RIPK3 promoter hypermethylation in hepatocytes protects from bile acid induced inflammation and necroptosis

Background & Aims Necroptosis facilitates cell death in a controlled manner and is employed by many cell types following injury. It plays a major role in various liver diseases, albeit the cell type-specific regulation of necroptosis in the liver and especially hepatocytes has not yet been conceptualized. Approaches & Results Here, we demonstrate that DNA methylation suppresses RIPK3 expression in human hepatocytes and HepG2 cells. In diseases leading to cholestasis the RIPK3 expression is induced in mice and humans in a cell-type specific manner. Over-expression of RIPK3 in HepG2 cells leads immediately to RIPK3 activation by phosphorylation that is further modulated by different bile acids. Conclusion Bile acids mediated RIPK3 activation facilitates the secretion and expression of IL-8 via the JNK-pathway, suggesting hepatocytes suppress RIPK3 expression to protect themselves from bile acid induced necroptosis and inflammation but in chronical liver diseases associated with cholestasis induction of RIPK3 expression may be an early event signaling danger and repair through release of IL-8. Graphical abstract


Graphical abstract Introduction
Regulated cell death signaling events, e.g. apoptosis, necrosis, ferroptosis, necroptosis or pyroptosis, are crucial events for maintenance of tissue homeostasis.
These pathways are involved in the induction and modulation of the immune response [1,2] and tissue regeneration [3,4], where each component contains their specific regulatory mechanism and molecular components. [5] Necroptosis is a pro-inflammatory cell death type that signals damage and repair to many cell types following injury. [6,7] Mechanistically, formation of the necroptosis-defining necrosome requires phosphorylation of receptor-interacting serine/threonine-protein kinase 3 (RIPK3) [8] and activation of mixed lineage kinase domain-like protein (MLKL) leading to pore formation and cell death. [9][10][11] The first steps in the induction of necroptosis are shared with the apoptotic pathway. It has been reported that receptor-interacting serine/threonine-protein kinase 1 (RIPK1) is recruited to an active cytokine receptor, such as tumor necrosis factor receptor type-1 (TNFR1) where it is phosphorylated (pRIPK1) in a multiprotein complex, also known as complex I. [12,13] pRIPK1 interacts with Fas-associated death domain protein to activate caspase 8 (CASP8) inducing apoptosis. In the presence of CASP8 inhibition, RIPK3 is phosphorylated (pRIPK3) and this represents the first step of the necroptosis pathway. [14] pRIPK3 then associates in a multi-protein complex with MLKL, resulting in its phosphorylation (pMLKL). The pRIPK3/pMLKL complex is then recruited to a membrane where multiple pMLKL proteins oligomerize into a pore, ultimately leading to cell lysis. [15] Necroptosis is a promising target for future tumor therapeutics [16], however its role in the liver is controversial. [17][18][19][20][21][22] Studies have cast doubt on hepatocyte RIPK3 expression under physiological conditions. [23] By contrast, hepatocyte injury was reduced in RIPK3 knockouts in varied murine models of acute and chronic liver injury (ethanol-induced [18], ischemia-reperfusion [20], non-alcoholic fatty liver disease [22], and concanavalin-A hepatitis [24]), albeit no protection was seen against injury induced by acetaminophen. [25,26] Also, necroptosis is emerging as a critical mechanism in pathogenesis of cholestasis. [27] In cholestatic liver disease bile acids are retained and the bile flow is disrupted. [27] Bile acids are very abundant metabolites in hepatocytes that are known to be involved in the development of different liver diseases. [28,29] Altogether 15 bile acids are detected in humans and their formation is the primary pathway of cholesterol catabolism which is tightly regulated within the liver parenchyma to prevent the cytotoxic accumulation of bile acids. [30,31] Cholic acid (CA) and chenodeoxycholic acid (CDCA) are primary bile acids, described as the dominant but not exclusive forms present in the liver. Hepatocytes conjugate CA and CDCA to taurine or glycine during biotransformation before they are secreted into the bile. [31] In the colon bile acids are subjected to various microbial-mediated transformations including deconjugation and transformation of primary to secondary bile acids (lithocholic acid (LCA), ursodeoxycholic acid (UDCA)). [32] Owing to their structural formation, bile acids are classified by their hydrophobicity (hydrophobic: LCA > CDCA > CA > UDCA). [33] The hydrophobic bile acids (LCA, CDCA) are commonly known as toxic bile acids and potent inducers of apoptotic or necrotic cell death, whereas the hydrophilic bile acids are often described as cytoprotective. [34][35][36] Besides the well-known function of bile acids as detergents in the digestive tract and signaling under physiologic conditions, they are also highly active signaling molecules for eukaryotic cells in supraphysiological concentrations as they occur in various liver diseases. The importance of regulating inflammation has been 6 highlighted, for instance, in their ability to trigger inflammation and cell death. [37,38] Presence of pathological concentrations of bile acids in hepatocytes, by accumulation, induces different cell death mechanisms (e.g. apoptosis, necrosis or necroptosis). [36] This implies that hepatocytes must have acquired an endogenous mechanism to counteract bile acids' pro-inflammatory and cell-toxic properties, e.g.
due to the loss of key mediators of inflammation and cell death.
Here, we investigate the expression profile of hepatocellular RIPK3 protein, responsible for the induction of necroptosis, under physiological and pathological liver conditions. Further, we investigate RIPK3 regulatory mechanisms using bile acids in vitro that may prevent and trigger hepatocellular inflammation and tissue regeneration.

Cell isolation and culture
HepG2 cells were cultured in Dulbecco's Modified Eagle Medium containing F12 nutrient mix (DMEM:F12; Biozym) supplemented with 10% fetal calf serum (FCS) and 100 iU penicillin and 100 iU streptomycin at 37°C in a humidified atmosphere of 5% CO2. Prior to the day of the experiment, cells were washed with phosphatebuffered saline without calcium and magnesium (PBS) and cultured into either 6-, 12-or 96-well tissue culture plates depending on the experimental conditions. Plateable, cryopreserved primary human hepatocytes (pHep) from one male donor (18 years, BMI 28.7) were purchased from Lonza, Switzerland. Cells were thawed and cultured according to the manufacturer's instructions using the recommended hepatocyte culture media provided by Lonza. Hepatocytes were seeded at a density of 2x10 6 cells per well in a 6-well plate coated with collagen type I (Corning) at 10 µg cm -2 .
Primary human macrophages (pMФ) were isolated from the whole blood of healthy volunteers with Biocoll separating solution (Merck) according to the manufacturer's protocol and seeded at a density of 2x10 6 cells per well in a 6-well plate. After 5 days of differentiation by cultivating in Dulbecco's Modified Eagle Medium (DMEM; Lonza) supplemented with 10% FCS, 10 ng mL -1 M-CSF (ReproTech) and 10 µg mL -1 ciprofloxacin (Fresenius Kabi), cells were washed with PBS and subsequently used.
Purification of the primary cells was characterized by specific markers (Fig. 1A).

Mice
Male and female FVB/N and FVB/NRj mice at 8-12 weeks of age were used for all experiments. FVB/N and FVB/NRj mice were partially bred within the animal facility of the Jena University Hospital under specific pathogen-free conditions and purchased from Janvier Labs. The animals had access to conventional rodent chow and water ad libitum. They were maintained under a constant humidity (50-60%), 12hour light/dark cycle (incl. 20 min dim phases) and a constant temperature (24°C).
All experimental protocols were approved by the ethics committee and local government authorities in Thuringia, Germany.

Surgical animal models
Surgery was performed under anesthesia inhalation (1-2% Isoflurane, CP-Pharma and 100 mL min -1 carbogen) and 1 mg kg -1 body weight p.o. Meloxicam (0.5 mg mL -1 suspension, CP-Pharma) for pain-relief was given 1 h before surgery. A midline incision was used to open the abdomen and expose the bile duct and liver. After the surgical procedure (details for different models are given below) the abdominal layers were closed with 4-0 antibacterial suture (Ethicon) and Bupivacaine (2-4 mg kg -1 body weight, PUREN Pharma) was administered intra-incisional for topical anesthesia. The animals received Ringer acetate (20 mL kg -1 , Berlin Chemie AG) subcutaneously for fluid resuscitation and were offered an additional heat source (warm lamp) during the recovery phase. Normal and soaked food was available on the ground for the animals at all times after surgery. The animals were scored and weighed a minimum of twice daily. The scores were designed to reflect post-surgical conditions as well as specific symptoms of surgical intervention.
General surgery: This group was used as a sham group. After exposure of the liver and bile duct, the abdominal layers were closed without additional intervention.

Warm Ischemia Reperfusion injury (IR):
Intrasurgical a microvascular clamp was placed above the left lateral branch of the portal vein to interrupt the blood flow to the left lateral lobe. The liver was covered and kept moist with Ringer acetate while the body temperature was maintained at 37°C with a heating plate. After 60 min of partially hepatic ischemia, the clamp was removed to initiate the reperfusion.

Non-surgical animal models
Acetaminophen induced liver injury (APAP): Animals had been fastened for 16 h prior APAP injection to reduce metabolic activity and glutathione levels in the liver.
Acetaminophen (Sigma Aldrich) was dissolved in pre-warmed PBS and mice were treated with 300 mg kg -1 body weight intraperitoneally. [40] After the APAP injection, mice had free access to food and water. Sham animals received an intraperitoneal injection of PBS.

Hepatocyte isolation
Animals subjected to surgical or non-surgical models of liver injury were sacrificed after defined time points by an overdose of ketamine (> 300 mg kg -1 body weight, CP-Pharma) and xylazine (> 50 mg kg -1 body weight, Bayer). For hepatocyte isolation a liver perfusion was performed. The system is composed of a peristaltic pump with adjustable speed, a silicone tubing immersed in different buffers in the water bath (38°C) and a cannula (we used 25 G) at the tube outlet. The pump speed was set to a maximum of 5 mL min -1 (beginning) and was increased to 12 mL min -1 for perfusion after the cannulation of the portal vein. The perfusion was started using Krebs Henseleit Buffer (KHB, Biochrom) supplemented with 8 U mL -1 heparin (Heparin-Natrium-2500-ratiopharm, stock solution: 5000 I.E. mL -1 , Ratiopharm).
When the liver appeared pale, the perfusion medium was changed to Liver Digest Medium (LDM, ThermoFisher Scientific). Perfusion was maintained until the liver appeared digested. Livers were separated and kept in a cell culture dish with some LDM after perfusion and incubated for an additional 5 min at 37°C. The tissue was strained through a 70 µm nylon cell strainer (Corning) into a conical tube using approximately 30 mL of DMEM (ThermoFisher Scientific). Hepatocytes were isolated and purified by three times centrifugation for 4 min at 40 g and 4°C. Between each centrifugation step the supernatant was removed and replaced with 30-40 mL DMEM to wash the hepatocyte pellet. Aliquots and lysates were prepared after counting hepatocytes in 10 mL DMEM.

Methylation Analysis
Genomic DNA was extracted from cell cultures using the DNeasy blood & tissue kit (Qiagen) according to the manufacturer's protocol. Cells were then re-suspended in the proteinase K digestion buffer (40 µg proteinase K) and incubated at 50°C for 30 min. Cell debris was pelleted by centrifugation at 14000 g for 10 minutes. For DNA methylation analysis, 500 ng of the purified genomic DNA was bisulfite converted using the EZ DNA Methylation-Direct kit (Zymo Research). Bisulfitetreated DNA was purified according to the manufacturer's protocol and was eluted to a final volume of 46 µL. PCRs were performed using 1 µL bisulfite-treated DNA and 10 µmol L -1 of each target-validated primer from the EpigenDx Assay (ADS1678FP, ADS1678RPB). One primer was biotin-labeled and HPLC-purified to purify the final PCR product using sepharose beads. PCR products were bound to Streptavidin Sepharose HP (GE Healthcare Life Sciences) for immobilization. Afterwards the immobilized PCR products were purified, washed, denatured with 0.

RIPK3 staining of Human Samples
To investigate the relationship between RIPK3 expression and cholestasis, patients

Quantification of Bile Acids by Mass Spectrometry
Using a LC-MS/MS in-house assay concentration of 15 bile acids was determined in HepG2 cells and isolated primary murine hepatocytes with two different sample preparations. First, 3-fold (w/v) ethanol-phosphate-buffer (15% 0.01 mol L -1 phosphate buffer solution pH 7.5, 85% ethanol) was added to pre-weighed isolated primary murine hepatocytes. Samples were homogenized in a pebble mill (QiaShredder) and centrifuged at 16000 g for 5 min. The supernatant was used for bile acid quantification. Second, 1x10 6 HepG2 cells were seeded and incubated for two days. Afterwards, cells were washed two times with 500 µL PBS, trypsinized with 100 µL 0.05% trypsin-EDTA for 3 min at 37°C and 5% CO2 and incubated with 1 mL DMEM to stop the trypsin reaction. The cell suspension was then centrifuged for 3 min at 4°C and 260 g. Afterwards the cell pellet was washed by resuspending it twice in 500 µL 4°C cold PBS and centrifuging for 3 min at 4°C and 260 g. The washed pellet was then resuspended in 100 µL PBS and homogenized in a pebble mill (QiaShredder). After centrifugation for 5 min at 4°C and 13000 g, the supernatant was used for bile acid quantification. The sample preparation was then followed by protein precipitation and filtration of the samples. For quantification an Agilent 1200 high performance liquid chromatography system (Agilent Technologies GmbH, Germany) with a CTC-PAL autosampler coupled to an API 4000 Triple Quadrupole mass spectrometer with electrospray ionization source (AB Sciex, Germany) was used. All chromatographic separations were performed with a reverse-phase analytical column. The mobile phase consisted of water and methanol, both containing formic acid and ammonia acetate, at a total flow rate of 300 µL min -1 . Fold induction of IL-8 gene expression was calculated using the Pfaffl method. [42] Primers used are found in Table S1.

RIPK3 promoter methylation protects hepatocytes from necroptosis under physiological conditions
Important cellular signaling events during necroptosis are the phosphorylation of RIPK1, RIPK3 and MLKL which ultimately form an auto-lysing pore-complex.
Previous studies discussed the presence of necroptosis signaling and particularly the expression of RIPK3 in liver tissue versus hepatocytes in health and liver diseases with different results. [18,[43][44][45] In the hepatocellular carcinoma cell line HepG2 as well as primary human hepatocytes (pHep) RIPK3 was not expressed (Fig. 1A, Fig.   S1), which had been suggested previously from investigations on murine liver tissue and liver cell lines. [46] As expected primary human macrophages (pMФ), which served as a positive control in this experiment, expressed high levels of RIPK3 (Fig.   1A). [47][48][49] To analyze the purity of the isolated primary hepatocytes and macrophages, a western blot against CD68 and albumin was performed from the same lysates.
CD68, a protein highly expressed by monocytes and macrophages, thereby served as a macrophage specific marker whereas albumin, which has great abundance in hepatocytes, was used to characterize hepatocytes. We did not detect any expression of CD68 in the hepatocyte lysate, indicating a high purity of the samples (Fig. 1B). A low expression of albumin, however, was found in macrophages (Fig.   1B). This may be attributed to the cultivation medium that is bound to the macrophages despite washing steps or is the result of low expressed (advanced glycation end-product)-albumin in macrophages. [50] While RIPK3 expression was absent in hepatocytes, RIPK1 and MLKL were highly expressed in three cell types HepG2, pHep, and pMФ (Fig. 1C). Activation of necroptosis signaling in pMФ is often achieved using TNF-α, the pan-caspase inhibitor zVAD-fmk and an inhibitor of the inhibitor of apoptosis protein (IAP) family.
As previously described, this stimulation induces expression and phosphorylation of the necroptotic genes (e.g. RIPK1) (Fig. 1C). [51] After 6 h stimulation, pRIPK3 and pMLKL levels were elevated in pMФ, indicating initiation and propagation of necroptosis (Fig. 1C). [52] It is known that hepatocytes are susceptible to cytokine signaling and express TNFR1. [53] However, after 6 h of stimulation with the necroptosis inducing TNF-α/zVAD-fmk/IAP (TBZ)-mix only RIPK1 expression in the pHep increased. [54] pRIPK1, pRIPK3 and pMLKL were not detected while MLKL expression remained stable (Fig. 1C). Thus, in comparison to pMФ, pHep and HepG2 cell line lack an important condition to form MLKL-dependent pores and undergo necroptosis under physiological conditions after stimulation. As all signaling molecules necessary for necroptosis were present other than RIPK3, we performed DNA methylation-specific sequencing analysis of predicted CpG islands in human RIPK3 promoter regions to elucidate a cause for this absence (Fig. 1D). Eight promoter elements located at the 5' untranslated region (UTR) and initiation site of transcription of the RIPK3 gene were analyzed, located approximately -65 to +89 base pairs around the transcriptional start site. All analyzed regions in pHep were hypermethylated with relative methylated cytosine levels ranging from 37%-83% and a global methylation level over all analyzed areas of 65% (Fig. 1D). Similarly, HepG2 cells also showed hypermethylation in the promoter region of the RIPK3 gene. Yet, with a relative methylated cytosine amount of 21%-50% (and 35% globally), the hypermethylation is less pronounced in HepG2 cells than in pHep (Fig. 1D) and reflects the previous observation of global hypermethylation loss in hepatocellular carcinoma. [55] Nevertheless, the hypermethylation in HepG2 was still sufficient to suppress RIPK3 protein expression (Fig. 1D). In contrast, the same regions in primary pMФ expressing RIPK3 showed a relative methylation of 0%-10% and only 4% globally (Fig. 1D). Thus, hepatocytes and HepG2 cells have a hypermethylated RIPK3 promoter silencing protein expression, preventing these cells from undergoing necroptosis under physiological conditions (Fig. 1D).

Hepatocytes express RIPK3 under pathological conditions
To investigate the RIPK3 expression in liver diseases different murine models of liver injury were utilized (Tab. S2) and the expression of RIPK3 in comparison to human liver diseases was analyzed (Fig. 2). We chose the bile duct ligation model (BDL), which resembles chronic cholestasis as a consequence of a post-hepatic, sterile injury in many ways. [56] Further, we induced warm ischemia reperfusion injury (IR) as it is a common model with pericentral cell death and injury due to generation of oxidative stress and production of inflammatory cytokines and chemokines. [57] The model is used to mimic the pathophysiological events occurring during liver transplantation. [58] The acetaminophen induced liver injury (APAP) is an acute model of liver damage that mimics a drug induced liver injury that is a common adverse effect encountered in clinical practice. [59,60]  We isolated hepatocytes from all models and analyzed the purity and expression of RIPK3. The purity of each individual isolated mouse hepatocyte lysate was assessed by a western blot against the melanoma cell adhesion molecule (MCAM) also known as CD146 which is highly expressed in (liver sinusoidal) endothelial cells (LSEC) [62][63][64] as well as various immune cells (e.g. lymphocytes, macrophages) [65][66][67]. In most samples MCAM was not detected (Fig. S2) and only a few lysates (3 out of 72) we found contamination with MCAM expressing cells. The amount of samples contaminated for each individual group is depicted as "MCAM frequency" in Figure   2A.

In hepatocyte lysates obtained from animals after 24 h BDL and a systemic infection
(PCI), RIPK3 was highly expressed. Also, RIPK3 was detectable in the hepatocytes from the ischemic lobe of all animals after IR and in circa 60% of the respective nonischemic lobes (Fig. 2B). In contrast, only a low number of animals depicted a RIPK3 expression after abdominal surgery. Further, the RIPK3 expression levels found in this model was the lowest compared to the other disease models (Fig. 2B).
In the acutely toxic APAP liver injury model, as well as in the control groups receiving intraperitoneal injections of sterile PBS and Ringer acetate, RIPK3 expression was not detectable in isolated hepatocytes after 24 h ( Fig. 2A, Fig. 2B). All individual results of RIPK3 expression in the primary murine hepatocytes are shown in Figure   S3. These data demonstrate that under specific pathological conditions hepatocytes regulate RIPK3 expression and may undergo necroptosis. Further, the data suggest that this effect is more pronounced in chronic and cholestatic liver injury than in acute diseases. We then analyzed the bile acid composition in primary murine hepatocytes from the same animals which had been subjected to different models of liver injury. Consistent with literature reports, we identified taurine-conjugated CA (TCA) as the predominant bile acid in primary murine hepatocytes. [30] Taurineconjugated primary bile acids (TCA > TCDCA > TUDCA and TLCA) were the principle accumulated bile acids in our varied models of liver injury whereas unconjugated as well as glycine-conjugated bile acids were mostly below the lower limit of quantitation (LLOQ) (Fig. 2C). TCA concentration increased slightly in bacterially infected (PCI) animals whereas the animals with bile duct ligation (BDL) shows a significant increase compared to the PBS control group (Fig. 2D). The absence of unconjugated primary bile acids shows that the liver function of biotransformation and conjugation is functional whereas hepatic clearance of bile acids is decreased. This was seen before in a study analyzing bile acid composition in liver cirrhotic patients where TCA was the most changed bile acid in the liver. [68,69] We then analyzed the RIPK3 expression in pathological liver sections from patients It is of note that RIPK3 expression was found in hepatocytes, LSECs and immune cells of all specimens regardless of their bilirubin level, indicating that the RIPK3 expression is induced in hepatocytes upon various liver injuries (Fig. 2E). In all groups the mean RIPK3 fluorescence intensity (FI) in LSECs was lower than in hepatocytes. Further, the mean RIPK3 FI was significantly increased in specimens from patients suffering hyperbilirubinemia compared to the reference group. This effect was present in both cell types, but more pronounced in hepatocytes compared to LSECs (Fig. 2F).

Bile acids are sensitive to affect RIPK3 activation that induces inflammation
After confirming the expression of RIPK3 in cholestatic liver diseases we sought further the effects of bile acids on the activation of RIPK3 and subsequent signaling events. Therefore we overexpressed a previously characterized human RIPK3 (NM_006871.3) construct with an N-terminal FLAG-tag driven by a CMV promoter (RIPK3-FLAG, [39]) in HepG2 cells, that do not express RIPK3 endogenously (Fig.   3A). The transfection with the vector backbone alone did not lead to an expression of RIPK3, thus the detected RIPK3 is fully attributed to the RIPK3-FLAG construct in the expression vector (Fig. 3A). Further it is of note that the basal amount of bile acids in these HepG2 cells were mostly not detectable (nd) and therefore results with external bile acid stimulation may be attributed to the different bile acids supplemented for stimulation (Fig. 3B). We incubated HepG2 cells with bile acids at 50 µmol L -1 dissolved in methanol, a non-physiological concentration reported to be present in human liver tissue during cholestasis, [70], which they may take up by specialized transporters such as organic anion transporter (pumps) or through passive diffusion. [71] LCA and glycine-LCA (GLCA) exhibited cell toxicity and caused significant membrane damage at 50 µmol L -1 , as examined by quantifying lactate dehydrogenase in the supernatant (Supplementary Methods), and were therefore used at their highest non-toxic concentration for stimulation (LCA: 5 µmol L -1 , GLCA: 10 µmol L -1 ) (Fig. S4).
Overexpression of RIPK3-FLAG leads to a basal phosphorylation in HepG2 cells (Fig. 3A, Fig. 3C-F), suggesting that various endogenous mechanisms may trigger its activation. The unconjugated hydrophilic bile acids CA and UDCA increased RIPK3-FLAG phosphorylation and RIPK3-FLAG expression itself after 6 and 24 h (Fig. 3C, Fig. 3D). Also, CDCA shows a minor effect on the RIPK3-FLAG phosphorylation while not affecting the RIPK3-FLAG expression (Fig. 3E). In contrast LCA decreased both, RIPK3-FLAG phosphorylation and expression within 24 h (Fig. 3F). TCA, glycine-conjugated CA (GCA) as well as glycine-and taurineconjugated UDCA (TUDCA and GUDCA) reduced the stimulatory effects of RIPK3-FLAG phosphorylation of CA or UDCA but did not abolish them (Fig. 3C, Fig. 3D).
These findings indicate a high complexity of the underlying metabolic signaling network affecting RIPK3 expression and necroptosis on multiple levels. In the context of the heterogeneous bile acid toxicity these effects are also well known as the bile acid paradox. [72,73] In connection with the ability to induce RIPK3dependent necroptosis we hypothesized that epigenetic RIPK3 suppression allows hepatocytes to resist noxious stimulation from endogenous metabolites as bile acids and avoid resulting chronic pro-inflammatory and death signaling. Hepatic synthesis and metabolism of bile acids, and perhaps other metabolites, could lead to permanent activation of RIPK3 and subsequent inflammation or even cell death. [74] As noted above, bile acids modulate the expression and phosphorylation of RIPK3.
We observed, consistent with previous literature, a phospho-serine (Ser) 227 RIPK3-FLAG positive double band also detectable using the RIPK3 antibody. [8,75] As shown by Chen 2013, RIPK3 contains multiple phosphorylation sites (Ser199, Ser227) that exhibit different functions and are both indispensable for the necroptosis induction. [75] Ser199 is the important residue for induction of kinase activity [76] whereas Ser227 is crucial for the induction of necroptosis [75].
Additionally, ubiquitination was reported to be regulated during RIPK3 activation. [77] Both post translational modifications may lead to the observed mass shift in HepG2 cells. [78] We were able to exclude the ubiquitination of RIPK3-FLAG processing in our model as a consequence for the mass shift (Fig. S5). To investigate a hyperphosphorylation we incubated the protein lysate with calf intestinal phosphatase (CIP) or Lambda phosphatase (LP), which possess a high specificity for phospho-serine, -threonine and -tyrosine residues. CIP and LP, both abolished the pSer227 RIPK3-FLAG signal and diminished the mass shift observed in the total RIPK3-FLAG blot (Fig. 4A). Thus, bile acids are not only able to modulate the RIPK3 phosphorylation but further modulate its hyperphosphorylation, which is a necessity for the activation of necroptosis signaling. [79] The bile acid dependent activation and modification of RIPK3 primes the necroptotic pathway and stimulates a phosphorylation of MLKL which then translocate to the plasma membrane and induces cell rupture followed by an inflammatory reaction. In hepatocytes, the secretion of interleukin-8 (IL-8) is specific as an early inflammatory insult to recruit immune cells, especially neutrophils, and trigger secondary tissue damage as well as repair in various diseases. [80] Therefore human hepatoma HepG2 cell line was used as a representative model of the human liver that display a high degree of morphological and functional differentiation to generate reproducible results. [81] To mimicking a permanent activation of both important phosphorylation sites of RIPK3 (Ser199, Ser227), mutation of the serine residue to aspartic acid (Asp) promoted the IL-8 secretion significantly especially when both phosphorylation sites were mutated to activating Asp residue (Fig. 4B). This demonstrates that expression, activation and hyperphosphorylation of RIPK3, as under in vitro cholestatic conditions, results in a necroptosis-related inflammatory response in HepG2 cells. The IL-8 level increases significantly upon expression of RIPK3-FLAG in HepG2 cells and is further enhanced by stimulation with the bile acid CA, the bile acid with the strongest RIPK3-FLAG activating and phosphorylating function on protein level (Fig. 4C). This shows that the bile acid CA is able to induce the necroptotic pathway by activating RIPK3 resulting in inflammation. Besides secretion, RIPK3-FLAG expression and stimulation with different bile acids affected the gene expression of IL-8 (Fig. 4D). HepG2 cells that don't express detectable levels of endogenous RIPK3 decrease the expression of IL-8 after stimulation with different unconjugated bile acids which could be interpreted as a protective effect of the bile acids. RIPK3-FLAG transfection immediately leads to a strong expression of IL-8 that correlates with the direct phosphorylation of RIPK3-FLAG seen on protein level (Fig. 3, Fig. 4D). Stimulation of RIPK3-FLAG expressing HepG2 cells with unconjugated bile acids hardly affected the IL-8 gene expression. Solely the hydrophobic bile acid CDCA induced IL-8 gene expression significantly (Fig. 4D).
Strikingly, the hydrophilic bile acid CA which induced activation and phosphorylation of RIPK3-FLAG as well as secretion of IL-8 did not affect the gene expression of IL-8 ( Fig. 3C, Fig. 4C, Fig. 4D). Incubation with all glycine-and especially taurineconjugated bile acids however reduced the IL-8 gene expression in comparison to their unconjugated counterparts. This is attributable to the reduction of the hydrophobicity and in turn the increased cell protective properties due to increased hydrophilicity and impermeability to the cell membrane. [36,82] As stated in previous literature it is known that IL-8 secretion may be controlled by different pathways whereby the Jun-(N)-terminal kinase (JNK) signaling pathway is one prominent example. [83,84] Therefore we used the four primary, unconjugated bile acids (CA, CDCA, UDCA, LCA) as well as TCDCA, the taurine-conjugated form of the strongest IL-8 gene expression activator, to investigate JNK activation by phosphorylation at Thr183/ Tyr185. RIPK3-FLAG overexpression directly stimulated JNK phosphorylation that agreed with the immediately seen RIPK3-FLAG phosphorylation and increased IL-8 secretion and gene expression (Fig. 3, Fig. 4C-E). Stimulation with the hydrophilic bile acids (CA, UDCA), that lead to activation and phosphorylation of RIPK3-FLAG, further enhanced the phosphorylation of JNK (Fig.   4E). Similar to the reducing effects on RIPK3-FLAG expression and phosphorylation, TCDCA strongly suppressed JNK phosphorylation even in the presence of RIPK3-FLAG (Fig. 4E). These results demonstrate that RIPK3 expression and phosphorylation induces necroptosis, which in turn induces IL-8 secretion regulated by JNK. This mechanism may then lead to both: an activation of repair mechanisms or progression of the liver disease.

Discussion
Our findings indicate that RIPK3 silencing is a hallmark mechanism within hepatocytes to avoid necroptosis, initiation of pro-inflammatory signaling, and pore formation. [85] We have identified bile acids as a class of molecules able to trigger the activation and phosphorylation of RIPK3 that in turn results in an RIPK3dependent IL-8 response in hepatocytes, which is known to trigger local tissue remodeling and infiltration. [86] Increased IL-8 secretion and gene expression is ascribed to the activation of JNK, a pathway involved in several physiological and pathological processes. [87] Thus, the RIPK3-mediated activation of the JNK pathway in hepatocytes may not only lead to an increased local IL-8 signaling but also trigger subsequent pro-inflammatory adaptations such as cell death, cell survival and proliferation in a cell type-specific manner. [88] The JNK pathway is one of the three major groups of mitogen-activated protein kinases (MAPK) which plays a significant role in acute as well as chronic liver injuries by regulating the metabolism and cell death pathway in the liver. [89] Previous studies show that bile acids are able to activate the JNK pathway which results in an inhibition of the bile acid synthesis. [90,91] Besides this, activation of JNK in general is also known to contribute to the expression of pro-inflammatory cytokines (IL-8, IL-6, IL-17). [84,92,93] Taken together, the results of this manuscript describe a hepatocytic pRIPK3-pMLKL-pJNK-IL-8 axis whereby the effect of other MAPK (p38, extracellular-signalregulated kinase) need to be evaluated further.
In conclusion, the observed RIPK3 promoter methylation which suppresses RIPK3 expression in hepatocytes prevents metabolite (e.g. bile acid) induced hepatocellular injury under physiological conditions and may be viewed as a key-protective mechanism allowing hepatocytes to synthesize and transform metabolites that otherwise would constantly trigger inflammatory signaling in hepatocytes under physiological conditions. Under pathophysiological conditions RIPK3 expression however was rapidly induced in mice and in all liver-related pathologies the expression of RIPK3 was found in hepatocytes and LSECs. Hepatocytes seemingly employ epigenetic RIPK3 silencing as an endogenous master switch protecting them from endobiotic metabolites, e.g. bile acids. An increase of conjugated bile acids, due to the great capacity for biotransformation and storage function of hepatocytes, is often observed first in the pathophysiology of diseases. [94][95][96] High concentrations of especially CA and TCA, causes RIPK3 hyperphosphorylation which is crucial for the consecutive activation of necroptosis, and inflammatory JNK-signaling. [97,98] TCA and TUDCA are able to facilitate cell survival and act as anti-cholestatic metabolites. [99] Our results further demonstrate that the same choleretic bile acids (TCA, TUDCA) are able to induce necroptosis in hepatocytes in various liver diseases associated with a hepatocellular RIPK3 expression. This demonstrates a novel pathophysiological mechanism in the progression of liver diseases.
The hepatocellular accumulation of TCA, and in some cases also CA, during various liver injuries supports the notion that hepatocytes employ necroptosis signaling to induce tissue remodeling and inflammation. Hyperphosphorylation of RIPK3 had been recognized previously as a key event in the activation cascade of RIPK3 and necroptosis. [79] As mentioned before, Ser199 (kinase activity) and Ser227 (interaction with MLKL and induction of necroptosis) contains specific functions during the activation of RIPK3. Activation of RIPK1 induces the activation of RIPK3.
The following hyperphosphorylation of RIPK3 induces the interaction with MLKL via the RIP homotypic interaction motif. The full mechanism of RIPK3 hyperphosphorylation has yet to be elucidated, but it seems a likely possibility that activation of RIPK3 induces a phosphorylation at Ser199 which in turn autophosphorylates Ser227 for the formation of a stable complex with MLKL. [75,100] Hydrophobic TCDCA on the other hand, also frequently increased during cholestasis, reduces the expression and phosphorylation of RIPK3. This suggests that other unconjugated bile acids besides CA exert their cell toxicity primarily by the regulation of apoptosis and due to their function as detergent in high concentrations may lead to direct tissue necrosis. [36,[101][102][103] The accumulation of bile acids may not represent the primary cause of liver injury but likely promote disease progression and chronification due to a chronic inflammatory response. [70] Further, this study' findings support previous controversial results obtained from RIPK3 knockout mice after various types of experimental liver injury. Knockout mice were especially protected from injuries that commonly result in chronic liver diseases (e.g. obstructive cholestasis, ethanol induced liver injury), inflammatory liver diseases (e.g. concanavalin-A hepatitis, fecal induced sepsis) and ischemia reperfusion damage [18,20,24,104], but not in the situation of acute toxic damage as caused by e.g. acetaminophen, that may lead to an accumulation of unconjugated bile acids in blood [25,26].
As RIPK3 knockout mice are protected from liver damage during different types of chronic injury, we postulate that the epigenetic profile, which is regulated in a highly dynamic manner during injury and liver regeneration, may be remodeled rapidly during cell stress and will modify hepatocyte susceptibility to endogenous metabolites, inflammatory signaling and cell death.
In summary, the RIPK3 expression is highly dynamic and is physiologically silenced in hepatocytes allowing them to carry out biotransformation of endogenous and exogenous substances even in high concentrations that may otherwise be toxic to the cells without triggering chronic inflammation. On the other hand, our results indicate that RIPK3 expression is rapidly increased during cholestatic liver injuries and the accumulation of bile acids, particularly hydrophilic ones (unconjugated and conjugated CA and UDCA), triggers the activation of inflammation through the necroptosis pathway. Especially in chronic situations where tissue regeneration and resolution of the injury is not achieved, the chronic activation of RIPK3 is another player contributing to the persistence of consistent liver injury.

Limitation of the study
The cell death mechanism referred to as necroptosis attracted a great attention over the time especially due to the controversial discussion about its function in the liver.
It remains to be elucidated whether RIPK3 is expressed in all types of liver diseases or just in case of obstructive cholestasis. In human specimens the investigation of RIPK3 signaling is particularly challenging due to the lack of healthy reference material. We have used bilirubin as a biomarker to depict differences, however we are aware that bilirubin rises late in the cause of liver disease and a prospective evaluation with a differential analysis of hepatocellular vs. plasma bile acids concentration may shine further light on the regulation of necroptosis signaling in these cells.