**4. Discussion**

A balanced bile acid homeostasis is a basic requirement for healthy normal liver function, including enzymes involved in the production and detoxification of bile acids, as well as sinusoidal and canalicular transporters mediating the bile acid flow. Recent research results point to an involvement of impaired bile acid homeostasis in PA-induced hepatotoxicity [16,17]. Therefore, the present study initially investigated the influence of PA treatment on the gene expression of 14 hepatobiliary transporters by qRT-PCR. This revealed a significant downregulation of gene expression of five ABC transporters and six sinusoidal SLC transporters. Additionally, the expression of several enzymes involved in bile acid homeostasis, synthesis, and detoxification was analyzed, also showing a significant PA-dependent downregulation. Furthermore, the effects of PA treatment on the gene expression of various transcriptional regulators were determined. a clear repression of the transcription regulators was shown, especially for *NR1I3* and *NR1I2*, followed by *HNF4A* and *NR0B2*. In general, these effects were more pronounced after 24 h of PA treatment than after 14 days of PA treatment. In regard to the chemical structure of the PA, the strongest effects were observed for the retronecine-type PAs echimidine and senecionine, whereas heliotridine-type PA heliotrine exerted the weakest effects. Similar results regarding the influence on gene expression of hepatobiliary transporters and also of enzymes involved in bile acid metabolism could be observed by Luckert et al. (2015) in primary human hepatocytes [16]. After treatment with 100 μM of the PAs echimidine, heliotrine, senecionine, and senkirkine, a downregulation of the gene expression of *SLC10A1*, *SLC22A7*, *SLCO1B1*, and *SLCO2B1* was observed. In accordance with the results presented here, Luckert et al. (2015) also observed a PA-related downregulation of the canalicular transporters *ABCC2* and *ABCB11* in primary human hepatocytes in their microarray analysis. The authors also observed a PA-induced downregulation for the enzymes *BAAT*, *CYP7A1*, *CYP8B1*, *CYP27A1*, and *SULT2A4* in primary human hepatocytes. Furthermore, the bioinformatic analysis of the microarray data performed by Luckert et al. (2015) predicted an inhibition of the transcriptional regulators NR1I3 and HNF4A, whereas for NR0B2, an activation was predicted. In agreement with the results of the present study, Xiong et al. [39] observed a reduction in the gene expression of *Slc10a1* and *Slco1b2* in rat livers after treatment with senecionine (35 mg per kg body weight, 36 h). In contrast, no effect on the expression of *ABCB11* was observed in rat livers. However, after treatment with *Senecio vulgaris*, a PAproducing plant, Xiong et al. (2014) also observed a reduced gene expression of *Abcb11* and the SLC transporters *Slc10a1*, *Slco1a2*, and *Slco1b2* in rat livers [40]. a reduced expression of *Cyp7a1* and *Baat* was detected in rat livers after treatment with senecionine, whereas the expression of *Cyp8b1*, in contrast to the results presented here, was not affected [39]. After treatment with *Senecio vulgaris*, reduced gene expression of *Baat*, but not of *Cyp7a1* and *Cyp8b1*, was observed in rat livers [40]. Consistent with the results of the present study, the two in vivo studies mentioned above also described a reduced gene expression of *NR0B2* in rat livers after treatment with senecionine or *Senecio vulgaris* [39,40].

The reduced expression of the transporters *SLC10A1*, *SLC22A7*, *SLC22A9*, *SLC51A*, *SLCO1B1*, and *SLCO2B1*, determined in the present study, may possibly lead to a reduced uptake of bile acids, since these transporters are located in the basolateral membrane of hepatocytes and are responsible for the uptake of bile acids and other organic anions from sinusoidal blood. In contrast, the transporters ABCC2 and ABCB11 are located on the canalicular side, and are responsible for the secretion of bile acids into the bile ducts. Therefore, a reduced gene expression of these transporters may result in a reduced secretion of bile acids. In this context, in the present study, an impairment of hepatobiliary transport was detected in vitro via fluorescence microcopy. Similar effects were also observed in vivo in mice [17]. Furthermore, inhibition of ABCB11, in particular by accumulation of bile acids in the hepatocytes, has already been described for the development of cholestasis [25,41,42]. Regarding the PA-induced downregulation of various transcription factors, PAs may also affect bile acid, cholesterol homeostasis, and associated signaling pathways via interaction with transcriptional regulators. Furthermore, the observed PA-related downregulation of the enzymes *CYP27A1*, *CYP39A1*, *CYP7A1*, and *CYP8B1* may inhibit bile acid synthesis, which could be due to a possible feedback regulation induced by accumulating toxic bile acids. In addition, the decreased gene expression of *CYP3A4*, *BAAT*, *SULT2A1*, and *UGT2B4* indicates a potential decrease in bile acid metabolism and conjugation, which in turn would lead to a reduced detoxification of bile acids.

With regard to the downregulation of *CYP7A1*, the rate-limiting enzyme of bile acid synthesis, the present study used a reporter gene assay to investigate the inhibition of

transcriptional activity of the *CYP7A1* promoter in the hepatocyte cell line HepG2. The observed inhibition of the *CYP7A1* promoter activity after PA treatment was not concentrationdependent. Additionally, the effect of PA treatment on the gene expression of *CYP7A1* in HepG2 cells was investigated. Whereas in the metabolically highly active HepaRG cells PA treatment caused a significantly reduced *CYP7A1* expression, in HepG2 cells, only a slight effect on the gene expression of *CYP7A1* was observed after 24 h of PA exposure. Since HepG2 cells generally exert low metabolic activity, the results suggest that the PA metabolites may be responsible for the reduced *CYP7A1* gene expression shown in HepaRG cells.

Impairment of tight junctions, changes in the cytoskeleton, oxidative stress, and induction of apoptosis/necrosis are also key events associated with the development of cholestasis [20,24–26]. The induction of apoptotic processes by PA treatment, both intrinsic and sensitized to extrinsic apoptosis, has been shown in various studies [29,43]. Damage to tight junctions can lead to an increase in paracellular permeability, reflux of bile constituents into the sinusoidal space of Disse and plasma, and a reduction of the osmotic gradient, which is normally the driving force for bile secretion [24,44,45]. With regard to a possible influence on the tight junction protein ZO-1, protein location was investigated by immunofluorescence staining. ZO-1 showed discontinuities in the staining only after 24 h of treatment with senecionine (35 μM), indicating a potential damage of the tight junctions. For the other three PAs echimidine, heliotrine, and senkirkine, no influence on ZO-1 could be detected, suggesting that this mechanism might only be marginally involved in a potential PA-associated cholestasis. To the best of our knowledge, no further studies investigating a PA-related effect on hepatocellular tight junction proteins are available.

A key event in the development of cholestasis is the accumulation of bile acids, leading to the induction of apoptosis/necrosis, oxidative stress, and a regulation of nuclear receptors and gene expression. Various in vivo studies have shown elevated bile acid levels in rat serum after administration of both senecionine and *Senecio vulgaris* [17,39,40]. The authors concluded that PA treatment resulted in an excess accumulation of bile acids in hepatocytes, which led to an adaptive response to avoid bile acid overload in hepatocytes. This adaptive response includes suppression of bile acid de novo synthesis (reduced gene expression of e.g., *Cyp7a1*), reduced sinusoidal reabsorption (reduced expression of e.g., *Slc10a1*), and reduction of bile acid accumulation by activation of alternative basolateral export pumps (Abcc3 and Abcc4). In the present in vitro study, a potential suppression of bile acid synthesis and sinusoidal uptake by regulation of corresponding enzymes and transporters (e.g., *SLC10A1* and *CYP7A1*) was shown in HepaRG cells at the mRNA level. However, no increased mRNA expression of alternative export pumps was detected, but, on the contrary, a downregulation of *ABCC3* was detected. In accordance with the studies of Xiong et al. (2014) [39,40], further studies showed an increase in bile acids in the serum and an increase in liver toxic enzyme markers and bilirubin in horses, as well as in calves, after ingestion of Senecio species or *Cynoglossum officinale* [46–49]. All of these data suggest that PA treatment leads to increased serum bile acid levels. However, in the present in vitro study, reduced bile acid levels were measured intra- and extracellularly in PA-treated cells compared to solvent-treated cells. The strength of the reduction is associated with the previously observed cytotoxicity [29] and the described potency of the tested PA [29,50]. a possible explanation for the observed reduced bile acid levels may be that PAs intracellularly disturb bile acid metabolism in HepaRG cells so dramatically that significantly less bile acids are formed and thus secreted.

In summary, we observed a strong PA-mediated impairment of bile acid homeostasis in human HepaRG cells, comprising a downregulation of numerous genes involved in bile acid uptake, synthesis, detoxification, and secretion. Furthermore, bile acid secretion seems to be disturbed, and thus might contribute to the development of cholestatic liver disease. Generally, the most pronounced effects were observed for senecionine and echimidine, representing PAs of the retronecine type. Using the HepaRG cell model and the molecular

initiating and key events addressed in the AOP for cholestasis, we could confirm the evidence from in vivo studies for the development of cholestasis in our in vitro cell model.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2304-815 8/10/1/161/s1, Figure S1: Influence of PA on CYP7A1 gene expression in HepG2 cells, Figure S2: PAdependent disturbance of ABCC-2 driven efflux in HepaRG cells (5 μM echimidine or senecionine), Figure S3: PA-dependent disturbance of ABCC-2 driven efflux in HepaRG cells (5 μM heliotrine or senkirkine), Figure S4: PA-dependent disturbance of ABCC-2 driven efflux in HepaRG cells (35 μM heliotrine or senkirkine), Table S1: Alterations of gene expressions of hepatobiliary transport proteins after 24 h PA treatment in HepaRG cells, Table S2: Alterations of gene expressions of hepatobiliary transport proteins after 14 days PA treatment in HepaRG cells, Table S3: Alterations of gene expressions of enzymes involved in bile acid/cholesterol metabolism after 24 h PA treatment in HepaRG cells, Table S4: Alterations of gene expressions of enzymes involved in bile acid/cholesterol metabolism after 14 days PA treatment in HepaRG cells, Table S5: Alterations of gene expressions of transcription regulators involved in bile acid/cholesterol metabolism after 24 h PA treatment in HepaRG cells, Table S6: Alterations of gene expressions of transcription regulators involved in bile acid/cholesterol metabolism after 14 days PA treatment in HepaRG cells.

**Author Contributions:** Conceptualization, J.W., A.B., and S.H.-P.; methodology and investigation, J.W., J.G., M.H., and C.L.; software, J.W. and S.H.-P.; writing—original draft preparation, J.W. and S.H.-P.; writing—review and editing, M.H., C.L., and A.B.; visualization, J.W. and S.H.-P.; supervision, A.B. and S.H.-P.; project administration, S.H.-P.; funding acquisition, S.H.-P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the German Research Foundation (Grant number LA1177/12-1) and by the German Federal Institute for Risk Assessment (grant numbers 1322-591 and 1322-624).

**Data Availability Statement:** The data presented in this study are available in the Supplementary Materials.

**Conflicts of Interest:** The authors declare no conflict of interest.
