The degree of liver injury determines the role of p21 in liver regeneration and hepatocarcinogenesis in mice

Authors


  • Potential conflict of interest: Nothing to report.

  • This work was supported by grants from the Wilhelm-Sander-Stiftung (to A.V.), the Deutsche Krebshilfe (a Max-Eder Grant to A.V.), and the Deutsche Forschungsgemeinschaft (TRR-77 [Liver Cancer—From Molecular Pathogenesis to Targeted Therapies] to A.V.) and the Cluster of Excellence REBIRTH II (to A.V.).

Address reprint requests to: Arndt Vogel, M.D., Department of Gastroenterology, Hepatology and Endocrinology, Medical School Hannover, Carl-Neuberg-Str. 1, D- 30625 Hannover, Germany. E-mail: vogel.arndt@mh-hannover.de; fax: +49-511-532-8392.

Abstract

Hepatocellular carcinoma (HCC) frequently arises in the context of chronic injury that promotes DNA damage and chromosomal aberrations. The cyclin-dependent kinase inhibitor p21 is an important transcriptional target of several tumor suppressors, which promotes cell cycle arrest in response to many stimuli. The aim of this study was to further delineate the role of p21 in the liver during moderate and severe injury and to specify its role in the initiation and progression of HCC. Deletion of p21 led to continuous hepatocyte proliferation in mice with severe injury allowing animal survival but also facilitated rapid tumor development, suggesting that control of compensatory proliferation by high levels of p21 is critical to the prevention of tumor development. Unexpectedly, however, liver regeneration and hepatocarcinogenesis was impaired in p21-deficient mice with moderate injury. Mechanistically, loss of p21 was compensated by activation of Sestrin2, which impaired mitogenic mammalian target of rapamycin (mTOR) signaling and activated cytoprotective Nrf2 signaling. Conclusion: The degree of liver injury and the strength of p21 activation determine its effects on liver regeneration and tumor development in the liver. Moreover, our data uncover a molecular link in the complex mTOR, Nrf2, and p53/p21-signaling network through activation of Sestrin2, which regulates hepatocyte proliferation and tumor development in mice with liver injury. (Hepatology 2013;53:1143–1152)

Abbreviations
CDK

cyclin-dependent kinase

ER

endoplasmic reticulum

ERK

extracellular signal-regulated kinase

FAA

fumarylacetoacetate

Fah

fumarylacetoacetate hydrolase

HCC

hepatocellular carcinoma

HT1

hereditary tyrosinemia type 1

mTOR

mammalian target of rapamycin

NTBC

2-(2-nitro-4-trifluoromethylbenzoyl)−1,3-cyclohexanedione

PCR

polymerase chain reaction

TUNEL

terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling

Hepatocellular carcinoma (HCC) is frequently associated with exposure to extrinsic factors that directly or indirectly induce DNA damage and chromosomal aberrations. Accumulation of DNA damage in hepatocytes ultimately leads to expanding foci of dysplastic hepatocytes, which progress to liver cancer if not rigorously controlled. ATM and ATR are serine/threonine kinases that sense DNA damage and coordinate DNA damage response pathways, most importantly p53. Activated p53 can inhibit proliferation to allow repair of DNA damage or trigger apoptosis if DNA damage is irreparable. p21 is one of the main effectors of p53 that induces cell cycle arrest and senescence in response to triggers such as DNA damage and telomere shortening by inhibiting the activity of cyclin-dependent kinase (CDK)–cyclin complexes and proliferating cell nuclear antigen.[1] Due to its ability to induce growth arrest and as one of the main targets of several tumor suppressors, p21 was also considered as a potential tumor suppressor. Furthermore, several genetic studies in mice confirmed the importance of p21 for the regulation of liver regeneration and its ability to delay tumor development in the liver.[2-5] The simple view on p21 as a tumor suppressor has been complicated, however, by findings that p21 can exhibit oncogenic activities in certain contexts. The first evidence for a protumorigenic role of p21 came from observations that p21 suppresses apoptosis of thymic lymphoma cells, thereby accelerating tumor growth.[6] More recent data suggest that p21 may also induce proliferation of cancer cells by promoting the assembly of type D cyclins with CDK4 and CDK6.[7]

The aim of this study was to further delineate the role of p21 in the liver during acute and chronic injury and to specify its role for the initiation and progression of HCC. Mice with a targeted genetic deletion of p21 were crossed into a mouse model of hereditary tyrosinemia type 1 (HT1). HT1 is an autosomal-recessive human disease caused by a genetic inactivation of the enzyme fumarylacetoacetate hydrolase (Fah), which carries out the last step in the tyrosine catabolism pathway. Inactivation of Fah leads to an accumulation of toxic metabolites, such as fumarylacetoacetate (FAA), which subsequently causes acute or chronic liver injury.[8] HT1 is characterized by an extremely high susceptibility to liver cancer. A murine model of Fah deficiency has been developed that represents all phenotypic and biochemical manifestations of the human disease on an accelerated time scale.[9-12]

Materials and Methods

Mice

C57Bl6-FahDexon5 and C57Bl6-Cdkn1atm1Ty1/J mice were crossed to generate Fah+//p21+/ breeders from which all Fah/ and Fah//p21/ animals used in these studies were derived. Drinking water was supplemented with 2-(2-nitro-4-trifluoromethylbenzoyl)−1,3-cyclohexanedione (NTBC) at a concentration of 7.5 mg/mL; 2.5% percent of this normal dose was used for low-dose NTBC treatment.

Survival Curves

Ten-week-old Fah-deficient mice were monitored for survival after NTBC was reduced (2.5%) or withdrawn (0%). Fah/ and Fah/p21/ double-knockout mice on 2.5% NTBC were followed for 400 days and Fah/p21/ 0% NTBC for 90 days.

Partial Hepatectomy

Briefly, mice were injected intraperitoneally with a ketamine (100 mg/kg body weight)/xylazine (4 mg/kg body weight) solution and subjected to a midline incision. The left and median lobes of the liver were ligated and resected. After closing the peritoneal and skin wounds, mice recovered from anesthesia on a warming pad. Thirty-eight hours or 1 week after PH, mice were sacrificed and livers were collected.

Immunoblotting

Frozen liver tissue was homogenized using an ultraturax (10 seconds) in cell lysis buffer [50 mM 4-(2-hydroxyethyl)−1-piperazine ethanesulfonic acid, 50 mM potassium chloride, 50 mM sodium fluoride, 5 mM tetrasodium diphosphate decahydrate, 1 mM ethylene diamine tetraacetic acid, 1 mM ethylene glycol tetraacetic acid, 5 mM β-glycerophosphate, 1 mM dithiothreitol, 1 mM vanadate, 1% (vol/vol) Nonidet P40] containing a Complete Protease Inhibitor mixture (Roche) and centrifuged for 10 minutes at 16,000g. Protein concentration was measured using the Bio-Rad Protein Assay Dye Reagent, and equal amounts of protein extracts were separated via sodium dodecyl sulfate–polyacrylamide gel electrophoresis and blotted to activated-polyvinylidene difluoride membrane (Bio-Rad).

Aminotransferase and Bilirubin Levels

Mouse blood was collected from the orbital sinus in lithium heparin tubes (LH1.3, Sarstedt, Germany) and processed according to the manufacturer's instructions. Aminotransferase and bilirubin levels were measured using an Olympus AU 400 system (Beckman Coulter, Switzerland).

RNA Isolation and Semiquantitative Reverse-Transcriptase Polymerase Chain Reaction

Total RNAs from liver tissue (n = 4) were extracted using the Qiagen RNAeasy kit. A Transcriptor High Fidelity cDNA Synthesis kit (Roche) was used to synthetize complementary DNA. Sequences of polymerase chain reaction (PCR) primers are provided upon request.

Statistical Analysis

Data are presented as the mean ± SD. Data were analyzed via analysis of variance followed by a Student t test to determine significance. P values were considered statistically significant at P < 0.05, P < 0.01, or P < 0.001, depending on the experiment.

Results

Loss of p21 Permits Survival of Fah-Deficient Mice With Severe Liver Injury

In order to determine the role of p21 in acute and chronic liver injury, Fah/ and Fah/p21/ mice in the C57BL/6 background were generated. The body weight of healthy double-knockout mice on 100% NTBC treatment was lower compared with Fah/ mice; however, the liver/body weight ratio was not significantly different, and there was no overt morphologic or biochemical phenotype. Next, NTBC treatment was completely stopped to induce severe liver injury. Following NTBC withdrawal, the mean survival of Fah/ mice was around 32 days (n = 20) until they eventually died from liver failure accompanied by progressive weight loss. In agreement with our previous observation with Fah/p21/ mice in the 129S background,[2] double-knockout mice survived the NTBC withdrawal for more than 4 months (n = 20; P < 0.0001) (Fig. 1A).

Figure 1.

Loss of p21 allows survival of Fah-deficient mice with severe liver damage but accelerates hepatocarcinogenesis. (A) Kaplan-Meyer survival curves of Fah/ (n = 20) and Fah/p21/ (n = 20) mice on 0% NTBC. (B-F) Eight-week-old Fah/ and Fah/p21/ mice were on 0% NTBC for 14 days or 2 months and compared with their 100% NTBC counterparts. (B) Representative liver tissue photographs for hematoxylin and eosin, TUNEL, Ki67, and p21 immunohistochemistry stainings. (C) GOT and bilirubin levels were increased significantly after NTBC withdrawal, especially in Fah/p21/ mice. (D) Representative photographs of livers and the indicated immunohistochemistry stainings from mice on 0% NTBC for 2 months. (E) Quantification of TUNEL- and Ki67-positive hepatocytes. Western blots of total liver lysates from pooled samples (n = 4) and hepatocyte size. (F) Tumor incidence and liver weight from Fah/ and Fah/p21/ mice on 0% NTBC. *P < 0.05; **P < 0.01; ***P < 0.001.

To further delineate the role of p21 in acute liver injury, mouse livers were collected 14 days after NTBC withdrawal. This time point was chosen because Fah/ mice on 0% NTBC still had the same weight and overall health as mice on 100% NTBC despite hepatic dysfunction.[10] As expected, histological examination revealed multiple small foci of necroinflammation in Fah/ mice on 0% NTBC (Fig. 1B). Furthermore, a few scattered terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling (TUNEL)-positive hepatocytes were detectable in these livers (Fig. 1B,E). A similar picture was evident in the surviving double knockout mice suggesting that loss of p21 did not significantly modulate acute FAA-induced liver injury in this early phase.

As expected, a strong activation of p21 and almost no Ki67 positive hepatocytes were seen in Fah/ mice on 0% NTBC despite a clear induction of cyclin D (Fig. 1B,E). Loss of p21 caused continuous hepatocyte proliferation in mice on 0% NTBC, thereby allowing survival of these mice in line with our previous observation (Fig. 1A,B,E).[2]

Loss of p21 Causes Rapid Tumor Development in Fah-Deficient Mice With Severe Liver Injury

To study the role of p21 signaling in severe FAA-induced liver injury at later time points, livers were analyzed 2 months after NTBC withdrawal. Histologic examination of the surviving mice revealed moderate to severe acinar inflammation and numerous ballooned and dysplastic hepatocytes (Fig. 1D). Biochemically liver injury measured by aminotransferase and bilirubin levels increased accordingly over time (Fig. 1C). Almost no TUNEL-positive hepatocytes were detectable in any mouse on 0% NTBC (Fig. 1D,E).

In the absence of p21, proliferation of hepatocytes with DNA damage further increased compared with the earlier time point (Ki67 labeling index of 47% at 2 months compared with 14% at 14 days; P = 0.005). In contrast, proliferation of hepatocytes was still markedly inhibited in the few surviving Fah/ mice, and almost no Ki67-labeled hepatocytes were detectable (n = 4 out of 41 mice) (Fig. 1D,E). Similar results were obtained with bromodeoxyuridine as a DNA synthesis marker and with phosphorylated histone H3 as a mitosis-specific marker (data not shown). In agreement with the proliferation assays, liver weight was significantly reduced in Fah/ mice compared with Fah/p21/ mice (P = 0.01) (Fig. 1F). Interestingly, however, the average hepatocyte cross-sectional area measured by β-catenin staining increased by 55% in Fah/ mice, suggesting a switch from proliferation-based liver regeneration to a regenerative process mediated by cell hypertrophy to at least partially compensate the strong p21-induced cell cycle arrest (Fig. 1E).

Due to the ongoing proliferation of hepatocytes with DNA damage, 85% of Fah/p21/ mice (n = 17) developed macroscopic detectable HCCs within 2-3 months. Interestingly, 25% of the few surviving Fah/ mice (one out of four) also developed liver tumors despite the profound cell cycle arrest induced by p21 (Fig. 1F). Overall, however, tumor incidence was significantly higher in double-knockout mice (P = 0.006).

p21 Is Required for Proliferation of Hepatocytes in Fah-Deficient Mice With Moderate Chronic Liver Injury

To analyze the role of p21 in chronic liver injury and its potential involvement in cancer formation under moderate hepatocellular damage, mice were exposed to a reduced treatment regimen of NTBC (2.5%) for up to 12 months. This suboptimal treatment closely mimics human liver disease leading to HCC formation in HT1 patients.[10, 13] Fah/ and Fah/p21/ mice survived the low-dose NTBC treatment (Fig. 2A).

Figure 2.

p21 is required for proliferation of hepatocytes with DNA damage. (A) Kaplan-Meyer survival curves of Fah/ and Fah/p21/ mice on 2.5% NTBC treatment. (B-F) Fah/ and Fah/p21/ mice were given 2.5% NTBC for 3 months and compared with their 100% NTBC counterparts. (B) Representative pictures of the indicated immunohistostainings. (C) Blood serum glutamic oxalic transaminase (GOT) and bilirubin levels. (D) Quantification of TUNEL- and Ki67-positive hepatocytes. (E) Western blots of total liver lysates from pooled samples (n = 4). (F) Hepatocyte size. *P < 0.05; ***P < 0.001.

Three months following NTBC reduction, histological examination revealed only mild acinar inflammation (Fig. 2B). Aminotransferase and bilirubin levels were accordingly not significantly increased in both groups (Fig. 2C). In contrast to Fah-deficient mice on 0% NTBC, multiple proliferating hepatocytes were found in livers of Fah/ mice on 2.5% NTBC treatment. In agreement with the Ki67 staining, cyclin D levels were elevated, and p21 was only slightly induced (Fig. 2B,D,E). TUNEL staining did not reveal any apoptotic hepatocytes (Fig. 1B,D). Surprisingly, the number of Ki67-labeled hepatocytes was significantly reduced in livers of Fah/p21/ mice under 2.5% NTBC treatment compared with Fah/ mice (Fig. 2B,D). Similar results were obtained with bromodeoxyuridine as a DNA synthesis marker and with phosphorylated histone H3 as a mitosis-specific marker (data not shown). Thus, proliferation-based liver regeneration was unexpectedly impaired in p21-deficient livers, suggesting that loss of p21 may actually impair hepatocyte proliferation during chronic liver injury. Similar to mice on 0% NTBC, the hepatocyte cross-sectional area measured by β-catenin staining increased in Fah/ mice (P = 0.05).

Loss of p21 Impairs Tumor Development in Fah-Deficient Mice With Moderate Liver Injury

To examine tumor onset and progression in Fah/ and Fah/p21/ mice under moderate chronic liver injury, livers of Fah-deficient mice were examined after 6, 9, and 12 months on 2.5% NTBC treatment. At 6 months, liver tumors were evident on macroscopic and histological examination in 50% of Fah/ mice (n = 10); tumor incidence increased over time, reaching 76% after 9 months (n = 20) and 100% after 12 months (n = 20) (Fig. 3A,B). Surprisingly, loss of p21 significantly delayed tumor development; no tumors were detectable after 6 (n = 15) and 9 months (n = 15), and only 50% of Fah/p21/ mice developed liver tumors after 12 months on 2.5% NTBC treatment (n = 10; P = 1.7E-2). Furthermore, Fah/ livers displayed a significantly greater number and size of tumors than Fah/p21/ livers (Fig. 3C,D). In contrast to the findings described here, Fah/p21/ mice in the 129S background still displayed a higher tumor incidence on 5% NTBC.[2] The background-specific differences are most likely due to a higher sensitivity of Fah/ mice in the 129S background to the NTBC reduction compared to mice in the C57Bl6 background. Additionally, we cannot rule out that the higher tumor incidence in the 129S background might also be related to a generally higher tumor susceptibility of these mice, epigenetic adaptations, which might occur in the back-crossed mice and/or cleanliness of the mouse facilities, which has been shown to significantly modulate hepatocarcinogenesis.[14]

Figure 3.

Loss of p21 delays tumor formation in Fah-deficient mice with moderate liver damage. Fah/ and Fah/p21/ mice were treated with either 100% or 2.5% NTBC for 6, 9, and 12 months. (A) Representative photographs of livers at indicated time points. Surprisingly, loss of p21 delayed tumor development in Fah-deficient livers. (B) Graphs representing tumor incidence and liver weight of Fah-deficient mice at different time points. Interestingly, only 50% of Fah/p21/ mice exhibited tumors at 12 months. (C,D) Scatter plots displaying the size of tumors (C) and tumor numbers (D) in Fah-deficient livers at the indicated time points. *P < 0.05; **P < 0.01; ***P < 0.001.

Taken together, these data indicate that loss of p21 dramatically accelerates tumor development in Fah/ mice with severe liver injury, but surprisingly delays tumor development in mice with moderate liver injury.

Differential Regulation of Cell Cycle–Related Genes in Fah−/− and Fah/p21−/− Mice

FAA is a highly electrophilic compound that induces DNA damage, mitotic abnormalities, chromosomal instability, and endoplasmic reticulum (ER) stress in vitro and in vivo.[15, 16] To better understand how loss of p21 modulates the cellular stress response in Fah-deficient mice, microarray analysis was performed with mice on 0% and 2.5% NTBC before visible tumor nodule development and compared with their respective controls on 100% NTBC. First, transcriptional profiles from tumor-prone mice (Fah−/− mice on 2.5% NTBC and Fah/p21−/− mice on 0% NTBC) and from Fah−/− mice were compared with profiles from healthy mice (Fah−/− and Fah/p21−/− mice on 100% NTBC) and Fah/p21−/− mice on 2.5% NTBC. KEEG Pathway analysis identified 334 genes that were regulated significantly. The most significant category modified in tumor-prone mice was related to cell cycle (P = 9.55E-5), followed by DNA repair (P = 1.1E-3) (Fig. 4A). Interestingly, direct comparison of gene expression from Fah−/− and Fah/p21−/− mice revealed a similar profile in tumor-prone Fah−/− mice on 2.5% NTBC, Fah−/− tumors, and Fah/p21−/− mice on 0% NTBC mice. In contrast, the expression profiles of Fah/p21−/− mice with moderate liver injury (2.5% NTBC), in which liver regeneration was impaired and tumor development delayed, clustered with expression profiles from healthy mice (Fig. 4A). Together, the pathway analysis identified cell cycle–related genes as modified by p21 and as most significantly associated with tumor development.

Figure 4.

The role of p21 for liver regeneration after PH depends on overall liver injury. (A) Hierarchical clustering of mouse cohorts, where cohorts with similar expression profiles were grouped together. Each row represents a value of four pooled RNA samples of each cohort. The top heat map corresponds to cell cycle–related genes identified by subtracting disease-associated genes from tumor-associated genes. The bottom heat map corresponds to expression profiles related to the Gene Ontology term “DNA repair” extracted from cancer-related genes (see Supporting Information). (B-D) p21+/+ and p21/ mice underwent 70% PH. Livers were collected 38 hours and 1 week after PH. (B) Ki67 immunostaining and quantification of positive hepatocytes. (C) Liver/body weight ratio of p21+/+ and p21/ mice 1 week after PH. (D) Western blots of total liver lysates from pooled samples. (E,F) Fah/ and Fah/p21/ mice were given 2.5% NTBC for 3 months and underwent 70% PH. Livers were collected 38 hours after PH. (E) Ki67 immunostaining and quantification of positive hepatocytes. (F) Western blots of total liver lysates from pooled samples. *P < 0.05.

The Role of p21 for Hepatocyte Proliferation After Partial Hepatectomy Depends on Preexisting Liver Injury

The above data strongly suggest that p21 modulates liver regeneration and hepatocarcinogenesis differently in mice with moderate and severe liver injury. To further analyze the role of p21 for hepatocyte proliferation, partial hepatectomy (PH) was performed.

First, the role of p21 was analyzed in p21+/+ and p21/ mice. Multiple Ki67-positive cells were clearly visible in p21+/+ and p21/ mice 38 hours after PH, and there was no significant difference between both groups (Fig. 4B). Liver mass recovery monitored by body/liver weight ratio was slightly accelerated in p21/ mice 1 week after PH (Fig. 4C). At this time point, almost no Ki67-positive cells were detectable in either group. Overall, there were only minor differences between knockout and wild-type hepatocytes, suggesting that p21 does not play a major role for the initiation and termination of hepatocyte proliferation in healthy mice. Next, partial hepatectomies were performed with Fah/ and Fah/p21/ mice with preexisting liver injury. We have shown that Fah/ mice on 0% NTBC do not survive PH due to the complete p21-mediated block of hepatocyte proliferation.[2] Here, Fah-deficient mice on 2.5% NTBC for 3 months with moderate liver injury were used. Surprisingly, hepatocyte proliferation following PH was markedly inhibited in Fah/ mice in which basal liver regeneration before PH was not impaired (Fig. 4E). Importantly, the profound cell cycle arrest was associated with a strong induction of p21 (Fig. 4F). In contrast to Fah/ mice, multiple Ki67-positive cells were clearly visible in Fah/p21/ mice on 2.5% NTBC 38 hours after PH (Fig. 4E). Together, these data indicate that p21 has no lasting effect on liver regeneration in healthy mice after PH. In contrast, PH in mice with preexisting liver injury leads to a strong induction of p21, which subsequently impairs liver regeneration.

Interaction of the Mammalian Target of Rapamycin and p21 Pathways Facilitates Tumor Development in the Liver

Several molecular pathways, in particular mitogen-activated protein kinase and mammalian target of rapamycin (mTOR), have been implicated in hepatocarcinogenesis in previous clinical and experimental studies.[3, 17, 18] Interestingly, most of these pathways are also important for liver regeneration, suggesting that they are likely candidates contributing to the cell cycle gene expression profile in tumor-prone Fah-deficient mice. To determine the role of these pathways in Fah-deficient mice, activation of JNK/c-jun, extracellular signal-regulated kinase (ERK), p38, and mTOR was analyzed 14 days after NTBC withdrawal and after 3 months on 2.5% NTBC. Activation of the JNK/c-jun, ERK, and p38 stress kinases did not correlate with the phenotype of Fah-deficient mice (Fig. 5A). A strong activation of the mTOR pathway, as monitored by immunoblot analysis of phosphorylated S6, was evident in Fah/ and Fah/p21/ mice on 0% NTBC. Similarly, a moderate phosphorylation/activation of S6 was seen in Fah/ mice with moderate liver injury (2.5% NTBC). Interestingly, however, S6 phosphorylation was significantly reduced by 50% in Fah/p21/ mice on 2.5% NTBC, in which hepatocyte proliferation was reduced (n = 6) (P < 0.05) (Fig. 5A,B). Moreover, reduction of NTBC induced an increase of 4-EPB1 protein levels in Fah-deficient mice, albeit without significantly changing the ratio between the phosphorylated and nonphosphorylated protein.

Figure 5.

Cross-talk between mTOR and p21 signaling pathways during tumor formation in the liver. (A-E) Fah/ and Fah/p21/ mice were given 0% NTBC for 14 days (A) or 2.5% NTBC for 3 months (A-E) and compared with their 100% NTBC counterparts (n = 4). (B) Quantification of S6 phosphorylation. (C) Expression of Sestrin2 at the RNA and protein level. (D) RNA levels from microarray analysis and semiquantitative reverse-transcriptase PCR (E) of different Nrf2 target genes. F, Fah/; Fp21, Fah/p21/. (F) Schematic representation of how hepatocyte proliferation and tumor formation are reduced by inhibition of mTOR and activation of Nfr2 pathways via Sestrin2 activation in p21-deficient livers with moderate injury. *P < 0.05.

Very recently, it has been shown that genotoxic and ER stress can inhibit mTOR activity in the liver through induction of Sestrin2.[19, 20] Here, a significantly stronger induction of Sestrin2 was evident in Fah/p21/ mice 3 months after NTBC reduction (increase of 50%) (Fig. 5C), suggesting that loss of p21 leads to a compensatory activation of Sestrin2, which subsequently inhibits mTOR activity. Moreover, Sestrin2 has been shown to activate Nrf2 signaling in mouse livers by promoting p62-dependent autophagic degradation of Keap1.[20] Accordingly, microarray and reverse-transcriptase PCR analysis revealed a significant stronger activation of several known downstream targets genes of Nrf2 including HO-1, Nqo1, and GSTm4 in livers of Fah/p21/ mice compared with Fah/ mice (Fig. 5D,E).

Discussion

Liver injury is often accompanied by severe DNA damage of hepatocytes, which leads to an activation of DNA repair pathways, including p53 and p21. Subsequent development of preneoplastic lesions and their progression to HCC reflects the convergence of genetic and epigenetic defects that provoke dysregulation of pathways controlling cell cycle progression. Several previous studies have shown that p21 regulates liver regeneration and hepatocarcinogenesis. JNK1-dependent down-regulation of p21, for example, is required for proliferation of hepatocytes and tumor progression in chemically induced carcinogenesis.[3] Similarly, we confirmed our findings in Fah-deficient mice that loss of p21 permits proliferation of hepatocytes with severe DNA damage, which rapidly progresses to dysplastic hepatocytes and HCC.[2] These studies established p21 as a negative regulator of hepatocyte proliferation and as a tumor suppressor. Paradoxically, however, we report here that hepatocyte proliferation was significantly reduced and, more importantly, tumor development was profoundly delayed in p21-deficient mice with moderate liver injury, providing further insight into the complex regulation of cellular processes required for liver regeneration and tumor development. The late spontaneous tumor onset in p21-deficient mice and the rarity of p21 loss of function mutations in cancer already provided some evidence that p21 is not a classical tumor suppressor. Here, we provide evidence that loss of p21 may actually promote or delay tumor development in the same disease and the same organ depending on the degree of preexisting injury.

Previous studies and our own observation suggest that the ability of p21 to modulate liver tumor development is closely linked to its ability to control cell cycle progression of hepatocytes. Interestingly, however, the role of p21 for liver regeneration appears to depend on the degree of liver injury and the strength of subsequent induction of p21. In healthy mice, p21 is not required for normal liver development and does not significantly affect initiation and termination of liver regeneration following PH. On the other hand, abundantly expressed transgenic p21 dramatically reduced hepatocyte cell cycle progression in an otherwise healthy and normal environment. Moreover, this function even overrides the powerful mitogenic signals induced by 70% PH.[21] Similarly, high levels of p21 in wild-type mice following extended PH or in Fah-deficient mice on 0% NTBC following 70% PH almost completely inhibit liver regeneration, resulting in a dramatically increased mortality.[2, 4] Here, we provide evidence that 70% PH induces to a strong and robust induction of p21 in mice with preexisting liver injury, subsequently impairing liver regeneration. Together, these data indicate that the degree of overall (acute and chronic) liver injury determines the strength of p21 induction in the liver and, subsequently, its effect on hepatocyte proliferation.

Interestingly, gene set enrichment analysis revealed that proliferation-related genes were most significantly, differently regulated between tumor-prone Fah-deficient mice and Fah/p21/ mice on 2.5% NTBC, suggesting that other mitogens might be affected by loss of p21. The factors that drive proliferation of hepatocytes and hepatocarcinogenesis in chronic liver injury are not completely understood. The mTOR pathway is increasingly recognized to regulate growth and proliferation of hepatocytes and tumor cells.[11, 22-24] In contrast to 4E-BP1, which appears to play only a minor role in mediating the effects of mTOR on mitogen-stimulated hepatocyte proliferation,[23] pharmacological and genetic studies revealed that, specifically, S6k1 promotes hepatocyte proliferation by regulating cyclin D1 promoter activity and messenger RNA levels in hepatocytes. Moreover, the biological importance of S6 ribosomal-mediated translation has been shown in adult mouse livers that have a conditionally deleted S6 gene and which fail to proliferate due to a block in cyclin E messenger RNA expression. Here, we observed a striking correlation between mTOR activation/S6 phosphorylation and hepatocyte proliferation/tumor development. Importantly, we have shown that activation of the mTOR pathway is required for proliferation of hepatocytes during FAA-induced liver injury. Moreover, pharmacological inhibition of mTOR signaling and specifically S6 phosphorylation impaired cell cycle progression of Fah/ hepatocytes following NTBC withdrawal and markedly suppressed liver regeneration and tumor development in Fah/p21/ mice.[11] mTOR activity can be inhibited by multiple mechanisms, including nutrient limitations and DNA damage. Very recently, Sestrin2 has been identified to suppress mTOR activity in the liver following genotoxic and ER stress.[19, 20] Here, the strong compensatory induction of Sestrin2 significantly inhibited mTOR activity, thereby impairing baseline liver regeneration in Fah/p21/ mice with moderate liver injury. Moreover, Sestrin2 has also been to shown to activate Nrf2 in the liver.[20] Accordingly, a stronger activation of Nrf2 target genes was evident in the livers of Fah/p21/ mice. Nrf2 is a transcription factor, which regulates a battery of antioxidants and other cytoprotective genes in many tissues.[25] Importantly, we have shown a high mortality and accelerated tumor development in mice with a targeted deletion of Nrf2 in Fah-deficient mice.[12] Thus, our data suggest that the compensatory induction of Sestrin2 does not only inhibit mTOR-mediated hepatocyte proliferation, it also enhances the Nrf2-regulated oxidative stress response, thereby protecting mice against subsequent injury and tumor development.

In conclusion, we provide evidence that the degree of liver injury and the strength of p21 activation determine its effects on hepatocyte proliferation and hepatocarcinogenesis. Moreover, our data uncover a molecular link in the complex mTOR, Nrf2, and p53/p21-signaling network through activation of Sestrin2, which can compensate for the loss of p21 in the liver during chronic injury.

Ancillary