Can Transcriptomic Profiles from Cancer Cell Lines Be Used for Toxicity Assessment?

A strategy to detect metabolic changes induced by exposure to chemicals from large sets of condition-specific metabolic models computed with enumeration techniques

BACKGROUND

The growing abundance of in vitro omics data, coupled with the necessity to reduce animal testing in the safety assessment of chemical compounds and even eliminate it in the evaluation of cosmetics, highlights the need for adequate computational methodologies. Data from omics technologies allow the exploration of a wide range of biological processes, therefore providing a better understanding of mechanisms of action (MoA) related to chemical exposure in biological systems. However, the analysis of these large datasets remains difficult due to the complexity of modulations spanning multiple biological processes.

RESULTS

To address this, we propose a strategy to reduce information overload by computing, based on transcriptomics data, a comprehensive metabolic sub-network reflecting the metabolic impact of a chemical. The proposed strategy integrates transcriptomic data to a genome scale metabolic network through enumeration of condition-specific metabolic models hence translating transcriptomics data into reaction activity probabilities. Based on these results, a graph algorithm is applied to retrieve user readable sub-networks reflecting the possible metabolic MoA (mMoA) of chemicals. This strategy has been implemented as a three-step workflow. The first step consists in building cell condition-specific models reflecting the metabolic impact of each exposure condition while taking into account the diversity of possible optimal solutions with a partial enumeration algorithm. In a second step, we address the challenge of analyzing thousands of enumerated condition-specific networks by computing differentially activated reactions (DARs) between the two sets of enumerated possible condition-specific models. Finally, in the third step, DARs are grouped into clusters of functionally interconnected metabolic reactions, representing possible mMoA, using the distance-based clustering and subnetwork extraction method. The first part of the workflow was exemplified on eight molecules selected for their known human hepatotoxic outcomes associated with specific MoAs well described in the literature and for which we retrieved primary human hepatocytes transcriptomic data in Open TG-GATEs. Then, we further applied this strategy to more precisely model and visualize associated mMoA for two of these eight molecules (amiodarone and valproic acid). The approach proved to go beyond gene-based analysis by identifying mMoA when few genes are significantly differentially expressed (2 differentially expressed genes (DEGs) for amiodarone), bringing additional information from the network topology, or when very large number of genes were differentially expressed (5709 DEGs for valproic acid). In both cases, the results of our strategy well fitted evidence from the literature regarding known MoA. Beyond these confirmations, the workflow highlighted potential other unexplored mMoA.

CONCLUSION

The proposed strategy allows toxicology experts to decipher which part of cellular metabolism is expected to be affected by the exposition to a given chemical. The approach originality resides in the combination of different metabolic modelling approaches (constraint based and graph modelling). The application to two model molecules shows the strong potential of the approach for interpretation and visual mining of complex omics in vitro data. The presented strategy is freely available as a python module ( https://pypi.org/project/manamodeller/ ) and jupyter notebooks ( https://github.com/LouisonF/MANA ).

State of the Art of Genomic Technology in Toxicology: A Review

The rapid growth of genomics techniques has revolutionized and impacted, greatly and positively, the knowledge of toxicology, ushering it into a "new era": the era of genomic technology (GT). This great advance permits us to analyze the whole genome, to know the gene response to toxicants and environmental stressors, and to determine the specific profiles of gene expression, among many other approaches. The aim of this work was to compile and narrate the recent research on GT during the last 2 years (2020-2022). A literature search was managed using the PubMed and Medscape interfaces on the Medline database. Relevant articles published in peer-reviewed journals were retrieved and their main results and conclusions are mentioned briefly. It is quite important to form a multidisciplinary taskforce on GT with the aim of designing and implementing a comprehensive, collaborative, and a strategic work plan, prioritizing and assessing the most relevant diseases, so as to decrease human morbimortality due to exposure to environmental chemicals and stressors.

From vision toward best practices: Evaluating in vitro transcriptomic points of departure for application in risk assessment using a uniform workflow

The growing number of chemicals in the current consumer and industrial markets presents a major challenge for regulatory programs faced with the need to assess the potential risks they pose to human and ecological health. The increasing demand for hazard and risk assessment of chemicals currently exceeds the capacity to produce the toxicity data necessary for regulatory decision making, and the applied data is commonly generated using traditional approaches with animal models that have limited context in terms of human relevance. This scenario provides the opportunity to implement novel, more efficient strategies for risk assessment purposes. This study aims to increase confidence in the implementation of new approach methods in a risk assessment context by using a parallel analysis to identify data gaps in current experimental designs, reveal the limitations of common approaches deriving transcriptomic points of departure, and demonstrate the strengths in using high-throughput transcriptomics (HTTr) to derive practical endpoints. A uniform workflow was applied across six curated gene expression datasets from concentration-response studies containing 117 diverse chemicals, three cell types, and a range of exposure durations, to determine tPODs based on gene expression profiles. After benchmark concentration modeling, a range of approaches was used to determine consistent and reliable tPODs. High-throughput toxicokinetics were employed to translate in vitro tPODs (µM) to human-relevant administered equivalent doses (AEDs, mg/kg-bw/day). The tPODs from most chemicals had AEDs that were lower (i.e., more conservative) than apical PODs in the US EPA CompTox chemical dashboard, suggesting in vitro tPODs would be protective of potential effects on human health. An assessment of multiple data points for single chemicals revealed that longer exposure duration and varied cell culture systems (e.g., 3D vs. 2D) lead to a decreased tPOD value that indicated increased chemical potency. Seven chemicals were flagged as outliers when comparing the ratio of tPOD to traditional POD, thus indicating they require further assessment to better understand their hazard potential. Our findings build confidence in the use of tPODs but also reveal data gaps that must be addressed prior to their adoption to support risk assessment applications.

Benchmarking causal reasoning algorithms for gene expression-based compound mechanism of action analysis

BACKGROUND

Elucidating compound mechanism of action (MoA) is beneficial to drug discovery, but in practice often represents a significant challenge. Causal Reasoning approaches aim to address this situation by inferring dysregulated signalling proteins using transcriptomics data and biological networks; however, a comprehensive benchmarking of such approaches has not yet been reported. Here we benchmarked four causal reasoning algorithms (SigNet, CausalR, CausalR ScanR and CARNIVAL) with four networks (the smaller Omnipath network vs. 3 larger MetaBase™ networks), using LINCS L1000 and CMap microarray data, and assessed to what extent each factor dictated the successful recovery of direct targets and compound-associated signalling pathways in a benchmark dataset comprising 269 compounds. We additionally examined impact on performance in terms of the functions and roles of protein targets and their connectivity bias in the prior knowledge networks.

RESULTS

According to statistical analysis (negative binomial model), the combination of algorithm and network most significantly dictated the performance of causal reasoning algorithms, with the SigNet recovering the greatest number of direct targets. With respect to the recovery of signalling pathways, CARNIVAL with the Omnipath network was able to recover the most informative pathways containing compound targets, based on the Reactome pathway hierarchy. Additionally, CARNIVAL, SigNet and CausalR ScanR all outperformed baseline gene expression pathway enrichment results. We found no significant difference in performance between L1000 data or microarray data, even when limited to just 978 'landmark' genes. Notably, all causal reasoning algorithms also outperformed pathway recovery based on input DEGs, despite these often being used for pathway enrichment. Causal reasoning methods performance was somewhat correlated with connectivity and biological role of the targets.

CONCLUSIONS

Overall, we conclude that causal reasoning performs well at recovering signalling proteins related to compound MoA upstream from gene expression changes by leveraging prior knowledge networks, and that the choice of network and algorithm has a profound impact on the performance of causal reasoning algorithms. Based on the analyses presented here this is true for both microarray-based gene expression data as well as those based on the L1000 platform.

Using chemical and biological data to predict drug toxicity

Various sources of information can be used to better understand and predict compound activity and safety-related endpoints, including biological data such as gene expression and cell morphology. In this review, we first introduce types of chemical, in vitro and in vivo information that can be used to describe compounds and adverse effects. We then explore how compound descriptors based on chemical structure or biological perturbation response can be used to predict safety-related endpoints, and how especially biological data can help us to better understand adverse effects mechanistically. Overall, the described applications demonstrate how large-scale biological information presents new opportunities to anticipate and understand the biological effects of compounds, and how this can support predictive toxicology and drug discovery projects.

Development of in silico prediction models for drug-induced liver malignant tumors based on the activity of molecular initiating events: Biologically interpretable features

Liver malignant tumors (LMTs) have recently been reported as severe and life-threatening adverse drug events associated with drug-induced liver injury (DILI). DILIs are the most common adverse drug event and can cause the withdrawal of medicinal products or major regulatory action. To reduce the attrition rate and cost of drug discovery, various quantitative structure-toxicity relationship models have been proposed to predict the probability of a DILI based on the chemical structure of a drug. However, there are many unresolved issues regarding the predictors of LMT-inducing drugs, and biologically interpretable prediction models for LMT have not been developed. Here, we constructed prediction models for whether a drug is LMT-inducing based on the activity of molecular initiating events (MIEs), which are biologically interpretable features and are defined as the initial interaction between a molecule and biosystem. We then constructed five machine learning models (i.e., LightGBM, XGBoost, random forest, neural network, and support vector machine) and evaluated their predictive performances. LightGBM achieved the best performance among the tested models. The MIEs making the highest contribution to the model construction for drug-induced LMT were inducement of Enhanced Level of Genome Instability Gene 1 (human ATAD5), nuclear factor-κ B, and activation of thyrotropin-releasing hormone receptor. These results support the previous literature and can be related to the mechanism onset of drug-induced LMT. Our findings may provide useful knowledge for drug development, research, and regulatory decision-making and will contribute to building more accurate and meaningful DILI prediction models by increasing understanding of biological predictors.

Randomized trial drug controlled compendious transcriptome analysis supporting broad and phase specific therapeutic potential of multiple candidates in COVID-19

Effective therapies for coronavirus disease 2019 (COVID-19) are urgently needed. Maladaptive hyperinflammation and excessive cytokine release underlie the disease severity, with antiinflammatory and cytokine inhibiting agents expected to exert therapeutic effects. A major present challenge is identification of appropriate phase of the illness for a given intervention to yield optimum outcomes. Considering its established disease biomarker and drug discovery potential, a compendious analysis of existing transcriptomic data is presented here toward addressing this gap. The analysis is based on COVID-19 data related to intensive care unit (ICU) and non-ICU admissions, discharged and deceased patients, ventilation and non-ventilation phases, and high oxygen supplementation. It integrates transcriptomic data related to the effects of, in various cellular treatment models, the COVID-19 randomized clinical trial (RCT) successful drug dexamethasone, and the failed drug, with a potential to harm, hydroxychloroquine/chloroquine. Similarly, effects of various COVID-19 candidate drugs/anticytokines as well as proinflammatory cytokines implicated in the illness are also examined. The underlying assumption was that compared to COVID-19, an effective drug/anticytokine and a disease aggravating agent would affect gene regulation in opposite and same direction, in that order. Remarkably, the assumption was supported with respect to both the RCT drugs. With this control validation, etanercept, followed by tofacitinib and adalimumab, showed transcriptomic effects predictive of benefits in both ventilation and non-ventilation ICU stages as well as in non-ICU phase. On the other hand, canakinumab showed potential for effectiveness in high oxygen supplementation phase. These findings may inform experimental and clinical studies toward drug repurposing in COVID-19.

Inferring molecular mechanisms of dexamethasone therapy in severe COVID-19 from existing transcriptomic data

Dexamethasone, a synthetic glucocorticoid, has previously shown mortality benefit in severe coronavirus disease 2019 (COVID-19) in a randomized controlled trial. As the illness is considered to reflect a hyperinflammatory state, this therapeutic effectiveness is presumably ascribed to broad anti-inflammatory activities of glucocorticoids. Here, an unbiased analysis of available transcriptomic data on lung and blood immune cells from severe COVID-19 patients and matching cellular models of dexamethasone treatment is presented that supports this presumption. Comparison of differentially expressed genes in severe COVID-19 with that in dexamethasone treated cells reveals a small set of genes that are regulated in opposite direction between the disease and the drug, and are enriched for genes and processes related to glucocorticoid pathway and receptor binding. This expression signature differentiates as a whole various cytokines from a set of anti-cytokine/anti-inflammatory agents, with the former resembling COVID-19 and the latter dexamethasone in gene regulation. The signature apparently relates to TNF- α, IL-1α, IL-1β, IFN-α, IFN-β, and IFN-γ signaling, but not IL-6 signaling, suggesting that therapeutic effect of dexamethasone in COVID-19 does not involve IL-6 pathway. However, as all these observations are purely based on bioinformatic analysis, experimental evidence will be required to validate the inferences drawn. In conclusion, the present analysis seems to provide a proof of concept for therapeutic mechanisms of dexamethasone in COVID-19.

Biomaterials for Three-Dimensional Cell Culture: From Applications in Oncology to Nanotechnology

Three-dimensional cell culture has revolutionized cellular biology research and opened the door to novel discoveries in terms of cellular behavior and response to microenvironment stimuli. Different types of 3D culture exist today, including hydrogel scaffold-based models, which possess a complex structure mimicking the extracellular matrix. These hydrogels can be made of polymers (natural or synthetic) or low-molecular weight gelators that, via the supramolecular assembly of molecules, allow the production of a reproducible hydrogel with tunable mechanical properties. When cancer cells are grown in this type of hydrogel, they develop into multicellular tumor spheroids (MCTS). Three-dimensional (3D) cancer culture combined with a complex microenvironment that consists of a platform to study tumor development and also to assess the toxicity of physico-chemical entities such as ions, molecules or particles. With the emergence of nanoparticles of different origins and natures, implementing a reproducible in vitro model that consists of a bio-indicator for nano-toxicity assays is inevitable. However, the maneuver process of such a bio-indicator requires the implementation of a repeatable system that undergoes an exhaustive follow-up. Hence, the biggest challenge in this matter is the reproducibility of the MCTS and the associated full-scale characterization of this system's components.

Deep Learning on High-Throughput Transcriptomics to Predict Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is one of the most cited reasons for the high drug attrition rate and drug withdrawal from the market. The accumulated large amount of high throughput transcriptomic profiles and advances in deep learning provide an unprecedented opportunity to improve the suboptimal performance of DILI prediction. In this study, we developed an eight-layer Deep Neural Network (DNN) model for DILI prediction using transcriptomic profiles of human cell lines (LINCS L1000 dataset) with the current largest binary DILI annotation data [i.e., DILI severity and toxicity (DILIst)]. The developed models were evaluated by Monte Carlo cross-validation (MCCV), permutation test, and an independent validation (IV) set. The developed DNN model achieved the area under the receiver operating characteristic curve (AUC) of 0.802 and 0.798, and balanced accuracy of 0.741 and 0.721 for training and an IV set, respectively, outperforming the conventional machine learning algorithms, including K-nearest neighbors (KNN), Support Vector Machine (SVM), and Random Forest (RF). Moreover, the developed DNN model provided a more balanced sensitivity of 0.839 and specificity of 0.603. Besides, we found the developed DNN model had a superior predictive performance for oncology drugs. Also, the functional and network analysis of genes driving the predictions revealed their relevance to the underlying mechanisms of DILI. The proposed DNN model could be a promising tool for early detection of DILI potential in the pre-clinical setting.