Genetic mouse studies suggest that the NF-κB pathway regulator NEMO (also known as IKKγ) controls chronic inflammation and carcinogenesis in the liver. However, the molecular mechanisms explaining the function of NEMO are not well defined. Here, we report that overexpression of the cell-cycle regulator p21 is a critical feature of liver inflammation and carcinogenesis caused by the loss of NEMO. NEMOΔhepa mice develop chronic hepatitis characterized by increased hepatocyte apoptosis and proliferation that causes the development of fibrosis and hepatocellular carcinoma (HCC), similar to the situation in human liver disease. Having identified p21 overexpression in this model, we evaluated its role in disease progression and LPS-mediated liver injury in double mutant NEMOΔhepa/p21−/− mice. Eight-week-old NEMOΔhepa/p21−/− animals displayed accelerated liver damage that was not associated with alterations in cell-cycle progression or the inflammatory response. However, livers from NEMOΔhepa/p21−/− mice displayed more severe DNA damage that was further characterized by LPS administration correlating with higher lethality of the animals. This phenotype was attenuated by genetic ablation of the TNF receptor TNF-R1 in NEMOΔhepa/p21−/− mice, demonstrating that DNA damage is induced via TNF. One-year-old NEMOΔhepa/p21−/− mice displayed greater numbers of HCC and severe cholestasis compared with NEMOΔhepa animals. Therefore, p21 overexpression in NEMOΔhepa animals protects against DNA damage, acceleration of hepatocarcinogenesis, and cholestasis. Taken together, our findings illustrate how loss of NEMO promotes chronic liver inflammation and carcinogenesis, and they identify a novel protective role for p21 against the generation of DNA damage. Cancer Res; 75(6); 1144–55. ©2015 AACR.

Hepatocellular carcinoma (HCC) is the fifth most common cancer worldwide and represents the third leading cause of cancer-related mortality (1). HCC develops when the balance between cell proliferation and cell death is disrupted, and the subsequent aberrant proliferation leads to tumor growth. Cell proliferation is controlled by protein kinase complexes consisting of cyclins and cyclin-dependent kinases (Cdks), which regulate distinct phases of the cell cycle.

p21 is involved in numerous growth-inhibitory pathways as shown in cell culture systems and in vivo (2). For example, embryonic fibroblasts lacking p21 expression are significantly impaired in their ability to arrest in G1-phase in response to DNA damage (3), and in vivo experiments with p21 knockout mice (p21−/−) revealed earlier hepatocyte DNA synthesis, cyclin/CDK kinase activation, and S-phase gene expression compared with wild-type mice in the partial hepatectomy (PH) model (4). In contrast, after PH mice with hepatic p21 overexpression displayed impaired hepatocyte proliferation (5).

p21 has been considered as a tumor suppressor because p21−/− mice display spontaneous tumor formation after 16 months and additionally these mice are more sensitive to chemically induced carcinogenesis (6, 7).

However, deletion of p21 does not necessarily promote tumor growth and a potential function as an oncogene has also been described. These assumptions are based on human studies as it has been shown that p21 is upregulated during inflammation and fibrosis in chronic liver diseases, and high p21 expression is linked to hepatocarcinogenesis in cirrhotic patients (8, 9). An oncogenic function for p21 has also been proposed in animal models. Mice deficient for p53 spontaneously develop multiple tumors while additional deletion of p21 in p53-deficient mice increases the survival of these mice from 6 to 9 months and for instance leads to a significant reduction of thymic lymphomas (10).

From these data at present the role of p21 in tumorigenesis is not clearly defined. In this study, we thus used the NEMOΔhepa model to investigate the role of p21 during inflammation-driven liver carcinogenesis. The NEMOΔhepa model is of specific interest as loss of NEMO immunoreactivity in liver cells is also found in a substantial proportion of human HCCs and in the mouse model, it resembles stage-dependent evolution of chronic liver disease as also found in humans, leading to fibrosis progression and the occurrence of HCC (1, 11, 12).

Our group showed that deletion of NEMO in hepatocytes triggers increased p21 expression (13). In this study, we aimed to investigate whether p21 overexpression in NEMOΔhepa livers has a tumor suppressive or oncogenic function. We demonstrate here that p21 overexpression is protective in this model, as hepatocarcinogenesis and cholestasis are significantly enhanced upon deletion of p21 in NEMOΔhepa mice. Therefore, our study defines p21 as a tumor suppressor in this inflammation-triggered model for liver carcinogenesis.

Animal approach

Mice were housed and treated in accordance with the guidelines of the National Academy of Sciences (NIH publication 86-23 revised 1985) and breed in the animal facility of the University Hospital RWTH Aachen. Animal studies were approved by the regional authorities for nature, environment, and consumer protection of the state North Rhine-Westphalia (LANUV).

NEMOΔhepa and p21−/− mice were generated as previously described (3, 12). Double knockout mice (NEMOΔhepa/p21−/− mice) were generated by crossing NEMOΔhepa mice with p21−/− mice. NEMOΔhepa/TNF-R1−/− mice were generated as previously described (14) and crossed with NEMOΔhepa/p21−/− mice to generate triple knockout mice (NEMOΔhepa/p21−/−/TNF-R1−/− mice). All strains were kept in a C57BL/6/background.

Examinations were performed at the age of 8, 26, and 52 week. Liver weight and body weight were recorded; serum and liver tissue were collected.

Liver injury model

Experiments were performed on 8-week-old male mice. Lipopolysaccharide (LPS; Sigma) was injected i.p. at a concentration of 25 μg/10 g body weight and mice were sacrificed 7 hours later. For the survival experiments, female mice at the age of 8 to 10 weeks were used and observed for 12 hours.

Immunofluorescence stainings

Immunofluorescence staining was performed on frozen liver sections. For bromodeoxyuridine (BrdUrd) staining mice had to be injected with BrdUrd (AppliChem) 2 hours before sacrificing.

Hepatocyte proliferation, cell death, and DNA double-strand breaks (DSB) were quantified by counting the number of nuclei positive for ether Ki67, BrdUrd, terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL), or phospho-H2AX relative to the total nuclei per power field (200×) stained with 4′,6-diamidino-2-phenylindole (DAPI). Stained microscopic images were acquired with a Zeiss Axio Imager.Z1 microscope, AxioCam MRm or AxioCam ICc3 camera using Axiovision 4.8 software (all from Carl Zeiss, Inc.).

Microarray analysis

One hundred nanogram of total RNA was labeled with the Ambion WT expression kit (Life Technologies) and hybridized on Affymetrix GeneChip Mouse Gene 1.0 ST arrays according to the manufacturer's instructions. Data were normalized with the Robust Multi-Array Average (RMA) method and probe sets were annotated to Entrez ID's according to Dai and colleagues (15–17). Genes were filtered on an intensity level of at least 20 in at least two samples and an interquartile range >0.2 (log2 scale). Intensity based–moderated t-statistics were used to calculate significant differences (18). All data were analyzed using an in-house developed system (19). Data are available in the Gene Expression Omnibus (GEO accession number: GSE61100).

Statistical analysis

Data are presented as mean ± standard deviation of the mean. Statistical significance was determined by the Students t test.

p21 overexpression protects NEMOΔhepa mice from severe liver injury

Deletion of NEMO in hepatocytes triggers apoptosis and compensatory proliferation in the liver. Unexpectedly, the Cdk inhibitor, p21, known to inhibit cell-cycle proliferation, is overexpressed under these conditions (Fig. 1A). To characterize the role of p21 for progression of chronic liver injury, we generated NEMOΔhepa/p21−/− double knockout mice (Fig. 1A and Supplementary Fig. S1A).

Figure 1.

Increased injury in livers of NEMOΔhepa/p21−/− mice is not associated with changes in hepatocyte proliferation. A, Western blot analysis of whole liver extracts was performed using p21 antibody. GAPDH was used as a loading control. AST serum levels (B) and Western blot analysis (C) of whole liver extracts using E2F1, CcnA, and CcnD antibody are shown. GAPDH was used as a loading control. D, BrdUrd stainings are presented (lighter nuclei are BrdUrd-positive cells; darker nuclei are nuclei that were stained with DAPI; magnification, ×200). E, Cdk2 kinase assay is depicted. F, Western blot analysis of whole liver extracts using p-p27 and p18 antibody were performed. GAPDH was used as a loading control. Values are mean ± SD; n = 5 animals/time point; *, P < 0.05 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Figure 1.

Increased injury in livers of NEMOΔhepa/p21−/− mice is not associated with changes in hepatocyte proliferation. A, Western blot analysis of whole liver extracts was performed using p21 antibody. GAPDH was used as a loading control. AST serum levels (B) and Western blot analysis (C) of whole liver extracts using E2F1, CcnA, and CcnD antibody are shown. GAPDH was used as a loading control. D, BrdUrd stainings are presented (lighter nuclei are BrdUrd-positive cells; darker nuclei are nuclei that were stained with DAPI; magnification, ×200). E, Cdk2 kinase assay is depicted. F, Western blot analysis of whole liver extracts using p-p27 and p18 antibody were performed. GAPDH was used as a loading control. Values are mean ± SD; n = 5 animals/time point; *, P < 0.05 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Close modal

p21 overexpression in NEMOΔhepa livers is protective as its loss leads to an increase in serum aspartate-aminotransferase (AST) levels in NEMOΔhepa/p21−/− compared with NEMOΔhepa animals (Fig. 1B).

Because of its known role in cell-cycle control, we next investigated the impact on hepatocyte proliferation by studying the expression of proteins involved in cell-cycle regulation. Quantitative real-time PCR (qRT-PCR) analysis revealed increased mRNA expression levels of cell-cycle mediators controlling late G1- and S-phase progression such as cyclin A2 (CcnA2), cyclin E2 (CcnE2), E2F1, and E2F2 in NEMOΔhepa/p21−/− livers (Supplementary Fig. S2A–S2E). Overexpression of CcnA2 and E2F1 was confirmed by Western blot analysis, which also showed a downregulation of the mitogen-inducible cyclin D (CcnD) in p21-deleted NEMOΔhepa livers (Fig. 1C).

Unpredicted, elevation of the S-phase marker CcnA2 was not accompanied by an increase in hepatocyte proliferation, as evidenced by quantification of BrdUrd incorporation (Fig. 1D) and Ki67 staining (Supplementary Fig. S2F), showing no significant differences between NEMOΔhepa and NEMOΔhepa/p21−/− livers. To further explore this finding, we measured kinase activity of Cdk2, which is the binding partner of CcnA2 and CcnE2. Cdk2 kinase activity was only slightly detectable in NEMOf/f and p21−/− controls. In contrast, we detected increased kinase activity in both NEMOΔhepa and NEMOΔhepa/p21−/− livers to approximately the same extend, thereby confirming that loss p21 did not affect basal hepatocyte proliferation in NEMO-deficient livers (Fig. 1E). PCNA is a master regulator of DNA-synthesis–associated processes such as DNA replication and nucleotide excision repair. Western blot analysis for PCNA did not show any significant differences in its expression between NEMOΔhepa and NEMOΔhepa/p21−/− livers (Supplementary Fig. S2G).

As deletion of p21 had no impact on DNA replication, we examined potential mechanisms, which might explain the lack of an increase in DNA replication in NEMOΔhepa/p21−/− livers. We thus investigated the protein expression of the Cdk inhibitors p27 and p18 by Western blot analysis. NEMOΔhepa/p21−/− livers revealed slightly increased p27 phosphorylation and additionally p18 was upregulated (Fig. 1F). Thus, loss of p21 is likely compensated by modulation of alternate cell-cycle inhibitors, which best explains the unchanged basal cell-cycle activity in NEMOΔhepa/p21−/− livers.

Lack of p21 expression in NEMOΔhepa mice triggers increased DNA damage

DNA damage induces p53 phosphorylation, which consecutively increases p21 transcription, thereby blocking G1–S-phase transition and replication of damaged DNA. Thus, we next investigated whether this mechanism might be activated in NEMOΔhepa/p21−/− livers. We found an upregulation of p53 phosphorylation in NEMOΔhepa/p21−/− compared with NEMOΔhepa livers. In addition, the phosphorylation of c-Jun N-terminale kinases (JNK) was increased most likely as a response to increased genotoxic stress (Fig. 2A).

Figure 2.

DNA damage is enhanced in the absence of p21 in NEMOΔhepa mice. A, Western blot analysis of whole liver extracts using p-p53 and p-SAPK/JNK (p-p54/p-p46) antibody was performed. GAPDH was used as a loading control. B, pH2AX stainings are shown (lighter nuclei are pH2AX-positive cells; darker nuclei are nuclei that were stained with DAPI; magnification, ×200). C, microarray analysis shows upregulation of DNA repair genes (Rad51, FOXM1, Check2, and Brca2) mainly in p21-deficient NEMOΔhepa mice. D, BRCA1 qRT-PCR of liver mRNA was determined. Values are mean ± SD; n = 5 animals/time point; **, P < 0.01; ***, P < 0.001 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Figure 2.

DNA damage is enhanced in the absence of p21 in NEMOΔhepa mice. A, Western blot analysis of whole liver extracts using p-p53 and p-SAPK/JNK (p-p54/p-p46) antibody was performed. GAPDH was used as a loading control. B, pH2AX stainings are shown (lighter nuclei are pH2AX-positive cells; darker nuclei are nuclei that were stained with DAPI; magnification, ×200). C, microarray analysis shows upregulation of DNA repair genes (Rad51, FOXM1, Check2, and Brca2) mainly in p21-deficient NEMOΔhepa mice. D, BRCA1 qRT-PCR of liver mRNA was determined. Values are mean ± SD; n = 5 animals/time point; **, P < 0.01; ***, P < 0.001 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Close modal

We previously demonstrated that NEMOΔhepa livers display mild DNA double-strand breakage (13). On the basis of the shown differences in NEMOΔhepa/p21−/− livers, we next focused more specifically on the relevance of p21 deletion for DNA damage in these mice. Loss of p21 expression strongly and significantly enhanced DNA damage in NEMOΔhepa/p21−/− compared with NEMOΔhepa livers as evidenced by pH2AX staining (Fig. 2B). To better characterize these findings, we performed array analysis to detect changes in the expression of genes involved in DNA damage. Interestingly, we found an upregulation of the DNA checkpoint kinase Chk2 and further genes involved in DNA repair such as the transcription factor FoxM1 and the recombinase Rad51 in NEMOΔhepa/p21−/− compared with NEMOΔhepa livers (Fig. 2C). FoxM1 controls the expression of the DNA repair gene Rad51, and Rad51, in turn, interacts with the tumor-suppressor BRCA1. A strong and significant mRNA upregulation of the DNA damage marker BRCA1 was observed in NEMOΔhepa/p21−/− livers (Fig. 2D). Hence, our results demonstrated that after loss of p21 expression in NEMOΔhepa livers the DNA damage as well as the expression of genes involved in DNA repair were increased. The enhanced DNA damage in the double knockout mice also explains the increased liver damage (Fig. 1B).

Inflammation is ameliorated in the absence of p21 in the NEMOΔhepa liver

The inflammatory response leading to chronic hepatitis is important to drive liver disease progression in NEMOΔhepa mice. CD45 immunohistochemistry pointed to a strong inflammatory response in NEMOΔhepa livers. Unexpectedly, p21 deletion reduced CD45 staining in NEMOΔhepa/p21−/− livers (Fig. 3A). This finding was further confirmed by qRT-PCR (Supplementary Fig. S3). The infiltrating cells were also positive for the macrophage marker F4/80 (Fig. 3B). In addition, these findings were associated with a reduced expression of TNF and upregulation of its negative regulator TIMP3 (Fig. 3C and D). Together, these findings suggest that the inflammatory response is reduced in NEMOΔhepa/p21−/− livers arguing that the changes in DNA damage might be relevant to explain the phenotype of NEMOΔhepa/p21−/− livers.

Figure 3.

Inflammation is attenuated in p21-deficient NEMOΔhepa mice. A, CD45 staining was performed (brown, CD45-positive cells; blue, nuclei; magnification, ×100). B, F4/80 staining (red, F4/80-positive cells; blue, nuclei that were stained with DAPI; magnification, ×200). C and D, qRT-PCR of liver mRNA was determined for TNF (C) and TIMP3 (D). Values are mean ± SD; n = 5 animals/time point; *, P < 0.05; **, P < 0.01 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Figure 3.

Inflammation is attenuated in p21-deficient NEMOΔhepa mice. A, CD45 staining was performed (brown, CD45-positive cells; blue, nuclei; magnification, ×100). B, F4/80 staining (red, F4/80-positive cells; blue, nuclei that were stained with DAPI; magnification, ×200). C and D, qRT-PCR of liver mRNA was determined for TNF (C) and TIMP3 (D). Values are mean ± SD; n = 5 animals/time point; *, P < 0.05; **, P < 0.01 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Close modal

p21 deletion increases the sensitivity toward DNA damage

LPS has been shown to induce DNA strand breakage (20). Therefore, we tested whether loss of p21 expression increases sensitivity against LPS-driven DNA damage. In WT animals, (NEMOf/f) minor DNA double-strand breakage, as detected by pH2AX staining, was observed after LPS stimulation (Fig. 4A). In p21−/− livers, the amount of pH2AX-positive cells was significantly increased, indicating that p21 is involved in mediating protection against DNA damage. In NEMOΔhepa livers, LPS triggered strong DNA damage, which was significantly enhanced in NEMOΔhepa/p21−/− livers. These findings further support our idea that p21 counteracts DNA damage in NEMO-deficient hepatocytes.

Figure 4.

p21 overexpression is beneficial to reduce LPS-mediated DNA damage in NEMOΔhepa mice. A, pH2AX stainings are depicted (red, pH2AX-positive cells; blue, nuclei that were stained with DAPI; magnification, ×100) of frozen liver sections from mice 7 hours after LPS treatment. B, ALT serum levels were evaluated. C, H&E stainings from liver sections of LPS-treated mice are shown (magnification, ×200). D, TUNEL staining (green, TUNEL-positive cells; blue, nuclei that were stained with DAPI; magnification, ×100) of frozen liver sections from mice 7 hours after LPS treatment are presented. Values are mean ± SD; n = 5 animals/time point; **, P < 0.01; ***, P < 0.001 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Figure 4.

p21 overexpression is beneficial to reduce LPS-mediated DNA damage in NEMOΔhepa mice. A, pH2AX stainings are depicted (red, pH2AX-positive cells; blue, nuclei that were stained with DAPI; magnification, ×100) of frozen liver sections from mice 7 hours after LPS treatment. B, ALT serum levels were evaluated. C, H&E stainings from liver sections of LPS-treated mice are shown (magnification, ×200). D, TUNEL staining (green, TUNEL-positive cells; blue, nuclei that were stained with DAPI; magnification, ×100) of frozen liver sections from mice 7 hours after LPS treatment are presented. Values are mean ± SD; n = 5 animals/time point; **, P < 0.01; ***, P < 0.001 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Close modal

Next, we tested the impact of LPS stimulation on liver injury. In agreement with our previous studies, NEMOΔhepa mice showed a severe increase in ALT and AST levels after LPS treatment, which was nearly doubled in NEMOΔhepa/p21−/− animals (Fig. 4B and Supplementary Fig. S5A). This observation was also reflected in liver histology as determined by H&E staining. Here, NEMOΔhepa/p21−/− animals showed severe hemorrhages within the liver tissue, which was not evident in NEMOΔhepa livers and the respective controls (Fig. 4C). To further characterize the mode of cell death, we performed TUNEL staining and found that the number of TUNEL-positive areas was significantly highest in NEMOΔhepa/p21−/− livers, suggesting that the increase in DNA damage and higher liver injury was associated with increased apoptosis (Fig. 4D).

To characterize the inflammatory response after LPS stimulation, we performed qRT-PCR analysis for IL1β expression. Interestingly, we found significant IL1β downregulation in p21 single knockout mice after LPS stimulation. However, no significant difference was found between NEMOΔhepa single and double knockout mice showing a comparable reactivity after LPS stimulation (Supplementary Fig. S4A). Furthermore, we analyzed IL6 expression levels by qRT-PCR and found that NEMOΔhepa and NEMOΔhepa/p21−/− livers also showed comparable IL6 levels (Supplementary Fig. S4B). As IL1β and IL6 regulation are not different between both mouse strains after LPS stimulation, these results indicate that the higher sensitivity of NEMOΔhepa/p21−/− hepatocytes also explain the higher mortality of these animals.

The oxidative stress response in the liver of LPS-treated mice was measured by CM-H2DCFDA staining to detect reactive oxygen species (ROS; Supplementary Fig. S4C). The presence of oxidative stress in NEMOΔhepa livers has been already described in our previous studies (11, 13). Our present result shows that p21 deletion itself does not cause ROS accumulation. We did observe by CM-H2DCFDA staining that NEMOΔhepa and NEMOΔhepap21−/− mice have elevated ROS levels in comparison with control mice (NEMOf/f and p21−/−). However, deleting p21 in NEMOΔhepa mice did not enhance ROS production in the liver. This has been observed in LPS-treated (Supplementary Fig. S4C) and untreated mice (data not shown). The microscopic pictures of the LPS-treated group are included in Supplementary Fig. S4. Therefore, we conclude that oxidative stress does not account for higher DNA damage in NEMOΔhepap21−/− livers.

Recently, we have described that TNF-R1 deletion in NEMOΔhepa livers rescues hepatocyte injury and apoptosis (14). Therefore, we crossed NEMOΔhepa/p21−/− with TNF-R1−/− animals to generate NEMOΔhepa/p21−/−/TNF-R1−/− mice (Supplementary Fig. S1B).

We stimulated NEMOΔhepa, NEMOΔhepa/p21−/−, NEMOΔhepa/p21−/−/TNF-R1−/−, and respective controls with LPS. Seven hours after LPS stimulation, TNF-R1−/− deletion resulted in a substantial improvement in liver injury compared with NEMOΔhepa and NEMOΔhepa/p21−/− mice (Supplementary Fig. S5B). H&E staining demonstrated that acute liver hemorrhage was completely prevented in triple knockout mice (Supplementary Fig. S5C). In addition, DNA double-strand breakage was significantly reduced after TNF-R1 deletion in NEMOΔhepa/p21−/− livers (Supplementary Fig. S5D). Importantly, 57% of NEMOΔhepa/p21−/− mice did not survive the first 7 hours after LPS stimulation, while none of the WT mice and only 14% of the NEMOΔhepa animals died during this observation period. Interestingly, deletion of TNF-R1 was highly beneficial as all NEMOΔhepa/p21−/−/TNF-R1−/− animals survived the challenge (Supplementary Fig. S5E). This suggests that the LPS-hypersensitivity due to lack of p21 is mediated via TNF-dependent signaling.

NEMOΔhepa/p21−/− livers show increased cholestasis associated with changes in biliary architecture

Macroscopic analysis of livers from 52-week-old NEMOΔhepa/p21−/− animals revealed an overall yellowish appearance (Fig. 5A). As a result, we studied markers of cholestasis and found that alkaline phosphatase (AP; Fig. 5B), total and direct bilirubin levels (Supplementary Fig. S6A and S6B) were significantly increased in the serum of NEMOΔhepa/p21−/− animals. Hence, NEMOΔhepa/p21−/− mice suffered from severe cholestasis. Histologic examination of NEMOΔhepa livers revealed enhanced steatotic lesion, whereas NEMOΔhepa/p21−/− livers showed large necrotic areas frequently located around bile ducts (Fig. 5C and D).

Figure 5.

Development of severe cholestasis is dependent on p21 expression. A, macroscopic images of a cholestatic liver obtained from NEMOΔhepa/p21−/− mice are shown. B, serum levels of AP are depicted. C, H&E staining (magnification, ×100); error bars in the NEMOΔhepa liver indicate areas of steatosis and in the NEMOΔhepa/p21−/− liver areas of necrosis. D, H&E staining showing large areas of necrosis and cell death in double knockout livers (magnification, ×400). E, CK19 staining of HCC areas are presented (green, CK19-positive cells; blue, nuclei that were stained with DAPI; magnification, ×200). F, CK19 staining of nontumor areas are shown (green, CK19-positive cells; blue, nuclei that were stained with DAPI; magnification, ×200). Values are mean ± SD; n = 5–10 animals/time point; **, P < 0.01 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−].

Figure 5.

Development of severe cholestasis is dependent on p21 expression. A, macroscopic images of a cholestatic liver obtained from NEMOΔhepa/p21−/− mice are shown. B, serum levels of AP are depicted. C, H&E staining (magnification, ×100); error bars in the NEMOΔhepa liver indicate areas of steatosis and in the NEMOΔhepa/p21−/− liver areas of necrosis. D, H&E staining showing large areas of necrosis and cell death in double knockout livers (magnification, ×400). E, CK19 staining of HCC areas are presented (green, CK19-positive cells; blue, nuclei that were stained with DAPI; magnification, ×200). F, CK19 staining of nontumor areas are shown (green, CK19-positive cells; blue, nuclei that were stained with DAPI; magnification, ×200). Values are mean ± SD; n = 5–10 animals/time point; **, P < 0.01 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−].

Close modal

These significant differences prompted us to study the impact on CK19-positive cells and we thus performed immunofluorescence staining. This staining revealed that CK19-positive cells in NEMOΔhepa livers are located in HCC samples around bile ducts, while in NEMOΔhepa/p21−/− HCCs single CK19-positive cells were equally distributed in the tissue (Fig. 5E). In contrast, no differences in CK19 staining were found in nontumorous areas of NEMOΔhepa and NEMOΔhepa/p21−/− livers (Fig. 5F).

p21 deletion promotes hepatocarcinogenesis

The long-term consequences of increased injury in NEMOΔhepa/p21−/− livers were assessed in 52-week-old animals. Macroscopically NEMOΔhepa/p21−/− livers showed more tumor nodules compared with NEMOΔhepa mice (Fig. 6A). The tumors displayed broadening of liver cell cords and loss of the reticulin network. To microscopically classify the observed nodules, we performed stainings with established HCC markers like Golgi protein-73 (GP73) and glutamine synthetase (GS; Fig. 6B), demonstrating that tumors in both strains classified as HCCs. Liver/body weight ratio was significantly enhanced in NEMOΔhepa/p21−/− livers (Fig. 6C) as well as liver injury markers like ALT and AST (Fig. 6D and Supplementary Fig. S7). Quantification of hepatocarcinogenesis revealed a significantly higher number of tumors per liver in NEMOΔhepa/p21−/− compared with NEMOΔhepa mice (Fig. 6E), while the average size of these tumors was similar between the groups (Fig. 6F).

Figure 6.

Enhanced hepatocarcinogenesis due to loss of p21 expression. A, macroscopic images of livers from 52-week-old mice are shown. B, paraffin-embedded liver tumor sections were stained for H&E, GP73, and GS. C, liver weight/body weight ratio was calculated. D, ALT serum levels were determined. E, tumor number per mice is shown. F, largest tumor per mice is shown. Values are mean ± SD; n = 5–10 animals/time point; *, P < 0.05; **, P < 0.01; ***, P < 0.001 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−].

Figure 6.

Enhanced hepatocarcinogenesis due to loss of p21 expression. A, macroscopic images of livers from 52-week-old mice are shown. B, paraffin-embedded liver tumor sections were stained for H&E, GP73, and GS. C, liver weight/body weight ratio was calculated. D, ALT serum levels were determined. E, tumor number per mice is shown. F, largest tumor per mice is shown. Values are mean ± SD; n = 5–10 animals/time point; *, P < 0.05; **, P < 0.01; ***, P < 0.001 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−].

Close modal

Earlier tumor onset in 26-week-old double knockout mice

After we found increased tumor burden in 52-week-old NEMOΔhepa/p21−/− livers, we studied whether changes in tumor growth and cell proliferation were evident at earlier time points. Twenty-six-week-old NEMOΔhepa/p21−/− livers showed already macroscopically significant differences compared with NEMOΔhepa mice. The liver surface of NEMOΔhepa/p21−/− animals showed a more irregular surface and small nodules could be detected on the hepatic surface (Fig. 7A). These changes were associated with a higher liver:body weight ratio (Fig. 7B). In addition, at this time point more Ki67-positive cells were found in the liver of NEMOΔhepa/p21−/− animals, suggesting higher proliferation in 26-week-old livers (Fig. 7C).

Figure 7.

Twenty-six-week-old double knockout mice show earlier tumor formation. A, macroscopic images of livers from 26-week-old mice are depicted. B, liver weight/body weight ratio was calculated. C, Ki67 stainings are presented (lighter nuclei are Ki67-positive cells; darker nuclei are nuclei that were stained with DAPI; magnification, ×200). Values are mean ± SD; n = 5 animals/time point; **, P < 0.01; ***, P < 0.001 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Figure 7.

Twenty-six-week-old double knockout mice show earlier tumor formation. A, macroscopic images of livers from 26-week-old mice are depicted. B, liver weight/body weight ratio was calculated. C, Ki67 stainings are presented (lighter nuclei are Ki67-positive cells; darker nuclei are nuclei that were stained with DAPI; magnification, ×200). Values are mean ± SD; n = 5 animals/time point; **, P < 0.01; ***, P < 0.001 [NEMOf/f vs. p21−/−, NEMOΔhepa vs. NEMOΔhepa/p21−/−]. For immunofluorescence staining, 6 to 10 sections were quantified.

Close modal

To better characterize the 26-week tumors, we performed staining for CK19, Gp73, and GS. Our result shows that in NEMOΔhepa as well as in NEMOΔhepap21−/− livers GP73-positive cells are partially present in the tumor area, whereas the tumors are in both cases negative for GS (Supplementary Fig. S8). As shown in Supplementary Fig. S8, we did not observe any signs of malignant growth at this age. Therefore, these tumors have been classified as adenomas.

The role of p21 for tumorigenesis is not fully understood. Convincing studies demonstrated that p21 can act as an oncogene as well as a tumor suppressor. In this study, we aimed to investigate the relevance of p21 for hepatocarcinogenesis in the NEMOΔhepa mouse model. This model is of clinical interest as these animals develop a cascade of events as also found in humans, which leads from chronic hepatitis to liver cirrhosis and growth of hepatocellular carcinomas (HCC). Interestingly, we found in human HCC that NEMO is downregulated thus indicating that results found in NEMOΔhepa livers might have a direct relevance for human liver diseases (1).

In former studies we found in NEMOΔhepa livers an overexpression of the cell-cycle inhibitor p21 (13). To study its relevance for disease progression, we generated NEMOΔhepa/p21−/− double knockout mice and examined the relevance of p21 deletion for disease progression and its sensitivity toward LPS-induced liver injury.

Because p21 is a cell-cycle inhibitor, we first investigated the impact on cell proliferation in NEMOΔhepa/p21−/− livers. Interestingly in 8-week-old animals p21 deletion had no influence on the proliferation of hepatocytes. This was an unexpected result because p21 binds to CcnE/cdk2 and CcnA/cdk2 complexes thereby preventing progression from G1- to S-phase. At present we propose that loss of p21 expression is compensated by the increased activity of other cell-cycle inhibitors such as p-p27 and p18, which is supported by our present data.

Furthermore, we found that deletion of p21 in NEMOΔhepa mice was beneficial in terms of inflammation as there was a clear amelioration of inflammatory cells and cytokines (TNF) in NEMOΔhepa/p21−/− livers. Reduced TNF expression is most likely caused by increased TIMP3 levels in the livers. TIMP3 is able to inhibit TACE, a protease that generates soluble TNF from the cell surface form of the cytokine (21). Hence, TIMP3 is an important negative TNF regulator (22). As a consequence of reduced cytokine production, the mitogenic stimulation of CcnD was decreased in 8-week-old NEMOΔhepa/p21−/− livers.

Despite a reduced inflammatory response, the loss of p21 expression triggered increased liver injury in NEMOΔhepa mice, which was reflected by elevated serum transaminases (AST). Further analysis revealed that increased liver injury in the double knockout mice is mediated through enhanced DNA damage. As a consequence of higher DNA damage, the expression of important mediators for DNA repair (BRCA1, RAD51 and FoxM1) were enhanced in NEMOΔhepa/p21−/− livers. However, several recent studies suggested that overexpression of DNA repair proteins is associated with a disadvantage for the treatment of breast cancer, glioblastoma multiforme and human soft tissue sarcoma cells, as they mediate resistance to chemotherapy. This has been nicely shown for FoxM1 and RAD51 mediating resistance against chemotherapeutics like Doxorubicin (23–26). Thus, these findings indicated that loss of p21 in NEMOΔhepa liver may lead to more aggressive disease progression and likely a higher rate of HCCs.

A responsible mechanism for exacerbation of DNA damage in p21-deficient NEMOΔhepa mice could be the elevated expression of CcnA2 and CcnE2, which did not cause enhanced hepatocyte proliferation in 8-week-old animals. Recent publications showed that ectopic overexpression of CcnA or CcnE in mouse embryonic fibroblast (MEFs) lead to an increase in DNA double-strand breakage (27). The observed downregulation of CcnD, a protein also known to be involved in DNA-repair mechanisms (28), might additionally contribute to enhanced DNA damage in NEMOΔhepa/p21−/− livers (28).

Hepatocyte-specific deletion of NEMO inhibits activation of NF-κB and causes substantial apoptosis in the liver after LPS-injection (11). p21 has an additional function in the suppression of autoimmunity and in the inhibition of apoptosis (29, 30). We therefore investigated if p21 overexpression in NEMOΔhepa livers can provide protection after challenging the animals with LPS.

Our LPS experiments strongly suggest that p21 has a yet unknown function in protecting from DNA damage. As LPS-dependent cytotoxicity is mainly mediated through TNF, we tested if the p21-dependent effect on DNA damage is also related to this pathway and deleted TNF-R1 in NEMOΔhepa/p21−/− mice. The triple knockout mice (NEMOΔhepa/p21−/−/TNF-R1−/−) show a massive reduction in DNA damage and cell death. In addition, AST serum values and liver histology was significantly improved. Finally, all triple knockout mice survived the LPS challenge, suggesting that the LPS-dependent effect on DNA damage in p21 deleted animals is mediated via TNF.

As p21 deletion triggered higher DNA damage, resulting in higher liver injury, we examined the relevance for spontaneous hepatocarcinogenesis in aging mice. Here, the protective role of p21 was already visible in NEMOΔhepa mice at the age of 26 weeks. Double knockout mice at this age had already a higher liver weight/body weight ratio and more frequently showed small tumors. At this age p21 deletion also resulted in enhanced hepatocyte proliferation as evidenced and quantified by Ki67 staining. Finally, loss of p21 overexpression caused exacerbation of hepatocarcinogenesis, which resulted in a significantly increased number of HCCs in 52-week-old animals. Interestingly, only the number of HCC nodules was increased and not the size of the tumors itself, suggesting that the loss of p21 has more impact on tumor initiation than on tumor progression.

In a study of Maeda and colleagues (31) was shown that a brief oral administration of an antioxidant (BHA, butylated hydroxyanisole) around the time of DEN exposure prevented excessive DEN-induced carcinogenesis in IkkβΔhep mice. In our chronic model, ROS production is not the cause for enhanced hepatocarcinogenesis in NEMOΔhepap21−/− mice, because we did not observe enhanced ROS levels compared with NEMOΔhepa single knockout mice.

p21 has been recently shown to promote hepatocarcinogenesis in chronic cholestatic liver injury, as loss of p21 in Mdr2−/− mice significantly delayed tumor development (32). Mdr2−/− mice are lacking the Abc4 protein, which is encoded by the multidrug resistance-2 gene; as a consequence, they develop chronic periductular inflammation and cholestatic liver disease, resulting in the development of HCC. Inhibition of NF-κB by expression of an IκB super-repressor (IκBαSR) transgene in hepatocytes has been shown to strongly reduce HCC development in Mdr2−/− mice, suggesting that NF-κB acts as a tumor promoter in the Mdr2−/− model (33). Whereas in the NEMOΔhepa model used in our study, the lack of NFκB activation in hepatocytes leads to HCC development. This suggests that the role of p21 to act as an oncogene or tumor suppressor depends on NFκB.

In addition, disease progression in NEMOΔhepa/p21−/− livers was associated with the occurrence of a cholestatic phenotype. As oval cells can differentiate into hepatocytes and cholangiocyctes, we analyzed CK19-positive cells. Analysis of HCCs obtained from NEMO single and double knockout mice showed an expression of CK19-positive cells, which in NEMOΔhepa HCCs were located around bile ducts, while was spread diffusely throughout the tissue in HCCs of NEMOΔhepa/p21−/− livers. CK19-positive HCCs are known to be more malignant than CK19-negative HCCs, due to the high recurrence frequency after operative resection in patients (34). Especially primary liver cancers with a more biliary phenotype are known to have a poorer prognosis after surgical resection (35). Furthermore, it is assumed that activation of the oval cell compartment in a setting of chronic injury initiates or promotes HCC development (36). Strong oval cell activation has been found in preneoplastic livers of Mdr2−/− mice, which was severely impaired in Mdr2−/−/Rage−/− livers and results in reduced HCC formation (36).

In summary, our results demonstrate that p21 overexpression in NEMOΔhepa mice reduces liver disease progression. Importantly, our results show that p21 is involved in protecting from DNA damage triggered by the inflammatory stress found in these livers, which was further supported by our LPS experiments. As evidenced in the aging experiments, these changes are associated with increased tumor initiation but not increased tumor progression. Thus, we describe for the first time an additional protective function of p21 against DNA damage.

M.V. Boekschoten is a project leader at TI Food & Nutrition. C. Doler is a Ph.D. student at Technical University of Graz. No potential conflicts of interest were disclosed by the other authors.

Conception and design: H. Ehedego, C. Trautwein

Development of methodology: W. Hu, C. Liedtke

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Ehedego, M.V. Boekschoten, J. Haybaeck, N. Gaβler, M. Müller, C. Trautwein

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Ehedego, M.V. Boekschoten, C. Doler, J. Haybaeck, M. Müller, C. Liedtke, C. Trautwein

Writing, review, and/or revision of the manuscript: H. Ehedego, M.V. Boekschoten, J. Haybaeck, M. Müller, C. Liedtke, C. Trautwein

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.V. Boekschoten, N. Gaβler, C. Liedtke

Study supervision: N. Gaβler, C. Trautwein

The authors thank Bettina Jansen for technical assistance.

This work was supported by a grant of the German Cancer Foundation (Deutsche Krebshilfe) Grant Nr. 107682 and the Interdisciplinary Centre for Clinical Research (IZKF Aachen) within the Faculty of Medicine; RWTH Aachen.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Aigelsreiter
A
,
Haybaeck
J
,
Schauer
S
,
Kiesslich
T
,
Bettermann
K
,
Griessbacher
A
, et al
NEMO expression in human hepatocellular carcinoma and its association with clinical outcome
.
Hum Pathol
2012
;
43
:
1012
9
.
2.
Crary
GS
,
Albrecht
JH
. 
Expression of cyclin-dependent kinase inhibitor p21 in human liver
.
Hepatology
1998
;
28
:
738
43
.
3.
Deng
C
,
Zhang
P
,
Harper
JW
,
Elledge
SJ
,
Leder
P
. 
Mice lacking p21CIP1/WAF1 undergo normal development, but are defective in G1 checkpoint control
.
Cell
1995
;
82
:
675
84
.
4.
Albrecht
JH
,
Poon
RY
,
Ahonen
CL
,
Rieland
BM
,
Deng
C
,
Crary
GS
. 
Involvement of p21 and p27 in the regulation of CDK activity and cell cycle progression in the regenerating liver
.
Oncogene
1998
;
16
:
2141
50
.
5.
Wu
H
,
Wade
M
,
Krall
L
,
Grisham
J
,
Xiong
Y
,
Van Dyke
T
. 
Targeted in vivo expression of the cyclin-dependent kinase inhibitor p21 halts hepatocyte cell-cycle progression, postnatal liver development and regeneration
.
Genes Dev
1996
;
10
:
245
60
.
6.
Martín-Caballero
J
,
Flores
JM
,
García-Palencia
P
,
Serrano
M
. 
Tumor susceptibility of p21(Waf1/Cip1)-deficient mice
.
Cancer Res
2001
;
61
:
6234
8
.
7.
Topley
GI
,
Okuyama
R
,
Gonzales
JG
,
Conti
C
,
Dotto
GP
. 
p21(WAF1/Cip1) functions as a suppressor of malignant skin tumor formation and a determinant of keratinocyte stem-cell potential
.
Proc Natl Acad Sci U S A
1999
;
96
:
9089
94
.
8.
Shiraki
K
,
Wagayama
H
. 
Cytoplasmic p21(WAF1/CIP1) expression in human hepatocellular carcinomas
.
Liver Int
2006
;
26
:
1018
9
.
9.
Wagayama
H
,
Shiraki
K
,
Sugimoto
K
,
Ito
T
,
Fujikawa
K
,
Yamanaka
T
, et al
High expression of p21WAF1/CIP1 is correlated with human hepatocellular carcinoma in patients with hepatitis C virus—associated chronic liver diseases
.
Hum Pathol
2002
;
33
:
429
34
.
10.
De la Cueva
E
,
García-Cao
I
,
Herranz
M
,
López
P
,
García-Palencia
P
,
Flores
JM
, et al
Tumorigenic activity of p21Waf1/Cip1 in thymic lymphoma
.
Oncogene
2006
;
25
:
4128
32
.
11.
Luedde
T
,
Beraza
N
,
Kotsikoris
V
,
van Loo
G
,
Nenci
A
,
De Vos
R
, et al
Deletion of NEMO/IKKgamma in liver parenchymal cells causes steatohepatitis and hepatocellular carcinoma
.
Cancer Cell
2007
;
11
:
119
32
.
12.
Beraza
N
,
Malato
Y
,
Sander
LE
,
Al-Masaoudi
M
,
Freimuth
J
,
Riethmacher
D
, et al
Hepatocyte-specific NEMO deletion promotes NK/NKT cell- and TRAIL-dependent liver damage
.
J Exp Med
2009
;
206
:
1727
37
.
13.
Malato
Y
,
Ehedego
H
,
Al-Masaoudi
M
,
Cubero
FJ
,
Bornemann
J
,
Gassler
N
, et al
NF-κB essential modifier is required for hepatocyte proliferation and the oval cell reaction after partial hepatectomy in mice
.
Gastroenterology
2012
;
143
:
1597
608
.
14.
Cubero
FJ
,
Singh
A
,
Borkham-Kamphorst
E
,
Nevzorova
YA
,
Al Masaoudi
M
,
Haas
U
, et al
TNFR1 determines progression of chronic liver injury in the IKKγ/Nemo genetic model
.
Cell Death Differ
2013
;
20
:
1
13
.
15.
Bolstad
BM
,
Irizarry
RA
,
Åstrand
M
,
Speed
TP
. 
A comparison of normalization methods for high density oligonucleotide array data based on variance and bias
.
Bioinformatics
2003
;
19
:
185
93
.
16.
Irizarry
RA
,
Bolstad
BM
,
Collin
F
,
Cope
LM
,
Hobbs
B
,
Speed
TP
. 
Summaries of Affymetrix GeneChip probe level data
.
Nucleic Acids Res
2003
;
31
:
e15
.
17.
Dai
M
,
Wang
P
,
Boyd
AD
,
Kostov
G
,
Athey
B
,
Jones
EG
, et al
Evolving gene/transcript definitions significantly alter the interpretation of GeneChip data
.
Nucleic Acids Res
2005
;
33
:
e175
.
18.
Sartor
MA
,
Tomlinson
CR
,
Wesselkamper
SC
,
Sivaganesan
S
,
Leikauf
GD
,
Medvedovic
M
. 
Intensity-based hierarchical Bayes method improves testing for differentially expressed genes in microarray experiments
.
BMC Bioinformatics
2006
;
7
:
538
.
19.
Lin
K
,
Kools
H
,
de Groot
PJ
,
Gavai
AK
,
Basnet
RK
,
Cheng
F
, et al
MADMAX—management and analysis database for multiple ∼omics experiments
.
J Integr Bioinform
2011
;
8
:
160
.
20.
Hoyt
DG
,
Mannix
RJ
,
Gerritsen
ME
,
Watkins
SC
,
Lazo
JS
,
Pitt
BR
. 
Integrins inhibit LPS-induced DNA strand breakage in cultured lung endothelial cells
.
Am J Physiol
1996
;
270
:
L689
94
.
21.
Black
RA
. 
TIMP3 checks inflammation
.
Nat Genet
2004
;
36
:
934
5
.
22.
Mohammed
FF
,
Smookler
DS
,
Taylor
SEM
,
Fingleton
B
,
Kassiri
Z
,
Sanchez
OH
, et al
Abnormal TNF activity in Timp3−/− mice leads to chronic hepatic inflammation and failure of liver regeneration
.
Nat Genet
2004
;
36
:
969
77
.
23.
Zhang
N
,
Wu
X
,
Yang
L
,
Xiao
F
,
Zhang
H
,
Zhou
A
, et al
FoxM1 inhibition sensitizes resistant glioblastoma cells to temozolomide by downregulating the expression of DNA repair gene Rad51
.
Clin Cancer Res
2012
;
18
:
5961
71
.
24.
Park
Y
,
Jung
SY
,
Jennings
NB
,
Rodriguez-aguayo
C
,
Peng
G
,
Lee
S
, et al
FOXM1 mediates Dox resistance in breast cancer by enhancing DNA repair
.
Carcinogenesis
2012
;
33
:
1843
53
.
25.
Klein
HL
. 
The consequences of Rad51 overexpression for normal and tumor cells
.
DNA Repair
2008
;
7
:
686
93
.
26.
Hannay
JAF
,
Liu
J
,
Zhu
Q
,
Bolshakov
SV
,
Li
L
,
Pisters
PWT
, et al
Rad51 overexpression contributes to chemoresistance in human soft tissue sarcoma cells: a role for p53/activator protein 2 transcriptional regulation
.
Mol Cancer Ther
2007
;
6
:
1650
60
.
27.
Tane
S
,
Chibazakura
T
. 
Cyclin A overexpression induces chromosomal double-strand breaks in mammalian cells
.
Cell Cycle
2009
;
8
:
3900
3
.
28.
Jirawatnotai
S
,
Hu
Y
,
Michowski
W
,
Elias
JE
,
Becks
L
,
Bienvenu
F
, et al
A function for cyclin D1 in DNA repair uncovered by protein interactome analyses in human cancers
.
Nature
2011
;
474
:
230
4
.
29.
Trakala
M
,
Arias
CF
,
Garcıa
MI
,
Moreno-Ortiz
MC
,
Tsilingiri
K
,
Fernández
PJ
, et al
Regulation of macrophage activation and septic shock susceptibility via p21(WAF1/CIP1)
.
Eur J Immunol
2009
;
39
:
810
9
.
30.
Gartel
AL
,
Tyner
AL
. 
The role of the cyclin-dependent kinase inhibitor p21 in apoptosis
.
Mol Cancer Ther
2002
;
1
:
639
49
.
31.
Maeda
S
,
Kamata
H
,
Luo
J
,
Leffert
H
,
Karin
M
. 
IKKβ couples hepatocyte death to cytokine-driven compensatory proliferation that promotes chemical hepatocarcinogenesis
.
Cell
2005
;
121
:
977
90
.
32.
Marhenke
S
,
Buitrago-Molina
LE
,
Endig
J
,
Orlik
J
,
Schweitzer
N
,
Klett
S
, et al
p21 promotes sustained liver regeneration and hepatocarcinogenesis in chronic cholestatic liver injury
.
Gut
2014
:
63
:
1501
12
.
33.
Pikarsky
E
,
Porat
RM
,
Stein
I
,
Abramovitch
R
,
Amit
S
,
Kasem
S
, et al
NF-kappaB functions as a tumour promoter in inflammation-associated cancer
.
Nature
2004
;
431
:
461
6
.
34.
Uenishi
T
,
Kubo
S
,
Yamamoto
T
,
Shuto
T
,
Ogawa
M
,
Tanaka
H
, et al
Cytokeratin 19 expression in hepatocellular carcinoma predicts early postoperative recurrence
.
Cancer Sci
2003
;
94
:
851
7
.
35.
Lee
JI
,
Lee
JW
,
Kim
JM
,
Kim
JK
,
Chung
HJ
,
Kim
YS
. 
Prognosis of hepatocellular carcinoma expressing cytokeratin 19: comparison with other liver cancers
.
World J Gastroenterol
2012
;
18
:
4751
7
.
36.
Pusterla
T
,
Nèmeth
J
,
Stein
I
,
Wiechert
L
,
Knigin
D
,
Marhenke
S
, et al
Receptor for advanced glycation endproducts (RAGE) is a key regulator of oval cell activation and inflammation-associated liver carcinogenesis in mice
.
Hepatology
2013
;
58
:
363
73
.