Hepatocyte-specific inhibitor-of-kappaB-kinase deletion triggers the innate immune response and promotes earlier cell proliferation during liver regeneration


  • Potential conflict of interest: Nothing to report.


Nuclear factor κB (NF-κB) is one of the main transcription factors involved in liver regeneration after partial hepatectomy (PH). It is activated upon IκB phosphorylation by the IκB kinase (IKK) complex comprising inhibitor of kappaB kinase 1 (IKK1), inhibitor of kappaB kinase 2 (IKK2), and nuclear factor-B essential modifier (NEMO). We studied the impact of hepatocyte-specific IKK2 deletion during liver regeneration. A 70% PH was performed on IKK2f/f (wild-type) and IKK2ΔLPCmice (hepatocyte-specific IKK2 knockout mice). PH in IKK2ΔLPC compared with IKK2f/f mice resulted in weaker and delayed NF-κB activation in hepatocytes, while nonparenchymal liver cells showed earlier NF-κB activation and higher tumor necrosis factor expression. Additionally, these animals showed increased and earlier serum amyloid A and chemotactic cytokine L-1 levels followed by enhanced polymorphonuclear cell recruitment to the liver. These results correlated with earlier Jun kinase activity, c-myc expression, and matrix metalloproteinase-9 activity, suggesting earlier priming in IKK2ΔLPC mice after PH. These data preceded a more rapid cell cycle progression and earlier hepatocyte proliferation as evidenced through cyclin and 5-bromo-2-deoxyuridine analysis. Interestingly, despite faster G1/S progression, IKK2ΔLPC mice exhibited an enduring mitosis phase, because mitotic bodies were still observed at later stages after PH. Conclusion: We demonstrate that PH in IKK2ΔLPC mice triggers a more rapid and pronounced inflammatory response in nonparenchymal liver cells, which triggers earlier hepatocyte proliferation. (HEPATOLOGY 2008.)

The healthy liver is a quiescent organ with low cell turnover. However, after injury caused by surgical resection, chemicals, or viral infections, the liver can restore its mass and cell function upon hepatocyte proliferation. The most common method of studying liver regeneration in rodents is partial hepatectomy (PH) of two-thirds (70%) of the liver. Liver regeneration after PH is a compensatory hyperplasic response and not a complete restoration of the resected lobes.1 PH causes a significant stress to the liver as it dramatically disrupts organ homeostasis, and the proinflammatory response, triggered by surgery, primes the remnant hepatocytes to proliferate and recover liver mass.1

Activation of the innate immune system through Toll-like receptor (TLR) and other pattern-recognition molecules plays a key role in host defense against invading pathogens.2 Recent studies have highlighted the crucial contribution of these mechanisms to the inflammatory response after PH.3–5 During the priming phase of liver regeneration, Kupffer cells are activated and secrete proinflammatory cytokines, most prominently tumor necrosis factor (TNF)-α, interleukin (IL)-6, and IL-1β, which can initiate the acute phase response in hepatocytes.1, 6, 7 Serum amyloid A (SAA), C-reactive protein, and the complement system are proteins involved in this physiologic response to restore liver homeostasis.6–8 Proinflammatory cytokine secretion from Kupffer cells during liver injury requires nuclear factor κB (NF-κB) activation and triggers a cascade leading to influx of inflammatory cells.9

Lipopolysaccharide, IL-1β, or TNF can induce production of chemotactic cytokines (chemokines [CXC]) such as CXC ligand-1 (CXCL-1 or KC). These chemokines belong to the Glu-Leu-Arg positive (ELR+) CXC family, which differ from ELR CXC chemokines both structurally (due to the presence of a three–amino acid sequence [Glu-Leu-Arg]) and functionally, because they can exert antagonistic effects.10 ELR+ CXC chemokines contribute to wound healing and resolution of inflammatory events by stimulating polymorphonuclear cell (PMNC) recruitment to the site of injury.10 Furthermore, they have protective and proregenerative properties as shown during acetaminophen-induced liver injury.11

Earlier work defined the proinflammatory cytokine TNF as an essential molecule involved in liver regeneration, with TNF-R1 knockout mice showing impaired hepatocyte proliferation.12 These results indicate that intracellular pathways of the TNF/TNF-R1 system are involved in mediating hepatocyte proliferation in vivo. Shortly after PH, TNF is rapidly secreted from Kupffer cells, inducing NF-κB activation.1 NF-κB is held in the cytoplasm in its inactive form when bound to IκB. Upon stimulation, the inhibitor of κB kinase (IKK) complex (consisting of IKKα/IKK1, IKKβ/IKK2, and IKKγ/nuclear factor-B essential modifier (NEMO)) phosphorylates IκB, leading to its degradation and resulting in NF-κB nuclear translocation and target gene transcription.13

The role of NF-κB during liver regeneration has been previously investigated although it remains controversial.14–16 Moreover, the implication of IKK2-driven NF-κB signaling in hepatocytes during liver regeneration remains to be elucidated. Because IKKβ/IKK2 knockout mice die in utero due to massive liver apoptosis, we used hepatocyte-specific IKK2 knockout mice (IKK2ΔLPC).17 Our previous work has shown that hepatocyte-specific IKK2 deletion exerts different outcomes depending on the type of liver injury. Whereas NF-κB is still activated in IKK2ΔLPC hepatocytes after exogenous TNF administration, NF-κB activation is attenuated during ischemia/reperfusion-induced liver injury.17 Therefore, we used our IKK2ΔLPC mice to elucidate the impact of hepatocyte-specific IKK2 deletion on NF-κB activation after PH and its implication during liver regeneration.


BrdU, 5-bromo-2-deoxyuridine; CXC, chemotactic cytokines; DAPI, 4′,6′-diamidino-2-phenylindole; ECM, extracellular matrix; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; IL, interleukin; JNK, c-Jun N-terminal kinase; MMP-9, matrix metalloproteinase-9; mRNA, messenger RNA; NF-κB, nuclear factor κB; NPC, nonparenchymal cell; PCR, polymerase chain reaction; PH, partial hepatectomy; PMNC, polymorphonuclear cell; SAA, serum amyloid A.

Materials and Methods

Generation of Conditional IKK2 Knockout Mice.

Mice carrying the loxP site-flanked [floxed (f)] IKK2 gene (IKK2f/f) were generated as described previously.17 IKK2f/f mice were then crossed with Alfp-cre transgenic mice which have the Cre recombinase under control of an albumin promoter.18 Crossings rendered Cre-positive (IKK2ΔLPC) and Cre-negative (IKK2f/f) mice that were genotyped via polymerase chain reaction (PCR) analysis as described previously.17 All mice were backcrossed to a pure Bl6/C57 background, and 8-week-old mice were used for the experiments. Animals were held according to the human care criteria prepared by the National Academy of Sciences (NIH publication 86-23, revised 1985).

Systemic NF-κB inhibition.

To inhibit IKK2 in all cell compartments, we administrated 10 μg/g body weight of the specific pharmaceutical IKK2 inhibitor AS60286817 (kindly provided by Merck-Serono International SA) or vehicle (0.5% carboxymethylcellulose/Tween 20 0.25%). Treatment was given orally 24 hours and 1 hour before PH, and the dose was repeated 24 hours after liver resection.

Partial Hepatectomy.

Pathogen-free 7-week-old to 9-week-old male mice were used for PH. PH operations were performed under ketamin/xylazine anesthesia as described previously.19 Each time point provides at least four mice per genotype. Remaining livers were removed, rinsed in ice-cold phosphate-buffered saline and individually frozen in liquid nitrogen and Tissue Tek (Sakura Europe).

RNA Isolation and Quantitative Real-Time PCR.

Livers were harvested and snap frozen in liquid nitrogen. RNA was isolated with PeqGold-RNA pure kit (Peqlab). First strand synthesis was performed with Oligo dT primers and reverse transcription with M-MLV Reverse Transcriptase (Invitrogen). Quantitative real-time PCR was performed using SYBR Green reagent (Invitrogen) in an ABI-Prism 7300 real-time PCR system (Applied Biosystems). Reactions were performed twice in triplicate, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) values were used to normalize gene expression which was expressed in times versus control basal expression and quantified via delta-delta CT calculation. Primers can be provided under request.

Quantification of Liver Cytokine Levels.

Frozen livers were homogenized in ice-cold lysis buffer containing 10 mM [4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid], 2 mM ethylene diamine tetraacetic acid (pH 8.0), 5 mM dithiothreitol, 1 mM Pefabloc, and 1 tablet of cocktail of proteinase inhibitors (Roche). TNF and IL-6 liver expression were analyzed in whole liver extracts via enzyme-linked immunosorbent assay that was performed following the manufacturer's instructions (R&D Systems).

c-Jun N-Terminal Kinase Activity Assay.

c-Jun N-terminal kinase (JNK) activity was assessed as described previously.20 As a substrate, we used a purified recombinant protein containing the N-terminal amino acids 1-223 of the human c-Jun fused to glutathion-S-transferase. For immobilization, 15 μg recombinant glutathion-S-transferase-Jun protein was attached to glutathion agarose beads (Sigma) in buffer containing 1.0% Triton X-100 and 1 mM dithiothreitol followed by incubation with 50 μg of total liver protein extract at 4°C overnight. After extensive washing in 20 mM [4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid] (pH 7.5); 20 mM MgCl2; 20 mM glycerophosphate; and 2 mM dithiothreitol, the bead-protein complexes were incubated with [32P]γ–adenosine triphosphate and 40 pM adenosine triphosphate for 20 minutes at 30°C. After kinase reaction, the proteins were fractionated on a 10% sodium dodecyl sulfate–polyacrylamide gel and visualized on X-ray film (Amersham). A single prominent band of 46 kDa represented phosphorylated glutathione-S-transferase-c-Jun.

Western Blot Analysis.

Protein extracts were resolved in a 10% sodium dodecyl sulfate–polyacrylamide gel and transferred to nitrocellulose membranes (Whatman). Membranes were blocked with 5% milk in Tris-buffered saline and probed with the following primary antibodies: p-IKK1/2 (Ser176/180), p-IκB-α, p-STAT-3, p-cdc2 (Cell Signaling Technology), cyclin D1, and cyclin A (Santa Cruz Biotechnology). As a loading control, we used GAPDH antibody (Biogenesis) and α-tubulin (Sigma). As secondary antibodies, we used anti-rabbit IgG–HRP-linked (Cell Signaling Technology) and anti-mouse IgG–HRP-linked (Santa Cruz Biotechnology).


All immunofluorescence described in this study was performed on liver cryosections. Slides were fixed in appropriate fixating agent, 70% methanol–30% acetone 10 minutes at room temperature for p65 (sc-109; Santa Cruz Biotechnology) and F4/80 (Serotec) and in 4% paraformaldehyde for CD11b (Becton Dickinson) 10 minutes at room temperature. As a secondary antibody, we used a goat anti-rabbit labeled with Cy-3 for p65, goat anti-rat with fluorescein isothiocyanate for F4/80 and goat anti-rat Cy-3 labeled for CD11b (all from Jackson Research).

Hepatocyte proliferation was detected via immunofluorescence through 5-bromo-2-deoxyuridine (Amersham Pharmacia) incorporation to the nuclei of hepatocytes undergoing DNA synthesis and anti-phospho-Histone H3 antibody (Cell Signaling Technology) was used to assess the presence of mitotic bodies after PH on liver sections. As secondary antibodies we used goat-anti-mouse Cy-3 labeled and goat anti-rabbit fluorescein isothiocyanate labeled respectively (Jackson Research). Quantification of proliferating and mitotic hepatocytes was performed by counting 5-bromo-2-deoxyuridine (BrdU) or phospho-Histone H3–positive nuclei relative to the total nuclei per power field (×200) stained with 4′,6′-diamidino-2-phenylindole (DAPI).


Whole liver proteins, extracted with an Nonidet P-40 buffer, were subjected to electrophoresis in an 8% sodium dodecyl sulfate–polyacrylamide gel supplemented with 10 mg/mL of Gelatinase A (Sigma). Gel was then washed twice for 15 minutes each time in Triton 2.5% in a rocking surface. Next, the gel was incubated at 37 °C for 24 hours in a buffer (pH 7.4) containing 50 mM Tris HCl, 10 mM CaCl2 · 2H2O, 100 mM NaCl and 0.05% Brij 35 (Sigma). Gel was fixed and stained with 0.5% Coomassie blue in 10% acetic acid, 30% isopropanol. After destaining the gel, gelatinase digestion via matrix metalloproteinase-9 (MMP-9) activity was detected as a nonstained band. This experiment was repeated three times in all samples (n = 4).

Cell Isolation and Flow Cytometry.

Single-cell suspensions were yielded through mechanical enzymatic digestion of freshly isolated livers 12 hours after PH. Briefly, whole liver samples were minced with scissors and subsequently digested at 37°C for 30 minutes in 0.3225 U/mL collagenase D (Roche Applied Science) in RPMI medium. After digestion, the extract was filtered twice though a 12-μm filter (Becton Dickinson). The suspension was centrifuged at 300g for 10 minutes to remove cell conglomerates. The supernatant was washed twice in Hank's balanced salt solution supplemented with 1% bovine serum albumin and 2 mM ethylene diamine tetraacetic acid. Cells were stained for CD45, F4/80, Gr1 (BD Biosciences), and CD11b (Becton Dickinson), as well as NK1.1 and CD3 (eBioscience) and analyzed via flow cytometry (FACSCanto II; Becton Dickinson).

For analysis of blood leukocytes, blood was taken via the retro-orbital venous plexus. Red blood cell lysis was performed using 1X Pharmlyse buffer (Becton Dickinson) following the manufacturer's instructions. Leukocytes were then stained for CD11b, Gr1, CD19 (Becton Dickinson), CD115, and CD3 (eBioscience) and subjected to flow cytometric analysis. White blood cells were counted with an automat in the animal facility of the university, and the results were expressed as the number of cells per microliter.

Statistical Analysis.

Data are expressed as the mean ± standard deviation. Statistical significance was determined via 2-way analysis of variance followed by Student t test.


IKK2 Hepatocyte-Specific Deletion Modifies I-κBα Phosphorylation and NF-κB Activation After PH.

After stimulation, the IKK complex gets activated by phosphorylation of the catalytic subunits IKK1 and IKK2. These kinases phosphorylate I-κB, eventually triggering NF-κB nuclear translocation and target gene activation. We previously generated hepatocyte-specific IKK2 ko mice (IKK2ΔLPC), which show effective IKK2 deletion in hepatocytes.17 PH was performed in these knockout mice and wild-type controls (IKK2f/f), and we first studied the impact of IKK2 deletion in the IKK complex using an IKK1/2 phospho-specific antibody (Ser176/180). In IKK2f/f, we observed transient IKK1/2 phosphorylation 1 hour after PH (Fig. 1A), whereas strong phospho-IKK1/2 expression was found in IKK2ΔLPC 1, 3, and 6 hours after PH (Fig. 1A). No significant differences in IKK1 protein level were evident between the 2 mice strains at different time points after PH (Fig. 1A).

Figure 1.

Hepatocyte-specific IKK2 deletion attenuates parenchymal NF-κB activation after PH. (A) Western blot analysis of p-IKK1/2, total IKK1, and p-I-κBα before and 1, 3, and 6 hours after PH. GAPDH and α-tubulin (lower panel) were used as loading controls. (B) Immunohistochemistry using p65+secondary Cy3-labeled and F4/80+secondary fluorescein isothiocyanate antibodies in livers of IKK2f/f and IKK2ΔLPC mice before and 1 and 3 hours after PH (left panel). DAPI staining was performed to detect nuclei. (C) Merged picture (right panel) of p65 (left panel) and F4/80 stainings (middle panel). Thin arrows pointing up show Kupffer cell p65 staining; full arrows pointing down show hepatocyte p65 staining.

We then tested if lack of hepatocytic IKK2 expression and strong IKK activation alters I-κBα phosphorylation after PH. In IKK2f/f mice, high p-I-κBα expression was already found 1 hour after PH, and lower I-κBα phosphorylation was evident 6 hours after PH (Fig. 1B). In IKK2ΔLPC mice, p-I-κBα expression was detected starting 3 hours after PH, indicating delayed NF-κB activation (Fig. 1B). These data suggest that IKK2 deletion and PH trigger increased IKK1 activation leading to delayed IκBα phosphorylation, because IKK1 is known to mainly activate the NF-κB noncanonical pathway.21

Because these results demonstrated altered IKK complex activity in IKK2ΔLPC mice, we now investigated NF-κB nuclear translocation in both animal strains via immunofluorescence using a p65 antibody. In order to dissect the cellular distribution of NF-κB activation we costained with a F4/80 antibody, which labels Kupffer cells. In IKK2f/f mice, p65 staining was positive 1 hour after liver resection in hepatocytes (full arrows pointing down), whereas 3 hours after PH, nonparenchymal cells became positive as evidenced by p65-F4/80 colocalization (Fig. 1C). In IKK2ΔLPC mice, no p65 response could be detected in hepatocytes 1 hour after resection, whereas positive signals were detected in nonparenchymal cells—most likely Kupffer cells as they are also positive for F4/80 (Fig. 1C). Three hours after resection, NF-κB positive hepatocytes were detected in IKK2ΔLPC (Fig. 1C). The p65 antibody used in this study binds to the N-terminal part of p65, which interacts with IκB; therefore, unstimulated cells remain p65-negative.22

In summary, lack of IKK2 expression in hepatocytes alters the NF-κB activation pattern in both parenchymal and nonparenchymal liver cells after PH.

Lack of Hepatocyte IKK2 Promotes Earlier Cell Cycle Entry After PH.

Kupffer cell–derived TNF production is induced after PH triggering NF-κB activation.1 As expected, liver TNF expression increased in IKK2f/f animals up to 6 hours after PH, whereas in IKK2ΔLPC mice it was already increased 1 hour after resection, remaining elevated 3 and 6 hours after PH (Fig. 2A).

Figure 2.

(A) IKK2ΔLPC mice exhibit earlier cell cycle priming after PH. TNF protein expression detected by enzyme-linked immunosorbent assay in livers of IKK2f/f and IKK2ΔLPC mice. Solid lines represent IKK2f/f; dashed lines represent IKK2ΔLPC (n = 4 animals/time point; ***P < 0.001, **P < 0.01 [IKK2f/f versus IKK2ΔLPC]). (B) JNK activity was detected in whole liver extracts showing stronger and sustained kinase activity in IKK2ΔLPC mice after PH. (C) Real-time PCR for c-myc liver mRNA expression in IKK2ΔLPC and IKK2f/f mice before and after PH. (D) Zymography to determine MMP-9 gelatinase activity in IKK2ΔLPC mice and IKK2f/f mice before and after PH. Two representative samples from each time point and genotype are shown. The position of the digested band is shown by an arrow.

TNF-dependent JNK activation is involved in triggering cell cycle progression after PH.12, 23, 24 Kinase assays of IKK2f/f mice whole-liver extracts showed transient JNK activation 1 hour after PH, whereas IKK2ΔLPC mice exhibited sustained JNK activity during the initial time points after liver resection (Fig. 2B).

Because c-myc is shortly activated after PH as part of the immediate-early gene response, it mediates early cell cycle progression of hepatocytes.25 In IKK2f/f mice c-myc messenger RNA (mRNA) expression increased already 3 hours after PH and peaked 6 hours after surgery (Fig. 2C). Unexpectedly, faster c-myc expression was evident in IKK2ΔLPC animals, because levels were found a maximum 3 hours after PH.

In the early phase of liver regeneration, extracellular matrix (ECM) degradation allows release of growth factors that promote hepatocyte proliferation.26, 27 MMP-9 is produced in response to TNF and mediates ECM degradation.28 Zymography analysis showed that in IKK2f/f animals MMP-9 proteinase activity was transiently activated 6 hours after PH, while in IKK2ΔLPC mice earlier and prolonged MMP-9 activity was observed (Fig. 2D). These results suggest that lack of IKK2 expression triggers hepatocytes to leave earlier the quiescent (G0) state.

Earlier S-Phase Progression and Hepatocyte Proliferation in IKK2ΔLPC Mice After PH.

Next, we analyzed G1/S progression of hepatocytes in IKK2ΔLPC mice after PH. Immunofluorescence-based BrdU analysis and quantification of proliferating cells were performed in IKK2f/f and IKK2ΔLPC mice between 24 and 72 hours after PH. Between 24 and 40 hours after liver resection, significantly more BrdU-positive hepatocytes were found in IKK2ΔLPC mice and maximum DNA synthesis was evident 48 hours after PH (Fig. 3A -B). In IKK2f/f mice a single and pronounced peak of DNA synthesis was found 48 hours after PH, and the number of proliferating cells was higher than IKK2ΔLPC mice at this time point.

Figure 3.

IKK2 deletion in hepatocytes accelerates hepatocyte proliferation after PH. (A) Immunohistochemistry using a BrdU antibody followed by a Cy3-labeled secondary antibody in livers of IKK2f/f and IKK2ΔLPC mice before and between 24 and 72 hours after PH (upper panel). Merged DAPI and BrdU staining is shown in the lower panels. (B) Quantification of BrdU-positive cells relative to total nuclei (DAPI) per ×200 power field. Solid lines represent IKK2f/f; dashed lines represent IKK2ΔLPC (n = 4 animals/time point; **P < 0.01, ***P < 0.001 [IKK2f/f versus IKK2ΔLPC]).

To further confirm these results, we studied expression of factors involved in cell cycle control. Differential cyclin D1 expression was found between mouse strains, because protein levels in IKK2ΔLPC peaked 24 hours after PH, whereas in wild-type littermates it was found a maximum of 40 hours after resection (Fig. 4A). Cyclin E mRNA analysis showed maximum up-regulation in IKK2ΔLPC 40 hours after PH that was delayed to 48 hours in wild-type mice (Fig. 4B). Cyclin A mRNA expression was first elevated 40 hours after PH in IKK2ΔLPC mice, while in wild-type mice no significant increase could be observed at this time point (Fig. 4C). At 48 hours, cyclin A expression peaked in both groups but was higher in IKK2f/f animals (Fig. 4C). Western blot analysis confirmed earlier cyclin A activation in IKK2ΔLPC mice (Fig. 4D). These results demonstrate earlier G1/S-phase progression and DNA synthesis in IKK2ΔLPC mice after PH.

Figure 4.

Lack of hepatocyte IKK2 expression triggers earlier G1/S phase transition after PH. (A) Western blot analysis for cyclin D1 expression of whole protein liver extracts of IKK2f/f and IKK2ΔLPC mice at time points before and after PH. (B) Real-time PCR for liver cyclin E mRNA expression of IKK2f/f and IKK2ΔLPC animals at time points before and after PH. (C) Liver cyclin A mRNA and (D) protein expression in IKK2f/f and IKK2ΔLPC mice before and after PH. Solid lines represent IKK2f/f; dashed lines represent IKK2ΔLPC [n = 4 animals/time point; *P < 0.05, **P < 0.01 (IKK2f/f versus IKK2ΔLPC)].

Prolonged Mitotic Phase in IKK2ΔLPC Mice after PH.

Cells remain arrested in G2 phase when cyclin B1 forms a complex with cdc2 (also known as Cdk1) that is inactive upon Tyr-15 and Thr-14 phosphorylation.29 G2/M phase progression occurs after dephosphorylation of cdc2 by Cdc25C phosphatase.30 Cyclin B1 mRNA expression was not different between IKK2ΔLPC and IKK2f/f mice (Fig. 5A). However, a different kinetic in cdc2 activation was observed. In IKK2ΔLPC, faster progression into the mitotic phase occurred, becuase we found earlier phosphorylation of cdc2 that was dephosphorylated (and thus activated) 60 hours after PH, whereas in IKK2f/f mice a certain degree of phosphorylated cdc-2 (inactive form) could still be detected at this time point (Fig. 5B).

Figure 5.

Misregulation of mitosis in hepatocyte-specific IKK2-deleted mice after PH. (A) Real-time PCR for liver cyclin B1 mRNA expression in IKK2f/f and IKK2ΔLPC mice before and after PH. (B) Phospho-cdc2 expression as evidenced via Western blot analysis in IKK2f/f and IKK2ΔLPC mice. (C) Immunofluorescence with a phospho-Histone H3 antibody before and between 60 and 96 hours after PH in IKK2f/f and IKK2ΔLPC mice (upper panel). Merged DAPI and anti-phospho-Histone H3 staining is shown in the lower panel. (D) Quantification of anti-phospho-Histone H3 positive cells in livers of IKK2f/f and IKK2ΔLPC mice. Solid lines represent IKK2f/f; dashed lines represent IKK2ΔLPC (n = 4 animals/time point; ***P < 0.001 [IKK2f/f versus IKK2ΔLPC]). (E) Zymography for MMP-9 gelatinase activity in IKK2ΔLPC mice occurred later than in IKK2f/f mice after PH. Two representative samples from each time point and genotype are shown. The position of the digested band is shown by an arrow.

Phosphorylation of Histone H3 is essential for chromosome condensation and progression from G2 to mitosis.31 Immunohistochemistry using a phospho-Histone H3 antibody and its quantification evidenced initiation of mitosis in both mouse strains 60 hours after PH (Fig. 5C-D). However, although the mitotic rate remained at the same level in IKK2f/f mice 72 hours after PH, we observed a significant increase in IKK2ΔLPC mice. After 96 hours, few phospho-Histone H3–positive cells could be detected, whereas no mitosis could be detected in IKK2f/fanimals at this time point.

ECM modulation is important in order to shift hepatocytes in a differentiated and quiescent state.26 Differences in ECM degradation were evident between the 2 mouse strains at late time points after PH. In IKK2f/f animals, MMP-9 activity was detected between 24 and 48 hours (Fig. 5E), while in IKK2ΔLPC mice it was found later (48-60 hours after PH). Together, these results suggest that IKK2ΔLPC mice require a prolonged mitotic phase in order to reach a quiescent state after PH.

Stronger CXCL1 and Acute Phase Gene Expression in IKK2ΔLPC Mice After PH.

Recent studies highlighted the role of the innate immune response during liver regeneration3–5 where proinflammatory cytokines and acute phase genes play a main role. Our first results demonstrated that dysregulation of NF-κB activation was accompanied by high TNF expression in IKK2ΔLPC mice leading to earlier priming. In addition to TNF, IL-6 is involved in triggering hepatocytes to leave their quiescent state. Liver IL-6 expression increased in both IKK2f/f and IKK2ΔLPC mice during the first 6 hours after PH; however, no significant differences were found between the two mouse strains (Fig. 6A). This result was further confirmed after no significant difference in nuclear phospho-STAT-3 expression was found between both groups after PH (Fig. 6B).

Figure 6.

Earlier and stronger inflammatory response in hepatocyte-specific IKK2 knockout mice. (A) IL-6 protein expression detected via enzyme-linked immunosorbent assay at time points before and after PH in livers of IKK2f/f and IKK2ΔLPC mice. (B) Western blot analysis of whole liver extracts of IKK2f/f and IKK2ΔLPC animals for phospho-STAT-3 expression before and after PH. GAPDH was used as a loading control. Real-time PCR to analyze liver mRNA levels in IKK2f/f and IKK2ΔLPC mice for (C) SAA and (D) CXCL-1 before and after PH. Solid lines represent IKK2f/f; dashed lines represent IKK2ΔLPC (n = 4 animals/time point; *P < 0.05, **P < 0.01, ***P < 0.001 [IKK2f/f versus IKK2ΔLPC]).

SAA is one of the main acute phase response proteins and is part of the unspecific defense in mice after liver injury.7, 26 Maximal SAA levels were found in IKK2f/f and IKK2ΔLPC mice 24 hours after PH. However, earlier and more intense acute phase response was detected in IKK2ΔLPC livers as evidenced by the 200-fold increase in SAA expression found already 6 hours after liver resection (Fig. 6C).

CXC chemokines like CXCL-1 attract neutrophils to sites of inflammation during early stages of wound healing facilitating tissue repair.10 We found that basal CXCL-1 mRNA expression was higher in livers of IKK2ΔLPC compared with IKK2f/f animals (Fig. 6D). After PH, CXCL-1 expression in IKK2f/f mice increased over time during the first 24 hours, whereas a dramatically faster and stronger up-regulation was evident in IKK2ΔLPC animals (Fig. 6D). Together, these data indicate an earlier and stronger chemoattractant and acute phase response in IKK2ΔLPC after PH.

IKK2ΔLPC Mice Show Enhanced Recruitment of Inflammatory Cells After PH.

To study if the increased chemoattractant response observed in IKK2ΔLPC mice correlated with stronger mobilization of PMNCs, we performed FACS analysis and white blood cell counts collected from IKK2f/f and IKK2ΔLPC mice before and 12 hours after PH. In untreated animals from both strains, a comparable percentage of CD11b+/Gr1+ cells, mainly PMNCs, were detected in peripheral blood (Fig. 7A). This cell population increased 12 hours after PH in IKK2ΔLPC, while no regulation was evident in IKK2f/f mice (Fig. 7A).

Figure 7.

Mobilization and recruitment of inflammatory cells after PH. (A) FACS analysis of peripheral blood leukocytes before (left panel) and 12 horus after PH (right panel). Fraction of CD11b+ cells is given in percentage of total leukocytes. Diagram shows total number of CD11b+/Gr1+/CD115− cells (PMNCs) in the peripheral blood at the given time points. (B) Fluorescence-activated cell sorting analysis of immune cells isolated from hepatic tissue before and 12 hours after PH. Dot plots show CD11b and F4/80 expression of cells pregated for CD45+ respectively. Bar chart shows the absolute frequency of CD11b+/Gr1+/F4/80− cells (mainly PMNCs) displayed as the percentage of total liver cells (normalized to CD45+ cells). (C) Immunofluorescent staining of frozen liver sections for CD11b+ confirmed the results obtained via fluorescence-activated cell sorting (red, CD11b-positive cells; blue, DAPI-stained nuclei). Black bars represent IKK2f/f; grey bars represent IKK2ΔLPC (n = 4 animals/time point; *P < 0.05 [IKK2f/f versus IKK2ΔLPC]).

We then investigated if increased PMNC mobilization in IKK2ΔLPC mice after PH correlated with a higher recruitment to the liver (Fig. 7B). Similar numbers of liver resident CD45+/CD11b+/Gr1+/F4/80− cells—mainly PMNCs—could be detected in both strains before PH. Twelve hours after PH, the amount of these cells did not increase significantly in IKK2f/f animals, whereas livers of IKK2ΔLPC mice already revealed a 3.5-fold increase of CD45+/CD11b+/Gr1+/F4/80− cells compared with the basal level and control wild-type littermates. These results suggest that the earlier inflammatory and chemoattractant response in IKK2ΔLPC after PH had a direct impact on mobilization of neutrophils to the blood stream and recruitment to the liver. This finding was further confirmed by immunofluorescence of liver sections using a CD11b antibody, supporting a more pronounced infiltration of CD11b+ cells after PH in IKK2ΔLPC mice (Fig. 7C).

Systemic NF-κB Inhibition with AS602868 Does Not Affect Hepatocyte Proliferation After PH.

Our data show that hepatocyte-specific IKK2 deletion is associated with a proliferative advantage after PH. Therefore, we tested if systemic inhibition is also associated with earlier proliferation after PH. We thus applied an IKK2 inhibitor AS602868 in a dose we used in previous experiments to block IKK2 after ischemia/reperfusion injury17 and to block progression of NASH.32

We administered AS602868 24 hours and 1 hour before PH as well as 24 hours after PH. AS602868 treatment compared with controls (vehicle) had no impact on the increase and the maximum of BrdU synthesis 40 and 48 hours after PH (Fig. 8A -B). These results indicate that systemic, in contrast to hepatocytic IKK2 inhibition, has no impact on cell cycle progression after PH.

Figure 8.

Systemic pharmaceutical IKK2 inhibition with AS602868 had no impact on hepatocyte proliferation after PH. (A) Immunohistochemistry using BrdU and a Cy3-labeled antibody in livers orally treated with AS602868 or vehicle showed no significant differences between treatment groups (AS602868 versus vehicle). (B) Solid lines represent quantification of BrdU-positive cells relative to vehicle; dashed lines represent AS602868 (n = 5 animals/time point).


NF-κB is an essential transcription factor to maintain liver homeostasis.13, 33 Additionally, knockout mice of members of the IKK complex controlling NF-κB activation die from hepatocyte apoptosis during embryonal development.34, 35 Thus the complex regulation between the different factors involved in this pathway is of central relevance for the liver. In the present study, we used hepatocyte-specific IKK2 knockout mice to further elucidate the role of the IKK complex during liver regeneration.

After PH in IKK2ΔLPC mice, major differences in NF-κB activation were found compared to wild-type IKK2f/f animals. In wild-type mice NF-κB activation was first found in hepatocytes, while in IKK2ΔLPC animals p65 was first positive in nonparenchymal cells (NPCs), resulting in a stronger inflammatory response. In our previous study using IKK2ΔLPC animals the inflammatory response after I/R injury was attenuated compared with wild-type controls.17 Both studies show that IKK2 expression in hepatocytes is essential to direct the crosstalk between hepatocytes and NPCs in a stimulus-dependent manner during inflammation, which is of major relevance for liver physiology. However, at present, the intercellular crosstalk between IKK2-deficient hepatocytes and NPCs during liver regeneration is not fully understood.

Analysis of the IKK complex activity revealed that IKK2ΔLPC mice showed stronger IKK1 activation after PH despite later p-IκBα expression that correlated with delayed kinetics of p65 activation in hepatocytes. A potential explanation for the stronger IKK1 activation observed could be that lack of IKK2 protein expression results in increased activation (phosphorylation) of IKK1 and consequently, enhanced signaling via the noncanonical pathway in IKK2ΔLPC mice after PH. At present, the role of IKK1 during liver regeneration has not been studied. However, knockout mice for typical IKK1 activators, such as lymphotoxin-α (LTα) and its receptor (LTαR), show impaired liver regeneration after PH36 pointing to the relevance of the noncanonical pathway during liver regeneration. Moreover, we cannot exclude that signaling via the noncanonical pathway in IKK2ΔLPC mice may cooperate in the production of ligands that activate NPCs, contributing to the crosstalk between cell compartments after PH.37

In order to further test the relevance of IKK2 for the crosstalk between hepatocytes and NPCs, we used the IKK2 inhibitor AS602868 to systemically block IKK2 activation during liver regeneration. Interestingly, inhibition of IKK2 in all cell compartments did not significantly alter DNA synthesis after PH. Earlier results demonstrated that IKK2 inhibition in myeloid cells blocks NF-κB activation38 and that by using AS602868 the inflammatory response is also blocked in NPCs in the liver.32 This, together with our present data suggests that AS602868 administration during PH also reduces the inflammatory response in NPCs which reverts the proliferative advantage shown in IKK2ΔLPC mice.

Our results using IKK2ΔLPC mice and AS602868 cannot be compared to previous experiments using either an adenoviral IκB superrepressor or transgenic mice carrying an inducible IκB mutant in hepatocytes.15, 16 Both approaches blocked NF-κB activation downstream of the IKK complex; however, the outcome of the studies was different, because apoptosis and impaired hepatocyte proliferation were found when the adenoviral vector was applied,15 whereas no significant effects on liver regeneration were observed using transgenic mice carrying the inducible IκBα mutant.16 A major problem using the adenoviral vector is the finding that this approach triggers apoptosis and DNA synthesis already before PH.15 In the Chaisson et al.16 study, NF-κB was completely blocked and not delayed, as found in our study. This difference between the 2 studies might also contribute to the observation why the imbalance in NF-κB activation as found in IKK2ΔLPC mice triggers the strong inflammatory response in the NPC compartment.

We show that earlier NF-κB activation in the NPC compartment observed in mice lacking IKK2 correlated with a rapid and strong inflammatory response. PMNCs (neutrophils) have differential roles depending on the type of damage, because they can contribute to liver injury or display beneficial effects by mediating resolution of wound healing, tissue repair, and cell proliferation.11, 39 Targeting immune cells to the damaged tissue relies on the production of chemoattractants. ELR+CXC chemokines mediate chemotaxis of neutrophils. After PH, higher expression of CXCL1 in IKK2ΔLPC mice correlated with a faster mobilization of neutrophils and recruitment to the liver that may contribute to the earlier hepatocyte proliferation in IKK2ΔLPC mice as ELR+CXC chemokines contribute to liver regeneration after acetaminophen-induced injury.11

In order to elucidate the molecular mechanism by which ELR+ chemokines are more rapidly induced in the liver of IKK2ΔLPC mice we studied the activation of relevant signaling pathways, namely the STAT-3 pathway in wild-type and IKK2ΔLPC mice after PH. We have previously demonstrated that hepatocytes are a potent source of CXCL1, which is strongly induced in response to gp130/STAT3 signaling.40 Levels of gp130 ligand IL-6 and pSTAT3 were equal in the 2 groups. However, there are evidences that the JAK/STAT-3 and NF-κB pathways communicate41, 42 in an IKK2-dependent fashion.43 We therefore propose that increased CXCL-1 levels observed in IKK2ΔLPC mice after PH are a result of disrupted inhibition of STAT-3 by IKK2. This is further supported by enhanced SAA transcription in IKK2ΔLPC mice as SAA gene expression is tightly controlled by gp130/STAT3. The interaction of both STAT-3 and NF-κB has both synergistic and inhibitory consequences during acute phase gene transcription, because they share binding sites in the promoters of these genes, rendering enhanced or competitive overlapping interactions.41–43 Thus, it is very likely that despite comparable IL-6 production and STAT-3 phosphorylation after PH, IKK2ΔLPC mice exhibit earlier, stronger, and sustained SAA production than wild-type littermates due to the lower presence of competitive NF-κB, allowing more efficient STAT-3 binding to the SAA promoter in hepatocytes.

Additionally, we investigated the degree of ECM degradation in the liver, because this promotes the bioavailability of growth factors such as hepatocyte growth factor28 and contributes to leukocyte extravasation during inflammatory response.44 MMP-9 is a TNF-inducible molecule45 that plays a key role in ECM degradation and deficient mice exhibit slower and delayed hepatocyte proliferation.28 In consonance, we found that enhanced TNF expression in livers of hepatocyte-IKK2 deleted mice correlated with earlier MMP-9 activity, which together could contribute to earlier priming as observed in IKK2ΔLPC mice.

These results correlated with prolonged JNK activation, a known target of TNF that can be negatively regulated by NF-κB,46 and earlier c-myc expression in livers of IKK2ΔLPC mice after PH. Both pathways are directly involved to translate stronger stimulation via extracellular stimuli into earlier cell cycle progression. Because cyclin D is a direct target gene of JNK/AP-1 signaling, these results provide the molecular link between the stronger inflammatory response and earlier G1/S-phase transition in IKK2ΔLPC mice.24

During mitosis, differences were found between IKK2ΔLPC and wild-type animals. Whereas cyclin B levels peaked at the same time point, exit of mitosis was prolonged in IKK2ΔLPC animals as evidenced by longer Histone H3 phosphorylation and delayed ECM degradation. At present, we have no obvious explanation for this observation. However, these differences could either be a direct effect of lack of IKK2 expression in hepatocytes or a consequence of impaired synchronization during an earlier time point of hepatocyte proliferation in IKK2ΔLPC animals.

The present work contributes essentially to the understanding of the role of IKK2 and consequently NF-κB signaling in injury induced liver regeneration. Additionally, it highlights the relevance of NPCs and their crosstalk with hepatocytes during this process and supports our previous work depicting the importance of preserving the balance of NF-κB activity between the liver cell compartments in order to maintain liver homeostasis and efficiently counteract injury.17, 32, 47 Our results suggest that interference with this crosstalk might represent a future target for regenerative or anti-inflammatory treatment strategies in liver diseases.


We want to thank Dr Christelle Guyot for her expert help in establishing the zymography analysis. We are very grateful to Dr. Michel Dreano for promptly providing us the AS602868. We are also very thankful to Prof. Dr. Manolis Pasparakis (University of Cologne, Germany) for kindly providing us with the IKK2ΔLPC mice.