Reactive oxygen species mediate liver injury through parenchymal nuclear factor-kappaB inactivation in prolonged ischemia/reperfusion

Abstract

Nuclear factor (NF)-kappaB participates in ischemia/reperfusion (I/R) hepatic signaling, stimulating both protective mechanisms and the generation of inflammatory cytokines. After analyzing NF-kappaB activation during increasing times of ischemia in murine I/R, we observed that the nuclear translocation of p65 paralleled Src and IkappaB tyrosine phosphorylation, which peaked after 60 minutes of ischemia. After extended ischemic periods (90 to 120 minutes) however, nuclear p65 levels were inversely correlated with the progressive induction of oxidative stress. Despite this profile of NF-kappaB activation, inflammatory genes, such as tumor necrosis factor (TNF) and interleukin (IL)-1beta, predominantly induced by Kupffer cells, increased throughout time during ischemia (30 to 120 minutes), whereas protective NF-kappaB-dependent genes, such as manganese superoxide dismutase (Mn-SOD), expressed in parenchymal cells, decreased. Consistent with this behavior, gadolinium chloride pretreatment abolished TNF/IL-1beta up-regulation during ischemia without affecting Mn-SOD levels. Interestingly, specific glutathione (GSH) up-regulation in hepatocytes by S-adenosylmethionine increased Mn-SOD expression and protected against I/R-mediated liver injury despite TNF/IL-1beta induction. Similar protection was achieved by administration of the SOD mimetic MnTBAP. In contrast, indiscriminate hepatic GSH depletion by buthionine-sulfoximine before I/R potentiated oxidative stress and decreased both nuclear p65 and Mn-SOD expression levels, increasing TNF/IL-1beta up-regulation and I/R-induced liver damage. Thus, the divergent role of NF-kappaB activation in selective liver cell populations underlies the dichotomy of NF-kappaB in hepatic I/R injury, illustrating the relevance of specifically maintaining NF-kappaB activation in parenchymal cells.

Figure 1

Variation of intracellular oxidative stress and TNF levels in liver after ischemia. Samples were taken from animals exposed to different times of ischemia (30, 60, 90, and 120 minutes or sham-operated and used as controls). After 1 hour of reperfusion, ROS production (A), levels of reduced and oxidized glutathione (B), and MDA (C) were measured in liver samples, whereas TNF levels (D) were analyzed in serum from I/R-treated mice (n ≥ 3). *P < 0.05 versus sham-operated mice.

Figure 2

Time-dependent increase of nuclear p65, Src phosphorylation, and κB-dependent gene expression. A: Nuclear translocation of p65, phosphorylation of Src, and tyrosine phosphorylation of IκB were measured after 1 hour of reperfusion in liver samples from mice exposed to different times of ischemia. A representative blot is shown (n = 3). Quantification of the levels of p65 and ratio of p-Src/Src are included (n = 3). A: Protein levels of Mn-SOD and TNF were examined in total liver extracts after 6 hours of sham operation or reperfusion. mRNA levels of NF-κB-dependent proinflammatory (B) and antiapoptotic (C) genes were measured 6 hours after different times of ischemia and normalized by 18S mRNA levels. *P < 0.05 versus sham-operated mice.

Figure 3

Primary hepatocytes and KCs display different responses in NF-κB activation after I/R-related stimuli. A: Increased Src phosphorylation was observed in hypoxic hepatocytes (0.5% O₂), but not after LPS exposure (2 hours, 0.1 ng/ml), as shown in the top panel. B: The levels of pTyr416Src were not increased in KCs after exposure to hypoxia or LPS. Representative blots are shown (n = 2). C and D: Changes in IL-1β mRNA levels in hepatocytes and KCs after LPS (4 hours, 0.1 ng/ml) and TNF (4 hours, 250 ng/ml) treatment (full bars) with respect to nontreated cells (empty bars), normalized by 18S mRNA levels (n = 3). *P < 0.05 versus control cells.

Figure 4

Effect of KC inactivation by gadolinium chloride on hepatic I/R injury and κB-dependent gene expression. Mice pretreated with gadolinium chloride or vehicle for 24 hours were subjected to 90 minutes of ischemia and 6 hours of reperfusion. ALT levels in serum (A) and mRNA liver levels of inflammatory (B) and protective (C) genes were analyzed (n = 3). *P < 0.05 versus sham-operated mice; ^#P < 0.05 versus I/R-treated mice.

Figure 5

GSH levels in hepatocytes and KCs. GSH concentration was measured in primary hepatocytes (A) and KCs (B) after 4 hours of exposure to SAM (2 mmol/L) or BSO (1 mmol/L) (n = 2 to 3). *P < 0.05 versus control cells.

Figure 6

Differential effects of SAM and BSO in GSH levels and liver injury after I/R. Hepatic GSH levels (A), MDA generation (B), and liver damage measured by liver H&E staining (C) and ALT in serum (D) from vehicle-, SAM-, and BSO-treated animals exposed to 90 minutes of ischemia after 6 hours of reperfusion. Neutrophil infiltration by myeloperoxidase activity (E) and immunohistochemistry (F) were analyzed after 6 hours of reperfusion. No changes in liver damage or neutrophil infiltration were observed in animals treated with SAM or BSO alone (n ≥ 3). *P < 0.05 versus sham-operated mice; ^#P < 0.05 versus I/R-treated mice.

Figure 7

GSH restoration in liver after I/R increased p65 nuclear levels and the transcription of protective genes. SOD mimetic protected the liver against I/R-induced damage. A: Representative Western blots of nuclear p65, p-Src, and Src examined after 1 hour of reperfusion (n = 3). Mn-SOD and TNF protein levels were analyzed 6 hours after sham operation or reperfusion. mRNA quantification of protective (B) and proinflammatory (C) NF-κB-dependent genes in liver samples from mice exposed to 90 minutes of ischemia and 6 hours of reperfusion and pretreated with vehicle, SAM, or BSO. D: ALT levels in serum from vehicle- or MnTBAP-treated mice and exposed to I/R as above (n = 3). *P < 0.05 versus sham-operated mice; ^#P < 0.05 versus I/R-treated mice.

Figure 8

Diagram showing different pathways of NF-κB signaling induced by I/R. In the hepatocyte, NF-κB controls protective genes after ischemia through a ROS-sensitive Src-mediated mechanism. However, high production of ROS by extended ischemia or by a deficient capacity to cope with oxidant production, as in GSH-depleted animals, reduces the transcription of NF-κB-responsive genes, and Src-mediated activation is ineffective. In contrast, KCs are mainly responsible for the production of NF-κB-dependent inflammatory proteins and show no evidence of reduction because of increased ROS production. In fact, levels of mRNAs from inflammatory genes rise as the time of ischemia was extended.