The Japanese toxicogenomics project: application of toxicogenomics

Reconciled rat and human metabolic networks for comparative toxicogenomics and biomarker predictions

The laboratory rat has been used as a surrogate to study human biology for more than a century. Here we present the first genome-scale network reconstruction of Rattus norvegicus metabolism, iRno, and a significantly improved reconstruction of human metabolism, iHsa. These curated models comprehensively capture metabolic features known to distinguish rats from humans including vitamin C and bile acid synthesis pathways. After reconciling network differences between iRno and iHsa, we integrate toxicogenomics data from rat and human hepatocytes, to generate biomarker predictions in response to 76 drugs. We validate comparative predictions for xanthine derivatives with new experimental data and literature-based evidence delineating metabolite biomarkers unique to humans. Our results provide mechanistic insights into species-specific metabolism and facilitate the selection of biomarkers consistent with rat and human biology. These models can serve as powerful computational platforms for contextualizing experimental data and making functional predictions for clinical and basic science applications.

The rat is a widely-used model for human biology, but we must be aware of metabolic differences. Here, the authors reconstruct the genome-scale metabolic network of the rat, and after reconciling it with an improved human metabolic model, demonstrate the power of the models to integrate toxicogenomics data, providing species-specific biomarker predictions in response to a panel of drugs.

Prediction of drug-induced liver injury using keratinocytes

Drug-induced liver injury (DILI) is one of the most common adverse drug reactions. DILI is often accompanied by skin reactions, including rash and pruritus. However, it is still unknown whether DILI-associated genes such as S100 calcium-binding protein A and interleukin (IL)-1β are involved in drug-induced skin toxicity. In the present study, most of the tested hepatotoxic drugs such as pioglitazone and diclofenac induced DILI-associated genes in human and mouse keratinocytes. Keratinocytes of mice at higher risk for DILI exhibited an increased IL-1β basal expression. They also showed a higher inducibility of IL-1β when treated by pioglitazone. Mice at higher risk for DILI showed even higher sums of DILI-associated gene basal expression levels and induction rates in keratinocytes. Our data suggest that DILI-associated genes might be involved in the onset and progression of drug-induced skin toxicity. Furthermore, we might be able to identify individuals at higher risk of developing DILI less invasively by examining gene expression patterns in keratinocytes. Copyright © 2017 John Wiley & Sons, Ltd.

Stem Cell Transcriptome Responses and Corresponding Biomarkers That Indicate the Transition from Adaptive Responses to Cytotoxicity

Analysis of transcriptome changes has become an established method to characterize the reaction of cells to toxicants. Such experiments are mostly performed at compound concentrations close to the cytotoxicity threshold. At present, little information is available on concentration-dependent features of transcriptome changes, in particular, at the transition from noncytotoxic concentrations to conditions that are associated with cell death. Thus, it is unclear in how far cell death confounds the results of transcriptome studies. To explore this gap of knowledge, we treated pluripotent stem cells differentiating to human neuroepithelial cells (UKN1 assay) for short periods (48 h) with increasing concentrations of valproic acid (VPA) and methyl mercury (MeHg), two compounds with vastly different modes of action. We developed various visualization tools to describe cellular responses, and the overall response was classified as "tolerance" (minor transcriptome changes), "functional adaptation" (moderate/strong transcriptome responses, but no cytotoxicity), and "degeneration". The latter two conditions were compared, using various statistical approaches. We identified (i) genes regulated at cytotoxic, but not at noncytotoxic, concentrations and (ii) KEGG pathways, gene ontology term groups, and superordinate biological processes that were only regulated at cytotoxic concentrations. The consensus markers and processes found after 48 h treatment were then overlaid with those found after prolonged (6 days) treatment. The study highlights the importance of careful concentration selection and of controlling viability for transcriptome studies. Moreover, it allowed identification of 39 candidate "biomarkers of cytotoxicity". These could serve to provide alerts that data sets of interest may have been affected by cell death in the model system studied.

Application of dynamic topic models to toxicogenomics data

Background

All biological processes are inherently dynamic. Biological systems evolve transiently or sustainably according to sequential time points after perturbation by environment insults, drugs and chemicals. Investigating the temporal behavior of molecular events has been an important subject to understand the underlying mechanisms governing the biological system in response to, such as, drug treatment. The intrinsic complexity of time series data requires appropriate computational algorithms for data interpretation. In this study, we propose, for the first time, the application of dynamic topic models (DTM) for analyzing time-series gene expression data.

Results

A large time-series toxicogenomics dataset was studied. It contains over 3144 microarrays of gene expression data corresponding to rat livers treated with 131 compounds (most are drugs) at two doses (control and high dose) in a repeated schedule containing four separate time points (4-, 8-, 15- and 29-day). We analyzed, with DTM, the topics (consisting of a set of genes) and their biological interpretations over these four time points. We identified hidden patterns embedded in this time-series gene expression profiles. From the topic distribution for compound-time condition, a number of drugs were successfully clustered by their shared mode-of-action such as PPARɑ agonists and COX inhibitors. The biological meaning underlying each topic was interpreted using diverse sources of information such as functional analysis of the pathways and therapeutic uses of the drugs. Additionally, we found that sample clusters produced by DTM are much more coherent in terms of functional categories when compared to traditional clustering algorithms.

Conclusions

We demonstrated that DTM, a text mining technique, can be a powerful computational approach for clustering time-series gene expression profiles with the probabilistic representation of their dynamic features along sequential time frames. The method offers an alternative way for uncovering hidden patterns embedded in time series gene expression profiles to gain enhanced understanding of dynamic behavior of gene regulation in the biological system.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-016-1225-0) contains supplementary material, which is available to authorized users.

Can a systems approach produce a better understanding of mood disorders?

BACKGROUND

One in twenty-five people suffer from a mood disorder. Current treatments are sub-optimal with poor patient response and uncertain modes-of-action. There is thus a need to better understand underlying mechanisms that determine mood, and how these go wrong in affective disorders. Systems biology approaches have yielded important biological discoveries for other complex diseases such as cancer, and their potential in affective disorders will be reviewed.

SCOPE OF REVIEW

This review will provide a general background to affective disorders, plus an outline of experimental and computational systems biology. The current application of these approaches in understanding affective disorders will be considered, and future recommendations made.

MAJOR CONCLUSIONS

Experimental systems biology has been applied to the study of affective disorders, especially at the genome and transcriptomic levels. However, data generation has been slowed by a lack of human tissue or suitable animal models. At present, computational systems biology has only be applied to understanding affective disorders on a few occasions. These studies provide sufficient novel biological insight to motivate further use of computational biology in this field.

GENERAL SIGNIFICANCE

In common with many complex diseases much time and money has been spent on the generation of large-scale experimental datasets. The next step is to use the emerging computational approaches, predominantly developed in the field of oncology, to leverage the most biological insight from these datasets. This will lead to the critical breakthroughs required for more effective diagnosis, stratification and treatment of affective disorders.

Connection Map for Compounds (CMC): A Server for Combinatorial Drug Toxicity and Efficacy Analysis

Drug discovery and development is a costly and time-consuming process with a high risk for failure resulting primarily from a drug's associated clinical safety and efficacy potential. Identifying and eliminating inapt candidate drugs as early as possible is an effective way for reducing unnecessary costs, but limited analytical tools are currently available for this purpose. Recent growth in the area of toxicogenomics and pharmacogenomics has provided with a vast amount of drug expression microarray data. Web servers such as CMap and LTMap have used this information to evaluate drug toxicity and mechanisms of action independently; however, their wider applicability has been limited by the lack of a combinatorial drug-safety type of analysis. Using available genome-wide drug transcriptional expression profiles, we developed the first web server for combinatorial evaluation of toxicity and efficacy of candidate drugs named "Connection Map for Compounds" (CMC). Using CMC, researchers can initially compare their query drug gene signatures with prebuilt gene profiles generated from two large-scale toxicogenomics databases, and subsequently perform a drug efficacy analysis for identification of known mechanisms of drug action or generation of new predictions. CMC provides a novel approach for drug repositioning and early evaluation in drug discovery with its unique combination of toxicity and efficacy analyses, expansibility of data and algorithms, and customization of reference gene profiles. CMC can be freely accessed at http://cadd.tongji.edu.cn/webserver/CMCbp.jsp .

ToxDB: pathway-level interpretation of drug-treatment data

Motivation: Extensive drug treatment gene expression data have been generated in order to identify biomarkers that are predictive for toxicity or to classify compounds. However, such patterns are often highly variable across compounds and lack robustness. We and others have previously shown that supervised expression patterns based on pathway concepts rather than unsupervised patterns are more robust and can be used to assess toxicity for entire classes of drugs more reliably.

Results: We have developed a database, ToxDB, for the analysis of the functional consequences of drug treatment at the pathway level. We have collected 2694 pathway concepts and computed numerical response scores of these pathways for 437 drugs and chemicals and 7464 different experimental conditions. ToxDB provides functionalities for exploring these pathway responses by offering tools for visualization and differential analysis allowing for comparisons of different treatment parameters and for linking this data with toxicity annotation and chemical information.

Database URL:http://toxdb.molgen.mpg.de

Mind the Gap! A Journey towards Computational Toxicology

Computational methods have advanced toxicology towards the development of target-specific models based on a clear cause-effect rationale. However, the predictive potential of these models presents strengths and weaknesses. On the good side, in silico models are valuable cheap alternatives to in vitro and in vivo experiments. On the other, the unconscious use of in silico methods can mislead end-users with elusive results. The focus of this review is on the basic scientific and regulatory recommendations in the derivation and application of computational models. Attention is paid to examine the interplay between computational toxicology and drug discovery and development. Avoiding the easy temptation of an overoptimistic future, we report our view on what can, or cannot, realistically be done. Indeed, studies of safety/toxicity represent a key element of chemical prioritization programs carried out by chemical industries, and primarily by pharmaceutical companies.

Characterization of Conserved Toxicogenomic Responses in Chemically Exposed Hepatocytes across Species and Platforms

BACKGROUND

Genome-wide expression profiling is increasingly being used to identify transcriptional changes induced by drugs and environmental stressors. In this context, the Toxicogenomics Project-Genomics Assisted Toxicity Evaluation system (TG-GATEs) project generated transcriptional profiles from rat liver samples and human/rat cultured primary hepatocytes exposed to more than 100 different chemicals.

OBJECTIVES

To assess the capacity of the cell culture models to recapitulate pathways induced by chemicals in vivo, we leveraged the TG-GATEs data set to compare the early transcriptional responses observed in the liver of rats treated with a large set of chemicals with those of cultured rat and human primary hepatocytes challenged with the same compounds in vitro.

METHODS

We developed a new pathway-based computational pipeline that efficiently combines gene set enrichment analysis (GSEA) using pathways from the Reactome database with biclustering to identify common modules of pathways that are modulated by several chemicals in vivo and in vitro across species.

RESULTS

We found that some chemicals induced conserved patterns of early transcriptional responses in in vitro and in vivo settings, and across human and rat genomes. These responses involved pathways of cell survival, inflammation, xenobiotic metabolism, oxidative stress, and apoptosis. Moreover, our results support the transforming growth factor beta receptor (TGF-βR) signaling pathway as a candidate biomarker associated with exposure to environmental toxicants in primary human hepatocytes.

CONCLUSIONS

Our integrative analysis of toxicogenomics data provides a comprehensive overview of biochemical perturbations affected by a large panel of chemicals. Furthermore, we show that the early toxicological response occurring in animals is recapitulated in human and rat primary hepatocyte cultures at the molecular level, indicating that these models reproduce key pathways in response to chemical stress. These findings expand our understanding and interpretation of toxicogenomics data from human hepatocytes exposed to environmental toxicants.

ZBIT Bioinformatics Toolbox: A Web-Platform for Systems Biology and Expression Data Analysis

Bioinformatics analysis has become an integral part of research in biology. However, installation and use of scientific software can be difficult and often requires technical expert knowledge. Reasons are dependencies on certain operating systems or required third-party libraries, missing graphical user interfaces and documentation, or nonstandard input and output formats. In order to make bioinformatics software easily accessible to researchers, we here present a web-based platform. The Center for Bioinformatics Tuebingen (ZBIT) Bioinformatics Toolbox provides web-based access to a collection of bioinformatics tools developed for systems biology, protein sequence annotation, and expression data analysis. Currently, the collection encompasses software for conversion and processing of community standards SBML and BioPAX, transcription factor analysis, and analysis of microarray data from transcriptomics and proteomics studies. All tools are hosted on a customized Galaxy instance and run on a dedicated computation cluster. Users only need a web browser and an active internet connection in order to benefit from this service. The web platform is designed to facilitate the usage of the bioinformatics tools for researchers without advanced technical background. Users can combine tools for complex analyses or use predefined, customizable workflows. All results are stored persistently and reproducible. For each tool, we provide documentation, tutorials, and example data to maximize usability. The ZBIT Bioinformatics Toolbox is freely available at https://webservices.cs.uni-tuebingen.de/.

Supporting read-across using biological data

Read-across, i.e. filling toxicological data gaps by relating to similar chemicals, for which test data are available, is usually done based on chemical similarity. Besides structure and physico-chemical properties, however, biological similarity based on biological data adds extra strength to this process. In the context of developing Good Read-Across Practice guidance, a number of case studies were evaluated to demonstrate the use of biological data to enrich read-across. In the simplest case, chemically similar substances also show similar test results in relevant in vitro assays. This is a well-established method for the read-across of e.g. genotoxicity assays. Larger datasets of biological and toxicological properties of hundreds and thousands of substances become increasingly available enabling big data approaches in read-across studies. Several case studies using various big data sources are described in this paper. An example is given for the US EPA's ToxCast dataset allowing read-across for high quality uterotrophic assays for estrogenic endocrine disruption. Similarly, an example for REACH registration data enhancing read-across for acute toxicity studies is given. A different approach is taken using omics data to establish biological similarity: Examples are given for stem cell models in vitro and short-term repeated dose studies in rats in vivo to support read-across and category formation. These preliminary biological data-driven read-across studies highlight the road to the new generation of read-across approaches that can be applied in chemical safety assessment.

Toward Good Read-Across Practice (GRAP) guidance

Grouping of substances and utilizing read-across of data within those groups represents an important data gap filling technique for chemical safety assessments. Categories/analogue groups are typically developed based on structural similarity and, increasingly often, also on mechanistic (biological) similarity. While read-across can play a key role in complying with legislation such as the European REACH regulation, the lack of consensus regarding the extent and type of evidence necessary to support it often hampers its successful application and acceptance by regulatory authorities. Despite a potentially broad user community, expertise is still concentrated across a handful of organizations and individuals. In order to facilitate the effective use of read-across, this document presents the state of the art, summarizes insights learned from reviewing ECHA published decisions regarding the relative successes/pitfalls surrounding read-across under REACH, and compiles the relevant activities and guidance documents. Special emphasis is given to the available existing tools and approaches, an analysis of ECHA's published final decisions associated with all levels of compliance checks and testing proposals, the consideration and expression of uncertainty, the use of biological support data, and the impact of the ECHA Read-Across Assessment Framework (RAAF) published in 2015.

In Silico Models for Hepatotoxicity

In this chapter we review the challenges of predicting human hepatotoxicity. Principally, this is our partial understanding of a very complex biochemical system and our ability to emulate that in a predictive capacity. We give an overview of the published modeling approaches in this area to date and discuss their design, strengths, and weaknesses. It is interesting to note the shift during the period of this review in the direction of evidenced-based approaches including structural alerts and pharmacophore models. Proposals on how best to utilize the data emerging from modeling studies are also discussed.

Discovery of Transcriptional Targets Regulated by Nuclear Receptors Using a Probabilistic Graphical Model

Nuclear receptors (NRs) are ligand-activated transcriptional regulators that play vital roles in key biological processes such as growth, differentiation, metabolism, reproduction, and morphogenesis. Disruption of NRs can result in adverse health effects such as NR-mediated endocrine disruption. A comprehensive understanding of core transcriptional targets regulated by NRs helps to elucidate their key biological processes in both toxicological and therapeutic aspects. In this study, we applied a probabilistic graphical model to identify the transcriptional targets of NRs and the biological processes they govern. The Tox21 program profiled a collection of approximate 10 000 environmental chemicals and drugs against a panel of human NRs in a quantitative high-throughput screening format for their NR disruption potential. The Japanese Toxicogenomics Project, one of the most comprehensive efforts in the field of toxicogenomics, generated large-scale gene expression profiles on the effect of 131 compounds (in its first phase of study) at various doses, and different durations, and their combinations. We applied author-topic model to these 2 toxicological datasets, which consists of 11 NRs run in either agonist and/or antagonist mode (18 assays total) and 203 in vitro human gene expression profiles connected by 52 shared drugs. As a result, a set of clusters (topics), which consists of a set of NRs and their associated target genes were determined. Various transcriptional targets of the NRs were identified by assays run in either agonist or antagonist mode. Our results were validated by functional analysis and compared with TRANSFAC data. In summary, our approach resulted in effective identification of associated/affected NRs and their target genes, providing biologically meaningful hypothesis embedded in their relationships.

Toxicogenomics-based prediction of acetaminophen-induced liver injury using human hepatic cell systems

Primary human hepatocytes (hHEP), human HepaRG and HepG2 cell lines are the most used human liver-based in vitro models for hepatotoxicity testing, including screening of drug-induced liver injury (DILI)-inducing compounds. hHEP are the reference hepatic in vitro system, but their availability is limited and the cells available for toxicology studies are often of poor quality. Hepatic cell lines on the other hand are highly proliferative and represent an inexhaustible hepatic cell source. However, these hepatoma-derived cells do not represent the population diversity and display reduced hepatic metabolism. Alternatively, stem cell-derived hepatic cells, which can be produced in high numbers and can differentiate into multiple cell lineages, are also being evaluated as a cell source for in vitro hepatotoxicity studies. Human skin-derived precursors (hSKP) are post-natal stem cells that, after conversion towards hepatic cells (hSKP-HPC), respond to hepatotoxic compounds in a comparable way as hHEP. In the current study, four different human hepatic cell systems (hSKP-HPC, hHEP, HepaRG and HepG2) are evaluated for their capacity to predict hepatic toxicity. Their hepatotoxic response to acetaminophen (APAP) exposure is compared to data obtained from patients suffering from APAP-induced acute liver failure (ALF). The results indicate that hHEP, HepaRG and hSKP-HPC identify comparable APAP-induced hepatotoxic functions and that HepG2 cells show the slightest hepatotoxic response. Pathway analyses further points out that HepaRG cells show the highest predicted activation of the functional genes related to 'damage of liver', followed by hSKP-HPC and hHEP cells that generated similar results. HepG2 did not show any activation of this function.

In silico ADME/T modelling for rational drug design

In recent decades, in silico absorption, distribution, metabolism, excretion (ADME), and toxicity (T) modelling as a tool for rational drug design has received considerable attention from pharmaceutical scientists, and various ADME/T-related prediction models have been reported. The high-throughput and low-cost nature of these models permits a more streamlined drug development process in which the identification of hits or their structural optimization can be guided based on a parallel investigation of bioavailability and safety, along with activity. However, the effectiveness of these tools is highly dependent on their capacity to cope with needs at different stages, e.g. their use in candidate selection has been limited due to their lack of the required predictability. For some events or endpoints involving more complex mechanisms, the current in silico approaches still need further improvement. In this review, we will briefly introduce the development of in silico models for some physicochemical parameters, ADME properties and toxicity evaluation, with an emphasis on the modelling approaches thereof, their application in drug discovery, and the potential merits or deficiencies of these models. Finally, the outlook for future ADME/T modelling based on big data analysis and systems sciences will be discussed.

Identification of in vitro and in vivo disconnects using transcriptomic data

BACKGROUND

Integrating transcriptomic experiments within drug development is increasingly advocated for the early detection of toxicity. This is partly to reduce costs related to drug failures in the late, and expensive phases of clinical trials. Such an approach has proven useful both in the study of toxicology and carcinogenicity. However, general lack of translation of in vitro findings to in vivo systems remains one of the bottle necks in drug development. This paper proposes a method for identifying disconnected genes between in vitro and in vivo toxicogenomic rat experiments. The analytical framework is based on the joint modeling of dose-dependent in vitro and in vivo data using a fractional polynomial framework and biclustering algorithm.

RESULTS

Most disconnected genes identified belonged to known pathways, such as drug metabolism and oxidative stress due to reactive metabolites, bilirubin increase, glutathion depletion and phospholipidosis. We also identified compounds that were likely to induce disconnect in gene expression between in vitro and in vivo toxicogenomic rat experiments. These compounds include: sulindac and diclofenac (both linked to liver damage), naphtyl isothiocyanate (linked to hepatoxocity), indomethacin and naproxen (linked to gastrointestinal problem and damage of intestines).

CONCLUSION

The results confirmed that there are important discrepancies between in vitro and in vivo toxicogenomic experiments. However, the contribution of this paper is to provide a tool to identify genes that are disconnected between the two systems. Pathway analysis of disconnected genes may improve our understanding of uncertainties in the mechanism of actions of drug candidates in humans, especially concerning the early detection of toxicity.

Toxicogenomic assessment of liver responses following subchronic exposure to furan in Fischer F344 rats

Furan is a widely used industrial chemical and a contaminant in heated foods. Chronic furan exposure causes cholangiocarcinoma and hepatocellular tumors in rats at doses of 2 mg/kg bw/day or greater, with gender differences in frequency and severity. The hepatic transcriptional alterations induced by low doses of furan (doses below those previously tested for induction of liver tumors) and the potential mechanisms underlying gender differences are largely unexplored. We used DNA microarrays to examine the global hepatic mRNA and microRNA transcriptional profiles of male and female rats exposed to 0, 0.03, 0.12, 0.5 or 2 mg/kg bw/day furan over 90 days. Marked gender differences in gene expression responses to furan were observed, with many more altered genes in exposed males than females, confirming the increased sensitivity of males even at the low doses. Pathway analysis supported that key events in furan-induced liver tumors in males include gene expression changes related to oxidative stress, apoptosis and inflammatory response, while pathway changes in females were consistent with primarily adaptive responses. Pathway benchmark doses (BMDs) were estimated and compared to relevant apical endpoints. Transcriptional pathway BMDs could only be examined in males. These median BMDs ranged from 0.08 to 1.43 mg/kg bw/day and approximated those derived from traditional histopathology. MiR-34a (a P53 target) was the only microRNA significantly increased at the 2 mg/kg bw/day, providing evidence to support the importance of apoptosis and cell proliferation in furan hepatotoxicity. Overall, this study demonstrates the use of transcriptional profiling to discern mode of action and mechanisms involved in gender differences.

Chemical biology databases: from aggregation, curation to representation

Systems chemical biology offers a novel way of approaching drug discovery by developing models that consider the global physiological environment of protein targets and their perturbations by drugs. However, the integration of all these data needs curation and standardization with an appropriate representation in order to get relevant interpretations. In this mini review, we present some databases and services, which integrated together with computational tools and data standardization, could assist scientists in decision making during the different drug development process.

Release of (and lessons learned from mining) a pioneering large toxicogenomics database

AIM

We release the Janssen Toxicogenomics database. This rat liver gene-expression database was generated using Codelink microarrays, and has been used over the past years within Janssen to derive signatures for multiple end points and to classify proprietary compounds.

MATERIALS & METHODS

The release consists of gene-expression responses to 124 compounds, selected to give a broad coverage of liver-active compounds. A selection of the compounds were also analyzed on Affymetrix microarrays.

RESULTS

The release includes results of an in-house reannotation pipeline to Entrez gene annotations, to classify probes into different confidence classes. High confidence unambiguously annotated probes were used to create gene-level data which served as starting point for cross-platform comparisons. Connectivity map-based similarity methods show excellent agreement between Codelink and Affymetrix runs of the same samples. We also compared our dataset with the Japanese Toxicogenomics Project and observed reasonable agreement, especially for compounds with stronger gene signatures. We describe an R-package containing the gene-level data and show how it can be used for expression-based similarity searches.

CONCLUSION

Comparing the same biological samples run on the Affymetrix and the Codelink platform, good correspondence is observed using connectivity mapping approaches. As expected, this correspondence is smaller when the data are compared with an independent dataset such as TG-GATE. We hope that this collection of gene-expression profiles will be incorporated in toxicogenomics pipelines of users.

Activation of the Nrf2 response by intrinsic hepatotoxic drugs correlates with suppression of NF-κB activation and sensitizes toward TNFα-induced cytotoxicity

Drug-induced liver injury (DILI) is an important problem both in the clinic and in the development of new safer medicines. Two pivotal adaptation and survival responses to adverse drug reactions are oxidative stress and cytokine signaling based on the activation of the transcription factors Nrf2 and NF-κB, respectively. Here, we systematically investigated Nrf2 and NF-κB signaling upon DILI-related drug exposure. Transcriptomics analyses of 90 DILI compounds in primary human hepatocytes revealed that a strong Nrf2 activation is associated with a suppression of endogenous NF-κB activity. These responses were translated into quantitative high-content live-cell imaging of induction of a selective Nrf2 target, GFP-tagged Srxn1, and the altered nuclear translocation dynamics of a subunit of NF-κB, GFP-tagged p65, upon TNFR signaling induced by TNFα using HepG2 cells. Strong activation of GFP-Srxn1 expression by DILI compounds typically correlated with suppression of NF-κB nuclear translocation, yet reversely, activation of NF-κB by TNFα did not affect the Nrf2 response. DILI compounds that provided strong Nrf2 activation, including diclofenac, carbamazepine and ketoconazole, sensitized toward TNFα-mediated cytotoxicity. This was related to an adaptive primary protective response of Nrf2, since loss of Nrf2 enhanced this cytotoxic synergy with TNFα, while KEAP1 downregulation was cytoprotective. These data indicate that both Nrf2 and NF-κB signaling may be pivotal in the regulation of DILI. We propose that the NF-κB-inhibiting effects that coincide with a strong Nrf2 stress response likely sensitize liver cells to pro-apoptotic signaling cascades induced by intrinsic cytotoxic pro-inflammatory cytokines.

Asymmetric author-topic model for knowledge discovering of big data in toxicogenomics

The advancement of high-throughput screening technologies facilitates the generation of massive amount of biological data, a big data phenomena in biomedical science. Yet, researchers still heavily rely on keyword search and/or literature review to navigate the databases and analyses are often done in rather small-scale. As a result, the rich information of a database has not been fully utilized, particularly for the information embedded in the interactive nature between data points that are largely ignored and buried. For the past 10 years, probabilistic topic modeling has been recognized as an effective machine learning algorithm to annotate the hidden thematic structure of massive collection of documents. The analogy between text corpus and large-scale genomic data enables the application of text mining tools, like probabilistic topic models, to explore hidden patterns of genomic data and to the extension of altered biological functions. In this paper, we developed a generalized probabilistic topic model to analyze a toxicogenomics dataset that consists of a large number of gene expression data from the rat livers treated with drugs in multiple dose and time-points. We discovered the hidden patterns in gene expression associated with the effect of doses and time-points of treatment. Finally, we illustrated the ability of our model to identify the evidence of potential reduction of animal use.

Comparative metabolome analysis of cultured fetal and adult hepatocytes in humans

The liver is the central organ of metabolism, but its function varies during development from fetus to adult. In this study, we comprehensively analyzed and compared metabolites in fetal and adult hepatocytes, the major parenchymal cell in the liver, from human donors. We identified 211 metabolites (116 anions and 95 cations) by capillary electrophoresis-time-of-flight mass spectrometry (CE-TOFMS) in the hepatocytes cultured in vitro. Principal component analysis and hierarchical clustering analysis of the relative amounts of metabolites clearly classified hepatocytes into 2 groups that were consistent with their origin, i.e., the fetus and adult. The amounts of most metabolites in the glycolysis/glyconeogenesis pathway, tricarboxylic acid cycle and urea cycle were lower in fetal hepatocytes than in adult hepatocytes. These results suggest different susceptibility of the fetal and adult liver to toxic insults affecting energy metabolism.

In vitro assessment of drug-induced liver steatosis based on human dermal stem cell-derived hepatic cells

Steatosis, also known as fatty liver disease (FLD), is a disorder in which the lipid metabolism of the liver is disturbed, leading to the abnormal retention of lipids in hepatocytes. FLD can be induced by several drugs, and although it is mostly asymptomatic, it can lead to steatohepatitis, which is associated with liver inflammation and damage. Drug-induced liver injury is currently the major cause of postmarketing withdrawal of pharmaceuticals and discontinuation of the development of new chemical entities. Therefore, the potential induction of steatosis must be evaluated during preclinical drug development. However, robust human-relevant in vitro models are lacking. In the present study, we explore the applicability of hepatic cells (hSKP-HPCs) derived from postnatal skin precursors, a stem cell population residing in human dermis, to investigate the steatosis-inducing effects of sodium valproate (Na-VPA). Exposure of hSKP-HPC to sub-cytotoxic concentrations of this reference steatogenic compound showed an increased intracellular accumulation of lipid droplets, and the modulation of key factors involved in lipid metabolism. Using a toxicogenomics approach, we further compared Na-VPA-treated hSKP-HPC and Na-VPA-treated primary human hepatocytes to liver samples from patients suffering from mild and advanced steatosis. Our data show that in hSKP-HPC exposed to Na-VPA and liver samples of patients suffering from mild steatosis, but not in primary human hepatocytes, "liver steatosis" was efficiently identified as a toxicological response. These findings illustrate the potential of hSKP-HPC as a human-relevant in vitro model to identify hepatosteatotic effects of chemical compounds.

Systematic toxicity mechanism analysis of proton pump inhibitors: an in silico study

Proton pump inhibitors (PPIs) are extensively used for the treatment of gastric acid-related disorders. PPIs appear to be well tolerated and almost have no short-term side effects. However, the clinical adverse reactions of long-term PPI usage are increasingly reported in recent years. So far, there is no study that elucidates the side effect mechanisms of PPIs comprehensively and systematically. In this study, a well-defined small molecule perturbed microarray data set of 344 compounds and 1695 samples was analyzed. With this high-throughput data set, a new index (Identity, I) was designed to identify PPI-specific differentially expressed genes. Results indicated that (1) up-regulated genes, such as RETSAT, CYP1A1, CYP1A2, and UGT, enhanced vitamin A's metabolism processes in the cellular retinol metabolism pathway; and that (2) down-regulated genes, such as C1QA, C1QC, C4BPA, C4BPB, CFI, and SERPING1, enriched in the complement and coagulation cascades pathway. In addition, strong association was observed between these PPI-specific differentially expressed genes and the reported side effects of PPIs by the gene-disease association network analysis. One potential toxicity mechanism of PPIs as suggested from this systematic PPI-specific gene expression analysis is that PPIs are enriched in acidic organelles where they are activated and inhibit V-ATPases and acid hydrolases, and consequently block the pathways of antigen presentation, the synthesis and secretion of cytokines, and complement component proteins and coagulation factors. The strategies developed in this work could be extended to studies on other drugs.

Computational toxicology, friend or foe?

Computational approaches offer the attraction of being both fast and cheap to run being able to process thousands of chemical structures in a few minutes. As with all new technology, there is a tendency for these approaches to be hyped up and claims of reliability and performance may be exaggerated. So just how good are these computational methods?

An integrative model of multi-organ drug-induced toxicity prediction using gene-expression data

Background

In practice, some drugs produce a number of negative biological effects that can mitigate their effectiveness as a remedy. To address this issue, several studies have been performed for the prediction of drug-induced toxicity from gene-expression data, and a significant amount of work has been done on predicting limited drug-induced symptoms or single-organ toxicity. Since drugs often lead to some injuries in several organs like liver or kidney, however, it would be very useful to forecast the drug-induced injuries for multiple organs. Therefore, in this work, our aim was to develop a multi-organ toxicity prediction model using an integrative model of gene-expression data.

Results

To train our integrative model, we used 3708 in-vivo samples of gene-expression profiles exposed to one of 41 drugs related to 21 distinct physiological changes divided between liver and kidney (liver 11, kidney 10). Specifically, we used the gene-expression profiles to learn an ensemble classifier for each of 21 pathology prediction models. Subsequently, these classifiers were combined with weights to generate an integrative model for each pathological finding. The integrative model outputs the likeliness of presenting the trained pathology in a given test sample of gene-expression profile, called an integrative prediction score (IPS). For the evaluation of an integrative model, we estimated the prediction performance with the k-fold cross-validation. Our results demonstrate that the proposed integrative model is superior to individual pathology prediction models in predicting multi-organ drug-induced toxicities over all the targeted pathological findings. On average, the AUC of the integrative models was 88% while the AUC of individual pathology prediction models was 68%.

Conclusions

Not only does this integrative model produce comparable prediction performance to existing approaches, but also it produces very stable performance overall. In addition, our approach is easily expandable to a variety of other multi-organ toxicology applications.

Whole genome transcript profiling of drug induced steatosis in rats reveals a gene signature predictive of outcome

Drug induced steatosis (DIS) is characterised by excess triglyceride accumulation in the form of lipid droplets (LD) in liver cells. To explore mechanisms underlying DIS we interrogated the publically available microarray data from the Japanese Toxicogenomics Project (TGP) to study comprehensively whole genome gene expression changes in the liver of treated rats. For this purpose a total of 17 and 12 drugs which are diverse in molecular structure and mode of action were considered based on their ability to cause either steatosis or phospholipidosis, respectively, while 7 drugs served as negative controls. In our efforts we focused on 200 genes which are considered to be mechanistically relevant in the process of lipid droplet biogenesis in hepatocytes as recently published (Sahini and Borlak, 2014). Based on mechanistic considerations we identified 19 genes which displayed dose dependent responses while 10 genes showed time dependency. Importantly, the present study defined 9 genes (ANGPTL4, FABP7, FADS1, FGF21, GOT1, LDLR, GK, STAT3, and PKLR) as signature genes to predict DIS. Moreover, cross tabulation revealed 9 genes to be regulated ≥10 times amongst the various conditions and included genes linked to glucose metabolism, lipid transport and lipogenesis as well as signalling events. Additionally, a comparison between drugs causing phospholipidosis and/or steatosis revealed 26 genes to be regulated in common including 4 signature genes to predict DIS (PKLR, GK, FABP7 and FADS1). Furthermore, a comparison between in vivo single dose (3, 6, 9 and 24 h) and findings from rat hepatocyte studies (2 h, 8 h, 24 h) identified 10 genes which are regulated in common and contained 2 DIS signature genes (FABP7, FGF21). Altogether, our studies provide comprehensive information on mechanistically linked gene expression changes of a range of drugs causing steatosis and phospholipidosis and encourage the screening of DIS signature genes at the preclinical stage.

A transcriptomics-based hepatotoxicity comparison between the zebrafish embryo and established human and rodent in vitro and in vivo models using cyclosporine A, amiodarone and acetaminophen

The zebrafish embryo (ZFE) is a promising alternative, non-rodent model in toxicology, which has an advantage over the traditionally used models as it contains complete biological complexity and provides a medium to high-throughput setting. Here, we assess how the ZFE compares to the traditionally used models for liver toxicity testing, i.e., in vivo mouse and rat liver, in vitro mouse and rat hepatocytes, and primary human hepatocytes. For this comparison, we analyzed gene expression changes induced by three model compounds for cholestasis, steatosis, and necrosis. The three compounds, cyclosporine A, amiodarone, and acetaminophen, were chosen because of their relevance to human toxicity and these compounds displayed hepatotoxic-specific changes in the mouse in vivo data. Compound induced expression changes in the ZFE model shared similarity with both in vivo and in vitro. Comparison on single gene level revealed the presence of model specific changes and no clear concordance across models. However, concordance was identified on the pathway level. Specifically, the pathway "regulation of metabolism - bile acids regulation of glucose and lipid metabolism via FXR" was affected across all models and compounds. In conclusion, our study with three hepatotoxic model compounds shows that the ZFE model is at least as comparable to traditional models in identifying hepatotoxic activity and has the potential for use as a pre-screen to determine the hepatotoxic potential of compounds.

The opportunities of mining historical and collective data in drug discovery

Vast amounts of bioactivity data have been generated for small molecules across public and corporate domains. Biological signatures, either derived from systematic profiling efforts or from existing historical assay data, have been successfully employed for small molecule mechanism-of-action elucidation, drug repositioning, hit expansion and screening subset design. This article reviews different types of biological descriptors and applications, and we demonstrate how biological data can outlive the original purpose or project for which it was generated. By comparing 150 HTS campaigns run at Novartis over the past decade on the basis of their active and inactive chemical matter, we highlight the opportunities and challenges associated with cross-project learning in drug discovery.

Identification of potential biomarkers of hepatitis B-induced acute liver failure using hepatic cells derived from human skin precursors

Besides their role in the elucidation of pathogenic processes of medical and pharmacological nature, biomarkers can also be used to document specific toxicological events. Hepatic cells generated from human skin-derived precursors (hSKP-HPC) were previously shown to be a promising in vitro tool for the evaluation of drug-induced hepatotoxicity. In this study, their capacity to identify potential liver-specific biomarkers at the gene expression level was investigated with particular emphasis on acute liver failure (ALF). To this end, a set of potential ALF-specific biomarkers was established using clinically relevant liver samples obtained from patients suffering from hepatitis B-associated ALF. Subsequently, this data was compared to data obtained from primary human hepatocyte cultures and hSKP-HPC, both exposed to the ALF-inducing reference compound acetaminophen. It was found that both in vitro systems revealed a set of molecules that was previously identified in the ALF liver samples. Yet, only a limited number of molecules was common between both in vitro systems and the ALF liver samples. Each of the in vitro systems could be used independently to identify potential toxicity biomarkers related to ALF. It seems therefore more appropriate to combine primary human hepatocyte cultures with complementary in vitro models to efficiently screen out potential hepatotoxic compounds.

ToxDBScan: Large-scale similarity screening of toxicological databases for drug candidates

We present a new tool for hepatocarcinogenicity evaluation of drug candidates in rodents. ToxDBScan is a web tool offering quick and easy similarity screening of new drug candidates against two large-scale public databases, which contain expression profiles for substances with known carcinogenic profiles: TG-GATEs and DrugMatrix. ToxDBScan uses a set similarity score that computes the putative similarity based on similar expression of genes to identify chemicals with similar genotoxic and hepatocarcinogenic potential. We propose using a discretized representation of expression profiles, which use only information on up- or down-regulation of genes as relevant features. Therefore, only the deregulated genes are required as input. ToxDBScan provides an extensive report on similar compounds, which includes additional information on compounds, differential genes and pathway enrichments. We evaluated ToxDBScan with expression data from 15 chemicals with known hepatocarcinogenic potential and observed a sensitivity of 88 Based on the identified chemicals, we achieved perfect classification of the independent test set. ToxDBScan is publicly available from the ZBIT Bioinformatics Toolbox.

Open TG-GATEs: a large-scale toxicogenomics database

Toxicogenomics focuses on assessing the safety of compounds using gene expression profiles. Gene expression signatures from large toxicogenomics databases are expected to perform better than small databases in identifying biomarkers for the prediction and evaluation of drug safety based on a compound's toxicological mechanisms in animal target organs. Over the past 10 years, the Japanese Toxicogenomics Project consortium (TGP) has been developing a large-scale toxicogenomics database consisting of data from 170 compounds (mostly drugs) with the aim of improving and enhancing drug safety assessment. Most of the data generated by the project (e.g. gene expression, pathology, lot number) are freely available to the public via Open TG-GATEs (Toxicogenomics Project-Genomics Assisted Toxicity Evaluation System). Here, we provide a comprehensive overview of the database, including both gene expression data and metadata, with a description of experimental conditions and procedures used to generate the database. Open TG-GATEs is available from https://toxico.nibiohn.go.jp/english/index.html.

Toward predictive models for drug-induced liver injury in humans: are we there yet?

Drug-induced liver injury (DILI) is a frequent cause for the termination of drug development programs and a leading reason of drug withdrawal from the marketplace. Unfortunately, the current preclinical testing strategies, including the regulatory-required animal toxicity studies or simple in vitro tests, are insufficiently powered to predict DILI in patients reliably. Notably, the limited predictive power of such testing strategies is mostly attributed to the complex nature of DILI, a poor understanding of its mechanism, a scarcity of human hepatotoxicity data and inadequate bioinformatics capabilities. With the advent of high-content screening assays, toxicogenomics and bioinformatics, multiple end points can be studied simultaneously to improve prediction of clinically relevant DILIs. This review focuses on the current state of efforts in developing predictive models from diverse data sources for potential use in detecting human hepatotoxicity, and also aims to provide perspectives on how to further improve DILI prediction.

Characterization of chemically induced liver injuries using gene co-expression modules

Liver injuries due to ingestion or exposure to chemicals and industrial toxicants pose a serious health risk that may be hard to assess due to a lack of non-invasive diagnostic tests. Mapping chemical injuries to organ-specific damage and clinical outcomes via biomarkers or biomarker panels will provide the foundation for highly specific and robust diagnostic tests. Here, we have used DrugMatrix, a toxicogenomics database containing organ-specific gene expression data matched to dose-dependent chemical exposures and adverse clinical pathology assessments in Sprague Dawley rats, to identify groups of co-expressed genes (modules) specific to injury endpoints in the liver. We identified 78 such gene co-expression modules associated with 25 diverse injury endpoints categorized from clinical pathology, organ weight changes, and histopathology. Using gene expression data associated with an injury condition, we showed that these modules exhibited different patterns of activation characteristic of each injury. We further showed that specific module genes mapped to 1) known biochemical pathways associated with liver injuries and 2) clinically used diagnostic tests for liver fibrosis. As such, the gene modules have characteristics of both generalized and specific toxic response pathways. Using these results, we proposed three gene signature sets characteristic of liver fibrosis, steatosis, and general liver injury based on genes from the co-expression modules. Out of all 92 identified genes, 18 (20%) genes have well-documented relationships with liver disease, whereas the rest are novel and have not previously been associated with liver disease. In conclusion, identifying gene co-expression modules associated with chemically induced liver injuries aids in generating testable hypotheses and has the potential to identify putative biomarkers of adverse health effects.

Diagnostic and predictive performance and standardized threshold of traditional biomarkers for drug-induced liver injury in rats

Traditional biomarkers such as alanine and aspartate aminotransferase (ALT, AST) and total bilirubin (TBIL) have been widely used for detecting drug-induced liver injury (DILI). Although the Food and Drug Administration (FDA) proposed standardized thresholds for human as Hy's law, those for animals have not been determined, and predictability of these biomarkers for future onset of hepatic lesions remains unclear. In this study, we investigated these diagnostic and predictive performance of 10 traditional biomarkers for liver injury by receiver-operating characteristic (ROC) curve, using a free-access database where 142 hepatotoxic or non-hepatotoxic compounds were administrated to male rats (n=5253). Standardization of each biomarker value was achieved by calculating the ratio to control mean value, and the thresholds were determined under the condition of permitting 5% false positive. Of these 10 biomarkers, AST showed the best diagnostic performance. Furthermore, ALT and TBIL also showed high performance under the situation of hepatocellular necrosis and bile duct injury, respectively. Additionally, the availability of the diagnostic thresholds in difference testing facility was confirmed by the application of these thresholds to in-house prepared dataset. Meanwhile, incorrect diagnosis by the thresholds was also observed. Regarding prediction, all 10 biomarkers showed insufficient performance for future onset of hepatic lesions. In conclusion, the standardized diagnostic thresholds enable consistent evaluation of traditional biomarkers among different facilities, whereas it was suggested that novel biomarker is required for more accurate diagnosis and prediction of DILI.

Discovering functional modules by topic modeling RNA-Seq based toxicogenomic data

Toxicogenomics (TGx) endeavors to elucidate the underlying molecular mechanisms through exploring gene expression profiles in response to toxic substances. Recently, RNA-Seq is increasingly regarded as a more powerful alternative to microarrays in TGx studies. However, realizing RNA-Seq's full potential requires novel approaches to extracting information from the complex TGx data. Considering read counts as the number of times a word occurs in a document, gene expression profiles from RNA-Seq are analogous to a word by document matrix used in text mining. Topic modeling aiming at to discover the latent structures in text corpora would be helpful to explore RNA-Seq based TGx data. In this study, topic modeling was applied on a typical RNA-Seq based TGx data set to discover hidden functional modules. The RNA-Seq based gene expression profiles were transformed into "documents", on which latent Dirichlet allocation (LDA) was used to build a topic model. We found samples treated by the compounds with the same modes of actions (MoAs) could be clustered based on topic similarities. The topic most relevant to each cluster was identified as a "marker" topic, which was interpreted by gene enrichment analysis with MoAs then confirmed by compound and pathways associations mined from literature. To further validate the "marker" topics, we tested topic transferability from RNA-Seq to microarrays. The RNA-Seq based gene expression profile of a topic specifically associated with peroxisome proliferator-activated receptors (PPAR) signaling pathway was used to query samples with similar expression profiles in two different microarray data sets, yielding accuracy of about 85%. This proof-of-concept study demonstrates the applicability of topic modeling to discover functional modules in RNA-Seq data and suggests a valuable computational tool for leveraging information within TGx data in RNA-Seq era.

Of text and gene--using text mining methods to uncover hidden knowledge in toxicogenomics

Background

Toxicogenomics studies often profile gene expression from assays involving multiple doses and time points. The dose- and time-dependent pattern is of great importance to assess toxicity but computational approaches are lacking to effectively utilize this characteristic in toxicity assessment. Topic modeling is a text mining approach, but may be used analogously in toxicogenomics due to the similar data structures between text and gene dysregulation.

Results

Topic modeling was applied to a very large toxicogenomics dataset containing microarray gene expression data from >15,000 samples associated with 131 drugs tested in three different assay platforms (i.e., in vitro assay, in vivo repeated dose study and in vivo single dose experiment) with a design including multiple doses and time points. A set of “topics” which each consist of a set of genes was determined, by which the varying sensitivity of three assay systems was observed. We found that the drug-dependent effect was more pronounced in the two in vivo systems than the in vitro system, while the time-dependent effect was most strongly reflected in the in vitro system followed by the single dose study and lastly the repeated dose experiment. The dose-dependent effect was similar across three assay systems. Although the results indicated a challenge to extrapolate the in vitro results to the in vivo situation, we did notice that, for some drugs but not for all the drugs, the similarity in gene expression patterns was observed across all three assay systems, indicating a possibility of using in vitro systems with careful designs (such as the choice of dose and time point), to replace the in vivo testing strategy. Nonetheless, a potential to replace the repeated dose study by the single-dose short-term methodology was strongly implied.

Conclusions

The study demonstrated that text mining methodologies such as topic modeling provide an alternative method compared to traditional means for data reduction in toxicogenomics, enhancing researchers’ capabilities to interpret biological information.

Genomic models of short-term exposure accurately predict long-term chemical carcinogenicity and identify putative mechanisms of action

BACKGROUND

Despite an overall decrease in incidence of and mortality from cancer, about 40% of Americans will be diagnosed with the disease in their lifetime, and around 20% will die of it. Current approaches to test carcinogenic chemicals adopt the 2-year rodent bioassay, which is costly and time-consuming. As a result, fewer than 2% of the chemicals on the market have actually been tested. However, evidence accumulated to date suggests that gene expression profiles from model organisms exposed to chemical compounds reflect underlying mechanisms of action, and that these toxicogenomic models could be used in the prediction of chemical carcinogenicity.

RESULTS

In this study, we used a rat-based microarray dataset from the NTP DrugMatrix Database to test the ability of toxicogenomics to model carcinogenicity. We analyzed 1,221 gene-expression profiles obtained from rats treated with 127 well-characterized compounds, including genotoxic and non-genotoxic carcinogens. We built a classifier that predicts a chemical's carcinogenic potential with an AUC of 0.78, and validated it on an independent dataset from the Japanese Toxicogenomics Project consisting of 2,065 profiles from 72 compounds. Finally, we identified differentially expressed genes associated with chemical carcinogenesis, and developed novel data-driven approaches for the molecular characterization of the response to chemical stressors.

CONCLUSION

Here, we validate a toxicogenomic approach to predict carcinogenicity and provide strong evidence that, with a larger set of compounds, we should be able to improve the sensitivity and specificity of the predictions. We found that the prediction of carcinogenicity is tissue-dependent and that the results also confirm and expand upon previous studies implicating DNA damage, the peroxisome proliferator-activated receptor, the aryl hydrocarbon receptor, and regenerative pathology in the response to carcinogen exposure.

Composition and applications of focus libraries to phenotypic assays

The wealth of bioactivity information now available on low-molecular weight compounds has enabled a paradigm shift in chemical biology and early phase drug discovery efforts. Traditionally chemical libraries have been most commonly employed in screening approaches where a bioassay is used to characterize a chemical library in a random search for active samples. However, robust curating of bioassay data, establishment of ontologies enabling mining of large chemical biology datasets, and a wealth of public chemical biology information has made possible the establishment of highly annotated compound collections. Such annotated chemical libraries can now be used to build a pathway/target hypothesis and have led to a new view where chemical libraries are used to characterize a bioassay. In this article we discuss the types of compounds in these annotated libraries composed of tools, probes, and drugs. As well, we provide rationale and a few examples for how such libraries can enable phenotypic/forward chemical genomic approaches. As with any approach, there are several pitfalls that need to be considered and we also outline some strategies to avoid these.