BackgroundPluripotent stem cells (PSCs) such as embryonic stem cells and induced pluripotent stem cells are promising target cells for cell regenerative medicine together with recently advanced technology of in-vitro differentiation. However, residual undifferentiated stem cells (USCs) during in-vitro differentiation are considered a potential risk for development of cancer cells and nonspecific lineage cell types. In this study we observed that USCs still exist during hepatic differentiation, consequently resulting in poor quality of the hepatic population and forming teratoma in vivo. Therefore, we hypothesized that effectively removing USCs from in-vitro differentiation could improve the quality of the hepatic population and guarantee safety from risk of teratoma formation.MethodsHuman PSCs were differentiated to hepatocytes via four steps. YM155, a known BIRC5 inhibitor, was applied for removing the residual USCs on the hepatic differentiation. After YM155 treatment, hepatocyte development was evaluated by measuring gene expression, immunostaining and hepatic functions at each stage of differentiation, and forming teratomas were confirmed by cell transplantation with or without YM155.ResultsThe selected concentrations of YM155 removed USCs (NANOG+ and OCT4+) in a dose-dependent manner. As a result, expression of endodermal markers (SOX17, FOXA2 and CXCR4) at stage II of differentiation and hepatic markers (ALB, AFP and HNF4A) at stage III was up-regulated by YM155 treatment as well as the hepatic population (ALB+), and functions (ALB/urea secretion and CYP450 enzyme activity) were enhanced at the final stage of differentiation (stage IV). Furthermore, we demonstrated that NANOG and OCT4 expression remaining until stage III (day 15 of differentiation) completely disappeared when treated with YM155 and teratoma formation was effectively prevented by YM155 pretreatment in the in-vitro study.ConclusionsWe suggest that the removal of USCs using YM155 could improve the quantity and quality of induced hepatocytes and eliminate the potential risk of teratoma formation.Electronic supplementary materialThe online version of this article (doi:10.1186/s13287-017-0517-2) contains supplementary material, which is available to authorized users.

In vitro prediction of hepatotoxicity can enhance the performance of non-clinical animal testing for identifying chemical hazards. In this study, we assessed high-content analysis (HCA) using multi-parameter cell-based assays as an in vitro hepatotoxicity testing model using various hepatotoxicants and human hepatocytes such as HepG2 cells and human primary hepatocytes (hPHs). Both hepatocyte types were exposed separately to multiple doses of ten hepatotoxicants associated with liver injury whose mechanisms of action have been described. HCA data were obtained using fluorescence probes for nuclear size (Hoechst), mitochondrial membrane potential (TMRM), cytosolic free calcium (Fluo-4AM), and lipid peroxidation (BODIPY). Cellular alterations were observed in response to all hepatotoxicants tested. The most sensitive parameter was TMRM, with high sensitivity at a low dose, next was BODIPY, followed by Fluo-4AM. HCA data from HepG2 cells and hPHs were generally concordant, although some inconsistencies were noted. Both hepatocyte types showed mild or severe mitochondrial impairment and lipid peroxidation in response to several hepatotoxicants. The results demonstrate that the application of HCA to in vitro hepatotoxicity testing enables more efficient hazard identification, and further, they suggest that certain parameters could serve as sensitive endpoints for predicting the hepatotoxic potential of chemical compounds.

Biofluid‐based biomarkers provide an efficient tool for hazard identification of chemicals. Here, we explored the potential of microRNAs (miRNAs) as biomarkers for hepatotoxicity of chemicals by linking in vitro to in vivo animal models. A search of the literature identified candidate circulating miRNA biomarkers of chemical‐induced hepatotoxicity. The expression of candidate miRNAs (miR‐122, miR‐151a, miR‐192, miR‐193a, miR‐194, miR‐21, miR‐29c), was determined by real‐time reverse transcription‐polymerase chain reaction in in vivo acute liver injury induced by acetaminophen, and then were further compared with those of in vitro cell assays. Candidate miRNAs, except miR‐29c, were significantly or biologically upregulated by acetaminophen, at a dose that caused acute liver injury as confirmed by hepatocellular necrosis. Except miR‐122 and miR‐193a, other miRNAs elevated in in vivo models were confirmed by in vitro models using HepG2 cells, whereas they failed by in vitro models using human primary hepatocytes. These findings indicate that certain miRNAs may still have the potential of toxicological biomarkers in linking in vitro to in vivo hepatotoxicity.