Chemokines and mitochondrial products activate neutrophils to amplify organ injury during mouse acute liver failure


  • Potential conflict of interest: Nothing to report.

  • See Editorial on Page 1599


Acetaminophen (APAP) is a safe analgesic and antipyretic drug. However, APAP overdose leads to massive hepatocyte death. Cell death during APAP toxicity occurs by oncotic necrosis, in which the release of intracellular contents can elicit a reactive inflammatory response. We have previously demonstrated that an intravascular gradient of chemokines and mitochondria-derived formyl peptides collaborate to guide neutrophils to sites of liver necrosis by CXC chemokine receptor 2 (CXCR2) and formyl peptide receptor 1 (FPR1), respectively. Here, we investigated the role of CXCR2 chemokines and mitochondrial products during APAP-induced liver injury and in liver neutrophil influx and hepatotoxicity. During APAP overdose, neutrophils accumulated into the liver, and blockage of neutrophil infiltration by anti–granulocyte receptor 1 depletion or combined CXCR2-FPR1 antagonism significantly prevented hepatotoxicity. In agreement with our in vivo data, isolated human neutrophils were cytotoxic to HepG2 cells when cocultured, and the mechanism of neutrophil killing was dependent on direct contact with HepG2 cells and the CXCR2-FPR1–signaling pathway. Also, in mice and humans, serum levels of both mitochondrial DNA (mitDNA) and CXCR2 chemokines were higher during acute liver injury, suggesting that necrosis products may reach remote organs through the circulation, leading to a systemic inflammatory response. Accordingly, APAP-treated mice exhibited marked systemic inflammation and lung injury, which was prevented by CXCR2-FPR1 blockage and Toll-like receptor 9 (TLR9) absence (TLR9−/− mice). Conclusion: Chemokines and mitochondrial products (e.g., formyl peptides and mitDNA) collaborate in neutrophil-mediated injury and systemic inflammation during acute liver failure. Hepatocyte death is amplified by liver neutrophil infiltration, and the release of necrotic products into the circulation may trigger a systemic inflammatory response and remote lung injury. (HEPATOLOGY 2012;56:1971–1982)

Acetaminophen (APAP) is a widely used over-the-counter analgesic and antipyretic drug used to relieve symptoms of mild inflammatory conditions. However, after accidental or intentional overdose, accumulation of the reactive metabolite, N-acetyl-p-benzoquinone-imine, in hepatocytes leads to intracellular oxidative stress, mitochondria dysfunction, and DNA damage.1 The combination of these phenomena culminates in generalized oncotic necrosis and liver dysfunction. In fact, APAP poisoning is the most frequent etiology of drug-induced liver injury (DILI), which can progress to acute liver failure (ALF), a disorder associated with high mortality and hospitalization costs. If treatment with the antidote, N-acetyl-cysteine, and intensive care support fail to allow for spontaneous recovery, the only effective treatment is liver transplantation (LT).2 However, even in those patients that fulfill the required transplantation criteria,3 more than 20% will die in 1 year after LT, and half of the patients will succumb in the next 10 years.4 In this sense, interventions intended to restrict organ damage may stop or prevent the progression to liver failure, allowing the recovery of organ function, which may ultimately reduce the necessity of liver transplants.

It is reasonable to consider that the total amount of organ injury after APAP overdose may be predicted as a summation of drug-induced cytotoxicity plus any additional cell death promoted by the inflammatory response to necrosis.5 In fact, necrosis-derived products, including adenosine triphosphate, mitochondrial DNA (mitDNA; unmethylated CpG motifs), and formyl peptides, may activate their innate immunity receptors in resident and parenchymal cells (e.g., P2X7, Toll-like receptor 9 [TLR9], and formyl peptide receptor 1 [FPR1], respectively), amplifying the inflammatory response.6 Notably, during severe organ injury, release of necrotic products to the systemic circulation initiates a “septic-like” status and remote organ injury, which was previously observed in patients after severe trauma.7 In this context, a similar situation might be occurring during acute liver injury, which would explain the systemic inflammatory response that ALF patients display.2 Accordingly, it was recently described that fragments of genomic DNA (gDNA) and mitDNA are released into the circulation as an indication of cell death,8 suggesting that necrosis products may reach remote organs through the systemic circulation.

Different chemotatic mediators choreograph the complex voyage of neutrophils to sites of inflammation. We have demonstrated previously that an intravascular gradient of CXC chemokine receptor (CXCR)2 chemokines and mitochondria-derived formyl peptides collaborate to precisely guide neutrophils to necrotic tissues.9 Despite the growing body of evidence that inappropriate neutrophilic inflammation contributes to the pathogenesis and intensification of several diseases, including gout,10 arthritis,11 and reperfusion injury,12 there are no established clinical therapeutic protocols aimed in controlling liver neutrophil influx and activation after acute hepatocyte necrosis, and the few clinical trials performed using corticosteroids have not encouraged their use in patients.13, 14 In the present work, we investigated the mechanisms involved in neutrophil accumulation into the liver during APAP overdose and how they could contribute to injury amplification. In addition, we aimed to elucidate whether necrosis-derived products are released into the circulation after ALF in mice and humans, hence leading to systemic inflammatory response and remote organ injury. In this context, the development of therapies aimed at restricting the additional immune-mediated injury may offer novel, effective treatment for drug-induced liver injury (DILI).


ALF, acute liver failure; ALT, alanine aminotransferase; ANOVA, analysis of variance; APAP, acetaminophen; BAL, bronchoalveolar lavage; CXCR2, CXC chemokine receptor 2; DCF-DA, 2′,7′-dichlorodihydrofluorescein diacetate; DILI, drug-induced liver injury; ELISA, enzyme-linked immunosorbent assay; FPR1, formyl peptide receptor 1; gDNA, genomic DNA; GR1, granulocyte receptor 1; IgG, immunoglobulin G; IL-1β, interleukin-1 beta; I/R, ischemia/reperfusion; IV, intravenous; IVM, intravital microscopy; LT, liver transplantation; Lysm-eGFP, lysozyme M promoter for enhanced green fluorescent protein; mitDNA, mitochondrial DNA; MPO, myeloperoxidase; nDNA, nuclear DNA; NO, nitric oxide; PCR, polymerase chain reaction; PE, phycoerythrin; PECAM-1, platelet-endothelial cell adhesion molecule-1; ROS, reactive oxygen species; SEM, standard error of the mean; TLR9, Toll-like receptor 9; TNF-α, tumor necrosis factor alpha; UFMG, Universidade Federal de Minas Gerais.

Materials and Methods


C57BL/6 and TLR9−/− mice were from Centro de Bioterismo in Universidade Federal de Minas Gerais (UFMG; Belo Horizonte, Brazil). Lysozyme M promoter for enhanced green fluorescent protein (Lysm-eGFP) mice were donated by Dr. Paul Kubes (University of Calgary, Calgary, Alberta, Canada). All procedures were approved by the Animal Care and Use Committee at UFMG (CEBIO no.: 051/2011). The investigation conformed to the standards of the 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-Aldrich, St. Louis, MO) or vehicle (warm sterile saline). After different time points, mice were anesthetized and euthanized for blood (serum), liver, bronchoalveolar lavage (BAL), and lung harvesting. Alanine aminotransferase (ALT) determination was performed using a kinetic test (Bioclin, Belo Horizonte, Brazil), and cytokines and chemokines were quantified by enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems, Minneapolis, MN) both in serum and tissues. Fragments of liver and lung were fixed and sectioned for histology (hematoxilin and eosin). Serum mitDNA was estimated by real-time polymerase chain reaction (PCR), as previously described.7 Neutrophil infiltration into the liver and lung was estimated by the myeloperoxidase (MPO) activity assay, standardized to the number of neutrophils obtained from the peritoneal cavity of casein-injected mice.10 In a separated set of experiments, the liver was imaged using confocal intravital microscopy (IVM), as described previously.9 Mice were imaged during 1 hour, and the whole video was accelerated to facilitate cell-movement visualization. BAL was collected for leukocyte counting as well as protein and nitrite quantification.15 All experimental groups included N ≥ 5.

In Vitro HepG2 and Neutrophil Culture.

HepG2 cells were maintained at 37°C under an atmosphere of 5% CO2 in complete RPMI 1640 medium containing 10% fetal bovine serum and cultured in 105 cells/well.16 APAP (5 mM), BOC-1 (10−5 M; MP Biomedicals LLC, Solon, OH), and DF2156a (10−6 M; Dompé Pharmaceutical, Milan, Italy) were dissolved in dimethyl sulfoxide, then in culture medium, and then incubated throughout the experiments. Cell viability was assessed by tetrazolium (Sigma-Aldrich) metabolism assay or by the trypan blue exclusion test (in neutrophil viability tests). Neutrophils were harvested from healthy donors and purified by dextran sedimentation and Ficoll gradient, as described elsewhere.17 Coculture assays were performed by adding neutrophils in different concentrations (102-105 cells/well) to a fixed HepG2 number (105 cells/well). In transwell experiments, HepG2 cells were cultured in a 5 × 105 cells/well density in 24-well plaques (0.4 μm/pore; Corning Incorporated, Corning, NY) separated from neutrophils. Reactive oxygen species (ROS) production ex vivo was measured using 2′,7′-dichlorodihydrofluorescein diacetate (DCF-DA), as described previously.10 Briefly, neutrophils (105 cells/well) were incubated with 10 μM of DCF-DA for 30 minutes at 37°C, and fluorescence was determined in a fluorometer (Synergy 2; BioTek, Winooski, VT) at an excitation wavelength of 485 nm and emission at 530 nm. Data are expressed as mean fluorescence intensity mean ± standard error of the mean (SEM).

In Vivo Drugs and Treatments.

BOC-1 (2 mg/kg, intravenous; IV), DF2156a (30 mg/kg, per os), or the respective vehicles were given 1 hour before APAP. Anti-GR1 antibody (RB6-8C5, anti-Ly-6G, IV; eBioscience, San Diego, CA) and immunoglobulin G (IgG) isotype control (Rat IgG; Sigma-Aldrich) were given at 24 and 4 hours before APAP challenge at the dose of 100 μg/mouse. Neutrophil depletion was confirmed by specific leukocyte counting under light microscopy. Liver sinusoids were fluorescently stained by IV injection of phycoerythrin (PE)-coupled anti-PECAM-1 (platelet-endothelial cell adhesion molecule-1) (8 μg/mice, anti-CD31; eBioscience).

Human Subjects.

Fifteen patients diagnosed with nonviral and suspected of drug-induced ALF addressed to the Liver Clinic of Hospital Federal de Bonsucesso of Rio de Janeiro were enrolled for this study. Peripheral blood samples were obtained from 5 healthy controls for the experiments. Serum CXCL8 was quantified by ELISA kits (R&D Systems), and mitDNA and nuclear DNA (nDNA) were quantified by real-time PCR, as described previously.7 This study was approved by institutional review boards (CEP-Fiocruz 22/03) and performed under an informed consent by all participants. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki.

Statistical Analysis.

Statistical analyzes were performed using one-way analysis of variance (ANOVA) (Tukey's post-test) and the Student t test. P values less than 0.05 were considered statistically significant. All data are presented as mean ± SEM.


Neutrophils Are Recruited to the Liver and Promote Cytotoxicity During APAP-Induced Liver Injury.

To establish the relationship between neutrophil accumulation and ALF, we orally overdosed mice with APAP (500 mg/kg) to induce hepatocyte death.5 Lysm-eGFP mice (eGFP-expressing neutrophils) were anesthetized, and the liver microcirculation was visualized under confocal IVM.9 Liver sinusoids were stained with PE-coupled anti-PECAM-1. Control mice presented a completely perfused sinusoidal microvasculature (Fig. 1A; Supporting Video 1) and a small number of adhered neutrophils through the imaging time. APAP administration led to extensive liver necrosis (Supporting Fig. 1), with a progressive, time-dependent neutrophil accumulation, which was observed 6 hours after APAP administration (Fig. 1B). At this time point, neutrophils diffusely accumulated within the field of view and crawled within areas that were weakly stained by PE-coupled anti-PECAM-1 antibody (presumably necrotic areas) (Fig. 1A, white arrowheads; Supporting Video 2). After 24 hours of APAP overdose, neutrophil accumulation within liver tissue was increased (Fig. 1A,B), and at that point clusters of neutrophils were observed by liver IVM (Supporting Videos 3 and 4). Moreover, there was a cooccurrence of the peak of the serum ALT activity with the maximum values of liver neutrophil infiltration (Fig. 1B) and CXCR2 chemokine production after APAP administration (CXCL1, Fig. 1C; CXCL2, Fig. 1D).

Figure 1.

Neutrophils accumulate into liver microvasculature after APAP overdose. (A) Lysm-eGFP mice (green fluorescent neutrophils) were used to visualize liver neutrophil migration after APAP (500 mg/kg) administration. Sinusoids were stained by PE-coupled anti-PECAM-1 antibody (red). After 6-24 hours of APAP overdose, neutrophil accumulation was observed in liver tissue. White arrowheads indicate areas of low blood perfusion, which were weakly stained by anti-PECAM-1 antibody. Images are representative of three independent experiments. (B) Time course of serum ALT levels confirmed liver injury after APAP challenge. Neutrophil quantification by MPO assay revealed the cooccurrence of maximum liver neutrophil migration (24 hours) with highest levels of serum ALT in liver CXCL1 (C) and CXCL2 (D). ELISA was used to estimate CXCL1 and CXCL2 in liver extracts, and baseline (control) values are expressed as dashed line in the graph. Scale bars: (A) 100 μm and (B) 50 μm. *Indicates statistical significance in comparison to controls or baseline values (ANOVA, Tukey's post-test; P < 0.05).

To investigate whether neutrophils contribute to ALF, we immunologically depleted circulating neutrophils in mice (IV RB6-8C5 anti-GR1 antibody, 100 μg/mouse 24 hours plus 100 μg/mouse 4 hours before APAP). Neutrophil depletion was associated with reduced liver necrosis, reflected by a significant reduction in serum ALT activity (Fig. 2A), in comparison to controls (P < 0.05). To investigate whether neutrophils promote direct hepatocyte cytotoxicity, we coincubated HepG2 cells with purified human neutrophils (Fig. 2B) using different cell proportions (1:1 and 10:1 HepG2 cells/neutrophils). For that, we first verified whether freshly isolated neutrophils remained viable throughout the incubation period. We observed that at least 85%-90% of the neutrophils are viable after 24 hours of incubation (105-106 cells/well; assessed by trypan blue exclusion test; Supporting Fig. 2A). In agreement with the ability of neutrophils to injure hepatocytes in vivo, results in HepG2 coculture showed that the addition of neutrophils significantly reduced HepG2 viability (Fig. 2B). When neutrophils were separated from HepG2 cells by using a transwell coculture system (0.4 μm/pore), we observed an inhibition of neutrophil-mediated hepatocyte death (Fig. 2B; P < 0.05). To investigate whether necrotic hepatocytes may increase neutrophil killing ability, we first established an in vitro APAP-toxicity model using HepG2 cells. APAP incubation reduced HepG2 viability time and dose dependently (Supporting Fig. 3A,B), reaching 50% of cytotoxicity within 24 hours (Fig. 2C; P < 0.05). The addition of neutrophils (10:1 HepG2/neutrophil ratio) to necrotic HepG2 resulted in further cell death (Fig. 2C; P < 0.05). Measurement of ROS production by neutrophils confirmed that the activation of neutrophils by necrosis products can enhance cytoxicity, because the addition of supernatant from APAP-treated HepG2 (5 mM; incubation for 24 hours) cells significantly increased neutrophil ROS production, in comparison to the supernatant from viable HepG2 cells (incubated with medium for 24 hours; Fig. 2D).

Figure 2.

Neutrophil contribution to liver injury: direct cytotoxicity to hepatocytes leads to additional cell necrosis after APAP overdose. (A) Neutrophil depletion using anti-GR1 antibody protected mice from APAP-induced liver injury, as reflected by a significant reduction in serum ALT. **Statistical significance, in comparison to vehicle. (B) Purified human neutrophils are directly cytotoxic to HepG2 cells when incubated in a 10:1 and 1:1 proportion (HepG2 cells/neutrophils; 105 HepG2 cells/well). Neutrophil cytotoxic ability was completely dependent on neutrophil-HepG2 contact, because separation of these cells by a transwell coculture system reverted HepG2 cytotoxicity mediated by neutrophils. **Statistical significance, in comparison to direct incubation. (C) Addition of neutrophils to APAP-incubated HepG2 caused significant supplementary HepG2 death. **Statistical significance, in comparison to APAP or neutrophil incubation alone. (D) Incubation with supernatant from APAP-treated HepG2 cells induced significantly more ROS production, in comparison to control. (E) Incubation with APAP enhanced CXCL8 production by HepG2 cells. *Indicates statistical significance, in comparison to controls (C) (ANOVA, Tukey's post-test; P < 0.05).

Collaboration Between CXCR2 Chemokines and Formyl Peptides in Neutrophil Recruitment and Hepatocyte Injury.

ELISA from liver fragments revealed that CXCR2-active chemokines (e.g., CXCL1 and CXCL2) were significantly increased at 24 hours after APAP administration. This is consistent with studies showing that hepatocytes express, and are an important source of, these chemokines.18 In agreement with the latter finding, HepG2 cells spontaneously release significant amounts of CXCL8, an ortolog of murine CXCL1 and CXCL2, which were increased after incubation with APAP (Fig. 2E). DF2156a, a CXCR1/CXCR2 inhibitor,19 was used to investigate the role of CXCR2 in the migration of neutrophils into the liver. CXCR2 blockade caused a partial reduction in liver neutrophil infiltration (Fig. 3A), which was not followed by protection of liver injury (Fig. 3B).

Figure 3.

CXCR2 chemokines and formyl peptides collaborate in liver neutrophil recruitment and hepatocyte injury. Mice were overdosed with a toxic dose of APAP (500 mg/kg), and samples were collected after 24 hours. (A) Mice treated with a CXCR2 antagonist (DF: DF2156a; 30 mg/kg) had a partial reduction in liver neutrophil migration and (B) were not protected from liver injury. (C) Similarly, single blockage of FPR1 by its antagonist, BOC-1 (2 mg/kg), reduced only partially liver neutrophil infiltration, (D) with no detected protection from liver injury. (E) However, combined CXCR2-FPR1 blockage by simultaneous administration of DF2156a and BOC-1 to mice led to significant inhibition of liver neutrophil emigration and (F) liver injury protection. *Indicates statistical significance, in comparison to controls (C); **indicates statistical significance, in comparison to vehicle-treated group (ANOVA, Tukey's post-test; P < 0.05).

During necrosis, mitochondria are spilled out into the extracellular milieu, and both mitochondria-derived formyl peptides and DNA can attract and activate neutrophils through their putative receptors, FPR1 and TLR9, respectively.20 Herein, we showed that similarly to CXCR2 blockade, single pretreatment with a FPR1 antagonist (e.g., BOC-1) caused only a partial reduction in the recruitment of neutrophils to the liver after APAP challenge (Fig. 3C), and there was no significant protection against liver damage (Fig. 3D). Because we have previously shown that chemokines and formyl peptides work together to guide neutrophils to sites of necrotic inflammation,9 we used a combined strategy to block both FPR1 and CXCR2 receptors. Strikingly, a robust reduction in liver neutrophil migration (Fig. 3E; P < 0.05) was then accompanied by a significant inhibition of liver damage (Fig. 3F; P < 0.05). Liver histology analysis confirmed that lower histological scores were attributed to CXCR2-FPR1–treated mice (Supporting Fig. 1). Next, we tested whether the CXCR2-FPR1–signaling pathway was contributing to neutrophil killing ability upon HepG2 cells in vitro. Combined CXCR2-FPR1 blockage in doses that were previously shown to reduce neutrophil activation,12, 21 led to a significant inhibition of neutrophil-mediated HepG2 cytotoxicity and ROS production (Supporting Fig. 2B,C).

MitDNA and CXCL8 Are Released Into the Systemic Circulation After ALF in Mice and Humans.

Cellular injury may sustain potentially toxic levels of circulating proinflammatory mediators and necrosis-derived products.7 To verify whether chemokines and mitDNA were released into the systemic circulation during liver injury, we analyzed blood samples from patients diagnosed with fulminant hepatitis (by nonviral and nonseptic causes). Inclusion criteria demanded the development of coagulopathy (prothrombin time activity >15 seconds) or an international normalized ratio ≥1.5 as well as hepatic encephalopathy within 8 weeks of jaundice onset in the absence of preexisting liver disease (Table 1). Real-time PCR quantification for nDNA or specific mitochondrial genome sequences revealed a significant elevation in the serum levels of mitDNA in ALF patients, in comparison to healthy volunteers (Fig. 4A), which is also an indication of mitochondrial formyl peptide release.7 Furthermore, elevation in serum mitDNA levels were more conspicuous than nDNA (Fig. 4B). Fulminant hepatitis patients also had significantly higher serum levels of the neutrophil chemokine, CXCL8, in comparison to controls (Fig. 4C; P < 0.05). Parallel studies in mice proved that mitDNA was released in the systemic circulation after APAP overdose, in comparison to controls (Fig. 4D), which also was more evident than serum nuclear DNA (Fig. 4E). Serum mitDNA levels were strongly correlated with serum ALT (Fig. 4F). Chemokines (CXCL1 and CXCL2) and proinflammatory cytokines (interleukin-1 beta [IL-1β] and tumor necrosis factor alpha [TNF-α]) were also significantly increased in the sera (Fig. 5). Combined blockage of CXCR2-FPR1 significantly reduced CXCL2 and IL-1β levels (Fig. 5B,C), whereas lower serum levels of CXCL1 were found in mice treated with FPR1 antagonist (Fig. 5A). No significant changes were detected in serum TNF-α levels throughout the treatments with antagonists.

Figure 4.

Increased serum levels of mitDNA and chemokines after ALF in humans and mice. (A) Real-time PCR quantification of specific mitochondria genome sequences (NADH dehydrogenase, cytocrome B, and cytocrome C oxidase) in serum from ALF patients (circles) and healthy controls (HC; squares) revealed a significant increase in circulating levels of mitDNA in ALF patients, (B) which highly increased, in comparison to serum nDNA concentration. Also, ALF patients displayed elevated levels of serum CXCL8. (D) In parallel, mice overdosed with APAP (500 mg/kg) displayed a time-dependent increase of serum mitDNA levels, in comparison to controls (dashed line; time zero), (E) which was also more prominent than serum nDNA elevation. (F) Serum mitDNA concentration was significantly corelated with ALT levels (R2 = 0.86; Pearson's correlation). *Indicates statistical significance, in comparison to healthy controls (Student t test; P < 0.05). NADH, nicotinamide adenine dinucleotide.

Figure 5.

Serum levels of chemokines and proinflammatory cytokines during ALF in mice: effect of CXCR2 and FPR1 blockage. Mice were overdosed with a toxic dose of APAP (500 mg/kg), and blood samples were collected after 24 hours. In different groups, mice were previously treated with the CXCR2 antagonist, DF2156a (DF; 30 mg/kg), FPR1 antagonist (BOC-1; 2 mg/kg), or with a combined dose of these two antagonists (BOC-1+DF). (A) Quantification by ELISA of serum levels of CXCL1, (B) CXCL2, (C) IL-1β, and (D) TNF-α. ND, nondetected. *Indicates statistical significance, in comparison to control; **indicates statistical significance, in comparison to vehicle-treated mice. (ANOVA, Tukey's post-test; P < 0.05).

Table 1. Clinical and Biochemical Parameters of Nonviral ALF Patients
 ALF Group (n = 15)Healthy Volunteers (n = 5)
  1. Abbreviations: F, female; ND, nondetermined.

Age6-5528  25-4027  
Serum ALT (U/L)122-7,3001,331Sex (F)653-117Sex (F)60
Serum bilirubin (mg/dL)3.5-37.026.5Survived11/15NDND  
Encephalopathy grade ≥1   10/15    

ALF Triggers Remote Lung Injury: Collaboration of CXCR2 Chemokines and Mitochondrial Products.

We hypothesized that increased serum levels of several proinflammatory mediators found during ALF may cause remote organ injury. In fact, mice treated with a toxic dose of APAP displayed a marked pulmonary inflammation, with increased leukocyte infiltration, shown in lung fragments, lavage fluid, and histopathology analysis (Supporting Fig. 4). After APAP overdose, leukocytes (mainly macrophages and lymphocytes; Fig. 6A-C) emigrated to the alveolar space (collected during the BAL procedure), whereas neutrophil accumulation remained within lung parenchyma (Fig. 6D,E). Also, increased levels of chemokines (CXCL1 and CXCL2; Supporting Fig. 5A,B), cytokines (IL-1β and TNF-α; Supporting Fig. 5C,D), and nitric-oxide (NO) production (Fig. 6F) were found in lungs from APAP-treated mice. Remarkably, protection from liver injury by combined CXCR2-FPR1 was accompanied by less-remote lung inflammation. Leukocyte infiltration into the lung (Fig. 6A-E) and NO production (Fig. 6F) were completely avoided by CXCR2-FPR1 blockage before APAP administration. Interestingly, ELISA from lung fragments revealed that despite the reduced signs of inflammation shown by CXCR2-FPR1–treated mice, no differences in chemokines and cytokines levels were detected (Supporting Fig. 5).

Figure 6.

Contribution of the CXCR2-FPR1–signaling pathway to remote lung injury. Lung samples were collected from mice challenged with APAP (500 mg/kg), which were pretreated with a combined CXCR2 (DF2156a; 30 mg/kg) and FPR1 antagonist (BOC-1; 2 mg/kg). Lungs were perfused to collect BAL and lung parenchyma processed to further analysis. (A) Total leukocyte number in BAL was counted in a Neubauer chamber. (B) Macrophages, (C) lymphocytes, and (D) neutrophils were counted under light microscopy using cytospin-prepared slides, stained with a regular hematological dye kit. (E) Neutrophil accumulation into the lung parenchyma was estimated by MPO activity assay. (F) Lung NO production was quantified by BAL nitrite concentration using Griess' assay. BOC-1+DF indicates combined CXCR2-FPR1 antagonism. *Indicates statistical significance, in comparison to control; **indicates statistical significance, in comparison to vehicle-treated mice (ANOVA, Tukey's post-test; P < 0.05).

Taking into account the elevated serum levels of mitDNA during ALF, we next investigated whether mice lacking the endosomal receptor for unmethylated CpG motifs (TLR9−/−), a sensor for mitDNA, are resistant to APAP-induced inflammation. When TLR9−/− mice were challenged with APAP, we observed a significant protection from liver damage, confirmed by lower serum ALT levels (Fig. 7A), reduced liver neutrophil migration (Supporting Fig. 6F), and histopathology scores (Supporting Fig. 1). Systemic markers of inflammation were also reduced in APAP-treated TLR9−/− mice, including lower serum levels of mitDNA (Fig. 7B), chemokines (CXCL1 and CXCL2; Fig. 7C,D), and cytokines (IL-1β and TNF-α; Fig. 7E,F). In agreement, TLR9−/− had no detectable lung inflammation after APAP challenge, reflected by healthy histological findings (Supporting Fig. 4), unaltered lung-leukocyte population (Supporting Figure 6A-E), and physiological levels of pulmonary CXCL1, CXCL2, IL-1β, and TNF-α (Supporting Figure 7A-D).

Figure 7.

TLR9−/− mice are completely protected from APAP-induced liver injury and systemic inflammatory response. Mice lacking the endossomal receptor for unmethylated CpG motifs (TLR9) were overdosed with a toxic dose of APAP (acetaminophen), and blood samples were collected after 24 hours. (A) Serum ALT levels were significantly reduced in APAP-treated TLR9−/− mice, (B) which was accompanied by lower circulating levels of mitDNA (assessed by real-time PCR; NADH dehydrogenase primer). (C) Serum levels of the chemokines, CXCL1 and (B) CXCL2, as well as proinflammatory cytokines (E) IL-1β and (F) TNF-α were also reduced in TLR9−/− mice overdosed with APAP (estimated by ELISA). ND, nondetected. *Indicates statistical significance, in comparison to control; **indicates statistical significance, in comparison to wild type (WT) treated mice. (ANOVA, Tukey's post-test; P < 0.05). NADH, nicotinamide adenine dinucleotide.


After APAP overdose, chemokines and mitochondrial products collaborated in liver neutrophil migration and activation, causing additional hepatotoxicity in mice. Also, injury amplification by neutrophils contributed to the systemic inflammatory response, which elicited remote lung injury during ALF (Supporting Fig. 8). Consequentially, blockage of relevant receptors in the context of cell necrosis, namely CXCR2, FPR1, and TLR9, culminated in liver injury reduction, avoided systemic inflammatory response, and suppressed remote lung injury.

Neutrophils are the first line of defense against invading pathogens, being one of the pillars of the innate immune response.22 Also, after sterile cell death, neutrophils emigrate to sites of necrosis where they may participate in the healing and removal of cell debris.23 However, uncontrolled neutrophil migration and activation is detrimental in several inflammatory diseases, such as endotoxemic shock,24 alcoholic hepatitis,25 gout,10 arthritis,11 obstructive cholestasis,26 ischemia-reperfusion (I/R) injury,12 and several others.27 In spite of previous reports regarding the detrimental role of neutrophils in liver injury, data from distinct groups are controversial. In fact, because of the different experimental approaches used throughout the experiments (i.e., different APAP doses, mouse strains, and neutrophil depletion strategies), the findings are not accurately comparable.5, 28-30 Also, neutrophil depletion by anti-GR1 antibody injection may provide resistance to APAP toxicity independently of neutrophil circulating numbers.23 In our study, we used three different approaches to investigate neutrophil contribution to ALF: (1) neutrophil depletion by anti-GR1 antibody; (2) chemokine and formyl peptide receptor antagonism; and (3) in vitro coculture of human neutrophils with HepG2 cells. We demonstrated that CXCR2 chemokines and formyl peptides work together to guide neutrophils to sites of liver necrosis, where they can cause additional liver injury, contributing to a systemic inflammatory response and further remote organ injury. Our data are in agreement with previous findings that reported that during neutrophil chemotaxis to sites of necrosis, CXCR2 chemokines (CXCL1 and CXCL2) can guide these cells only until the area surrounding the necrosis focus, because the intravascular chemokine gradient is abruptly interrupted approximately 150 μm away from these sites.9 Once adhered within this “chemotaxis zone,” neutrophils shift their chemotatic mechanism and precisely infiltrate into the inner site of necrosis by then following a mitochondria-derived formyl peptide gradient, mainly through FPR1 signaling.9 In this way, concomitant blockage of complementary pathways of chemotaxis and activation emerges as an alternative strategy, because an almost complete inhibition of neutrophil migration or function seems to be necessary to significantly reduce liver damage.5, 30

We have shown that human neutrophils were cytotoxic to HepG2 cells, which was completely dependent on cell-cell contact, because separation by transwell inserts completely reverted neutrophil-mediated necrosis. During emigration to the liver parenchyma, activated neutrophils may need to encounter hepatocytes to promote unwanted cytotoxicity. Also, recognition of necrosis products by neutrophils may improve their cytotoxicity, because the addition of these leukocytes to necrotic hepatocytes led to more cell death and ROS production. The simplest explanation is that neutrophils may need an intimate contact with the target to exert killing behavior, generating a microenvironment in which ROS production facilitates cytotoxicity by granule contents.31 Considering liver cell populations, FPR1 is constitutively expressed by neutrophils, Kupffer cells, and hepatocytes,32 which also express CXCR2.33 In this sense, blockage of these receptors in vivo may also directly affect these cells and may account partially for the observed effects. In our hands, we could not observe any significant protection by adding these antagonists to isolated HepG2 cells. However, in vitro blockage of CXCR2-FPR1 reverted neutrophil-mediated cytotoxicity, suggesting that despite the effect in other cell types, the activation of these receptors in neutrophils seems to be an important pathway during hepatocyte injury. Consequently, we suggest that because a significant fraction of neutrophil killing ability was contact dependent, both activation and cell contact are necessary to substantiate cytotoxicity.

After necrosis, cellular products are released in the interstitium and may be collected into the systemic circulation. Zhang et al. have shown that severe trauma patients may display a “septic-like” status, even in the absence of infection, which is partly triggered by the activation of the innate immune system by circulating mitochondrial damage-associated molecular patterns, mainly by FPR1 and TLR9 receptors.7 In accord with this, we have demonstrated that significantly increased serum levels of mitDNA and CXCL8 were found in patients with ALF, which was more prominent than nDNA release. Our parallel experiments in mice confirmed that mitDNA and chemokines are released in the course of APAP-induced liver failure, suggesting that a “necrotic shock” might be elicited by hepatocyte death. Thus, limiting the recognition of necrosis-related products may aid in the treatment of ALF patients. Accordingly, McGill et al. have recently confirmed that the mechanism by which APAP promotes cell death involves mitochondria damage,8 suggesting the assessment of mitDNA serum as a novel clinical predictive element of disease severity. Our findings in mice lacking TLR9 supported the relevance of mitDNA in the pathogenesis of liver injury, as shown by the reduced necrosis and inflammation observed in APAP-treated TLR9−/− mice. Previous reports have described apoptotic gDNA as the major activator of TLR9 in APAP-induced liver injury.34 However, we and others8 have provided evidence that mitDNA also plays a key role in immune system activation by TLR9 signaling during ALF.

Because of the unique pulmonary microvasculature architecture, the blood within the lung capillary network contains approximately 50-fold more neutrophils, and also more lymphocytes and monocytes, than other organs, which is a protective strategy to avoid pathogen dissemination during alveoli gas exchange.35 However, such abundance of lung leukocytes may be also harmful during systemic inflammatory responses. Systemic release of CXCR2 agonists (TNF-α and IL-1β) were directly related to an increased pulmonary neutrophilia and remote lung lesion during intestinal I/R injury.12 In fact, lung macrophages and dendritic cells express several receptors for necrosis-derived products,36 including FPR1,37 CXCR2,38 and TLR9 (39).39 In spite of the efficacy of the combined CXCR2-FPR1 antagonism in reducing lung leukocyte infiltration and ROS production, significant differences in lung cytokines and chemokines were not observed. Consequently, further studies may clarify whether the protection observed was a result of direct lung CXCR2-FPR1 blockage or the result of reduced liver injury. However, APAP-challenged TLR9−/− presented healthy levels of all lung-inflammatory markers evaluated in our study, which points out mitDNA as a central mediator of remote injury during ALF.

In summary, we propose that during ALF, circulating necrotic products may extend the liver-inflammatory response to remote organs, including the lungs. Taking into account our murine model and fulminant hepatic-failure patient data, we propose that controlling additional neutrophil-mediated hepatocyte injury and the further recognition of released necrotic products and chemokines by the immune system may consist in a promising alternative to slow down the progression of ALF and systemic inflammatory response.


The authors thank the transplantation staff of Hospital Federal Bonsucesso (Rio de Janeiro, Brazil) and Dr. Daniel Santos Mansur for revising the final manuscript for this article.