Animal care (housing, husbandry conditions) and animal procedures were performed in accordance with the European Council Directive of 24 November, 1986 (86/609/EEC), and the present study was approved by the state animal care commission (Koblenz; approval number, 23 177-07/G 14-1-010). Mice received standard food for rodents (Altromin Lage, Nr. 1314) with free access to food and water. They were kept in groups of five siblings of the same sex per cage with constant temperatures of 22-24˚C and humidity of 55±10% as well as a 12-h day and night rhythm. Male Oct3-knockout (FVB.Slc22a3tm1Dpb, Oct3^-/-) (17), their WT littermates (FVB) and C57BL/6 mice (in total n=51), 4-6 weeks old with an average body weight of 20 g at the start of the experiment, were used in this study. Oct3^-/- mice were kindly provided by Prof. Schinkel, Cancer Centre Amsterdam. C57BL/6 and WT mice were bred by the Translational Animal Research Centre (TARC) of the University Medical Centre, Johannes Gutenberg-University Mainz. To investigate the relevance of Oct3 expression and the effects on cholestasis and fibrosis, two different animal models of fibrosis were analysed: i) Chemically induced liver fibrosis by the application of pro-fibrotic carbon tetrachloride (CCl₄) or thioacetamide (TAA) for 6 weeks; and ii) cholestasis-associated fibrosis after 7 days of bile duct ligation (BDL).

Total RNA was extracted from livers of three 5-week-old untreated WT and Oct3^-/- mice using the High Pure RNA tissue kit (cat. no./ID: 11828665001; Roche Diagnostics) following the manufacturer's instructions. RNA quantity and purity were estimated using a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies) and integrity was assessed by Agilent 2100 Bioanalyzer (Agilent Technologies). cDNA libraries were generated using the QuantSeq 3'mRNA-Seq Library Prep kit for Illumina (Lexogen, Vienna, Austria) following the manufacturer's instructions (21). RNA sequencing was performed using Illumina HiSeq Rapid Mode by the Institute of Human Genetics, Department of Genomics, Life & Brain Center, University of Bonn. The sequencing kit was HiSeq 3000/4000 SBS Kit (single read, 50 cycles) (cat. no./ID: FC-410-1001; Illumina). Coverage was standard 3' Seq. The loading concentration of DNA was 0.06-0.44 nmol assuming a nucleotide length of 100-300 bp. Data were deposited at the BioProject database (http://www.ncbi.nlm.nih.gov/bioproject/685115, BioProject ID PRJNA685115). The read sequences were aligned to the Mus_musculus.GRCm38.74 reference genome followed by read mapping and read counting, as described before using the Bioconductor package Rsubread (V 1.24.2) (22). Before aligning reads, low quality reads were filtered, reads containing adapter sequences, and duplicate mapping reads using Bioconductor package ShortRead (V 1.32.1) (23). For differential expression analysis (WALD-Test) the Bioconductor package DESeq2 (V 1.14.1) with an adjusted P-value <0.01 was used (24). All data analysis was performed using R programming language and related packages.

Functional classification and network analysis were performed using Ingenuity Pathway Analysis (Ingenuity Systems Inc.). The significance of each network, function and pathway was determined by the scoring system provided by Ingenuity Pathway Analysis tool. Data will be provided on demand.

C57BL/6, WT and Oct3^-/- mice, 4-6 weeks old, were treated with pro-fibrotic thioacetamide (TAA) or CCl₄ for 6 weeks (25). TAA was injected intraperitoneally three times a week in escalating doses, starting with 50 mg/kg (doses 1 and 2, week 1), 100 mg/kg (doses 2 to 5, weeks 1-2), 200 mg/kg (doses 6 to 10, weeks 2-4), 300 mg/kg (doses 11 to 15, weeks 4-5), and 400 mg/kg (dose 16 onwards, week 6). Placebo intraperitoneal injection served as the control. CCl₄ was administered three times a week by oral gavage in escalating doses 50/50 vol/vol mixed with mineral oil: 0.875 ml/kg (dose 1 dose, week 1), 1.75 ml/kg (doses 2 to 7, weeks 1-2), 2.5 ml/kg (doses 8 to 13, weeks 3-4), and 3.25 ml/kg (after week 4). Oral gavage of mineral oil served as the control. Animals were culled by cervical dislocation after 6 weeks of treatment or after 1 to 4 weeks of reversal, death was confirmed by loss of heartbeat through direct cardiac palpation and tissues were harvested for qPCR and histological analysis.

WT and Oct3^-/- mice, 7-10 weeks old (body weight 18-20 g), underwent bile duct ligation (BDL) or placebo surgery (sham operation) as previously described under anaesthesia with 100 mg/kg Ketamine and 20 mg/kg Rompun (i.p) (26-28). Animals were sacrificed by cervical dislocation after 7 days; death was confirmed by loss of heartbeat and tissues were harvested for qPCR and histological analysis.

Total RNA was extracted from liver tissue using the High Pure RNA Tissue Kit (Roche Diagnostics) and cDNA synthesis was performed using the iScript cDNA Synthesis kit (Bio-Rad) according to the manufacturer's recommendations. Quantitative analysis of Oct1 (Slc22A1) transcripts was performed by quantitative real-time reverse transcriptase (RT-) polymerase chain reaction (qPCR). The Quantitect SYBR-Green PCR Kit (Qiagen) and validated primers of a Quantitect Primer Assay with the primer sets Mm_SLC22A1_2_SG (OCT1; 84 bp fragment), Mm_HNF4α (HNF4α; 100 bp fragment forward, 5'-GGATATGGCCGACTACAGCG-3' and reverse, 5'-AGATGGGGACGTGTCATTGC-3') and Mm_GAPDH_3_SG (GAPDH; 144 bp fragment) (Qiagen) were used according to the manufacturer's instructions. For the amplification, an initial denaturation at 95˚C for 15 min, followed by 15 sec at 94˚C, 30 sec at 55˚C and 30 sec at 72˚C for 40 cycles was used. Samples were run on a LightCycler^® 480 real-time PCR system (Roche Diagnostics). The relative expression levels were calculated by normalisation to GAPDH gene expression using the LightCycler^® 480 software Release 1.5.0.

Total protein extracts were prepared in sample buffer pH 8.0 containing 20 mM Tris, 5 mM EDTA, 0.5% Triton X-100 and EDTA-free protease inhibitors (Complete Mini, 1:25; Roche Diagnostics). For western blot analysis 60 µg total protein was separated by a 12% SDS-PAGE gel. The gel was transferred onto a nitrocellulose transfer membrane (OPTITRAN BA-S85/Whatman) following separation. Rabbit anti-HNF4α monoclonal antibody (1:1,000; Abcam) or goat anti-actin polyclonal antiserum (1:1,000; Santa Cruz Biotechnology, Inc.) were used as the primary antibodies. Horseradish peroxidase (HRP)-conjugated anti-rabbit or anti-goat IgG (Santa Cruz Biotechnology, Inc.) was used as the secondary antibody at a 1:10,000 dilution. Protein bands were visualised using Western Lightning^® Plus-ECL enhanced chemiluminescent substrate (Perkin Elmer).

Primary murine hepatocytes were incubated with rabbit-polyclonal-anti Hnf4α (Bioss Antibodies Inc.) as the primary antibody after preincubation with hydrogen peroxide for blocking of endogenous peroxidase. Endogenous biotin was blocked with the Avidin-Biotin Blocking kit (Vector Laboratories) and contaminating proteins were inhibited by ROTI^®-Immunoblock solution (ROTH). After incubation with the secondary antibody (goat anti-rabbit IgG-Biotin, 1:1,000; Dako Cytomation), the TSA™ Cyanine system (Perkin Elmer) was added. For the negative control, the primary antibody was omitted. The images were evaluated under a fluorescence microscope (Olympus BX51, Olympus U-RFL-T).

HepG2 (ATCC^® HB-8065™), a human liver cancer cell line, and HuH7 (RRID: CVCL_0336), a well differentiated hepatocyte-derived carcinoma cell line, were grown at 37˚C in a humidified atmosphere (5% CO₂) in plastic culture flasks (Falcon 3112; Becton-Dickinson). The medium was Dulbecco's modified Eagle's medium (31885-023; Life Technologies) supplemented with 10% foetal calf serum (Life Technologies). Medium was changed every 2-3 days and the culture was split every 7 days.

The pcDNAOCT1 and pcDNAOCT3 plasmids and an empty vector (Invitrogen; Thermo Fisher Scientific, Inc.) were stably transfected into HepG2 and HuH7 cells by mixing with the Attractene Transfection Reagent (Qiagen) according to the instructions of the manufacturer. Primary hepatocytes were isolated from Oct3^-/- and WT mice and cultured in collagen-coated 24-well culture plates (2.5x10⁵/ml) as previously described (29). For functional inhibition of the transporters, primary murine hepatocytes were treated with different doses (0, 50, 100 and 150 µM) of the standard non-selective OCT inhibitor quinine (Sigma-Aldrich; Merck KGaA) for 48 h (30-35).

Data management and statistical analysis were performed with Prism version 7.0 (GraphPad Software, Inc.). Results are expressed as means ± SEM and represent data from a minimum of three independent experiments assessed in triplicates. Three biological replicates were assumed being the minimum for any inferential analysis (biological repetition). As sample numbers were small, normal distribution was assumed. Therefore, no normality test was necessary. When two groups were compared, unpaired Student's t-test was used. Data with more than two groups were analysed by one-way or two-way ANOVA with Dunnett's multiple comparisons test after one-way ANOVA and Tukey-Kramer test after two-way ANOVA. For Pearson's correlation analysis SPSS program (version 23.0; IBM Corp.) was used. P<0.05 was considered statistically significant.

Transcriptome analysis showed that Hnf4α is one of the top upstream regulators in Oct3^-/- mice (P<0.001), with 110 target molecules. Hnf4α plays a pivotal role in regulating various transmembrane proteins and enzymes in Oct3^-/- mice (Fig. 1A). The majority of genes regulated by Hnf4α were upregulated in Oct3^-/- mice (Fig. 1B). Other significantly upregulated (positive z-score) upstream regulators were the (proto-)oncogenes myc (P=1.59x10^-13; z=2.21) and kras (P=5.43x10^-7; z=0.77), while the tumour suppressor tp53 was significantly downregulated (negative z-score) in Oct3^-/- mice (P=1.1x10^-7; z=-3.15) (Fig. 1C).

Untreated Oct3^-/- mice did not show differences in Hnf4α mRNA expression in comparison to WT littermates at the age of 4 weeks (Fig. 2A). Hnf4α mRNA expression was significantly downregulated in cholestatic Oct3^-/- mice (n=6) in comparison to WT mice (n=8) 7 days after BDL (P<0.01) (Fig. 2B).

Also, after chemical fibrosis induction with 6 weeks of CCl₄ treatment, Hnf4α mRNA expression was significantly downregulated in Oct3^-/- mice (n=7) as compared to WT mice (n=9) (P<0.001) (Fig. 2C).

Fibrosis was induced with TAA and CCl₄ treatment for 6 weeks in C57BL/6 mice (n=5), which are susceptible to conventional toxin-induced fibrosis progression and reversal models. Hnf4α mRNA expression was quantified by qPCR at the end of the treatment period and after up to four weeks of reversal. After 6 weeks of TAA and CCl₄ treatment, Hnf4α mRNA expression was significantly downregulated in fibrotic mouse livers (P<0.01 compared to baseline). After reversal for one and four weeks, the Hnf4α mRNA level increased again (Fig. 2D and E). Hnf4α mRNA expression correlated well with Oct1 mRNA expression (Fig. 2F).

Oct regulation cannot be easily studied, as the transporters are not relevantly expressed in cell lines (36). Therefore, experiments with stably OCT1- and OCT3-transfected tumour cell lines (HepG2 and HuH7, n=4) and primary hepatocytes isolated from Oct3^-/- (n=6) and WT (n=4) mice were performed. Proof that transfection with pcDNAOCT1 and pcDNAOCT3 induced overexpression of OCT1 and OCT3 compared with the empty vector was provided as Fig. S1. Hnf4α mRNA expression was significantly upregulated in OCT1- and OCT3-transfected HepG2 and HuH7 cells compared with in tumour cells transfected with empty vector (Fig. 3A) and primary Oct3^-/- hepatocytes (Fig. 3B) after treatment with the Oct inhibitor quinine (P<0.01). Western blots and immunofluorescence in primary WT and Oct3^-/- hepatocytes showed an increase of Hnfα protein expression with escalating quinine doses (Figs. 3C and S2-4). These data clearly show that functional loss of Oct induces the expression of Hnf4α. Interestingly, immunofluorescence of primary murine hepatocytes showed that Hnf4α was not only increased with escalating quinine doses, but theHnf4α distribution also differed between Oct3^-/- and WT hepatocytes. While Hnf4α was located in the cytosol of WT hepatocytes, Oct3^-/- hepatocytes showed nuclear Hnf4α expression, indicating that Oct3 affects Hnf4α in vivo (Figs. 3C and S5).