Intracellular survival pathways in the liver


Christian Trautwein, MD, Medizinische Klinik III, University Hospital, RWTH Aachen, Pauwelsstr. 30, 52074 Aachen, Germany.
Tel: +49 241 80 80 860
Fax: +49 241 80 82455


Abstract: Recent studies have drawn attention to cytokines as important modulators of hepatocyte cell death during acute and chronic liver disease. Through interaction with cell surface receptors, they activate specific intracellular pathways that influence cell fate in different manners. For example, tumor necrosis factor not only induces proapoptotic signals via the caspase cascade but also activates intracellular survival pathways, namely the nuclear factor (NF)-κB pathway. In this article, we will focus on the function of the NF-κB pathway in liver physiology and pathology. Especially, recent data based on experiments with genetically modified mice will be discussed, which demonstrated important and controversial functions of this pathway e.g. in cytokine-mediated hepatocyte apoptosis, ischemia-reperfusion injury, liver regeneration and the development of hepatocellular carcinoma. Moreover, the role of the interleukin-6 pathway and its possible protective function in the context of liver failure will be summarized.

Cytokines are small-molecular-weight messengers secreted by one cell to alter its own behavior (autocrine messenger), a closely related cell (paracrine messenger) or cells in different organs (endocrine messenger). Thereby, they enable different organs to react properly to changes in their surrounding and thus to maintain the organism in homeostasis. The liver is an exceptional organ in terms of its metabolic, synthetic and detoxifying function. It has the unique potential to regenerate after tissue loss and for example plays an important role in the regulation process that keeps blood glucose stable. All these and many other functions represent the organ's ability to execute the proper reaction toward the body's demands. Therefore, the liver appears to be a preferred source for and target of cytokine signalling, respectively, and at present one can only imagine the incredible complexity of the network connecting cytokines, receptors and signalling pathways in the liver.

Cytokine action is mediated by the interaction of cellular receptors, that signal internally to the nucleus, and external factors that are able to bind these receptors. These networks have evolved early in evolution; pathways with strong homology to human cytokine networks are already found in Drosophila and mollusks (1). For example, in Drosophila, nuclear factor (NF)-κB-like transcription factors are activated in order to combat infections (2), and this function represents one major role of cytokine networks in higher organisms like humans. Maintaining the ordered balance between proliferation and controlled cell death (apoptosis) during embryonic development and organogenesis represents another important function of cytokines under physiologic conditions (3). As these functions are preserved in the adult organism, a disturbance of the critical balance might have deleterious effects.

Hepatic failure takes place when the amount of functioning hepatocytes decreases until the organ is not capable of fulfilling both its metabolic and synthetic functions anymore. Excessive apoptosis has been implicated in a number of acute and chronic liver diseases, e.g. viral and autoimmune hepatitis, cholestatic disease, alcoholic or drug/toxin-induced liver injury and transplantation-associated liver damage, including ischemia–reperfusion injury and graft rejection (4). Numerous studies in patient samples, cell culture and animal models point to the fact that e.g. acute and chronic liver failure due to various agents like Hepatitis virus is aggravated or even facilitated by the excess signalling of so-called death receptor ligands (5). However, there is also a group of cytokine-triggered pathways that mediate a survival signal to the hepatocytes. Under certain conditions, the action of these survival pathways can counteract the death signal mediated by e.g. death receptor ligands. In fact, in many instances hepatic failure might represent the deregulation of hepatic homeostasis as a result of an imbalance between damaging and protective signals that are very tightly regulated under physiologic conditions.

In this article, we will focus on the NF-κB pathway. It represents an intracellular survival pathway that is activated in parallel to the proapoptotic caspase cascade by death receptor ligands. Recent studies in animal models could highlight a fundamental role of the NF-κB pathway in the regulation of processes like cell death and proliferation of liver cells, which, when deregulated, form the basis for the occurence of hepatocellular carcinoma (HCC). Moreover, we will introduce the interleukin (IL)-6 signalling pathway which withholds a protective role in many experimental liver disease models.

The NF-κB family of transcription factors

NF-κB is a dimer of members of the Rel family of DNA-binding proteins. The mammalian NF-κB family includes five cellular DNA-binding subunits proteins: p50 (NF-κB1), p52 (NF-κB2), c-Rel (Rel), p65 (RelA) and RelB (6). The NF-κB DNA- binding subunits share an N-terminal Rel homology domain (RHD). This region is responsible for DNA binding, dimerization, nuclear translocation and interaction with the inhibitory IκB proteins (7).

p65 (RelA), RelB and c-Rel contain C-terminal transactivation domains that trigger target gene transcription. Of these proteins, p65, which contains two potent transactivation domains (TADs) within its C-terminus, mediates strongest gene activation (8). NF-κB commonly refers to a p50/p65 heterodimer. It is one of the most avidly forming dimers and is the major Rel complex in most cells. The activity of NF-κB is controlled by IκBs (IκBα, IκBβ, IκBε, IκBγ, IκBNS and Bcl-3), a family of cytoplasmic inhibitory proteins that share a number of protein/protein interaction domains called ankyrin repeats. The precursor forms p105 and p100 are also included in this family, as they contain IκB-like repeats and, therefore, inhibit NF-κB activation (9). NF-κB is effectively sequestered in the cytoplasm by IκB in an inactive state via complex formation and the ability of IκB to mask the nuclear localization site (NLS) of NF-κB. As IκBα is an NF-κB target gene, it also terminates NF-κB activation at the transcriptional level: increased synthesis of IκBα shuts down NF-κB-induced gene expression by IκBα-mediated nuclear export of the DNA-binding subunits, thereby acting within a negative feedback loop (10).

Control of NF-κB activation: general aspects

In order to activate NF-κB, different pathways have evolved, which all lead to the generation of DNA-binding dimers. At present, the so-called canonical and non-canonical pathways and the DNA damage-induced NF-κB pathway have been identified. Moreover, further mechanisms like p65 posttranslational modifications regulate the activity of this transcription factor.

The canonical pathway is the best-described and probably the most important mediator of NF-κB activation in response to cytokines. An essential step in this pathway is the disruption of cytoplasmic NF-κB: IκB complexes, initiated by the phosphorylation of the most important IκB family member, IκBα, at Serine 32 and 36 (S32 and S36) through a high-molecular-weight IκB kinase (IKK) complex. This phosphorylation is a prerequisite for the subsequent polyubiquitination of IκB-α by a specific, constitutively active ubiquitin ligase belonging to the SCF family (11). The ubiquitin-marked I-κB proteins are then rapidly degraded by the 26S proteasome, leading to the unmasking of the NLS of NF-κB and thus allowing nuclear entry, DNA binding and transcriptional activity of NF-κB. Numerous studies have implicated a central role of this pathway, in liver physiology and pathology.

Another form of NF-κB activation, the so-called non-canonical pathway has been described particularly in B cells. The activation of this IκB-independent pathway involves the IKK subunit IKK1 and results in the release of p52/RelB and p50/RelB dimers (12–14). It is induced e.g. by lymphotoxin β (LTβ) and leads to NIK- and IKK1-dependent processing of the p100 precursor protein, which results in the release of p52 (12, 15). DNA-damage-induced NF-κB activation, in contrast to the previously described pathways, occurs in an IKK-independent manner. It has been observed e.g. after doxorubicin stimulation or ultraviolet (UV) radiation, and involves mitogen-activated protein kinase (MAPK)-dependent alternative IκBα phosphorylation (16–19). A specific function of these pathways in hepatocytes in vivo remains to be elucidated.

A growing number of studies have suggested that besides the formation and nuclear translocation of NF-κB dimers, post-translational modifications of NF-κB subunits might also influence NF-κB activation. Especially, the phosphorylation of p65 and also its acetylation appear to modify its transcriptional activity significantly (20, 21).

NF-κB in the regulation of hepatocyte apoptosis

Death receptor cytokines in liver failure

Studies in patients and animal models have strongly implicated that death receptor ligands such as tumor necrosis factor (TNF) or Fas ligand (FasL) are involved in the induction of apoptosis and in triggering destruction of the liver (5). TNF was originally identified by its capacity to induce hemorrhagic necrosis in mice tumors (22), but severe side effects led to a failure of its use as a systemic anticancer chemotherapeutic agent (23, 24). A very prominent effect was the direct cytotoxic role of TNF for human hepatocytes, resulting in increased levels of serum amino transferases and bilirubin. Since then, many clinical studies have underlined the crucial role of TNF in fulminant hepatic failure (FHF) and other liver diseases. TNF participates in many forms of hepatic pathology, including ischemia/reperfusion (I/R) injury, alcoholic and viral hepatitis and injury from hepatotoxins (25–28).

Exogenous TNF induces FHF and hepatocyte apoptosis in combination with other toxins (28). TNF serum levels are clearly elevated in patients with FHF (29). In another study, it was shown that serum TNF levels were significantly higher in FHF patients who died than in patients who survived (30). We could show that serum TNF, TNF-receptor 1 (TNF-R1) and TNF-R2 levels were increased markedly in patients with FHF and that these changes directly correlated with disease activity. In explanted livers of patients with FHF, infiltrating mononuclear cells expressed high amounts of TNF and hepatocytes overexpressed TNF-R1. Moreover, we found that apoptotic hepatocytes were significantly increased in FHF, and there was a strong correlation with TNF-α expression (31). Thus, it is very likely that the TNF system is involved in the pathogenesis of FHF in humans, and its significance has also been shown clearly in several animal models of hepatic failure, e.g. the endotoxin/d-galactosamine (GalN) and the concanavalin A (ConA) model (32, 33).

TNF-dependent NF-κB activation

TNF facilitates programmed cell death by activation of caspases. It signals through two distinct cell surface receptors, TNF-R1 and TNF-R2, of which TNF-R1 initiates the majority of TNF's biological activities. Binding of TNF to its receptor leads to the release of the inhibitory protein silencer of death domains (SODD) from TNF-R1's intracellular domain. This leads to the recognition of the intracellular TNF-R1 domain by the adaptor protein TNF receptor-associated death domain (TRADD), which in turn recruits the Fas-associated death domain (FADD). FADD recruits caspase-8 to the TNF-R1 complex, where it becomes activated and initiates the protease cascade leading to activation of executioner caspases and apoptosis (Fig. 1) (34).

Figure 1.

 Tumor necrosis factor (TNF)-dependent activation of nuclear factor-κB (NF-κB).

Next to activation of caspases, ligand binding of TNF to its receptor also leads to the activation of the NF-κB pathway (Fig. 1). As outlined before, TNF-induced activation of NF-κB relies on phosphorylation of two conserved serines (S32 and S36 in human IκBα) in the N-terminal regulatory domain of IκBs. After phosphorylation, the IκBs undergo a second posttranslational modification: polyubiquitination by a cascade of enzymatic reactions, followed by the degradation of IκB proteins by the proteasome, thus releasing NF-κB from its inhibitory IκB-binding partner, so it can translocate to the nucleus and activate transcription of NF-κB-dependent target genes (11, 35). As the enzymes that catalyze the ubiquitination of I-κB are constitutively active, the only regulated step in NF-κB activation appears to be in most cases the phosphorylation of I-κB molecules.

A high-molecular-weight complex that mediates the phosphorylation of I-κB has been purified and characterized. This complex consists of three tightly associated IKK polypeptides: IKK1 (also called IKKα) and IKK2 (IKKβ) are the catalytic subunits of the kinase complex and have very similar primary structures with 52% overall similarity (11, 36, 37). Moreover, it contains a regulatory subunit called NF-κB essential modulator (NEMO), IKKγ or IKKAP-1 (38–40). In vitro, IKK1 and IKK2 can form homo and heterodimers (41). Both IKK1 and IKK2 are able to phosphorylate I-κB in vitro, but IKK2 has a higher kinase activity in vitro compared with IKK1 (38, 42–45).

Activation of the IKK complex upon TNF stimulation involves IKK recruitment to the TNF-R1 (46–48). Next to TNF-R1, this process involves TNF-receptor-associated factor 2 (TRAF2) and the death-domain kinase receptor-interacting protein (RIP). In response to TNF treatment, TRAF2 recruits the IKK complex to TNF-R1 via the interaction of the RING-finger motifs of TRAF2 with the leucin-zipper motif of both IKK1 and IKK2 (46, 47). RIP can directly interact with NEMO and mediate IKK activation, although the enzymatic activity of RIP is not required for this process (48). The mechanism by which recruitment of the IKK complex to the TNF receptor leads to IKK activation is not clear, but might involve NEMO-induced autophosphorylation of the IKK complex. Moreover, ubiquitination of multiple factors that regulate the IKK complex, like TRAF6/TAK1 or c-IAP1, an inhibitor of apoptosis that is also part of the TNF receptor complex, modulates the activity of the NF-κB pathway (35).

Antiapoptotic function of NF-κB in the liver

Numerous studies have provided evidence that NF-κB provides survival signals in the context of death receptor-induced apoptosis in the liver. This process is assumed to involve the transcriptional induction of various apoptotic suppressors (49). Evidence that NF-κB governs critical antiapoptotic proteins comes from well-described animal models. Injection of TNF into mice and addition of TNF to hepatic cells resulted in activation of nuclear translocation and of DNA binding of NF-κB (50). In contrast to Fas-mediated apoptosis, hepatocytes are resistant to apoptosis induced by TNF or LPS, a potent inductor for endogenous TNF in the liver, unless they are treated with inhibitors of transcription of (antiapoptotic) proteins like cycloheximide or actinomycin D (28, 51, 52). Specific blockade of NF-κB activation by adenoviral directed overexpression of the NF-κB superrepressor IκB (which retains the dimeric NF-κB in the cytoplasm as it contains mutations at the phosphorylation sites) significantly enhanced TNF-mediated apoptosis of hepatocytes (53). A similar result was obtained by treatment of hepatocytes with the proteasome inhibitor lactacystin, which prevents degradation of IκBs, and an antibody against p65 protein (54).

Further evidence for the central role of the NF-κB pathway in preventing apoptosis in hepatocytes came from genetic experiments. Knockout mice lacking the p65 subunit of NF-κB die between days E15 and E16 postcoitum as a result of fetal hepatocyte apoptosis (55). This is caused by increased sensitivity towards TNF, as TNF/p65 double-deficient mice are rescued from embryonic lethality (56). c-Rel may partially compensate for p65 in blocking liver apoptosis in that c-rel/p65 double knockout mice show liver degeneration approximately 1.5 days earlier than p65 single knockout mice (57).

Genetic experiments have also highlighted the differential functions of the IKK subunits in TNF-mediated liver apoptosis. Mice lacking Ikk1 die shortly after birth and display a phenotype marked by thickening of the skin and limb as well as skeletal defects. Stimulation of embryonic fibroblasts or liver tissue with IL-1, TNF or endotoxin results in normal IKK activation and I-κBα degradation (58, 59). In contrast, Ikk2−/− mice die in utero approximately at embryonic day (E) 12.5 as a result of massive apoptosis in the liver, and fibroblasts from these mice show no activation of NF-κB in response to TNF and IL-1 (60–62). A similar phenotype was noted in mice lacking the regulatory subunit NEMO, which also die from massive apoptosis in the liver and show a defect in NF-κB activation upon TNF stimulation in primary murine embryonic fibroblast (MEF) cells (63). Therefore, at least during embryogenesis, IKK2 and NEMO appear to be the critical subunits for NF-κB activation and protection of liver cells from proinflammatory cytokines like TNF.

The role of the IKK subunits in the adult animal is controversial, but might vary from the situation during embryogenesis. Experiments with adenovirus expressing dominant-negative IKK forms in primary hepatocytes showed that overexpression of a dominant-negative IKK2 mutant can totally block TNF-dependent NF-κB activation and leads to hepatocyte apoptosis (64). Conditional knockout mice based on the cre/loxP system have emerged as new powerful tools to study gene functions in the adult animal in vivo. In a recent study, we could show that hepatocyte-specific ablation of IKK2 did not lead to a strongly impaired activation of NF-κB or increased apoptosis after TNF stimulation, probably because IKK1 homodimers can take over this function in the absence of IKK2 in the adult mouse. In contrast, conditional knockout of NEMO resulted in complete block of NF-κB activation and massive hepatocyte apoptosis (65), underlining that NEMO is the only irreplaceable IKK subunit for prevention of TNF-mediated liver apoptosis.

Besides its antiapoptotic function in TNF-mediated liver apoptosis, recent reports suggest that NF-κB also plays a role in other apoptotic pathways. Concanavalin A (ConA) activates T lymphocytes in vitro and causes T-cell-dependent hepatic injury in mice, characterized by apoptotic cell death (66). ConA exerts cytotoxic effects through cell-bound TNF, that activates TNF-R1 and TNF-R2 (67). ConA stimulation in mice leads to activation of NF-κB, and this pathway is blocked by anti-TNF (68). Inhibition of NF-κB by degradation-resistant IκBα (IκBα super-repressor) increases susceptibility to ConA-mediated liver apoptosis (69). Fas, also called APO-1 or CD95, is a cell surface receptor belonging to the TNF receptor that is expressed in hepatocytes and plays a role in liver failure (70, 71). The function of NF-κB in Fas-mediated apoptosis in hepatocytes is controversial. In hepatocyte-derived cell lines, inactivation of NF-κB made these cells more susceptible to apoptosis induced by Fas stimulation (72). In contrast, in a model of adenoviral hepatitis in mice, NF-κB mediates the proapoptotic function of Fas (73), underlining that the pro or antiapoptotic role of NF-κB is determined by the nature of the death stimulus.

Mediators of the antiapoptotic NF-κB function

After binding to its responsive DNA elements, NF-κB induces a variety of target genes that mediate its antiapoptotic function. Among these genes are the cellular inhibitors of apoptosis (c-IAPs), e.g. c-IAP1 and c-IAP2, which directly bind and inhibit effector caspases, such as caspase-3 and caspase-7, and prevent activation of procaspase-6 and procaspase-9 (74). Another cIAP regulated by NF-κB is X chromosome-linked IAP (XIAP), which inhibits caspase-3 and caspase-7 and might prevent activation of procaspase-9 (75). FLICE inhibitory protein (c-FLIP) is an NF-κB regulated-protein that inhibits apoptosis by interfering with procaspase-8 activation (76). Moreover, antiapoptotic Bcl-2 family members (A1 and Bcl-xL) and TRAF1 and TRAF2 are induced upon TNF stimulation in an NF-κB-dependent manner (76, 77).

Recent studies have revealed that the NF-κB and the c-Jun N-terminal kinase (JNK) pathways are functionally connected. The JNK cascade is important in regulating cell death decisions and is strongly activated in TNF-mediated apoptosis via TRAF2 and RIP (5, 78). The effects of JNK on TNF-induced apoptosis have been enigmatic, and numerous studies have suggested pro or antiapoptotic functions (79–82). Functional interconnection between NF-κB and JNK was revealed by two studies, showing that TNF-induced activation of JNK is prolonged in cells that are deficient in NF-κB activation (p65 and Ikk2 knockout MEFs and cells stably expressing degradation-resistant IκBα), and prolonged JNK activation in these cells promoted apoptosis (79, 82). Only ectopic expression of XIAP, but not other antiapoptotic molecules, could inhibit prolonged JNK activation in TNF-treated p65 knockout MEFs (82). This interaction was further confirmed in primary rat hepatocytes, where inhibition of NF-κB by an IκBα superrepressor led to prolonged JNK activation. Inhibition of JNK by a chemical inhibitor could inhibit TNF-induced, but not Fas-induced apoptosis in these cells (83), underlining that one important mechanism of how NF-κB mediates its antiapoptotic function in hepatocytes is suppression of JNK activity.

Role of NF-κB in hepatic I/R injury

Despite these linear, death receptor-dependent pathways, which can be well examined in experimental models like TNF or FasL stimulation in mice, experimental hepatic I/R injury in rodents represents a complex model that reflects liver damage after organ transplantation, tissue resections or hemorrhagic shock and whose molecular mechanisms are poorly understood. In most studies, the injury detected after temporal clamping of hepatic blood flow and the subsequent reperfusion leads to an excessive inflammatory response, followed by necrotic cell death (84). Instead of linear apoptotic signalling, it has been proposed that I/R injury consists of two different phases: one displaying acute cellular injury and a secondary, subacute phase resulting from inflammatory responses (85–87). Genetic experiments identified TNF but not Fas as a critical component in the mediation of I/R injury to the liver (88). The role of NF-κB in this model is not clear, but recent studies have shown that NF-κB is activated after I/R in the liver and might withhold a specific function in this context.

NF-κB DNA binding occurs quickly upon hepatic I/R (89), but it has been unclear how NF-κB is activated in this model. It has been suggested that this process occurs alternatively to that during death receptor-dependent signalling. Whereas TNF stimulation leads to IKK-dependent phosphorylation of both I-κBα and I-κBβ on serine residues, recent studies have suggested that during I/R injury, NF-κB becomes activated by c-Src-dependent tyrosine phosphorylation of I-κBα but not I-κBβ on Tyr42, and this process takes place in the absence of I-κBβ ubiquitin-dependent degradation (90, 91), thus representing an alternative NF-κB activation pathway to the ones mentioned before. Data from our own laboratory suggest that this process is still dependent on IKK2, as conditional knockout mice for IKK2 show a defect in NF-κB activation after I/R (65). It has also been unclear whether NF-κB-dependent signalling withholds a protective or damaging role in ischemia–reperfusion injury. As NF-κB withholds both proinflammatory and antiapoptotic properties in the liver (92), inhibition of its function after I/R might turn the fate of hepatocytes towards both directions. In fact, inhibition of NF-κB activation protects from liver injury due to I/R (65, 91), thus underlining that, depending on the experimental model, the NF-κB pathway does not serve as a survival pathway, but instead can aggravate hepatocyte death and liver damage.

NF-κB in liver regeneration

In mammals, the liver stands out against other organs by its capacity to regenerate lost parenchymal mass in response to e.g. surgical resection, viral injury or toxic exposure. A well-established model for studying liver regeneration is that of partial (two-thirds) hepatectomy (PH) in rodents. It has been a major challenge to evaluate the extra- and intracellular events that orchestrate this highly complex process that drives quiescent, fully differentiated hepatocytes through distinct stages including priming of hepatocytes, cell cycle progression, proliferation and cessation of regeneration (93, 94). The model of liver regeneration is an in vivo model of cellular proliferation and can only, to a very limited extent, be reproduced in cell culture models. Therefore, it is a good of example how transgenic and knockout animal technology was providing major advances in understanding the function of cytokine signalling and NF-κB in this model.

NF-κB was first identified in the liver as a factor that is rapidly activated within 30 min after PH (95). A possible involvement of TNF-related signalling cascades was further suggested by the fact that liver regeneration is defective in TNFR-1 knockout mice, which do not show hepatic NF-κB activation after PH (96). Therefore, further efforts were aimed at a more specific inhibition of the NF-κB pathway in this model. Treatment of mice with the fungal metabolite gliotoxin led to an inhibition of NF-κB activation and a defect in cell proliferation after PH, which was accompanied by increased apoptosis (97). However, again, it is not clear whether this compound only inhibits NF-κB or other crucial factors as well. A more specific inhibition of this pathway was again achieved by adenoviral gene transfer of the I-κBα super-repressor into mouse livers. This treatment led to a defect in cellular proliferation and increased apoptosis after PH (98). Given these clear indications for a proproliferative function of NF-κB in the liver, it was rather surprising that, in a transgenic mouse model where the I-κBα super-repressor was expressed under a hepatocyte-specific promoter, neither a defect in hepatocyte proliferation nor increased apoptosis was seen (99). The lack of apoptosis seen in this study in contrast to the previous one with adenoviral gene transfer could be explained by the fact that the adenoviral infection itself might have driven hepatocytes into a proapoptotic state. However, comparison of these different experimental models points toward another possible explanation for the different phenotypes. Numerous previous studies have highlighted the important role that non-parenchymal liver cells like Kupffer cells, a microphage-like cell type in the liver, might play in regulating the proliferative response of hepatocytes after PH (93). In fact, adenoviral vectors infect both parenchymal and non-parenchymal liver cells (100, 101), whereas the I-κBα superrepressor in the transgenic mouse model was by the nature of its promoter only expressed in hepatocytes (99).

The theory of a diverse role of NF-κB in the different cell types of the liver after PH was confirmed recently by application of the conditional knockout technology. This technique allows the specific targeting of a certain allele in different cell types using the same floxed mouse line by crossing it to transgenic mouse lines expressing cre-recombinase under different, cell-type-specific promoters (102). In a study by Maeda et al., (103) the NF-κB-activating kinase IKK2 was conditionally knocked out in hepatocytes and this did not result in a defect in liver regeneration after PH compared with control animals. Furthermore, the authors also created a mouse line that, upon induction with interferon, had a conditional knockout of IKK2 in all liver cells including Kupffer cells by crossing the same IKK2-floxed mouse line with a transgenic cre line that depends on the interferon-responsive Mx-promoter. In this mouse line, liver regeneration was strongly impaired. Therefore, NF-κB appears not to act as a direct mitogen on hepatocytes during liver regeneration, but is more likely involved in the production of cytokines like TNF and IL-6 in non-parenchymal liver cells, which subsequently act on hepatocytes and promote their proliferation.

NF-κB in the development of HCC

According to Hanahan and Weinberg (104), tumorigenesis requires six essential alterations to normal cell physiology: self-sufficiency in growth signals; insensitivity to growth inhibition; evasion of apoptosis; immortalization; sustained angiogenesis; and tissue invasion and metastasis. NF-κB is able to induce several of these cellular alterations and thus is an interesting target to study in this context. The development of HCC has been studied in different animal models. One experimental approach is to induce tumors by chemical carcinogens, which, after a certain amount of time, leads to liver tumors that reassemble closely malignancies found in humans, e.g. in terms of genetic alterations and their histological and clinical properties (105, 106). In another model, mice develop cancer on the basis of a chronic inflammation. The Mdr2-knockout mouse develops spontaneously a cholestatic hepatitis and subsequently HCC (107) and thus represents a good equivalent to the clinically often seen sequence of carcinogenesis based on chronic inflammation e.g. caused by chronic viral hepatitis.

As in these animal models, progression to liver cancer normally takes more than 6 months and an initial event e.g. in the cholestasis model cannot be pinpointed to a defined time point; the modulation of pathways with drugs or even adenoviral vectors is hardly possible. In contrast, transgenic/knockout technology allows a long-term modulation of certain pathways. The following two studies have examined the role of the NF-κB pathway in HCC development.

Pikarsky et al. (108) used a hepatocyte-specific inducible I-κB-super-repressor transgene to study the role of NF-κB in the Mdr2-knockout model. In this study, inhibition of NF-κB resulted in a dramatic reduction of tumor development in this model. As their mouse model allowed an induction of the transcript and thus an inhibition of NF-κB at different stages of tumorigenesis, the authors could demonstrate that hepatocyte transformation and initiation of HCCs in this model were rather unaffected by inhibition of NF-κB, whereas an inhibition of NF-κB during the later stages of tumor promotion contributed mainly to the observed phenotype. From their study, the authors convincingly concluded that transformed hepatocytes are extremely vulnerable to TNF-mediated apoptosis and that NF-κB is needed during this later stage of tumor promotion to protect the transformed liver cells from dying.

Another group published a study that suggested an opposite, tumour-suppressive role of NF-κB in the liver. Maeda et al. (103) used the conditional hepatocyte-specific IKK2 knockout mouse (IKK2Δhep) to study NF-κB function in a chemical carcinogenesis model. IKK2Δhep mice that were injected at the age of 2 weeks with the DNA-damaging agent diethyl nitrosamine (DEN) developed more and larger liver tumors than their respective wild-type litter mates. According to the authors, this correlated with enhanced reactive oxygen species (ROS) production, increased JNK activation and hepatocyte death, giving rise to augmented compensatory proliferation of surviving hepatocytes. In contrast, when the authors applied an inducible IKK2 knockout mouse that deletes the gene in hepatocytes and non-parenchymal liver cells like the cytokine-producing Kupffer cells, these mice were now protected from liver carcinogenesis. Therefore, the authors demonstrated that the function of NF-κB in the development of HCC might be varying between different cell types in the liver. However, given the different outcomes of the above-mentioned studies, which might be very well explained by the different natures of the tumor models that were applied, further examinations are needed to clarify the function of NF-κB in the development of HCC.

IL-6: a protective cytokine in the context of liver failure?

IL-6 belongs to a family of cytokines comprising of IL-6, IL11, leukemia inhibitory factor (LIF), oncostatin M, ciliary neurotropic factor, novel neurotrophin-1/B cell stimulating factor-3 and cardiotropin 1 (109). IL-6 binds to hepatocytes by interacting with an 80-kd membrane glycoprotein (gp80) that complexes with a signal-transducing molecule named gp130 (Fig. 2A). Binding of gp130 leads to dimerization of the intracellular domains of two gp130 molecules, which promotes association with receptor associated Janus kinases (JAKs; JAK1, JAK2 and TYK). The JAKs become activated and phosphorylate different tyrosine residues on the gp130 molecule. Depending on the location of the phosphorylated tyrosines, signal transducers and activators of transcription (STAT) proteins (mainly STAT-3) and also the Ras/MAPK pathway become activated and trigger numerous downstream effects mediated by the signalling of IL-6 and related cytokines (Fig. 2B) (109).

Figure 2.

 Interleukin-6 (IL-6)-dependent signalling. (A) IL-6 binding and receptor dimerization, (B) signal transduction through gp130.

As outlined before, IL-6 is one of the key regulators of the initial steps of liver regeneration. An important role of IL-6-dependent signalling in the liver is also attributed to the induction of the acute-phase response (110, 111). STAT3 participates in transcriptional activation. In terms of apoptosis, some experiments showed a role for gp130 in promoting antiapoptotic effects in different cell types. Activation of STAT3 in B cells and human myeloma cells causes activation of antiapoptotic genes such as bcl-2 and bcl-xl and protects these cells from Fas-dependent apoptosis (112). Similar results were found in T cells. STAT3-deficient T cells were severely impaired in IL-6-induced proliferation, which was due to the profound defect in IL-6-mediated prevention of apoptosis. In hepatocytes, IL-6 protects from transforming growth factor (TGF)-β-induced apoptosis by blocking TGF-β-induced activation of caspase-3 via rapid tyrosine phosphorylation of phosphatidylinositol 3 kinase (PI 3 kinase), that constitutively activated the protein kinase Akt (113).

In humans, there is strong evidence that IL-6 is directly involved in the pathogenesis of different diseases, including multiple myeloma and congestive heart disease (114, 115). Recently, our group analyzed the potential role of IL-6 in the development of acute and chronic liver injury in humans and examined the pathophysiological basis in animal models. We found a direct correlation of IL-6 expression in serum and liver tissue with disease progression in patients with FHF. Additionally, we could show an abolished acute phase response and an increased susceptibility to LPS-induced liver injury in mice deficient for functional gp130 in hepatocytes (116, 117) and define genes that are protective during T-cell-mediated liver injury (118).


As outlined above, a growing number of studies have implicated cytokines and cytokine-dependent pathways in the development of liver failure, chronic liver disease, hepatic inflammation and liver carcinogenesis. Survival pathways like NF-κB or IL-6 withhold a protective function in many experimental liver disease models and are thus an attractive target for a pharmacological intervention. Numerous chemical compounds, monoclonal antibodies and viral vectors that inhibit or modulate e.g. the TNF or the NF-κB pathway have been developed or are under development (119). However, the studies described above have highlighted the fact that e.g. an inhibition of NF-κB in the liver can have different outcomes depending on the experimental model applied: protecting from apoptosis in a model of TNF-dependent cell death vs. aggravating cellular necrosis in a model of I/R injury. Thus, a survival pathway is not necessarily protective in any context. Moreover, the study by Pikarsky et al. (108) highlighted that a survival pathway like NF-κB might even protect tumor cells from apoptosis during development of HCC, and thus a pharmacological modification in terms of an enhancement might have deleterious effects. Finally, studies by Maeda et al. (103) suggested that the function of NF-κB in different cell types of the liver like Kupffer cells vs. hepatocytes might differ. Therefore, systemic drugs that would systemically target members of the NF-κB signalling pathway might have beneficial effects in one cell type, but unpredictable side effects in other cell types and thus to the whole system. At present, technical means for a cell-type-specific drug targeting in the liver like in most other organs are not available. Until then, the conditional knockout technology is the only way to study gene and protein function in defined disease models and defined cell populations in the liver and will provide the knowledge that is needed for the application of the next generation of molecular drugs that will modify survival pathways in the liver.