Constitutive Activation of the Tumor Suppressor p53 in Hepatocytes Paradoxically Promotes Non-Cell Autonomous Liver Carcinogenesis

Abstract

In chronic liver diseases (CLD), p53 is constitutively activated in hepatocytes due to various etiologies as viral infection, ethanol exposure, or lipid accumulation. This study was aimed to clarify the significance of p53 activation on the pathophysiology of CLDs. In Kras-mutant liver cancer model, murine double minute 2 (Mdm2), a negative regulator of p53, was specifically deleted in hepatocytes [Alb-Cre KrasLSL-G12D Mdm2fl/fl (LiKM; KrasG12D mutation and Mdm2 loss in the liver)]. Accumulation of p53 and upregulation of its downstream genes were observed in hepatocytes in LiKM mice. LiKM mice showed liver inflammation accompanied by hepatocyte apoptosis, senescence-associated secretory phenotype (SASP), and the emergence of hepatic progenitor cells (HPC). More importantly, Mdm2 deletion promoted non-cell autonomous development of liver tumors. Organoids generated from HPCs harbored tumor-formation ability when subcutaneously inoculated into NOD/Shi-scid/IL2Rγ (null) mice. Treatment with acyclic retinoid suppressed growth of HPCs in vitro and inhibited tumorigenesis in LiKM mice. All of the phenotypes in LiKM mice, including accelerated liver tumorigenesis, were negated by further deletion of p53 in hepatocytes (Alb-Cre KrasLSL-G12D Mdm2fl/fl p53fl/fl). Activation of hepatic p53 was noted in liver biopsy samples obtained from 182 patients with CLD, in comparison with 23 normal liver samples without background liver diseases. In patients with CLD, activity of hepatic p53 was positively correlated with the expression of apoptosis, SASP, HPC-associated genes and tumor incidence in the liver after biopsy. In conclusion, activation of hepatocyte p53 creates a microenvironment prone to tumor formation from HPCs. Optimization of p53 activity in hepatocytes is important to prevent patients with CLD from hepatocarcinogenesis.

Significance: This study reveals that activation of p53 in hepatocytes promotes liver carcinogenesis derived from HPCs, which elucidates a paradoxical aspect of a tumor suppressor p53 and novel mechanism of liver carcinogenesis. See related commentary by Barton and Lozano, p. 2824.

Graphical abstract

Figure 1.

Hepatocyte-specific Mdm2 deletion in LiK (Kras^G12D mutation specifically in the liver) mice promoted the development of liver tumors. Hepatocyte-specific Kras^G12D-mutant mice [Kras^LSL-G12D/+ Alb-Cre; LiK (Kras^G12D mutation in the liver) mice] and Mdm2-floxed mice were mated to generate hepatocyte-specific Mdm2-knockout Kras^G12D mice. After the mating of Mdm2^fl/+ Kras^+/+Alb-Cre mice and Mdm2^fl/+ Kras^LSL-G12D/+ mice, the phenotypes of the following offspring groups were analyzed at 4 months of age (A, B, and D) or subjected to survival analysis (C): (i) Kras^G12D Mdm2^+/+ (LiK; Mdm2^+/+ Kras^LSL-G12D/+ Alb-Cre); (ii) Kras^G12D Mdm2^+/del (Mdm2^fl/+ Kras^LSL-G12D/+ Alb-Cre; LiKM [Kras^G12D mutation and Mdm2 loss in the liver)-hetero mice]; and (iii) Kras^G12D Mdm2^del/del (Mdm2^fl/fl Kras^LSL-G12D/+ Alb-Cre; LiKM mice) littermates. A, Macroscopic appearance of the livers. B, Macroscopic tumor number, maximum tumor diameter, and liver to body weight ratio (N = 10–21 per group). C, Kaplan–Meier survival analysis (N = 12–14 per group). D, Hematoxylin and eosin staining in the liver (×100). Arrowheads, tumors. E and F, LiKM mice were mated with ROSA26-LacZ mice, and the Kras^G12D Mdm2^del/del ROSA26-LacZ offspring mice were sacrificed at 4 months of age. E, β-Galactosidase staining of the liver (×100, left; ×200, right). F, Genotyping of the Mdm2 gene in tumorous (T) and nontumorous (NT) tissues in Kras^G12D Mdm2^del/del ROSA26-LacZ mice. Tissue DNA was extracted from both tumorous and nontumorous liver tissues collected via laser microdissection after β-galactosidase staining. The DNA band was separated by electrophoresis after PCR. *, P < 0.05.

Figure 2.

Mdm2 deletion in hepatocytes induced p53 activation and liver inflammation accompanied by hepatocyte apoptosis, SASP, and inflammatory cytokine production in LiKM mice. Liver samples were obtained from LiK, LiKM-hetero, and LiKM mice at 6 weeks of age. A, Serum ALT (N = 9 per group). B, Serum caspase-3/7 activity (N = 6 per group). C–E, RNA-seq of the liver tissues in control (LiK) and target (LiKM) mice (N = 3 per group). C, Heatmap of DEGs. D, Enrichment analysis of DEGs in LiKM mice by PANTHER pathways and WikiPathways. Pathways noted in red showed inflammation-associated pathways. E, mRNA expression of inflammation-associated molecules. Genes noted in red showed significant expression changes (P < 0.05). F, qPCR analysis of the mRNA expression in whole liver tissue (N = 7–9 per group). G, qPCR analysis of the mRNA expression in hepatocytes and NPCs isolated from the liver (N = 7–10 per group). H, IHC for p53 (×400), p21 (×400), F4/80 (×200), and CD3 (×200), as well as TUNEL staining (×400), and SA-β-gal staining (×400) of the liver and the number of positive hepatocytes (N = 4–7). Arrows, positive hepatocytes. *, P < 0.05.

Figure 3.

Mdm2 deletion in hepatocytes induced the emergence of HPCs in LiK mice. Liver samples were obtained from LiK, LiKM-hetero, and LiKM mice at 6 weeks of age (A–E) or 4 months of age (E). A and B, RNA-seq of liver tissues from LiK and LiKM mice (N = 3 per group). Enrichment analysis of DEGs by GenesigDB (A). Stem cell–related pathways are shown in red. mRNA expression of HPC markers (B). Genes noted in red showed significant expression changes (P < 0.05). C, qPCR analysis of the mRNA expression of markers of HPC and soluble factors involved in HPC activation in whole liver tissues (N = 7–9 per group). D, qPCR analysis of the mRNA expression of soluble factors involved in HPC activation in hepatocytes and NPCs isolated from the liver (N = 6–10 per group). E, IHC for pan CK, AFP, and CD133 in the liver at 6 weeks (×100) and 4 months (×200) of age and frequency of positive cells in the liver in LiKM mice (N = 3–4 per group). F, Gene set enrichment analysis (GSEA) of AFP-positive cells compared with AFP-negative cells in RNA-seq. AFP-positive and negative cells were extracted from liver tissues of LiKM mice at 4 months of age via laser microdissection. Total RNA was isolated and subjected to RNA-seq (N = 3 per group). *, P < 0.05.

Figure 4.

HPC-derived organoids generated from LiKM mice showed tumor-formation ability in an allograft model. Organoids were generated from 0.50 g liver samples in LiK and LiKM mice. A, Representative pictures and numbers of organoids larger than 200 μmol/L 10 days after isolation from the livers (N = 7 per group). B, PCR of the DNA isolated from the organoids, whole livers, and tails of the indicated genotypes and specimens. C, Karyotyping of organoid-consisting cells (N = 100 per group). Two lines of organoids were generated in each genotype. Fifty cells were examined for karyotyping in each line. D, Protein expression in organoids. E and F, Organoids were inoculated into NOG mice for allografts. E, Macroscopic appearance of mice 21 days after organoid inoculation and the tumorigenesis rate. Arrowheads, allograft. F, Hematoxylin and eosin (H&E) staining and IHC for pan CK, AFP, and CD133 in allografts (×400). *, P < 0.05.

Figure 5.

ACR inhibited HPC expansion and suppressed tumorigenesis in LiKM mice. LiK and LiKM mice were fed a normal diet (ND) or a diet containing 0.06% ACR from 5 weeks of age until sacrifice at 4 months of age. A, Representative images of the livers. B, Macroscopic tumor number, maximum tumor diameter, and liver to body weight ratio (N = 14–17 per group). C, Serum ALT (N = 7–17 per group). D and E, RNA-seq of the liver tissues in LiK and LiKM mice (N = 3 per group). Total RNA was extracted from nontumorous liver tissues at 4 months of age. Heatmap of DEGs and enrichment analysis between LiK and LiKM mice fed a normal diet (D) and LiKM mice fed a normal or ACR diet (E). F, Pathways noted in red showed inflammation- or stem cell–related pathways. mRNA expression in the liver (N = 5–14 per group). G, IHC for pan CK and CD133 in the liver (×100). *, P < 0.05.

Figure 6.

Concurrent deletion of hepatocyte p53 suppressed hepatocyte apoptosis and SASP, HPC expansion, and development of HCC in LiKM mice. LiKM mice and p53-floxed mice were mated to generate hepatocyte-specific Mdm2 and p53-double knockout Kras^G12D mice. After mating p53^fl/+ Mdm2^fl/+ Kras^+/+Alb-Cre mice and p53^fl/+ Mdm2^fl/+ Kras^LSL-G12D/+ mice, the Kras^G12D Mdm2^del/del p53^+/+ (p53^+/+ mdm2^fl/fl Kras^LSL-G12D/+ Alb-Cre; LiKM mice) and Kras^G12D Mdm2^del/del p53^del/del [p53^fl/fl mdm2^fl/fl Kras^LSL-G12D/+ Alb-Cre; LiKMP (Kras^G12D mutation and Mdm2/p53 double loss in the liver) mice] offspring littermates were sacrificed at 6 weeks (A–D) or 4 months (E–H) of age. LiKMP mice were mated with ROSA26-LacZ mice to generate Kras^G12D Mdm2^del/del p53^+/+ ROSA26-LacZ and Kras^G12D Mdm2^del/del p53^del/del ROSA26-LacZ mice and were sacrificed at 4 months of age (I). A, Serum ALT (N = 6–9 per group). B, Serum caspase-3/7 activity (N = 7–8 per group). C, qPCR analysis of the mRNA expression in whole liver tissue (N = 6–9 per group). D, IHC for p53 (×400), p21 (×400), F4/80 (×200), CD3 (×200), pan CK (×100), CD133 (×100), and AFP (×100), as well as TUNEL staining (×400) and SA-β-gal staining (×400) of the liver and the number of positive hepatocytes. E, Macroscopic images of the livers. F, Macroscopic tumor number, maximum tumor diameter, and liver to body weight ratio in the Mdm2-KO groups (N = 12–34 per group). G, Hematoxylin and eosin (H&E) staining of the liver tissues in LiKM mice and LiKMP mice (left, poorly differentiated tumor; right, cholangiocellular carcinoma; ×200). H, Genotyping of the trp53 and mdm2 genes in liver tumors in 4-month-old LiKMP and control mouse tail samples. I, β-Galactosidase staining of the livers of Kras^G12D Mdm2^del/del p53^+/+ ROSA26-LacZ and Kras^G12D Mdm2^del/del p53^del/del ROSA26-LacZ mice (left, poorly differentiated tumor; right, cholangiocellular carcinoma; ×200). The picture of cholangiocellular carcinoma in Kras^G12D Mdm2^del/del p53^del/del ROSA26-LacZ mice was obtained from a female mouse. *, P < 0.05.

Figure 7.

Hepatic p53 activity was associated with apoptosis, SASP, HPCs, and hepatocarcinogenesis in patients with CLD. Human liver biopsy samples were obtained from 182 patients with CLDs and 23 patients without underlying liver diseases as normal control livers. A, Flow chart of patient and sample selection and analysis. B, mRNA expression of p21. C, mRNA expression of p53 target, senescence/inflammation-related, and HPC-related genes. D, Correlation of the mRNA expression in the bulk liver tissues between p53 target, senescence/inflammation-related, and HPC-related genes and p21 in 182 patients with CLD. E, Cumulative liver tumor incidence rate of the p21 mRNA-high and p21 mRNA-low expression groups of patients with 144 HCV-related CLDs. Patients were divided into the p21-high expression group (N = 72) and the p21-low expression group (N = 72) according to the median p21 expression value in the liver. *, P < 0.05.