Although etiologies of ALF vary between Western countries and the Eastern developing world, the resulting clinical manifestation is remarkably similar. This reflects common patterns of innate immune responses to various pathogenic factors, such as bacteria toxins, cytokines, and free radicals [31
]. Among many others, proinflammatory cytokines (such as TNF-α, IL-1β, and IL-6) may play a common role in the pathophysiology of ALF.
HMGB1 is a nonclassical proinflammatory mediator that is secreted by, and activates proinflammatory responses in, phagocytes and endothelial cells [5
]. To appreciate a potential role for HMGB1 in ALF, we investigated whether HMGB1 can be released by hepatocytes in the liver of patients or animals with acute liver failure/injury.
At nontoxic concentrations, both exogenous (e.g., LPS) and endogenous (TNF) inflammatory stimuli induced HMGB1 nuclear-cytoplasmic translocation, and subsequent release in human hepatocyte HepG2 cells. In 1999, Wang et al first reported that monocytes/macrophages actively release HMGB1 in response to exogenous (e.g., LPS) or endogenous inflammatory stimuli (such as TNF-α, IL-1β, or IFN-γ) [2
]. Subsequently, active HMGB1 release has been shown in non-immune cells such as pituicytes and enterocytes [32
]. In the present study, we found that hepatocytes similarly translocate nuclear HMGB1 to cytoplasm, and release it following LPS or TNF-α stimulation. The active release of HMGB1 was not dependent on cell death, as other chemokines (such as IL-8 and GRO) were similarly released by hepatocyte following TNF stimulation.
Hepatocytes are responsible for multiple functions, including regulation of homeostasis, blood sugar, metabolisms of lipids and amino acids, bile formation, and detoxifying capacities. Our present study raises the possibility that activated hepatocytes could be a source of extracellular HMGB1, which may contribute to inflammatory response during ALF[34
]. Trying to understand if hepatocytes can release other cytokines besides HMGB1 when stimulated with LPS or TNF, human cytokine antibody array was employed to test 42 typical cytokines in the supernatant. Interestingly, both LPS and TNF-α stimulation failed to induce significant release of typical inflammatory cytokines, such as IL-1, TNF, IL-6, IFN-γ, in HepG2 cells. Nevertheless, a few chemokines (e.g., IL-8 and GRO) were released by hepatocyte following TNF stimulation but its relevance to HMGB1 release is a subject of on-going investigation.
In China, a large proportion of ALF are caused by HBV infection [35
], thus we evaluated HMGB1 cytoplasmic translocation in ALF patients caused by hepatitis B. Consistent with previous reports, we found histopathological changes; these included massive, sub-massive or bridging necrosis with immune cell infiltration and regeneration nodular of hepatocytes in liver sections of patients with ALF. Interestingly, HMGB1 cytoplasmic translocation is clearly observed in regenerated hepatocytes of patients with ALF caused by HBV infection, but not in patients with chronic HBV infection. One limitation of the clinical study is that liver tissue samples were obtained during liver transplantation surgery, and thus did not represent early clinical manifestation immediately after the onset of ALF. Consequently, we employed a murine model of ALF induced by co-administration of D-GalN and LPS to further investigate the translocation of HMGB1 in hepatocytes during liver failure/injury. In this model of ALF, liver injury is dependent on the induction of proinflammatory cytokines (such as TNF and IFN-γ) [28
], and loss of liver function and hepatic histology occur typically 6-12 h post administration of D-Gal and LPS [37
]. In the present study, we found that HMGB1 nuclear-cytoplasmic translocation occurs as early as 3 h after injection of D-GalN (600 mg/kg) and LPS (0.5 mg/kg). At this early time point, there was no significant necrosis of hepatocytes and the rate of hepatic apoptosis (detected by TUNEL assay) was still low (<10%), but the percentage of hepatocytes with HMGB1 cytoplasmic translocation was already rather high (27.42% ± 4.99%).
HMGB1 can bind to several potential receptors (e.g., RAGE and TLR2/4) and that may be highly expressed during inflammation or in primary hepatocelullar carcinoma. We can't exclude the possibility that the HMGB1 observed in cytoplasm of hepatocytes in ALF patients was HMGB1 from the nucleus of necrotic cells that combined with a receptor. This possibility is unlikely, because LPS- or TNF-α-induced cytoplasmic HMGB1 translocation was observed in hepatocytes in the absence of cell death in vitro. To examine HMGB1 translocation in vivo, we strategically chose a very early sampling time (after onset of acute liver injury/failure), when necrosis of hepatocytes was rare but the cytoplasmic HMGB1 was easily seen.
It has been widely suggested that re-localization and accumulation of HMGB1 in the cytoplasm is a necessary step for its extracellular release, raising the possibility that hepatocytes might be a source for extracellular HMGB1 during ALF. It is reasonable to propose that during liver injury/failure HMGB1 is released as a danger signal from activated or damaged hepatocytes as well as immune cells and necrotic cells. Collectively, extracellular HMGB1 itself, or in conjunction with other inflammatory stimuli, orchestrates a rigorous inflammatory response.
To our knowledge, this is the first study to link hepatic HMGB1 release and potential pathophysiology/prognosis of ALF. Indeed, circulating HMGB1 levels were found to be elevated in patients suffering from liver failure caused by hepatitis B (data not shown). The source of circulating HMGB1 may include activated innate immune cells or non-immune cells (such as hepatocytes).