1,2-unsaturated pyrrolizidine alkaloids (PAs) are the most widespread natural toxins. Humans are directly exposed to these substances through the consumption of contaminated food. PA poisoning can lead to severe liver damage, such as hepatic veno-occlusive disease (VOD). PAs are not toxic per se. In the liver, their enzymatic activation occurs to form highly reactive metabolites that can form DNA and protein adducts. So far, little is known about the molecular mechanisms of action of PAs. This project aims to contribute to a better understanding of the molecular mode of action and, in particular, the structure-activity relationships of PAs in order to further elucidate the development of PA-induced toxicity in humans and to better assess a potential hazard to humans from PA ingestion. The aims of this project included elucidation of structure-dependent uptake across the intestinal epithelium in vitro (Caco-2/MDCKII cell model) and structure-dependent induction of liver toxic effects in vitro. Endpoints for analysis of liver toxicity included induction of cytotoxicity and apoptosis and disruption of bile acid homeostasis. The passage of structurally different PAs across the intestinal epithelium was studied using the Caco 2 and MDCKII cell models. In preliminary work for this project, it was hypothesized that PAs pass through the intestinal epithelium depending on their structure, which should be directly associated with ABCB1-dependent efflux. This hypothesis could not be confirmed as a result of this project. Potential ABCB-1-dependent efflux was shown to occur in all structure types regardless of necine base or degree of esterification. Nevertheless, we were able to show that PAs pass the intestinal epithelium to different degrees. Transferred to the in vivo situation, this implies that PAs enter the liver with portal blood. The structure-dependent induction of liver toxic effects was systematically investigated in the second part of this project. A transcriptome study in primary human hepatocytes treated with four structurally different PAs provided evidence for biological endpoints affected by PA. After establishing an appropriate liver cell model, induction of cytotoxicity, induction of apoptosis, and disruption of bile acid homeostasis were examined with a set of first 22, then 14 PAs in the human hepatoma cell line HepaRG. Our results clearly show strong grouping of PAs across all endpoints. The PA monoesters have no inducing effect on any of the end points studied, whereas cyclic PAs and open-chain diesters (especially of the heliotridine type) are particularly toxic. Also, the disturbance of bile acid homeostasis goes along with the grouping elaborated in the cytotoxicity studies. Thus, our studies in the liver show strong structure-dependent differences in the induction of toxic effects. This is possibly associated with different cellular uptake mechanisms and/or enzyme affinities for bioactivation.In a final part of the project, the induction of liver toxicity in vivo was investigated in mice using the PA senecionine as representative. This study demonstrated a time- and dose-dependent induction of severe liver injury associated with an increase in liver parameters ALT and AST. By using a reporter mouse with green-fluorescent endothelial cells, endothelial cells were specifically confirmed as a target of PA-induced liver toxicity. These effects were accompanied by pericentral hepatocellular necrosis, severe hemorrhage, and impaired bile acid secretion. In addition, infiltration of immune cells was observed. To investigate the metabolism¬dependency of this effect, Por-knockout mice that no longer exhibited CYP activity were used. In these mice, no liver toxic effects could be identified after senecionine treatment. Thus, we clearly demonstrated here that metabolic toxification by CYP must occur to form the potent liver toxic properties of senecionine.In summary, our results show that structure-dependent properties of PAs cannot be generalized. Although we identified a pattern in the liver, which could be reproduced in all levels of the studies, we could not find this grouping again with regard to passage across the intestinal epithelium, which, however, may also be associated with the chosen models. Thus, caution is advised to follow a uniform grouping; it is more important to consider the endpoints and questions separately.