Search

Mouse models are frequently used to study chronic liver diseases (CLDs). To assess their translational relevance, we quantified the similarity of commonly used mouse models to human CLDs based on transcriptome data. Gene‐expression data from 372 patients were compared with data from acute and chronic mouse models consisting of 227 mice, and additionally to nine published gene sets of chronic mouse models. Genes consistently altered in humans and mice were mapped to liver cell types based on single‐cell RNA‐sequencing data and validated by immunostaining. Considering the top differentially expressed genes, the similarity between humans and mice varied among the mouse models and depended on the period of damage induction. The highest recall (0.4) and precision (0.33) were observed for the model with 12‐months damage induction by CCl4 and by a Western diet, respectively. Genes consistently up‐regulated between the chronic CCl4 model and human CLDs were enriched in inflammatory and developmental processes, and mostly mapped to cholangiocytes, macrophages, and endothelial and mesenchymal cells. Down‐regulated genes were enriched in metabolic processes and mapped to hepatocytes. Immunostaining confirmed the regulation of selected genes and their cell type specificity. Genes that were up‐regulated in both acute and chronic models showed higher recall and precision with respect to human CLDs than exclusively acute or chronic genes. Conclusion: Similarly regulated genes in human and mouse CLDs were identified. Despite major interspecies differences, mouse models detected 40% of the genes significantly altered in human CLD. The translational relevance of individual genes can be assessed at https://saezlab.shinyapps.io/liverdiseaseatlas/., In the present study, we observed that – although major interspecies differences remain – improved mouse models show up to 40% of the gene expression changes seen in liver tissue of human NAFLD or NASH, which is much higher than previously reported. Moreover, we identified gene sets consistently regulated in human and mouse chronic liver disease. Based on single‐cell RNA‐sequencing data set we mapped liver cell types to those genes and validated them by immunostaining.

Small-molecule flux in tissue-microdomains is essential for organ function, but knowledge of this process is scant due to the lack of suitable methods. We developed two independent techniques that allow the quantification of advection (flow) and diffusion in individual bile canaliculi and in interlobular bile ducts of intact livers in living mice, namely Fluorescence Loss After Photoactivation (FLAP) and Intravital Arbitrary Region Image Correlation Spectroscopy (IVARICS). The results challenge the prevailing 'mechano-osmotic' theory of canalicular bile flow. After active transport across hepatocyte membranes bile acids are transported in the canaliculi primarily by diffusion. Only in the interlobular ducts, diffusion is augmented by regulatable advection. Photoactivation of fluorescein bis-(5-carboxymethoxy-2-nitrobenzyl)-ether (CMNB-caged fluorescein) in entire lobules demonstrated the establishment of diffusive gradients in the bile canalicular network and the sink function of interlobular ducts. In contrast to the bile canalicular network, vectorial transport was detected and quantified in the mesh of interlobular bile ducts. In conclusion, the liver consists of a diffusion dominated canalicular domain, where hepatocytes secrete small molecules and generate a concentration gradient and a flow-augmented ductular domain, where regulated water influx creates unidirectional advection that augments the diffusive flux.

Cholestasis, the impairment of bile flux out of the liver, is a common complication of many pathological liver disorders, such as cholangiopathies, primary biliary sclerosis, and primary biliary cirrhosis. Besides accumulation of bile acids in the liver and blood, it leads to a proliferative response of the biliary tree termed as a ductular reaction. The ductular reaction is characterized by enhanced proliferation of cholangiocytes, which form the epithelial lining of bile ducts. This strong reaction of the biliary tree has been reported to generate a source of progenitor cells that can differentiate to hepatocytes or cholangiocytes during regeneration. On the other hand, it can cause periportal fibrosis eventually progressing to cirrhosis and death. In 2D histology, this leads to the appearance of an increased number of duct lumina per area of tissue. Yet, the biliary tree is a 3D vstructure and the appearance of lumina in thin slices may be explained by the appearance of novel ducts or by ramification or convolution of existing ducts in 3D. In many such aspects, traditional 2D histology on thin slices limits our understanding of the response of the biliary tree. A comprehensive understanding of architecture remodeling of the biliary network in cholestasis depends on robust 3D sample preparation and analysis methods. To that end, we describe pipe-3D, a processing and analysis pipeline visualization based on immunofluorescence, confocal imaging, surface reconstructions, and automated morphometry of the biliary network in 3D at subcellular resolution. This pipeline has been used to discover extensive remodeling of interlobular bile ducts in cholestasis, wherein elongation, branching, and looping create a dense ductular mesh around the portal vein branch. Surface reconstructions generated by Pipe-3D from confocal data also show an approximately fivefold enhancement of the luminal duct surface through corrugation of the epithelial lamina, which may increase bile reabsorption and alleviate cholestasis. The response of interlobular ducts in cholestasis was shown to be in sharp contrast to that of large bile ducts, de novo duct formation during embryogenesis. It is also distinct from ductular response in other models of hepatic injury such as choline-deficient, ethionine-supplemented diet, where parenchymal tissue invasion by ducts and their branches is observed. Pipe-3D is applicable to any model of liver injury, and optionally integrates tissue clearing techniques for 3D analysis of thick (>500 μm) tissue sections.

Since xenobiotics enter the organism via the liver, hepatocytes must cope with numerous perturbations, including modifications of proteins leading to endoplasmic reticulum stress (ER-stress). This triggers a signaling pathway termed unfolded protein response (UPR) that aims to restore homeostasis or to eliminate disturbed hepatocytes by apoptosis. In the present study, we used the well-established CCl4 hepatotoxicity model in mice to address the questions whether CCl4 induces ER-stress and, if so, whether the well-known ER-stress effector CHOP is responsible for CCl4-induced apoptosis. For this purpose, we treated mice with a high dose of CCl4 injected i.p. and followed gene expression profile over time using Affymetrix gene array analysis. This time resolved gene expression analysis allowed the identification of gene clusters with overrepresented binding sites for the three most important ER-stress induced transcription factors, CHOP, XBP1 and ATF4. Such result was confirmed by the demonstration of CCl4-induced XBP1 splicing, upregulation of CHOP at mRNA and protein levels, and translocation of CHOP to the nucleus. Two observations indicated that CHOP may be responsible for CCl4-induced cell death: (1) Nuclear translocation of CHOP was exclusively observed in the pericentral fraction of hepatocytes that deteriorate in response to CCl4 and (2) CHOP-regulated genes with previously reported pro-apoptotic function such as GADD34, TRB3 and ERO1L were induced in the pericentral zone as well. Therefore, we compared CCl4 induced hepatotoxicity in CHOP knockout versus wild-type mice. Surprisingly, genetic depletion of CHOP did not afford protection against CCl4-induced damage as evidenced by serum GOT and GPT as well as quantification of dead tissue areas. The negative result was obtained at several time points (8, 24 and 72 h) and different CCl4 doses (1.6 and 0.132 g/kg). Overall, our results demonstrate that all branches of the UPR are activated in mouse liver upon CCl4 treatment. However, CHOP does not play a critical role in CCl4-induced cell death and cannot be considered as a biomarker strictly linked to hepatotoxicity. The role of alternative UPR effectors such as XBP1 remains to be investigated.