Necroptosis is a key pathogenic event in human and experimental murine models of non-alcoholic steatohepatitis

INTRODUCTION

Necrosis is considered an accidental process characterized by cell swelling and early loss of plasma membrane integrity, with consequent leakage of pro-inflammatory mediators. The unregulated nature of necrosis precludes the design of thera- peutic interventions for protection of necrosis-associated cell injury and inflammation. Interestingly, regulated necrosis or necroptosis was recently described as a novel cell death path- way, morphologically similar to that of necrosis, but occurring in a programmed manner, through well-defined biochemical pathways [1].

Necroptosis may occur in response to multiple stimuli, tumour necrosis factor-α (TNF-α) being the most well-studied initiation signal. Typically, TNF-α-dependent necroptosis is initiated by interaction between receptor-interacting proteins 1 (RIP1) and 3 (RIP3), which originate in the necrosome, an oligomeric amyl- oid signalling complex. At the functional level, the auto- and trans-phosphorylation of RIP1 and RIP3 are required for necro- some assembly and activation of necroptotic signalling [2]. In that regard, necrostatin-1 (Nec-1), a tryptophan-based molecule that allosterically inhibits RIP1 kinase activity, blocks necroptosis [3]. However, in specific cell contexts, increased RIP3 levels [4– 6] or overexpression of a RIP3 phospo-mimetic mutant [7] can trigger necroptosis in the absence of RIP1, suggesting that RIP3 is a unique regulator of necroptotic cell fate. RIP3 recruits and phosphorylates the pseudokinase mixed lineage kinase domain- like (MLKL), which in turn oligomerizes and causes irreversible cellular membrane damage, resulting in necrotic cell death [8]. In addition, overproduction of reactive oxygen species (ROS) has been described as a possible contributing factor in some cellular contexts [3,4,9].

The physiological relevance of necroptosis has been demon- strated in a variety of paradigms. In particular, necroptosis is involved in the pathogenesis of several inflammatory disorders, including pancreatitis [10] and chronic inflammation of gut [11,12] and skin [13]. Likewise, necroptosis is arising as a likely pathological feature of inflammation-driven liver diseases. For instance, RIP3 mediates ethanol or drug-induced liver injury in vivo [8,14,15], whereas apoptosis and RIP3-dependent necrop- tosis are simultaneously activated in an animal model of chronic hepatic inflammation [16]. In addition, concanavalin A-induced hepatic failure is associated with hepatocyte necroptosis and ab- errant TNF-α signalling [17,18]. Curiously, as a potent proin- flammatory cytokine, TNF-α is involved in the pathogenesis of a broad range of liver diseases, including viral hepatitis, alco- holic hepatitis, acute liver failure and non-alcoholic fatty liver disease (NAFLD) [19]. NAFLD is the most common chronic liver disease and its prevalence and incidence is increasing due to strong association with obesity and the metabolic syndrome. NAFLD encompasses a spectrum of liver dysfunction ranging from simple fatty liver or hepatic steatosis to hepatic necrosis and inflammation, characteristic of non-alcoholic steatohepatitis (NASH). Further, NAFLD may potentially progress to severe long-term consequences, including cirrhosis, hepatocellular car- cinoma and, ultimately, premature death. Whereas NAFLD trig- gering is associated with lipid deposition within hepatocytes, the subsequent mechanisms of liver damage are multifactorial and relate to metabolic changes, oxidative stress, miRNA expression and cytokine production [20]. Their exact contribution and in- terplay during NAFLD progression remains far from completely understood.

We hypothesize that necroptosis represents a major pathway mediating the pathogenesis of inflammation-driven liver diseases. This prompted us to determine the involvement of necroptosis in NAFLD pathogenesis in humans and in experimental murine models of hepatic steatosis and NASH. Further, we aimed at establishing the involvement of TNF-α and oxidative stress in necroptotic signalling in hepatocytes.

METHODS

Patients, histological findings and grading

NAFLD liver specimens were obtained from morbidly obese pa- tients undergoing bariatric surgery, as previously described [21]. Moreover, liver tissue was also prospectively and sequentially collected from patients that fulfilled the clinical and pathological diagnostic features of alcoholic steatohepatitis (ASH; n=5), hep- atitis B (n=5) and hepatitis C (n=4). Control liver specimens were obtained from five individuals who were potential liver donors, with normal liver histology and biochemistry. No statist- ical differences were observed in age and gender between patient groups. All liver specimens were processed conventionally for diagnostic purposes and were blindly evaluated by an experi- enced pathologist.

Informed written consent was obtained from all patients and the study protocol conformed to the Ethical Guidelines of the 1975 Declaration of Helsinki, as reflected in a priori approval by the Hospital of Santa Maria (Lisbon, Portugal) Human Ethics Committee. Paraffin-embedded liver tissue sections from ASH and NAFLD patients were stained with Haematoxylin and Eosin (H&E). The Gordon and Sweet’s Silver Staining method was used for identification of reticular fibres, Chromotrope-Aniline Blue (CAB) for connective tissue and Perl’s Prussian Blue for iron. Steatosis was graded from 0 to 3 based on the percentage of steatotic hepatocytes (0, none; 1, <33 %; 2, 33 %–66 %; 3, >66 %). In addition, portal and lobular inflammation and portal and lobular fibrosis were semi-quantitatively graded on a scale of 0–4 (0, absence; 1, mild; 2, moderate; 3, severe degree and 4, cirrhosis), as previously reported [22]. Eleven patients were classified as having NASH and five as having simple steatosis. In control individuals (n=5), all parameters were graded 0.

Serum analyses

Serum samples from a cohort of morbidly obese patients with clinical and biopsy-proven diagnosis of steatosis (n=29) and NASH (n=15) were used for biochemical assays. Serum of liver disease-free individuals was also analysed (n=5). Routine laboratory assays included total cholesterol, high-density lipoprotein (HDL)-cholesterol, low-density lipoprotein (LDL)- cholesterol, triacylglycerol, aspartate aminotransferase (AST), alanine aminotransferase (ALT), γ -glutamyl transpeptidase (γ – GT), total bilirubin, fasting serum glucose and fasting insulin, us- ing the standard techniques of clinical chemistry laboratories. In addition, sandwich ELISA was used to determine cytokeratin 18 full-length (CK18-M65; 10020, Peviva), high-mobility group box 1 (HMGB1; ST51011, IBL International GmbH), cyclophilin A (CypA; E01C0613, BlueGene Biotech) and TNF-α (HSTA00D, R&D System Inc.) serum levels. In addition, phospho-MLKL (p-MLKL), total MLKL and RIP3 levels were analysed in the liver specimens of these patients by immunoblotting.

Animals and diets

Male C57BL/6 6-week-old mice (Harlan Laboratories) were fed either standard chow diet (control; n=5; Mucedola) or high-fat (35 % total fat, 54 % trans-fatty acid enriched) choline-deficient diet (HFCD; n=4–5; Harlan Laboratories) for 6 and 18 weeks. Alternatively, 5-month-old C57BL/6 wild-type (WT) or RIP3- deficient (RIP3—/—) mice were fed either a chow diet or a me- thionine and choline-deficient (MCD; n=5) diet (TestDiet) for 2 and 8 weeks. At the indicated time-points, animals were killed by exsanguination under isoflurane anaesthesia. The liver was removed; one lobe was collected, rinsed in normal saline and immediately flash-frozen in liquid nitrogen for protein and RNA extraction; another lobe was included in optimal cutting tem- perature compound (4583, Tissue Tek) for histochemistry of fat by Oil Red O staining (O-0625, Sigma–Aldrich Co.). Paraffin- embedded sections (3–4 μm) were stained with H&E, CAB or Masson’s Trichrome. Liver sections were scored in a blinded fash- ion by experienced pathologists, using a four-point severity scale (0, normal; 1, mild; 2, moderate; 3, severe), for steatosis, inflam- matory cell infiltration and fibrosis. In the MCD model, serum was also collected and ALT and AST determined using standard clinical chemistry techniques. All procedures were reviewed and approved by the local ethics committee and national competent authorities for animal protection. Animals received humane care in a temperature-controlled environment with a 12-h light–dark cycle, complying with the Institute’s guidelines and as outlined in the ‘Guide for the Care and Use of Laboratory Animals’ pre- pared by the National Academy of Sciences and published by the National Institutes of Health (NIH publication 86-23 revised 1985).

Quantitative RT-PCR

RNA was extracted from animal liver samples using the TRIzol reagent according to the manufacturer’s instructions (Life Tech-and HPRT were calculated based on the standard curve. TNF-α, IL-1β, collagen-1-α1 and TGFβ mRNA levels were normal- ized to the level of HPRT and expressed as fold change from controls.

Cell culture and treatments

Primary rat hepatocytes were isolated from male rats (100– 150 g) by collagenase perfusion, as previously described [23,24]. Primary mouse hepatocytes were isolated from male WT and RIP3—/— mice using liver perfusion and liver digest medium (Gibco, Life Technologies Corp.) according to the manufacturer’s protocols with some modifications. Briefly, mice were killed with isofluorane overdose, inferior vena cava cannulated and the liver perfused in situ with liver perfusion medium (37 ◦C), followed by perfusion with liver digest medium (pH 7.4, 37 ◦C). The liver was removed and then gently minced in 20 ml of liver digest medium. The liver cell suspension was homogenized with 20 ml of Williams-E Glutamax Complete Medium (Gibco) containing 10 % FBS (Gibco) and then filtered through a 70 μm cell strainer. Cell suspension was centrifuged at 50 g for 5 min. The pellet was resuspended in 10 ml of medium and added to 10 ml of buf- fered Percoll, containing 9 ml of Percoll (Sigma–Aldrich Co.) plus 1 ml 10× PBS (Life Technologies Corp.). Cell suspension was centrifuged at 50 g for 5 min and the pellet resuspended in Williams-E Glutamax Complete Medium containing penicillin (100 units/ml)/streptomycin (100 μg/ml; Gibco) and 4 % FBS. Cell viability, as determined by Trypan Blue exclusion, was gen- erally >85 %. After isolation, primary murine hepatocytes were plated on PrimariaTM tissue culture dishes (BD Biosciences) at 5 × 104 cells/cm2. Cells were maintained at 37 ◦C in a humidified atmosphere of 5 % CO2.

Primary murine hepatocytes were pre-treated with 50 μM of a pan-caspase inhibitor, zVAD-fmk (Enzo Life Sciences Inc.) and/or 100 μM Nec-1 (Sigma–Aldrich Co.) or DMSO (Sigma– Aldrich Co.) vehicle control (final concentration of 0.1 %). After 1 h, cells were exposed to 400 μM palmitic acid (PA; Sigma– Aldrich Co.) or TNF-α (10 ng/ml; PeproTech EC Ltd.) plus cyc- loheximide (CHX; 0.5 μg/ml; Sigma–Aldrich Co.) or vehicle control. PA was prepared in isopropyl alcohol at a stock concen- tration of 80 mM. PA was added to Complete William’s E medium containing 1 % BSA to ensure a physiologic ratio between bound and unbound PA in the medium, approximating the molar ratio present in the plasma [25]. The concentration of PA used in the experiments was less than the fasting total non-esterified fatty acid plasma concentrations observed in human NASH [26,27]. The concentration of isopropyl alcohol was 0.5 % in final incuba- tions; this concentration was used as vehicle control in PA exper- iments. After 24 h, hepatocytes were processed for cell death as- says, caspase-3/7 activity measurements and immunocytochem- istry analyses.

For functional analyses, primary rat hepatocytes were trans- fected at the moment of plating with 100 pM of a siRNA nt against ripk3 (siRIPK3; s140868, Ambion, Life Technologies Corp.) or with a siRNA control, using LipofectamineTM2000 (Invitrogen, Life Technologies Corp.), according to the manu- facturer’s instructions. Primary rat hepatocytes were harvested at 48 h post-transfection for protein extraction, to confirm RIP3 silencing. Alternatively, primary rat and mouse hepatocytes were pre-treated for 1 h with zVAD-fmk and/or Nec-1 before incuba- tion with TNF/CHX. After 3 and 24 h, cells were harvested for total ROS levels detection and cell-death assays, respectively.

Total and soluble/insoluble protein extraction

For isolation of total protein extracts, liver pieces and primary rat hepatocytes were homogenized using a glass dounce ho- mogenizer in ice-cold lysis buffer (10 mM Tris/HCl, pH 7.6, 5 mM MgCl2, 1.5 mM potassium acetate, 1 % Nonidet P-40, 2 mM DTT) and 1× Halt Protease and Phosphatase Inhib- itor Cocktail (Pierce, Thermo Fisher Scientific). The lysate was centrifuged at 3200 g for 10 min at 4 ◦C and the supernatant re- covered and stored at —80 ◦C. For soluble/insoluble protein ex- traction, livers were homogenized using a glass dounce homogen- izer in radio-immunoprecipitation assay (RIPA) buffer (50 mM Tris/HCl, pH 8; 150 mM NaCl; 1 % NP-40; 0.5 % sodium deoxy- cholate; 0.1 % SDS) and 1× Halt Protease and Phosphatase Inhib- itor Cocktail (Pierce). The suspension was centrifuged at 15 000 g for 20 min at 4 ◦C. The resultant supernatant was collected as the soluble protein fraction and was centrifuged a second time for complete removal of all insoluble proteins. Insoluble protein fractions contained in pellets were resuspended in RIPA buffer supplemented with 8 M urea and were sonicated at 30 %–40 % power for 7 s. Protein concentrations were determined using the Bio-Rad protein assay kit (Bio-Rad Laboratories), according to the manufacturer’s specifications.

Immunoblotting

Steady-state levels of RIP3 and MLKL were determined by im- munoblot analysis. Briefly, 60 μg of total protein extracts, in- soluble or soluble protein fractions were separated on an 8 % SDS/PAGE. Following electrophoretic transfer on to nitrocellu- lose membranes and blocking with 5 % milk solution, blots were incubated overnight at 4 ◦C with primary rabbit polyclonal an- tibodies against RIP3 (1:200, Santa Cruz Biotechnology Inc.), mouse MLKL (1:500, SAB1302339, Sigma–Aldrich), human p- MLKL (1:1000, Abcam plc) and human MLKL (1:1000, Abcam plc) and with a secondary antibody conjugated with horseradish peroxidase (Bio-Rad Laboratories) for 3 h at room temperature. Membranes were processed for protein detection using Super Signal substrate (Pierce). β-Actin (1:20000; Sigma–Aldrich) and Ponceau S staining (Merck) were used as loading controls. We have previously validated Ponceau S staining as a loading con- trol for immunoblot analysis [28], which is particularly relevant when regular housekeeping proteins are inadequate controls, like for mouse liver insoluble/soluble protein fractions.

Cell death and caspase activity measurements General cell death was evaluated using the lactate dehyd- rogenase (LDH) Cytotoxicity Detection KitPLUS (Roche Dia- gnostics GmbH), following the manufacturer’s instructions. In each experiment set-up, experimental LDH values were normal- ized with maximum releasable LDH activity in the cells, after cell disruption with the provided lysis solution. Hoechst 33258 (Sigma–Aldrich Co.) labelling of attached cells was used to de- tect apoptotic nuclei by morphological analysis, as previously described [29]. Caspase-3/-7 activity was measured using the Caspase-Glo 3/7 Assay (Promega Corp.), according to the man- ufacturer’s protocol.

Analysis of total ROS levels

ROS levels were analysed through the use of 2r,7r- dichlorodihydrofluorescein diacetate (H2DCFDA; Sigma– Aldrich Co.), a cell-permeant non-fluorescent molecule that is oxidized by ROS to form dichlorofluorescein, a fluorescent com- pound. Primary murine hepatocytes were incubated with 10 μM H2DCFDA at 37 ◦C for 30 min. Cells were then washed with PBS (Life Technologies Corp.). Twenty-five milligrams of liver tissue was homogenized using a glass dounce in 500 μl of ice-cold PBS. To remove insoluble particles, the lysate was centrifuged at 10 200 g for 5 min at 4 ◦C and the supernatant was recovered. Fifty microlitres of the lysate were incubated with 10 μM H2DCFDA at room temperature for 30 min. The emission of green fluores- cence was measured using the GloMax-Multi+ Detection System (Promega Corp.). Fluorescence values were corrected with total protein content.

Immunochemistry and image analysis

RIP3 expression was evaluated by immunohistochemistry in human liver tissue. Paraffin-embedded liver sections were de- paraffined, rehydrated and boiled three times in 10 mM citrate buffer, pH 6. Sections were then incubated for 1 h in blocking buffer, containing 0.3 % Triton X-100 (Sigma–Aldrich Co.), 1 % FBS (Life Technologies Corp.) and 10 % normal donkey serum (Jackson ImmunoResearch Laboratories). RIP3 expression and localization in primary rat hepatocytes was evaluated by immun- ocytochemistry. Cells were washed twice, fixed with paraform- aldehyde in PBS (4 %, w/v) and then blocked for 1 h at room temperature in PBS, containing 0.1 % Triton X-100, 1 % FBS and 10 % normal donkey serum. For both staining procedures, samples were then incubated with a primary antibody reactive to RIP3 (1:50; Santa Cruz Biotechnology) overnight at 4 ◦C. After rinsing, the primary antibody was developed by incubating with a 1:200 secondary DyLight 594-conjugated anti-rabbit antibody (Jackson ImmunoResearch Laboratories) for 2 h at room tem- perature. Hepatocyte nuclei were stained with Hoechst 33258 at 50 μg/ml in PBS for 6 min at room temperature. Samples were mounted using Fluoromount-G (Beckman Coulter). De- tection of RIP3 in cells was visualized using an AxioScope.A1 microscope (Carl Zeiss Microscopy GmbH). Images were ac- quired, under 400× magnification, using an AxioCam HRm camera with the AxioVision software (release 4.8; Carl Zeiss Microscopy GmbH). Semi-quantitative analysis of mean fluores- cence intensities of RIP3 was performed using the NIH ImageJ software. Eight images per sample were obtained. Images were converted into an 8-bit format and the background was sub- tracted. An intensity threshold was set and kept constant for all images analysed. RIP3 fluorescence intensity in primary rat hep- atocytes was normalized with the number of cells per microscopic field.

Densitometry and statistical analysis

The relative intensities of protein bands were analysed using the Image Lab densitometric analysis program (version 5.1; Bio- Rad Laboratories). Results from different groups were compared using the Student’s t test or Kruskal–Wallis non-parametric AN- OVA. Values of P < 0.05 were considered statistically signific- ant. All statistical analysis was performed with GraphPad Prism 5 software (GraphPad Software).

RESULTS

RIP3 is increased in the liver of patients with chronic liver disease

Fatty acids, alcohol, viruses and the immune response are well-known triggers of liver injury and chronic disease, at least in part by strongly activating liver cell death [1]. Nevertheless, the relative contribution of necroptosis to the pathogenesis of inflammation-associated human liver patholo- gies, namely NAFLD, ASH, hepatitis B and hepatitis C (Fig- ure 1A), has not been systematically explored.

RIP3 is a crucial player in the necroptotic signalling path- way. For instance, cell death shifts from apoptosis to necroptosis when RIP3 is overexpressed in NIH3T3 mouse fibroblasts [4]. Moreover, the expression of RIP3 in different cell types correl- ates with their responsiveness to induction of necroptosis [10]. Thus, to assess the presence of necroptosis in chronic liver dis- eases, RIP3 protein expression was evaluated by immunohisto- chemistry in liver biopsies from patients with steatosis, NASH, ASH, hepatitis B and hepatitis C, as well as in histologically normal subjects (Figure 1B). In normal livers, RIP3 was weakly expressed. Notably, all chronic liver disease patients showed a significant increase in liver RIP3 expression, when compared with healthy controls (at least, P < 0.05), except for steatosis. Of note, hepatic steatosis and steatohepatitis grading and sever- ity (Table 2) correlated with induction of RIP3. Particularly, a positive correlation was found between RIP3 expression and the relative degree of portal fibrosis in a cohort of control individuals, steatosis, NASH and ASH patients (r2 = 0.91, P < 0.05).

NAFLD patients display increased biological markers of necroptosis

Among chronic liver diseases, NAFLD is becoming increasingly recognized as an important public health concern with an im- portant economic burden. Although pathogenesis remains in- completely understood, hepatocyte cell death has been shown to play a key role in disease progression [20]. We next further assessed the contribution of necroptosis during NAFLD patho- genesis. Markers of hepatocellular injury, namely transaminases and γ -GT, were elevated in the serum of patients with steatosis and, more significantly, in patients with NASH (at least, P < 0.05; Figure 2A). Moreover, absolute ALT levels were higher than AST levels in NAFLD patients. To further explore the role of hepatic necrosis in NAFLD pathogenesis, we determined serum levels of diverse biomarkers of necrosis/necroptosis, previously shown to associate with liver damage [30,31]. Determination of cir- culating levels of cytokeratin 18 cleaved by caspases has been suggested as a blood biomarker for predicting the presence of NASH [30]. Similarly, CK18-M65, the major intermediate fil- ament in the liver, passively released from necrotic cells into the extracellular compartment, were found significantly elevated in steatosis (P < 0.05) and NASH (P < 0.01), compared with healthy volunteers (Figure 2B). HMGB1 protein is a ubiquitous nuclear protein that is released by necrotic cells into the extracel- lular space, thus triggering an inflammatory response by interact- ing with different cellular receptors, like the toll-like receptors (TLRs). In fact, NAFLD patients displayed increased protein systemic levels of HMGB1 (P < 0.01) and CypA (P < 0.05), a proposed early marker of necroptosis [32]. In addition, TNF-α is a well-established modulator of necroptosis and contributes not only to the liver inflammatory process, but also to hepatic insulin resistance and fat accumulation associated with NASH [33,34]. Accordingly, serum levels of TNF-α were significantly increased in NASH patients, compared with steatosis (P < 0.01) and con- trols (P < 0.05; Figure 2C). Finally, MLKL phosphorylation by RIP3 kinase is specifically required for ensuring necroptotic cell death. We found that increased expression of RIP3 in the liver of NASH patients (P < 0.01) is accompanied by an increase of MLKL phosphorylation (P < 0.05; Figure 2D). Taken together, the enhanced liver levels of RIP3, accompanied by an increase of MLKL phosphorylation and serum markers of hepatic nec- rosis and TNF-α in NASH patients strongly suggest a role for necroptosis in NAFLD progression.

Liver necroptosis is activated in HFCD diet-induced experimental NASH

To further explore whether necroptosis is actively involved in liver damage during NAFLD, we assessed the role of liver necrop- tosis in an experimental mouse model of NASH. Adult C57BL/6 mice were fed a standard diet (control) or a HFCD diet for 6 and 18 weeks. This dietary model does not induce significant changes in growth-rate and body weight of the mice (result not shown). No macroscopic abnormalities were detected in the liver of pair-fed control animals. Conversely, after 6 and 18 weeks, livers from HFCD-fed mice showed typical features of steato- hepatitis. Histological analysis of liver sections from mice fed a HFCD diet for 6 weeks revealed severe steatosis (P < 0.01) and inflammation (P < 0.01), without relevant fibrosis in mice fed a HFCD diet for 6 weeks, compared with controls. After 18 weeks, steatosis and inflammation were still severe, but also associated with hepatocellular ballooning and sinusoidal fibrosis (Figure 3A; Table 3), closely mimicking the histopathological features of hu- man NASH. In fact, HFCD-fed mice also exhibited physiological changes related to disease progression, namely increased mRNA levels of proinflammatory cytokines TNF-α and IL-1β in the liver (P < 0.05; Figure 3B).

RIP3-dependent signalling was very recently suggested to play a role in hepatic cell death and fibrosis in an animal model of NASH [35]. However, RIP3 also mediates non-necroptotic path- ways that could be involved in liver damage, including inflam- masome activation and inflammatory cytokine activation [36]. In turn, MLKL is a specific and critical RIP3-dowstream effector of necroptosis. Our results showed that RIP3 and MLKL expression was strongly increased in whole liver cell lysates from mice fed HFCD diet for 18 weeks (P < 0.05; Figure 3C), being preferen- tially sequestered in the insoluble protein fraction (Figure 3D). Although RIP3 and MLKL levels were similar between control and HFCD-diet groups at 6 weeks (Figure 3C), these proteins appeared to be more significantly retained in the insoluble pro- tein fraction of the HFCD-fed mice livers, compared with the soluble fraction (Figure 3D). These findings are consistent with the known insoluble amyloid structure of the necrosome, critical in the transmission of the pro-necroptotic signal [2,37]. Overall, our results suggest that HFCD diet-induced liver damage involves activation of necroptosis at both 6 and 18 weeks of feeding. Still, RIP3- and MLKL-dependent signalling is more significant at 18 weeks, concomitantly with aggravated liver injury, suggesting that necroptosis is involved in NAFLD progression.

Palmitic acid and TNF-α induce necroptosis in primary rat hepatocytes

The activation of necroptosis in human and mouse NAFLD prompted us to further explore the mechanisms governing these actions in vitro, determining whether distinct relevant stimuli are able to trigger necroptosis. PA is a saturated non-esterified fatty acid contributing to NASH pathogenesis [38] and widely used in in vitro models of NASH [39]. In addition, TNF-α also plays a key role in NAFLD pathogenesis. Because activ- ation of death receptors in cells with defective apoptotic ma- chinery can lead to necroptosis [3], primary rat hepatocytes were loaded with either PA or TNF-α plus CHX, in the presence or absence of zVAD-fmk, to inhibit caspase activation. In paral- lel, cells were incubated with or without Nec-1, to further con- firm whether necroptosis is a significant cell-death pathway in hepatocytes and, simultaneously, whether it can be effectively targeted, hinting at prospective novel therapeutic strategies for NAFLD and related inflammation-associated liver diseases. Both PA (Figure 4A) and TNF-α/CHX alone (Figure 4B) induced ap- optosis of primary rat hepatocytes, as revealed by activation of the executioner caspases-3/-7 (P < 0.05) and overall cell death (P < 0.01; Figures 4A and 4B). Co-incubation of either PA or TNF-α/CHX with Nec-1 had no effect on caspase 3-dependent cell death. Interestingly, suppression of caspase-3/-7 activities us- ing zVAD-fmk was not able to prevent overall cell death, despite completely inhibiting caspases activity (P < 0.01; Figures 4A and 4B). In agreement, TNF-α/CHX alone induced apoptosis, whereas co-incubation with zVAD-fmk completely abrogated it (result not shown). These results suggest that, in conditions were apoptosis is blocked, hepatocytes can effectively shift between different cell death pathways to still eliminate injured cells. In the present study, because Nec-1 was effective at inhibiting cell death induced by co-incubation of either PA or TNF-α/CHX with zVAD-fmk (at least, P < 0.05), the alternate cell-death pathway appears to be necroptosis. To confirm this, RIP3 expression was evaluated by immunocytochemistry in TNF-α/CHX-induced ap- optotic and necroptotic hepatocytes (Figure 4C). Not surpris- ingly, RIP3 expression was not affected by Nec-1, as it mainly targets RIP1 kinase activity. Curiously, TNF-α/CHX induced a slight increase in RIP3 expression levels (P < 0.05), compared
with control cells. Still, this increase more than doubled in the presence of zVAD-fmk (P < 0.05), consistent with the role of RIP3 in necroptosis. Further, in unstimulated or TNF-α/CHX- treated hepatocytes, RIP3 displayed a diffused pattern or occa- sionally accumulated in the nucleus (Figure 4C, upper panel). On the other hand and in agreement with the amyloid structure of the necrosome, a large number of cells displayed a RIP3 punctuated pattern within the cytoplasm, after TNF-α/CHX plus zVAD-fmk stimulation. Collectively, our results indicate that primary rat hepatocytes are vulnerable to necroptosis induced by insults involved in NAFLD pathogenesis, which in turn can be rescued by Nec-1.

TNF-α induced necroptosis involves RIP3-dependent ROS production

To evaluate the functional role of RIP3 during primary rat hepato- cyte necroptosis, cells were transfected with a siRNA specific for RIP3. RIP3 siRNA-transfected cells displayed a 60 % decrease in total RIP3 protein levels, compared with siRNA-control trans- fected cells (P < 0.01; Figure 5A). After RIP3 knockdown, cell death induced by TNF-α/CHX plus zVAD-fmk was abrogated (P < 0.01), whereas no effects were observed in cells incubated with TNF-α/CHX alone (Figure 5B). These results confirm the specific role of RIP3 in hepatocyte necroptosis.

Oxidative stress is a well-established mediator of liver injury in NAFLD pathogenesis [26,40]. Moreover, in specific cellular contexts, oxidative stress has also been suggested to play a crucial role in executing necroptosis [4,9]. A significant increase in ROS production was detected in primary rat hepatocytes exposed to TNF-α/CHX plus zVAD-fmk for 3 h, when compared with TNF- α/CHX alone and no addition (P < 0.05; Figure 5C). Consistent with its effects on necrosome assembly, Nec-1 pre-treatment sig- nificantly counteracted this increase (P < 0.05).

To further confirm the association between necroptosis, RIP3 and oxidative stress, cultured hepatocytes isolated from WT and RIP3—/— mice were used. Both TNF-α/CHX and TNF-α/CHX plus zVAD-fmk increased overall cell death and ROS levels in WT hepatocytes by ∼50 % (Figures 5C and 5D). Of note, cell death and oxidative stress were completely abrogated upon incubation of RIP3—/— mouse hepatocytes with TNF-α/CHX plus zVAD-fmk but not TNF-α/CHX alone, confirming the specific critical role of RIP3 in necroptotic signalling. Altogether, oxidative stress is a downstream event of RIP3 activation and appears to play a mechanistic role during necroptotic signalling induced by TNF-α in hepatocytes.

Absence of RIP3 ameliorates hepatic damage during MCD diet-induced experimental NASH

To further evaluate the functionality of RIP3-dependent necrop- tosis in NASH, C57BL/6 WT and RIP3—/— mice were fed a MCD diet for either 2 or 8 weeks. This represents a traditional dietary mouse model of NASH that induces steatosis, inflammation and hepatic fibrosis. Serum ALT and AST levels were significantly increased in MCD diet-fed WT mice compared with control mice on chow diet, at both time-points (at least, P < 0.05; Figure 6A). Importantly, circulating levels of hepatic enzymes were signific- antly reduced in RIP3—/— animals (P < 0.05), indicating a role for RIP3 during MCD diet-induced hepatic injury. Next, we scored liver lesions and accumulation of lipid droplets in the hepato- cytes, after MCD-diet feeding. WT and RIP3—/— mice on chow diet displayed normal liver morphology and minimal hepatocel- lular lipid accumulation. WT and RIP3—/— mice on the MCD diet for 2 weeks showed mild to moderate vacuolation of hepato- cytes, corresponding to accumulation of small to medium-size lipid droplets, as seen with the Oil Red O staining. These lesions were of similar severity in both groups. In turn, WT mice on the MCD diet for 8 weeks showed moderate to severe vacuola-
tion of hepatocytes, corresponding to accumulation of large lipid droplets, whereas RIP3—/— mice displayed significantly decreased hepatocyte vacuolation and smaller and scattered lipid droplets (P < 0.01; Figures 6B and 6C).

Because inflammation is a key pathological feature of NASH, we assessed hepatic expression of proinflammatory mediators in both WT and RIP3—/— mice. MCD feeding induced a ∼2- and (Figure 6F). Altogether, our results show that RIP3-dependent signalling promotes liver injury, hepatic steatosis, inflammation and fibrosis in murine NASH, thus contributing to NAFLD patho- genesis.

Activation of RIP3-dependent necroptosis and oxidative stress constitute early events in MCD-diet induced NASH

RIP3-dependent signalling can promote proinflammatory cy- tokine production independently of necroptosis activation [36]. The fact that RIP3 deficiency abrogates cytokine expression after 2 weeks of MCD feeding without significantly affecting lipid de- position prompted us to further confirm whether RIP3-dpendent necroptosis is, in fact, activated at early time-points of MCD diet-induced NASH. Because MLKL and RIP3 oligomerization is a pre-requisite for necroptosis execution, we evaluated total and soluble/insoluble levels of RIP3 and MLKL in liver cell lysates from MCD diet-fed mice for 2 weeks. The MCD diet in- creased RIP3 and MLKL expression in whole liver cell lysates from WT mice compared with controls (at least, P < 0.05; Fig- ure 7A). More importantly, RIP3 and MLKL were accumulated in the insoluble protein fraction of WT mouse livers after MCD feeding (P < 0.05; Figures 7B and 7C). Remarkably, although the increase in total hepatic MLKL expression was similar between WT and RIP3—/— mice after MCD feeding (Figure 7A), absence of RIP3 prevented sequestration of MLKL in the insoluble pro- tein fraction of the livers (Figures 7B and 7C). These findings suggest that RIP3 is critical to MLKL oligomerization in the liver in response to MCD diet and, hence, necroptosis is activated early in this NASH model.

Finally, since MCD diet-induced NASH is associated with in- creased oxidative stress and our mechanistic studies showed a link between RIP3-dependent necroptosis and ROS production, we further assessed whether RIP3 deficiency modulates the oxidative stress response in vivo. Our results showed that absence of RIP3 strongly decreased ROS generation in the liver after MCD for 2 weeks compared with WT mice (P < 0.05; Figure 7D). These findings confirm that oxidative stress is a downstream event res- ulting from activation of necroptosis during NAFLD-associated liver injury.

DISCUSSION

The relevance of apoptosis in the pathogenesis of inflammation- driven liver diseases, such as NAFLD, has been extensively doc- umented. Interestingly, necroptosis is a well-orchestrated form of programmed cell death, sharing upstream signalling elements of apoptosis and morphological characteristics of necrosis. Al- though necroptosis activation has recently been implicated in a variety of pathological conditions [3,10–13], its importance in the pathogenesis of inflammatory liver diseases has been poorly explored. In the present study, we show that necroptosis is also a common and important cell death pathway in hepatocytes and a pathological feature of NAFLD, contributing to inflammation and liver damage.

RIP3 kinase represents a crucial mediator of necroptosis and its protein expression levels correlate with cell sensitivity to nec- roptosis [10,11,16]. Interestingly, we found that RIP3 is weakly expressed in hepatocytes. Thus, it is likely that necroptosis does not occur under physiological conditions in the liver. However, our results also show that RIP3 is significantly induced in human hepatocytes from patients with chronic liver disease and, as such, it may contribute to their enhanced susceptibility to necroptosis under these pathological conditions, with concomitant effects in liver injury. In fact, we show that RIP3 induction strongly correl- ates with steatohepatitis severity. In support, recent reports found elevated RIP3 expression levels in liver biopsies from patients with alcoholic liver disease [14] and NASH [35], suggesting links between RIP3-dependent necroptosis and animal liver in- jury. In addition, activation of RIP3-dependent necroptosis in animal models of drug-induced hepatotoxicity and acute liver failure has also been reported [14,16]. Still, our results are the first to show that necroptosis is a common pathological feature in a range of human inflammation-associated liver diseases. Of note, because RIP3 may also execute necroptosis-independent functions during inflammation [36], we cannot yet fully exclude the possibility that other RIP3-associated events are simultan- eously activated in chronic liver diseases, contributing to disease pathogenesis.

Given the increasingly worldwide prominence of NAFLD, its association with inflammation and the poor knowledge of the pathogenic mechanisms governing its progression from simple steatosis to advanced NASH and, further, to cirrhosis and end- stage liver failure, we focused our analysis on the pathogenic role of hepatocyte necroptosis using human samples and experi- mental murine models of NAFLD. Our results showed that RIP3 expression was significantly increased in liver tissue from pa- tients with NASH, compared with control individuals and only slightly increased in patients with steatosis. The mechanism by which RIP3 executes necrotic cell death involves MLKL phos- phorylation and subsequent oligomerization and translocation to cellular membranes, causing its lysis. Thus, MLKL is a key player in necroptotic signalling and its increased phosphoryla- tion is a specific marker of necroptosis activation. Intriguingly, MLKL phosphorylation in liver and serum markers of necrosis were increased in both steatosis and NASH patients. These res- ults suggest that hepatocyte necroptosis is triggered in the ab- sence of significantly increased levels of RIP3 expression and that
increased RIP3 protein levels immediately follow. Further, they also hint at the possibility of using CypA as an early indicator of NAFLD in patients with metabolic syndrome. This was corrob- orated in two different dietary in vivo models of NAFLD. We first showed that the HFCD diet induced histopathological alterations in mouse liver that closely resembles those observed in human NASH, including severe steatosis, inflammation and fibrosis. In fact, similarly to the observations in the NAFLD human liver samples, mice on the HFCD diet displayed steatohepatitis after 6 weeks of feeding and, yet, RIP3 overall levels remained un- changed. However, RIP3 and MLKL were found significantly sequestered in the insoluble protein fraction of liver lysates, providing strong evidence of necrosome assembly and necrop- tosis activation [2,37]. At 18 weeks of feeding, steatohepatitis was aggravated and RIP3 and MLKL strongly expressed and retained in the insoluble protein fraction of mice liver lysates. Again, these results suggest that necroptosis is engaged prior to highly increased RIP3 protein expression levels, contributing to leak- age of pro-inflammatory mediators, such as CyPA and HMGB1. Upon a sustained stimulus, in this case the HFCD diet, it is likely that the increasing TNF-α production exacerbates necroptosis- associated liver injury and hence aggravates liver inflammation. Indeed, we demonstrated that RIP3 deficiency abrogates MCD- induced TNF-α expression, whereas TNF-α may induce necrop- tosis of cultured primary murine hepatocytes. Moreover, absence of RIP3 also attenuates hepatic inflammation, steatosis, liver in- jury and fibrosis after MCD feeding. Finally, it has been reported that Nec-1 is able to rescue MCD diet-mediated liver injury in mice [41]. Altogether, these findings strongly indicate that nec- roptosis is a key pathogenic factor involved in NAFLD triggering and progression. Importantly, we showed for the first time that mice on the MCD diet for 2 weeks, when steatohepatitis is quickly and strongly developing, display increased RIP3 protein levels, accompanied by RIP3-dependent MLKL sequestration in insol- uble liver protein fractions. These findings further suggest that the assessment of RIP3 and MLKL protein levels in soluble/insoluble proteins fractions might be considered a valuable approach to characterize and identify necroptosis activation in vivo, filling the gap of lack of available specific biomarkers of necrop- tosis and excluding the involvement of non-necroptotic RIP3 functions.

With evidences that necroptosis is an active process during NAFLD, we also sought to elucidate some of the mechanisms underlying its activation. Our results showed that TNF-α is in- creased in serum of obese NASH patients and in the liver of NASH animals, in agreement with previous reports [33,42]. Fur- ther, TNF-α levels correlate with liver disease severity [43] and tumour necrosis factor receptor 1 (TNFR1) is overexpressed in livers of patients with NASH [43,44], whereas TNFR1-deficient mice are resistant to steatosis and liver injury in animal mod- els of stetohepatitis [42]. Therefore, given its key role in nec- roptosis, TNF-α may trigger necroptotic signalling pathways in hepatocytes, through stimulation of TNFR1, thus mediating NAFLD-related hepatocyte injury. In agreement with this hy- pothesis, we show that incubation of primary hepatocytes with TNF-α/CHX plus a pancaspase inhibitor triggered cellular nec- roptosis. Moreover, it also promoted RIP3 up-regulation, which was also observed in human pathological liver samples and in in vivo models of steatohepatitis. Likewise, our results also show that co-incubation of PA and a pancaspase inhibitor also induce necroptosis in primary rat hepatocytes. Alongside TNF-α and PA, other stimuli are probably involved in triggering necroptosis in the NAFLD context. For instance, during NASH, hepatocytes are heavily exposed to TLR4 ligands, including the intestine- derived lipopolysaccharide (LPS), non-esterified fatty acids and HMGB1 [45]. In turn, activation of TLR4 in hepatocytes may trigger necroptosis by activating RIP3 through toll IL-1 receptor (TIR) domain-containing adaptor-inducing interferon-β (TRIF) [46].

Of note, in vitro, necroptosis appears to be mainly activated only when apoptosis is suppressed in the presence of zVAD- fmk, as Nec-1 had little protective effect in cells incubated with TNF-α/CHX alone. The apparent lack of relevance of necroptosis in vitro might have masked the importance of this cell-death path- way regarding the pathogenesis of liver diseases. Nevertheless, despite our evidence highlighting the role of necroptosis during NAFLD progression, apoptosis remains a well-established key feature of NAFLD [20,21]. In fact, necroptosis and apoptosis may be simultaneously activated in a liver disease context [16], specifically in NAFLD. In agreement, both genetic and pharma- cological inhibition of caspases, in mouse models of NASH, was shown to be unable to completely abrogate hepatocyte cell death and oxidative stress, as well as liver inflammation and hepatic steatosis [40,47].

RIP3 has been shown to be the main switch responsible for shifting TNF-α-induced cell death from apoptosis to necroptosis in NIH3T3 mouse fibroblasts, in part by increasing metabolism- associated ROS generation [4]. Accordingly, we found that expos- ure of primary murine hepatocytes to TNF-α/CHX plus zVAD- fmk also triggers a RIP3-dependent ROS burst. In fact, although the link between ROS production and necroptosis appears to be context- and cell-specific [3], it has been shown that necroptotic liver injury in response to acetaminophen involves mitochondrial dysfunction and oxidative stress [15]. In addition, ROS plays a pathogenic role in NAFLD and predictors of oxidative stress have been found in both patients [26] and animal models of NASH [40]. Interestingly, these markers of oxidative stress positively correlate with necroinflammation in NASH patients, independ- ently of the steatosis grade [48]. Free radicals may originate from activated lipid metabolism during NAFLD pathogenesis. How- ever, oxidative stress may have causes other than changes in lipid metabolism. In fact, our results also show that ROS generation is a downstream event of RIP3 activation after MCD feeding for 2 weeks. Altogether, these studies corroborate our findings that oxidative stress is critical for the necroptosis signalling pathway in hepatocytes and further highlight that this is a mechanism occurring in vivo, during NASH pathogenesis and progression.

Unlike a previous publication [35], we found that MCD- induced hepatic lipid accumulation is significantly attenuated by RIP3 deficiency at 8 weeks. In agreement with our findings, it has been reported that absence of RIP3 reduced ethanol-induced hepatic steatosis by unknown mechanisms [14]. A possible ex- planation is that absence of RIP3 prevents oxidative damage and cytokine up-regulation, altering hepatic fat metabolism [20], and hence attenuates steatosis. Despite the current NAFLD burden, no specific and effective pharmacological therapy is yet available. Steatosis is usually benign, but can progress to NASH, increasing the risk of liver-related morbidity and mortality. To develop novel therapeutic strategies for NAFLD treatment, a better understand- ing of the underlying molecular, cellular and biochemical mech- anisms implicated in the onset and progression of the disease is imperative. The present study underscores the presence and active role of RIP3-dependent necroptosis during NAFLD trig- gering and progression, among other inflammation-driven liver diseases. Further, mechanistic studies established the association between TNF-α-induced RIP3 expression, activation of necrop- tosis and oxidative stress in hepatocytes, shedding new light on disease pathogenesis. The targeting of liver necroptotic signalling pathways could arise as a promising therapeutic option to arrest NAFLD and other inflammation-associated liver diseases’ devel- opment and progression.

CLINICAL PERSPECTIVES

. Regulated necrosis or necroptosis, an immunogenic pro- grammed cell death type, morphologically similar to nec- rosis, has recently been implicated in the pathogenesis of inflammation-driven liver disease. In turn, necroinflammation is an important pathological feature of NASH, whereas TNF-α is a well-established trigger of necroptosis, involved in NASH pathogenesis.
. In the present study, we found that liver RIP3 levels are sig-
nificantly increased in several chronic liver diseases, correlat- ing with steatohepatitis histological severity. In addition, nec- roptosis, as defined by RIP3-dependent MLKL activation, is triggered in the liver of mouse models of NASH and in human NAFLD, whereas absence of RIP3 ameliorates liver injury, steatosis, inflammation and fibrosis in MCD-induced exper- imental NASH. Mechanistically, TNF-α-induced necroptosis of primary murine hepatocytes involves RIP3-dependent ROS production.
. Targeting necroptosis may provide an unprecedented oppor-
tunity to develop novel therapeutic strategies to attenuate or prevent liver injury, inflammation and fibrosis associated with NAFLD pathogenesis.