Rat and dog quad-culture liver chip models: characterization and use to interrogate a potential flavin-containing monooxygenase-mediated, species-specific toxicity of a histamine receptor antagonist

Microphysiological systems (MPS) contain multiple cell types in three dimensions and often incorporate fluidic shear forces. There is interest in MPS for disease and efficacy modeling, safety and disposition studies. Animal cell-based MPS are needed to provide confidence in the translation of data from human cell-based MPS. We developed rat and dog quad-culture liver MPS incorporating primary hepatocytes, sinusoidal endothelial, Kupffer, and stellate cells. Using cryopreserved primary cells, we established a protocol for co-culturing cells under physiological flow conditions. Cells were evaluated for viability, morphology, and function (e.g. albumin production, cytochrome P450, and flavin-containing monooxygenase [FMO] activity). Optimized culture conditions maintained high-quality rat and dog liver chips for up to 7 days. Model performance was evaluated with ABT-288, a histamine-3 receptor antagonist that caused elevated serum transaminases in dogs but not rats. This finding was partially attributed to the high levels of FMO-mediated N-oxide metabolites produced in the dog. Key findings in our study were (i) dog chips showed much higher FMO-mediated N-oxidation compared with rat, and (2) dog chips exhibited modestly higher sensitivity to ABT-288 toxicity endpoints (albumin, alanine transaminase, and lactate dehydrogenase) compared with rat. Species differences in N-oxidation were not observed in rat and dog liver microsomes or 2D hepatocyte monocultures, suggesting that properties of the quad-culture MPS were necessary to model higher FMO activity observed in dogs in vivo. The data suggest that this preclinical species liver chip model provides novel understanding of in vitro to in vivo translation of ABT-288 dog liver toxicity.