Ziwei Dai

Deep Learning for Drug-Induced Liver Injury

Drug-induced liver injury (DILI) has been the single most frequent cause of safety-related drug marketing withdrawals for the past 50 years. Recently, deep learning (DL) has been successfully applied in many fields due to its exceptional and automatic learning ability. In this study, DILI prediction models were developed using DL architectures, and the best model trained on 475 drugs predicted an external validation set of 198 drugs with an accuracy of 86.9%, sensitivity of 82.5%, specificity of 92.9%, and area under the curve of 0.955, which is better than the performance of previously described DILI prediction models. Furthermore, with deep analysis, we also identified important molecular features that are related to DILI. Such DL models could improve the prediction of DILI risk in humans. The DL DILI prediction models are freely available at http://www.repharma.cn/DILIserver/DILI_home.php.

The impact of cellular metabolism on chromatin dynamics and epigenetics

The substrates used to modify nucleic acids and chromatin are affected by nutrient availability and the activity of metabolic pathways. Thus, cellular metabolism constitutes a fundamental component of chromatin status and thereby of genome regulation. Here we describe the biochemical and genetic principles of how metabolism can influence chromatin biology and epigenetics, discuss the functional roles of this interplay in developmental and cancer biology, and present future directions in this rapidly emerging area.

A Flux Balance of Glucose Metabolism Clarifies the Requirements of the Warburg Effect

The Warburg effect, or aerobic glycolysis, is marked by the increased metabolism of glucose to lactate in the presence of oxygen. Despite its widespread prevalence in physiology and cancer biology, the causes and consequences remain incompletely understood. Here, we show that a simple balance of interacting fluxes in glycolysis creates constraints that impose the necessary conditions for glycolytic flux to generate lactate as opposed to entering into the mitochondria. These conditions are determined by cellular redox and energy demands. By analyzing the constraints and sampling the feasible region of the model, we further study how cell proliferation rate and mitochondria-associated NADH oxidizing and ATP producing fluxes are interlinked. Together this analysis illustrates the simplicity of the origins of the Warburg effect by identifying the flux distributions that are necessary for its instantiation.

Understanding metabolism with flux analysis: From theory to application.

Quantitative and qualitative knowledge of metabolic rates (i.e. fluxes) over a metabolic network and in specific cellular compartments gives insights into the regulation of metabolism and helps to understand the contribution of metabolic alterations to pathology. In this review we introduce methodology to resolve metabolic fluxes from stable isotope labeling and relevant techniques in model development, model simplification, flux uncertainty analysis and experimental design that together is termed metabolic flux analysis. Finally we discuss applications using metabolic flux analysis to elucidate mechanisms pertinent to tumor cell metabolism. We hope that this review gives the readers a brief introduction of how flux analysis is conducted, how technical issues related to it are addressed, and how its application has contributed to our knowledge of tumor cell metabolism.

Methionine metabolism influences genomic architecture and gene expression through H3K4me3 peak width

Nutrition and metabolism are known to influence chromatin biology and epigenetics through post-translational modifications, yet how this interaction influences genomic architecture and connects to gene expression is unknown. Here we consider, as a model, the metabolically-driven dynamics of H3K4me3, a histone methylation mark that is known to encode information about active transcription, cell identity, and tumor suppression. We analyze the genome-wide changes in H3K4me3 and gene expression in response to alterations in methionine availability in both normal mouse physiology and human cancer cells. Surprisingly, we find that the location of H3K4me3 peaks is largely preserved under methionine restriction, while the response of H3K4me3 peak width encodes almost all aspects of H3K4me3 biology including changes in expression levels, and the presence of cell identity and cancer-associated genes. These findings may reveal general principles for how nutrient availability modulates specific aspects of chromatin dynamics to mediate biological function.Methionine availability is known to affect the global levels of histone methylation. Here the authors investigate the metabolically driven dynamics of H3K4me3 and find that methionine availability influences peak width, which is linked to changes in expression of genes associated with cell fate and cancer.

Metabolic pattern formation in the tumor microenvironment.

Metabolic alterations including increased glycolysis are a common feature of many cancers. In their recent study, Lowengrub, Waterman, and colleagues (Lee et al, ) report a spatial pattern of glycolysis in solid tumors that occurs within the tumor microenvironment. This spatial organization is linked to gradients derived from Wnt signaling and nutrient availability that mediate a reaction‐diffusion mechanism and is consistent with a Turing‐type model of spatial localization.

Environmental factors shape methionine metabolism in p16/MTAP deleted cells

The co-deletion of a common tumor suppressor locus and neighboring metabolic gene is an attractive possible synthetic dependency of tumor suppression on metabolism. However, the general impact that these co-deletions have on metabolism, which also dependent on nutrient availability and the tissue of origin, is unknown. As a model to investigate this question, we considered a set of tissue-matched cancer cells with homozygous co-deletions in CDKN2a and MTAP, genes respectively encoding the most commonly deleted tumor suppressor p16 and an enzyme involved in methionine metabolism. A comparative metabolomics analysis revealed that while there exists a definite pan-cancer metabolic signature of MTAP-deletion, this signature was not preserved when cells were subjected to changes in the availability of methionine, serine, or cysteine, nutrients related to methionine metabolism. Notably, the heterogeneity exhibited by these cells in their responsiveness to nutrient availability dominated both MTAP status and tissue-of-origin. Furthermore, re-expression of MTAP exerted a modest effect on metabolism. Together these findings demonstrate that environmental factors, particularly nutrition and tissue identity, may overwhelm the genetic effects of collateral deletions of metabolic genes.

Methionine metabolism influences the genomic architecture of H3K4me3 with the link to gene expression encoded in peak width

Nutrition and metabolism are known to influence chromatin biology and epigenetics by modifying the levels of post-translational modifications on histones, yet how changes in nutrient availability influence specific aspects of genomic architecture and connect to gene expression is unknown. To investigate this question we considered, as a model, the metabolically-driven dynamics of H3K4me3, a histone methylation mark that is known to encode information about active transcription, cell identity, and tumor suppression. We analyzed the genome-wide changes in H3K4me3 and gene expression in response to alterations in methionine availability under conditions that are known to affect the global levels of histone methylation in both normal rodent physiology and in human cancer cells. Surprisingly, we found that the location of H3K4me3 peaks at specific genomic loci was largely preserved under conditions of methionine restriction. However, upon examining different geometrical features of peak shape, it was found that the response of H3K4me3 peak width encoded almost all aspects of H3K4me3 biology including changes in expression levels, and the presence of cell identity and cancer associated genes. These findings reveal simple yet new and profound principles for how nutrient availability modulates specific aspects of chromatin dynamics to mediate key biological features.

Identification of Cancer-associated Metabolic Vulnerabilities by Modeling Multi-objective Optimality in Metabolism

Computational modeling of the genome-wide metabolic network is essential for designing new therapeutics targeting cancer-associated metabolic disorder, which is a hallmark of human malignancies. However, previous studies generally assumed that metabolic fluxes of cancer cells are subjected to the maximization of biomass production, despite the wide existence of trade-offs among multiple metabolic objectives. To address this issue, we developed a multi-objective model of cancer metabolism with algorithms depicting approximate Pareto surfaces and incorporating multiple omics datasets. To validate this approach, we built individualized models for NCI-60 cancer cell lines, and accurately predicted cell growth rates and other biological consequences of metabolic perturbations in these cells. By analyzing the landscape of approximate Pareto surface, we identified a list of metabolic targets essential for cancer cell proliferation and the Warburg effect, and further demonstrated their close association with cancer patient survival. Finally, metabolic targets predicted to be essential for tumor progression were validated by cell-based experiments, confirming this multi-objective modelling method as a novel and effective strategy to identify cancer-associated metabolic vulnerabilities.