Beta-catenin–NF-κB interactions in murine hepatocytes: A complex to die for


  • Kari Nejak-Bowen,

    Corresponding author
    1. Departments of Pathology, University of Pittsburgh, Pittsburgh, PA
    • University of Pittsburgh, School of Medicine, 200 Lothrop Street, S433-BST, Pittsburgh, PA 15261
    Search for more papers by this author
    • fax: 412-648-1916;

  • Alexander Kikuchi,

    1. Departments of Pathology, University of Pittsburgh, Pittsburgh, PA
    Search for more papers by this author
  • Satdarshan P.S. Monga

    Corresponding author
    1. Departments of Pathology, University of Pittsburgh, Pittsburgh, PA
    2. Departments of Medicine, University of Pittsburgh, Pittsburgh, PA
    • University of Pittsburgh, School of Medicine, 200 Lothrop Street S-421 BST, Pittsburgh, PA 15261
    Search for more papers by this author
    • fax: 412-648-1916.

  • Potential conflict of interest: Dr. Monga consults for Bristol-Myers Squibb, Merck, and Phase Rx.

  • Supported by National Institutes of Health Grants 1R01DK62277 and 1R01CA124414 (to S. P. S. M.) and by Rango's Fund for the Enhancement of Pathology Research. This work was also supported in part by a Postdoctoral Research Fellowship from ALF/AASLD to KNB.


Wnt/β-catenin signaling plays an important role in hepatic homeostasis, especially in liver development, regeneration, and cancer, and loss of β-catenin signaling is often associated with increased apoptosis. To elucidate how β-catenin may be regulating hepatocyte survival, we investigated the susceptibility of β-catenin conditional knockout (KO) mice and their wild-type (WT) littermates to Fas and tumor necrosis factor-α (TNF-α), two common pathways of hepatocyte apoptosis. While comparable detrimental effects from Fas activation were observed in WT and KO, a paradoxical survival benefit was observed in KO mice challenged with D-galactosamine/lipopolysaccharide. KO mice showed significantly lower morbidity and liver injury due to early, robust, and protracted activation of NF-κB in the absence of β-catenin. Enhanced NF-κB activation in KO mice was associated with increased basal inflammation and Toll-like receptor 4 expression and lack of the p65/β-catenin complex in hepatocytes. The p65/β-catenin complex in WT livers underwent temporal dissociation allowing for NF-κB activation to regulate hepatocyte survival following TNF-α-induced hepatic injury. Decrease of total β-catenin protein but not its inactivation induced p65 activity, whereas β-catenin stabilization either chemically or due to mutations repressed it in hepatomas in a dose-dependent manner, whereas β-catenin stabilization repressed it either chemically or due to mutations. Conclusion: The p65/β-catenin complex in hepatocytes undergoes dynamic changes during TNF-α–induced hepatic injury and plays a critical role in NF-κB activation and cell survival. Modulation of β-catenin levels is a unique mode of regulating NF-κB activity and thus may present novel opportunities in devising therapeutics in specific hepatic injuries. (HEPATOLOGY 2013)

Hepatic inflammation is a common cause of hepatic injury and cell death, with etiologies such as viruses and alcohol. The liver is particularly susceptible to apoptosis due to a rich expression of death receptors. Of the six identified death receptors, Fas and tumor necrosis factor-α (TNF-α) are considered to have major pathologic significance in the liver.1, 2 Activation of the Fas-mediated apoptotic pathway has been linked to liver diseases such as hepatic inflammation, viral hepatitis, alcoholic hepatitis, nonalcoholic steatohepatitis, cholestasis, and Wilson's disease, while TNF-α mediated activation of apoptosis has been implicated in alcoholic hepatitis, ischemia/reperfusion, and fulminant hepatic failure.1, 2

Fas-mediated cell death, experimentally induced by intravenous injection of Jo-2 antibody, causes activation of Fas receptor and initiation of a downstream cascade that induces massive hepatocyte apoptosis. Treatment with lipopolysaccharide (LPS) also causes hepatotoxicity and lethality, which is primarily mediated by release of TNF-α from macrophages.3 However, binding of TNF-α to its receptor or binding of LPS to Toll-like receptor 4 (TLR-4) also activates nuclear factor kappa B (NF-κB),4, 5 which directly antagonizes the proapoptotic effects of TNF-α and prevents cell death.6 Thus, liver injury through the TNF-α pathway requires hepatocyte sensitization accomplished by pretreatment with D-galactosamine (GalN) that depletes uridine triphosphate and inhibits de novo RNA synthesis.7 NF-κB regulates expression of antiapoptotic genes such as IAPs, c-FLIP, TRAFs, and Bcl family members, among others.8

The Wnt/β-catenin pathway is an important player in liver biology with roles in development, regeneration, and tumorigenesis (reviewed in Nejak-Bowen and Monga9). However, little is known about its role in hepatocyte survival, although evidence exists that β-catenin ablation renders hepatocytes susceptible to apoptosis in development, regeneration, and more recently in hepatic ischemia-reperfusion injury.10 We used β-catenin conditional knockout (KO) mice and their wild-type (WT) littermates to test susceptibility to Fas and TNF-α. Whereas Fas activation had comparable effects in WT and KO mice, a paradoxical survival advantage was observed in KO mice after GalN/LPS treatment. We demonstrate that the p65/β-catenin complex in hepatocytes underwent dynamic changes to regulate NF-κB activation, and a decrease in β-catenin protein levels, both in vivo and in vitro, led to robust and protracted p65 nuclear translocation and activation. Conversely, β-catenin stabilization suppressed NF-κB activity. Thus, we provide evidence that β-catenin–NF-κB interactions may be altered in hepatic pathologies and that modulation of the complex may be uniquely exploited therapeutically for certain forms of liver injury.


ALT, alanine aminotransferase; AST, aspartate aminotransferase; CBP, β-catenin–CREB binding protein; cDNA, complementary DNA; EGFR, epidermal growth factor receptor; GalN, D-galactosamine; GS, glutamine synthetase; GSK-3β, glycogen synthase kinase-3β; H&E, hematoxylin and eosin; HCC, hepatocellular carcinoma; HGF, hepatocyte growth factor; IκB, inhibitor of κB; IHC, immunohistochemistry; KO, knockout; LiCl, lithium chloride; LPS, lipopolysaccharide; phospho-p65, Ser-536-phosphorylated p65; siRNA, small interfering RNA; TLR-4, Toll-like receptor 4; TNF-α, tumor necrosis factor-α; TUNEL, terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling; WB, western blotting; WT, wild-type.

Materials and Methods

Conditional β-catenin knockout mice (C57BL/6) were generated as described.11 Ctnnb1loxp/loxp; Alb-Cre+/− mice are referred to as KO mice and Ctnnb1loxp/loxp; Alb-Cre−/− or Ctnnb1loxp/Wt; Alb-Cre−/− mice are referred to as WT mice. All studies were approved by the University of Pittsburgh's Institutional Animal Care and Use Committee and were conducted in accordance with National Institutes of Health guidelines.

For complete methods, see the Supporting Information.


β-Catenin KO and WT Mice Exhibit Comparable Susceptibility to Fas-Mediated Apoptosis.

In agreement with our previous study,11 resting KO livers display a small increase in apoptosis, as demonstrated by an increase in caspase-3, active caspase-3, and caspase-8 proteins (Fig. 1A). To determine whether Fas may be mediating basal cell death in the absence of β-catenin, we examined changes in expression of two key receptor tyrosine kinases, epidermal growth factor receptor (EGFR) and the hepatocyte growth factor (HGF) receptor Met, as these signaling pathways are known to prevent Fas-induced liver injury and are also known β-catenin targets.11-14 We found a dramatic reduction in Met and EGFR protein in KO mice (Fig. 1B). Additionally, expression of HGF messenger RNA is up-regulated 9.27-fold in KO mice at baseline (Supporting Table 2).

Figure 1.

β-Catenin KO animals show alterations in components of the Fas pathway, which does not result in increased susceptibility to Fas-mediated apoptosis. (A) Caspase protein expression and activity was higher in KO livers than in WT livers at baseline. Actin represents loading control. (B) Expression of Met and EGFR was decreased in KO livers compared with WT livers at baseline as assessed via WB. (C) Immunoprecipitation revealed that Fas and β-catenin associated strongly in WT livers at baseline, and that this association was absent in KO livers. (D) Active caspase-3 activity was not significantly changed in Hepa 1-6 cells transfected with either WT, constitutively active, or inactive β-catenin followed by Jo-2 treatment, as measured via fluorometric assay. NS, not significant. (E) Kaplan-Meier survival analysis of WT and KO mice (n = 20) after intravenous injection with Jo-2 revealed no significant differences in overall survival between the two groups. (F) Grouping WT and KO animals treated with Jo-2 by time of death revealed that KO animals were slightly more resistant to Fas-mediated apoptosis than WT animals.

As shown previously, β-catenin is known to complex with Met,15 which in turn is known to complex with Fas13 in hepatocytes. We also observed a β-catenin/Fas complex via immunoprecipitation studies in WT livers, but not in KO livers (Fig. 1C). It has been shown that β-catenin phosphorylation by HGF/Met at tyrosine (Y) 654 and 670 dissociates it from Met.16 To determine whether mutation of β-catenin tyrosine residues destabilizes the Met/Fas/β-catenin interactions altering susceptibility of hepatoma cells to Fas-mediated apoptosis, we transfected Hepa 1-6 cells with WT, phospho-mimetic Y654/670E (glutamic acid), or phospho-null Y654/670F (phenylalanine) β-catenin followed by treatment with Jo-2 antibody. Determination of caspase-3 activity via fluorometric assay measuring cleavage of the caspase-3 peptide substrate DEVD-AFC 12 hours after Jo-2 treatment revealed insignificant differences in apoptosis between three conditions, suggesting that gain or loss of β-catenin from the Met/Fas complex does not alter susceptibility to Fas ligand (Fig. 1D).

Next, we challenged WT and KO mice with Jo-2. Insignificant differences in survival between WT and KO in response to Jo-2 were evident (Fig. 1E and F).

KO Mice Show Increased Resistance to TNF-α-Induced Apoptosis.

Next, we challenged KO and WT mice with GalN followed 30 minutes later by LPS to activate TNF-α-mediated liver injury.17, 18 As expected, all nine WT mice became lethargic and moribund approximately 6 hours after GalN/LPS administration, but surprisingly, most KO mice (14/15) survived past 6 hours, with some being uncompromised and healthy as late as 12 hours posttreatment (Supporting Table 1). Thus, although stimulation of the TNF-α pathway caused predictable morbidity in WT mice, KO mice showed a significant decrease in morbidity and mortality (Fig. 2A). The KO mice were also refractory to GalN pretreatment followed by intravenous injection of TNF-α, the major mediator of LPS-induced hepatotoxicity (data not shown).19

Figure 2.

β-Catenin KO mice are protected from injury induced via the TNF-α pathway. (A) Kaplan-Meier survival analysis of WT and KO mice after treatment with GalN/LPS revealed that KO mice survived significantly longer than their WT counterparts (n ≥ 9). **P < 0.01. (B) Gross liver specimens from WT and KO mice injected with GalN/LPS showed that WT livers became engorged with blood 6 hours postinjection, whereas the KO livers harvested at the same time appeared normal. The corresponding histology was characteristic of massive inflammation and apoptotic cell death in WT livers and near normal histology in KO livers (H&E stain) (magnification × 100). (C) Serum AST and ALT levels were significantly higher in WT mice than in KO mice 6 hours after GalN/LPS treatment (n = 3). *P < 0.05. (D) The number of apoptotic hepatocytes 6 hours after GalN/LPS injection (TUNEL IHC) was dramatically lower in KO mice compared with WT controls (magnification × 100). (E) Cleaved caspase-3 and caspase-8 were increased in WT mice compared with KO mice after 6 hours of GalN/LPS treatment, as shown via WB. GAPDH (glyceraldehyde 3-phosphate dehydrogenase) represents the loading control. (F) The level of active caspase-3 activity was increased in WT mice after 6 hours of GalN/LPS compared with KO mice as measured by fluorometric assay (n = 3). **P < 0.01.

Histological and Biochemical Analysis of WT and KO Livers After GalN/LPS Treatment Verifies that KO Mice Are Protected from Cell Death.

Livers from WT mice injected with GalN/LPS were harvested when they showed signs of morbidity and KO livers were harvested at comparable and later time points despite lack of any morbidity (Supporting Table 1). Grossly, WT livers were congested and enlarged, whereas KO livers, which were typically small and pale, remained relatively unaffected by treatment (Fig. 2B). Histologically, WT livers showed intense inflammation, massive cell death, and red blood cell sequestration in sinusoidal spaces, with only a few cells spared periportally (Fig. 2B). KO livers showed mostly healthy hepatocytes and intact liver with only occasional sinusoidal dilation (Fig. 2B). Serum biochemistry revealed a 40-fold increase in serum alanine aminotransferase (ALT) and a 20-fold increase in serum aspartate aminotransferase (AST) in WT livers compared with KO livers (Fig. 2C). Assaying the livers of both genotypes for apoptosis revealed that GalN/LPS-treated KO livers had dramatically fewer hepatocytes with terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labeling (TUNEL)-positive nuclei compared with WT livers (Fig. 2D). Finally, WT and KO livers at 6 hours were assessed for the presence of activated caspases by western blotting (WB) (Fig. 2E) and fluorometric measurement of caspase-3 activity (Fig. 2F), which showed WT livers to have significantly greater apoptosis compared with KO livers.

KO Mice Show Mild and Self-Limited injury, Whereas WT Mice Display Progressive Injury to GalN/LPS.

To characterize the initiation and progression of liver injury in WT and KO mice after GalN/LPS, livers and plasma were obtained 3, 4, and 5 hours after treatment. TUNEL assay showed few apoptotic cells in either WT or KO livers at 3 hours, whereas at 4 hours KO animals displayed more TUNEL-positive nuclei than WT livers. However, at 5 hours, TUNEL positivity in KO mice had not progressed and may have improved, whereas extensive apoptosis was evident in WT mice (Fig. 3A), which was also confirmed by hematoxylin and eosin (H&E) staining (Fig. 3B). Consistent with TUNEL, serum AST was low and comparable at 3 hours, greater in KO livers at 4 hours, and markedly higher in WT livers at 5 hours (Fig. 3C). These observations suggest comparable initiation of liver damage in KO and WT mice after GalN/LPS, and although the damage progresses in WT, it is self-limited in KO mice.

Figure 3.

Progressive damage occurs to WT livers after GalN/LPS, whereas damage in KO livers is arrested after an early onset. (A) KO livers had more TUNEL-positive cells compared with WT livers at 4 hours post-GalN/LPS. At 5 hours, there was a rapid increase in the number of apoptotic cells in WT mice compared with KO mice, whereas KO mice showed no progression in the numbers of TUNEL-positive cells (magnification × 100). (B) WT mice had significantly more damage than KO mice as early as 5 hours after GalN/LPS (H&E staining) (magnification × 100). (C) Serum AST levels were increased in KO mice at 4 hours, whereas at 5 hours the AST in WT mice surpassed that seen in the KO mice.

NF-κB and its Downstream Targets Are Up-regulated in KO Livers 6 Hours Post-GalN/LPS.

To determine the mechanism of protection in KOs, we examined the expression of NF-κB, a known antiapoptotic mediator of TNF-α injury. Six hours after GalN/LPS treatment, there was a clear increase in total p65 levels in KOs, as analyzed via WB (Fig. 4A). Similarly, we detected the presence of total and transcriptionally active Ser-536-phosphorylated p65 (phospho-p65) protein in hepatocyte nuclei of KO but not WT livers.20 An increase in glycogen synthase kinase (GSK-3β), a known NF-κB activator,21 was also observed in KO livers at 6 hours, suggesting a possible mechanism of p65 phosphorylation (Fig. 4A). Extensive cytoplasmic and nuclear p65 in KO livers was verified via immunohistochemistry (IHC) at 5 hours (Fig. 4B). NF-κB activation in KO livers at 6 hours after GalN/LPS administration was further substantiated by the increase of NF-κB downstream targets Traf-1 and Fas, as well as Stat3, which is a downstream effector of the NF-κB target gene interleukin-6 (Fig. 4C). In fact, many components of LPS-induced NF-κB protective machinery—including TLR-4 and tumor necrosis factor receptor type 1-associated DEATH domain protein (TRADD)—were increased in KO livers, whereas TNF-R1 and Fas-associated death domain protein (FADD) are unchanged (Fig. 4C). Measurement of NF-κB transcriptional activity via colorimetric assay confirmed an increase in KO livers (Fig. 4D). Additional targets, including inflammatory mediators and cytokines, were analyzed via complementary DNA (cDNA) array for NF-κB-regulated target genes and were also found to be up-regulated (Fig. 4E).

Figure 4.

Cytoprotective proteins and expression of downstream NF-κB targets are higher in KO livers than in WT livers 6 hours post-GalN/LPS. (A) WB revealed that both p65 and GSK-3β were increased in whole-liver lysates from KO mice compared with WT mice. Actin and Ponceau represent loading controls. Both p65 and GSK-3β were localized mainly to the nuclei of the KO livers, as demonstrated by WB of cell fractionation extracts. Additionally, p65 was present in its active, phosphorylated form in both the cytoplasm and nucleus of KO livers. (B) IHC for p65 revealed a marked increase in p65 expression in KO mice compared with WT mice 5 hours after GalN/LPS administration (magnification × 100). (C) WB of TNF-α/LPS pathway members and NF-κB target genes revealed that expression of targets and some effectors was higher in KO mice compared with WT mice after GalN/LPS. (D) Transcriptional activity of NF-κB was increased in KO mice compared with WT mice 6 hours after GalN/LPS, as measured via colorimetric assay. (E) cDNA analysis of selected NF-κB targets after treatment with GalN/LPS revealed that KO mice had a several-fold increase in the expression of many NF-κB-induced genes compared with WT mice. (F) Nuclear extracts from KO mice displaying differential susceptibility to GalN/LPS at 7.5 hours after treatment showed that p65 was higher in KO animals that were protected from apoptosis compared with those that were susceptible.

To further demonstrate that NF-κB plays a protective role in KO animals after LPS injury, we examined p65 nuclear expression in KO mice that displayed a range of susceptibility to GalN/LPS. As shown in Supporting Table 1, approximately 7.5 hours after GalN/LPS, six of 15 KO mice were partially susceptible to LPS-induced apoptosis, whereas nine of 15 showed prolonged survival. The KO mice that showed protection had nuclear p65 higher than those showing injury, indicating a direct correlation between p65 nuclear expression and protection from apoptosis (Fig. 4F). This finding suggests that protracted NF-κB activation following GalN/LPS injury is the mechanism of protection in β-catenin KO mice.

NF-κB Is Not Activated Basally in KO Animals.

The above observations led us to question the status of NF-κB in resting KO livers. A previous microarray analysis11 revealed an up-regulation of several TNF-α-dependent genes in KO livers at baseline (Supporting Table 2). Expression of TLR-4, whose activation induces NF-κB signaling, was also increased in KO livers at baseline (Fig. 5A). IHC for CD45, a cell surface marker of leukocytes, revealed greater numbers of inflammatory cells, including macrophages, in unstimulated KO livers (Fig. 5B).

Figure 5.

There is no basal activation of NF-κB in KO livers at baseline. (A) WB revealed a basal increase in TLR-4 in KOs. Actin and Ponceau show comparable loading. (B) KO livers contained higher numbers of CD45-positive inflammatory cells, as revealed via IHC (magnification × 200). (C) RIPA extracts from unstimulated WT and KO livers revealed that total p50/p65 was unchanged, as were downstream targets Traf-1 and Fas. Actin and Ponceau represent loading controls. (D) IHC for p65 confirmed the absence of active p65 signaling in both KO and WT livers at baseline (magnification × 100). (E) There was no difference in transcriptional activity of NF-κB between KO livers and WT livers at baseline, as measured via colorimetric assay. (F) Analysis of NF-κB antiapoptotic targets via cDNA array revealed no difference in expression between WT livers and KO livers at baseline. The scatter plot compares the expression of various known antiapoptotic genes between WT and KO mice and is graphed as fold change from WT.

We next wanted to address whether protection in KO livers was due to a basal increase in NF-κB activity. We examined whole-cell extracts from resting WT and KO livers for the expression of NF-κB subunits and downstream targets. As shown in Fig. 5C, there was no difference in total p65 or NF-κB activation between WT and KO livers at baseline. IHC also revealed an absence of activated p65 in both groups (Fig. 5D). Measurement of NF-κB transcriptional activity further confirmed these observations (Fig. 5E). Finally, expression of antiapoptotic NF-κB targets, such as IAP and Traf, were equivalent in WT and KO livers as measured by cDNA array (Fig. 5F). Therefore, despite an increase in inflammation and TLR-4, there appeared to be no frank NF-κB activation at baseline in the KO livers.

p65 Associates with β-Catenin in WT Livers but Not in KOs and May Be Part of an Innate Response to Injury.

In addition to its well-known interaction with the inhibitor of κB (IκB) complex, p65 has also been shown to physically associate with β-catenin in the context of colon, breast, and liver cancer.22, 23 To determine whether the p65/β-catenin complex is present under normal physiologic conditions in hepatocytes, we immunoprecipitated protein lysates from WT and KO livers with p65 and probed the blots for β-catenin. Fig. 6A shows coprecipitation of β-catenin and p65 in both WT and KO livers. However, the association in KO livers was dramatically reduced in KO livers, suggesting the presence of a NF-κB/β-catenin complex in hepatocytes and nonparenchymal cells.

Figure 6.

The decreased association of p65 and β-catenin in KO livers contributes to protection from injury after LPS treatment. (A) Immunoprecipitation revealed that p65 and β-catenin associate strongly in WT livers at baseline but less so in KO livers. (B) Association of p65/β-catenin decreased after administration of LPS in WT livers, especially at 1 hour and 3 hours, as assessed via immunoprecipitation. (C) Representative WB of WT liver nuclear lysates shows p65 translocation to the nucleus at 1 hour, whereas its peak activation (ser536-phospho-p65) occurred at 2 hours after LPS administration. Ponceau represents loading control. The right panel shows representative nuclear expression of ser536-phospho-p65 in approximately 50% of hepatocytes in WT at 2 hours after LPS via IHC (magnification × 200). (D) NF-κB was active in the nucleus 1 hour after LPS treatment in KO but not WT mice, as revealed via IHC for ser536-phospho-p65 (magnification × 200). (E) Representative WB shows increased p65 and p50 along with ser536-phospho-p65 in KO livers over WT livers at 1 hour after LPS. (F) Transcriptional activity of NF-κB was increased in KO livers compared with WTs 1 hour after LPS treatment, as measured via colorimetric assay. *P < 0.05.

Lack of β-Catenin in Hepatocytes Primes the Liver for NF-κB Activation.

Next, to investigate whether the p65/β-catenin complex undergoes changes and thus modulates NF-κB activation, we harvested WT livers at baseline and at 1, 2, and 3 hours after treatment with LPS only. Disruption of β-catenin and p65 association was observed as early as 1 hour after LPS (Fig. 6B) along with concomitant p65 nuclear translocation (Fig. 6C). Although p65 phosphorylation began to increase simultaneously, it peaked at 2 hours after LPS treatment, as shown by the appearance of ser536-phospho-p65 in the nuclei (Fig. 6C). IHC confirmed the presence of active p65 in approximately 50% of hepatocytes (Fig. 6C), consistent with previous reports.24, 25

We hypothesized that lack of β-catenin in hepatocytes may be lowering the threshold of p65 activation after apoptotic stimuli. To test this hypothesis, we treated both KO and WT with LPS to compare kinetics of p65 nuclear translocation and activation. While some animal-to-animal variation was evident, KO livers showed a greater increase in nuclear p65 at 1 hour after LPS treatment compared with WT livers (Fig. 6E). Additionally, at 1 hour after LPS, KO but not WT livers showed active ser536-phospho-p65 via both IHC and WB (Fig. 6D,E). These results were also confirmed by calorimetric measurement of NF-κB transcriptional activity after 1 hour of LPS, in which KO shows a significant increase over WT (Fig. 6F). Thus, loss of β-catenin lowers the threshold to prime the KO livers for early and robust p65 nuclear translocation and activation in response to TNF-α.

Modulation of β-Catenin Expression Alters NF-κB Activity in Hepatoma Cells.

To directly address how p65-β-catenin interactions may influence NF-κB activity, we first transfected HepG2 cells, which harbor a monoallelic exon-3-deleted constitutively active β-catenin,26 with control or β-catenin small interfering RNA (siRNA) concomitantly with either TOPflash (a luciferase reporter that measures β-catenin/Tcf-dependent transcriptional activation) or p65 luciferase reporter plasmid. Although RNA inhibition caused a reduction in full-length β-catenin at 48 hours, as shown by WB and TOPflash reporter assay, there was no significant change in p65 activity (Fig. 7A). While this was unexpected, further analysis of p65/β-catenin association in HepG2 cells by p65 immunoprecipitation revealed an association between p65 and the predominant truncated as well as the full-length form of β-catenin (Fig. 7B), suggesting that despite knockdown of the WT form, the presence of the truncated form was sufficient to bind and disallow p65 activation. However, when Hep3B cells that contain full-length, nonmutated β-catenin were transfected with siRNA and reporter plasmids, β-catenin was effectively suppressed, leading to a significant decrease in TOPflash reporter activity and an increase in p65 reporter activity (Fig. 7C). We next treated Hep3B cells with ICG-001, which inhibits β-catenin-CREB binding protein (CBP) association and down-regulates β-catenin-TCF4 transcriptional activity without reducing total β-catenin levels.27 While ICG-001 expectedly decreased TOPflash reporter activity, it unexpectedly reduced p65 reporter activity, which may be due to an increase in non-CBP-bound pool of β-catenin (Fig. 7D). These findings suggest that β-catenin modulation of NF-κB signaling is regulatable through manipulation of β-catenin expression; however, only agents that suppress total β-catenin levels may be useful to induce p65 activation.

Figure 7.

Modulation of β-catenin signaling in human hepatoma cell lines regulates p65 activity. (A) HepG2 cells transfected with either control or β-catenin siRNA showed no significant differences in p65 reporter plasmid activity, despite a significant decrease in TOPflash activity. **P < 0.01. WB demonstrated a decrease in full-length but not truncated β-catenin at 48 hours after siRNA treatment in HepG2 cells. GAPDH (glyceraldehyde 3-phosphate dehydrogenase) represents loading control NS, not significant. (B) Immunoprecipitation revealed that p65 associated with both the full-length and truncated form of β-catenin in HepG2 cells. (C) NF-κB activity as measured by p65 luciferase reporter was increased in Hep3B cells transfected with β-catenin siRNA. A decrease in TOPflash reporter activity confirmed knockdown of β-catenin. **P < 0.01. (D) p65 luciferase reporter activity was decreased in Hep3B cells transfected with β-catenin activity inhibitor ICG-001. Inhibition of β-catenin signaling was verified by a decrease in TOPflash reporter activity. **P < 0.01. (E) Transfection of Hep3B cells with mutant S33Y and S45Y β-catenin increases TOPflash activity but causes a decrease in p65 luciferase reporter activity compared with control. **P < 0.01. (F) Treatment of Hep3B cells with escalating doses of LiCl increased TOPflash activity but causes a decrease in p65 luciferase reporter activity compared with control. *P < 0.05; **P < 0.01.

Increased β-Catenin Levels Decrease p65 Activity and Expression in Hepatoma Cells and Hepatocellular Carcinoma Patients.

We next examined conditions in which β-catenin was overexpressed both in vitro and in vivo to determine the effect on p65 expression and activity. Hep3B cells were transfected with control plasmid or plasmid expressing constitutively active S33Y/β-catenin or S45Y/β-catenin, simultaneous with p65 or TOPflash reporters. While expression of mutated β-catenin induced TOPflash activity, it also resulted in significantly repressing p65 activity (Fig. 7E). We next treated Hep3B cells with an escalating dose of lithium chloride (LiCl), a known inhibitor of GSK-3β that in turn induces β-catenin protein stabilization. This led to a dose-dependent increase in TOPflash reporter activity and a concomitant and significant decrease in p65 reporter activity (Fig. 7F). Finally, we analyzed human hepatocellular carcinoma (HCC) tissue array via IHC. Tumor-wide glutamine synthetase (GS) staining is a good indicator of β-catenin mutations.28, 29 Of 93 HCC tumors on Biomax HCC tissue array, 30 were GS-positive, consistent with the numbers of HCC with β-catenin gene mutations (reviewed by Nejak-Bowen and Monga9). Of this subset, the majority of GS-positive HCC (63% [19/30]) were negative for p65 (Fig. 8A,B). These findings indicate that β-catenin activation in HCC negatively affects p65 expression and NF-κB activity.

Figure 8.

Overexpression of β-catenin in HCC correlates inversely with p65 expression. (A) The majority of patient HCC samples that were GS-positive were also p65-negative, as classified via IHC on Biomax HCC tissue array. Samples were grouped by p65 status secondary to GS status. (B) Representative IHC for β-catenin, GS, and p65 demonstrates the two different GS-dependent p65 classifications (magnification × 50).


β-Catenin is a crucial component of the Wnt pathway, which plays multiple roles in liver homeostasis through its regulation of proliferation, differentiation, and regeneration. However, its role in hepatic injury remains unexplored. The analysis was initiated to test two common modes of hepatocyte apoptosis: Fas- and TNF-α-mediated cell death. We have reported that β-catenin and the HGF receptor c-Met associate at the cell surface.15 c-Met sequestration of the Fas receptor that can prevent Fas-mediated apoptosis in hepatocytes has also been reported.13 We also identified the Fas/β-catenin complex in the liver. Because HGF messenger RNA up-regulation (along with a dramatic reduction in Met protein levels) was evident in KO livers, we hypothesized that basal apoptosis may be due to destabilization of c-Met/Fas/β-catenin complexes, which may enhance free-Fas levels available for engagement with Fas ligands like Jo-2. However, the KO mice were as susceptible as WT mice to Fas-activation.

Because several TNF-α-related genes were up-regulated basally in KO livers, we next reasoned that the prevalence of basal apoptosis may be a consequence of TNF-α activation. However, we found that β-catenin KO mice were paradoxically resistant to the effects of GalN/LPS-induced hepatocyte apoptosis. The time to morbidity in WT C57BL/6 mice following GalN/LPS treatment is generally 6-8 hours.17 We observed 100% of the WT mice displaying morbidity by 6 hours. In contrast, nearly all (93.3%) KO mice were still alive at this time, with times to morbidity ranging from 7.5 hours to 10.5 hours in a subset of mice, whereas many animals remained healthy until the outer limit of the predicted survival time of 12 hours. All of the hallmarks of apoptotic death that were present in the WT mice—including elevated liver enzymes, hepatic caspase activation, and TUNEL positivity—were absent or greatly diminished in KO.

To address the mechanism of relative resistance to TNF-α induced liver injury, we first proceeded to determine whether the metabolism of D-galactosamine used as a transcriptional suppressor prior to LPS administration could be a factor. We have shown a perturbation in vitamin C biosynthesis in β-catenin KO mice,30 which could result in accumulation of D-glucuronate, a precursor of vitamin C.31 D-Galactosamine is known to reduce the hepatic content of uridine diphosphate (UDP) glucose,32 which would decrease the amount of glucuronate, as well as inhibit the formation of de novo glucuronate by depleting UDP. However, pretreatment of KO and WT mice before LPS with actinomycin-D, an independent transcriptional inhibitor, in lieu of GalN, also recapitulated the observations in GalN/LPS-treated mice (data not shown).

The transcription factor NF-κB plays a key role in both innate and adaptive immunity. It plays a direct role in hepatocyte survival and regeneration33 and is known to positively regulate the transcription of antiapoptotic genes such as c-IAPs, Trafs, Bcl-XL, and c-FLIP,34 making it a likely candidate for a cytoprotective role in KO mice in response to TNF-α. Indeed, we found greater activation of NF-κB along with high expression of many of its downstream targets in KO mice after TNF-α. KO livers demonstrated an increase in basal inflammation and macrophages, which are a prominent source of TNF-α, which in turn may be due to higher total hepatic bile acids at baseline in chow-fed KO.35 In addition, levels of TLR-4, which has been shown to activate NF-κB, are higher in KO.36 These two factors may be the priming mechanisms for NF-κB activation, especially in the absence of the p65/β-catenin complex in hepatocytes. Interestingly, however, NF-κB was not active in KO under resting conditions, which may be because of negative feedback regulation due to NF-κB-dependent transcription of inhibitory targets such as IκB.37 Similarly, we observed heterogeneity in NF-κB activation, which explains interanimal variability in susceptibility of KO mice to TNF-α, although its basis remains undetermined.

Whereas others have described an association between β-catenin and p65 in cancer,22 we report this association under normal physiologic conditions and ascribe to it functionality within the cell and in the context of acute liver injury. We demonstrated that the p65/β-catenin complex changes dynamically over time in response to TNF-α stimulation and acute liver injury. Whereas basal p65/β-catenin association prevented NF-κB activation during resting conditions, upon TNF-α signaling, the p65/β-catenin complex underwent dissociation, allowing nuclear p65 translocation that regulated cell survival through expression of specific antiapoptotic downstream targets.

The relevance of p65/β-catenin association was demonstrated by manipulating β-catenin levels. We observed greater p65 activity upon silencing of the β-catenin gene. However, ICG-001, a known blocker of β-catenin's downstream transcriptional activity through blockade of the β-catenin/CBP complex, decreases p65 reporter activity.38 This is not surprising, because ICG-001 treatment does not decrease total β-catenin levels; in fact, its free pool is increased.27 This result thus shows that not all anti-β-catenin therapies will be effective in stimulating NF-κB signaling, and only those agents that decrease total β-catenin levels and not its activity alone may be useful, because it is the physical presence of β-catenin protein that directly affects p65 activity through formation of the inhibitory complex. Nonetheless, β-catenin inhibition to enhance p65 activation may be therapeutically exploitable to treat certain hepatic injuries where TNF-α signaling is the chief perpetrator.

We also demonstrate that stabilizing β-catenin by LiCl treatment represses p65 activity. Although inhibition of GSK-3β may inhibit p65 directly, this effect is likely due to increased β-catenin binding to p65 as reported.39 It is possible that the APC/Axin/GSK-3β/β-catenin complex may be in close proximity to or in direct association with the β-catenin/NF-κB/IκB complex. If true, this association could serve as a mode of cross-talk and integration of two distinct signaling pathways. Whether the p65/β-catenin complex exists as a part of a larger multimeric complex requires additional investigation.

β-Catenin gene mutations leading to its activation and nuclear translocation are frequent in HCC, and as we have shown, this in turn leads to decreased p65 activity and expression in hepatoma cells and tissues. This is in agreement with previous reports that enhanced β-catenin can bind to and inhibit NF-κB transcriptional activation in cancer cells.22, 23 Although the functional consequences of this observation need to be investigated more thoroughly, it has been reported that β-catenin may act as a negative regulator of inflammation through repression of NF-κB signaling.40, 41 This may explain why in certain cases of β-catenin-mutated HCC, lesser cirrhosis is evident, because inflammation induces hepatic injury and fibrosis.28, 42 However, the role of NF-κB in liver injury and hepatocyte survival is pleiotropic, because its activation can be contextually antiapoptotic or proinflammatory.8 Similarly, β-catenin antagonism that has been touted for cancer therapeutics9 may have unintended consequences of promoting tumor cell survival depending on NF-κB status. Further preclinical work would be necessary to determine the long-term effects of NF-κB activation in the context of β-catenin inhibition.