*2.8. Analysis of Bile Acid Content*

The amount of different bile acids was detected via ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) in cell culture supernatants and cell lysates. Therefore, HepaRG cells were seeded at a density of 0.2 × <sup>10</sup><sup>5</sup> cells/well in six-well plates. After the proliferation and differentiation period of four weeks, the FBS level in the medium was reduced to 2% for 48 h. The cells were then washed once with pre-warmed PBS and incubated with either 5, 21, or 35 μM of PA, the solvent (1.7% DMSO, 0.35% ACN), or the positive control (20 μM of cyclosporine A) in an FBS-free medium to avoid interactions with bovine bile acids occurring in FBS [21]. The amount of cell culture medium was reduced to 1 mL per well to yield a higher concentration of secreted bile acids. Supernatants were collected after incubation for 48 h. Cells were washed twice with PBS, trypsinized for 15 min at 37 ◦C, and collected by adding 1 mL of PBS. After centrifugation at 250× *g* for 5 min, the supernatant was discarded. To increase the amount of bile acids, the cells and medium of three wells were pooled.

Bile acids were quantified using UPLC-MS/MS as described previously [37]. Briefly, bile acids in cell lysates and cell culture supernatants were extracted by adding methanol containing deuterated internal standards. After intensive mixing and centrifugation, methanol was evaporated under a stream of nitrogen. The samples were resuspended in a 1:1 (*v/v*) methanol:water mixture and injected onto the UPLC system (Infinity1290, Agilent Technologies, Santa Clara, CA, USA). Separation was performed on a Kinetex C18 column (Phenomenex, Torrance, CA, USA) using water and acetonitrile as mobile phases. Detection was operated in negative mode using a QTRAP 5500 mass spectrometer (Sciex, Concord, ON, Canada).

### *2.9. Statistical Analysis*

Statistical analysis was performed with SigmaPlot 13.0 software. Statistical differences between the mean values of the treatments and the solvent control were determined by one-way ANOVA, followed by Dunnett's post hoc test. Statistically significant differences were assumed at *p* < 0.05.

#### **3. Results**

#### *3.1. PA-Dependent Alterations of Gene Expression of Transporters, Enzymes, and Transcription Regulators Involved in Bile Acid Homeostasis*

Considering the microarray data of Luckert et al. (2015) [16] and the proposed mechanisms for the development of cholestasis according to the AOP for cholestatic liver diseases [26], 32 target genes were selected to examine possible effects on their expression after 24 h and 14 days of treatment with the four structurally different PAs echimidine, heliotrine, senecionine, and senkirkine via qRT-PCR. These 32 target genes comprise 14 hepatobiliary transporters, eight enzymes, and ten transcriptional regulators involved in bile acid homeostasis. Data of cell viability were recently summarized in Waizenegger et al. (2018) [29]. Based on these results, the following three concentrations were chosen: 5 μM (non-cytotoxic), 35 μM (non- to slightly cytotoxic), and 70 μM (cytotoxic). The qRT-PCR results showed that only the higher concentrations (35 and 70 μM) affected gene expression, whereas no significant changes occurred after treatment with the lowest concentration (5 μM). Furthermore, the regulatory effects after 24 h of PA treatment seemed to be higher than after 14 days. Additionally, the strongest regulatory effects were found after treatment with the retronecine-type PA echimidine and senecionine, while the weakest were detected for the heliotridine-type PA heliotrine.

The gene expression data showed a significant downregulation of five ABC transporters (*ABCB4*, *ABCB11*, *ABCC2*, *ABCC3*, *ABCC6*) and six solute carrier (SLC) transporters (*SLC10A1*, *SLC22A7*, *SLC22A9*, *SLC51A*, *SLCO1B1*, and *SLCO2B1*). After 24 h of PA treatment (Figure 2A), the most pronounced effects were found for the three SLC transporters *SLC22A7*, *SLC22A9*, and *SLC51A*, followed by the two SLC transporters *SLC10A1* and *SLCO2B1*. a complete list of the gene expression values is provided in Table S1 in the Supplemental Materials. a lower but significantly reduced gene expression was observed for the three ABC transporters *ABCB4*, *ABCB11*, and *ABCC6*. The weakest significant regulatory effects were detected for the two ABC transporters *ABCC2* and *ABCC3*. For the two SLC transporters *SLC51B* and *SLCO1B3*, no significant changes in gene expression were observed, except for the treatment with 35 and 70 μM of senecionine, as well as 70 μM of heliotrine. In comparison, after 14 days of treatment, the strongest decrease in gene expression was also observed for the SLC transporters *SLC22A7* and *SLC51A*, while the regulatory effect on *SLC22A9*, as well as *SLCO1B1* and *SLCO2B1*, was significantly lower compared to the 24 h treatment (Figure 2A). In contrast, *ABCB11* was more downregulated after 14 days of PA treatment than after 24 h. The weakest significant downregulation was detected for the ABC transporters *ABCB4*, *ABCC2*, and *ABCC6* and the SLC transporter *SLC10A1*, whereas no significant effects were found for the three transporters *ABCC3*, *SLC51B*, and *SLCO1B3* after continuous PA treatment. Finally, for the transporter *ABCB1*, a significant downregulation of gene expression was only observed after 14 days of treatment with the retronecine-type PAs echimidine and senecionine.

Concerning the effects of PAs on several enzymes involved in bile acid homeostasis, gene expression data revealed the strongest downregulation for the three CYP monooxygenases *CYP3A4*, *CYP7A1*, and *CYP8B1* after 24 h of PA treatment, with the most prominent repression for *CYP7A1*, the rate-limiting enzyme in bile acid formation (see Figure 2B). Furthermore, a significant downregulation was found for the phase II enzymes *SULT2A1* (encoding sulfotransferase 2A1) and *UGT2B4* (encoding UDP-glucuronosyltransferase 2B4), as well as for *BAAT* (encoding bile acid-CoA:amino acid *N*-acyltransferase). All three enzymes are required for bile acid conjugation. Compared to the aforementioned enzymes, gene expression of the two CYP monooxygenases *CYP27A1* and *CYP39A1*, also involved in bile acid synthesis, was only slightly decreased after PA treatment. In line with the 24 h treatment, the 14-day treatment also led to the strongest decrease in gene expression of *CYP7A1*, followed by *CYP3A4* and *CYP8B1*. However, the downregulation of *BAAT*, *SULT2A1*, and *UGT2B4* was less pronounced after 14 days. In contrast to the 24 h PA treatment, no effect was found on the expression of *CYP27A1* and *CYP39A1* after 14 days.

Additionally, for both PA treatment schemes, gene expression data showed a downregulation of all 10 investigated transcription regulators (see Figure 2C). After 24 h, the strongest downregulation was observed for *NR1I3* (encoding the constitutive androstane receptor), followed by *NR1I2* (encoding the pregnane X receptor) and further on *HNF4A* (encoding the hepatocyte nuclear factor 4 alpha), *PPARA* (encoding the peroxisome proliferator activated receptor alpha), and *NR0B2* (encoding small heterodimer partner). The weakest downregulation was found for the five remaining transcription regulators *ESR1* (encoding the estrogen receptor alpha), *NR1H4* (encoding the farnesoid X receptor), *HNF1A* (encoding the hepatocyte nuclear factor 1 alpha), *INSIG2* (encoding the insulin induced gene 2), and *SREBF1* (encoding the sterol regulatory element-binding transcription factor 1). After 14 days of PA treatment, the strongest downregulation was also detected for

*NR1I3*, followed by *NR0B2*, *NR1I2*, and *SREBF1*. In line with the 24 h treatment, *NR1H4*, *HNF4A*, and *INSIG2* were only slightly downregulated after 14 days of PA treatment, whereas no regulation of gene expression was observed for *ESR1*, *HNF1A*, and *PPARA*.

**Figure 2.** PA-dependent alterations of bile acid homeostasis-associated gene expressions ((**A**): transporters, (**B**): enzymes, (**C**): nuclear receptors) in HepaRG cells. HepaRG cells were seeded and cultivated as described in the material and methods section. Subsequently, the cells were incubated for 24 h or 14 days with 5, 35, or 70 μM of PA or the solvent (Ctr; 1.7% DMSO and 0.7% ACN). The total RNA was isolated and transcribed into cDNA. Expression analysis was performed by qRT-PCR. Gene expression of the target gene was normalized to *ACTB* and referred to solvent control to obtain relative expression (2−ΔΔCt method). The heat map shows the relative expression values for up- and downregulation as positive and negative fold changes (means of three independent experiments with three replicates each). For better comparability, only the fold change range between −15 and 15 is shown. Values exceeding this range are marked with +. Statistical differences were evaluated using one-way ANOVA followed by Dunnett's test: \* *p* < 0.05, \*\* *p* < 0.005, \*\*\* *p* < 0.001. Em, echimidine; Sc, senecionine; Hn, heliotrine; Sk, senkirkine.

#### *3.2. PA-Dependent Inhibition of CYP7A1 Promoter Activity and Gene Expression*

Since PA treatment in HepaRG cells led to a strong decrease in *CYP7A1* gene expression (see Section 3.1), possible inhibitory effects of PAs on the transcriptional activity of *CYP7A1* were investigated using a reporter gene assay. Therefore, HepG2 cells were transfected with the reporter gene plasmid pGL4.14-CYP7A1-Prom and the control plasmid pcDNA3-RLuc.

Following 24 h of PA treatment (5, 35, 70, or 250 μM), the firefly and Renilla luciferase activities were determined. In HepG2 cells, even the highest PA concentration did not induce cytotoxicity. This might be due to the low expression level of phase I xenobioticmetabolizing enzymes. As shown in Figure 3, no concentration-dependent inhibition of *CYP7A1* promoter activity was observed for the four PAs. Although there was a significant decrease in firefly luciferase activity after exposure to the PAs echimidine and heliotrine, this inhibition was relatively constant across all four investigated concentrations (~1.5- and 1.9-fold for 5 μM and 250 μM of echimidine, respectively, and ~1.6- and 1.8-fold for 5 μM and 250 μM of heliotrine, respectively).

**Figure 3.** Interaction of PAs with *CYP7A1* promoter activity in HepG2 cells. HepG2 cells were cultivated and seeded as described in the materials and methods section. Cells were transfected with the reporter gene plasmid pGL4.14-CYP7A1-Prom (80 ng) and the control plasmid pcDNA3-Rluc (1 ng) for 6 h and subsequently treated with 5, 35, 70, or 250 μM of PA, solvent (Ctr; 2.5% ACN), or positive control (PC, 5 μM of PMA). After an incubation period of 24 h, the cells were lysed, and the firefly and Renilla luciferase activity was detected. The activity of the firefly luciferase was normalized to the activity of the Renilla luciferase and referred to solvent control (= 1) to obtain the x-fold induction. Shown are means ± standard deviations of three independent experiments with three replicates each. Statistical differences were evaluated using one-way ANOVA followed by Dunnett's test: \* *p* < 0.05, \*\* *p* < 0.005, \*\*\* *p* < 0.001. Em, echimidine; Sc, senecionine; Hn, heliotrine; Sk, senkirkine.

The treatment with the known inhibitor of *CYP7A1* promoter activity PMA resulted in a significant inhibition of transcriptional activity by about 2.5-fold. Furthermore, this substance showed a concentration-dependent repression. To investigate whether the effect on *CYP7A1* repression by PAs is dependent on the metabolism of the PA, the influence of the four PAs on *CYP7A1* gene expression in HepG2 cells was investigated. Treatment for 24 h with the retronecine-type PAs echimidine and senecionine did not lead to a significant change in *CYP7A1* gene expression in HepG2 cells, whereas treatment with 70 μM of the PAs heliotrine and senkirkine resulted in a weak but significant downregulation of 1.9- or 1.5-fold, respectively (data depicted in Supplemental Materials, Figure S1).
