Developing high-fidelity hepatotoxicity models from pluripotent stem cells

Abstract

Faithfully recapitulating human physiology "in a dish" from a renewable source remains a holy grail for medicine and pharma. Many procedures have been described that, to a limited extent, exhibit human tissue-specific function in vitro. In particular, incomplete cellular differentiation and/or the loss of cell phenotype postdifferentiation play a major part in this void. We have developed an interdisciplinary approach to address this problem, using skill sets in cell biology, materials chemistry, and pharmacology. Pluripotent stem cells were differentiated to hepatocytes before being replated onto a synthetic surface. Our approach yielded metabolically active hepatocyte populations that displayed stable function for more than 2 weeks in vitro. Although metabolic activity was an important indication of cell utility, the accurate prediction of cellular toxicity in response to specific pharmacological compounds represented our goal. Therefore, detailed analysis of hepatocellular toxicity was performed in response to a custom-built and well-defined compound set and compared with primary human hepatocytes. Importantly, stem cell-derived hepatocytes displayed equivalence to primary human material. Moreover, we demonstrated that our approach was capable of modeling metabolic differences observed in the population. In conclusion, we report that pluripotent stem cell-derived hepatocytes will model toxicity predictably and in a manner comparable to current gold standard assays, representing a major advance in the field.

Keywords: Embryonic stem cells; Hepatocyte differentiation; Stem cell; Toxicity; iPS.

Figure 1.

Constructing stable models of human hepatocyte function. (A): Human induced pluripotent stem cells differentiated efficiently to hepatocytes using an established procedure and expressed ALB, HNF4a, AFP, E-CAD, CYP3A, and A1AT. (B): Human induced pluripotent stem cell-derived hepatocytes displayed hepatocyte function on a polyurethane surface in two separate experiments. TTR, fibrinogen, and fibronectin were detected by enzyme-linked immunosorbent assay as previously described [4]. (C): Human iPSC-derived hepatocytes displayed cytochrome P450 drug metabolism (CYP3A and CYP1A2) on a polyurethane surface in two separate experiments in a manner comparable to hESC-derived hepatocytes. n = 3 for each experiment; error bars represent the SD. Scale bar = 50 μm. Abbreviations: A1AT, α-1-antitrypsin; AFP, α-fetoprotein; ALB, albumin; CYP3A, cytochrome P450 3A; E-CAD, E-cadherin; hESC, human embryonic stem cell; HNF4α, hepatocyte nuclear factor 4α; iPSC, induced pluripotent stem cell; RLU, relative light units; TTR, transthyretin.

Figure 2.

Identifying a specific and pharmacologically relevant panel of compounds. (A, B): Cytochrome P450 gene overexpression was used to deliver metabolically active THLE5B (Tc5) hepatocyte populations. The different clones' metabolic activity was measured using high-performance liquid chromatography. CYP2C9 activity was assessed using the conversion of diclofenac to 4OH-diclofenac, and CYP2D6 activity was measured by conversion of dextromethorphan to dextrophen. (C): The human cell line THLE5B and transfected derivatives were used to screen more than 1 million compounds in the search for compounds that were metabolized exclusively by particular cytochrome P450s. BMS 1 and 2 were metabolized exclusively by 2C9 and 2D6, respectively. These experiments were repeated in freshly isolated primary human hepatocytes for BMS 1 and 2. IC₅₀ values are quoted in μM. Abbreviations: Conc, concentration; CYP, cytochrome P450.

Figure 3.

Comparison of stem cell-based model with industry gold standard. To ascertain the sensitivity of stem cell-derived hepatocytes to the BMS compounds, human embryonic stem cell (H9)-derived hepatocytes were incubated with BMS 1 and 2 at different concentrations (0, 5, 10, 20, and 50 μM) for 72 hours. Cell viability was measured at 72 hours after incubation using ATP Cell-Titer Glo. We observed a significant reduction in cell viability for both compounds at a concentration of 50 μM. n = 3; error bars represent the SD. Statistical significance was determined using Student's t test; *, p < .05. Abbreviation: RLU, relative light units.

Figure 4.

Model optimization, reproducibility, and translation to iPSC-derived hepatocytes. (A): Hepatocyte differentiation was extended for a further 4 days to improve sensitivity. Twenty days after replating, human embryonic stem cell (H9)-derived hepatocytes (n = 3) were incubated with BMS 1 and 2 at 50 μM for 72 hours. Cell viability was measured using ATP Cell-Titer Glo. Cell viability was significantly reduced in response to BMS 1 and 2. (B): Human induced pluripotent stem cell (33D6)-derived hepatocytes (n = 10) were incubated with BMS 1 and 2 at 50 μM for 72 hours. Cell viability was measured using ATP Cell-Titer Glo. iPSC hepatocyte viability was significantly reduced in response to BMS 2 and not BMS 1. (C): CYP2C9 activity was measured using the pGlo assay. CYP2C9 activity was greatest in hESC-derived hepatocytes (2,341 RLU/ml per milligram of protein). CYP2C9 activity was barely detectable in iPSC hepatocytes at 3.4 RLU/ml per milligram of protein. Error bars represent the SD. Statistical significance was determined using Student's t test; ***, p < .01. Abbreviations: hESC, human embryonic stem cell; iPSC, induced pluripotent stem cell; RLU, relative light units.