Altered responsiveness to extracellular ATP enhances acetaminophen hepatotoxicity

Background Adenosine triphosphate (ATP) is secreted from hepatocytes under physiological conditions and plays an important role in liver biology through the activation of P2 receptors. Conversely, higher extracellular ATP concentrations, as observed during necrosis, trigger inflammatory responses that contribute to the progression of liver injury. Impaired calcium (Ca2+) homeostasis is a hallmark of acetaminophen (APAP)-induced hepatotoxicity, and since ATP induces mobilization of the intracellular Ca2+ stocks, we evaluated if the release of ATP during APAP-induced necrosis could directly contribute to hepatocyte death. Results APAP overdose resulted in liver necrosis, massive neutrophil infiltration and large non-perfused areas, as well as remote lung inflammation. In the liver, these effects were significantly abrogated after ATP metabolism by apyrase or P2X receptors blockage, but none of the treatments prevented remote lung inflammation, suggesting a confined local contribution of purinergic signaling into liver environment. In vitro, APAP administration to primary mouse hepatocytes and also HepG2 cells caused cell death in a dose-dependent manner. Interestingly, exposure of HepG2 cells to APAP elicited significant release of ATP to the supernatant in levels that were high enough to promote direct cytotoxicity to healthy primary hepatocytes or HepG2 cells. In agreement to our in vivo results, apyrase treatment or blockage of P2 receptors reduced APAP cytotoxicity. Likewise, ATP exposure caused significant higher intracellular Ca2+ signal in APAP-treated primary hepatocytes, which was reproduced in HepG2 cells. Quantitative real time PCR showed that APAP-challenged HepG2 cells expressed higher levels of several purinergic receptors, which may explain the hypersensitivity to extracellular ATP. This phenotype was confirmed in humans analyzing liver biopsies from patients diagnosed with acute hepatic failure. Conclusion We suggest that under pathological conditions, ATP may act not only an immune system activator, but also as a paracrine direct cytotoxic DAMP through the dysregulation of Ca2+ homeostasis.


Lay abstract
The abusive use of medications is a major health issue and accounts every year to huge hospitalization costs and several deaths. Analgesics are one of the first alternatives to treat fever and pain, and acetaminophen is the most frequent drug found in these formulations. It is not surprising that the cases of acetaminophen overdose are still frequent in the clinics, and since the liver is a central organ in acetaminophen detoxification, hepatocytes are directly damaged during overdose. Despite the liver ability to regenerate after injury, a massive cell death may trigger an inflammatory response that accounts to additional injury. One of the strategies to restrict organ injury is to control liver inflammation, avoiding organ failure. Here we show that ATP, a key molecule in cell bioenergetics, is also involved in liver inflammation. When cells die, they release ATP to the extracellular environment and this may directly cause additional cell hepatocyte death. These effects may be explained by ATP ability to directly cause intracellular ionic dysregulation in acetaminophen-treated cells. Mice that received an overdose of acetaminophen had significantly less liver damage when extracellular ATP actions were inhibited. Also, human-derived cells cultivated in vitro were also protected from these toxic effects when the same blockage strategy was employed. Finally, we established that liver samples from patients suffering from acute hepatitis expressed more receptors to ATP, which suggests that a similar amplifying effect happened during their disease. In this sense, we provided evidence that managing liver response to extracellular ATP released from dead cells may hold future opportunities to avoid liver failure, transplantations and death.

Background
Drug-induced liver injury (DILI) is an adverse drug reaction that causes acute hepatocyte death. There are several different grades of DILI, which range from an asymptomatic lesion (detectable only by serum transaminases analysis) to severe cases that require liver transplantation [1]. However, 20% to 50% of eligible patients die before a transplant becomes available as a result of hepatic encephalopathy and multiple organ failure [2], indicating that novel therapies aimed to control the progression of liver damage are extremely necessary. The most common cause of DILI is the overdose of acetaminophen (APAP), a popular antipyretic and analgesic drug. Following APAP administration, its reactive metabolite NAPQI (N-acetyl-p-benzoquinone imine) accumulates within hepatocytes, causing cell death mainly by oncotic necrosis [3].
When cells die under such stressing situations, their intracellular contents are spilled to the interstitium and trigger inflammation by directly causing damage to adjacent cells or activating resident cells to release pro-inflammatory mediators. In the latter case, these molecules are called damage-associated molecular patterns (DAMPs) [4]. In general, immune cells express receptors to almost all molecules that originally inhabit the intracellular compartment [5,6], but not all DAMPs are exclusively associated with immune responses. For example, cells can secrete ATP to modulate intracellular functions, including cytosolic calcium (Ca 2+ ) concentration and energetic balance [7][8][9]. However, extracellular ATP concentration significantly increases during necrosis, which in turn activates inflammasome assembling via P2X7 receptor, leading to release of IL-1β [5,10]. The inflammatory response triggered by necrosis-derived ATP was recently described as an important factor to liver injury progression, and activation of P2X7 receptor is required for manifestations of APAPinduced injury [11].
Impaired intracellular Ca 2+ management is also observed during APAP-induced hepatotoxicity, and it is closely related to the onset of cell death [12]. Moreover, intracellular Ca 2+ accumulation, particularly into the nucleus, causes DNA fragmentation by endonucleases, accelerating the progression of APAP-dependent cellular necrosis [13], indicating that molecules with ability to increase Ca 2+ signaling may cause catastrophic consequences to APAP-challenged cells. ATP induces Ca 2+ mobilization from intracellular stocks [14] and also by opening Ca 2+ permeable channels in the membrane via P2 receptors [15]. Taking into account the increased extracellular ATP concentration found during necrosis, we hypothesized that excessive interstitial ATP might contribute to liver injury progression not only via immune system stimulation, but also by worsening intracellular Ca 2+ imbalance observed during APAP administration, acting as a direct cytotoxic DAMP.

Results
Acetaminophen-induced liver damage, but not remote lung inflammation, is dependent on extracellular ATP signaling Previous data from our group showed that ATP is released following liver necrosis [16]. Initially, the participation of extracellular ATP in APAP-induced liver injury was investigated by liver confocal intravital microscopy as previously described [16,17]. Control mice presented a fully perfused liver microvasculature, as shown by the regular staining of sinusoids by phycoeritrin (PE)-coupled anti-CD31 ( Figure 1A; red channel; Control), and a few neutrophils were found within sinusoids (green channel; Additional file 1: Video 1). However, marked liver necrosis and increased neutrophil infiltration (an indicative of liver inflammation) were observed following 24 hours of APAP administration (500 mg/Kg), revealing also large non-perfused areas (poorly stained by PE-anti-CD31, Figure 1A; APAP; Additional file 2: Video 2). Histopathology analysis (H&E stained slides) confirmed liver necrosis induced by APAP treatment ( Figure 1B; control in comparison to APAP; arrow heads), which was diminished following extracellular ATP metabolism by exogenous ATPase (Figure 1B; Apy; apyrase grade VII; 25 U/mice; 24 h). In addition, apyrase significantly reduced liver injury (assessed by serum levels of ALT; Figure 1C) and neutrophil infiltration ( Figure 1A; Apy; Additional file 3: Video 3). After APAP administration, increased serum levels of pro-inflammatory cytokines (including TNF-α and IL-1β) were observed, which were significantly reduced by apyrase treatment (Figure 1D-E). Likewise, blockage of P2X (TNP-ATP; 1 mg/Kg; 24 h) or P2X7 (oxi-ATP; 9 mg/Kg; 24 h) ( Figure 1F) caused significant reduction in serum levels of ALT and liver injury, which was not reproduced in vivo by selective P2Y receptor antagonism (reactive blue-2; 10-100 mg/Kg; 24 h -data not shown). All pharmacological strategies directed to dampen extracellular ATP signaling, including cleavage by apyrase or different P2 receptor antagonists (TNP-ATP and oxi-ATP), reduced liver inflammation, necrosis (as assessed by histological score from H&E slides; Additional file 4: Figure S1), and neutrophil infiltration ( Figure 1G). Acute liver injury led to remote lung inflammation ( Figure 2A; control in comparison to APAP), with concomitant pulmonary leukocyte accumulation ( Figure 2B). Leukocytes recovered from BAL were predominately macrophages ( Figure 2C). While dampening ATP sensing resulted in significantly less liver damage and inflammation, no detectable effects on pulmonary injury (Figure 2A) or lung leukocyte infiltration ( Figure 2B-C) were observed.

HepG2 cells release ATP following acetaminophen incubation
To expand our in vivo findings, we established an in vitro model of APAP cytotoxicity using both primary mouse hepatocytes and a human lineage of hepatocytic GFP-expressing neutrophils (in green). Mice were treated with acetaminophen (500 mg/Kg; 24 h) and a group received apyrase (25 U/mice) 2 hours after APAP treatment. Control mice were wild type C57 or Lysm-eGFP (in intravital microscopy studies). No significant differences regarding liver injury were observed between C57 and C57-Lysm-eGFP expressing mice. Note the reduced liver injury and neutrophil recruitment in apyrase-treated mice. (B) H&E stained liver sections confirmed liver injury and partial protection promoted by extracellular ATP metabolism by apyrase. (C) Circulating levels of liver transaminase (ALT) and (D-E) pro-inflammatory cytokines (TNF-α and IL-1β). (F) Serum ALT levels of APAPchallenged mice treated with different purinergic receptors antagonists TNP-ATP (selective P2X, 1 mg/Kg) and oxi-ATP (selective P2X7, 9 mg/Kg). * P < 0.05 in comparison to control group and ** in comparison to vehicle-treated group. N = 5/group. Data are mean ± SEM. Scale: 100 μm. Figure 3A; 5-20 mM; 24 h), which started in the 6 th hour post-incubation, reaching 50% of cytotoxicity following 24 hours ( Figure 3B; assessed by MTT metabolism). Coincubation with the standard APAP antidote N-acetyl cysteine significantly prevented APAP effects on HepG2 cells, confirming that the major cause of cytotoxicity in our in vitro model was due to APAP bioactivation to toxic metabolites (data not shown). APAP-mediated cytotoxicity was also evaluated by an alternative cell viability test (ethidium bromide/acridine orange staining). Incubation with APAP (20 mM; 24 h) caused significant reduction in the number of viable cells in comparison to controls ( Figure 3C-E). Therefore, these settings were chosen to all subsequent in vitro experiments.

cells (HepG2 cells). APAP incubation caused cell death (
To prove that stressed/necrotic hepatocytes could be a relevant source of extracellular ATP, we incubated HepG2 cells with APAP and measured ATP and ADP concentrations in the supernatant by high-performance liquid chromatography (HPLC). Supernatant from untreated cells had undetectable amounts of both ATP and ADP. However, following APAP incubation, extracellular ATP concentration rapidly increased, while ADP was detected in later timepoints ( Figure 3F). Interestingly, cleavage of extracellular ATP reverted APAP cytotoxicity ( Figure 3G), which was reproduced by unspecific blockage of P2 receptors by suramin (0.01-0.1 mM; 24 h) ( Figure 3H). Selective participation of different P2 receptors subfamilies was also investigated. Specific P2X receptors antagonism by PPADS (0.1-100 μM; 24 h) or TNP-ATP (0.1-100 μM) also partially prevented APAP cytotoxicity, as well as selective blockage of P2Y receptors by reactive blue-2 (3-30 μM) ( Figure 3H). While selective P2X7 blockage (by oxi-ATP) prevented APAP hepatotoxicity in vivo, no detectable effects were observed in vitro using a large dose range (10-100 μM; 24 h) (Additional file 5: Figure S2). These data suggest that not a specific subtype, but several ATP and ADP receptors may be involved in extracellular purinergic signaling during necrosis, and their combined stimulation may enhance cell death independently of the immune system activation.
In this context, we hypothesized that ATP might be directly harmful in concentrations that are biologically relevant. In fact, in the same titers found in medium recovered from APAP-challenged cells, both ATP and ADP were directly cytotoxic to "naïve" HepG2 cells (10 μM, Figure 3I, Additional file 4: Figure S1). ATPmediated toxicity was detected in early timepoints (6 h), persisting until the end of incubation period (24 h, Figure 3I), which was also confirmed by ethidium bromide/acridine orange staining ( Figure 3J; control in comparison to APAP). Subsequently, we examined whether the effects observed following ATP incubation were derived only from its metabolism to ADP. For this, we used a non-hydrolysable ATP (ATP-γ-S) in the same dose range and confirmed that this stable analogue was equally able to deflagrate cell death ( Figure 3I).
Increased intracellular Ca 2+ availability underlies the mechanisms by which ATP/P2 receptors activation contribute to APAP-mediated cell death To investigate the biological relevance of extracellular purines during necrosis, we measured the Ca 2+ signal amplitude by confocal microscopy when HepG2 cells were challenged with ATP or ADP in the same concentration range found during APAP-mediated cytotoxicity (10 μM). Following ATP or ADP administration, marked intracellular Ca 2+ signals were observed in comparison to controls, which were sustained and repeated throughout the incubation time ( Figure 4A). Strikingly, "naïve" HepG2 cells that received supernatant collected from APAP-treated cells ( Figure 4B) also displayed enhanced intracellular Ca 2+ signal, which was completely abolished by incubation with apyrase or unspecific P2 receptor antagonist suramin ( Figure 4C). These data suggest that such extracellular concentrations of ATP and ADP may play an important role in intracellular Ca 2+ mobilization in during APAP incubation. Next, we investigated if acetaminophen challenge was also able to modify cell responsiveness to extracellular ATP released during necrosis. For this, we cultured HepG2 cells in the presence or absence of APAP (20mM) for 24 hours ( Figure 4D). Original medium was removed and intracellular Ca 2+ signal was triggered by the same extracellular ATP concentration found during APAP treatment (10 μM). Following ATP administration, the remaining viable APAP-treated cells displayed higher, sustained and repeated Ca 2+ signals ( Figure 4E-F, Additional file 6: Video 4), while control cells had lower intracellular Ca 2+ increase and rapidly returned to baseline values ( Figure 4E-F, Additional file 7: Video 5). Additionally, we observed that not only APAP-treated cells were hyper-responsive to ATP, but the majority of the reactive cells reached higher fluorescence values than untreated cells, indicating a both quantitative and qualitative change in Ca 2+ dynamics during APAP incubation ( Figure 4G).
To confirm that unbalanced calcium signal was involved in APAP cytotoxicity, we treated HepG2 cells with an intracellular Ca 2+ scavenger (BAPTA-AM; 1 nM) throughout APAP challenge. Incubation with BAPTA-AM reduced APAP-mediated cell death in 50% ( Figure 4H), suggesting that increased intracellular calcium availability contributed to APAP-mediated cell death.
Primary mouse hepatocytes also developed hyperresponsiveness to extracellular ATP following APAP exposure, which is prevented by P2 receptor antagonism To validate our data obtained in HepG2 cells, we obtained primary hepatocytes from mice ( Figure 5A) and incubated with different APAP and ATP doses (5-40 mM and 10-100 μM, respectively). As shown in Figure 5B, APAP incubation decreased cell viability (18 h) in a dose dependent manner. Incubation with APAP (20 mM) caused 50% of cell death, and this dose was chosen for subsequent experiments. Following exogenous ATP administration in the same dose found during necrosis (10 μM), naïve mouse hepatocytes incubated with APAP (20 mM; 6 h) increased intracellular calcium signal, and returned to baseline values after 60 seconds ( Figure 5C and 5G; Additional file 8: Video 6). However, APAP-treated primary hepatocytes presented prolonged (Additional file 9: Video 7) or repeated intracellular calcium signal ( Figure 5D and 5G; Additional file 10: Video 8), reaching higher fluorescence values in both cases. APAP-treated primary hepatocytes remained responsive to exogenous ATP for longer periods (~200 seconds; 3-fold increase; Figure 5D) in comparison to controls (~70 seconds, Figure 5C). ATP-triggered calcium increase was totally abrogated when APAPchallenged cells were treated with a P2R blocker (suramin; Figure 5E; Additional file 11: Video 9), suggesting that, analogous to HepG2 cells, APAP incubation also caused hyper-responsiveness to extracellular ATP in primary mouse hepatocytes. Moreover, incubation of primary hepatocytes with exogenous ATP in the dose range found during necrosis (10-100 μm, 18 h) significantly reduced cell viability, confirming that in biological relevant concentrations extracellular ATP may be also directly cytotoxic to mouse liver cells ( Figure 5H).

Increased expression of several purinergic receptors may explain the hyper-responsiveness to ATP during necrosis
In order to elucidate the mechanisms involved in the increased sensitivity to extracellular purines displayed by APAP-treated cells, we measured the expression of several purinergic receptors. Quantitative PCR analysis revealed that numerous P2 receptors (P2X1, P2X2, P2X7, P2Y2, P2Y4) and ectonucleotidases (NTPDase 1 and 6) were upregulated during APAP incubation in comparison to controls ( Figure 6A), which could be one of the reasons for the elevated Ca 2+ signal triggered by ATP in APAPincubated cells. Taking into account the detrimental effects of an exacerbated extracellular ATP signaling during necrosis, we hypothesized that the elevated expression of NTPDases might consist in a mechanism for ATP signaling restriction and cell protection, driving extracellular ATP metabolism ultimately to adenosine. In fact, in vitro incubation with adenosine was not only innocuous to HepG2 cells ( Figure 6B), but also partially reverted APAP-mediated cytotoxicity ( Figure 6C). Corroborating this protective loop, an increased expression of adenosine A2a receptor (A2aR) was observed in APAP-challenged cells. In fact, A2aR is described to reduce inflammatory reactions and accelerate healing when binding to adenosine [18]. Such protective profile was also confirmed in vivo, since blockage of adenosine receptors with an unspecific P1 antagonist (theophylline) worsened APAP-mediated liver injury ( Figure 6D).
To validate our findings in humans, we investigated the profile of purinergic receptors expression in patients diagnosed with drug-induced acute hepatitis. In agreement, several P2 receptors were also upregulated in acute hepatitis patients in comparison to healthy donors ( Figure 6E), suggesting that altered cell responsiveness to extracellular purines may be also relevant in the context of liver injury progression. In addition, enhanced purinergic receptor expression (P2X2, P2X7 and P2X1; Figure 6F) was correlated with higher grades of liver injury (assessed by serum ALT levels) in several acute hepatitis patients (Pearson's correlation; r).

Discussion
Extracellular ATP is a well-characterized damage-associated molecular pattern (DAMP), which activates the NLRP3 inflammasome via P2X7 receptors, inducing production of inflammatory cytokines, including interleukin-1β [5,11,19,20]. The majority of in vitro studies commonly use elevated concentrations of ATP (ranging from 1 to 10 mM) and short incubation regimes to induce cell activation, particularly leukocytes [21][22][23]. However, these conditions are unlikely to represent any in vivo environment found under either physiological or pathological circumstances [24]. For instance, extracellular ATP concentration increases during inflammatory responses, but it reaches maximum concentrations in the range of hundred micromolars [25]. So far, the damage exerted by increased extracellular concentrations of ATP has been solely attributed to indirect, inflammatory effects over immune cells [11,26]. In this study, we showed that ATP and ADP have direct cytotoxic effects in hepatic cells and may have profound influence in the pathogenesis of acute liver failure. We provided evidence that ATP signaling during necrosis might have a double-faceted action by i) enhancing inflammatory response via IL-1β release and ii) directly causing hepatotoxicity due to a hyper-responsiveness behavior to ATP and increased intracellular Ca 2+ availability.
In this study, we used a murine model of APAP poisoning to show that activation of several P2 receptors is detrimental during liver injury progression. Confocal intravital microscopy revealed remarkable changes in the liver environment during APAP overdose that were dependent on extracellular ATP signaling. Also, harmful effects of P2 receptor agonists seems to be restricted to the liver, since blockage of ATP/ADP sensing was not effective in preventing remote lung injury. In fact, despite the hepatoprotection and the reduced levels of circulating cytokines promoted by apyrase treatment, DAMPs released probably from remote injury were sufficient to support lung inflammation [17]. In line with this interpretation, recent work demonstrated that increased serum levels of ATP found during hepatectomy were only transient (~5 minutes) and rapidly returned to baseline values [27]. These data indicate that while extracellular ATP and ADP play a key role in local liver injury, they are not mediators of remote inflammatory responses.
Also, we performed a series of in vitro experiments to determine if hepatocytes could constitute a relevant source of extracellular ATP. Subsequently to APAP incubation, higher concentrations of ATP and ADP were found in the culture medium recovered from stressed/ necrotic HepG2 cells in comparison to controls. Taking into account that these titers were sufficient to directly induce HepG2 death, we postulated that ATP released from a suffering or necrotic cell binds to P2 receptors in a challenged, neighboring cell and increases its intracellular Ca 2+ concentration in sufficiently high levels to accelerate cell damage or even death. It is possible that off-target factors, including alterations in probe hydrolysis or intracellular pH variations could contribute to Fluo-4AM fluorescence variations observed in the current study. However, these factors possibly had a minor impact in our results, as we were able to significantly abrogate ATP-mediated increase in intracellular calcium signal with P2R antagonist treatment. Therefore, our data suggest that the main pathway involved in altered cell responsiveness in our model is related to P2R signaling. In fact, paracrine communication between hepatocytes mediated by ATP was previously reported [28], and blockage of cell-cell interaction via gap junctions may hold opportunities to restrict liver damage during APAP overdose [29]. Also, recent data have confirmed that extracellular ATP may enhance APAPmediated liver damage in vivo via activation of P2X7 and P2Y2 [30,31]. Thus, we suggest that necrotic/suffering hepatocytes may efficiently supply extracellular ATP to fuel both immune system activation and direct hepatotoxicity. Likewise, in vitro effects were not restricted to single receptor activation (e.g. P2X7), but rather mediated by different P2 receptors subfamilies, suggesting that larger spectrum inhibition might be necessary to promote appreciable cytoprotection. ATP induces intracellular signals by mobilizing Ca 2+ from cytosolic and nuclear stocks [14] or allowing its influx across the membrane [32]. Interestingly, in situations of deprivation of mitochondrial ATP, as seen throughout APAP challenge, cells may upregulate P2 receptors to increase intracellular Ca 2+ and stimulate ATP synthesis within mitochondria as a tentative to escape from an irreversible damage [7][8][9]. Accordingly, both primary mouse hepatocytes and HepG2 cells developed a hyper-responsive behavior to exogenous ATP following APAP administration, which may be correlated receptors were compared to serum ALT levels from healthy and acute hepatitis patients. Note that increased P2R expression was correlated with higher serum ALT levels in several patients. Pearson's correlation was calculated comparing fold increase of P2R expression to serum ALT levels (r). * P < 0.05 in comparison to control group and ** in comparison to vehicle treated group.
with increased expression of different purinergic receptors. Therefore, we propose that such change in P2R levels observed in both APAP-treated cells and in liver biopsies from acute hepatitis patients may be a strategy to restrict cell suffering through P2 receptors-mediated intracellular Ca 2+ increase. However, APAP challenge may disturb important intracellular mechanisms that manage Ca 2+ compartmentalization, since challenged cells displayed elevated intracellular Ca 2+ signal in our in vitro model. In this context, cytoprotection promoted by intracellular calcium sequestration (using BAPTA-AM) provided interesting insights for future investigations focused on elucidate the most important calcium sources that extracellular ATP recruits to boost APAP-triggered hepatotoxicity. It is worth taking into consideration that although we observed P2 receptor upregulation in HepG2 cells following APAP challenge, it is conceivable that purinergic signaling in other cell types (i.e. liver resident cells and infiltrating leukocytes) is also involved in the development of acute liver failure caused by APAP administration. Furthermore, although we confirmed APAP cytotoxicity by two different methodologies, alterations in cell metabolic pathways following APAP overdose (i.e. mitochondrial activity and cell proliferation rate) may also contribute to the mechanisms involved in APAP cytotoxicity.

Conclusions
We provided novel evidence that within the liver environment ATP acts not only as an immune system activator, but also as a direct cytotoxic DAMP by increasing intracellular Ca 2+ concentration. Also, while upregulation of P2 receptors may consist in a physiological strategy to regulate cell functions, the hyper-responsiveness behavior to biologically relevant concentrations of ATP accounted to additional hepatotoxicity (Figure 7). These findings have clear implications for liver disease pathogenesis and therapy, providing rationale to alternative pharmacological approaches for management of acute hepatitis.

Methods
Mice C57BL/6 mice were from Centro de Bioterismo in UFMG (Brazil). Lysm-eGFP mice were donated by Dr. Paul Kubes (University of Calgary, Canada). All procedures were approved by Animal Care and Use Committee in UFMG (CEBIO n°051/2011). The investigation conformed to the standards of Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication 85-23, 1996 revision).

Model of APAP-induced liver injury
Mice were fasted for 15 hours before oral APAP administration (500 mg/kg; Sigma, USA) or vehicle (warm sterile saline). After different time-points, mice were anesthetized and euthanized for blood (serum), liver, BAL (bronchial-alveolar lavage) and lung harvesting. ALT determination was performed using a kinetic test (Bioclin, Brazil) and cytokines and chemokines were quantified by ELISA kits (R&D systems, USA) both in serum and tissues. Fragments of liver and lung were fixed in formalin and sectioned for histology (H&E). Histological score was assessed by an experienced pathologist, in which 0: no lesion present; 1/2: individual necrotic cells seen at the first cell layer adjacent to the central vein, and hyaline degeneration present; 1: necrotic cells extending two or three cell layers from the central veins; 2: necrotic cells extending three to six cell layers from the central veins, but limited in peripheral distribution; 3: the same as 2, but with necrosis extending from one central vein to another; 4: more severe than 3, with extensive centrilobular necrosis throughout the section. A final score was given for each liver section [33]. Serum mitDNA (cytochrome C primer) was estimated by Real-Time PCR as previously described [34]. Neutrophil infiltration into the liver was estimated by the myeloperoxidase activity assay (MPO) [35]. In a separated set of experiments, the liver was imaged using confocal intravital microscopy as described previously [16]. Three-dimensional video reconstructions were made using confocal Z-stacks (40 μm; 1 μm step) and mounted using Volocity software (Perkin-Elmer, USA). BAL was collected for leukocyte counting [36]. All experimental groups included N ≥ 5.

In vitro HepG2 assays
HepG2 (American Type Culture Collection) cells were maintained at 37°C under an atmosphere of 5% CO2 in complete RPMI1640 medium containing 10% FBS and cultured in 10 5 cells/well in 96 wells plates. After 24 hours of incubation the supernatant was replaced by medium without FBS containing the treatments [38].  Figure  S2). Cell viability was assessed by MTT metabolism assay (Sigma, USA) or by ethidium bromide/acridine orange viability assay [39]. Purinergic receptors and ectonucleotidases expression was quantified by Real-Time PCR as described previously [40,41]. Experiments were repeated at least three times, using 12 replicates per group.
Primary mouse hepatocytes (PMH) isolation and culture PMH were isolated as previously described [42,43]. Briefly, the portal vein was cannulated, and a solution collagenase B (4 mg/ml; Roche; from C. histolyticum) was perfused during 8 minutes (5 ml/min). Following Figure 7 Proposed mechanism: 1: Under physiological conditions, extracellular ATP regulates several intracellular signaling pathways, which involves also calcium compartmentalization. 2-3: Acetaminophen incubation directly causes hepatocyte necrosis, calcium imbalance and further ATP release. 4: In parallel, challenged viable hepatocytes up-regulate several purinergic receptors, probably as a regulatory homeostatic strategy, causing ATP hyper-responsiveness. Binding of extracellular ATP to purinergic receptors increases intracellular Ca 2+ and pulses, which accounted to additional cell necrosis, reverberating APAP-induced death. Dampening of extracellular ATP signalling or reducing intracellular Ca 2+ availability significantly reduced hepatocyte necrosis. Data from FHF patients suggest that a similar necrosis-amplification pathway may be involved in organ injury progression. initial digestion, liver was separated from the diaphragm and stomach, and carefully minced in a plastic Petri dish in Williams's medium and the resultant cells were double filtered in serial nylon mesh filters. The cell suspension was washed twice with William's medium and ressuspended in William's medium containing 50 U of penicillin and 50 mg of streptomycin. This protocol yielded a high rate of viable hepatocytes (>95% as assessed by commercial fast hematological staining kit and Trypan blue exclusion). Cells were then seeded on collagen I-treated glass coverslips (3 × 10 5 cells/well) and cultured using 6-wells plates to further MTT and calcium signal assays (as described previously for HepG2 cells).

Analysis of extracellular nucleotides by HPLC
Cells were incubated in presence or absence of APAP and supernant was recovered in different timepoints. Aliquots of 40 μL were applied to a reverse phase HPLC system using a C18 Shimadzu column (Shimadzu, Japan) with absorbance measured at 260 nm. The mobile phase was 60 mM KH 2 PO 4 , 5 mM tetrabutylammonium chloride, pH 5.9, in 15% methanol. Retention times were assessed using standard samples of ATP and its metabolites.

Detection of intracellular Ca 2+ signals with confocal microscopy
Nuclear and cytosolic Ca 2+ were monitored in individual cells by using time-lapse confocal microscopy, as described [14]. Cells were incubated with 4 μM cell permeant fluo-4-AM (fluo-4 acetoxymethyl ester; Molecular Probes) and all fluorescence analyses were performed offline using ImageJ software (NIH) as previously described [14].

Human patients
Fifteen patients diagnosed with non-viral and suspected of drug-induced acute liver failure addressed to the Liver Clinic of Hospital Federal de Bonsucesso of Rio de Janeiro were enrolled for this study (

Additional files
Additional file 1: Video 1. Three-dimensional Z-stack rendering from liver confocal intravital microscopy -Control mouse. Lysm-eGFP mice were used to visualize neutrophil infiltration and sinusoidal perfusion. Sinusoids were stained by i.v. injection of PE-coupled anti-PECAM-1. Zstacks were made (depth: 40 μm) and TIFF-acquired images were mounted by using Volocity software (NIH, USA). Mice were imaged by sequential laser scans (2.71 seconds).
Additional file 4: Figure S1. H&E slides and histological score from mice treated with different P2R antagonists and challenged with APAP. Mice were treated (2 hours after APAP challenge; 500 mg/Kg; 24 h) with apyrase (25 U/mice), TNP-ATP (1 mg/Kg; i.v.), oxidized-ATP (oxi-ATP; 9 mg/ Kg; i.v.) or reactive blue-2 (10-100 mg/Kg; i.p.). Histological score was assessed using 0: no lesion present; 1/2: individual necrotic cells seen at the first cell layer adjacent to the central vein, and hyaline degeneration present; 1: necrotic cells extending two or three cell layers from the central veins; 2: necrotic cells extending three to six cell layers from the central veins, but limited in peripheral distribution; 3: the same as 2, but with necrosis extending from one central vein to another; 4: more severe than 3, with extensive centrilobular necrosis throughout the section. A final score was given for each liver section. * -P < 0.05 in comparison to control (C) group and ** in comparison to vehicle treated group. Data are mean ± SEM. Scale: 100 μm.