Qualitative and quantitative concentration-response modelling of gene co-expression networks to unlock hepatotoxic mechanisms for next generation chemical safety assessment.

Next generation risk assessment of chemicals revolves around the use of mechanistic information without animal experimentation. In this regard, toxicogenomics has proven to be a useful tool to elucidate the underlying mechanisms of adverse effects of xenobiotics. In the present study, two widely used human in vitro hepatocyte culture systems, namely primary human hepatocytes (PHH) and human hepatoma HepaRG cells, were exposed to liver toxicants known to induce liver cholestasis, steatosis or necrosis. Benchmark concentration-response modelling was applied to transcriptomics gene co-expression networks (modules) to derive benchmark concentrations (BMCs) and to gain mechanistic insight into the hepatotoxic effects. BMCs derived by concentration-response modelling of gene co-expression modules recapitulated concentration-response modelling of individual genes. Although PHH and HepaRG cells showed overlap in deregulated genes and modules by the liver toxicants, PHH demonstrated a higher responsiveness, based on the lower BMCs of co-regulated gene modules. Such BMCs can be used as transcriptomics point of departure (tPOD) for assessing module-associated cellular (stress) pathways/processes. This approach identified clear tPODs of around maximum systemic concentration (Cmax) levels for the tested drugs, while for cosmetics ingredients the BMCs were 10-100-fold higher than the estimated plasma concentrations. This approach could serve next generation risk assessment practice to identify early responsive modules at low BMCs, that could be linked to key events in liver adverse outcome pathways. In turn, this can assist in delineating potential hazards of new test chemicals using in vitro systems and used in a risk assessment when BMCs are paired with chemical exposure assessment.

Risk assessment of chemicals has traditionally been focused on animal experiments. In contrast, next generation risk assessment uses biological information obtained from experiments in cell culture models without animals to identify potential hazards. Since the liver is the main target organ of toxicity, many liver cell (hepatocyte) models have been developed and applied for hazard assessment. In this study, two widely used human hepatocyte cell models, PHH and HepaRG, were exposed to liver toxic chemicals. Biological changes in gene expression were measured in a concentration range to identify the concentration at which a biological response was perturbed using concentration response modelling. Genes belonging to the same biological process were joined based on co-expression to derive an average concentration of this process. This animal-free approach could be applied for risk assessment when biological response concentrations were related to the expected human exposure to identify potential hazard of the test chemicals.

A systems approach reveals species differences in hepatic stress response capacity.

To minimize the occurrence of unexpected toxicities in early phase preclinical studies of new drugs, it is vital to understand fundamental similarities and differences between preclinical species and humans. Species differences in sensitivity to acetaminophen (APAP) liver injury have been related to differences in the fraction of the drug that is bioactivated to the reactive metabolite N-acetyl-p-benzoquinoneimine (NAPQI). We have used physiologically based pharmacokinetic modeling to identify oral doses of APAP (300 and 1000 mg/kg in mice and rats, respectively) yielding similar hepatic burdens of NAPQI to enable the comparison of temporal liver tissue responses under conditions of equivalent chemical insult. Despite pharmacokinetic and biochemical verification of the equivalent NAPQI insult, serum biomarker and tissue histopathology analyses revealed that mice still exhibited a greater degree of liver injury than rats. Transcriptomic and proteomic analyses highlighted the stronger activation of stress response pathways (including the Nrf2 oxidative stress response and autophagy) in the livers of rats, indicative of a more robust transcriptional adaptation to the equivalent insult. Components of these pathways were also found to be expressed at a higher basal level in the livers of rats compared with both mice and humans. Our findings exemplify a systems approach to understanding differential species sensitivity to hepatotoxicity. Multiomics analysis indicated that rats possess a greater basal and adaptive capacity for hepatic stress responses than mice and humans, with important implications for species selection and human translation in the safety testing of new drug candidates associated with reactive metabolite formation.

Identifying multiscale translational safety biomarkers using a network-based systems approach.

Animal testing is the current standard for drug and chemicals safety assessment, but hazards translation to human is uncertain. Human in vitro models can address the species translation but might not replicate in vivo complexity. Herein, we propose a network-based method addressing these translational multiscale problems that derives in vivo liver injury biomarkers applicable to in vitro human early safety screening. We applied weighted correlation network analysis (WGCNA) to a large rat liver transcriptomic dataset to obtain co-regulated gene clusters (modules). We identified modules statistically associated with liver pathologies, including a module enriched for ATF4-regulated genes as associated with the occurrence of hepatocellular single-cell necrosis, and as preserved in human liver in vitro models. Within the module, we identified TRIB3 and MTHFD2 as a novel candidate stress biomarkers, and developed and used BAC-eGFPHepG2 reporters in a compound screening, identifying compounds showing ATF4-dependent stress response and potential early safety signals.

Application of high-throughput transcriptomics for mechanism-based biological read-across of short-chain carboxylic acid analogues of valproic acid.

Chemical read-across is commonly evaluated without specific knowledge of the biological mechanisms leading to observed adverse outcomes in vivo. Integrating data that indicate shared modes of action in humans will strengthen read-across cases. Here we studied transcriptomic responses of primary human hepatocytes (PHH) to a large panel of carboxylic acids to include detailed mode-of-action data as a proof-of-concept for read-across in risk assessment. In rodents, some carboxylic acids, including valproic acid (VPA), are known to cause hepatic steatosis, whereas others do not. We investigated transcriptomics responses of PHHs exposed for 24 h to 18 structurally different VPA analogues in a concentration range to determine biological similarity in relation to in vivo steatotic potential. Using a targeted high-throughput screening assay, we assessed the differential expression of ~3,000 genes covering relevant biological pathways. Differentially expressed gene analysis revealed differences in potency of carboxylic acids, and expression patterns were highly similar for structurally similar compounds. Strong clustering occurred for steatosis-positive versus steatosis-negative carboxylic acids. To quantitatively define biological read-across, we combined pathway analysis and weighted gene co-expression network analysis. Active carboxylic acids displayed high similarity in gene network modulation. Importantly, free fatty acid synthesis modulation and stress pathway responses are affected by active car­boxylic acids, providing coherent mechanistic underpinning for our findings. Our work shows that transcriptomic analysis of cultured human hepatocytes can reinforce the prediction of liver injury outcome based on quantitative and mechanistic biological data and support its application in read-across.

The human hepatocyte TXG-MAPr: gene co-expression network modules to support mechanism-based risk assessment.

Mechanism-based risk assessment is urged to advance and fully permeate into current safety assessment practices, possibly at early phases of drug safety testing. Toxicogenomics is a promising source of mechanisms-revealing data, but interpretative analysis tools specific for the testing systems (e.g. hepatocytes) are lacking. In this study, we present the TXG-MAPr webtool (available at https://txg-mapr.eu/WGCNA_PHH/TGGATEs_PHH/ ), an R-Shiny-based implementation of weighted gene co-expression network analysis (WGCNA) obtained from the Primary Human Hepatocytes (PHH) TG-GATEs dataset. The 398 gene co-expression networks (modules) were annotated with functional information (pathway enrichment, transcription factor) to reveal their mechanistic interpretation. Several well-known stress response pathways were captured in the modules, were perturbed by specific stressors and showed preservation in rat systems (rat primary hepatocytes and rat in vivo liver), with the exception of DNA damage and oxidative stress responses. A subset of 87 well-annotated and preserved modules was used to evaluate mechanisms of toxicity of endoplasmic reticulum (ER) stress and oxidative stress inducers, including cyclosporine A, tunicamycin and acetaminophen. In addition, module responses can be calculated from external datasets obtained with different hepatocyte cells and platforms, including targeted RNA-seq data, therefore, imputing biological responses from a limited gene set. As another application, donors' sensitivity towards tunicamycin was investigated with the TXG-MAPr, identifying higher basal level of intrinsic immune response in donors with pre-existing liver pathology. In conclusion, we demonstrated that gene co-expression analysis coupled to an interactive visualization environment, the TXG-MAPr, is a promising approach to achieve mechanistic relevant, cross-species and cross-platform evaluation of toxicogenomic data.

An ensemble learning approach for modeling the systems biology of drug-induced injury.

BACKGROUND: Drug-induced liver injury (DILI) is an adverse reaction caused by the intake of drugs of common use that produces liver damage. The impact of DILI is estimated to affect around 20 in 100,000 inhabitants worldwide each year. Despite being one of the main causes of liver failure, the pathophysiology and mechanisms of DILI are poorly understood. In the present study, we developed an ensemble learning approach based on different features (CMap gene expression, chemical structures, drug targets) to predict drugs that might cause DILI and gain a better understanding of the mechanisms linked to the adverse reaction. RESULTS: We searched for gene signatures in CMap gene expression data by using two approaches: phenotype-gene associations data from DisGeNET, and a non-parametric test comparing gene expression of DILI-Concern and No-DILI-Concern drugs (as per DILIrank definitions). The average accuracy of the classifiers in both approaches was 69%. We used chemical structures as features, obtaining an accuracy of 65%. The combination of both types of features produced an accuracy around 63%, but improved the independent hold-out test up to 67%. The use of drug-target associations as feature obtained the best accuracy (70%) in the independent hold-out test. CONCLUSIONS: When using CMap gene expression data, searching for a specific gene signature among the landmark genes improves the quality of the classifiers, but it is still limited by the intrinsic noise of the dataset. When using chemical structures as a feature, the structural diversity of the known DILI-causing drugs hampers the prediction, which is a similar problem as for the use of gene expression information. The combination of both features did not improve the quality of the classifiers but increased the robustness as shown on independent hold-out tests. The use of drug-target associations as feature improved the prediction, specially the specificity, and the results were comparable to previous research studies.

Comparative transcriptomics of shear stress treated Pkd1-/- cells and pre-cystic kidneys reveals pathways involved in early polycystic kidney disease.

Mutations in the PKD1 or PKD2 genes are the cause of autosomal dominant polycystic kidney disease (ADPKD). The encoded proteins localize within the cell membrane and primary cilia and are proposed to be involved in mechanotransduction. Therefore, we evaluate shear stress dependent signaling in renal epithelial cells and the relevance for ADPKD. Using RNA sequencing and pathway analysis, we compared gene expression of in vitro shear stress treated Pkd1-/- renal epithelial cells and in vivo pre-cystic Pkd1del models. We show that shear stress alters the same signaling pathways in Pkd1-/- renal epithelial cells and Pkd1wt controls. However, expression of a number of genes was slightly more induced by shear stress in Pkd1-/- cells, suggesting that Pkd1 has the function to restrain shear regulated signaling instead of being a mechano-sensing activator. We also compared altered gene expression in Pkd1-/- cells during shear with in vivo transcriptome data of kidneys from Pkd1del mice at three early pre-cystic time-points. This revealed overlap of a limited number of differentially expressed genes. However, the overlap between cells and mice is much higher when looking at pathways and molecular processes, largely due to altered expression of paralogous genes. Several of the altered pathways in the in vitro and in vivo Pkd1del models are known to be implicated in ADPKD pathways, including PI3K-AKT, MAPK, Hippo, calcium, Wnt, and TGF-ß signaling. We hypothesize that increased activation of selected genes in renal epithelial cells early upon Pkd1 gene disruption may disturb the balance in signaling and may contribute to cyst formation.

Comprehensive transcriptome analysis of fluid shear stress altered gene expression in renal epithelial cells.

Renal epithelial cells are exposed to mechanical forces due to flow-induced shear stress within the nephrons. Shear stress is altered in renal diseases caused by tubular dilation, obstruction, and hyperfiltration, which occur to compensate for lost nephrons. Fundamental in regulation of shear stress are primary cilia and other mechano-sensors, and defects in cilia formation and function have profound effects on development and physiology of kidneys and other organs. We applied RNA sequencing to get a comprehensive overview of fluid-shear regulated genes and pathways in renal epithelial cells. Functional enrichment-analysis revealed TGF-ß, MAPK, and Wnt signaling as core signaling pathways up-regulated by shear. Inhibitors of TGF-ß and MAPK/ERK signaling modulate a wide range of mechanosensitive genes, identifying these pathways as master regulators of shear-induced gene expression. However, the main down-regulated pathway, that is, JAK/STAT, is independent of TGF-ß and MAPK/ERK. Other up-regulated cytokine pathways include FGF, HB-EGF, PDGF, and CXC. Cellular responses to shear are modified at several levels, indicated by altered expression of genes involved in cell-matrix, cytoskeleton, and glycocalyx remodeling, as well as glycolysis and cholesterol metabolism. Cilia ablation abolished shear induced expression of a subset of genes, but genes involved in TGF-ß, MAPK, and Wnt signaling were hardly affected, suggesting that other mechano-sensors play a prominent role in the shear stress response of renal epithelial cells. Modulations in signaling due to variations in fluid shear stress are relevant for renal physiology and pathology, as suggested by elevated gene expression at pathological levels of shear stress compared to physiological shear.

High-Throughput Phenotypic Screening of Kinase Inhibitors to Identify Drug Targets for Polycystic Kidney Disease.

Polycystic kidney disease (PKD) is a prevalent disorder characterized by renal cysts that lead to kidney failure. Various signaling pathways have been targeted to stop disease progression, but most interventions still focus on alleviating PKD-associated symptoms. The mechanistic complexity of the disease, as well as the lack of functional in vitro assays for compound testing, has made drug discovery for PKD challenging. To identify modulators of PKD, Pkd1-/- kidney tubule epithelial cells were applied to a scalable and automated 3D cyst culture model for compound screening, followed by phenotypic profiling to determine compound efficacy. We used this screening platform to screen a library of 273 kinase inhibitors to probe various signaling pathways involved in cyst growth. We show that inhibition of several targets, including aurora kinase, CDK, Chk, IGF-1R, Syk, and mTOR, but, surprisingly, not PI3K, prevented forskolin-induced cyst swelling. Additionally, we show that multiparametric phenotypic classification discriminated potentially undesirable (i.e., cytotoxic) compounds from molecules inducing the desired phenotypic change, greatly facilitating hit selection and validation. Our findings show that a pathophysiologically relevant 3D cyst culture model of PKD coupled to phenotypic profiling can be used to identify potentially therapeutic compounds and predict and validate molecular targets for PKD.

Fluid shear stress-induced TGF-ß/ALK5 signaling in renal epithelial cells is modulated by MEK1/2.

Renal tubular epithelial cells are exposed to mechanical forces due to fluid flow shear stress within the lumen of the nephron. These cells respond by activation of mechano-sensors located at the plasma membrane or the primary cilium, having crucial roles in maintenance of cellular homeostasis and signaling. In this paper, we applied fluid shear stress to study TGF-ß signaling in renal epithelial cells with and without expression of the Pkd1-gene, encoding a mechano-sensor mutated in polycystic kidney disease. TGF-ß signaling modulates cell proliferation, differentiation, apoptosis, and fibrotic deposition, cellular programs that are altered in renal cystic epithelia. SMAD2/3-mediated signaling was activated by fluid flow, both in wild-type and Pkd1 -/- cells. This was characterized by phosphorylation and nuclear accumulation of p-SMAD2/3, as well as altered expression of downstream target genes and epithelial-to-mesenchymal transition markers. This response was still present after cilia ablation. An inhibitor of upstream type-I-receptors, ALK4/ALK5/ALK7, as well as TGF-ß-neutralizing antibodies effectively blocked SMAD2/3 activity. In contrast, an activin-ligand trap was ineffective, indicating that increased autocrine TGF-ß signaling is involved. To study potential involvement of MAPK/ERK signaling, cells were treated with a MEK1/2 inhibitor. Surprisingly, fluid flow-induced expression of most SMAD2/3 targets was further enhanced upon MEK inhibition. We conclude that fluid shear stress induces autocrine TGF-ß/ALK5-induced target gene expression in renal epithelial cells, which is partially restrained by MEK1/2-mediated signaling.

Inhibition of Activin Signaling Slows Progression of Polycystic Kidney Disease.

Autosomal dominant polycystic kidney disease (ADPKD), characterized by the formation of numerous kidney cysts, is caused by PKD1 or PKD2 mutations and affects 0.1% of the population. Although recent clinical studies indicate that reduction of cAMP levels slows progression of PKD, this finding has not led to an established safe and effective therapy for patients, indicating the need to find new therapeutic targets. The role of TGF-ß in PKD is not clearly understood, but nuclear accumulation of phosphorylated SMAD2/3 in cyst-lining cells suggests the involvement of TGF-ß signaling in this disease. In this study, we ablated the TGF-ß type 1 receptor (also termed activin receptor-like kinase 5) in renal epithelial cells of PKD mice, which had little to no effect on the expression of SMAD2/3 target genes or the progression of PKD. Therefore, we investigated whether alternative TGF-ß superfamily ligands account for SMAD2/3 activation in cystic epithelial cells. Activins are members of the TGF-ß superfamily and drive SMAD2/3 phosphorylation via activin receptors, but activins have not been studied in the context of PKD. Mice with PKD had increased expression of activin ligands, even at early stages of disease. In addition, treatment with a soluble activin receptor IIB fusion (sActRIIB-Fc) protein, which acts as a soluble trap to sequester activin ligands, effectively inhibited cyst formation in three distinct mouse models of PKD. These data point to activin signaling as a key pathway in PKD and a promising target for therapy.

Altered Hippo signalling in polycystic kidney disease.

Autosomal dominant polycystic kidney disease (ADPKD) is characterized by progressive deterioration of renal function and formation of cysts, and is an important cause of end-stage renal disease. Previously we showed that tubular epithelial injury accelerates cyst formation in inducible Pkd1-deletion mice. In these mice, expression of the planar cell polarity (PCP) component Four-jointed (Fjx1) is decreased during epithelial repair, while in control mice Fjx1 expression is increased and may be required during tissue regeneration. In cystic kidneys, however, Fjx1 expression is also increased. Besides a PCP component, Four-jointed is also implicated in the Hippo-signalling pathway. This pathway is involved in organ size control by regulating proliferation and apoptosis. The role of Hippo signalling, together with the opposing expression pattern of Fjx1 during epithelial repair and at cystic stages, triggered us to investigate the activity of the Hippo pathway during these processes. Therefore, we examined its final effector molecule, the transcriptional co-activator Yes-associated protein (YAP) and observed that during tissue repair, YAP expression was not different between Pkd1-deletion mice and controls, ie during tissue regeneration YAP expression was increased and predominantly localized in the cytoplasm but normalized after tissue repair. At a later stage, however, in cystic epithelia and epithelia of dilated tubules, strong nuclear YAP accumulation was observed, accompanied by up-regulation of the YAP transcriptional targets Birc-3, Ctgf, InhbA, and Fjx1. Altered activity of the Hippo pathway was confirmed in renal tissues from human ADPKD and ARPKD patients, as well as in cystic renal tumours. Our data strengthen the concept that during epithelial repair Four-jointed is involved in PCP signalling, while in cystic kidneys it is related to Hippo signalling and cyst growth.

Curcumin inhibits cystogenesis by simultaneous interference of multiple signaling pathways: in vivo evidence from a Pkd1-deletion model.

Autosomal dominant polycystic kidney disease (ADPKD) caused by mutations in either the PKD1 or PKD2 gene is a major cause of end-stage renal failure. A number of compounds targeting specific signaling pathways were able to inhibit cystogenesis in rodent models and are currently being tested in clinical trials. However, given the complex signaling in ADPKD, an ideal therapy would likely have to comprise several pathways at once. Therefore, multitarget compounds may provide promising therapeutic interventions for the treatment of ADPKD. To test this hypothesis, we treated Pkd1-deletion mice with diferuloylmethane (curcumin), a compound without appreciable side effects and known to modulate several pathways that are also altered in ADPKD, e.g., mammalian target of rapamycin (mTOR) and Wnt signaling. After conditional inactivation of Pkd1, mTOR signaling was indeed elevated in cystic kidneys. Interestingly, also activation of signal transducers and activator of transcription 3 (STAT3) strongly correlated with cyst progression. Both pathways were effectively inhibited in vitro by curcumin. Importantly, Pkd1-deletion mice that were treated with curcumin and killed at an early stage of PKD displayed improved renal histology and reduced STAT3 activation, proliferation index, cystic index, and kidney weight/body weight ratios. In addition, renal failure was significantly postponed in mice with severe PKD. These data suggest that multitarget compounds hold promising potential for safe and effective treatment of ADPKD.