HDAC1 Cooperates with C/EBPα in the Inhibition of Liver Proliferation in Old Mice*

Epigenetic control of liver proliferation involves cooperation between transcription factors and chromatin-remodeling proteins. In this work, we found that the levels of HDAC1 (histone deacetylase 1) are increased in quiescent livers of old mice. The elevation of HDAC1 in liver is mediated by the RNA-binding protein CUGBP1. We found that the age-associated CUGBP1-eIF2 complex binds to the 5′ region of HDAC1 mRNA and increases translation of HDAC1 in the liver. Further analyses showed that CUGBP1 also increases expression of HDAC1 in cultured cells, in the livers of CUGBP1 transgenic mice, and in the livers of mice injected with cyclin D3, which enhances the formation of the CUGBP1-eIF2 complex. In livers of old mice, HDAC1 interacts with the transcription factor C/EBPα and is recruited by this protein to E2F-dependent promoters as a component of high Mr C/EBPα-Brm complexes. The recruitment of HDAC1 to c-Myc and FoxM1B promoters leads to deacetylation of histone H3 at Lys-9 on these E2F-dependent promoters. We show that HDAC1 is an important mediator of growth-inhibitory activity of C/EBPα and that small interfering RNA-mediated inhibition of HDAC1 reduces the ability of C/EBPα to inhibit cell proliferation. In addition, we have found that both elevation of HDAC1 and interaction of C/EBPα with HDAC1 are controlled by cyclin D3-dependent mechanisms. Treatment of old mice with growth hormone, which reduces cyclin D3 levels, leads to the reduction of the CUGBP1-eIF2 complex, normalization of HDAC1 levels, and inhibition of interactions of HDAC1 with C/EBPα-Brm complexes. Thus, our data demonstrate that translational elevation of HDAC1 in livers of old mice is involved in the assembly of high Mr protein-protein complexes that inhibit liver proliferation.

Although the liver is a quiescent tissue, it is able to completely regenerate itself in response to injury and after partial hepatectomy. It has been shown that aging significantly reduces regenerative capacities of the liver and proliferative response after partial hepatectomy (1)(2)(3). The first fundamental work was performed by Bucher et al. (1) in 1964 and showed that old animals have a delayed and reduced peak of DNA synthesis. This phenomenon was later confirmed by a number of publications from other groups (2)(3)(4)(5). Despite intensive work, very little is known about biochemical pathways by which aging inhibits liver proliferation. Recent studies revealed that expression and activities of two transcription factors, C/EBP␣ and FoxM1B, are altered in livers of old mice (4 -7). The C/EBP family member C/EBP␣ is expressed at high levels in the liver and is involved in the regulation of liver growth and differentiation (8 -10). C/EBP␣ inhibits liver proliferation in young mice through the interactions with Cdk2 and p21, leading to the inhibition of Cdk2 kinase activity (11)(12)(13). In old livers, C/EBP␣ is recruited to Brm complexes to repress E2F targets through the occupation of E2F-dependent promoters (5,14). FoxM1B is not detectable in quiescent livers but is elevated after partial hepatectomy and is required for proper liver proliferation (4,15). Several publications from Costa and co-workers (4,15) have shown that aged livers fail to elevate FoxM1B after partial hepatectomy, and this failure correlates with the reduced proliferation. We have recently found that the age-associated appearance of the C/EBP␣-Brm complex inhibits expression of FoxM1B in livers of old mice after partial hepatectomy (7). Although the elevation of the C/EBP␣-Brm complex in old livers is well described (5,14), the composition of such complex and mechanisms by which this complex represses promoters are not known.
Proliferation of the liver is controlled by a complex cooperation of several networks that determine expression of genes through remodeling of the chromatin structure. One of the proteins that control chromatin structure is HDAC1. HDAC1 is involved in the regulation of many biological processes, including cellular senescence (16,17). A recent paper has demonstrated that the induction of HDAC1 in melanocytes inhibits cell proliferation and triggers cellular senescence through formation of complexes with Brg1/Brm1 and Rb proteins (17). Despite the established role of HDAC1 in many tissues, very little is known about the role of HDAC1 in the biological processes of the liver. Transgenic mice with liver-specific overexpression of HDAC1 have been generated and have shown that the elevation of HDAC1 leads to down-regulation of p21 (18). It has recently been shown that hepatitis B virus-induced development of hepatocellular carcinoma is mediated by the HBxdependent repression of estrogen receptor and that this repression involves HDAC1 (19). Aging changes expression of the proteins on different levels, including alterations of protein translation. We have previously shown that the RNA-binding protein CUGBP1 regulates translation of certain mRNA in different tissues of mice (20) and that aging enhances translational activity of CUGBP1 by increasing its interactions with eukaryotic initiation translation factor eIF2 (21). It has also been shown that CUGBP1 binds to the 5Ј region of C/EBP␤ mRNA and increases translation of C/EBP␤ in regenerating livers (22,23) and in livers of old mice (21). Although the translational function of CUGBP1 has been described by many groups, little is known about translational targets of CUGBP1 and their roles in liver biology.
In this work, we have identified an additional translational target of CUGBP1, HDAC1. We found that HDAC1 protein is elevated in old quiescent livers, whereas HDAC1 mRNA is not changed. Searching for the mechanisms of HDAC1 elevation in the livers of old mice, we found that the age-specific CUGBP1-eIF2 complex is associated with HDAC1 mRNA and increases translation of HDAC1. Using tissue culture systems and animal models, we showed that activation of the CUGBP1-eIF2 pathway leads to elevation of HDAC1. Furthermore, we present evidence that, in livers of old mice, HDAC1 is associated with C/EBP␣-Brm and is required for the repression of E2F targets. Thus, the studies described in this work have identified a role of translational control in the appearance of the age-specific complex repressor C/EBP␣-Brm-HDAC1.

EXPERIMENTAL PROCEDURES
Animals-In this study, we have used young mice (4 -6 months) and old mice (20 -22 months). Generation of CUGBP1 TR mice was described in our previous work (20). In these studies, we have used 4 -6-month-old CUGBP1 TR 2 mice. To increase the CUGBP1-eIF2 complex in livers of young mice, cyclin D3 plasmid was injected into 4-month-old mice, as previously described (7,21). Animal experiments were approved by the Institutional Animal Care and Use Committee at Baylor College of Medicine (protocol AN-1439).
Isolation of Protein Extracts-Cytoplasmic and nuclear extracts were isolated as previously described (5,7,10). Briefly, livers were homogenized in buffer A containing 20 mM Tris-HCl, pH 7.5, 30 mM KCl, 10 mM ␤-mercaptoethanol, and inhibitors of phosphatases. Nuclei were spun down at 12,000 rpm for 15 min, and supernatant (cytoplasm) was kept in the Ϫ80°C freezer. The pellet (nuclei) was treated with buffer B containing 20 mM Tris-HCl, pH 7.5, 0.42 M NaCl, 10 mM ␤-mercaptoethanol, 25% sucrose, 5 mM MgCl 2 , and inhibitors of phosphatases. After a 30-min incubation on ice, nuclei were spun down at 12,000 rpm for 10 min, and supernatant (nuclear extract) was frozen and kept in a Ϫ80°C freezer.
Partial Purification of CUGBP1 from Cytoplasm Using Chromatography on a DEAE Column-Since the RNA binding activity of CUGBP1 is significantly inhibited in cytoplasm by endog-enous RNA (22), we have partially purified CUGBP1 from cytoplasm using a DEAE column. The detailed procedure has been described in our previous work (22). Briefly. 1 mg of cytoplasmic proteins were loaded onto a DEAE-Sephadex column and washed from unbound proteins. The bound proteins were eluted with different concentrations of NaCl, including 0.1, 0.2, 0.3, and 0.4 M NaCl. Under these conditions, endogenous RNA remains bound to the column, whereas the majority of proteins are eluted. The location of CUGBP1 within elution fractions was determined by Western blotting and by gel shift-supershift as shown in Fig. 2, C and D. CUGBP1 is usually eluted with 0.3 M NaCl.
Western Blotting and Co-immunoprecipitations-50 g of proteins were loaded on gradient 8 -16% or 4 -20% SDS-polyacrylamide gel (Bio-Rad). Proteins were transferred on nitrocellulose membrane, and the membrane was blocked with 10% dry milk on TTBS for 1 h. The membrane was incubated with primary antibodies for 2-4 h, washed, and incubated with secondary antibodies for 1 h. After wash, the signals were detected by detection reagents (Amersham Biosciences). Protein loading was verified by a reprobe of the membranes with ␤-actin and by Coomassie stain. Antibodies to C/EBP␣ (A144), CUGBP1 (B1), HDAC1, and eIF2␣ were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Antibodies to ␤-actin were from Sigma True-Blot secondary antibodies and IP beads from Ebioscience were used for Western blotting and co-immunoprecipitation studies.
Gel Shift and UV Cross-link Assays-Gel shift and UV crosslink assays were performed with RNA probes containing C/EBP␤ mRNA probe or the 5Ј region of HDAC1 mRNA (Fig.  2). UV cross-linking analysis was performed as described in our previous studies (22)(23)(24). RNA probes were labeled by [␥-32 P]ATP and T4 kinase. Equal amounts of radioactive RNA probes (50 -200,000 cpm) were incubated with proteins for 30 min at room temperature and subjected to UV treatment for 5 min at 125 mJ. Reaction mixtures were loaded on denaturing SDS-polyacrylamide gel. Proteins were transferred onto the membrane and autoradiographed. The membranes were stained with Coomassie Brilliant Blue to verify protein loading.
Size Exclusion Chromatography and Analyses of Protein-Protein Complexes-This procedure was used for the isolation of CUGBP1-eIF2 and C/EBP␣-HDAC1-Brm complexes. CUGBP1-eIF2 complex was separated from free CUGBP1 by size exclusion chromatography of the cytoplasmic extracts on a SEC400 column (DuoFlow; Bio-Rad), as previously described (5)(6)(7). The complex was immunoprecipitated from high M r fractions with Abs to CUGBP1 and RNA was isolated from the complex and used for RT-PCR with primers to HDAC1 mRNA. For the isolation of C/EBP␣-Brm complex, nuclear extracts from livers of old mice were fractionated on an SEC400 column. Five runs were combined, and C/EBP␣ was immunoprecipitated from fractions containing the complex. C/EBP␣ IPs were separated by two-dimensional gel as described (6,10) and stained with Coomassie Brilliant Blue or transferred on the filters and probed with Abs to HDAC1 and HP1␣.
Examination of HDAC1 mRNA-Total RNA was isolated from livers and from cultured cells using TRIzol reagent (Invitrogen). The RNA was reverse transcribed and used for real time PCR with primers specific for HDAC1 mRNA. The sequence of these primers is as follows: forward, 5Ј-ATTCCT-GCGTTCTATTCGCC CAGA-3Ј; reverse, 5Ј-TTAGCAGT-TCCAGGATGGCCAAGA-3Ј. The levels of HDAC1 mRNA were calculated as ratios to ␤-actin or to glyceraldehyde-3phosphate dehydrogenase. A summary of three or four independent experiments is shown.
Chromatin Immunoprecipitation (ChIP) Assay-The chromatin immunoprecipitation assay was performed using the ChIP-IT kit (Active Motif). The chromatin solutions were isolated from livers of young and old animals, and DNA was sheared by enzymatic digestion according to the instruction manual. Antibodies against C/EBP␤, C/EBP␣, E2F4, Brm, HDAC1, Rb (Santa Cruz Biotechnology), acetyl-histone H3-Lys-9 (Cell Signaling), and histone H3-trimethyl Lys-9 (Abcam) were added to each aliquot of precleared chromatin and incubated overnight. Protein G beads were added and incubated for 1.5 h at 4°C. DNA was isolated and used for PCRs with primers specific for E2F binding sites within the c-Myc and FoxM1B promoters. The sequences of the primers for these promoters are described in our previous studies (5,7). PCR mixtures were amplified for one cycle of 95°C for 5 min, annealing temperature for primers (62°C) for 5 min, and 72°C for 2 min. Then PCR mixtures were amplified for 34 cycles of 95°C for 1 min, annealing temperature for 2 min, and 72°C for 1.5 min. PCR products were separated by 1.5% agarose gel electrophoresis or by 4% PAGE.

Protein Levels of HDAC1 Are Increased in Livers of Aged Mice-
Since HDAC1 plays a critical role in cellular senescence (17), we examined expression of HDAC1 in livers of old and young mice. Western blotting analysis showed that protein levels of HDAC1 are 2.5-3-fold higher in livers of old mice (Fig. 1, A  and B). Consistent with the elevation of protein levels, enzymatic activity of HDAC1 was also increased in old livers (Fig. 1C). Examination of mRNA levels of HDAC1 by RT-PCR, however, showed no significant differences (Fig. 1D), suggesting that HDAC1 is elevated in livers of old mice at the level of translation or protein stability.
We have previously shown that RNA-binding protein, CUGBP1, activates translation of C/EBP␤ mRNA in the liver of old mice (21). The aged livers increase translational activities of CUGBP1 by enhancement of the interaction of CUGBP1 with eIF2 (21). Given correlations between activation of CUGBP1 and induction of protein levels of HDAC1 in livers of old mice, we asked if the CUGBP1-eIF2 complexes might regulate translation of HDAC1. We first determined whether HDAC1 mRNA is associated with the CUGBP1-eIF2 complexes. Cytoplasm of old livers was fractionated by size exclusion chromatography, and the CUGBP1-eIF2 complex was localized within fractions by EMSA (Fig. 1E) and by UV cross-link with C/EBP␤ probe (Fig. 1F). The complex was isolated from high M r fractions by immunoprecipitation with Abs to CUGBP1 as shown in Fig. 1G. RT-PCR with RNA isolated from the complex showed that HDAC1 mRNA was associated with the CUGBP1-eIF2 complex in old livers (Fig. 1G). The association of the CUGBP1-eIF2 complex with HDAC1 mRNA in the aged livers is specific, since phosphoenolpyruvate car- Western blotting was performed with nuclear extracts from livers of young (4 months) and old (22 months) mice. The filter was reprobed with Abs to ␤-actin and stained with Coomassie Brilliant Blue (bottom). B, bar graphs show levels of HDAC1 protein as a summary of three experiments with six animals of each age group. C, HDAC1 activity is increased in livers of old mice. The HDAC1 activity was examined using HDAC1 kit (Upstate Biotechnology, Inc.). D, levels of HDAC1 mRNA are not altered in livers of old mice. Semiquantitative RT-PCR was performed with RNA isolated from livers of young and old mice. The levels of HDAC1 mRNA were calculated as ratios to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) control. A summary of two experiments with three animals of each age group is shown. E, purification of the CUGBP1-eIF2 complex by size exclusion chromatography. The CUGBP1-eIF2 complex was separated from free CUGBP1 by size exclusion chromatography. The location of the complex was determined by gel shift assay with the C/EBP␤ probe, and the complex was immunoprecipitated with monoclonal Abs to CUGBP1. F, examination of the size exclusion fractions by UV cross-link with the C/EBP␤ probe. The position of CUGBP1 is shown by an arrow. Positions of molecular weight markers are shown on the right. G, HDAC1 mRNA is associated with the CUGBP1-eIF2 in livers of old mice. RNA was isolated from the CUGBP1-eIF2 complex and examined by RT-PCR with primers specific to HDAC1 mRNA. RT-PCR with primers to phosphoenolpyruvate carboxykinase (PEPCK) served as a negative control. SEPTEMBER 19, 2008 • VOLUME 283 • NUMBER 38

JOURNAL OF BIOLOGICAL CHEMISTRY 26171
boxykinase mRNA used as the control was not observed in these complexes (Fig. 1G).
CUGBP1 Binds to the 5Ј Region of HDAC1 mRNA and Increases Translation of HDAC1-We next examined if CUGBP1 interacts with the HDAC1 mRNA and if CUGBP1 increases translation of HDAC1. In the initial experiments, we determined a region of HDAC1 mRNA to which CUGBP1 binds. The 5Ј region of the mouse HDAC1 mRNA contained several GCN repeats ( Fig. 2A). Previous studies have shown that CUGBP1 binds to GCN repeats and that mutations of GC islands within RNAs abolished interactions of CUGBP1 with RNAs (23)(24)(25). Therefore, we synthesized an RNA probe, the sequence of which is shown in Fig. 2A, and examined interac-tions of this probe with bacterially expressed purified MBP-CUGBP1 and with CUGBP1 purified from proliferating livers using fractionation on a DEAE column as described (22). Initial UV cross-link experiments with bacterially expressed CUGBP1 showed that CUGBP1 binds to the 5Ј region of HDAC1 mRNA and that this binding is specific, since the addition of cold HDAC1 probe completely blocks the binding (Fig. 2B). We have previously shown that the DEAE steps of purification significantly increase RNA binding activity of CUGBP1 due to a removal of endogenous RNA, which competes for the interactions with the probe (22). Therefore, we purified CUGBP1 from the liver by chromatography on DEAE column and used this protein for further investigations. UV cross-link assay with  CUGBP1 purified from the liver showed that CUGBP1 (located in the 0.3 M NaCl elution fraction) interacts with the HDAC1 mRNA (Fig. 2C). To confirm the identity of CUGBP1 in the 0.3 M NaCl fraction, we performed EMSA in which antibodies to CUGBP1 were incorporated into the binding reactions. As can be seen in Fig. 2D, antibodies to CUGBP1 supershifted the complex, confirming that CUGBP1 binds to the 5Ј region of HDAC1 mRNA. Given the association of the CUGBP1-eIF2 complex with HDAC1 mRNA in the livers of old mice (Fig. 1), we suggested that the CUGBP1-eIF2 complex might increase translation of HDAC1 through the interaction with the 5Ј region of HDAC1 mRNA. To examine this suggestion, the biologically active CUGBP1-eIF2 complex was isolated from the cytoplasm of old livers using three steps of HPLC-based chromatography, as described in our previous work (21). Elution fractions from the last step of purification (chromatography on CHT1 column) were examined by a UV cross-link assay with the HDAC1 mRNA probe. The CUGBP1-eIF2 complex consists of three RNA-binding proteins, one of which is CUGBP1 (21). Consistent with these observations, three proteins of the complex interacted with the HDAC1 mRNA (Fig. 2E). To determine if CUGBP1 increases translation of HDAC1, we overexpressed GFP-CUGBP1 in Hep3B2 cells and examined expression of HDAC1 protein.

5'-G C C G G G C G G C G A G C A AG AU G AU G G C G C A G AC U C AG G G C AC C A AG AG G A A AG U C U G U U AC U AG U E M S
Western blotting with antibodies to CUGBP1 detects both endogenous and GFP-CUGBP1, which allows us to monitor a ratio of transfected and endogenous CUGBP1 (see Fig.  2F). As can be seen in Fig. 2F, CUGBP1 increased expression of HDAC1 up to 2.5-3-fold. This increase is specific, since the levels of ␤-actin are not changed by CUGBP1. Examination of HDAC1 mRNA in cells transfected with CUGBP1 showed no change, suggesting that CUGBP1-mediated elevation of HDAC1 does not involve activation of transcription or stabilization of the HDAC1 mRNA (Fig.  2F, bar graphs). Since the levels of ectopically expressed GFP-CUGBP1 are above physiological levels of endogenous CUGBP1 (Fig.  2F), we next examined if the inhibition of endogenous CUGBP1 in HepB2 cells will affect expression of HDAC1. The endogenous CUGBP1 was inhibited with siRNA, as previously described (21), and the levels of CUGBP1 and HDAC1 proteins were determined by Western blotting. Transfection of unrelated short RNA was used as the control. Fig. 2G shows that the inhibition of CUGBP1 leads to the significant reduction of HDAC1, whereas transfection of the control RNA does not change the expression of HDAC1 and CUGBP1. Examination of the HDAC1 mRNA demonstrated that the inhibition of CUGBP1 does not affect levels of HDAC1 mRNA (Fig. 2G, bar graphs). Thus, these studies showed that CUGBP1 binds to the HDAC1 mRNA and that ectopic expression of CUGBP1 increases translation of HDAC1 in cultured hepatocytes, whereas the inhibition of CUGBP1 reduces translation of HDAC1.
The CUGBP1-eIF2 Complex Increases Translation of HDAC1 in the Liver-To examine if the binding of CUGBP1-eIF2 complex to the HDAC1 mRNA increases translation of HDAC1 in the liver, we used two mouse models that up-regulate the CUGBP1-eIF2 complex in young livers: CUGBP1 TR mice and mice injected with cyclin D3, which phosphorylates CUGBP1 and increases amounts of CUGBP1-eIF2 complex (21). To ensure that CUGBP1 TR livers used for these studies contained the CUGBP1-eIF2 complex, we performed co-IP studies and found that CUGBP1 is associated with eIF2 (Fig. 3A). Western blotting showed that protein levels of HDAC1 are also increased in these animals (Fig. 3, C and D). To examine if the CUGBP1-mediated elevation of HDAC1 in CUGBP1 transgenic mice occurs on the level of translation, we determined levels of HDAC1 mRNA in WT and CUGBP1 TR mice. Real time PCR showed no significant differences in the levels of HDAC1 mRNA (Fig. 3E). Since the formation of CUGBP1-eIF2 complex is controlled by cyclin D3-Cdk4 (21), we injected cyclin D3 into 4-month-old mice and examined the CUGBP1-eIF2 complex and levels of HDAC1 protein. Cytoplasmic extracts were fractionated on a size exclusion column SEC400 (Bio-Rad), and the fractions were examined by Western blotting with antibodies to CUGBP1 and eIF2␣. Fig. 3B shows that, in cyclin D3-injected mice, both CUGBP1 and eIF2␣ are shifted to fractions containing high M r complexes. To confirm that this shift is mediated by the formation of CUGBP1-eIF2 complex, CUGBP1 was immunoprecipitated from each fraction, and the IPs were probed with antibodies to eIF2␣. As one can see in Fig.  3B, eIF2␣ is associated with CUGBP1 in fractions containing high M r complexes, whereas eIF2␣ is very weak or undetectable in CUGBP1 IPs from fractions of control livers. Thus, this HPLC-based examination showed that cyclin D3 increases the amounts of the CUGBP1-eIF2 complex. Western blotting with antibodies to HDAC1 showed that the elevation of the CUGBP1-eIF2 complex correlated with a significant increase of HDAC1 (Fig. 3, C and D). Calculations of HDAC1 levels as ratios to ␤-actin showed that HDAC1 is 2.5-fold increased in livers of CUGBP1 TR mice and 3-4-fold increased in livers of mice injected with cyclin D3 (Fig. 3D). These data show that the elevation of the CUGBP1-eIF2 complex in young livers leads to the increase of protein levels of HDAC1.

HDAC1 Is a Component of the C/EBP␣-Brm Complex in Livers of
Old Mice-Given established mechanisms of the elevation of HDAC1 in the mouse liver, we next determined biological consequences of this elevation. Livers of old animals have reduced regenerative capacities (1-3), which are associated with the elevation of a complex repressor C/EBP␣-Brm (5,14). Since previous studies have shown that HDAC1 interacts with C/EBP␣ (26), we examined if the elevation of HDAC1 in livers of old mice might be involved in the repression of genes through interactions with the C/EBP␣-Brm complex. We also tested heterochromatin-associated protein 1␣, HP1␣, because HP1␣ is associated with both C/EBP␣ and HDAC1 (27). Several approaches were applied to examine the association of C/EBP␣ with HDAC1 and HP1␣. Co-IP studies revealed that interactions of C/EBP␣ with HDAC1 and HP1␣ were increased in livers of old mice (Fig. 4A). To confirm these data, C/EBP␣ was immunoprecipitated from young and old livers, and enzymatic activity of HDAC1 was determined in C/EBP␣ IPs. We found that HDAC1 activity is 2.5-3-fold higher in C/EBP␣ IPs from old livers than in IPs from young livers. The HDAC1 activity is specific, since trichostatin A inhibitor blocks the HDAC1 activity in C/EBP␣ IPs (Fig. 4B). We next examined the interactions of HDAC1 and PH1␣ with C/EBP␣ using HPLC-based techniques. In these studies, we have also asked if HDAC1 and HP1␣ might be associated with the C/EBP␣-Brm complex, which is abundant in livers of old mice. We have fractionated nuclear extracts of old livers by size exclusion chromatography and examined C/EBP␣, HDAC1, and HP1␣ in size exclusion fractions. HDAC1 and HP1␣ colocalize with C/EBP␣ in the high M r fractions that contain the C/EBP␣-Brm complex (Fig. 4C). To determine if HDAC1 is associated with C/EBP␣ in high M r fractions, we have immunoprecipitated C/EBP␣ from each fraction and examined HDAC1 activity in size exclusion fractions and in C/EBP␣ IPs from each fraction. These studies demonstrated that the enzymatic activity of HDAC1 is observed in the C/EBP␣-Brm complex (Fig. 4C). To obtain additional evidence that HDAC1 is associated with the C/EBP␣-Brm complex, we purified the complex from old livers by gel-filtration and immunoprecipitations as shown in Fig. 5A, separated components by two-dimensional gel, and performed Western blotting with antibodies to HDAC1 and to HP1␣. Coomassie stain of two-dimensional gel showed that the purified C/EBP␣-Brm complex contains known components (determined by Western blotting; data not

. HDAC1 is a component of the C/EBP␣-Brm complex in old livers.
A, the interaction of HDAC1 with C/EBP␣ is increased in livers of old mice. C/EBP␣ was immunoprecipitated from young and old livers, and HDAC1, C/EBP␣, and HP1␣ were examined in these IPs. Input, levels of HDAC1 and HP1␣ proteins were determined by Western blotting with protein extracts used for Co-IP studies. B, C/EBP␣-HDAC1 complexes possess deacetylase activity. C/EBP␣ IPs were examined for the deacetylase activity. Trichostatin A (TSA) was added to reaction mixtures to examine the specificity of the assay. C, HDAC1 activity is associated with the high M r C/EBP␣-Brm complex. Nuclear extracts from old livers were fractionated by size exclusion chromatography. Western blotting with Abs to C/EBP␣, HDAC1, and HP1␣ was performed. Gel filtration fractions and C/EBP␣ IPs from each fraction were examined by an HDAC1 enzymatic assay with and without the inhibitor of HDAC1, trichostatin A.
shown) and several additional unknown proteins (Fig. 5A). Western blotting revealed that two of these new proteins interact with antibodies to HDAC1 and with antibodies to HP1␣ (Fig. 5B). Taken together, these studies revealed that HDAC1 and HP1␣ are components of the C/EBP␣-Brm complex in old livers and that the C/EBP␣-Brm-HDAC1 complex possesses deacetylase activity. The Growth Hormone-Cyclin D3 Pathway Regulates CUGBP1-eIF2 Complexes and Interactions of HDAC1 with C/EBP␣ in the Livers of Old Mice-We have previously found that aging liver increases cyclin D3-Cdk4 in cytoplasm and nuclei and that this accumulation leads to the formation of C/EBP␣-Brm and CUGBP1-eIF2 complexes (6, 21). Given our new data for the role of the translational CUGBP1-eIF2 complex in the elevation of HDAC1, we asked if the reduction of cyclin D3 by growth hormone (GH) will affect the CUGBP1-eIF2 complexes, expression of HDAC1, and interactions of HDAC1 with the C/EBP␣-Brm complex. We first examined if GH changed cyclin D3 levels in cytoplasm. Fig. 6A shows that protein levels of cyclin D3 were higher in cytoplasm from livers of old mice compared with levels of cyclin D3 in the cytoplasm of livers of young mice. However, the levels of cyclin D3 were significantly reduced in the cytoplasm of old mice by treatment with GH. Examination of CUGBP1 and eIF2␣ showed that, consistent with previous reports, protein levels and phosphorylation of CUGBP1 were increased in the cytoplasm of old livers but were reduced by the treatment with GH. The levels of eIF2␣ did not differ significantly between young and old mice. We next precipitated CUGBP1 from cytoplasmic extracts and examined the association of eIF2␣ and CRT (components of the age-specific CUGBP1-eIF2 complex (21)) with CUGBP1. Fig.  6B shows that the CUGBP1-eIF2 complexes were abundant in old livers and that treatment of the mice with GH significantly reduced amounts of the CUGBP1-eIF2 complexes. Western blotting of nuclear proteins showed that protein levels of HDAC1 were also reduced in old livers after treatment with GH (Fig. 6C). Thus, these data revealed that elimination of the CUGBP1-eIF2 complex correlates with the reduction HDAC1 in GH-treated mice. We next examined if the association of HDAC1 with C/EBP␣-Brm complex was affected by GH. Coimmunoprecipitation studies revealed that the C/EBP␣-Brm complex is abundant in livers of old mice and that GH eliminates the complex. This GH-mediated reduction of the C/EBP␣-Brm complex also reduced interactions of HDAC1 with C/EBP␣ (Fig. 6D). Taking together these data, we conclude that HDAC1 interacts with the C/EBP␣-Brm complex in livers of old mice. These studies also revealed that GH-cyclin D3 pathway controls formation of the translational CUGBP1-eIF2 complex and interactions of HDAC1 with C/EBP␣-Brm complex in the liver. These data are consistent with the hypothesis that the increase of HDAC1-C/EBP␣-Brm complexes in old livers is mediated by a reduction of GH and following elevation of cyclin D3.

HDAC1 Occupies E2F-dependent Promoters in Livers of Old Mice as the Component of C/EBP␣ Complexes and Inhibits Cell
Proliferation-We next examined if HDAC1 is associated with E2F promoters and if this association requires formation of the C/EBP␣-Brm complex. The chromatin immunoprecipitation assay was performed with c-Myc and FoxM1B promoters, since these two genes are critical regulators of liver proliferation and are repressed by the C/EBP␣-Brm complex in livers of old mice (5,7). We found that E2F4 and Rb were observed on the c-Myc and FoxM1B promoters in young animals, whereas E2F4, Rb, C/EBP␣, and HDAC1 were associated with these promoters in old livers (Fig. 6E). GH-mediated elimination of the C/EBP␣-Brm-HDAC1 complexes led to the removal of C/EBP␣, Rb, and HDAC1 from the promoters. In these studies, Brm was not detectable on the c-Myc and FoxM1 promoters. We suggest that the lack of Brm on the E2F-dependent promoters might be due to masking the site of interaction with antibodies within the C/EBP␣-Brm complex. It is also possible that Brm is required for the assembly of the complex, but it is dissociated from the complex on the chromatin. Since HDAC1 inhibits transcription of the genes via deacetylation of histone H3, we have examined the occupation of c-Myc and FoxM1B promoters by Lys-9-acetylated histone H3 and by histone H3 trimethylated at Lys-9. The acetylation of histone H3 at Lys-9 is a mark of transcriptionally active regions, whereas the methylation of histone H3 at Lys-9 is associated with transcriptional repression. ChIP assays with specific antibodies showed that, in livers of young mice, acetyl-Lys-9-H3 is bound to the FoxM1B and c-Myc promoters; however, this histone mark is not detectable on these promoters in livers of old mice. On the contrary, trimethyl-Lys-9-H3 is not detectable on these promoters in livers of young mice, but it is present on the promoters in livers of old mice.
These patterns of occupation of the FoxM1B and c-Myc promoters are consistent with the hypothesis that the C/EBP␣-mediated recruitment of HDAC1 to E2F-dependent promoters represses transcription of these genes. Interestingly, treatments of old mice with GH reverse the patterns of binding to those observed in livers of young mice. Thus, these studies showed that HDAC1 is a component of C/EBP␣ complexes that occupy E2Fdependent promoters in old livers and that occupation of the E2F-dependent promoters by HDAC1 leads to transcriptional repression of these genes. C/EBP␣ is one of the key proteins of the C/EBP␣-Brm complex that determines age-specific formation of the complex in the liver (5,7). To examine the role of HDAC1 in growth-inhibitory activity of C/EBP␣, the expression of HDAC1 was inhibited in Hep3B2 cells by siRNA (Fig. 7A, top), and C/EBP␣ was transfected in cells with reduced levels of HDAC1. A colony formation assay showed that the ability of C/EBP␣ to arrest cell proliferation was significantly reduced in cells with inhibited HDAC1 (Fig. 7, A and B). Note that we obtained around 50% inhibition of the activity of C/EBP␣ in cells with reduced HDAC1. The remaining activity of C/EBP␣ might be due to additional pathways of C/EBP␣ growth arrest, such as inhibition of Cdk2 and stabilization of p21 (11,12), or due to incomplete inhibition of HDAC1 by siRNA.

DISCUSSION
The Translational CUGBP1-eIF2 Complex Elevates HDAC1 in the Liver-Previous studies have established the role of transcriptional control in the age-associated loss of proliferative capacities of the liver (4,5,7,14). In this work, we have presented evidence for the contribution of translational pathways in the formation of age-specific complexes that inhibit liver proliferation. We have found that HDAC1 protein levels are increased in quiescent old livers by a translational mechanism that involves CUGBP1 and eIF2. These two proteins form a complex that binds to HDAC1 mRNA in vitro and is associated with the HDAC1 mRNA in the livers of old mice. We have also found that elevation or inhibition of CUGBP1 in cultured hepatocytes increases or reduces HDAC1 expression correspondingly. The formation of the . The reduction of cyclin D3 by GH inhibits the CUGBP1-eIF2-HDAC1 pathway and removes HDAC1-C/EBP␣-Brm complexes from E2F-dependent promoters. A, treatment of old mice with GH reduces protein levels of cyclin D3 and CUGBP1 in cytoplasm of the liver. Western blotting with cytoplasmic extracts from livers of young, old, and old GH-treated mice was performed using Abs to cyclin D3, CUGBP1, eIF2␣, and ␤-actin. Data with two animals of each group are shown. ␤-Actin controls are shown for cyclin D3 and eIF2␣. The bar graphs show protein levels of cyclin D3 and CUGBP1 calculated as ratios to ␤-actin. B, GH reduces the CUGBP1-eIF2 complex. CUGBP1 was immunoprecipitated from cytoplasmic extracts, and the IPs were probed with antibodies to eIF2␣, CRT, and CUGBP1. C, the reduction of CUGBP1-eIF2 complexes in cytoplasm of GH-treated mice correlates with the reduction of HDAC1 in nuclear extracts. Nuclear extracts of young, old, and old GH-treated mice were probed with antibodies to HDAC1. The membrane was reprobed with antibodies to ␤-actin. D, elimination of the CUGBP1-eIF2 complex by the growth hormone-cyclin D3 pathway disrupts the interaction of HDAC1 with the C/EBP␣-Brm complex. C/EBP␣ was immunoprecipitated from livers of young, old, and old GH-treated mice. The IPs were probed with antibodies to Brm, C/EBP␣, and HDAC1. E, HDAC1 occupies E2F-dependent promoters as the component of C/EBP␣ complexes. GH-mediated disruption of the C/EBP␣-Brm complex removes HDAC1 from the E2F-dependent promoters. Chromatin solutions were isolated from young, old, and old GH-treated mice. E2F4, C/EBP␣, Rb, Brm, HDAC1 (HD1), acetyl-Lys-9-H3, and trimethyl-Lys-9-H3 were immunoprecipitated, and the IPs were examined by PCR with primers for c-Myc and FoxM1B promoters. In, 1:100 input. Ag. incubation with agarose beads (negative control).  SEPTEMBER 19, 2008 • VOLUME 283 • NUMBER 38

JOURNAL OF BIOLOGICAL CHEMISTRY 26177
bition of HDAC1 reduces the ability of C/EBP␣ to cause growth arrest. Consistent with these observations, treatment of old mice with GH also normalizes expression of HDAC1 (Fig. 6C) and reduces the C/EBP␣-Brm-HDAC1 complex. Taken together, these observations revealed that the elevation of HDAC1 in livers of old mice is involved in the inhibition of proliferation.