Nuclear factor κB (NF-κB) has a central role in coordinating the expression of a wide variety of genes that control immune responses and is also recognized as an antiapoptotic transcription factor. Here, we focused on the role of the NF-κB signaling pathway in the interaction between inflammatory cells and hepatocytes in liver inflammation. We found that pretreatment of mice with adenoviruses expressing a mutant form of the inhibitor κB superrepressor (Ad5IκB), a NF-κB inhibitor, reduced the migration of inflammatory cells and cytokine and chemokine expression in the liver 12 hours after a single intravenous injection of an anti-CD40 antibody (αCD40) compared with mice infected with control adenoviruses (Ad5LacZ). We also confirmed reductions in cytokine production by macrophages, T cells, and natural killer (NK) cells in the liver of Ad5IκB-treated mice by FACS analysis. However, αCD40 treatment in Ad5IκB-infected mice induced elevation of serum alanine aminotransferase at 24 hours, and the liver injury was associated with massive hepatocyte apoptosis. Furthermore, interferon gamma (IFN-γ) production by NK cells and T cells was increased and stimulated tumor necrosis factor alpha (TNF-α) production by macrophages in the Ad5IκB-infected liver. Moreover, the liver injury was completely suppressed by the administration of anti–IFN-γ and anti–TNF-α. These results suggest that inhibition of NF-κB activity suppressed αCD40-induced liver inflammation at an early phase, resulting in a reduction in cytokine and chemokine production, whereas it sensitized hepatocytes to TNF-α–induced apoptosis and exacerbated liver injury at the late phase. In conclusion, NF-κB exerts pivotal activities at inflammatory sites, and caution should be exercised in NF-κB–targeted therapy of liver disease. (HEPATOLOGY 2004;40:1180–1189.)
Nuclear factor kappa B (NF-κB) is a ubiquitous transcription factor that is activated by a variety of cytokines and mitogens1 and is thought to be a key regulator of genes involved in inflammation, responses to infection, and stress. The targets for transcriptional activation by NF-κB include genes for cytokines, acute phase response proteins, immunoglobulins, and cell adhesion molecules.1, 2 NF-κB is one of the most important regulators of proinflammatory gene expression. The synthesis of cytokines such as tumor necrosis factor alpha (TNF-α), interleukin (IL)-1, IL-6, and IL-8 is mediated by NF-κB.1, 3 NF-κB activation also increases the expression of the adhesion molecules E-selectin, VCAM-1, and ICAM-1, while NF-κB inhibition reduces leukocyte adhesion and transmigration.4 Classic NF-κB is a heterodimer of p50 and p65 (Rel-A), but the proteins that constitute the NF-κB family form a variety of homodimers and heterodimers. NF-κB is retained in an inactive form in the cytoplasm through association with one of the inhibitor κB (IκB) proteins, including IκBα or IκBβ.3 After cellular stimulation, the phosphorylation, ubiquination, and subsequent proteolysis of IκB in proteosomes enables NF-κB to translocate into the nucleus, where it regulates the transcription of NF-κB-responsive genes by interacting with B binding sites.1, 3, 5
Recently, many studies have clearly shown that NF-κB activation is involved in apoptosis,1, 3, 4 although its role is still unknown. Nevertheless, NF-κB activation may lead to induction of apoptosis in some cell types via activator protein 1 signaling. However, in other lineages such as hepatocytes and synovial cells, NF-κB plays an antiapoptotic role.6, 7 Thus, it is generally accepted that NF-κB plays a pivotal role in the crossroads between life and death in a variety of cells. The first evidence for a critical antiapoptotic role of NF-κB in the liver was reported by Beg et al., who demonstrated that mice lacking the p65 subunit of NF-κB died embryonically accompanied by massive TNF-induced destruction of the liver via apoptosis.8 In the absence of NF-κB activity, cellular susceptibility to TNF-induced apoptosis increases, whereas enforced activation of NF-κB protects against apoptosis. Therefore, the exact relationship between NF-κB and inflammation must be identified after due consideration of the roles of NF-κB in inflammatory cells and target cells.
CD40 was first identified in B lymphocytes but has since been found on many other cell types, including macrophage lineage cells, stromal cells, endothelium, and epithelium.9–13 There is now compelling evidence that CD40 is pivotal in regulating inflammatory responses, and recent studies have shown widespread expression of CD40 during inflammatory liver diseases, including allograft rejection, autoimmune disease, and viral hepatitis.14–19
In this study, we determined the role of NF-κB in anti-CD40–induced liver injury, which depends on activated macrophages in the liver. Here, we show that inhibition of NF-κB suppressed cytokine and chemokine production by inflammatory cells at an early phase, but reinforced liver injury via massive apoptosis at the late phase. These results suggest that NF-κB has a dual role in the regulation of liver inflammation and provide useful information for therapeutic strategies against NF-κB.
The CB6F1 mice used in this study were purchased from The Japan SLC (Shizuoka, Japan). In all experiments, the mice were matched for age (7-8 weeks) and sex (male). All animals were housed in pathogen-free rooms under strict barrier conditions and received humane care according to the guidelines of the Animal Care Committee of Gifu University School of Medicine.
Anti-CD40 and Anti-cytokine Antibodies.
The FGK45 hybridoma producing a rat immunoglobulin (Ig) G2a monoclonal antibody against mouse CD40 (αCD40) was kindly provided by Dr. A. Rolink (Basel Institute for Immunology, Basel, Switzerland).20 The mice were injected intravenously with either 100 μg of αCD40 or 100 μg of purified rat IgG2a (BD Pharmingen, San Diego, CA) as a control antibody. Mice were injected intraperitoneally (250 μg/mouse) at day 0 and day 2 with: (1) a hamster monoclonal antibody (mAb) specific for murine interferon gamma (IFN-γ); (2) a hamster mAb specific for murine TNF-α; or (3) a control hamster IgG. All antibodies were purchased from Genzyme (Cambridge, MA).
Adenovirus Construction and Infection.
A recombinant replication-deficient adenovirus expressing mutant inhibitor κB (Ad5IκB)21 and the control adenovirus (Ad5LacZ) were kindly provided by Dr. D. Brenner (Columbia University, New York, NY). The viral solution containing 2 × 109 plaque-forming units (0.2 mL) was injected into mice via the tail vein. Three days after the viral infection, the mice were treated intravenously with an αCD40 injection.
Nuclear Extract Preparation and Electrophoretic Mobility Shift Assay.
As described previously,21 nuclear proteins were prepared from liver tissues using a modification of a method by Dignam et al.4 Protein concentrations were measured using the Bradford method. A double-stranded NF-κB consensus oligonucleotide probe (5′-TAGTTGAGGGGACTTTCCCAGGCA-3′) was labeled with [32P]-deoxycytidine 5′-triphosphate (ICN Biomedicals, Costa Mesa, CA) using Klenow DNA polymerase. Protein–DNA binding reactions were performed for 30 minutes at room temperature using 20 μg of nuclear proteins and 2.5 ng of 32P-labeled, double-stranded oligonucleotide probe.
Tissue DNA and RNA Analyses.
Frozen liver was mechanically pulverized under liquid nitrogen and total RNA was isolated for ribonuclease protection assay (RPA) as described previously.15 All reagents for RPA were purchased from BD Pharmingen.
Biochemical and Histological Analyses.
The extent of hepatocellular injury was monitored histologically and biochemically by measuring serum alanine aminotransferase (sALT) activity at multiple time points after the injection. sALT activity was measured using a standard clinical automatic analyzer. For histological analysis, liver tissue was fixed in 10% zinc-buffered formalin, embedded in paraffin, sectioned (3 μm) and then stained with hematoxylin-eosin.
Serum TNF-α Concentration.
Serum TNF-α concentrations were assayed using an enzyme-linked immunosorbent assay kit (Genzyme Techne Co., Minneapolis, MN) according to the manufacturer's protocol. Absorbance was measured at 450 nm for both the standards and experimental samples. The cytokine concentrations in the experimental samples were calculated by reference to the standard curve.
Terminal Deoxynucleotidyl Transferase-Mediated dUTP Nick End Labeling Assay.
Apoptotic cells were estimated using terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay, which relies on the incorporation of labeled dUTP at sites of DNA breaks. All the reagents, including the buffers, were part of a TUNEL assay kit (Apop Tag; Oncor, Gaithersburg, MD), and the procedure was performed according to the manufacturer's instructions.
Preparation of Cells and Hepatocytes.
To isolate intrahepatic leukocytes (IHLs), perfused livers were digested with 10 mL of RPMI 1640 (Life Technologies, Gaithersburg, MD) containing 0.02% (wt/vol) collagenase IV (Sigma-Aldrich, St. Louis, MO) and 0.002% (wt/vol) DNase I (Sigma-Aldrich) for 40 minutes at 37°C. Cells were washed with RPMI 1640 and then overlaid on 24% (wt/vol) metrizamide (Sigma-Aldrich) in phosphate-buffered saline. After centrifugation for 20 minutes at 1,500g, IHLs were isolated at the interface.
To prepare hepatocytes, livers were perfused sequentially with 5 mL Hank's balanced salt solution without Ca2+ and Mg2+, containing 1% wt/vol glucose and 0.5 mmol/L egtazic acid, followed by 10 mL collagenase-(hepatocyte-qualified) solution (GIBCO BRL, Grand Island, NY). Livers were removed and filtered through a sterile 70-μm cell strainer. The cells were washed twice via centrifugation at 50g for 2 minutes and resuspended in RPMI 1640 containing 10% fetal calf serum.
Single-cell suspensions of IHLs were washed in phosphate-buffered saline (containing 1% BSA and 0.02% sodium azide), and incubated for 20 minutes on ice with a culture supernatant from the hybridoma cell line 2.4G2 (American Type Culture Collection) to block Fc receptors. The cells were surface-stained with a fluorochrome-conjugated monoclonal antibody for 20 minutes on ice. Anti-CD3, anti-NK1.1, anti-CD11b, anti-CD11c, and anti–Gr-1 antibodies were used, along with IFN-γ– and TNF-α–allophycocyanin for intracellular cytokine detection (all from BD Pharmingen). Samples were acquired using a FACScalibur flow cytometer (BD Immunocytometry Systems, San Jose, CA), and the data were analyzed using CELLQuest software (BD Immunocytometry Systems).
Annexin V Staining.
Apoptosis of IHLs was determined using annexin V-FITC and Via-Probe (a 7-AAD (7-amino-actinomycin D) viability probe) (Pharmingen, La Jolla, CA) according to the manufacturer's instructions.
All values in the figures and text are expressed as the mean ± SD. The significance of differences among mean values was evaluated using the Mann-Whitney U test.
Role of NF-κB in αCD40-Induced Liver Inflammation.
We have already shown that αCD40-activated antigen-presenting cells, especially macrophages, trigger liver injury by producing inflammatory cytokines.15 To determine the role of NF-κB in this liver injury model, we used an adenovirus expressing the IκB superrepressor (Ad5IκB), which inhibits NF-κB DNA binding activity and transcriptional activity.21, 22 Seven-week-old male CB6F1 mice received an intravenous injection of αCD40 (100 μg) or rat IgG (100 μg) following pretreatment with Ad5IκB or control adenoviruses (Ad5LacZ), and were sacrificed at 4, 12, 16, and 24 hours after the injection. As shown in Fig. 1A, there was a slight increase in the sALT activity in the Ad5IκB- and Ad5LacZ-infected mice after rat IgG injection as demonstrated previously.23 We found that sALT activity levels were similar between the Ad5IκB- and Ad5LacZ-infected mice 12 hours after the αCD40 injection (75.7 ± 22.5 and 49.7 ± 6.1, respectively). However, the sALT activity was remarkably elevated in the Ad5IκB-infected mice at 24 hours after the αCD40 injection compared with the Ad5LacZ-infected mice (1170 ± 287.5 and 215.7 ± 34.6, respectively; *P < .01) (see Fig. 1A).
To determine the effect of NF-κB activity on the inflammatory cell infiltration in the same livers, we counted the absolute number of IHLs and calculated the number of cells in each IHL subset at each time point via FACS analysis. As shown in Fig. 1B, the total numbers of IHLs were at the same level for the Ad5IκB- and Ad5LacZ-infected mice after rat IgG injection. On the other hand, the total number of IHLs rose markedly after αCD40 injection in Ad5LacZ-infected mice. This increase in IHLs was inhibited by Ad5IκB infection at 4, 12, 16, and 24 hours (P < .05) (Fig. 1B), indicating that NF-κB activity is involved in the infiltration of inflammatory cells into the liver. We also showed that the cell number in all subsets of IHLs was suppressed in the Ad5IκB-infected mice 12, 16, and 24 hours after the αCD40 injection. It is of note that the number of IHLs was still low at 24 hours after the αCD40 injection, even though severe liver injury had occurred in the Ad5IκB-infected mice to a greater extent at this time point.
Next, we analyzed whether or not αCD40 induced NF-κB activation in the liver and whether NF-κB activity was inhibited in the liver of Ad5IκB-infected mice by an electrophoretic mobility shift assay. Nuclear extracts from the IHLs and hepatocytes were used for the assay to test the induction of NF-κB activity. As shown in Fig. 1C, we observed induction of NF-κB activity in the IHLs and hepatocytes of the Ad5LacZ-infected mice 12 and 24 hours after the αCD40 injection. As expected, we observed inhibition of NF-κB activation in the IHLs and hepatocytes of Ad5IκB-infected mice 12 and 24 hours after the αCD40 injection (Fig. 1C).
Furthermore, to determine cytokine and chemokine expression in the liver in the presence or absence of NF-κB activity, we conducted a RPA of the total liver RNA extract. As shown in Fig. 1D, TNF-α, macrophage inflammatory protein 2, CXC chemokine ligand 10, and CC chemokine ligand 5 messenger RNA (mRNA) expression was upregulated in the livers of Ad5LacZ-infected mice 12 and 16 hours after injection, while the expression of these mRNAs was reduced in the Ad5IκB-infected mice. In contrast, IFN-γ mRNA expression in the liver was detected at the same level 12 hours with or without NF-κB activity. These results suggest that inhibition of NF-κB activity suppressed inflammatory cytokine and chemokine expression in the liver at this time point. However, we showed that TNF-α, IFN-γ, CXC chemokine ligand 10, CC chemokine ligand 4, CC chemokine ligand 3, and macrophage inflammatory protein 2 mRNA expression was increased in the livers of Ad5IκB-infected mice at 24 hours after the αCD40 injection to a greater extent than in the Ad5LacZ-infected mice. These results coincided with the high level of sALT activity at 24 hours (Fig. 1D, bottom), indicating that the upregulation of particular inflammatory cytokines and chemokines may be associated with liver injury.
Production of Inflammatory Cytokines.
To determine which cell population produced IFN-γ and TNF-α after the αCD40 injection, we stained the intrahepatic macrophages (CD11b+), natural killer (NK) cells (NK1.1+), and T cells (CD3+) with antibodies to IFN-γ and TNF-α using an intracellular cytokine detection method. As shown in Fig. 2, NK cells in Ad5IκB-infected mice produced a small amount of IFN-γ after the rat IgG injection, while those in Ad5LacZ-infected mice did not. In contrast, IFN-γ was produced by NK cells and T cells in the Ad5LacZ-infected mice 12 hours after the αCD40 injection. Consistent with RPA results, inhibition of the NF-κB activity reduced the production of IFN-γ by NK cells and T cells in the liver 12 hours after the injection. Although the macrophage populations in the liver produced a small amount of TNF-α after the rat IgG injection in either Ad5IκB- or Ad5LacZ-infected mice, a high level of TNF-α was detected in the Ad5LacZ-infected mice 12 hours after the αCD40 injection (see Fig. 2). In a similar manner to NK cells and T cells, TNF-α production by macrophages was also suppressed when NF-κB activity was inhibited.
On the other hand, at 24 hours after the injection the NK cells and T cells in the Ad5IκB-infected mice produced a larger amount of IFN-γ. However, we found that the macrophages in Ad5IκB-infected mice produced TNF-α at a lower level than those in the control Ad5LacZ-infected mice.
Serum Level of TNF-α.
To examine the inflammatory cytokine production systematically, we measured serum TNF-α levels in the Ad5IκB- and Ad5LacZ-infected mice 12 and 24 hours after the αCD40 or rat IgG injection. As shown in Fig. 3, the serum TNF-α concentration was reduced in the Ad5IκB-infected mice 12 hours after the αCD40 injection compared with the Ad5LacZ-infected mice (P < .05). On the other hand, 24 hours after the αCD40 injection, serum TNF-α concentration was increased in the Ad5IκB-infected mice (P < .05).
Ad5IκB Treatment Induced Apoptosis of Hepatocytes After αCD40 Injection.
To determine the histological changes in the livers of Ad5IκB- and Ad5LacZ-infected mice after αCD40 or rat IgG injection, we stained liver tissues with hematoxylin-eosin and in situ TUNEL. The histological analysis revealed widely scattered inflammatory foci in the liver parenchyma and around the portal tract containing mostly lymph mononuclear cells and a few apoptotic hepatocytes in the Ad5LacZ-infected mice 12, 16, and 24 hours after the αCD40 injection (Fig. 4A). On the other hand, in the Ad5IκB-infected mice, liver necrosis was detected in the parenchyma (see Fig. 4A) and lymph mononuclear cells infiltrated around the necrotic area. In addition, the Ad5IκB-infected livers demonstrated a marked increase in the number of TUNEL-positive cells after the αCD40 injection (see Fig. 4A), as quantified in Fig. 4B. These results suggested that the Ad5IκB-infected livers underwent massive apoptosis of hepatocytes after the αCD40 injection, and this observation was associated with the high elevation of sALT activity. It is noteworthy that the absolute cell number of the inflammatory cell infiltrate was reduced in the Ad5IκB-infected livers after αCD40 treatment compared with the Ad5LacZ-infected livers, as shown in Fig. 1B. Collectively, blocking of NF-κB activity by the IκB superrepressor was associated with inhibition of inflammatory cell recruitment but with a strong stimulus for hepatocyte apoptosis after the αCD40 injection.
Does Ad5IκB Treatment Induce Apoptosis of Intrahepatic Leukocytes?
We showed that Ad5IκB treatment suppressed the migration of inflammatory cells into the liver after the αCD40 injection (Fig. 1B). To test the possibility that the inflammatory cells underwent apoptosis, we analyzed annexin V and 7-AAD expression in IHLs at 16 and 24 hours after the αCD40 injection (Fig. 5). Ad5IκB infection increased annexin V+/7-AAD− fraction in IHLs to a greater extent than Ad5LacZ infection 24 hours after the αCD40 injection. These results suggest that inhibition of NF-κB activity enhanced the apoptosis of IHLs as well as hepatocytes (see Fig. 4), and furthermore that IHLs can also be induced to undergo apoptosis by cytokines as reported previously.24
Roles of IFN-γ and TNF-α.
To evaluate which cytokine was responsible for the induction of apoptosis in the liver, we injected CB6F1 mice with either anti–IFN-γ or anti–TNF-α mAb, or an irrelevant hamster IgG, at −2 and 0 days relative to the injection of αCD40 in Ad5IκB- or Ad5LacZ-infected mice. These mice were sacrificed 24 hours after the injection, and the sALT activity was measured. As shown in Fig. 6A, the administration of anti–IFN-γ and anti–TNF-α mAb completely protected against liver injury after αCD40 injection in Ad5IκB-infected mice, indicating that IFN-γ and TNF-α were responsible for the αCD40-induced liver injury. Consistent with this result, anti–IFN-γ and anti–TNF-α mAb treatment completely blocked the induction of apoptosis in the liver 24 hours after the αCD40 injection, as assessed with TUNEL assays (data not shown). These results demonstrate that IFN-γ and TNF-α signaling pathways activated the αCD40-induced hepatocyte apoptosis.
As shown in Figs. 1D and 2, we found that IFN-γ production by NK or T cells was increased 24 hours after αCD40 injection under NF-κB inactivation. We analyzed TNF-α expression by macrophages 24 hours after injection of αCD40 with anti–IFN-γ mAb in Ad5IκB-infected mice to determine whether IFN-γ affected TNF-α production by these cells. As shown in Fig. 6B, we found that TNF-α expression by macrophages after the injection of αCD40 was remarkably inhibited with anti–IFN-γ mAb in NF-κB-inactivated mice, indicating that IFN-γ stimulated TNF-α production at this time point.
Recent studies have demonstrated that NF-κB plays important roles in the resolution of lung inflammation25 and the control of inflammation in intestinal ischemia-reperfusion injury.26 It has also been well established that hepatocytes become sensitized to cell death through the inactivation of NF-κB.3 However, the interaction between NF-κB and liver injury caused by inflammatory cells has not yet been investigated in vivo. Here, we evaluated whether NF-κB activity was involved in the regulation of liver inflammation in αCD40-induced liver injury. We have already shown that this type of liver injury depends on activated intrahepatic macrophages and NK cells. In addition, this liver injury has been shown to be mediated by IFN-γ and TNF-α.15 In this study, we found that inhibition of NF-κB activity has the dual effect of suppression of cell infiltration and exacerbation of apoptosis at inflammatory sites in the αCD40-induced liver injury model.
We showed that inhibition of NF-κB activity initially exerted an immunosuppressive role by reducing cytokine production by inflammatory cells at the 12-hour time point (see Fig. 2). It has been clearly established that NF-κB is one of the most important regulators of proinflammatory gene expression.1, 2 Actually, NF-κB signaling is known to regulate the synthesis of IL-1, IL-6, and TNF-α.5 It has also been shown that CD40 signaling activates the NF-κB pathway.10, 12 In this context, the production of inflammatory cytokines and chemokines derived from αCD40-activated macrophages was suppressed by inhibition of NF-κB activity at this time point. We suggest that inflammatory cells, including NK cells and T cells, also show reductions in migration and cytokine production as the same consequence. However, further experiments will be required to confirm such direct effects of Ad5IκB on T cells or NK cells, although adenoviruses efficiently infect many different organs and cells in vivo, including macrophages and fibroblasts.27 Consistent with these results, we showed that mRNA expression of inflammatory cytokines and chemokines in the whole liver was reduced at 12 and 16 hours in the Ad5IκB-infected mice compared with the Ad5LacZ-infected mice (see Fig. 1D). The changes in serum TNF-α concentration also support the suppression of αCD40-induced inflammation in the Ad5IκB-infected mice at this time point. Collectively, these data suggest that NF-κB activity plays an important role in cytokine and chemokine production by inflammatory cells after αCD40 injection. However, it is noteworthy that even though the absolute early cytokine production is suppressed in the Ad5IκB-infected mice at 12 and 16 hours, sALT activity had already been increased, indicating that the destruction of hepatocytes was beginning at this time (see Fig. 1D, bottom).
On the other hand, liver injury was exacerbated 24 hours after the αCD40 injection under inhibition of the NF-κB activity, as indicated by the remarkable increase in sALT activity in the Ad5IκB-infected mice. The mRNA expression of intrahepatic inflammatory cytokines and chemokines was increased in the Ad5IκB-infected mice compared with the Ad5LacZ-infected mice at this time point (see Fig. 1D), demonstrating a completely reversed condition compared with that at 12 and 16 hours. In addition, we also showed that this severe liver injury was associated with the accumulation of apoptotic hepatocytes in the parenchyma (see Fig. 4). Such an observation in this model supports that inhibition of NF-κB activity sensitizes hepatocytes to apoptosis.22 Taking the findings at 12 and 16 hours together, inflammatory cytokines were suppressed in the liver under the inhibition of NF-κB activity, but severe liver injury had occurred by 24 hours nonetheless. We propose a possible mechanism to explain such dual observations (Fig. 7):
1a lower amount of TNF-α is produced from the Ad5IκB-infected macrophages after αCD40 stimulation;
2hepatocytes prepare for sensitization as a result of the inhibition of NF-κB activity;
3TNF-α is sufficient to cause apoptosis of the sensitized hepatocytes; and
4apoptotic hepatocytes trigger cytokine production from macrophages, exacerbating severe liver injury.
This hypothesis, particularly step 4, is supported by a recent report that hepatocyte apoptosis and phagocytosis of apoptotic bodies by Kupffer cells promote liver injury by stimulating these cells to further express death ligands and cytokines.28
It is curious that chemokine expression was upregulated in the liver at 24 hours in the Ad5IκB-infected mice, but the number of infiltrating inflammatory cells was reduced compared with the Ad5LacZ-infected mice (see Fig. 1B). We consider this result as two hypotheses: (1) Ad5IκB-infected inflammatory cells are sensitive to apoptosis; and (2) after 24 hours, inflammatory cells migrate into the liver as a result of chemokine expression. Although further experiments are required to clarify these hypotheses, we found a clue to support the former in that the Ad5IκB-infected IHLs were strongly annexin V+/7-AAD− compared with the Ad5LacZ-infected IHLs (see Fig. 5). In addition, we showed that cell recruitment into the liver was reduced at 16 hours as well as 12 hours in Ad5IκB-infected mice (see Fig. 1B). Consistent with the results of the RPA (Fig. 1C), we consider that the reduced migration of IHLs is due to chemokine suppression by NF-κB inactivation until 16 hours after the injection, whereas the reduction in IHL recruitment at 24 hours is due to increased apoptosis of IHLs.
Furthermore, we found the unexpected result that anti–IFN-γ treatment completely protected against liver injury and induction of apoptosis after inhibition of NF-κB activity. Actually, anti–IFN-γ treatment blocked TNF-α production by macrophages (see Fig. 6). This result demonstrates that IFN-γ plays an important role in the stimulation of TNF-α production by macrophages, indicating that crosstalk between NK cells or T cells, which are the main IFN-γ producers, and macrophages is an important event for causing severe liver injury in this model. Interestingly, although the absolute numbers of NK cells and T cells were reduced at 24 hours in the Ad5IκB-infected mice (see Fig. 1B), we found that these cells produced much more IFN-γ (see Fig. 2). Although we have no direct evidence, a possible hypothesis is that IFN-γ production by these cells might be stimulated by cytokines (e.g., IL-12 and IL-18 produced by macrophages), which are potent IFN-γ inducers, as described previously.29, 30 We also found another unclear result: inflammatory cytokine production by NK cells and T cells was increased 24 hours after the injection of αCD40, despite suppressed NF-κB activity in IHLs (see Fig. 1C). Future experiments will be required to elucidate the mechanism.
NF-κB is highly activated at sites of inflammation in a wide range of diseases, including rheumatoid arthritis, inflammatory bowel disease, multiple sclerosis, psoriasis, and asthma.31 These changes are accompanied by enhanced recruitment of inflammatory cells and production of proinflammatory mediators. Specific inhibition of NF-κB activity has been shown to be consistently effective at controlling inflammatory diseases in several animal models.32–34 It has been demonstrated that blocking NF-κB activity by overexpression of IκB inhibits both the inflammatory response and tissue destruction in the rheumatoid synovium.35 Administration of NF-κB decoys also seems to be effective in animal models of rheumatoid arthritis.6 Because TNF-α is believed responsible for liver inflammation, including viral, autoimmune, and alcoholic hepatitis,36–38 blocking upstream NF-κB activity might be one of the ideal therapeutic strategies for these diseases.
Overall, our findings provide several important insights into the interaction between NF-κB and liver injury. Inhibition of NF-κB activity may be harmful to the host defense and may exacerbate liver injury even though the absolute early cytokine production by inflammatory cells is suppressed, because hepatocytes become susceptible to TNF-α–induced apoptosis. Thus, care must be taken to use an anti-inflammatory intervention to inhibit NF-κB activity in the case of liver disease.
The authors thank Dr. Antonius Rolink (Basel Institute for Immunology, Basel, Switzerland) for providing the anti–mouse agonistic CD40 monoclonal antibody; Dr. David Brenner (Columbia University, New York, NY) for providing the adenoviruses; Tomomi Morita and Mika Izawa for technical assistance; and Dr. Motoshi Sawada for helpful discussions.