, 2006), the 3D liver model used here appears to capture these drug toxicities in the absence of an additional stimulus such as LPS. One possible explanation is that drugs such as trovafloxacin and APAP may be capable of directly or indirectly (via e.g. a metabolite formed) activate Kupffer or HSC which then can exacerbate drug-induced toxicity by the release of pro-inflammatory

mediators. While in the 3D model the potential contribution of inflammation is part of the model itself, cultures where e.g. cytokine mixes are added on top of the drug bear the risk of inducing inflammation where in an in vivo situation there would not be such an effect and thus creating in vitro artifacts. The presence of the NPC in addition to hepatocytes increases the 3D liver culture sensitivity for detection of Apitolisib drug-induced toxicities

with a mode of action involving learn more inflammatory pathways triggered by e.g. Kupffer cells and thus are suggested to more accurately reflect physiological conditions. As expected, human 3D liver cells show higher donor-to-donor variability of protein secretion, CYP induction and response to drug-induced toxicity. These results are suggested to reflect the in vivo situation where inter-subject variability for example in induction of CYP1A1 by omeprazole ( Rost et al., 1994), CYP3A4 by rifampicin cAMP ( Ged et al., 1989) and drug-induced toxicity ( Sioud and Melien, 2007) are well-known phenomena ( Lehmann et al., 1998 and Sioud and Melien, 2007). In summary, we could provide experimental evidence

that the described 3D liver models of human and rat contain at least four main liver cell types, that the cell populations retain their functionality, and that they are stable during 3 months periods in culture. Our results demonstrate that 3D liver co-cultures can detect species-specific differences of drugs-induced toxicity which was not possible using hepatocyte monolayer cultures. We believe that the presence of NPC in addition to hepatocytes increased the sensitivity of the 3D liver model as such as that drug toxicity can be detected with therapeutically relevant concentrations. Furthermore the possibility of treating cells for long-periods of time allowed us to study time-dependent drug effects in vitro and to more accurately detect DILI compared to other commonly used cell culture models. This might help in the future to better assess possible drug-induced toxicities in animals and man. There is a strong need for robust long-term in vitro screening models, the use of which could reduce in the future the number of animals used in drug development. Taken together, our results demonstrated that the 3D liver model shown here can capture aspects of tissue physiology in vitro other cell models lack.