Nuclear factor κB up-regulation of CCAAT/enhancer-binding protein β mediates hepatocyte resistance to tumor necrosis factor α toxicity


  • Potential conflict of interest: Nothing to report.


The sensitization of hepatocytes to cell death from tumor necrosis factor α (TNFα) underlies many forms of hepatic injury, including that from toxins. Critical for hepatocyte resistance to TNFα toxicity is activation of nuclear factor κB (NF-κB) signaling, which prevents TNFα-induced death by the up-regulation of protective proteins. To further define the mechanisms of hepatocyte sensitization to TNFα killing, immunoblot analysis comparing livers from mice treated with lipopolysaccharide (LPS) alone or LPS together with the hepatotoxin galactosamine (GalN) was performed to identify TNFα-induced protective proteins blocked by GalN. Levels of CCAAT/enhancer-binding protein β (C/EBPβ) were increased after LPS treatment but not GalN/LPS treatment. In a nontransformed rat hepatocyte cell line, TNFα-induced increases in C/EBPβ protein levels were dependent on NF-κB–mediated inhibition of proteasomal degradation. Pharmacological inhibition of c-Jun N-terminal kinase (JNK) did not affect C/EBPβ degradation, indicating that the process was JNK-independent. C/EBPβ functioned to prevent cell death as adenoviral C/EBPβ overexpression blocked TNFα-induced apoptosis in cells sensitized to TNFα toxicity by NF-κB inhibition. C/EBPβ inhibited TNFα-induced caspase 8 activation and downstream mitochondrial cytochrome c release and caspase 3 and caspase 7 activation. Studies in primary hepatocytes from c/ebpβ−/− mice confirmed that loss of C/EBPβ increased death from TNFα. c/ebpβ−/− mice were also sensitized to liver injury from a nontoxic dose of LPS or TNFα. The absence of jnk2 failed to reverse the GalN-induced block in C/EBPβ induction by LPS, again demonstrating that C/EBPβ degradation was JNK-independent. Conclusion: C/EBPβ is up-regulated by TNFα and mediates hepatocyte resistance to TNFα toxicity by inhibiting caspase-dependent apoptosis. In the absence of NF-κB signaling, proteasomal degradation of C/EBPβ is increased by a JNK-independent mechanism and promotes death from TNFα. (HEPATOLOGY 2010;.)

Tumor necrosis factor α (TNFα) is a critical mediator of multiple forms of liver injury, including that resulting from toxins,1, 2 ischemia/reperfusion,3, 4 viral hepatitis,5, 6 and cholestasis.7, 8 Central to TNFα's role as a hepatotoxic factor is its ability under certain pathophysiological conditions to induce apoptotic cell death. TNFα binding to the type 1 TNFα receptor recruits a series of intracellular proteins that ultimately activate initiator caspase 8.9 Caspase 8 activation triggers the sequential release of lysosomal cathepsin B,10 cleavage of the proapoptotic Bcl-2 family member Bid,11 initiation of the mitochondrial death pathway with release of cytochrome c, and activation of downstream effector caspases that induce apoptosis.12 Hepatocytes are normally resistant to TNFα cytotoxicity through TNFα-induced activation of the transcription factor nuclear factor κB (NF-κB).13, 14 Loss of NF-κB activity in hepatocytes in culture,14 or in vivo,15 sensitizes the cells to death through this apoptotic pathway.10, 13, 14 TNFα-dependent liver injury from hepatotoxins such as carbon tetrachloride, galactosamine, and alcohol may result from a block in protective NF-κB signaling.

A mechanism of NF-κB–mediated resistance to TNFα toxicity is down-regulation of the mitogen-activated protein kinase c-Jun N-terminal kinase (JNK).16–18 In the absence of NF-κB signaling, TNFα-induced JNK activation is converted from a transient to a prolonged response that triggers cell death. Although the central function of JNK is to increase gene transcription through the phosphorylation and activation of the activator protein-1 component c-Jun, JNK regulates TNFα toxicity through nontranscriptional effects on protein degradation. The induction of cell death by JNK overactivation occurs in part from alterations in the half-lives of two proteins that mediate hepatocyte TNFα resistance: cFLIP and Mcl-1.19, 20 Loss of the transcription factor NF-κB therefore sensitizes hepatocytes to TNFα cytotoxicity in part through JNK-dependent effects on protein degradation.

CCAAT/enhancer-binding protein β (C/EBPβ) is a member of a family of transcription factors that regulate several critical cellular functions, including apoptosis.21 C/EBPβ has three protein forms that in rodents are termed LAP1 (38 kDa), LAP2 (34 kDa), and LIP (20 kDa).21 LAP1 and LAP2 act as transcriptional activators and LIP as a dominant negative. C/EBPβ promotes cell survival in several nonhepatic cell types after DNA damage.22–24 In addition, a novel nontranscriptional function of C/EBPβ as a caspase inhibitor has been described in hepatic stellate cells.22 Whether C/EBPβ performs this function in other cell types is unknown. In hepatocytes, C/EBPβ promotes apoptosis from the death receptor Fas.25 C/EBPβ regulation by TNFα has been shown to occur in hepatocytes at the level of subcellular localization as TNFα induces C/EBPβ translocation to the nucleus, where it regulates hepatocyte differentiation and proliferation through effects on gene transcription.26–28 It is unknown whether C/EBPβ functions in hepatocyte death from TNFα.

Galactosamine/lipopolysaccharide (GalN/LPS)-induced liver injury is a well-established experimental model of TNFα-dependent hepatic injury.29, 30 GalN is a hepatocyte-specific transcriptional inhibitor that at subtoxic doses sensitizes hepatocytes to death from LPS-induced TNFα. A feature of this model is that protein changes induced by LPS alone can be contrasted with those from combined GalN/LPS treatment to identify protective proteins whose induction by TNFα is blocked by GalN. Using this approach, we identified C/EBPβ as a factor whose induction by LPS was blocked by the hepatotoxin GalN. In vitro and in vivo studies demonstrated that C/EBPβ blocks TNFα-induced apoptosis in hepatocytes at the level of caspase 8 activation. These findings identify C/EBPβ as an NF-κB–regulated antiapoptotic factor that mediates hepatocyte resistance to TNFα toxicity.


C/EBPβ, CCAAT/enhancer-binding protein β; GalN, galactosamine; JNK, c-Jun N-terminal kinase; LPS, lipopolysaccharide; mRNA, messenger RNA; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; NF-κB, nuclear factor κB; TNFα, tumor necrosis factor α; TUNEL, terminal deoxynucleotidyl transferase–mediated dUTP nick-end labeling.

Materials and Methods

Materials and Methods are available in the Supporting Information online.


GalN Blocks LPS-Induced Increase in C/EBPβ in Mouse Liver.

As a strategy to identify novel factors mediating hepatocyte resistance to hepatotoxin-induced, TNFα-dependent liver injury, immunoblot analysis was performed on hepatic proteins isolated from LPS- and GalN/LPS-treated mice. An increase in hepatic levels of a specific protein by LPS alone that is blocked by GalN cotreatment identifies that protein as a potential TNFα-inducible protective factor. C/EBPβ levels were examined because of the known function of this protein as a caspase inhibitor. LPS administration markedly increased both the LAP1 and LAP2 forms of C/EBPβ protein in mouse liver within 4 hours (Fig. 1A). However, cotreatment with GalN blocked this LPS-induced increase in C/EBPβ (Fig. 1A). In contrast, levels of C/EBPα were unaffected by LPS or GalN treatment and served as a loading control (Fig. 1A). These findings suggested that a hepatotoxin-mediated inhibition of C/EBPβ induction may sensitize hepatocytes to death from TNFα.

Figure 1.

LPS induction of C/EBPβ protein is blocked by GalN cotreatment. (A) Total hepatic protein was isolated from the livers of an untreated mouse and mice treated for the hours indicated with LPS alone, or LPS together with GalN, and immunoblotted with antibodies against C/EBPβ or C/EBPα. Results are representative of three independent experiments. (B) Fold increase in hepatic C/EBPβ mRNA levels as determined by real-time polymerase chain reaction at the times indicated after LPS or GalN/LPS treatment. C/EBPβ mRNA levels were normalized to those for TATA-box binding protein (n = 3).

GalN inhibits hepatocyte transcription, suggesting that this hepatotoxin may block C/EBPβ induction by this mechanism. To determine whether GalN blocked C/EBPβ up-regulation at the messenger RNA (mRNA) level, hepatic levels of C/EBPβ mRNA after LPS or GalN/LPS treatment were determined by real-time polymerase chain reaction. Levels of C/EBPβ mRNA increased three- to eight-fold at 1-6 hours after treatment with LPS alone (Fig. 1B). GalN did block the increase in C/EBPβ mRNA at 2 hours, but at the other time points C/EBPβ mRNA levels increased two- to four-fold despite GalN cotreatment (Fig. 1B). Thus, GalN reduced but did not block completely the LPS-induced increase in C/EBPβ mRNA. These findings are in contrast to the complete absence of any increase in C/EBPβ protein in GalN/LPS-treated mice, suggesting that the lack of C/EBPβ protein induction in these mice was mediated at least in part through changes in protein translation or degradation.

TNFα and LPS Induce an NF-κB–Dependent Increase in C/EBPβ in RALA Hepatocytes.

To determine whether TNFα and LPS regulate C/EBPβ specifically in hepatocytes, the effects of TNFα and LPS on C/EBPβ levels were examined in RALA hepatocytes cultured under nontransformed conditions. TNFα treatment increased cellular C/EBPβ protein levels within 2 hours (Fig. 2A). The increase in C/EBPβ was further augmented by cotreatment with LPS (Fig. 2A), indicating that TNFα and LPS both up-regulate C/EBPβ protein content in hepatocytes. The normal up-regulation of C/EBPβ by TNFα was dependent in part on protein synthesis as the induction was partially blocked by the protein synthesis inhibitor cycloheximide (Fig. 2B).

Figure 2.

TNFα induction of C/EBPβ is mediated by NF-κB–dependent inhibition of proteasomal degradation. (A) RALA hepatocytes were untreated or treated with TNFα alone or together with LPS for the indicated number of hours. Total protein was isolated and aliquots immunoblotted with antibodies for C/EBPβ or β-actin. (B) RALA hepatocytes received no pretreatment or were pretreated with cycloheximide (CHX) for 1 hour, then treated with TNFα for the number of hours shown. Total protein was immunoblotted for C/EBPβ and β-actin. (C) Following infection with Ad5LacZ or Ad5IκB, cells were left untreated or treated with TNFα in the absence or presence of 1 hour of MG132 (MG) pretreatment. Total protein was isolated 4 hours after TNFα administration and immunoblotted with C/EBPβ or β-actin antibodies. Levels of β-actin demonstrated equal loading among protein samples. Immunoblots are representative of four independent experiments. Numerical data under the immunoblots represents the relative signal intensity by densitometry scanning of four experiments.

TNFα-induced activation of the transcription factor NF-κB is a critical protective response for hepatocyte resistance to TNFα toxicity.14 To investigate the role of NF-κB in TNFα up-regulation of C/EBPβ, NF-κB activation was inhibited with the adenovirus Ad5IκB which expresses a mutant IκB that irreversibly binds and inactivates NF-κB.15 The TNFα-mediated increase in C/EBPβ was abrogated in Ad5IκB-infected cells, but not in control Ad5LacZ-infected hepatocytes (Fig. 2C), indicating that NF-κB activation mediated the TNFα-induced increase in C/EBPβ.

The total block in induction of C/EBPβ protein in GalN/LPS-treated mice, despite an increase in C/EBPβ mRNA, suggested that NF-κB signaling regulates C/EBPβ in vivo at the level of protein degradation. To test this possibility, cells were treated with TNFα in the absence or presence of the proteasomal inhibitor MG132.31 MG132 treatment alone in Ad5LacZ- or Ad5IκB-infected cells increased cellular C/EBPβ protein content to a level equivalent to that in TNFα-treated, Ad5LacZ-infected cells (Fig. 2C), demonstrating constitutive regulation of C/EBPβ levels by proteasomal degradation. Cotreatment with MG132 had no effect on C/EBPβ levels in TNFα-treated, Ad5LacZ-infected cells (Fig. 2C), indicating that C/EBPβ was not regulated by proteasomal degradation in these cells. In contrast, MG132 had a marked effect on C/EBPβ levels in cells lacking NF-κB. Inhibition of proteasomal function in Ad5IκB-infected cells increased TNFα-induced C/EBPβ content to levels equivalent to those in TNFα-treated, Ad5LacZ-infected cells (Fig. 2C). Thus, despite the fact that the TNFα-induced increase in C/EBPβ depended in part on protein synthesis (Fig. 2B), the up-regulation of C/EBPβ levels by TNFα treatment was largely dependent on an NF-κB–dependent inhibition of C/EBPβ protein degradation.

As previous studies have demonstrated that JNK overactivation resulting from a block in NF-κB signaling alters protein degradation,19, 20 the possible involvement of JNK in the increased degradation of C/EBPβ with NF-κB inhibition was examined. Pretreatment of cells with the pharmacological JNK inhibitor SP60012532 failed to reverse the block in C/EBPβ up-regulation that occurred in the absence of NF-κB signaling (data not shown). Taken together, these findings demonstrate that the up-regulation of hepatocyte levels of C/EBPβ in response to TNFα is dependent on NF-κB-mediated inhibition of proteasomal degradation by a JNK-independent mechanism.

C/EBPβ Regulates Hepatocyte Death from TNFα.

Studies in nonhepatic cells have demonstrated an antiapoptotic function for C/EBPβ.22–24 The ability of proteasomal inhibition to increase levels of C/EBPβ led us to investigate whether MG132 was able to block hepatocyte death from TNFα. Despite mild toxicity from MG132 treatment alone, proteasomal inhibition significantly decreased the amount of cell death in Ad5IκB-infected cells treated with TNFα (Fig. 3A). To determine whether C/EBPβ functioned to prevent RALA hepatocyte death from TNFα, the effect of C/EBPβ overexpression on TNFα-induced apoptosis in RALA hepatocytes with an inhibition of NF-κB activation was assessed. Cells infected with the C/EBPβ-expressing adenovirus WT-C/EBPβ alone or coinfected with WT-C/EBPβ and either Ad5LacZ or Ad5IκB expressed increased levels of C/EBPβ compared with cells infected with Ad5LacZ alone (Fig. 3B). Cells were coinfected with Ad5IκB and either Ad5LacZ as a control or WT-C/EBPβ and treated with TNFα. When compared with Ad5IκB/Ad5LacZ-coinfected cells, the amount of cell death after TNFα treatment was significantly decreased in Ad5IκB/WT-C/EBPβ–coinfected cells at 6 and 12 hours by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (Fig. 3C). The ability of C/EBPβ expression to block cell death from TNFα was confirmed by fluorescence microscopic studies of cells costained with acridine orange/ethidium bromide to quantify the numbers of apoptotic and necrotic cells. As previously established, death from NF-κB inactivation and TNFα was predominantly apoptotic, and no significant increase occurred in the numbers of necrotic cells (data not shown). The marked increase in apoptotic cells with TNFα administration was significantly reduced by adenoviral expression of C/EBPβ (Fig. 3D). Thus, the NF-κB–dependent increase in C/EBPβ in TNFα-treated RALA hepatocytes is a mechanism of cellular resistance to TNFα-induced apoptosis.

Figure 3.

C/EBPβ blocks RALA hepatocyte death from TNFα. (A) Percentage cell death at 6 hours by MTT assay in Ad5IκB-infected cells treated with TNFα or MG132 (MG) alone or in combination. *P < 0.00001 versus cells treated with TNFα alone (n = 8). (B) Immunoblots of total protein from RALA hepatocytes infected with WT-C/EBPβ, Ad5LacZ, and Ad5IκB as indicated and probed for C/EBPβ or β-actin. (C) Percentage cell death by MTT assay at 6 and 12 hours after TNFα treatment in cells coinfected with Ad5IκB and either Ad5LacZ or WT-C/EBPβ *P < 0.01 versus TNFα-treated Ad5IκB/Ad5LacZ-coinfected cells (n = 3-4). (D) Percentage of apoptotic cells in Ad5IκB/Ad5LacZ (LacZ) and Ad5IκB/WT-C/EBPβ (C/EBP) coinfected cells untreated or TNFα-treated for 6 or 12 hours. *P < 0.01 versus TNFα-treated Ad5IκB/Ad5LacZ-infected cells (n = 3-4).

C/EBPβ Inhibits TNFα-Induced Caspase Activation.

The sensitization of hepatocytes to TNFα toxicity by NF-κB inhibition occurs through caspase-dependent apoptosis.17, 33 The ability of C/EBPβ to function as a caspase inhibitor suggested that the mechanism of C/EBPβ's inhibition of TNFα-induced apoptosis may be through blocking caspase activation. Adenoviral expression of C/EBPβ significantly decreased levels of activity of the initiator caspase 8 in both untreated and TNFα-treated cells in which NF-κB was inhibited by Ad5IκB (Fig. 4A). Inhibition of caspase 8 by C/EBPβ prevented TNFα-induced activation of the mitochondrial death pathway as WT-C/EBPβ decreased the amount of truncated Bid that translocated to the mitochondria and blocked the cytochrome c release from mitochondria into cytoplasm that occurred in Ad5IκB/Ad5LacZ-coinfected cells (Fig. 4B). In contrast, levels of cytochrome oxidase, a mitochondrial protein not released during apoptosis, were equivalent in Ad5LacZ- and WT-C/EBPβ–infected cells after TNFα treatment and indicated equal protein loading (Fig. 4B). As a result of the inhibition of cytochrome c release, downstream effector caspase 3 and caspase 7 activation was blocked in cells overexpressing C/EBPβ as detected by decreases in the active, cleaved caspase forms on immunoblots (Fig. 4C). The block in caspase 3 activation was confirmed through measurement of caspase 3 activity, which was significantly decreased by C/EBPβ overexpression in both untreated and TNFα-treated cells (Fig. 4D). These results indicate that C/EBPβ blocks TNFα-induced apoptosis by the inhibition of caspase activation.

Figure 4.

C/EBPβ blocks TNFα-induced caspase activation. (A) Relative levels of caspase 8 activity in untreated control (Con) and 4-hour TNFα-treated cells. Cells were coinfected with Ad5IκB and either Ad5LacZ or WT-C/EBPβ as indicated. *P < 0.0002 versus Ad5IκB/Ad5LacZ-coinfected cells with the same treatment (n = 7). (B) Immunoblots of mitochondrial (Mit) or cytosolic (Cyt) proteins from coinfected cells untreated or treated with TNFα and probed with antibodies for truncated Bid (tBid), cytochrome c (Cyt c), cytochrome oxidase (Cyt ox) or β-actin. (C) Immunoblots of total protein from cells coinfected with Ad5IκB and Ad5LacZ or WT-C/EBPβ as indicated. Cells were untreated or treated with TNFα for the indicated number of hours. Proteins were immunoblotted with antibodies for caspase 3 (Casp 3), caspase 7 (Casp 7) or β-actin. Arrows indicate the cleaved forms of caspase 3 and caspase 7. Immunoblots are representative of results from three independent experiments. (D) Relative levels of caspase 3 activity in untreated and 6-hour TNFα-treated cells coinfected with Ad5IκB and either Ad5LacZ or WT-C/EBPβ. *P < 0.0002 versus Ad5IκB/Ad5LacZ-coinfected cells with the same treatment (n = 4).

Absence of C/EBPβ Sensitizes Primary Hepatocytes to Death from TNFα.

Our findings suggested that the loss of C/EBPβ would result in an increase in hepatocyte sensitivity to TNFα. To investigate this possibility, primary hepatocytes were isolated from littermate control wild-type mice and c/ebpβ knockout mice, placed in culture, and examined for their sensitivity to TNFα-induced death. TNFα treatment alone was not sufficient to induce death in either wild-type or C/EBPβ null hepatocytes (data not shown). When the hepatocytes were sensitized to TNFα by infection with Ad5IκB, however, cell death at 10 and 24 hours in the knockout cells was two-fold greater than in wild-type cells (Fig. 5A). Knockout cells had greater levels of the cleaved active forms of caspase 3 and caspase 7 that resulted in increased caspase activity as indicated by cleavage of the caspase substrate poly(ADP-ribose) polymerase (Fig. 5B). We have therefore been able to demonstrate with both overexpression and loss-of-function approaches that C/EBPβ mediates hepatocyte resistance to TNFα cytotoxicity.

Figure 5.

C/EBPβ knockout hepatocytes undergo increased cell death from TNFα. (A) Percentage cell death by MTT assay at the indicated times after TNFα treatment in Ad5IκB-infected wild-type and c/ebpβ knockout hepatocytes. *P < 0.00005 versus untreated cells (n = 12). (B) Total protein from untreated or TNFα-treated Ad5IκB-infected primary mouse hepatocytes from wild-type (WT) and c/ebpβ knockout (KO) mice were immunoblotted for C/EBPβ, caspase 3 (Casp 3), caspase 7 (Casp 7), poly(ADP-ribose) polymerase (PARP), or β-actin. The cleaved forms of caspase 3 and caspase 7 and poly(ADP-ribose) polymerase are indicated by arrows. Data are representative of three independent experiments.

Sensitization to LPS- and TNFα-Induced Liver Injury Occurs in the Absence of C/EBPβ.

The in vivo function of C/EBPβ in LPS-induced liver injury was determined. The ability of C/EBPβ to block TNFα-dependent liver injury in vivo was examined by comparing the degree of liver injury in wild-type and c/ebpβ−/− mice after the administration of a usually nontoxic dose of LPS. Wild-type mice had normal ALT levels after treatment with low-dose LPS, but ALT levels were increased in knockout mice (Fig. 6A). Reflective of the predominantly apoptotic nature of TNFα-induced hepatocyte death, a much greater increase occurred in the numbers of terminal deoxynucleotidyl transferase–mediated dUTP nick-end labeling (TUNEL)-positive cells in LPS-treated c/ebpβ−/− mice compared with littermate controls (Fig. 6B). The steady-state numbers of TUNEL-positive cells in the liver were increased eight-fold at 6 hours and four-fold at 24 hours in null mice compared with control mice. To ensure that injury from LPS represented toxicity from TNFα, C/EBPβ null mice were examined for sensitivity to injury from TNFα. An injection of TNFα led to liver injury in knockout but not wild-type mice as demonstrated by increased serum ALT levels (Fig. 6C) and numbers of apoptotic cells (Fig. 6D) at 6 hours. C/EBPβ therefore mediates hepatocyte resistance to TNFα toxicity in vivo as well as in vitro.

Figure 6.

C/EBPβ knockout mice are sensitized to liver injury from LPS and TNFα. (A) Serum ALT levels in control wild-type (WT) and c/ebpβ knockout (KO) mice 6 and 24 hours after LPS injection. *P < 0.03, **P < 0.001 versus wild-type mice (n = 10-12). (B) Numbers of apoptotic cells by TUNEL staining in the livers of the same animals. *P < 0.0001 versus wild-type mice (n = 5-7). (C) Serum ALT levels in wild-type mice and c/ebpβ knockout mice 6 hours after injection with TNFα. *P < 0.05 versus wild-type mice (n = 4). (D) Numbers of apoptotic cells in the livers of the same animals. *P < 0.01 versus wild-type mice (n = 4). (E) Immunoblots of total hepatic protein from untreated and GalN/LPS-treated wild-type and jnk2 null mice for C/EBPβ and β-actin. The last lane contains protein from a wild-type mouse injected with LPS alone to demonstrate normal levels of C/EBPβ induction by LPS. Blots are representative of three experiments.

In the absence of NF-κB signaling, TNFα-induced JNK activation is converted from a transient to prolonged response that triggers cell death in part through altered protein degradation of antiapoptotic proteins. To examine whether the proapoptotic effects of JNK during TNFα-dependent injury in vivo are mediated via degradation of C/EBPβ, we investigated the effect of loss of jnk2 on C/EBPβ induction after GalN/LPS treatment. Our previous investigations have demonstrated that the jnk2 gene expresses the JNK isoforms that promote liver injury from GalN/LPS.34 Western blots of total liver protein from GalN/LPS-treated wild-type and jnk2 null mice for C/EBPβ revealed that the absence of jnk2 failed to reverse the GalN-induced inhibition of C/EBPβ induction by LPS (Fig. 6E). Mice null for jnk1 are not protected from GalN/LPS toxicity34 and also failed to up-regulate C/EBPβ (data not shown). Thus, consistent with the in vitro findings in cells with NF-κB inhibition, C/EBPβ degradation that occurred in vivo during GalN/LPS-induced liver injury was not mediated by JNK.


Significant progress has been made in defining the mechanisms by which hepatocytes lose resistance to TNFα toxicity and undergo TNFα-induced cell death. Critical for resistance to TNFα-induced apoptosis is the ability of the hepatocyte to activate the NF-κβ signaling pathway in response to TNFα stimulation.13-15 Prominent among the forms of hepatic injury mediated by sensitization to TNFα toxicity are those induced by hepatotoxins.1, 2 Hepatotoxins invariably impair macromolecular synthesis, suggesting that they may sensitize hepatocytes to TNFα-dependent injury from a toxin-induced block in the transcriptional or translational induction of protective signals by NF-κB. The identification of the protective protein effectors of NF-κB signaling may therefore increase our understanding of the mechanisms of toxic liver injury and suggest new therapies for its prevention.

These studies identify for the first time that C/EBPβ is an NF-κB-regulated mediator of hepatocellular resistance to TNFα toxicity. C/EBPβ is one member of a family of leucine-zipper transcription factors that regulate cell proliferation, differentiation, and metabolism through effects on gene expression. In addition to its role in transcription, Buck et al.22 have demonstrated a novel nontranscriptional function of C/EBPβ as a caspase inhibitor. In the present studies, C/EBPβ was up-regulated by LPS/TNFα in vitro and in vivo, which suggested that this protein may have an antiapoptotic function in TNFα-induced liver injury. Although TNFα has been shown to alter the subcellular localization of C/EBPβ,26-28 TNFα regulation of C/EBPβ protein levels has not been reported previously in hepatocytes. Consistent with a function for C/EBPβ as a protective factor against TNFα-induced cell apoptosis was that C/EBPβ up-regulation was NF-κB–dependent. Although LPS/TNFα increased C/EBPβ mRNA levels and protein synthesis, the primary mechanism by which NF-κB regulated cellular C/EBPβ content was through a decrease in proteasomal degradation of C/EBPβ. Findings from both gain-of-function studies in RALA hepatocytes and loss-of-function studies in primary mouse hepatocytes demonstrated that C/EBPβ mediates hepatocyte resistance to TNFα toxicity. The absence of C/EBPβ alone was insufficient to sensitize mouse hepatocytes to death from TNFα; however, the significance of this finding is unclear, because NF-κB activation occurs in primary hepatocytes in response to the stress of the liver perfusion and cell isolation.35 Other NF-κB–dependent protective factors may have been up-regulated by this perfusion-induced NF-κB activation that compensated for the loss of C/EBPβ. Alternatively, the null mice may have up-regulated other compensatory protective factors that negated the loss of C/EBPβ. Nonetheless, the findings identify C/EBPβ as a new antiapoptotic protein regulated by NF-κB at the level of protein degradation.

Confirmatory of the in vitro hepatocyte data were findings that C/EBPβ was up-regulated and functioned in hepatotoxic liver injury in vivo. Identical to results in RALA hepatocytes, hepatic C/EBPβ protein levels were markedly increased by the TNFα inducer LPS. Consistent with the ability of GalN to block the up-regulation of NF-κB–induced protective signaling, mice cotreated with GalN and LPS failed to up-regulate C/EBPβ. C/EBPβ was protective against TNFα cytotoxicity, because null mice but not wild-type mice developed liver injury from low-dose LPS or TNFα alone. Injury in C/EBPβ null mice was far less than that elicited by the combination of GalN and LPS in wild-type mice. These results suggest that C/EBPβ functions as one of a redundant set of NF-κB–regulated antiapoptotic factors in the hepatocyte. Alternatively, as with the studies in cultured hepatocytes from these mice, the null mice may have responded to the knockout of C/EBPβ by up-regulating other antiapoptotic factors in compensation for the loss of C/EBPβ that in part masked the true importance of C/EBPβ as an antiapoptotic factor in vivo.

The mechanism of the antiapoptotic effect of C/EBPβ was at least in part at the level of initiator caspase 8 activation, because C/EBPβ blocked the activation of this caspase and therefore the downstream mitochondrial death pathway and effector caspase cleavage. However, further studies must be performed to delineate the mechanism by which C/EBPβ blocks caspase 8 activation to confirm this possibility. Our finding is consistent with that of Buck et al.,22 who similarly found that C/EBPβ inhibited caspase 8 activation in hepatic stellate cells. This effect in hepatocytes appears to be specific for the TNFα death pathway. In contrast to the present finding of an antiapoptotic function for C/EBPβ in TNFα-mediated hepatocyte injury, studies in C/EBPβ null mice demonstrated that C/EBPβ promotes hepatocyte apoptosis from the Fas death receptor.25 Fas-mediated cell death is also caspase 8 mediated, yet C/EBPβ promoted this form of apoptosis. The mechanism of the differential effect of C/EBPβ on the TNFα and Fas death receptor pathways remains to be determined, but the current study suggests the interesting possibility that TNFα, through induction of C/EBPβ, may potentiate Fas toxicity.

A protective mechanism of NF-κB signaling is its inhibition of proapoptotic JNK overactivation.16-18 JNK signaling alters the half-lives of proteins that mediate hepatocyte resistance to TNFα toxicity. JNK1 has been reported to promote TNFα-induced death by mediating degradation of the antiapoptotic factor cFLIP.19 Conversely, other studies have suggested an antiapoptotic effect of JNK1 through an increase in the half-life of Mcl-1.20 NF-κβ is therefore known to regulate death from TNFα through JNK-dependent effects on protein degradation. Levels of C/EBPβ were similarly regulated through NF-κβ–dependent effects on the rate of C/EBPβ protein degradation. However, this effect was JNK-independent, because it was not blocked in vitro by pharmacological JNK inhibition. The absence of jnk2 in vivo, which prevented GalN/LPS-induced liver injury,34 also failed to restore the LPS-induced increase in C/EBPβ, indicating that jnk2 potentiation of liver injury does not occur through degradation of C/EBPβ. This study is the first to demonstrate a JNK-independent effect of NF-κB on protein degradation that modulates hepatocyte resistance to death from TNFα.

The new identification of C/EBPβ as an NF-κB–regulated antiapoptotic factor in the TNFα death pathway adds to the mechanistic complexity of TNFα-induced hepatocyte injury. This complexity results in part from the presence of both JNK-dependent and JNK-independent effects of NF-κB on proteasomal degradation. The existence of multiple mechanisms of resistance against the TNFα-activated apoptotic death pathway attests to the importance of hepatic resistance to TNFα toxicity in maintaining normal liver function.


The authors thank David Brenner for providing the Ad5LacZ and Ad5IκB adenoviruses and Xiao-Ming Yin for providing the anti-Bid antibody.