p300 Regulates Liver Functions by Controlling p53 and C/EBP Family Proteins through Multiple Signaling Pathways

Abstract

The histone acetyltransferase p300 has been implicated in the regulation of liver biology; however, molecular mechanisms of this regulation are not known. In this paper, we examined these mechanisms using transgenic mice expressing a dominant negative p300 molecule (dnp300). While dnp300 mice did not show abnormal growth within 1 year, these mice have many alterations in liver biology and liver functions. We found that the inhibition of p300 leads to the accumulation of heterochromatin foci in the liver of 2-month-old mice. Transcriptome sequencing (RNA-Seq) analysis showed that this inhibition of p300 also causes alterations of gene expression in many signaling pathways, including chromatin remodeling, apoptosis, DNA damage, translation, and activation of the cell cycle. Livers of dnp300 mice have a high rate of proliferation and a much higher rate of proliferation after partial hepatectomy. We found that livers of dnp300 mice are resistant to CCl₄-mediated injury and have reduced apoptosis but have increased proliferation after injury. Underlying mechanisms of resistance to liver injury and increased proliferation in dnp300 mice include ubiquitin-proteasome-mediated degradation of C/EBPα and translational repression of the p53 protein by the CUGBP1-eukaryotic initiation factor 2 (eIF2) repressor complex. Our data demonstrate that p300 regulates a number of critical signaling pathways that control liver functions.

INTRODUCTION

Liver is a one of the largest tissues that has the ability to regenerate itself upon stimulation and performs a variety of complex functions. Hepatocellular carcinoma (HCC) is one of the leading causes of death, and surgical resection is the main approach to eliminate tumor sections (1). Liver proliferation after surgery (partial hepatectomy [PH] in mouse models) is impaired in aging mice (2). Our laboratory and other groups have shown that key genes in liver function include CCAAT enhancer-binding protein (C/EBP) family, retinoblastoma (Rb) family, histone deacetylase 1 (HDAC1), p300 (3–5), and RNA CUG-binding protein 1 (CUGBP1) (4) genes. C/EBPα is involved in many aspects of liver function, and it inhibits liver proliferation by direct interactions with cell cycle proteins (2, 6, 7). C/EBPα is expressed in the liver as two isoforms with molecular masses of 42 kDa and 30 kDa. The growth-inhibitory activity of C/EBPα in liver of young animals is mediated through direct interactions with cdk2, repression of E2F-dependent transcription, and interactions with chromatin remodeling proteins (2, 6). Aging liver hyperphosphorylates C/EBPα at S193, which results in inhibition of liver proliferation by promoting the formation of HDAC1 and C/EBPα complexes (4, 8). Expression of constitutively active mutant C/EBPα-S193D in mice strongly inhibits liver proliferation, while mutation of S193 to Ala leads to increased liver proliferation after partial hepatectomy and a failure to stop liver regeneration (5–7). Examination of C/EBPα complexes with chromatin remodeling proteins revealed that C/EBPα-S193D knock-in (KI) mice have increased levels of complexes with p300 and HDAC1, while S193A mice have a reduction in the level of these complexes (5). Consistent with the role of C/EBPα-p300 complexes in liver biology, C/EBPα-S193D mice exhibit altered chromatin structures and age-associated dysfunctions in the liver (6, 7), one of which is the development of hepatic steatosis (9). Another C/EBP family member, C/EBPβ, also regulates liver proliferation, with effects being dependent on the levels of C/EBPβ isoforms. A single C/EBPβ mRNA produces three isoforms, full-length protein (C/EBPβ-FL), liver-activating protein (C/EBPβ-LAP), and liver-inhibitory protein (C/EBPβ-LIP), through alternative translation from three AUG codons (10). These three isoforms possess differential activities; thus, a balance is important for proper regulation of cell functions (11). C/EBPβ-LIP is a truncated molecule that contains a DNA-binding domain but lacks activation domains. Since C/EBPβ-LIP binds to the same regions of DNA as C/EBPβ-FL and because it heterodimerizes with C/EBP family proteins, it works as a dominant negative molecule. It has been shown that overexpression of C/EBPβ-LIP in livers leads to stronger expression of cell cycle genes encoding PCNA and cyclins A and E (12). In addition to this activity, C/EBPβ-LIP directly interacts with Rb and disrupts E2F-Rb complexes, leading to derepression of E2F-dependent promoters and to proliferation (13).

The role of p300 in liver biology has not been fully elucidated. High-level expression of p300 is associated with poor prognoses and epithelial-to-mesenchymal transition in HCC (14, 15). p300 forms complexes with C/EBP proteins and activates promoters of genes involved in triglyceride synthesis during the development of hepatic steatosis (9). Another study showed that inhibition of histone acetyltransferase (HAT) activity of p300 decreased the incidence of hepatic steatosis (16). Mice lacking the CH1 domain of p300 are resistant to high-fat-diet-mediated elevations of triglyceride levels, insulin resistance, and glucose intolerance (17). Recent studies of animal models with hepatocyte-specific deletion of p300 and with p300 (G422S) knock-in mice have shown that p300 plays an important role in glycogen synthesis through maintaining basal gluconeogenesis (18, 19). In addition, p300 has been implicated in the regulation of several transcription factors, such as Foxo1 and farnesoid X receptor (FXR), which are highly expressed in the liver (20, 21). Interestingly, alterations of the p300-mediated regulation of FXR are involved in the development of metabolic disorders (22). Despite these important observations, very little is known about the mechanisms by which p300 regulates liver functions.

In this paper, we examined the role of p300 in the regulation of liver biology using transgenic mice that express dominant negative p300 (dnp300). Transcriptome sequencing (RNA-Seq) analysis revealed multiple gene expression alterations in several pathways in livers with inhibited activity of p300. The main consequences of these alterations are increased proliferation of the liver and an impaired response to liver injury. Among multiple changes, translational repression of p53 appears to be a key event in the impaired response to liver injury.

MATERIALS AND METHODS

Mice, partial hepatectomy, and CCl₄-induced liver injury.

The generation of transgenic mice expressing dominant negative CH3-p300 was described in our previous report (9). Multiple lines of dnp300 mice were examined in this study. Experiments in this paper were performed with 2- to 3-month-old mice. For partial hepatectomy, surgery was performed as described in our previous reports (6, 9). Acute treatments of mice with CCl₄ were performed as described in our previous study (23). Data represent summaries of data from work with 4 or 5 mice per time point. Experiments with animals have been approved by the Institutional Animal Care and Use Committee at Baylor College of Medicine (protocol AN-2503).

Liver histology and immunohistochemistry.

Livers were fixed overnight in buffered 10% formalin, embedded in paraffin, and sectioned. Bromodeoxyuridine (BrdU) was injected intraperitoneally 4 h before the mice were sacrificed. BrdU staining was performed by using a BrdU uptake assay kit (Invitrogen). Immunohistochemistry was performed by using the HistoMouse-Plus kit (Invitrogen) for H3K9TriMe and Ki67. Apoptosis was measured by using a terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) apoptosis detection kit (GenScript).

Antibodies.

Antibodies (Abs) used were specific for C/EBPα (14AA), C/EBPβ (C-19), p300 (C-20), the c-myc tag (9E10), CREB-binding protein (CBP) (C-20), cyclin D1 (H-295), cyclin E (M-20), PCNA (FL-261), Cdc2 (C-19), Chk2 (H-300), CUGBP1 (3B1), the α subunit of eukaryotic initiation factor 2 (eIF2α) phosphorylated at Ser52 (ph-S51-eIF2α), eIF2α (FL-315), heterochromatin protein 1α (HP1α), and ubiquitin (Ub) (Santa Cruz Biotechnology); p53 (clone DO1; Active Motif); and Ki67 (ab16667) and histone H3 trimethylated at K9 (H3K9TriMe) (ab8898) (Abcam). Monoclonal anti-β-actin antibody was obtained from Sigma.

Protein isolation and Western blotting.

Protein extracts were isolated from livers as described previously (6, 7). Proteins (50 to 100 μg) were loaded onto gradient (4 to 20%) gels (Bio-Rad), and membranes were probed with the corresponding antibodies.

Biochemical assays.

Liver panel analysis using collected serum was performed by facilities at Baylor College of Medicine. Blood was taken from fasted mice for analysis.

RNA-Seq and analysis.

Total RNA was prepared from mouse livers by using an RNeasy kit (Qiagen) according to the manufacturer's instructions. Library preparation and sequencing on the Illumina HiSeq2000 instrument were performed by Genewiz, Inc. Target genes were verified by synthesizing cDNA using a qScript synthesis kit (Quanta) according to the manufacturer's instruction. The real-time PCR mixtures contained 2× QuantiTect SYBR green PCR master mix (Qiagen), 10× QuantiTect primer assay mix, and synthesized cDNA. Amplification was performed in triplicate for each sample on a Realplex Mastercycler (Eppendorf). Relative expression levels were normalized to β-actin levels. For RNA-Seq data analysis, we trimmed the low-quality 11 nucleotides from the 5′ ends of the reads. The resulting 40-nucleotide single-ended reads were mapped to the mouse genome (UCSC mm10) by using Tophat (24) with NCBI RefSeq genes as the reference and up to two possible mismatches. The read duplicates were removed by using samtools (25) in order to reduce possible biases caused by reverse transcription-PCR (RT-PCR). Cufflinks (26) was used to estimate gene expression levels (transcript counts). DESeq 2 (27) was used to test if genes were differentially expressed between the wild-type (WT) and p300 groups. The Benjamini-Hochberg method, which controls the false discovery rate (FDR), was used to correct for multiple comparisons. Alterations of genes shown in the maps are all significant (FDR < 0.1). These genes were clustered into functional groups by using DAVID bioinformatics tools (NIH). Differential expressions of genes selected from each category were verified by RT-quantitative PCR (QRT-PCR). The list of primers is given in Table S1 in the supplemental material.

Statistical analysis.

All values are presented as means ± standard deviations (SD). Statistical analyses were performed by using the Student t test. Statistical significance was assumed when the P value was <0.05.

RESULTS

Characterization of dnp300 mice.

The histone acetylase p300 displays its activity through several domains. One of these domains, CH3, interacts with many transcription factors and is a key portion of the molecule. We have recently generated transgenic mice that express only the CH3 domain of p300 (9), which are called dnp300 mice. These mice were previously used for breeding with C/EBPα-S193D mice. It was shown previously that dnp300 mice and double-dnp300-transgenic C/EBPα-S193D knock-in mice have inhibited hepatic steatosis (9). In this paper, we present a detailed characterization of these mice and an examination of the effects of inhibition of p300 on liver biology. To test if the truncated p300 protein works as a dominant negative molecule, we examined the effects of the dnp300 molecule on the activity of WT p300 toward the transcription factor E2F1. A luciferase reporter construct linked to an E2F1-dependent cyclin E promoter was cotransfected with E2F1 or with E2F1 and WT p300 plus increasing amounts of dnp300 into HEK293 cells. Figure 1A shows that E2F1 activates the cyclin E promoter and that WT p300 increases this E2F1-dependent activation. However, dnp300 dramatically reduces the activation of the promoters by E2F1 and p300. This dnp300 construct was used for the generation of transgenic mice with the genotypes shown in Fig. 1B. RT-PCR with primers recognizing dnp300 mRNA showed that dnp300 is expressed in heart, liver, kidney, muscle, testis, and lung (Fig. 1C). Examination of the expression of endogenous p300 in the tissues showed that the dnp300 molecule did not change the levels of WT p300 mRNA in brain, spleen, heart, and kidney but reduced the levels of p300 in liver and muscle (Fig. 1D). Our previous studies suggested a critical role of p300 in the liver (9); therefore, we focused further studies on a detailed reexamination of liver biology. Expression of c-myc-tagged dnp300 in the liver was confirmed by Western blotting with Abs to c-myc (Fig. 1E). Although the levels of endogenous p300 protein were slightly reduced in livers of dnp300 mice, this reduction was not statistically significant. Examination of a homolog of p300, the lysine acetyltransferase CREB-binding protein (CBP), showed a 2-fold elevation of CBP levels in livers of dnp300 mice (Fig. 1E). A slower-migrating CBP isoform was also detected, suggesting possible posttranslational modifications of CBP. Thus, these results show that livers of dnp300 mice express high levels of dnp300. More details regarding the generation of these mice and examination of hepatic steatosis can be found in our recent report (9). To obtain preliminary information for liver functions and for metabolic changes in dnp300 mice, we examined blood parameters that are provided by liver and other tissues. Examination of 8 fasted animals of each genotype revealed multiple alterations of these parameters in serum. While the levels of alanine aminotransferase (ALT)/aspartate transaminase (AST), albumin, glucose, and cholesterol were not changed, we determined significant changes in the levels of triglycerides (reduction), direct bilirubin (induction), and indirect bilirubin (reduction) (Fig. 1F). Since dnp300 mice have reduced levels of C/EBPα-p300 complexes, which upregulate enzymes of triglyceride (TG) synthesis (9), we examined the levels of these enzymes and found significant reductions in the levels of the DGAT1 and DGAT2 proteins in livers of dnp300 mice (Fig. 1G). This finding suggests that the reduction in the level of TGs in the serum is the result of the reductions of DGAT1 and DGAT2 levels. Since p300 plays an important role in gluconeogenesis (18, 19), we examined the expression of enzymes of glucose metabolism but did not detect alterations in their expression (data not shown). As shown in Fig. 1F, levels of creatinine (CRE) and gamma glutamyltransferase (GGT) were reduced in the serum of dnp300 mice. Serum levels of CRE are regulated by skeletal muscle and kidney; therefore, the dramatic reduction of CRE levels in serum of dnp300 mice demonstrated that the inhibition of p300 changes the functions of liver and other tissues, leading to an alteration of metabolism.

FIG 1.

Characterization of dnp300 mice. (A) dnp300 inhibits p300-dependent activation of the cyclin E promoter. The cyclin E-luciferase construct was cotransfected in HEK293 cells with E2F1 and p300 and with increasing amounts of the dnp300 construct. Luciferase activity was examined 16 h after transfections. (B) Typical picture of genotyping of dnp300 TR mice. TR, transgenic; W, wild type. (C) Expression of dnp300 transcripts in different tissues (shown at the top) was determined by QRT-PCR. H20, control without DNA; Pl, positive control using plasmid DNA. (D) Expression levels of endogenous WT p300 mRNA in tissues were examined by using QRT-PCR and normalized to values for the β-actin control. (E) Western blotting of the c-myc tag, endogenous p300, and the p300 homolog CBP. Bar graphs show the levels of the proteins as ratios to β-actin levels. (F) Blood parameters for dnp300 mice. Alterations of blood parameters are shown in red. BUN, blood urea nitrogen. (G) Levels of DGAT1 and DGAT2 proteins are reduced in livers of dnp300 mice. Western blotting was performed with Abs to DGAT1 and DGAT2. The filter was reprobed with β-actin. Bar graphs show levels of proteins as ratios to β-actin levels.

Livers of 2-month-old dnp300 mice have an increased abundance of heterochromatin structures and have increased proliferation.

Since the main activity of p300 is associated with chromatin remodeling, we examined the structure of the chromatin in livers of dnp300 mice by 4′,6-diamidino-2-phenylindole (DAPI) staining and by costaining with a marker of heterochromatin, histone H3 trimethylated at K9 (H3K9TriMe). DAPI staining showed that nuclei of hepatocytes of dnp300 mice have abundant focus-like structures (Fig. 2A), compared to age-matched WT mice. Costaining of the same slides with antibodies to H3K9Me revealed that these focus-like structures contain marker of heterochromatin H3K9TriMe (Fig. 2A, bottom). We calculated the number of H3K9Me-positive hepatocytes and found that the abundances of heterochromatin structures are significantly increased in livers of 2-month-old dnp300 mice (Fig. 2A, bar graphs). In these studies, we also noted that hepatocytes of dnp300 mice were larger than hepatocytes of age-matched WT mice. To obtain additional proof of an increased abundance of heterochromatin structures in livers of dnp300 mice, we examined protein levels of another marker of heterochromatin, heterochromatin protein 1α (HP1α). Western blotting showed a 3- to 4-fold elevation of HP1α levels in livers of dnp300 mice (Fig. 2A, bottom right image). We next examined the morphology of the livers of 2-month-old dnp300 mice by staining with hematoxylin and eosin (H&E). In addition to the changes in the sizes of hepatocytes, we detected a significant number of mitotic figures in livers of dnp300 mice (Fig. 2B). In age-matched WT mice, no mitotic figures were observed. These data suggested that livers of adult dnp300 mice proliferate. To further test this suggestion, we examined BrdU uptake and found that livers of dnp300 mice had increased DNA replication. Up to 3% of hepatocytes incorporated BrdU in dnp300 mice (Fig. 2C). In WT mice, the percentage of BrdU-positive hepatocytes was 0.1% or lower. In agreement with these findings, examination of cell cycle proteins revealed that levels of cyclin D1 and cyclin E were increased in livers of 2-month-old dnp300 mice (Fig. 2D). We have previously shown that dnp300 disrupts complexes of WT p300 with C/EBP family proteins (9). Therefore, we examined the expressions of these proteins in livers of WT and dnp300 mice. C/EBPα is expressed in the liver as 42-kDa and 30-kDa isoforms. The 42-kDa isoform possesses growth-inhibitory activity (2, 3). We found that the level of this isoform of C/EBPα is significantly reduced in livers of dnp300 mice (Fig. 2E). Examination of another member of the C/EBP family, C/EBPβ, showed that the levels of both the C/EBPβ-LAP and C/EBPβ-LIP isoforms are significantly increased in livers of dnp300 mice. Since C/EBPβ-LIP promotes liver proliferation (13), the elevation of C/EBPβ-LIP levels in dnp300 livers might contribute to the increased liver proliferation. The changes of C/EBPα and C/EBPβ occur at the protein level, while levels of the corresponding mRNAs are not changed (data not shown). Taken together, these results show that livers of 2-month-old dnp300 mice proliferate and have a significant portion of mitotic and replicating hepatocytes, which is associated with high expression levels of cell cycle proteins and C/EBPβ isoforms and with reduced expression levels of the 42-kDa isoform of C/EBPα.

FIG 2.

Livers of dnp300 mice have alterations of chromatin structure and increased proliferation. (A, top) Immunofluorescence staining with anti-H3K9TriMe antibody and DAPI in livers of 2-month-old WT and dnp300 mice. (Bottom) Immunohistochemistry of livers with anti-H3K9TriMe antibody. The bar graph represents the percentage of hepatocytes with multiple positive foci. Data represent means ± SD (n = 3 to 5). *, P < 0.05. The right bottom section shows an examination of HP1α by Western blotting. Bar graphs show levels of HP1α as ratios to β-actin levels. (B) H&E staining of livers of 2-month-old WT and dnp300 mice. The bar graph represents the percentage of mitotic figures, which are denoted by arrows. Data represent means ± SD (n = 3). *, P < 0.05. (C) Examination of DNA replication by BrdU uptake and quantitation of the percentage of BrdU-positive hepatocytes. Bar graphs show a summary of data from three independent experiments with four mice of each genotype. Data represent means ± SD (n = 4). *, P < 0.05. (D and E) Protein levels of cyclin D1 and cyclin E (D) and of C/EBPα and C/EBPβ (E) were examined by Western blotting and densitometry calculations (bar graphs). Data represent means ± SD (n = 3 to 5). *, P < 0.05. CRM, cross-reactive molecules observed on cyclin E and C/EBPα filters. Bars, 20 μm.

Livers of dnp300 mice have alterations of gene expression in several signaling pathways.

To determine which pathways are affected by the inhibition of p300, we performed RNA-Seq analysis of livers of 2-month-old dnp300 mice and age-matched WT mice. This analysis identified alterations in the expressions of 1,015 genes. Clustering of these altered genes revealed that the inhibition of p300 causes changes in the expression levels of genes coding for chromatin remodeling proteins; phosphoproteins; DNA damage proteins; oncoproteins; and enzymes that control fatty liver, apoptosis, the cell cycle, and translation (Fig. 3A). The most significant alterations in each pathway were confirmed by QRT-PCR. The most significant pathway with differentially expressed genes showed reduced expression levels of genes coding for phosphoproteins (Fig. 3B), suggesting that phosphorylation is changed in livers of dnp300 mice. Levels of several mRNAs coding for chromatin remodeling proteins (Smard2, Sirt3, and Smarcal1) were increased in livers of dnp300 mice (Fig. 3C), suggesting that additional changes in chromatin might be mediated by this increase. It is also important to note that genes of the apoptotic pathway with elevated expression levels are mainly related to the inhibition of apoptosis (Fig. 3E). We have detected a significant reduction in the level of oncogenes (Fig. 3D), suggesting that dnp300 mice might be resistant to the development of liver cancer. Alterations in enzymes that regulate fatty liver are consistent with our observations that dnp300 mice are resistant to the development of hepatic steatosis under conditions of a high-fat diet (9). In summary, RNA-Seq results show that the inhibition of p300 causes significant alterations of gene expression in several signaling pathways.

FIG 3.

Inhibition of p300 affects several signaling pathways in the liver. (A) List of pathways with altered gene expression determined by RNA-Seq analysis of livers of three WT and three dnp300 mice. (B) Differentially expressed genes that code for phosphoproteins. The names of these genes are shown in Table S2 in the supplemental material. Bar graphs show verification of the expression of genes selected from this category by QRT-PCR. (C) List of differentially expressed genes involved in chromatin remodeling. Bar graphs show verification of the expression of genes selected from this category by QRT-PCR. (D) List of oncogenes differentially expressed in dnp300 versus WT mice. Bar graphs show verification of the expression of genes selected from this category by QRT-PCR. (E) List of differentially expressed genes clustered in the apoptosis pathway. Bar graphs show verification of the expression of genes selected from this category by QRT-PCR. For all bar graphs, data represent means ± SD (n = 3). *, P < 0.05.

Liver regeneration after surgical resections is increased in dnp300 mice.

Currently, the main approach employed for the elimination of liver cancer is surgical resection of the tumor portion of the liver. This approach is based on the unique capability of liver to regenerate to its original size and functions. To examine the role of p300 in liver regeneration after surgical resections, we performed a two-thirds partial hepatectomy in dnp300 mice. Examination of BrdU uptake showed that hepatocytes of dnp300 mice enter the cell cycle early and have a significantly higher rate of proliferation within 72 h after surgery. Livers of dnp300 mice contained 2 to 3% proliferating hepatocytes before surgery (Fig. 2); however, after PH, an additional portion of hepatocytes started to proliferate at 24 h, bringing the percentages of BrdU-positive hepatocytes up to 10% and up to 65% 36 h after PH (Fig. 4A and B). Although there was variability in the number of BrdU-positive hepatocytes in livers of dnp300 mice, the total percentage of BrdU-positive hepatocytes within 72 h after PH was 142% ± 8%, suggesting that a significant portion of hepatocytes replicated twice within this time period (Fig. 4B). Under the same conditions, livers of WT mice had 15 to 20% proliferating hepatocytes. To confirm this observation, we examined liver proliferation by staining of livers with Ki67 and by counting mitotic figures. Figure 4C and D show that the number of Ki67-positive hepatocytes is significantly higher in livers of dnp300 mice within 72 h after PH and that mitotic figures are more abundant in these mice after PH (Fig. 4E). Examination of cell cycle proteins showed that levels of PCNA, cdc2, cyclin D1, and cyclin E are elevated early in dnp300 livers (Fig. 4F and G). Note that cyclins D1 and E were already activated in livers of dnp300 mice (Fig. 2). Thus, these studies demonstrate that liver proliferation is significantly increased in dnp300 mice after PH. Regarding the molecular mechanisms by which dnp300 increases liver regeneration, our previous findings showed that C/EBPα-S193A mice have very low levels of C/EBPα-p300 complexes and have a very high level of proliferation after PH (5). In addition, dnp300 livers show elevated levels of the C/EBPβ-LIP isoform, which is a promoter of liver proliferation. Therefore, we suggest that the reduction of C/EBPα levels and the elevation of C/EBPβ-LIP levels (Fig. 2) as well as the reduction of the levels of C/EBPα-p300 complexes (5, 9) contribute to the increased liver regeneration in dnp300 mice.

FIG 4.

Livers of dnp300 mice have increased proliferation after partial hepatectomy. (A) BrdU uptake in the livers at different time points after PH. Black arrows show BrdU-positive hepatocytes; red arrows show mitotic figures. (B) Percentages of BrdU-positive hepatocytes at different time points after PH. (C) Examination of liver proliferation by Ki67 staining. Typical images of Ki67 staining are shown. Arrows show mitotic figures. (D) Numbers of Ki67-positive hepatocytes at each time point after PH. Bars for BrdU and Ki67 staining are 20 μm. (E) Percentages of mitotic figures at each time point after PH. (F) Expression of cell cycle proteins (shown on the left) after PH, determined by Western blotting. (G) Levels of PCNA protein calculated as ratios to the level of the β-actin control. Calculation of the ratios of cdc2, cyclin D1, and cyclin E levels to the β-actin level also confirmed the elevated levels of these proteins after PH in livers of dnp300 mice (data not shown). For panels B, D, E, and G, data represent means ± SD (n = 3 to 5). *, P < 0.05.

Livers of dnp300 mice are resistant to injury mediated by carbon tetrachloride treatments.

Liver is the main organ of drug and toxicant metabolism, making it the primary target of chemically induced injury. To examine the role of p300 in liver injury, WT and dnp300 mice were injected with CCl₄, and livers were examined at different time points after CCl₄ injections. H&E staining shows that livers of dnp300 had much less injury at all time points examined (Fig. 5A). In agreement with this observation, examination of ALT and AST levels revealed that dnp300 mice had reduced ALT levels in serum (Fig. 5B), while the AST level was identically elevated in WT and dnp300 mice (data not shown). The treatment of mice with CCl₄ leads to DNA damage and activation of apoptosis to eliminate dead cells. Therefore, we examined apoptosis in dnp300 mice by using a TUNEL assay. Apoptosis was dramatically reduced in livers of dnp300 mice, especially 24 h after CCl₄ treatments. While WT mice had up to 12 to 14% apoptotic hepatocytes, livers of dnp300 mice had only 2 to 3% TUNEL-positive hepatocytes (Fig. 5C and D). The CCl₄ treatments also initiate liver proliferation to replace dead hepatocytes. Examination of DNA replication by measuring BrdU uptake showed significant increases in the numbers of replicating hepatocytes in livers of dnp300 mice, with maximum levels 48 h after CCl₄ treatments (Fig. 5E and F). We also observed early and high-level mitosis in dnp300 livers compared to that in WT mice (Fig. 5G). Taken together, these studies showed that the inhibition of p300 leads to the reduction of liver injury and apoptosis in response to CCl₄ treatments and stimulates liver proliferation to a higher degree.

FIG 5.

Livers of dnp300 mice have decreased injury and increased proliferation after acute injection of CCl₄. (A) H&E staining of livers at different time points after CCl₄ injections. Bars, 20 μm. (B) Examination of serum ALT as a measure of liver injury at different time points after CCl₄ injections. (C) Immunohistochemistry of apoptotic cells in livers of WT and dnp300 mice was performed by using a TUNEL assay. Arrows show apoptotic hepatocytes. (D) Percentages of TUNEL-positive hepatocytes at each time point after CCl₄ injections. (E) Typical images showing BrdU uptake at different time points after injection of CCl₄. Arrows show BrdU-positive hepatocytes. (F) Calculation of percentages of BrdU-positive hepatocytes in livers of WT and dnp300 mice after injury. (G) Numbers of mitotic figures per 1,000 hepatocytes. For panels B, D, F, and G, data represent means ± SD (n = 4 or 5). *, P < 0.05.

Levels of p53 protein are reduced in livers of dnp300 mice before and after CCl₄ treatments.

We next examined the molecular pathways by which dnp300 changes the response of the liver to CCl₄ treatments. Given the multilevel alterations (see data from the RNA-Seq approach, above), we examined the expressions of key regulators of proliferation and apoptosis, the p53 and C/EBP proteins, using a Western blot assay. We found that the levels of C/EBPα were significantly lower in dnp300 mice during the whole course of CCl₄ treatments (except at 6 h), while the levels of both the C/EBPβ-LAP and C/EBPβ-LIP isoforms were significantly elevated. We also found that protein levels of p53 were significantly lower in livers of dnp300 mice before treatments with CCl₄ and during the whole course of CCl₄ treatments (Fig. 6A and B). Levels of p53 mRNA were not reduced in livers of dnp300 mice (data not shown), suggesting that the reduction of the p53 protein level is mediated at the level of translation or protein stability. Therefore, we performed experiments to determine if stability and/or translation of the p53 protein is increased or reduced in livers of dnp300 mice. It has been shown that p53 is ubiquitinated by MDM2 ligase, which triggers ubiquitin-proteasome-mediated degradation of p53 (28, 29). C/EBPα is also degraded in the liver by the ubiquitin-proteasome pathways (6). Given the reduced levels of both C/EBPα and p53 in livers of dnp300 mice, we asked if the levels of these proteins might be reduced by ubiquitination and subsequent degradation by the proteasome. Therefore, we immunoprecipitated C/EBPα and p53 from WT and dnp300 livers and examined the amounts of Ub-C/EBPα and Ub-p53 conjugates. As shown in Fig. 6C, Ub-p53 conjugates were not detectable in livers of dnp300 mice; while Ub-C/EBPα conjugates were abundant. Based on these observations, we conclude that the reduced level of C/EBPα is associated with an increased level of its degradation, while the reduction of the level of p53 is mediated by a different mechanism.

FIG 6.

p53 is reduced in livers of dnp300 mice and is a target for regulation by CUGBP1. (A) Expression of C/EBPβ, C/EBPα, and p53 proteins after injection with CCl₄ was determined by Western blotting. The β-actin control is shown for C/EBPα and p53 membranes. (B) Protein levels of C/EBPα and p53 were calculated as ratios to β-actin levels. (C) Amounts of Ub-C/EBPα conjugates are increased in livers of dnp300 mice, while Ub-p53 conjugates are not detected. C/EBPα and p53 were immunoprecipitated from livers of three dnp300 mice, and the IPs were examined by Western blotting with Abs to ubiquitin. Incubation of beads with proteins from WT and dnp300 mice served as a control for nonspecific absorption. Bottom images show Western blotting of p53 and C/EBPα IPs with the corresponding antibodies to p53 and C/EBPα. (D) Protein levels of CUGBP1 are increased in livers of dnp300 mice compared to levels in livers of WT mice. Bar graphs show levels of CUGBP1 as ratios to the β-actin level. (E) Nucleotide sequence of the 5′ UTR of p53 mRNA (top), RNA probes (middle), and diagram of CUGBP1 constructs (bottom). The binding site for CUGBP1 is shown in blue. (F) UV cross-linking analyses of the interactions of WT and mutant (Mut) p53 probes with CUGBP1 proteins. (Left) Competition of WT and mutant oligonucleotides for binding to WT CUGBP1. (Middle) Interactions of CUGBP1 mutants (shown on the top) with the WT p53 probe. (Right) Bar graphs showing binding of CUGBP1 proteins as ratios to the amounts of proteins determined by densitometry of Coomassie stains.

Therefore, we next examined if translation of p53 mRNA might be inhibited in livers of dnp300 mice. In parallel studies of liver biology of dnp300 mice, we examined the expressions of several additional key regulators of liver biology, including RNA CUG-binding protein 1 (CUGBP1). This protein might activate or repress the translation of mRNAs depending on the interactions with the “active” form of eIF2α or with the “inactive” form of eIF2α phosphorylated at Ser51 (30). We first examined the levels of CUGBP1 in liver cytoplasmic extracts of dnp300 mice. We found that protein levels of CUGBP1 were 3- to 4-fold higher in livers of dnp300 mice than in livers of WT mice (Fig. 6D). Because CUGBP1 binds to CUG- and UG-rich repeats, we searched the 5′ untranslated region (UTR) of p53 mRNA for the potential binding site and found a consensus sequence that is located upstream of the AUG start codon (Fig. 6E). Therefore, we synthesized an RNA WT probe covering the consensus sequence and a mutant RNA oligomer with mutations of UG repeats (Fig. 6E) and examined the interactions of these RNAs with CUGBP1. For the initial studies, we used bacterially expressed, purified WT CUGBP1 and the CUGBP1 S302A and S302D mutants, which are linked to maltose-binding protein (Fig. 6E). These studies showed that CUGBP1 strongly interacts with the WT 5′ UTR of p53 mRNA, while its interactions with the mutant 5′ UTR of p53 mRNA are very weak (Fig. 6F). In agreement with this result, incorporation of a cold WT competitor inhibited the binding of CUGBP1 to the WT probe, while the mutant competitor did not affect binding (Fig. 6F). Since unphosphorylated CUGBP1 works as a repressor of translation, but phosphorylated CUGBP1 works as an activator of translation (30), we examined the interactions of the CUGBP1 S302A (mimicking the repressor form) and S302D (mimicking the activator form) mutants with the 5′ UTR of p53 mRNA. In these UV cross-linking experiments, we calculated the binding of CUGBP1 proteins to the 5′ UTR of p53 mRNA as a ratio of the p32 signals to the signals of proteins added to the binding reaction mixtures, which were determined by Coomassie staining on the same membrane. Figure 6F shows that the S302A mutant binds 3 to 4 times more strongly to the 5′ UTR of p53 mRNA than does WT CUGBP1 or the S302D mutant of CUGBP1. Thus, these studies revealed that the 5′ UTR of p53 mRNA contains the binding site for CUGBP1 and that the CUGBP1 S302A mutant binds much more strongly to this region than does WT CUGBP1.

CUGBP1–ph-S51-eIF2α complexes repress translation of p53 in livers of dnp300 mice.

CUGBP1 displays its translational activities through interactions with eIF2α. It has been shown that CUGBP1 activates the translation of mRNAs when it associates with active eIF2α but represses translation when it is in a complex with inactive eIF2α phosphorylated at Ser51 and that bacterially expressed (nonphosphorylated) CUGBP1 preferentially interacts with inactive eIF2α in vitro (30). To test if the unphosphorylated CUGBP1 S302 or S302A mutant interacts with inactive eIF2α in vivo, we transfected the green fluorescent protein (GFP) control and GFP-linked WT and S302A and S302D mutant CUGBP1 proteins (Fig. 7A) into Hep3B2 cells, and CUGBP1 was immunoprecipitated with monoclonal Abs to GFP. These immunoprecipitations (IPs) were then examined by Western blotting with antibodies against total eIF2α and against ph-S51-eIF2α. We found that the S302D mutant, which mimics the phosphorylated form of CUGBP1, preferentially interacts with active, unphosphorylated eIF2α, while the S302A mutant binds more strongly to ph-S51-eIF2α (Fig. 7B). Thus, these data show that the S302A mutant, or CUGBP1 unphosphorylated at S302, forms complexes-repressors with ph-S51-eIF2α in cultured cells. Given the fact that the S302A mutant also binds more strongly to the 5′ UTR of p53 mRNA, these data suggested that translation of p53 might be repressed by the CUGBP1–ph-S51-eIF2α complex. To test this suggestion, we generated constructs in which we placed the 5′ UTR of WT p53 mRNA and a mutated p53 5′ UTR upstream of the AUG codon of luciferase constructs (Fig. 7C) and cotransfected them with WT CUGBP1 and the CUGBP1 S302A and S302D mutants into Hep3B2 cells. Examination of luciferase activity showed that both WT CUGBP1 and the CUGBP1 S302D mutant activate the translation of a luciferase construct containing the 5′ UTR of WT p53 mRNA (Fig. 7C). On the contrary, the CUGBP1 S302A mutant dramatically represses the translation of the construct containing the 5′ UTR of WT p53 mRNA, but it does not affect the translation of a construct containing the mutant p53 5′ UTR (Fig. 7C). Since the CUGBP1 S302A mutant preferentially interacts with ph-S51-eIF2α and forms complexes-repressors, this result strongly suggests that the CUGBP1 S302A mutant and, perhaps, unphosphorylated WT CUGBP1 inhibit the translation of p53 mRNA.

FIG 7.

The CUGBP1–ph-S51-eIF2α complex represses p53 translation in cultured cells and in livers of dnp300 mice. (A) Diagram showing GFP-CUGBP1 constructs used in these studies. (B) Interactions of transfected CUGBP1 mutants with total eIF2α and ph-S51-eIF2α. GFP-CUGBP1 constructs were transfected into Hep3B2 cells, and CUGBP1 was immunoprecipitated with Abs to GFP. The IPs were probed with Abs to total eIF2α and to ph-S51-eIF2α. Bar graphs show ratios of eIF2α levels to CUGBP1 levels in CUGBP1 IPs. B, beads. (C) The CUGBP1 S302A mutant represses translation of p53 mRNA through the CUGBP1-binding site located in the 5′ UTR of p53 mRNA. The top shows 5′ sequences of luciferase constructs containing WT and mutant CUGBP1-binding sites from the 5′ UTR of p53 mRNA. Bar graphs show results of cotransfections of WT and mutant CUGBP1 with WT and mutant p53-luciferase constructs. Data represent means ± SD (n = 5). *, P < 0.05. (D) CUGBP1 shows a stronger interaction with p53 mRNA in livers of dnp300 mice than in livers of WT mice. (Top) Levels of CUGBP1 and eIF2α in cytoplasmic extracts after CCl₄ treatments. (Middle) Results of UV cross-linking assays of immunoprecipitated CUGBP1 with an RNA oligomer corresponding to the WT p53 probe (Fig. 6E). Coomassie stain of the UV-cross-linked membrane is shown at the bottom. (Bottom) Image showing results from two-dimensional gel electrophoresis of CUGBP1 from livers of WT and dnp300 mice 48 h after CCl₄ treatments. Red arrows on the top show the positions of phosphorylated CUGBP1, which works as an activator of translation. Black arrows show the positions of unphosphorylated CUGBP1, which represses the translation of mRNAs. (E) Amounts and associations of CUGBP1–ph-S51-eIF2α complexes with p53 mRNA are increased in livers of dnp300 mice after CCl₄ treatments. Inactive ph-S51-eIF2α was immunoprecipitated from protein extracts at different time points after CCl₄ injection, and CUGBP1 protein and p53 mRNA levels in these IPs were determined by Western blotting. (Top) Western blotting with Abs to CUGBP1. The IgG signals are shown at the bottom. (Middle) Amounts of CUGBP1 shown as ratios to the amount of IgG. (Bottom) Bar graphs showing amounts of p53 mRNA in the same ph-S51-eIF2α IPs as a summary of data from three independent experiments.

Given this observation and observations that CUGBP1–ph-S51-eIF2α complexes inhibit the translation of p53 in cultured cells, we next examined if this mechanism might operate in dnp300 mice after CCl₄ treatments. We first determined CUGBP1 expression after CCl₄ treatments and found that the levels of CUGBP1 are higher in dnp300 mice after CCl₄ injections during entire testing period (Fig. 7D). Levels of eIF2α were slightly higher at early time points after CCl₄ in dnp300 mice. We next examined CUGBP1-binding activity toward the 5′ UTR of p53 mRNA in cytoplasmic extracts using a UV cross-linking assay. To increase the sensitivity and specificity of the UV cross-linking assay, we immunoprecipitated CUGBP1 from the cytoplasmic extracts and incubated these immunoprecipitates with the 5′-UTR p53 probe. Figure 7D shows that the binding activity of CUGBP1 is significantly higher in dnp300 mice and is increased in livers of dnp300 mice 72 and 96 h after CCl₄ treatments. As shown in Fig. 7C, CUGBP1 unphosphorylated at Ser302 or the S302A mutant represses the translation of p53 mRNA. Therefore, we next examined the phosphorylation status of CUGBP1 in WT and dnp300 mice using a two-dimensional (2D) gel approach. For these studies, we used cytoplasmic extracts from mice 48 h after CCl₄ injections, the time when differences in the binding of CUGBP1 to p53 mRNA are more evident. Figure 7D (bottom) shows that phosphorylated forms of CUGBP1 were detected in WT mice in the acetic region of the 2D gel (CUGBP1 activator); however, these isoforms were not detected in livers of dnp300 mice. On the contrary, nonphosphorylated forms of CUGBP1, which work as repressors of translation, are abundant in livers of dnp300 mice. To further examine if CUGBP1 might repress p53 in livers of dnp300 mice after CCl₄ treatments, we determined the amounts of CUGBP–ph-S51-eIF2α complexes and their association with p53 mRNA. Inactive ph-S51-eIF2α was immunoprecipitated from cytoplasmic extracts, and the amounts of CUGBP1 protein and p53 mRNA were examined in these IPs. These co-IP studies showed that the amounts of these complexes-repressors are increased in livers of dnp300 mice before and after CCl₄ treatments and that these complexes are bound to p53 mRNA after CCl₄ treatments (Fig. 7E). This binding strongly correlates with repression of p53 mRNA translation. With the results from cultured cells and livers, we conclude that CUGBP1 represses p53 translation in livers of dnp300 mice.

DISCUSSION

The chromatin structure of cells in the liver undergoes dynamic changes during prenatal and postnatal development and in response to challenges such as partial hepatectomy and liver injury. Among many regulators of chromatin structure, p300 has been shown to be an important protein that regulates liver biology; however, the mechanisms of p300-dependent regulation of liver biology are not well understood. In this paper, we performed a careful examination of signaling pathways in livers of dnp300 mice. One of the critical observations is that the reduction of p300 activity leads to an elevated level of heterochromatin regions and to increased sizes of hepatocytes. It has been shown that the number of heterochromatin foci and the sizes of hepatocytes increase with age (7). We found that hepatocytes of 2-month-old mice are larger in livers of dnp300 mice than in livers of WT mice. These results suggest that a proper level of activity of p300 is involved in protection of the liver from the development of age-like changes. RNA-Seq analysis identified alterations of gene expression in multiple signaling pathways in livers of dnp300 mice. The most abundant alterations were detected in the expressions of genes coding for phosphoproteins, apoptotic proteins, and oncoproteins. Consistent with the changes of phosphoprotein signaling, livers of dnp300 mice have many alterations of gene expression that are controlled at the levels of protein stability and translation. For example, protein levels of C/EBPα and p53 are reduced in livers of dnp300 mice, while levels of the corresponding mRNAs do not change. It is also important that the levels of the C/EBPβ-LAP and -LIP isoforms and the levels of CUGBP1 protein are increased in dnp300 mice, but levels of C/EBPβ and CUGBP1 mRNAs are not changed. These observations show the complexity of the pathways that are under the control of p300 and demonstrate that p300 is a critical regulator of liver integrity and functions.

Liver is a quiescent tissue in adult mice, but it proliferates after birth to increase its size and differentiates to provide its functions to the body. In WT mice, liver proliferation and differentiation are complete at 2 months age. Examination of 2-month-old dnp300 mice showed that liver has a high rate of proliferation without additional challenges. These observations suggest that p300 is involved in the termination of liver proliferation after birth and that the inhibition of p300 by a dominant negative molecule leads to the failure of hepatocytes to stop proliferation. We also examined the role of p300 in the regulation of liver proliferation after surgical resections and after injury mediated by carbon tetrachloride. These studies demonstrated that an additional proliferative stimulus further increases the proliferation of livers in dnp300 mice and that livers of dnp300 mice do not stop proliferation in the examined time frame (96 h) after PH as well as after CCl₄-mediated injury. The precise mechanisms by which the inhibition of p300 increases liver proliferation are not known. However, our data in this paper and recently reported observations suggest that these mechanisms are associated with reduced levels of C/EBPα and p53 and an elevated level of the C/EBPβ-LIP isoform. Particularly, the reduction of the level of C/EBPα and the subsequent reduction of the level of C/EBPα-p300 complexes (9) might be responsible for the lack of termination of liver proliferation after PH. It is interesting to note that Nygard et al. examined gene profiles at later stages of liver regeneration and identified changes in the apoptotic pathway (31). Therefore, it is possible that alterations of gene expression in the apoptotic pathway in dnp300 mice also contribute to the lack of termination of liver regeneration. In addition to this, recent reports revealed that lipid metabolism and integrin-linked kinase are implicated in the termination of liver proliferation after PH (32, 33). Our data for altered expression of DGAT1/DGAT2 and genes coding for phosphoproteins are consistent with those previously reported observations. In this paper, we focused our mechanistic studies on the role of p300 in the response of the liver to CCl₄-mediated injury. Our data demonstrated that in mice with inhibited p300 activity, protein levels of C/EBPα and p53 are reduced and stay very low after CCl₄ treatments. The reduction of the level of p53 is mediated by a complex-repressor of translation, CUGBP1–ph-S51-eIF2α, which binds to p53 mRNA and inhibits the translation of p53 mRNA. In summary, we found that p300 plays a critical role in the regulation of liver biology by controlling multiple pathways to support liver proliferation, the response of the liver to CCl₄-mediated injury, and hepatic steatosis (Fig. 8).

FIG 8.

p300 regulates liver proliferation and recovery after injury through multiple pathways, which include p53 and C/EBP family proteins.