Generative modeling of single-cell gene expression for dose-dependent chemical perturbations.

Single-cell sequencing reveals the heterogeneity of cellular response to chemical perturbations. However, testing all relevant combinations of cell types, chemicals, and doses is a daunting task. A deep generative learning formalism called variational autoencoders (VAEs) has been effective in predicting single-cell gene expression perturbations for single doses. Here, we introduce single-cell variational inference of dose-response (scVIDR), a VAE-based model that predicts both single-dose and multiple-dose cellular responses better than existing models. We show that scVIDR can predict dose-dependent gene expression across mouse hepatocytes, human blood cells, and cancer cell lines. We biologically interpret the latent space of scVIDR using a regression model and use scVIDR to order individual cells based on their sensitivity to chemical perturbation by assigning each cell a "pseudo-dose" value. We envision that scVIDR can help reduce the need for repeated animal testing across tissues, chemicals, and doses.

Cystine/Glutamate Xc- Antiporter Induction Compensates for Transsulfuration Pathway Repression by 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) to Ensure Cysteine for Hepatic Glutathione Biosynthesis.

Exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) has been associated with the induction of oxidative stress and the progression of steatosis to steatohepatitis with fibrosis. It also disrupts metabolic pathways including one-carbon metabolism (OCM) and the transsulfuration pathway with possible consequences on glutathione (GSH) levels. In this study, complementary RNAseq and metabolomics data were integrated to examine the hepatic transsulfuration pathway and glutathione biosynthesis in mice following treatment with TCDD every 4 days for 28 days. TCDD dose-dependently repressed hepatic cystathionine ß-synthase (CBS) and cystathionine γ-lyase (CTH) mRNA and protein levels. Reduced CBS and CTH levels are also correlated with dose-dependent decreases in hepatic extract hydrogen sulfide (H2S). In contrast, cysteine levels increased consistent with the induction of Slc7a11, which encodes for the cystine/glutamate Xc- antiporter. Cotreatment of primary hepatocytes with sulfasalazine, a cystine/glutamate Xc- antiporter inhibitor, decreased labeled cysteine incorporation into GSH with a corresponding increase in TCDD cytotoxicity. Although reduced and oxidized GSH levels were unchanged following treatment due to the induction of GSH/GSSG efflux transporter by TCDD, the GSH:GSSG ratio decreased and global protein S-glutathionylation levels in liver extracts increased in response to oxidative stress along with the induction of glutamate-cysteine ligase catalytic subunit (Gclc), glutathione synthetase (Gss), glutathione disulfide reductase (Gsr), and glutathione transferase π (Gstp). Furthermore, levels of ophthalmic acid, a biomarker of oxidative stress indicating GSH consumption, were also increased. Collectively, the data suggest that increased cystine transport due to cystine/glutamate Xc- antiporter induction compensated for decreased cysteine production following repression of the transsulfuration pathway to support GSH synthesis in response to TCDD-induced oxidative stress.

Dioxin-elicited decrease in cobalamin redirects propionyl-CoA metabolism to the ß-oxidation-like pathway resulting in acrylyl-CoA conjugate buildup.

2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) is a persistent environmental contaminant that induces diverse biological and toxic effects, including reprogramming intermediate metabolism, mediated by the aryl hydrocarbon receptor. However, the specific reprogramming effects of TCDD are unclear. Here, we performed targeted LC-MS analysis of hepatic extracts from mice gavaged with TCDD. We detected an increase in S-(2-carboxyethyl)-L-cysteine, a conjugate from the spontaneous reaction between the cysteine sulfhydryl group and highly reactive acrylyl-CoA, an intermediate in the cobalamin (Cbl)-independent ß-oxidation-like metabolism of propionyl-CoA. TCDD repressed genes in both the canonical Cbl-dependent carboxylase and the alternate Cbl-independent ß-oxidation-like pathways as well as inhibited methylmalonyl-CoA mutase (MUT) at lower doses. Moreover, TCDD decreased serum Cbl levels and hepatic cobalt levels while eliciting negligible effects on gene expression associated with Cbl absorption, transport, trafficking, or derivatization to 5'-deoxy-adenosylcobalamin (AdoCbl), the required MUT cofactor. Additionally, TCDD induced the gene encoding aconitate decarboxylase 1 (Acod1), the enzyme responsible for decarboxylation of cis-aconitate to itaconate, and dose-dependently increased itaconate levels in hepatic extracts. Our results indicate MUT inhibition is consistent with itaconate activation to itaconyl-CoA, a MUT suicide inactivator that forms an adduct with adenosylcobalamin. This adduct in turn inhibits MUT activity and reduces Cbl levels. Collectively, these results suggest the decrease in MUT activity is due to Cbl depletion following TCDD treatment, which redirects propionyl-CoA metabolism to the alternate Cbl-independent ß-oxidation-like pathway. The resulting hepatic accumulation of acrylyl-CoA likely contributes to TCDD-elicited hepatotoxicity and the multihit progression of steatosis to steatohepatitis with fibrosis.

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) dysregulates hepatic one carbon metabolism during the progression of steatosis to steatohepatitis with fibrosis in mice.

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD), a persistent environmental contaminant, induces steatosis that can progress to steatohepatitis with fibrosis, pathologies that parallel stages in the development of non-alcoholic fatty liver disease (NAFLD). Coincidently, one carbon metabolism (OCM) gene expression and metabolites are often altered during NAFLD progression. In this study, the time- and dose-dependent effects of TCDD were examined on hepatic OCM in mice. Despite AhR ChIP-seq enrichment at 2 h, OCM gene expression was not changed within 72 h following a bolus dose of TCDD. Dose-dependent repression of methionine adenosyltransferase 1A (Mat1a), adenosylhomocysteinase (Achy) and betaine-homocysteine S-methyltransferase (Bhmt) mRNA and protein levels following repeated treatments were greater at 28 days compared to 8 days. Accordingly, levels of methionine, betaine, and homocysteic acid were dose-dependently increased, while S-adenosylmethionine, S-adenosylhomocysteine, and cystathionine exhibited non-monotonic dose-dependent responses consistent with regulation by OCM intermediates and repression of glycine N-methyltransferase (Gnmt). However, the dose-dependent effects on SAM-dependent metabolism of polyamines and creatine could not be directly attributed to alterations in SAM levels. Collectively, these results demonstrate persistent AhR activation disrupts hepatic OCM metabolism at the transcript, protein and metabolite levels within context of TCDD-elicited progression of steatosis to steatohepatitis with fibrosis.

A toxicogenomic approach for the risk assessment of the food contaminant acetamide.

Acetamide (CAS 60-35-5) is detected in common foods. Chronic rodent bioassays led to its classification as a group 2B possible human carcinogen due to the induction of liver tumors in rats. We used a toxicogenomics approach in Wistar rats gavaged daily for 7 or 28 days at doses of 300 to 1500 mg/kg/day (mkd) to determine a point of departure (POD) and investigate its mode of action (MoA). Ki67 labeling was increased at doses ≥750 mkd up to 3.3-fold representing the most sensitive apical endpoint. Differential gene expression analysis by RNA-Seq identified 1110 and 1814 differentially expressed genes in male and female rats, respectively, following 28 days of treatment. Down-regulated genes were associated with lipid metabolism while up-regulated genes included cell signaling, immune response, and cell cycle functions. Benchmark dose (BMD) modeling of the Ki67 labeling index determined the BMD10 lower confidence limit (BMDL10) as 190 mkd. Transcriptional BMD modeling revealed excellent concordance between transcriptional POD and apical endpoints. Collectively, these results indicate that acetamide is most likely acting through a mitogenic MoA, though specific key initiating molecular events could not be elucidated. A POD value of 190 mkd determined for cell proliferation is suggested for risk assessment purposes.

The food contaminant acetamide is not an in vivo clastogen, aneugen, or mutagen in rodent hematopoietic tissue.

Acetamide (CAS 60-35-5) is classified by IARC as a Group 2B, possible human carcinogen, based on the induction of hepatocellular carcinomas in rats following chronic exposure to high doses. Recently, acetamide was found to be present in a variety of human foods, warranting further investigation. The regulatory body JECFA has previously noted conflicting reports on acetamide's ability to induce micronuclei (MN) in mice in vivo. To better understand the potential in vivo genotoxicity of acetamide, we performed acute MN studies in rats and mice, and a subchronic study in rats, the target species for liver cancer. In the acute exposure, animals were gavaged with water vehicle control, 250, 1000, or 2000 mg/kg acetamide, or the positive control (1 mg/kg mitomycin C). In the subchronic assay, bone marrow of rats gavaged at 1000 mg/kg/day (limit dose) for 28 days was evaluated. Both acute and subchronic exposures showed no change in the ratio of polychromatic to total erythrocytes (P/E) at any dose, nor was there any increase in the incidence of micronucleated polychromatic erythrocytes (MN-PCE). Potential mutagenicity of acetamide was evaluated in male rats gavaged with vehicle control or 1500 mg/kg/day acetamide using the in vivoPig-a gene mutation assay. There was no increase in mutant red blood cells or reticulocytes in acetamide-treated animals. In both acute and sub-chronic studies, elevated blood plasma acetamide in treated animals provided evidence of systemic exposure. We conclude based on this study that acetamide is not clastogenic, aneugenic, or mutagenic in vivo in rodent hematopoietic tissue warranting a formal regulatory re-evaluation.

2,3,7,8-Tetrachlorodibenzo-p-dioxin dose-dependently increases bone mass and decreases marrow adiposity in juvenile mice.

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) and other aryl hydrocarbon receptor (AhR) agonists have been shown to regulate bone development and remodeling in a species-, ligand-, and age-specific manner, however the underlying mechanisms remain poorly understood. In this study, we characterized the effect of 0.01-30 µg/kg TCDD on the femoral morphology of male and female juvenile mice orally gavaged every 4 days for 28 days and used RNA-Seq to investigate gene expression changes associated with the resultant phenotype. Micro-computed tomography revealed that TCDD dose-dependently increased trabecular bone volume fraction (BVF) 2.9- and 3.3-fold in male and female femurs, respectively. Decreased serum tartrate-resistant acid phosphatase (TRAP) levels, combined with a reduced osteoclast surface to bone surface ratio and repression of femoral proteases (cathepsin K, matrix metallopeptidase 13), suggests that TCDD impaired bone resorption. Increased osteoblast counts at the trabecular bone surface were consistent with a reciprocal reduction in the number of bone marrow adipocytes, suggesting AhR activation may direct mesenchymal stem cell differentiation towards osteoblasts rather than adipocytes. Notably, femoral expression of transmembrane glycoprotein NMB (Gpnmb; osteoactivin), a positive regulator of osteoblast differentiation and mineralization, was dose-dependently induced up to 18.8-fold by TCDD. Moreover, increased serum levels of 1,25-dihydroxyvitamin D3 were in accordance with the renal induction of 1α-hydroxylase Cyp27b1 and may contribute to impaired bone resorption. Collectively, the data suggest AhR activation tipped the bone remodeling balance towards bone formation, resulting in increased bone mass with reduced marrow adiposity.

Convergence of hepcidin deficiency, systemic iron overloading, heme accumulation, and REV-ERBα/ß activation in aryl hydrocarbon receptor-elicited hepatotoxicity.

Persistent aryl hydrocarbon receptor (AhR) agonists elicit dose-dependent hepatic lipid accumulation, oxidative stress, inflammation, and fibrosis in mice. Iron (Fe) promotes AhR-mediated oxidative stress by catalyzing reactive oxygen species (ROS) production. To further characterize the role of Fe in AhR-mediated hepatotoxicity, male C57BL/6 mice were orally gavaged with sesame oil vehicle or 0.01-30µg/kg 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) every 4days for 28days. Duodenal epithelial and hepatic RNA-Seq data were integrated with hepatic AhR ChIP-Seq, capillary electrophoresis protein measurements, and clinical chemistry analyses. TCDD dose-dependently repressed hepatic expression of hepcidin (Hamp and Hamp2), the master regulator of systemic Fe homeostasis, resulting in a 2.6-fold increase in serum Fe with accumulating Fe spilling into urine. Total hepatic Fe levels were negligibly increased while transferrin saturation remained unchanged. Furthermore, TCDD elicited dose-dependent gene expression changes in heme biosynthesis including the induction of aminolevulinic acid synthase 1 (Alas1) and repression of uroporphyrinogen decarboxylase (Urod), leading to a 50% increase in hepatic hemin and a 13.2-fold increase in total urinary porphyrins. Consistent with this heme accumulation, differential gene expression suggests that heme activated BACH1 and REV-ERBα/ß, causing induction of heme oxygenase 1 (Hmox1) and repression of fatty acid biosynthesis, respectively. Collectively, these results suggest that Hamp repression, Fe accumulation, and increased heme levels converge to promote oxidative stress and the progression of TCDD-elicited hepatotoxicity.

Comparative analysis of TCDD-induced AhR-mediated gene expression in human, mouse and rat primary B cells.

2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) is a persistent environmental pollutant that activates the aryl hydrocarbon receptor (AhR) resulting in altered gene expression. In vivo, in vitro, and ex vivo studies have demonstrated that B cells are directly impaired by TCDD, and are a sensitive target as evidenced by suppression of antibody responses. The window of sensitivity to TCDD-induced suppression of IgM secretion among mouse, rat and human B cells is similar. Specifically, TCDD must be present within the initial 12h post B cell stimulation, indicating that TCDD disrupts early signaling network(s) necessary for B lymphocyte activation and differentiation. Therefore, we hypothesized that TCDD treatment across three different species (mouse, rat and human) triggers a conserved, B cell-specific mechanism that is involved in TCDD-induced immunosuppression. RNA sequencing (RNA-Seq) was used to identify B cell-specific orthologous genes that are differentially expressed in response to TCDD in primary mouse, rat and human B cells. Time course studies identified TCDD-elicited differential expression of 515 human, 2371 mouse and 712 rat orthologous genes over the 24-h period. 28 orthologs were differentially expressed in response to TCDD in all three species. Overrepresented pathways enriched in all three species included cytokine-cytokine receptor interaction, ECM-receptor interaction, focal adhesion, regulation of actin cytoskeleton and pathways in cancer. Differentially expressed genes functionally associated with cell-cell signaling in humans, immune response in mice, and oxidation reduction in rats. Overall, these results suggest that despite the conservation of the AhR and its signaling mechanism, TCDD elicits species-specific gene expression changes.

Ozone-Induced Type 2 Immunity in Nasal Airways. Development and Lymphoid Cell Dependence in Mice.

Inhalation exposures to ozone commonly encountered in photochemical smog cause airway injury and inflammation. Elevated ambient ozone concentrations have been epidemiologically associated with nasal airway activation of neutrophils and eosinophils. In the present study, we elucidated the temporal onset and lymphoid cell dependency of eosinophilic rhinitis and associated epithelial changes in mice repeatedly exposed to ozone. Lymphoid cell-sufficient C57BL/6 mice were exposed to 0 or 0.5 parts per million (ppm) ozone for 1, 2, 4, or 9 consecutive weekdays (4 h/d). Lymphoid cell-deficient, Rag2(-/-)Il2rg(-/-) mice were similarly exposed for 9 weekdays. Nasal tissues were taken at 2 or 24 hours after exposure for morphometric and gene expression analyses. C57BL/6 mice exposed to ozone for 1 day had acute neutrophilic rhinitis, with airway epithelial necrosis and overexpression of mucosal Ccl2 (MCP-1), Ccl11 (eotaxin), Cxcl1 (KC), Cxcl2 (MIP-2), Hmox1, Il1b, Il5, Il6, Il13, and Tnf mRNA. In contrast, 9-day ozone exposure elicited type 2 immune responses in C57BL/6 mice, with mucosal mRNA overexpression of Arg1, Ccl8 (MCP-2), Ccl11, Chil4 (Ym2), Clca1 (Gob5), Il5, Il10, and Il13; increased density of mucosal eosinophils; and nasal epithelial remodeling (e.g., hyperplasia/hypertrophy, mucous cell metaplasia, hyalinosis, and increased YM1/YM2 proteins). Rag2(-/-)Il2rg(-/-) mice exposed to ozone for 9 days, however, had no nasal pathology or overexpression of transcripts related to type 2 immunity. These results provide a plausible paradigm for the activation of eosinophilic inflammation and type 2 immunity found in the nasal airways of nonatopic individuals subjected to episodic exposures to high ambient ozone.

Pyruvate Kinase Isoform Switching and Hepatic Metabolic Reprogramming by the Environmental Contaminant 2,3,7,8-Tetrachlorodibenzo-p-Dioxin.

The environmental contaminant 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) elicits dose-dependent hepatotoxicity that includes fat accumulation, inflammation, and fibrosis that may progress to hepatocellular carcinoma. To further investigate these effects, RNA-Seq data were integrated with computationally identified putative dioxin response elements, and complementary targeted metabolomic and aryl hydrocarbon receptor (AhR) ChIP-Seq data from female C57BL/6 mice gavaged with TCDD every 4 days for 28 days. Data integration using CytoKEGG with manual curation identified dose-dependent alterations in central carbon and amino acid metabolism. More specifically, TCDD increased pyruvate kinase isoform M2 (PKM2) gene and protein expression. PKM2 has lower catalytic activity resulting in decreased glycolytic flux and the accumulation of upstream intermediates that were redirected to the pentose phosphate pathway and serine/folate biosynthesis, 2 important NADPH producing pathways stemming from glycolysis. In addition, the GAC:KGA glutaminase (GLS1) protein isoform ratio was increased, consistent with increases in glutaminolysis which serves an anaplerotic role for the TCA cycle and compensates for the reduced glycolytic flux. Collectively, gene expression, protein, and metabolite changes were indicative of increases in NADPH production in support of cytochrome P450 activity and ROS defenses. This AhR-mediated metabolic reprogramming is similar to the Warburg effect and represents a novel advantageous defense mechanism to increase anti-oxidant capacity in normal differentiated hepatocytes.

Comparisons of differential gene expression elicited by TCDD, PCB126, ßNF, or ICZ in mouse hepatoma Hepa1c1c7 cells and C57BL/6 mouse liver.

The aryl hydrocarbon receptor (AhR) is a promiscuous receptor activated by structurally diverse synthetic and natural compounds. AhR activation may lead to ligand-specific changes in gene expression despite similarities in mode of action. Therefore, differential gene expression elicited by four structurally diverse, high affinity AhR ligands (2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD; 10nM, 30 µg/kg), 3,3',4,4',5-pentachlorobiphenyl (PCB126; 100nM, 300µg/kg), ß-naphthoflavone (ßNF; 10 µM, 90 mg/kg), and indolo[3,2-b]carbazole (ICZ; 1µM)) in mouse Hepa1c1c7 hepatoma cells and C57BL/6 mouse liver samples were compared. A total of 288, 183, 119, and 131 Hepa1c1c7 genes were differentially expressed (|fold-change|≥ 1.5, P1(t)≥ 0.9999) by TCDD, ßNF, PCB126, and ICZ, respectively. Only ∼35% were differentially expressed by all 4 ligands in Hepa1c1c7 cells. In vivo, 661, 479, and 265 hepatic genes were differentially expressed following treatment with TCDD, ßNF, and PCB126, respectively. Similar to Hepa1c1c7 cells, ≤ 34% of gene expression changes were common across all ligands. Principal components analysis identified time-dependent gene expression divergence. Comparisons of ligand-elicited expression between Hepa1c1c7 cells and mouse liver identified only 11 common gene expression changes across all ligands. Although metabolism may explain some ligand-specific gene expression changes, PCB126, ßNF, and ICZ also elicited divergent expression compared to TCDD, suggestive of selective AhR modulation.

Assessment of energetic costs of AhR activation by ß-naphthoflavone in rainbow trout (Oncorhynchus mykiss) hepatocytes using metabolic flux analysis.

Exposure to environmental contaminants such as activators of the aryl hydrocarbon receptor (AhR) leads to the induction of defense and detoxification mechanisms. While these mechanisms allow organisms to metabolize and excrete at least some of these environmental contaminants, it has been proposed that these mechanisms lead to significant energetic challenges. This study tests the hypothesis that activation of the AhR by the model agonist ß-naphthoflavone (ßNF) results in increased energetic costs in rainbow trout (Oncorhynchus mykiss) hepatocytes. To address this hypothesis, we employed traditional biochemical approaches to examine energy allocation and metabolism including the adenylate energy charge (AEC), protein synthesis rates, Na(+)/K(+)-ATPase activity, and enzyme activities. Moreover, we have used for the first time in a fish cell preparation, metabolic flux analysis (MFA) an in silico approach for the estimation of intracellular metabolic fluxes. Exposure of trout hepatocytes to 1µM ßNF for 48h did not alter hepatocyte AEC, protein synthesis, or Na(+)/K(+)-ATPase activity but did lead to sparing of glycogen reserves and changes in activities of alanine aminotransferase and citrate synthase suggesting altered metabolism. Conversely, MFA did not identify altered metabolic fluxes, although we do show that the dynamic metabolism of isolated trout hepatocytes poses a significant challenge for this type of approach which should be considered in future studies.