Mechanisms of acetaminophen hepatotoxicity and their translation to the human pathophysiology.

Acetaminophen (APAP) overdose is the most common cause of acute liver failure in the United States and mechanisms of liver injury induced by APAP overdose have been the focus of extensive investigation. Studies in the mouse model, which closely reproduces the human condition, have shown that hepatotoxicity is initiated by formation of a reactive metabolite N-acetyl-p-benzoquinone imine (NAPQI), which depletes cellular glutathione and forms protein adducts on mitochondrial proteins. This leads to mitochondrial oxidative and nitrosative stress, accompanied by activation of c-jun N-terminal kinase (JNK) and its translocation to the mitochondria. This then amplifies the mitochondrial oxidant stress, resulting in translocation of Bax and dynamin related protein 1 (Drp1) to the mitochondria, which induces mitochondrial fission, and ultimately induction of the mitochondrial membrane permeability transition (MPT). The induction of MPT triggers release of intermembrane proteins such as apoptosis inducing factor (AIF) and endonuclease G into the cytosol and their translocation to the nucleus, causing nuclear DNA fragmentation and activation of regulated necrosis. Though these cascades of events were primarily identified in the mouse model, studies on human hepatocytes and analysis of circulating biomarkers from patients after APAP overdose, indicate that a number of mechanistic events are identical in mice and humans. Circulating biomarkers also seem to be useful in predicting the course of liver injury after APAP overdose in humans and hold promise for significant clinical use in the near future.

Introduction

Acetaminophen is an analgesic drug, which is safe at therapeutic doses, but can produce significant hepatotoxicity with an overdose. APAP hepatotoxicity is the most frequent cause of acute liver failure (ALF) in the US [1] and a recent study evaluating outcomes in adults with acute liver failure between 1998 and 2013 indicates that hepatotoxicity due to APAP accounted for almost half the cases of ALF for the entire 16-year period, with unintentional APAP overdoses, (those in which patients took excessive medication over several days for ailments like pain or fever) being more common than intentional (suicidal) overdoses [2]. This is similar to an earlier study, which showed that unintentional APAP overdose accounted for over 50 % of cases of acetaminophen-related ALF [3]. These unintentional overdoses are mainly driven by the increasing availability of combination drugs, which contain acetaminophen in addition to other drug classes such as opioids [4]. Once the hepatotoxicity due to APAP was recognized, the mechanisms behind this were extensively investigated, especially in the mouse model, which recapitulates key mechanistic aspects of liver injury in humans such as mitochondrial dysfunction [5,6]. These insights, especially regarding GSH depletion and protein adduct formation [7], resulted in development of N-acetylcysteine (NAC) as an antidote for APAP overdose [8,9]. NAC treatment is now the standard of care under these conditions, and in spite of the significant therapeutic potential of NAC in preventing APAP- induced ALF, it has to be administered early after APAP consumption to have the most benefit. Since this may not be possible for most patients with an APAP overdose, especially those with unintentional overdoses, new therapeutic options which could benefit when administered at later time points are continuously being investigated. These studies have led to significant additional insight into cellular signaling events driving hepatocyte death and liver injury after APAP overdose and will be reviewed in this article.

Acetaminophen metabolism initiates liver injury.

Therapeutic doses of APAP are typically conjugated with glucuronic acid and sulfate, which are the main metabolites of APAP, and are then excreted in the urine [10]. A minor component of APAP is also oxidized by the microsomal cyto-chrome P450 system, predominantly by Cyp2E1 and Cyp1A2 [11], to form a reactive metabolite, N-acetyl-p-benzoquinone imine (NAPQI) [12]. This minor metabolite is typically harmless, since it is mostly conjugated with glutathione and excreted in bile [13,14]. However, even at these low, therapeutic doses, there occurs very limited reaction with protein sulfhydryl groups leading to covalent binding and protein adduct formation [15,16]. The impact of these protein adducts is limited because they are effectively removed by autophagy [17]. It is only when high levels of APAP saturate the sulfation pathway and glucuronidation cannot keep up any longer [18], that there is an excess generation of the reactive NAPQI metabolite. This then consumes glutathione for its conjugation, resulting in depletion of glutathione stores. While the initial depletion is similar in both GSH and GSSG, without affecting the GSSG:GSH ratio (1:200), the recovery rates are different, with GSSG content increasing faster than that of GSH [19]. Recovery rate of GSH can significantly influence injury, since an induction of glutamate-cysteine ligase which correlated with faster recovery of GSH is one of the mechanisms by which female mice are protected against APAP-induced liver injury [20]. A differential metabolomics study suggests that the depletion of glutathione after low dose APAP (150mg/kg) is paralleled by elevation in the glutathione analogue ophthalmic acid, where the SH group of the cysteine residue of GSH is replaced with a CH3 group from 2-aminobutyrate [21]. Detection of ophthalmic acid in serum from APAP-induced acute liver failure patients was also more frequent in non-survivors [22]. As glutathione depletion occurs, there is an increasing reaction of NAPQI with sulfhydryl groups of proteins to form protein adducts [15]. In contrast to earlier assumptions, this generation of protein adducts can take place before GSH levels are depleted extensively [15,23] and also after therapeutic doses of APAP without relevant GSH depletion [15,16]. Interaction of NAPQI with mitochondrial proteins and formation of mitochondrial protein adducts is thought to be critical for the toxicity [24-26].

Mitochondrial protein adduct formation and APAP hepatotoxicity.

Though mitochondria were traditionally considered important cellular organelles due to their role in ATP generation, it is now evident that they play important roles in various cell signaling scenarios, including cell death by regulated necrosis [27]. Formation of NAPQI adducts on mitochondrial proteins was found to be unique to APAP when compared to its regioisomer 3’-hydroxyacetanilide (AMAP), which is non-toxic in mice [24,28,29]. In contrast, recent reports indicate that AMAP is cytotoxic in primary human hepatocytes or precision-cut human liver slices [30,31]. Interestingly, AMAP toxicity in primary human hepatocytes correlated with mitochondrial protein adduct formation and mitochondrial dysfunction [31]. Specific targets within the mitochondria, such as glutathione peroxidase and the alpha subunit of ATP synthase have been identified to undergo adduct formation by proteomic approaches [32]. Enzymes such as HMG CoA synthase have also been shown to be modified, accompanied by inhibition of enzyme activity [33]. APAP adducts on mitochondrial proteins such as glycine amidinotransferase, Parkinson disease protein 7 (PARK7), peroxiredoxin 6 and voltage-dependent anion-selective channel protein 2 (VDAC2) have also been detected in cultures of human hepatocytes [34], indicating that adduct formation is not a global phenomenon affecting all mitochondrial proteins, but rather selective with specific targets. While mitochondrial protein adducts are relevant to APAP hepatotoxicity since NAPQI binding to mitochondrial proteins correlates with the toxicity [35], the role of specific proteins targeted for adduct formation in the pathophysiology is not well understood.

Adduct formation initiates mitochondrial oxidative and nitrosative stress

Mitochondrial protein adduct formation results in increased superoxide production [36] (Figure 1) accompanied by compromised mitochondrial respiration [37]. The importance of mitochondrial superoxide production in APAP hepatotoxicity is illustrated by the fact that mice with a partial deficiency of manganese superoxide dismutase (SOD2) (which would usually scavenge superoxide) have exaggerated liver injury when exposed to APAP overdose [38,39]. The enhanced production of superoxide would allow its reaction with nitric oxide (NO) within the mitochondria producing the reactive radical peroxynitrite (ONOO–), which can nitrate protein tyrosine residues and compromise their function [40] (Figure 1). APAP hepatotoxicity results in early nitrotyrosine formation exclusively in hepatocyte mitochondria [41], suggesting that generation of elevated superoxide and its reaction with NO occurs within this organelle. The source of nitric oxide contributing to peroxynitrite formation within mitochondria, however, is not well characterized. It has been demonstrated that peroxynitrite formation subsequent to APAP overdose is independent of inducible nitric oxide synthase (iNOS) [42]. The lack of iNOS involvement was also suggested by the absence of protection with iNOS inhibitors [43] and in iNOS-deficient mice [44]. In contrast, neuronal NOS (nNOS) was shown to be present in hepatocytes [45] and a nNOS inhibitor was protective against APAP-induced cell death in isolated mouse hepatocytes [46]; likewise, nNOS-deficient mice showed delayed injury after APAP overdose [47], suggesting that nNOS could be a putative source of NO for peroxynitrite formation after APAP overdose.

Figure 1.

A Minardo representation [167] of acetaminophen hepatotoxicity in the mouse, illustrating the temporal separation of events after an APAP overdose. APAP hepatotoxicity is initiated by its conversion to the reactive intermediate NAPQI, which results in glutathione depletion and formation of APAP protein adducts. Adduct formation on mitochondrial proteins modulates respiratory chain function, producing elevated levels of free radicals such as superoxide. This, along with nitric oxide, generates peroxynitrite resulting in protein nitration within mitochondria. Mitochondrial oxidative stress results in oxidation of thioredoxin 1, releasing its partner ASK1, which activates JNK resulting in its phosphorylation and translocation to the mitochondrial outer membrane, where it interacts with Sab and subsequently stimulates free radical production from the mitochondrial electron transport chain. This in turn amplifies JNK activation and subsequent mitochondrial oxidant stress, which ultimately results in activation of RIP3 and translocation of Drp1 and Bax to the mitochondria. While Bax initiates outer membrane permeabilization, Drp1 induces mitochondrial fission and subsequent activation of the mitochondrial permeability transition. This then releases apoptosis inducing factor (AIF) and endonuclease, which translocate to the nucleus and initiates nuclear DNA fragmentation.

The relevance of peroxynitrite to APAP-induced hepatotoxicity is illustrated by the critical anti-oxidant proteins such as SOD2 it targets [48] and the protection afforded by its scavenging with delayed glutathione supplementation, which also replenished mitochondrial GSH levels [49-52]. In addition, the selective metabolism of superoxide by the mitochondrial targeted SOD-mimetic Mito-TEMPO effectively reduced APAP hepatotoxicity [53]. Furthermore, APAP overdose in SOD2 deficient mice also caused aggravated liver injury accompanied by exacerbated peroxynitrite and protein carbonyl formation [38,39].

The MAP kinase JNK and amplification of mitochondrial oxidative stress.

While formation of APAP-protein adducts on mitochondria initiates a mitochondrial nitrosative and oxidative stress, this effect alone does not seem to be sufficient to ultimately trigger the MPT and cell death. Thus, it was recognized that an oxidant stress-mediated activation of the MAP kinase JNK in the cytosol [54] is what ultimately seems to amplify the mitochondrial oxidant stress and results in downstream signaling events. JNK activation after APAP overdose occurs early after APAP overdose, and is then sustained during the signaling cascade inducing hepatocyte cell death. The apoptosis signal-regulating kinase 1 (ASK1) is involved in APAP-induced activation of JNK, with ASK1- deficient mice being protected against the sustained JNK elevation [55] and a specific ASK1 inhibitor decreasing JNK activation at 1.5 h and preventing JNK translocation to mitochondria [56]. In addition, it was found that the mixed-lineage kinase 3 (MLK3) was activated by oxidative stress and was required for JNK activation in response to oxidative stress [57]. It was also seen that JNK phosphorylation at one, three and six hours after APAP treatment was significantly attenuated in MLK3-KO mice [57]. Since MLK3 has been suggested to be part of a feedback mechanism that regulates cellular responses to ROS [58] and the activation of JNK can be prevented by anti-oxidants [42], it is possible that APAP-induced, ROS mediated JNK activation occurs through multiple mechanisms with temporal changes in their interaction.

Though activation of JNK in the cytosol seems to be an early event after APAP overdose, occurring within an hour after a 300 mg/kg dose in the mouse [56], it is also influenced by the dose of APAP. Lower APAP doses such as 150mg/kg were shown to induce transient JNK activation with reversible mitochondrial dysfunction in the mouse liver [26]. Once JNK is activated and phosphorylated in the cytosol, the next mechanistic step for amplification of mitochondrial injury is its translocation to mitochondria, where it binds to the Sab protein on the outer mitochondrial membrane [54,59]. Binding to and phosphorylation of Sab by p-JNK leads to inactivation of p-Src on the inner mitochondrial membrane, which then inhibits electron transport and increases reactive oxygen species release [54,60], thus amplifying oxidant stress and peroxynitrite formation [42]. While a few studies suggested that JNK was protective in acetaminophen toxicity [61,62], the data from one of them [62] could be influenced by the differing susceptibility to APAP toxicity of mice sub-strains used in the study [63]. It is however unclear why the knocking down of both JNK1 & 2 in hepatocytes resulted in a paradoxical exacerbation of APAP-induced liver injury [61], and this requires further study. The overall pathophysiological relevance of the JNK amplification loop in the murine models of APAP- induced liver injury has been extensively shown with various JNK inhibitors and JNK gene silencing [42,64,65] as well as the protection by inhibition of upstream signaling events [56,57,59]. Similar effects of APAP-induced JNK activation and mitochondrial p-JNK translocation was also observed in primary human hepatocytes although JNK inhibition only moderately reduced cell death [66]. It remains unclear if this is due to a species difference or reflects the greater dependence on oxidant stress amplification in vivo compared to cultured cells, which are generally kept under hyperoxic conditions resulting in more oxidant stress [36].

In parallel to activation and translocation of JNK, mitochondrial oxidant stress also results in the early translocation of the cytosolic protein Bax to the mitochondria [67,68]. While the initial mitochondrial oxidant stress seems to be required for Bax translocation [53], Bax does not seem to subsequently influence mitochondrial oxidant stress and peroxynitrite formation, since the protection against APAP-induced liver injury in Bax knockout mice was transient, only occurring at early time points [67]. However, recent information suggesting that Bax could be part of a large integrated network mediating various regulated forms of cell death through mitochondrial translocation [69] give tantalizing clues as to its implications in APAP-induced liver injury and will be further discussed below.

APAP overdose, mitochondrial dynamics and autophagy.

Mitochondria are dynamic organelles which undergo changes in morphology through cycles of fusion and fission. These processes are critical for mitochondrial homeostasis and bioenergetics [70]. Mitochondrial fusion and fission are regulated by a number of GTPase proteins such as optic atrophy 1 (OPA1), mitofusin 1/2 (Mfn1/2) and dynamin related protein 1 (Drp1) in mammalian cells [71]. While OPA1 and Mfn1 & 2 are involved in mitochondrial fusion, Drp1 mediates mitochondrial fission in mammals, where the role of mitochondrial fission 1 protein (Fis1) (a fission protein in yeast) is still not confirmed [72]. The impact of APAP overdose on mitochondrial dynamics was first identified when significant elevations in Drp1 and its translocation to the mitochondria were discovered after APAP overdose [73] (Figure 1). The role of Drp1 in mitochondrial fission after APAP overdose was subsequently confirmed by other studies [74], suggesting that alterations in mitochondrial dynamics after APAP overdose could have important mechanistic implications. Interactions between Bax and Drp1 have also been implicated during mitochondrial fission in pathophysiological conditions, with spatial and temporal association of Bax with mitochondrial fission sites, Drp1, and Mfn2 during apoptosis [75]. Bax has been suggested to be required for Drp1-mediated mitochondrial fission caused by photodynamic therapy in human lung adenocarcinoma cells [76], while Drp1 influenced Bax translocation to mitochondria in response to irradiation-induced apoptosis [77] and Drp1-induced membrane remodeling stimulates Bax oligomerization [78]. In addition, pharmacological inhibition of Drp1was shown to prevent Bax induced mitochondrial outer membrane permeabilization (MOMP) [79]. This interaction between Bax and Drp1, both of which translocate to the mitochondria after APAP overdose would suggest a scenario where APAP-induced Bax and Drp 1 translocation to mitochondria facilitate mitochondrial fission, which then initiates down-stream events such as opening of the mitochondrial permeability transition pore.

Removal of damaged mitochondria through autophagy (mitophagy) has been shown to limit APAP-induced injury [80], especially adjacent to the acute necrosis area [81]. Mitochondrial fission as observed during APAP hepatotoxicity may enhance this process [73]. The mitochondrial translocation of Parkin, an E3 ubiquitin ligase, is required for mitophagy induction after APAP overdose and acute knockdown of Parkin aggravates APAP-induced liver injury [82]. However, chronic deletion of Parkin renders animals resistant to APAP [82], possibly due to development of compensatory and adaptive mechanisms for the chronic loss of Parkin, which may contribute to the resistance to APAP-induced liver injury [82]. Parkin-independent mitochondrial spheroid formation may substitute for Parkin-dependent autophagy in removing damaged mitochondria [83]. On the other hand, deletion of the autophagy gene Atg5 leads to chronic injury, regeneration and inflammation, which also protected against APAP toxicity [84]. The mechanism of this protection involves the persistent activation of Nrf2 with higher GSH synthesis rates and increased hepatocyte proliferation [84]. These observations support the critical role of autophagy for cell survival under normal conditions and during APAP-induced liver injury.

Protein folding in the endoplasmic reticulum (ER) is a critical cellular function and various cellular stresses such as ROS or alterations in cellular calcium can impair protein folding and initiate ER stress. Mice treated with 200mg/kg of acetaminophen showed activation of ER stress with upregulation of GADD153/CHOP by 6 hours after APAP administration, accompanied by a decrease in Grp78 levels [85]. Higher doses of APAP also induce markers of ER stress, with doses of 450mg/kg APAP inducing activation of ER stress-responsive transcription factor ATF6 and transcriptional activation and elevated expression of GADD153/CHOP [86]. CHOP deficient mice were also shown to be protected against APAP-induced liver injury, though interestingly the protection was only seen in animals given APAP by gavage and not in those given APAP as an intra-peritoneal injection [87]. Hence, while ER stress does seem to occur after APAP overdose, the mechanisms by which APAP induces ER stress are poorly understood [88] and need more study.

APAP-induced mitochondrial permeability transition and regulated necrosis.

While the immediate consequence of amplification of mitochondrial oxidant stress is loss of mitochondrial protein function due to modification of thiols [33], peroxynitrite mediated nitrotyrosine formation [48] and oxidative mitochondrial DNA damage [41], the critical event resulting in escalation of cell wide damage is induction of the mitochondrial permeability transition (MPT). The mitochondrial permeability transition is an extensively studied phenomenon, which was initially linked to the apoptotic cell death pathway, though it is now recognized that it is activated in numerous forms of cell death. It involves initial permeabilization of the mitochondrial outer membrane, followed by an abrupt change in inner membrane permeabilization allowing exit of molecules less than 1500 daltons [89,90]. Though it has been suggested that Bax and Bak form the components of the mitochondrial permeability transition pore (MPTP) in the outer membrane [90,91], the components within the inner membrane are still being confirmed. While cyclophilin D is generally accepted as being one of the proven components and regulator of the MPT [92], recent evidence suggests that the c-subunit ring of the F1FO ATP synthase could also be a regulatory unit within the inner membrane [93]. The induction of the MPT subsequent to APAP overdose results in breakdown of the proton gradient across the membrane, loss of mitochondrial membrane potential, and consequently, cessation of ATP production, which eventually leads to cell death [94-97]. However, the role of cyclophilin D and the MPTP in APAP-induced liver damage seem to be dependent on the dose of APAP administered. For example, while use of a relatively low overdose of 200 mg/kg APAP demonstrated protection against liver injury in cyclophilin D-deficient mice [96], only transient protection was achieved when cyclophilin was inhibited using cyclosporine A in vitro [94]. Furthermore, treatment with a higher dose of APAP (600 mg/kg) in vivo also showed no protection against liver injury [98]. Though MPT was initially studied only in the context of pathophysiology, it is now being recognized that it could also have physiological roles [99], such as in the heart [100] and in calcium buffering in neuronal cells [101]. This may explain the transient MPTP opening seen after treatment of mice with low doses of APAP (150mg/kg) [26], when JNK activation was also seen only for a short term. However, sustained JNK activation resulted in irreversible activation of MPT [26]. The involvement of lysosomal iron translocating to the mitochondria through the calcium uniporter has also been suggested in APAP-induced MPTP opening [102,103] implying that reactive oxygen species formation through the iron-mediated Fenton reaction could also be an amplifying source to ultimately cause mitochondrial dysfunction. Hence, the activation of the mitochondrial permeability transition pore, which is a key mechanistic step in the cascade of cell signaling after APAP overdose, seems to be influenced by the dose of APAP, with lower doses producing a reversible activation, while higher doses produce a sustained effect.

The immediate consequence of Bax-induced permeabilization of the mitochondrial outer membrane would be release of a number of proteins from the inter-membrane space, which have been implicated in cell death pathways such as apoptosis. Release of these proteins including cytochrome c, second mitochondria-derived activator of caspase (Smac), apoptosis inducing factor (AIF) and endonuclease G have been extensively studied in the context of apoptotic cell death, and it is recognized that the mechanisms of their release may not be identical in all cases. Cytochrome c and Smac release occurs almost simultaneously, suggesting that they exit mitochondria by a similar mechanism [104], while AIF release from the mitochondria occurs at a different rate [105]. Also, while Bax was involved in release of cytochrome c and Smac, this was not the case for endonuclease G and AIF [106], suggesting that these two proteins may be released by slightly differing mechanisms in contrast to cytochrome c and Smac. It has also been shown that DRP1 can influence release of intermembrane proteins [104] implicating it in release of cytochrome c, Smac, AIF and endonuclease G which is evident after APAP [67,107]. Once in the cytosol, AIF and endonuclease G (both of which contain nuclear localization sequences [108]) translocate to the nucleus [107], where endonuclease G cleaves DNA every 50-300kb [109], thereby generating DNA fragments typically seen after APAP overdose [41,110] (Figure 1). AIF is also critical for DNA damage and liver injury due to its involvement in DNA fragmentation [111], a fact highlighted by the fact that AIF-deficient mice have lower DNA damage and liver injury after APAP [112].

Necrotic cell death after APAP overdose

As discussed, proteins released from the mitochondrial inter-membrane space after APAP are also seen in the cytosol after apoptotic cell death. However, despite the mitochondrial release of cytochrome c [107,113] during APAP-induced liver injury, there is no activation of caspases [114,115] and all morphological characteristics of necrosis including cell and organelle swelling, cell contents release and karyolysis are evident both in vivo and in vitro [116-118]. In addition, caspase inhibitors do not protect against APAP-induced liver injury [114,115,118, 119]. Recent studies have uncovered a number of molecular mediators dictating the mode of cell death after APAP overdose, which confirms that cells die by a controlled form of cell death, now termed regulated necrosis. Necroptosis, a form of regulated necrosis is typically mediated by activation of death receptors such as TNF receptor 1 (TNFR1), which ultimately lead to the assembly of a necrotic death complex (necrosome) [120], which consists of the receptor-interacting kinase 1 (RIP1), the receptor-interacting kinase 3 (RIP3), which interact through their homotypic interaction motif (RHIM) domains, as well as the pseudokinase mixed-lineage kinase domain-like protein (MLKL) [121]. The activation of these mediators such as the receptor interacting protein kinases 1 and 3 (RIPK 1 and 3), result in necrosis while inhibiting apoptosis [122], and RIP3 has been suggested to be a molecular switch between apoptosis and necrosis [123] implicating it as a unique molecular regulator dictating necrotic cell death. It has been suggested that death signals flows from RIP1 to RIP3 through their RHIM domains, resulting in recruitment of additional RIP3 molecules for signal propagation, implying that RIP3 oligomerization is the minimal functional unit that is required to drive necrosome assembly [121]. Activated RIP3 binds to MLKL, and subsequently phosphorylates MLKL [124]. Activated MLKL then traffics the necrosome to various phospholipid-rich cellular compartments, where MLKL interferes with membrane integrity, causing necrotic cell death [121].

Early induction of RIP3 levels were evident after APAP overdose [73,125], and genetic deletion of RIP3 delayed APAP-induced cell death [73], implicating the kinase in APAP-induced cell death pathways. This involvement of RIP3 in APAP-induced acute liver injury is supported by several reports showing upregulation of RIP3 protein expression in ethanol-induced liver injury. [126], non-alcoholic steatohepatitis [127] and furosemide-induced liver injury [128]. It was also recently reported that while RIP3 deletion was protective in ConA-induced autoimmune hepatitis, RIP1 inhibition exacerbated disease and accelerated animal death. In APAP-mediated liver injury however, blockade of either RIP1 or RIP3 was protective [125]. In addition, a recently developed, highly selective RIP3 inhibitor protected against APAP-induced cell death in isolated human hepatocytes in vitro and in an in vivo mouse model. [129]. In contrast to all this evidence, one group of researchers were unable to detect RIP3 expression in primary mouse hepatocytes under basal conditions or after treatment with APAP [74], a finding which could have been influenced by culture conditions [130].

It should be noted however, that the various physiological and pathophysiological processes may have slight differences in specific components of the necrosome, with inducers such as the murine cytomegalovirus (MCMV) and herpes simplex virus having other proteins with RHIM domains substituting for necrosome components[121]. Hence, requirements of specific necrosome components may be variable depending on the pathophysiology, with Toll-like receptor 3-mediated necrosis proceeding independent of RIP1 [131] and MLKL-mediated regulated necrosis proceeding independently of RIPK3 in inflammation after Con A-induced hepatitis [132]. The involvement of MLKL in APAP-induced liver injury is questionable however, since two studies found that MLKL knockout mice were not protected [74,132]. This information, coupled with the fact that APAP hepatotoxicity is not attenuated in absence of TNFR1 [133] or influenced by absence of TNF [134], leads to the conclusion that APAP-induced cell death probably does not involve classical necroptosis. However, reactive oxygen species, which are central players in the APAP-induced signaling cascade, have been shown to be involved in RIP3-induced necroptosis in the cardiomyocyte [135] and several studies have also implicated RIP1 in APAP-induced liver injury [57,73,74,136] further confirming that the mode of cell death after APAP is regulated necrosis. A recent report has identified an iron dependent non-apoptotic form of cell death termed ferroptosis [137] and studies on primary hepatocytes in culture seem to suggest that this form of cell death may play a role in APAP-induced hepatocyte injury [138]. However, the physiological significance of this form of cell death in APAP-induced liver injury in vivo needs further study. Thus, based on the current data in the literature, APAP-induced hepatocyte cell death does not represent classical apoptosis or necroptosis. While the exact sequence of events and identity of all the mediators involved in the process are still being investigated, hepatocytes after an APAP overdose die by regulated necrosis.

Role of inflammation and intercellular communication in APAP-induced liver injury

APAP-induced hepatocyte necrosis results in massive release of damage associated molecular patterns (DAMPs), which can then lead to recruitment of monocytes and neutrophils. [139]. While it is well established that inflammation is induced after APAP- induced liver injury, there has been some controversy in the literature regarding the biological role of this inflammation and whether it would be a useful therapeutic target. While a number of earlier studies suggested that inflammation played a role in hepatocyte necrosis with P2X7 receptor-mediated purinergic signaling thought to promote liver injury through the inflammasome [140,141], the identity of the cytotoxic cell type involved in immune mediated injury is not clear and the effect of P2X7 was shown to be due to inhibition of P450 isoenzymes by the inhibitor of P2X7 and not through inflammasome activation [142]. A considerable amount of data, reviewed in Woolbright et al. [139], exists that raises concerns about the role of sterile inflammation and the importance of inflammasome activation during APAP hepatotoxicity, and the majority of experimental evidence suggests that the extensive sterile inflammatory response during APAP hepatotoxicity is mainly beneficial by limiting the formation and the impact of pro-inflammatory mediators and by promoting tissue repair [143]. The role of inflammation in APAP-induced liver injury is further discussed in detail in the Woolbright article in this special issue [144].

Intercellular communication plays an important role in tissue homeostasis, and gap junctions between hepatocyte are important guardians of this process. The gap junctions are composed of connexin proteins [145] and a small molecule inhibitor of connexin 32 (2-aminoethoxy-diphenyl-borate) was suggested to be protective against APAP hepatotoxicity [146]. However, it was later shown that the 2-aminoethoxy-diphenyl-borate protects against acetaminophen hepatotoxicity by inhibiting cytochrome P450 enzymes and JNK activation [147]. Genetic deficiency of connexin32 was also found to have no effect on acetaminophen-induced cell death, inflammation or oxidative stress [148]. However, a study exploring the role of multiple connexins such as connexin26, connexin32 and connexin43 demonstrated that gap junction communication was compromised after APAP overdose, accompanied by a switch in connexin production from connexin32 and connexin26 to connexin43 [149]. Connexin43-deficient animals had aggravated liver injury after APAP overdose, with increased cell death, inflammation and oxidative stress, suggesting that hepatic connexin43-mediated signaling could protect against APAP-induced liver injury [149]. A recent study also suggests that inhibition of pannexins, a family of transmembrane channel forming proteins linking the cytosol to the extracellular environment, alleviates APAP-induced hepatotoxicity [150]. Thus, the intracellular signaling which results in hepatocyte necrosis after APAP could also influence inter-cellular communication through various gap junction proteins and further research in this area will provide additional insight into the transmission of APAP- induced signaling across the liver. Further details of the role of gap junctions in APAP hepatotoxicity are discussed in the Maes article in this special issue [151].

Translation of mechanisms of APAP-induced liver injury from mice to humans

The translation of mechanisms obtained from mouse models of APAP overdose to the human situation were only initiated recently. Exploration of intra-hepatic changes after APAP overdose in humans can only be studied using in vitro cell culture approaches, due to the lack of liver biopsies from these patients where the procedure is contraindicated. A major constraint towards these in vitro analyses was the metabolic incompetence of the commonly used liver cell line HepG2 [152]. However, recent studies in metabolically competent HepaRG cell lines [23] and primary human hepatocytes [66] have confirmed that exposure to high levels of APAP result in GSH depletion, APAP protein adduct formation in mitochondria and organelle dysfunction accompanied by oxidant stress, resulting in necrotic cell death in human cells as demonstrated earlier in mouse models. JNK activation and its translocation to the mitochondria were also evident in primary hepatocytes [66], while this was not seen in HepaRG cells [66]. Thus, many key intracellular signaling events discovered in mice are reproducible in human hepatocytes suggesting that common mechanisms are in play during progression of APAP-induced liver injury. Further confirmation of the similarity of mechanisms comes from the detection of APAP protein adducts in plasma from patients after APAP overdose implying that they generate a similar reactive metabolite as in mice [153]. Though APAP protein adduct levels could identify patients with potential APAP overdose [16,153], interpretation needs to be tempered by the fact that APAP protein adducts were detectable in circulation in humans taking therapeutic doses of acetaminophen and this persisted for over a week after dosing was stopped [154]. Also, protein adduct levels tend to peak at later time points in humans [18] when compared to mice [15], which correlates with the delayed cell injury in patients and human hepatocytes.

The central role of mitochondrial dysfunction and nuclear DNA fragmentation in APAP-induced liver injury in the mouse model is also replicated in humans as illustrated by the detection of a mitochondrial matrix enzyme, glutamate dehydrogenase, mitochondrial DNA and nuclear DNA fragments in patients with APAP overdose [155]. The importance of mitochondrial dysfunction is further confirmed by the correlation of these serum markers of mitochondrial damage with poor outcome, as defined by death or need for liver transplant to survive [156]. Confirmation of the mode of cell death in humans as being necrosis comes from the significant elevation in biomarkers of necrosis, such as full-length cytokeratin-18, HMGB1 and microRNA-122 in APAP overdose patients [157, 158] in contrast to apoptotic biomarkers such as the caspase-cleaved form of cytokeratin-18 and caspase-3 activity, which showed only a minor increase or no significant change, respectively, in these patients [155,157]. Thus, experiments in primary human hepatocytes and human HepaRG cells as well as measurement of specific circulating biomarkers from patients after APAP overdose indicate commonality of mechanisms of liver injury in mice and men. It is important to recognize that rats are generally less susceptible to APAP than mice and are not a good model for the human toxicity [25]. From an intervention standpoint, administration of antibodies against HMGB1 has been shown to be protective against APAP-induced liver injury, by enhancing liver regeneration and recovery [159]. HMGB1 neutralization also decreased bacterial translocation from the gut during APAP hepatotoxicity [160]. A recent study showed that use of a partly humanized anti-HMGB1 monoclonal antibody in mice resulted in a 50% reduction in liver injury after APAP overdose, with prolonged therapeutic efficacy when compared to NAC [161]. Interestingly, however, animals treated with the humanized antibody showed reduced regeneration [161], suggesting that mechanisms of protection were different from that seen with the rodent anti-HMGB1 treatment [159].

In addition to information of mechanisms of liver injury, prediction of the course of liver injury and severity in individual patients would be very important for APAP overdose patients from the clinical management standpoint. Measurement of liver enzymes such as ALT and AST are not very useful in this context, since they represent acute cell death and peak levels of ALT and AST are not predictive of clinical outcome [157,162]. Serum biomarkers which could predict acute liver failure and death, would hence be of immense clinical use. Towards this end, it is now evident that higher levels of cytokeratin-18, HMGB1 and glycodeoxycholic acid levels correlated with poor outcome [157,163]. At the other end of spectrum, in early presenting patients, elevations in miR-122, cytokeratin-18, HMGB1 and argininosuccinate synthetase could indicate liver injury before ALT/AST increases are measurable [162,164,165]. Profiles of circulating miRNA could also be useful in identifying APAP hepatotoxicity and in differentiating cause of cell damage [166]. Thus, key mechanistic steps in the cellular signaling cascade induced in the liver after APAP overdose first discovered in animal models have been corroborated in human hepatocytes and also indirectly by circulating biomarkers from human patients. Future studies in this area of biomarker discovery is likely to uncover additional molecules, which could identify therapeutic targets or help predict the course of liver injury after APAP overdose in the clinic [162].

In conclusion, significant advances have been made in recent years regarding cell signaling mechanisms involved in APAP-induced liver injury, with the role of mitochondrial dynamics and regulated necrosis being identified. A number of these key mechanistic events are also reproduced in humans, suggesting that cellular responses to APAP overdose are conserved between mice and men. However, significant lacunae exist in our knowledge regarding a number of steps in the process, justifying further studies on mechanisms as well as predictive biomarkers in the future.