemoinformatics in addressing basic biological questions.

Exploring the biologically relevant chemical space for drug discovery.

Both recent studies and our calculation suggest that the physicochemical properties of launched drugs changed continuously over the past decades. Besides shifting of commonly used properties, the average biological relevance (BR) and similarity to natural products (NPs) of launched drugs decreased, reflecting the fact that current drug discovery deviated away from NPs. To change the current situation characterized by high investment but low productivity in drug discovery, efforts should be made to improve the BR of the screening library and hunt drugs more effectively in the biologically relevant chemical space. Additionally, a multiple dimensional molecular descriptor, named the biologically relevant spectrum (BRS) was proposed for quantitative structure-activity relationships (QSAR) study or screening library preparation. Prediction models for 43 biological activity categories were developed with BRS and support vector machine (SVM). In most cases, the overall prediction accuracies were around 95% and the Matthew's correlation coefficients (MCC) were over 0.8. Thirty-seven out of 48 drug-activity associations were successfully predicted for drugs that launched from 2006 to 2012, which were not included in the training data set. A web-server named BioRel ( http://ibi.hzau.edu.cn/biorel ) was developed to provide services including BR, BRS calculation, activity class, and pharmacokinetic property prediction.

Sequence variety in the CC' loop of Siglec-8/9/3 determines the recognitions to sulfated oligosaccharides.

Siglecs are important lectins found in different types of immune cells and function as regulatory molecules by recognizing self-associated glycans and converting extracellular interactions into signals for inhibiting immune cell functions. Although many Siglecs have been found to show broad specificities and recognize different types of sulfated oligosaccharides, Siglec-8 and Siglec-9 displayed a high degree of specificity for sialyl N-acetyllactosamine (sLacNAc) with sulfations at O6-positions of the galactose (6'-sulfation) and N-acetylglucosamine (6-sulfation), respectively. Siglec-3 was recently discovered to bind sLacNAc both sulfations. In addition to a conserved arginine residue for binding to sialic acid residue, the sequence variety in the CC' loop may provide binding specificities to sulfated oligosaccharides in Siglecs. Thus, the present study employed molecular models to study the impact of different residues in the CC' loops of Siglec-8/9/3 to the recognitions of 6-sulfations in Gal and/or GlcNAc of sLacNAc. The negatively charged residues in the CC' loop of Siglec-9 formed unfavorable electrostatic repulsions with the 6-sulfate in Gal and resulted no recognitions, in contrast to the favorable interactions formed between the positively charged residues in the CC' loop of Siglec-8 and the 6-sulfate in Gal resulting strong specificity. A two-state binding model was proposed for Siglec-3 recognizing 6-sulfations in Gal and GlcNAc of sLacNAc, as the neutral residues in the CC' loop of Siglec-3 could not form strong favorable interactions to lock the 6-sulfate in Gal within a single binding pose or strong unfavorable interactions to repel the 6-sulfate in Gal. The oligosaccharide adopted two distinctive binding poses and oriented the sulfate groups to form interactions with residues in the CC' loop and G-strand. The present study provided a structural mechanism for the sequence variety in the CC' loop of Siglec-8/9/3 determining the recognitions to the sulfated oligosaccharides and offered insights into the binding specificities for Siglecs.

Yin/Yang associated differential responses to Psoralea corylifolia Linn. In rat models: an integrated metabolomics and transcriptomics study.

ETHNOPHARMACOLOGICAL RELEVANCE: Psoralea corylifolia Linn. (BGZ) is a commonly used traditional Chinese medicine (TCM) for the treatment of kidney-yang deficiency syndrome (Yangsyn) with good curative effect and security. However, BGZ was also reported to induce liver injury in recent years. According to TCM theory, taking BGZ may induce a series of adverse reactions in patients with kidney-yin deficiency syndrome (Yinsyn), which suggests that BGZ-induced liver damage may be related to its unreasonable clinical use. AIM OF THE STUDY: Liver injury caused by TCM is a rare but potentially serious adverse drug reaction, and the identification of predisposed individuals for drug-induced liver injury (DILI) remains challenging. The study aimed to investigate the differential responses to BGZ in Yangsyn and Yinsyn rat models and identify the corresponding characteristic biomarkers. MATERIALS AND METHODS: The corresponding animal models of Yangsyn and Yinsyn were induced by hydrocortisone and thyroxine + reserpine respectively. Body weight, organ index, serum biochemistry, and Hematoxylin and Eosin (HE) staining were used to evaluate the liver toxicity effect of BGZ on rats with Yangsyn and Yinsyn. Transcriptomics and metabonomics were used to screen the representative biomarkers (including metabolites and differentially expressed genes (DEGs)) changed by BGZ in Yangsyn and Yinsyn rats, respectively. RESULTS: The level changes of liver organ index, alanine aminotransferase (ALT), and aspartate aminotransferase (AST), suggested that BGZ has liver-protective and liver-damaging effects on Yangsyn and Yinsyn rats, respectively, and the results also were confirmed by the pathological changes of liver tissue. The results showed that 102 DEGs and 27 metabolites were significantly regulated related to BGZ's protective effect on Yangsyn, which is mainly associated with the glycerophospholipid metabolism, arachidonic acid metabolism, pantothenate, and coenzyme A (CoA) biosynthesis pathways. While 28 DEGs and 31 metabolites, related to the pathway of pantothenate and CoA biosynthesis, were significantly regulated for the BGZ-induced liver injury in Yinsyn. Furthermore, 4 DEGs (aldehyde dehydrogenase 1 family member B1 (Aldh1b1), solute carrier family 25 member 25 (Slc25a25), Pim-3 proto-oncogene, serine/threonine kinase (Pim3), out at first homolog (Oaf)) and 4 metabolites (phosphatidate, phosphatidylcholine, N-Acetylleucine, biliverdin) in the Yangsyn group and 1 DEG [galectin 5 (Lgals5)] and 1 metabolite (5-amino-1-(5-phospho-D-ribosyl)imidazole-4-carboxylate) in Yinsyn group were significantly correlated to the ALT and AST levels of BGZ treated and untreated groups (receiver operating characteristic (ROC) ≥ 0.9). CONCLUSIONS: Yinsyn and Yangsyn are the predisposed syndromes for BGZ to exert liver damage and liver protection respectively, which are mainly related to the regulation of amino acid metabolism, lipid metabolism, energy metabolism, and metabolism of cofactors and vitamins. The results further suggest that attention should be paid to the selection of predisposed populations when using drugs related to the regulation of energy metabolism, and the Yinsyn/Yangsyn animal models based on the theory of TCM syndromes may be a feasible method for identifying the susceptible population to receive TCM.

Chemical basis of metabolic network organization.

Although the metabolic networks of the three domains of life consist of different constituents and metabolic pathways, they exhibit the same scale-free organization. This phenomenon has been hypothetically explained by preferential attachment principle that the new-recruited metabolites attach preferentially to those that are already well connected. However, since metabolites are usually small molecules and metabolic processes are basically chemical reactions, we speculate that the metabolic network organization may have a chemical basis. In this paper, chemoinformatic analyses on metabolic networks of Kyoto Encyclopedia of Genes and Genomes (KEGG), Escherichia coli and Saccharomyces cerevisiae were performed. It was found that there exist qualitative and quantitative correlations between network topology and chemical properties of metabolites. The metabolites with larger degrees of connectivity (hubs) are of relatively stronger polarity. This suggests that metabolic networks are chemically organized to a certain extent, which was further elucidated in terms of high concentrations required by metabolic hubs to drive a variety of reactions. This finding not only provides a chemical explanation to the preferential attachment principle for metabolic network expansion, but also has important implications for metabolic network design and metabolite concentration prediction.