RIPK1 in Liver Parenchymal Cells Limits Murine Hepatitis during Acute CCl4-Induced Liver Injury

Some life-threatening acute hepatitis originates from drug-induced liver injury (DILI). Carbon tetrachloride (CCl4)-induced acute liver injury in mice is the widely used model of choice to study acute DILI, which pathogenesis involves a complex interplay of oxidative stress, necrosis, and apoptosis. Since the receptor interacting protein kinase-1 (RIPK1) is able to direct cell fate towards survival or death, it may potentially affect the pathological process of xenobiotic-induced liver damage. Two different mouse lines, either deficient for Ripk1 specifically in liver parenchymal cells (Ripk1LPC-KO) or for the kinase activity of RIPK1 (Ripk1K45A, kinase dead), plus their respective wild-type littermates (Ripk1fl/fl, Ripk1wt/wt), were exposed to single toxic doses of CCl4. This exposure led in similar injury in Ripk1K45A mice and their littermate controls. However, Ripk1LPC-KO mice developed more severe symptoms with massive hepatocyte apoptosis as compared to their littermate controls. A pretreatment with a TNF-α receptor decoy exacerbated liver apoptosis in both Ripk1fl/fl and Ripk1LPC-KO mice. Besides, a FasL antagonist promoted hepatocyte apoptosis in Ripk1fl/fl mice but reduced it in Ripk1LPC-KO mice. Thus, the scaffolding properties of RIPK1 protect hepatocytes from apoptosis during CCl4 intoxication. TNF-α and FasL emerged as factors promoting hepatocyte survival. These protective effects appeared to be independent of RIPK1, at least in part, for TNF-α, but dependent on RIPK1 for FasL. These new data complete the deciphering of the molecular mechanisms involved in DILI in the context of research on their prevention or cure.


Introduction
Over the past 20 years, there has been a steady increase in research on drug-induced liver injury (DILI) [1]. This stems not only from its high prevalence observed in the diagnosis of acute liver failure [2], but also to risks of drug withdrawal from the pharmaceutical market. Two types of DILI are distinguished. Thus, the intrinsic DILI is directly related to the physicochemical properties of drugs, is dose-dependent, and largely predictable. Conversely, in rare cases, idiosyncratic DILI (IDILI) occurs unpredictably because it is specific to individuals, their environment, and possible comorbidities [3]. After the specific acetaminophen-poisoning animal model, the most popular in vivo model to study intrinsic DILI is the carbon tetrachloride (CCl 4 )-induced acute hepatitis [4]. CCl 4 is an organic compound which was previously widely used as a precursor to refrigerants, cleaning agent, degreaser, and as a constituent in fire extinguishers until its hepatotoxic and carcinogenic Int. J. Mol. Sci. 2022, 23, 7367 2 of 15 potential was elucidated. Some clinical cases of intentional or accidental CCl 4 intoxication have been reported as early as the beginning of the last century with examples still occurring until recently [5,6]. Xenobiotic metabolism enzymes, in particular cytochromes P450, transform CCl 4 into the reactive metabolite trichloromethyl radical (CCl 3 •). This generated free radical alkylates nucleic acids, proteins, and lipids but reacts also with oxygen molecules to form trichloromethylperoxy radical (CCl 3 OO•). This highly reactive species then promotes lipid peroxidation. This combination of events results in hepatotoxicity [7,8].
CCl 4 at high doses induce hepatocyte death, which can be necrotic or apoptotic depending upon activation of different signalling pathways [9]. In hepatocyte, reactive metabolites trigger endoplasmic reticulum or mitochondrial stress. In case of overwhelmed organelle-specific adaptive stress responses, hepatocyte death occurs either by mitochondrial permeability transition (MPT) leading to necrosis or by regulated cell death mechanisms leading to mitochondrial outer membrane permeabilization (MOMP), which initiates the intrinsic apoptotic pathway [10,11]. In case of extensive liver damage, it may be difficult to differentiate the type of cell death by histology [12]. Cell swelling and plasma membrane rupturing (oncosis), which promotes an inflammatory response, is defined as necrosis, while all combined, caspase activation, chromatin fragmentation, and phagocytosis of apoptotic cell bodies that reduces inflammation, define apoptosis [13,14]. Cell death then promotes a pro-inflammatory environment where the tumor necrosis factor (TNF-α) plays a key role. The intracellular signalling pathways, prompted by the interaction of TNF-α with its cognate receptors (TNFR1 and TNFR2), are governed by the receptor-interacting protein kinase-1 (RIPK1) [15]. Thus, the scaffolding properties of RIPK1 direct signal transduction to an NF-κB dependent pro-survival pathway, while its kinase activity directs the cell to death by apoptosis or by necroptosis [16]. Thus, some works demonstrated the protective role of RIPK1 in hepatocytes during different acute hepatitis involving TNF-α [17][18][19]. Regarding CCl 4 -induced hepatotoxicity, some studies focused on the implication of TNFα [20,21], none investigate RIPK1. With two different mouse lines, either deficient for Ripk1 specifically in liver parenchymal cells (Ripk1 LPC-KO ), or for the kinase activity of RIPK1 (Ripk1 K45A ) plus their respective WT littermates (Ripk1 fl/fl and Ripk1 wt/wt ), we aimed to deepen the role of RIPK1 during CCl 4 -induced hepatotoxicity.

Scaffolding Properties of RIPK1 in Liver Parenchymal Cell Limit Apoptosis Occurrence in Acute CCl 4 -Induced Hepatitis
To investigate the role of RIPK1 in CCl 4 -induced hepatotoxicity, Ripk1 LPC-KO mice along with their WT littermates (Ripk1 fl/fl ) were first exposed to a toxic dose of CCl 4 by gavage. Liver damage was assessed by histopathological tissue staining and by serum transaminase measurement at 24, 48, and 72 h post-treatment ( Figure 1A). In both genotypes, CCl 4 treatment induced classical centrilobular liver damage as shown by H&E staining. The surface of necrotic areas appeared significantly larger in the livers of Ripk1 LPC-KO mice 24 h after the gavage. Besides, serum transaminase levels (ALT) significantly increased at 24 and 48 h and started to drop at 72 h. Even if the kinetics appeared similar between genotypes, the levels of serum transaminases reached significantly higher values in Ripk1 LPC-KO mice in samples taken at 24 h. On the other hand, samples taken 72 h after CCl 4 administration revealed lower levels of necrotic areas and of serum transaminases in Ripk1 LPC-KO mice than in Ripk1 fl/fl littermates. This is most probably due to the earlier hepatocyte death in challenged Ripk1 LPC-KO mice. Interestingly, labelling of cleaved caspase-3 (CC3) showed that much more apoptosis occurred 24 h post-treatment in the livers of Ripk1 LPC-KO mice than in those of Ripk1 fl/fl littermates ( Figure 1B).  For all graphs, each dot represents an individual and errors bars are expressed as means ± SEM (* p < 0.05; ** p < 0.01; *** p < 0.001 and **** p < 0.0001).
Next, mice deficient for the kinase activity of RIPK1 (Ripk1 K45A ) along with their WT littermates (Ripk1 wt/wt ) were subjected to the same CCl 4 treatment. Similar liver injury occurred at 24 h post-treatment in both genotypes as shown by the histological analysis of livers and by the determination of serum transaminase levels ( Figure 2). CC3 staining also showed no modification in apoptosis occurrence. Therefore, only the scaffolding properties of RIPK1 in liver parenchymal cells interfere with CCl 4 -hepatoxicity by avoiding the induction of extensive apoptosis. or containing CCl4 (dose of 2.4 g/kg body weight). Sample collections were conducted 24, 48 or 72 h post-gavage. (A) Representative pictures of liver tissue sections stained by H&E with quantification of necrotic areas. Levels of serum alanine aminotransferase (ALT). (B) Representative pictures of liver tissue sections stained by cleaved caspase 3 (CC3) antibody. Signal quantification of CC3 staining. For all graphs, each dot represents an individual and errors bars are expressed as means ± SEM (* p < 0.05; ** p < 0.01; *** p < 0.001 and **** p < 0.0001).
Next, mice deficient for the kinase activity of RIPK1 (Ripk1 K45A ) along with their WT littermates (Ripk1 wt/wt ) were subjected to the same CCl4 treatment. Similar liver injury occurred at 24 h post-treatment in both genotypes as shown by the histological analysis of livers and by the determination of serum transaminase levels ( Figure 2). CC3 staining also showed no modification in apoptosis occurrence. Therefore, only the scaffolding properties of RIPK1 in liver parenchymal cells interfere with CCl4-hepatoxicity by avoiding the induction of extensive apoptosis.

Ripk1 Deficiency in Liver Parenchymal Cells Potentiates Oxidative Stress, Inflammation and Immune Infiltration in CCl 4 -Induced Hepatitis
CCl 4 metabolism triggers the production of reactive oxygen species (ROS), which contribute to cell toxicity and ultimately result in the release of damage-associated molecular patterns (DAMPs) into the extracellular environment, a key aspect of inflammation that further activate Kupffer cells (KCs), monocyte-derived macrophages (MFs) and neutrophils. These activated immune cells will then release pro-inflammatory cytokines in liver that will exacerbate cell damage. To investigate these different features of the pathological process, the analysis of immune infiltration and the regulation of mRNA levels of several genes was compared in the livers of Ripk1 fl/fl and Ripk1 LPC-KO mice, untreated for 24 h after CCl 4 gavage. Thus, the analysis of immune infiltration by CD45 staining revealed much more infiltrated leukocytes in the hepatic tissue of CCl 4 -treated Ripk1 LPC-KO mice ( Figure 3). The specific analysis of the macrophage subpopulation (CD68+ cells) revealed a similar but less marked trend ( Figure S1). Accordingly, all tested mRNA levels of genes related to oxidative stress (Hmox-1, Nfe2l2 [Nrf-2], Cybb [Nox-2], and Nqo1) were more induced by CCl 4 treatment in Ripk1 LPC-KO mice ( Figure 4A). Conversely, mRNA level inductions of evaluated genes related to the NF-κB pathway (Rela [NFkBp65], Cflar [cFLIP], Tnfaip3 [A20], Ikk-γ [NEMO], Saa1, Cxcl1, and Ccl20) were significantly less pronounced in Ripk1 LPC-KO mouse samples (except for Nfkbia [Iκbα]) ( Figure 4B). However, a similar analysis on mRNA of genes involved in inflammation (Ifnγ, Il-6, Il-1β and Ccl2) systematically showed an exacerbated induction in the liver of Ripk1 LPC-KO mice after CCl 4 gavage ( Figure 4C). The same expression regulation of these cytokines was found at the protein level during their serum quantification ( Figure S2). These results supported the notion that the scaffolding properties of RIPK1 would protect hepatocytes from apoptosis during CCl 4 intoxication by activating the NF-κB pathway. The higher rate of hepatocyte mortality observed in Ripk1 LPC-KO mice is thus thought to be the cause of exacerbated oxidative stress, inflammation, and immune infiltration. In parallel, mRNA levels of death factors and their cognate receptors were also investigated in livers. Thus, the mRNA amounts for Tnfrsf1a [TNFR-1] and Tnfrsf1b      . For all graphs, each dot represents an individual and errors bars are expressed as means ± SEM (ns, non-significant; * p < 0.05; ** p < 0.01; *** p < 0.001, and **** p < 0.0001).

Discussion
Downstream of different death ligand receptors, RIPK1 has a pro-survival function through its scaffolding properties or it triggers cell death through its kinase activity [22]. Thus, RIPK1 can induce caspase-dependent apoptosis, or necroptosis by coordinating with RIPK3 and the mixed lineage kinase domain like pseudokinase (MLKL) [23]. We re-

Discussion
Downstream of different death ligand receptors, RIPK1 has a pro-survival function through its scaffolding properties or it triggers cell death through its kinase activity [22]. Thus, RIPK1 can induce caspase-dependent apoptosis, or necroptosis by coordinating with RIPK3 and the mixed lineage kinase domain like pseudokinase (MLKL) [23]. We recently demonstrated the protective role of RIPK1 in NASH induced fibrosis [24]. In parallel, RIPK1 interaction with RIPK3 has been described to increase with repeated injections of CCl 4 , suggesting the possible involvement of RIPK1 in chronic CCl 4 -induced hepatotoxicity [25]. In contrast to the chronic model of CCl 4 -induced hepatitis obtained by repeated expositions which is more focused on liver fibrosis pathophysiology [26], our investigations covered the acute CCl 4 liver toxicity after single dose administration. Thus, we demonstrated that RIPK1 depletion specifically in liver parenchymal cells sensitizes mice (Ripk1 LPC-KO ) to CCl 4 -induced hepatocyte apoptosis, while genetic inactivation of its kinase activity (Ripk1 K45A ) never affected CCl 4 -induced acute hepatitis. Further, pharmacological inhibition of RIPK1 by Nec1s also had no impact on CCl 4 -induced hepatitis in mice ( Figure S4). Therefore, the involvement of RIPK1 in CCl 4 -induced acute hepatitis engages its platform function but not its kinase activity. Thus, as in the specific case of acetaminophen intoxication [27], the kinase activity of RIPK1 does not represent a relevant therapeutic target.
CCl 4 -induced hepatotoxicity has been associated with increased expression of hepatic transcript and of plasma concentrations of TNF-α [20,28]. Besides, RIPK1 is known to play a protective role in different type of acute hepatitis involving TNF-α [17][18][19]. Although the TNF signalling pathway has been depicted as not critical for chronic exposure to CCl 4 [21], the cytokine is believed to contribute to the damage occurring in acute CCl 4induced hepatitis. Indeed, its neutralization or depletion of its receptor (TNFR1) has been described as leading to an improvement of clinical signs in animal models [28]. However, our experiments using a TNF-α decoy receptor (ETA) did not reproduce this protection. On the contrary, apoptosis was over-induced in the livers of Ripk1 fl/fl control mice as well as in those of Ripk1 LPC-KO mice in presence of ETA. This protective effect of TNF-α appeared, therefore, at least in part, independent of RIPK1 present in liver parenchymal cells, without excluding its potential involvement in other cell types present in the hepatic tissue. This limitation of the deleterious effects of CCl 4 by TNF-α has been also reported in acute intoxication in rats [20]. However, its role emerged as more complex [20]. Indeed, if low ETA concentrations limited liver injury caused by CCl 4 in rats, higher concentrations lost the protective effect and can even aggravate symptoms. Altogether, these different data probably reflect the coexistence of the protective and harmful properties of TNF-α in acute hepatitis induced by CCl 4 , a balance which could be influence depending on the experimental conditions.
The increased susceptibility of Ripk1 LPC-KO mice to acute hepatitis induced by CCl 4 , which was amplified in presence of ETA, was systematically associated with proportional mRNA upregulation of FasL/Fas couple in the liver. The ligand, FasL, exists in transmembrane or soluble form. Although both bind efficiently to the cognate receptor (Fas), interaction with the soluble form triggers rapid internalization of the complex, leading to downregulation of the Fas receptor on the cell surface [29,30]. Therefore, the soluble FasL compete with its membrane-bound counterpart for apoptosis induction. Thus, the conversion of the cytotoxic membrane-bound FasL to its protective soluble form has been proposed to represent an important regulatory mechanism by which susceptible tissues, such as the liver, protect themselves from excessive immune responses associated with infiltrating cytotoxic T cells harboring transmembrane FasL [30]. During our investigations, we found that an antagonist of FasL (anti-FasL antibody) interfered with the severity of CCl 4 -induced hepatitis. In control mice with the Ripk1 fl/fl genotype, the apoptotic death rate of hepatocytes increased, most probably reflecting the presence of a protective activity of FasL. Conversely, the same experiment conducted on Ripk1 LPC-KO mice revealed a predominant activity of FasL rather favorable to apoptosis induction. These results were consistent with previously published data showing a protective role of RIPK1 downstream of Fas in hepatocytes [31]. During our experimentations, it appeared that mRNA levels of TRAIL and TRAIL-R2 were upregulated during acute CCl 4 poisoning, highlighting the interest of further investigating this TRAIL/TRAILR axis. Overall, this study provides a deeper understanding of mechanisms occurring during acute liver hepatitis induced by CCl 4 .

Animals, Treatment Protocols
Both mice lines, C57BL/6J Ripk1 LPC-KO and C57BL/6N Ripk1 K45A mice along with their respective WT littermates (Ripk1 fl/fl and Ripk1 wt/wt ) were maintained in our animal facility. Homogeneous groups of female and male mice aged 9 to 12 weeks were used for each experiment. CCl 4 diluted in olive oil was administered orally at a dose of 2.4 g/kg body weight (10 µL/g body weight). Olive oil alone, administered in age and sex matched mice, was used as control. Etanercept (ETA, Pfizer, New-York, NY, USA) or anti-FasL antibody (Ultra-LEAF TM anti-CD178 clone MFL3, Biolegend, Amsterdam, The Netherlands) were eventually injected via intraperitoneal route, both at a dose of 10 mg/kg body weight, 1 h prior CCl 4 challenge and 12 h later only for ETA. Mice were euthanized at different indicated time points. A schematic diagram representing the experimental design is shown in Supplementary Data (Figure S3).

Biochemical Studies
Plasma alanine (ALT) and aspartate (AST) transaminases were measured according to the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) primary reference procedures using Olympus AU2700 Chemistry analyser ® (Olympus Optical, Tokyo, Japan). Only the ALT data are depicted since those of the AST systematically followed the same variations.

Histopathological Studies and Immunolocalization in Liver Tissues
Mouse liver samples were fixed in 4% paraformaldehyde and embedded in paraffin for immunohistochemistry. For histopathology, hematoxylin and eosin (H&E) staining of liver tissues was carried out to investigate the liver injury. For immunolocalization, paraffin-embedded mouse liver sections (5 µm) were dried for 1 h at 58 • C, followed by antigen retrieval and incubation with primary antibody (anti-cleaved caspase-3, Cell Signaling Technology, #9661, Danvers, MA, USA, or anti-CD45 antibody, BioLegend, #103107, Amsterdam, Netherlands) in a Ventana automated staining platform (Ventana Medical Systems, Illkirch-Graffenstaden, France). Revelation of primary antibody was carried out using horseradish peroxidase (HRP)-conjugated secondary antibody (Dako, Agilent Technologies, Les Ulis, France) and DAB substrate kit (Ventana Medical Systems, #760-124, Illkirch-Graffenstaden, France). Slides were then counterstained with hematoxylin. All paraffin-embedded liver sections were scanned with a digital slide scanner (Nanozoomer 2.0-RS, Hamamatsu Photonics, Massy, France) and files were analyzed with the NDP viewer 2.5 software (Hamamatsu, Hamamatsu City, Japan). Signal quantifications were performed with an image analysis software (NIS-Element AR analysis software, Nikon, Tokyo, Japan).

RNA Isolation and qPCR Analysis
Total RNA was extracted from frozen liver fragments using the Nucleospin RNA kit (Macherey-Nagel, #740955, Hoerdt, France) and an Ultra-Turrax ® homogenizer according to manufacturer's instructions. After RNA quantification using the NanoDrop (ND-1000 Spectrophotometer, Thermo Fisher Scientific, Illkirch, France), cDNA was synthetized using the SuperScriptTM II Reverse Transcriptase (Thermo Fisher Scientific, #18064022, Illkirch, France). Real-time quantitative PCR was performed using the fluorescent SYBR Green dye (Power SYBR ® Green PCR Master Mix, Thermo Fisher Scientific, Illkirch, France) and the CFX384 Touch TM Real-Time PCR Detection System (Bio-Rad, Marnes-La-Coquette, France). Each measurement was performed in triplicate. The relative gene expression was normalized against the 18S gene expression. Healthy control mice served as a reference for mRNA expression (control mRNA level was arbitrarily set at 1). All primer sequences are depicted in Table S1 in the supporting information.

Statistical Analysis
Data were expressed as means ± SEM for all mice treated similarly. Mean differences between experimental groups were assessed using the non-parametric Mann-Whitney Utest. All statistical analysis was achieved with the GraphPad Prism5 software. Significance is shown as follows: * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001.