Growth Hormone Corrects Proliferation and Transcription of Phosphoenolpyruvate Carboxykinase in Livers of Old Mice via Elimination of CCAAT/Enhancer-binding Protein α-Brm Complex*

Growth hormone (GH), which is reduced with age, corrects the impaired proliferative capacity of livers of old animals. In this paper, we present a mechanism by which GH eliminates age-dependent negative control of proliferation and increases transcription of liver-specific genes in livers of old mice. The reduced proliferative capacities of the liver of old animals are associated with the CCAAT/enhancer-binding protein α (C/EBPα)-Brm complex, which inhibits E2F-dependent promoters. We found that a sequestration of C/EBPα into complexes with Brm leads to a weak interaction of C/EBPα with promoters of liver-specific genes, expression of which is reduced in old animals. Injection of either GH or the regulator of the amplitude of endogenous GH release, ghrelin, reduces the C/EBPα-Brm complex in livers of old mice, leading to a derepression of E2F targets, to increased interactions of C/EBPα with promoters of liver-specific genes, and to correction of their expression. GH-dependent elimination of the complex is mediated by the inhibition of cyclin D3-CDK4 activity and by elevation of a phosphatase, protein phosphatase 2A, which dephosphorylates C/EBPα and dissociates the complex.

C/EBP 2 family proteins regulate liver proliferation and differentiation. Two members of C/EBP family, C/EBP␣ and C/EBP␤, are involved in molecular pathways that cause an aging phenotype in tissue culture systems and in the liver (1)(2)(3)(4). C/EBP␣ and C/EBP␤ interact with the promoters of liver-specific genes and activate expression of corresponding proteins. In addition to activation of liver-specific promoters, C/EBP␣ plays a key role in the inhibition of liver proliferation (5)(6)(7)(8). Examination of C/EBP␣ knock-out mice showed that C/EBP␣ is required for the inhibition of hepatocyte proliferation in newborn animals and after birth (7). Biological functions of C/EBP␣ during prenatal development of the liver depend on the cell type in which C/EBP␣ is expressed. For example, examination of proliferation in prenatal livers of C/EBP knockout mice at day 17 of gestation revealed that the deletion of C/EBP␣ causes a 3.5-fold increase of proliferation in hepatocytes, whereas billiarly epithelial cells do not show significant change in the rate of proliferation (9). C/EBP␣ is also required for the inhibition of liver growth in adult mice. Experimental work from Wang's group (6) demonstrated that growth-inhibitory activity of C/EBP␣ in adult animals protects livers from development of tumors and that this activity is mediated by inhibiting CDK2 and by an induction of protein levels of p21 in nuclei. Taken together, these studies revealed that liver-specific mechanisms of C/EBP␣ growth arrest in young animals involve direct interactions of C/EBP␣ with cell cycle-dependent kinases (10,11) and with p21 (6,7). Evaluation of C/EBP␣ animal models suggests a critical role for C/EBP␣ in aging. For example, deletion of C/EBP␣ in adult animals leads to the development of an aging phenotype in the liver (12,13). Consistent with these findings, we and others have observed that aging changes pathways of C/EBP␣ growth arrest in liver by recruitment of C/EBP␣ into complexes with Brm, Rb, and E2F, which inhibit transcription of E2F targets (1,2,4). It is well known that aging reduces the proliferative response of the liver (8,14,15). Recently, Rando's group (2) showed that exposure of old mice to a young systemic environment by parabiosis restored proliferative capacity of livers in old mice. This restoration involved inhibition of the age-specific C/EBP␣-Brm complex. One of the components of the young systemic environment is growth hormone. The amplitude of episodic growth hormone (GH) release declines during aging, and the impaired proliferative potential of old livers following partial hepatectomy (PH) can be corrected by administration of GH (16). Although GH treatment results in increased insulin-like growth factor-1 production by the liver, it has been shown that the effects of GH on proliferation following PH are directly induced by GH and not indirectly through insulin-like growth factor-1 (17). The positive role of GH in the regulation of liver proliferation has been demonstrated by several studies, but the molecular pathways by which GH promotes proliferation are not well understood.
In this paper, we investigated the effects of GH on the C/EBP␣-Brm complex and on the expression of liver-specific genes. In livers of old mice, the major portion of C/EBP␣ is associated with Brm and located on E2F-dependent promoters, where the complex represses transcription of cell cycle genes, such as c-Myc and FoxM1B. The sequestration of C/EBP␣ into these complexes leads to a significant reduction of free C/EBP␣ capable of activating liver-specific promoters. Particularly, the interactions of C/EBP␣ with PEPCK and HNF6 promoters are reduced in livers of old mice, which correlate with a reduction of PEPCK transcription. We have found that GH eliminates the C/EBP␣-Brm complex by a phosphorylation-dependent mechanism that involves alterations of at least two pathways: activation of PI3K-Akt-PP2A and a reduction of protein levels of cyclin D3, leading to a reduced activity of CDK4. The GH-mediated elimination of the complex leads to the derepression of E2F targets, to an increased association of C/EBP␣ with HNF6 and PEPCK promoters, and to correction of PEPCK expression. Ghrelin treatment of old mice, which provides a more physiological approach to GH replacement during aging, also restores the young liver phenotype in old mice.
Animals and Growth Hormone Studies-Animal experiments were approved by the Institutional Animal Care and Use Committee at Baylor College of Medicine (protocol AN-1439). In this paper, we have used young (4 -6-month-old) and old (22-24-month-old) mice. Examination of C/EBP␣-CDK and C/EBP␣-Brm complexes was performed with five or six animals of each age group. Data for expression of liver-specific mRNAs and proteins in livers of young and old mice were obtained with 5-9 animals of each age group. Mouse recombinant growth hormone or ghrelin (1 mg/kg) was injected into old animals subcutaneously twice a day (at 8:30 a.m. and 8:30 p.m.) for 7 days. GH was purchased from the National Hormone and Peptide Program, and physiological saline (0.9% NaCl) was used as vehicle for GH and ghrelin and as the control. Twice a day injections provides relatively stable circulating GH levels during the whole time period. Under these conditions, injection of ghrelin causes biologically relevant release of GH (18). Examination of the C/EBP␣-Brm complex, protein levels, and phosphorylation isoforms of C/EBP␣ were performed as described below. Data represent studies of four GH-treated, four ghrelintreated, and four control animals.
Protein Isolation and Western Blotting-Nuclear extracts were isolated as described in previous papers (1,4,10). Proteins (50 -100 g) were added on gradient (4 -20% or 8 -16%) polyacrylamide gel, transferred on the membrane, and probed with antibodies to proteins of interest. To verify protein loading, each filter was reprobed with ␤-actin and then stained with Coomassie Blue. Levels of protein expression were calculated as ratios to ␤-actin.
Gel Filtration Analysis of Protein-Protein Complexes in Mouse Liver-The detailed procedure for the analysis of protein-protein complexes is described in our previous papers (1,4). Nuclear extracts were isolated from livers of young and old mice as described (7) and fractionated by size exclusion column SEC450 (high pressure liquid chromatography; BioLogic HR; Bio-Rad). Locations of the proteins within gel filtration factions and C/EBP␣ complexes were examined by Western blotting and co-IP assays.
Gel Shift Assay-Examination of the interaction of E2F-Rb complexes with the FoxM1B promoter was performed as described in our previous papers (1,19). DNA probes (see Fig. 5) were incubated with nuclear extracts from 3T3-L1 cells in the presence of salmon DNA as a competitor and loaded on a native 5% polyacrylamide gel.
Examination of C/EBP␣ Phosphorylation by Two-dimensional Gel and by Western Blotting with Ser(P) 193 -specific Abs-To examine expression of the C/EBP␣-Ser(P) 193 isoform, C/EBP␣ was first immunoprecipitated from liver nuclear extracts with antibodies to total protein and then examined by Western blotting with antibodies to Ser(P) 193 . To confirm data of the IP-Western blot with Ser(P) 193 Abs, we have also used a two-dimensional Western assay as described in our previous papers (4,19).
Coimmunoprecipitation-C/EBP␣ was immunoprecipitated from nuclear extracts with polyclonal antibodies (14AA; Santa Cruz Biotechnology), and the presence of Rb, Brm, E2F4, and C/EBP␣ IPs was examined by Western blotting with monoclonal antibodies to the mentioned proteins.
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. The size of DNA fragments produced averaged between 500 and 1000 bp in length. Antibodies against C/EBP␣, E2F1, E2F4, and Rb 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. After reversing the crosslinking, DNA was isolated and used for PCRs with primers specific for E2F binding sites within the c-Myc promoter and hydroxysteroid transferase (1) and FoxM1B promoter (see Fig.  5) as well as with primers to C/EBP␣ binding sites within the PEPCK and HNF6 promoters. The sequences of the primers for these promoters are as follows: PEPCK forward, 5Ј-GGCCTC-CCAACATTCATTAAC-3Ј; PEPCK reverse, 5Ј-GTAGCCCG-CCCTCCTTGCTTTAA-3Ј; HNF6 forward, 5Ј-GCTCGAGC-TGGCGGGCGGCACAGGC-3Ј; HNF6 reverse, 5Ј-AGGAGT-CCAGTCTTCACATCGGCTG-3Ј. As a control for the appropriate shearing of DNA, we have used primers for a region ϳ1.5 kb upstream of the C/EBP site in each promoter. The sequences of these primers are as follows: PEPCK-NP1-F, 5Ј-AACCAACTGGCCCTAACTCACAGA-3Ј; PEPCK-NP2-R, 5Ј-GCTGCAGTCCAGCTAATGCAACAA-3Ј; HNF-NP1-F, 5Ј-AGGCAGGCACTTGGGTTAAGAGAT-3Ј; HNF-NP2, 5Ј-AGAGCCTTGTCTGCTTAGGTGCTT-3Ј. PCR mixtures were amplified for 1 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.

Sequestration of C/EBP␣ into Complexes with Brm Reduces Association of C/EBP␣ with PEPCK and HNF6 Promoters in
Livers of Old Mice-A number of recent publications revealed that C/EBP␣ displays its functions via formation of multiple protein-protein complexes. Accordingly, we performed a detailed examination of the amounts of C/EBP␣ in proteinprotein complexes in livers of young and old mice. In livers of young animals, C/EBP␣ forms complexes with CDK2 and CDK4 (11). Therefore, we determined the relative amounts of C/EBP␣ complexes with CDKs and free C/EBP␣ using size exclusion chromatography/Co-IP. After separation of C/EBP␣ and C/EBP␣-CDK complexes on the SEC450 column, CDK2 and CDK4 were simultaneously precipitated from each fraction, and C/EBP␣ was examined in CDK IPs and in supernatants. As seen in Fig. 1A, C/EBP␣ is observed in CDK complexes and in supernatant. Densitometric calculations demonstrated that 20 -25% of C/EBP␣ is in complexes containing CDKs, whereas 75-80% of C/EBP␣ is CDK-free and is observed in the supernatant.
We next performed an examination of C/EBP␣ in livers of old mice, and the results of these studies are presented in Fig.  1B. After fractionation on a SEC450 column, Brm was immu-noprecipitated from each fraction, and the amount of C/EBP␣ in IPs and in supernatants was determined by Western blotting. Consistent with previous studies, C/EBP␣-Brm complex was observed in the high M r region of gel filtration. Surprisingly, no C/EBP␣ was observed in supernatants after precipitation of Brm. Three independent experiments with protein extracts from livers of three old mice reproducibly showed no or very little C/EBP␣ in supernatants of the gel filtration fractions following immunoprecipitation of Brm complexes. Calculations of the amount of C/EBP␣ showed that around 94 -95% of C/EBP␣ is associated with Brm within the high M r C/EBP␣-Brm complex. Interestingly, the co-IP assay detected a small portion of C/EBP␣ (5-6%) in Brm IPs from gel filtration fractions with a molecular mass around 200 kDa. We suggest that this complex consists of C/EBP␣ and Brm only and might be a result of partial dissociation of the high M r C/EBP␣-Rb-E2F4-Brm complex.
We next examined interaction of C/EBP␣ with E2F-dependent promoters (as a component of the C/EBP␣-Brm complex) and the interactions of C/EBP␣ with PEPCK and HNF6 promoters in livers from old mice. It has been previously shown that C/EBP␣ binds directly to the PEPCK and HNF6 promoters (12,13,20). E2F1, E2F4, C/EBP␣, and Rb were immunoprecipitated from chromatin solutions isolated from livers of young and old mice, and these IPs were used in PCRs with primers to c-Myc (E2F-dependent), HNF6, and PEPCK (C/EBP-dependent) promoters. The results of these studies are shown in Fig.  1C. In agreement with previous observations, C/EBP␣ is not detectable on the c-Myc promoter in young livers, but in old liver it is associated with the E2F site of the c-Myc promoter as a component of the C/EBP␣-Rb-E2F4-Brm complex. Examination of these IPs with primers to the HNF6 and PEPCK promoters showed that only C/EBP␣ is associated with these promoters in livers of young mice. A weak association of Rb was also observed on the HNF6 promoter in livers from young mice. These data revealed that the C/EBP␣-Brm complex does not interact with PEPCK and HNF6 promoters in aging liver. Examination of C/EBP␣ complexes in livers from old mice by ChIP assays showed markedly different results. Within the sensitivity of our assays, we could not detect interactions of C/EBP␣ with HNF6 promoter in livers from old mice. Although C/EBP␣ is detected on the PEPCK promoter in livers from old mice after 35 PCR cycles, a quantitative examination of the amounts of C/EBP␣ showed a significant difference. The PCR product of the PEPCK promoter is detected after 22-24 PCR cycles in C/EBP␣ IPs from livers of young mice, whereas detection of the PEPCK promoter in C/EBP␣ IPs from livers of old animals requires up to 28 -30 cycles (Fig. 1D). To verify the quality of chip protocol, we have examined the association of C/EBP␣ complexes with hydroxysteroid transferase promoter. This promoter has been previously shown to interact with E2F1 and C/EBP␣ independently of E2F-Rb complexes (21) and age of mice (1). As can be seen in Fig. 1, C and D, the occupation of this promoter by E2F1 and C/EBP␣ is not altered in livers of old mice. Taken together, the ChIP results for the HNF6 and PEPCK promoters indicate that the interaction of C/EBP␣ with these promoters is reduced in livers from old mice, presumably due to sequestration of C/EBP␣ into the C/EBP␣-Brm complex.
Expression of Transcriptional Targets of C/EBP␣ in Old Livers-Given the reduced interaction of C/EBP␣ with promoters of HNF6 and PEPCK genes, we examined their expression in young and old livers. In addition to these proteins, we analyzed peroxisome proliferator-activated receptor ␣ and glucokinase, since C/EBP␣ regulates their expression through direct binding to the promoters (13). Western blotting showed that protein levels of glucokinase and peroxisome proliferator-activated receptor ␣ are not changed; however, protein levels of PEPCK are significantly reduced in old livers (Fig. 2). Examination of transcripts for these direct targets of C/EBP␣ showed that levels of PEPCK mRNA are also reduced to 25-30% compared with levels observed in livers of young mice. Since C/EBP␣ is a positive regulator of the PEPCK promoter (12,13) and because free C/EBP␣ is reduced in old livers, these data suggest that a recruitment of C/EBP␣ into the C/EBP␣-Brm complex might cause the reduction of PEPCK. To further test this hypothesis, we have examined whether elimination of the C/EBP␣-Brm complex in livers of old mice corrects expression of direct targets of C/EBP␣.

GH Eliminates C/EBP␣-Brm Complex in Livers from
Old Mice by Dephosphorylation of C/EBP␣-Considering possible mechanisms for elimination of the C/EBP␣-Brm complex, we sought a pathway that eliminates the complex but does not change C/EBP␣ protein levels. Several observations suggested that GH signaling might be involved. First, GH treatment corrects liver proliferation in old animals (16), suggesting that GH might inhibit C/EBP␣-Brm complex formation. Second, in mice, where GH action is blocked, the liver does not proliferate after PH in the absence of GH (17). Therefore, we tested whether GH treatment of old mice eliminates the C/EBP␣-Brm complex in the liver. 24-month-old animals were treated with GH for 7 days, and livers were harvested from control (salineinjected) and GH-treated mice. We first examined the age-specific C/EBP␣-Brm complex by size exclusion chromatography followed by examination of C/EBP␣ and Brm in the size exclusion fractions. Fig. 3A shows that in control animals, C/EBP␣ and Brm co-localize in high M r fractions; however, in liver

. The sequestration of C/EBP␣ into C/EBP␣-Brm complexes reduces amounts of free C/EBP␣ and decreases association of C/EBP␣ with PEPCK and HNF6 promoters.
A, determination of amounts of CDK-free C/EBP␣ and C/EBP␣ within complexes with CDK2 and CDK4 in livers of young mice. Top (Western blot), nuclear proteins from livers of young mice were separated by chromatography on SEC450 column. Locations of C/EBP␣ and CDK2 within fractions were determined by Western blotting with specific antibodies. Bottom, CDK2/4 were simultaneously immunoprecipitated with specific antibodies from each fraction, and C/EBP␣ was examined in CDK2/CDK4 IPs and in supernatants. Coomassie stain shows IgG on the filter with CDK2/4 IPs. Amounts of C/EBP␣ in the CDK2/CDK4 IPs and supernatants were calculated by densitometry as ratios to total signals of C/EBP␣. B, calculations of amounts of free C/EBP␣ and C/EBP␣ in complexes with Brm in livers of old mice. Nuclear proteins from livers of old mice were separated by size exclusion chromatography, and the locations of Brm and C/EBP␣ were determined by Western blotting. Brm was immunoprecipitated from the fractions, and C/EBP␣ was examined in Brm IPs and in supernatants as described above. C, occupation of c-Myc, HNF6, and PEPCK promoters by C/EBP␣ in livers of young and old mice was examined by ChIP analysis. E2F1, E2F4, C/EBP␣, and Rb were immunoprecipitated from chromatin solutions, and the IPs were examined by PCR with primers specific to E2F region of the c-Myc promoter and to C/EBP sites on HNF6, PEPCK, and hydroxysteroid transferase promoters. Ag, agarose mock control. D, the occupation of the PEPCK and hydroxysteroid transferase promoters by C/EBP␣ in livers of young and old mice determined by a quantitative PCR. The bottom part of each panel shows a similar examination of the input (1/100) DNA. extracts from GH-treated animals, C/EBP␣ is no longer detectable in high M r fractions and is shifted to low M r complexes or free proteins. Interestingly, the position of Brm within fractions was not changed by GH treatment. The GH-mediated shift of C/EBP␣ is specific, since the position of cross-reactive (CRM) protein (on C/EBP␣ blot) is not altered by GH (Fig. 3A). Coimmunoprecipitation showed that Brm is present in C/EBP␣ IPs from high M r fractions of control livers, but Brm is not detectable in C/EBP␣ IP from size exclusion fractions from livers of GH-treated mice, showing that GH disrupts the C/EBP␣-Brm complex. To confirm these observations, we performed direct co-immunoprecipitation studies. Fig. 3B shows that Rb, E2F4, and Brm are observed in C/EBP␣ IPs from control livers, whereas these proteins are not detectable by Western blotting in C/EBP␣ IPs from the livers of old animals treated with GH. Note that although the association of C/EBP␣ with Rb is reduced by GH, it might be detected after long exposure (see Fig. 7). Thus, these results demonstrate that treatment of animals with GH eliminates the C/EBP␣-Brm complex.
To examine pathways by which GH neutralizes the complex, we determined protein levels of the components of the C/EBP␣-Brm complex. Western blotting of nuclear extracts from control animals and from animals treated with GH showed no significant changes in the protein levels of Brm, C/EBP␣, Rb, and E2F4 (Fig. 3C). Since our previous studies showed that phosphorylation of C/EBP␣ at Ser 193 is critical for the interaction with Brm (4, 10), we next examined the phosphorylated status of C/EBP␣ in livers from old mice treated with GH. Two approaches were used for these studies: Western blotting with antibodies to the Ser 193 -phosphorylated form of C/EBP␣ and two-dimensional gel separation. C/EBP␣ was immunoprecipitated with antibodies to total protein and then examined by Western blotting with Ser(P) 193 Abs and with Abs to total C/EBP␣, as described in our previous paper (4). Fig. 3D shows that, although total amounts of C/EBP␣ are not altered in livers of GH-treated mice, the amounts of Ser(P) 193 isoform are dramatically reduced. In agreement with this observation, two-dimensional examination of C/EBP␣ showed no Ser(P) 193 in old GH-treated animals (Fig. 3E). Thus, these results of two independent methods showed that GH eliminates the Ser(P) 193 isoform of C/EBP␣, leading to a disruption of the age-specific complex.
GH Reduces CDK4/CDK6 Activity in Old Livers by Downregulation of Cyclin D3-We have shown that the presence of a phosphate on Ser 193 depends on the balance of activities of CDK4/CDK6 and PP2A and that aging increases phosphorylation of Ser 193 by activation of CDK4/CDK6 through elevation of cyclin D3 (4). Therefore, we have next determined the effects of GH on expression of cyclin D3 and on kinase activities associated with cyclin D3. Western blotting with specific antibodies showed that protein levels of cyclin D3 are reduced in livers from old mice treated with GH (Fig. 4A). A series of multiple experiments showed a 3-4-fold reduction compared with levels in control animals. We then examined kinase activity associated with cyclin D3 in livers from old mice following GH treatment. Cyclin D3 was immunoprecipitated with specific Abs and examined in a kinase assay using glutathione S-transferase-Rb as substrate. Fig. 4B shows a typical result of these studies. GH treatment reduces activity of kinases associated with cyclin D3, which correlates with the reduction of protein levels of cyclin D3. Thus, these studies demonstrate that GH reduces cyclin D3-CDK4/6 activity in livers of old mice.
GH Increases Nuclear Concentrations of PP2A, Which Dephosphorylates C/EBP␣-The phosphorylation of Ser 193 of C/EBP␣ is also controlled by PI3K-Akt-PP2A (10). Therefore, we examined this pathway in livers from old mice treated with GH. Since Akt-1 is a target of PI3K (22), we performed Western blotting of protein extracts with antibodies specific to "active," Ser 473 -phosphorylated Akt1. These studies showed that the amount of phosphorylated Akt1 is increased in animals treated with GH (Fig. 4C). Since activation of PI3K-Akt pathway increases levels of PP2A in nuclei (10), we next examined PP2A in nuclear extracts of control and GH-treated mice by Western blotting with antibodies to a catalytic subunit of PP2A, 36-kDa subunit C. As can be seen in Fig. 4D, 36-kDa PP2A subunit C is mainly a cytoplasmic protein in control animals; however, in GH-treated mice, nuclear concentration of PP2A is increased.
To determine if GH-mediated accumulation of PP2A in nuclei dephosphorylates C/EBP␣, PP2A was immunoprecipitated from livers of GH-treated mice and incubated with Ser(P) 193 -C/EBP␣ substrates obtained from different biological systems. The first substrate tested was C/EBP␣, which was overexpressed in 3T3-L1 cells, where it is phosphorylated at

Growth Hormone Corrects Pathways in Livers of Old Mice
Ser 193 (10). The second substrate was the C/EBP␣-Brm complex purified from livers of old mice; C/EBP␣-Ser(P) 193 is the major isoform of C/EBP␣ within this complex (4). As can be seen in Fig.  4E, PP2A IP from livers of old GH-treated mice dephosphorylates these C/EBP␣ substrates at Ser 193 . These data are consistent with the hypothesis that GH-mediated elevation of PP2A disrupts the C/EBP␣-Brm complex through dephosphorylation of C/EBP␣ at Ser 193 . Taken together, our data show that dephosphorylation of C/EBP␣ at Ser 193 is mediated by a change in the balance of cyclin D3/PP2A (Fig. 4F).
C/EBP␣-Brm Complex Inhibits Fox M1B Promoter-We next examined if elimination of the C/EBP␣-Brm complex by GH leads to derepression of genes that control liver proliferation. Several key proteins are required for a proper proliferation of the liver after PH (23). Studies from Costa's laboratory (24) demonstrated that transcription factor FoxM1B is a key positive regulator of liver proliferation after PH. Moreover, experiments with FoxM1B knock-out mice demonstrated that FoxM1B is involved in GH-mediated acceleration of liver proliferation in livers of old mice (16). Therefore, we hypothesized that expression of FoxM1B might be repressed in old animals by the C/EBP␣-Brm complex. To test our hypothesis, we carefully investigated the mouse FoxM1B promoter. Since it had been shown that 300 bp of the FoxM1B promoter are sufficient to drive cell cycle-dependent expression of this gene (25), we cloned 783 bp of the mouse FoxM1B promoter into a reporter vector and examined its regulation by the C/EBP␣-Brm complex. The nucleotide sequence of the proximal 300 nucleotides of the mouse FoxM1B promoter is shown in Fig.  5A. Because the C/EBP␣-Rb-E2F4-Brm complex binds to E2F consensus sequences, we first performed a search for these sequences and found an element (located between Ϫ93 and Ϫ115) (Fig. 5A) that interacts with E2Fs. A typical gel shift assay with a DNA probe containing this E2F site is shown in Fig. 5B. WT and E2F mutant (CGC to TAT) (Fig.  5B, top) were incubated with nuclear extracts from 3T3-L1 cells that contained free E2F4 and E2F4-p130 complexes. The results show that E2F4 and E2F4-p130 complexes interact with the wild type FoxM1B promoter, since antibodies to E2F4 and to p130 supershifted the corresponding complexes. Mutation of the E2F site completely abolished these interactions. We next examined if the identified E2F binding site of the FoxM1B promoter might serve as a negative regulator of transcription. In the course of these studies, we found that the surrounding region also contains a weak site which interacts with C/EBP family proteins and that this sequence acts as a positive element in cultured cells (see Fig. 5A). To distinguish the effect of the C/EBP␣-Brm complex from that of direct binding of C/EBP␣, we generated additional mutants of the FoxM1B promoter, in which C/EBP and E2F sites were mutated. Examination of a basal activity of the WT FoxM1B promoter and activities of these mutants in HEK293 cells is shown in Fig. 5C. Mutation of the E2F site leads to activation of the FoxM1B promoter, whereas the mutation of the C/EBP site reduces activity of the promoter. These studies showed that the E2F element within the FoxM1B promoter is involved in the negative regulation of the promoter.
We next examined the effects of the C/EBP␣-Brm complex on activity of the FoxM1B promoter. Three C/EBP␣ constructs FIGURE 3. Growth hormone eliminates the age-specific C/EBP␣-Brm complex by dephosphorylation of C/EBP␣ at Ser 193 . A, treatment of old animals with GH dissociates the age-specific C/EBP␣-Brm complex. For size exclusion analysis, liver protein extracts from old control and old GH-treated animals were fractionated by size exclusion chromatography, and locations of C/EBP␣ and Brm within gel filtration fractions were determined by Western blotting with specific antibodies. CRM is a cross-reactive molecule (28 kDa), which is detected on C/EBP␣ Western blotting and serves as an internal control for the GH-mediated shift of C/EBP␣. B, GH eliminates the age-specific C/EBP␣-Brm complex. For co-IP examination, C/EBP␣ was immunoprecipitated from livers of old and old GH-treated mice, and Rb, E2F4, and Brm were examined in C/EBP␣ IPs. C, GH does not change protein levels of the components of the age-specific C/EBP␣-Brm complex. Western blotting was performed with nuclear extracts isolated from two control and two GH-treated animals. The bottom shows a Coomassie Blue stain of a separate gel loaded with the same protein extracts. D, GH treatment dephosphorylates C/EBP␣ at Ser 193 . C/EBP␣ was immunoprecipitated from livers of control and GH-treated mice with antibodies to total protein, and these IPs were examined by Western blotting with antibodies to Ser(P) 193 . After strip, the membrane was probed with Abs to total C/EBP␣ and then with secondary Abs to IgG. E, two-dimensional gel examination of C/EBP␣ isoforms. Nuclear extracts from livers of old and old GH-treated mice were separated by two-dimensional gel electrophoresis and probed with antibodies to C/EBP␣. C/EBP␣ control, C/EBP␣ overexpressed in 3T3-L1 cells, in which it is hyperphosphorylated at Ser 193 (10). Positions of Ser(P) 193 isoforms of C/EBP␣ are shown by red arrows. Con, control.
were used to distinguish specific effects of the C/EBP␣-Brm complex (Fig. 5D, top): WT C/EBP␣, which forms a C/EBP␣-Brm complex and binds to DNA; C/EBP␣-S193A, which does not form complexes with Brm; and C/EBP␣-R290A mutant, which forms the C/EBP␣-Brm complex but does not interact with DNA (4,26). Fig. 5D (bottom) shows results of co-transfection of the WT FoxM1B promoter with C/EBP␣ constructs into Brm-positive HEK293 cells. In these cells, WT C/EBP␣ represses wild type FoxM1B promoter, whereas the C/EBP␣-S193A mutant (which does not form a C/EBP␣-Brm complex) activates the promoter presumably through a direct binding to the promoter. The mutation of R290A enhances the ability of C/EBP␣ to repress the FoxM1B promoter. These repressive effects of C/EBP␣ are consistent with the ability of the R290A mutant to form complexes with Brm.
To examine if Brm is required for the C/EBP␣-mediated inhibition of the FoxM1B promoter, the C/EBP␣-R290A mutant was co-transfected with the FoxM1B promoter into Brm-negative C33A cells. As seen in Fig. 5E, C/EBP␣ does not repress the FoxM1B promoter in C33A cells that lack functional Brm. Thus, these studies revealed that C/EBP␣ represses the FoxM1B promoter and that this repression requires both an intact E2F binding site within the promoter and expression of the functional Brm.
Livers of Old Animals Fail to Remove the C/EBP␣-Rb-E2F4 Complex from FoxM1B Promoter after PH-Given inhibition of the FoxM1B promoter by E2F-Rb family proteins and by the C/EBP␣-Brm complex in tissue culture systems, we next examined whether this inhibition occurs in the liver. For this goal, we examined occupation of the FoxM1B promoter by the E2F-Rb and C/EBP␣-Brm complexes in livers of young and old mice using chip assays. Fig. 6A shows results of these studies. In quiescent livers from young mice, E2F1, E2F4, and Rb are observed on the FoxM1B promoter, but in livers of old mice, the pro- Western blotting was performed with nuclear extracts isolated from livers of two control (injected with saline) and two GH-treated animals using antibodies against cyclin D3. Short (30 s) and long (2 min) exposures are shown. The membrane was reprobed with antibodies to ␤-actin and stained with Coomassie Blue to verify protein loading. B, growth hormone reduces the kinase activity associated with cyclin D3. Cyclin D3 was immunoprecipitated from livers of young, old, and old GH-treated animals. The kinase activity of cyclin D3 complexes was examined in an in vitro kinase assay with glutathione S-transferase-Rb substrate. Short (16-h) and long (36-h) exposures for Rb are shown. K6/d3, baculovirus-expressed, purified cyclin D3-CDK6 complex serves as a positive control. NS, a nonspecific band (molecular mass of 33 kDa) that appears in each reaction and serves as an additional control. C, GH activates the PI3K-Akt pathway in livers of old mice. Cytoplasmic extracts from livers of two control and two GH-treated mice were examined by Western blotting with phosphospecific Abs to Akt. The membrane was reprobed with antibodies to total Akt. D, nuclear concentration of a catalytic subunit C of PP2A is increased in livers of GH-treated mice. Cytoplasmic and nuclear extracts from livers of control and GH-treated mice were examined by Western blotting with antibodies to PP2A. The membrane was reprobed with Abs to ␤-actin. 3T3, a positive control; nuclear extracts from 3T3-L1 cells treated with insulin (10). E, PP2A precipitated from the liver of GH-treated mice dephosphorylates C/EBP␣. PP2A was immunoprecipitated with antibodies to catalytic subunit C and incubated with C/EBP␣ expressed in 3T3-L1 cells and with C/EBP␣-Brm complex isolated from livers of old mice. A half part of each substrate was incubated with a mock-agarose control. The samples were separated on two-dimensional gel and probed with antibodies to C/EBP␣. The positions of C/EBP␣ isoforms are shown at the top. F, a hypothetical pathway by which GH eliminates the C/EBP␣-Brm complex (see "Results"). moter is also occupied by the C/EBP␣-E2F4-Rb complex. Although the FoxM1B promoter contains a weak binding site for C/EBP␣, we could not detect association of C/EBP␣ with the FoxM1B promoter in young livers. The presence of C/EBP␣ on the promoter in livers of old mice seems to be mediated through the C/EBP␣-E2F4-Rb complex, since elimination of the complex by GH (which does not change protein levels of C/EBP␣) (see Fig. 3) removes C/EBP␣ from the promoter (see below).
Proliferation of the liver after PH is regulated by a complex cooperation of several networks (23). Since expression of FoxM1B after PH is reduced in livers of old mice (24) but is restored by GH treatment (16), we investigated whether the C/EBP␣-Brm complex might be involved in age-dependent repression of the FoxM1B after PH. Chromatin solutions were isolated from livers of young and old mice at 8 h after PH, and occupation of the FoxM1B promoter was examined by ChIP assay. Fig. 6A shows that in livers of young mice, Rb-E2F complexes are not detectable on the FoxM1B promoter 8 h after PH, whereas C/EBP␣-Brm complexes are observed on the FoxM1B promoter in livers of old mice after PH. Because the C/EBP␣-Brm complex inhibits the FoxM1B promoter in cultured cells, and in old livers transcription of FoxM1B is not increased following PH, these data suggest that the C/EBP␣-Brm complex represses the FoxM1B promoter in livers of old mice.
GH-mediated Elimination of C/EBP␣-Brm Complex Releases C/EBP␣ from E2F Targets, Directs C/EBP␣ to Liver-specific Promoters, and Corrects Transcription of PEPCK-We next asked if GH-mediated disruption of the C/EBP␣-Brm complex might lead to a redirection of C/EBP␣ to liver-specific promoters. To examine this hypothesis, ChIP assays were performed with E2Fdependent c-Myc and FoxM1B promoters and with C/EBP␣dependent PEPCK and HNF6 promoters. As seen in Fig. 6B, GH treatment releases C/EBP␣ from c-Myc and FoxM1B promoters and increases the amount of C/EBP␣ associated with PEPCK and HNF6 promoters. Interestingly, GH also releases Rb from the E2F-dependent promoters, whereas E2F1 and E2F4 remain associated with these promoters. The release of both C/EBP␣-E2F4-Rb and E2F4-Rb complexes from the E2Fdependent promoters suggests that GH removes a negative control of these E2F targets. To determine the amounts of C/EBP␣ on the PEPCK promoter in livers of young, old, and old GH-treated mice, chromatin IPs of C/EBP␣ from these livers were examined by a quantitative PCR. Fig. 6C shows that, whereas in control livers C/EBP␣ is detected on the PEPCK promoter after 28 PCR cycles, GH treatment increased association of C/EBP␣ with the PEPCK promoter so that C/EBP␣ is detectable on this promoter after 24 cycles. Examination of PEPCK and HNF6 transcripts in livers of old mice treated with GH by real time PCR showed a quite different result for these mRNAs. The GH-mediated redirection of C/EBP␣ to the PEPCK promoter correlates with the correction of its transcription (Fig. 6D). The GH-mediated induction of PEPCK transcription in livers of old mice was also detected by Northern blotting (data not shown). These data show that GH-mediated disruption of the C/EBP␣-Brm complex directs C/EBP␣ to the PEPCK promoter and increases its transcription in livers of old mice. On the contrary, GH-mediated redirection of C/EBP␣ to the HNF6 promoter correlates with the 3-4-fold repression of the HNF6 transcripts. This observation is consistent with previous publications showing that C/EBP␣ negatively regulates the HNF6 promoter through a direct binding (9,20). Recent work from Roesler's group (27) suggested that this C/EBP␣-mediated nega-tive regulation involves a recruitment of CA150 protein to the HNF6 promoter. Taken together, the studies of effects of GH on the expression of liver-specific genes showed that GH restores transcriptional activities of C/EBP␣.
Ghrelin-mediated Release of GH in Livers of Old Mice Reduces the C/EBP␣-Brm Complex-Physiologically, GH is released episodically, with major pulses occurring at 3-h intervals. The amplitude of these pulses declines during aging but can be restored by treatment with ghrelin receptor agonists (18). Therefore, we asked whether treatment of old mice with ghrelin is sufficient to eliminate formation of the C/EBP␣-Brm complex in the liver. The liver does not express the ghrelin receptor; therefore, any effect of ghrelin on the C/EBP␣-Brm complex will be mediated indirectly by increased amplitude of endogenous GH release. Old mice were treated with ghrelin, and levels of Brm, Rb, E2F4, and C/EBP␣ and the C/EBP␣-Brm complex were measured by Western blotting and by co-IP. Protein levels of Brm, Rb, E2F4, and C/EBP␣ were unaffected by ghrelin treatment, but like GH treatment, ghrelin reduces cyclin D3 and increases the nuclear concentration of PP2A (Fig. 7A). This result suggested that this ghrelin-dependent event might lead to dephosphorylation of C/EBP␣ at Ser 193 and elimination of the C/EBP␣-Brm complex. Co-IP studies showed that the interaction of C/EBP␣ with cyclin D3 is reduced similarly in ghrelin and GH-treated mice, whereas the interaction of C/EBP␣ with PP2A is increased. Indeed, these interactions caused dephosphorylation of C/EBP␣ at Ser 193 , as shown by Western blotting with Ser(P) 193 -specific Abs (Fig.  7B). Examination of C/EBP␣-Brm complex by co-IP (Fig. 7C) and by gel filtration (Fig. 7D) showed that ghrelin treatment of old mice is sufficient to disrupt the C/EBP␣-Brm complex. The results of real time PCR with RNA from livers of animals treated with ghrelin showed that, like GH treatment, ghrelin increases transcription of PEPCK to the levels observed in young livers (Fig. 7E). Taken together, these studies reveal that ghrelin-induced elevation of endogenous GH in old animals is sufficient to inhibit formation of the C/EBP␣-Brm complex and to correct transcription of PEPCK.

DISCUSSION
Administration of GH or Injection of the Regulator of the Amplitude of Endogenous GH, Ghrelin, Eliminates the Age-specific C/EBP␣-Brm Complex-In this paper, we have examined distribution of C/EBP␣ in protein complexes in livers from FIGURE 6. Growth hormone releases C/EBP␣-Brm (C/EBP␣-E2F4-Rb) complex from the E2F-dependent promoters and increases the interaction of C/EBP␣ with promoters of liver-specific genes. A, C/EBP␣-E2F4-Rb complex occupies the FoxM1B promoter in quiescent livers and in livers from old mice after partial hepatectomy. Chromatin solutions were isolated from quiescent livers (0) and from livers at 8 h after PH from young and old animals. The ChIP assay was performed as described in the legend to Fig. 1. In quiescent young livers, E2F-Rb complexes are detected on the FoxM1B promoter. In old mice, the C/EBP␣-E2F4-Rb complex occupies the FoxM1B promoter in quiescent livers and in livers after PH. B, GH treatment releases the C/EBP␣-E2F4-Brm complex from E2F-dependent promoters. E2F1, E2F4, C/EBP␣, and Rb were immunoprecipitated from chromatin solutions of the livers of young, old saline-treated (control), and old GH-treated mice. These IPs were used for PCR with primers specific to FoxM1B, c-Myc, PEPCK, and HNF6 promoters. Ag, agarose-mock control. C, GH increases interactions of C/EBP␣ with the PEPCK promoter in livers of old mice. C/EBP␣ IPs from chromatin solutions of livers of young, old control, and old GH-treated mice were examined by 22-28 cycles of PCR with primers to the PEPCK promoter. D, GH increases transcription of PEPCK and represses transcription of HNF6 in livers of old mice. Real time PCR was performed with RNA isolated from livers of young and old mice and from livers of animals treated with saline (control) or with GH. young and old mice and found that livers from old mice contain a very small portion of Brm-free C/EBP␣. In complexes with Brm, C/EBP␣ is mainly associated with E2F-dependent promoters, such as c-Myc and FoxM1B, and inhibits the activities of these promoters. Within the sensitivity of our assays, we could not detect the C/EBP␣-Brm complexes on the promoters, to which C/EBP␣ binds directly and activates transcription. Examination of direct targets of C/EBP␣ showed that expression of PEPCK is reduced in livers of old animals. These observations are consistent with previous data showing the reduced expression of PEPCK in old hepatocytes (28).
In seeking pathways that could eliminate the C/EBP␣-Brm complex but not change levels of C/EBP␣, we examined whether GH might disrupt the complex. Our studies revealed that treatment of old animals with GH prevents the formation of the complex without changing production of the protein components of the complex; therefore, elimination of the complex is regulated by post-translational modifications. This observation is consistent with findings that components in the blood of young animals (which contains higher concentrations of GH) reduce the complex by a similar mechanism (2). Previous investigations demonstrated that the C/EBP␣-Brm complex is supported in livers of old mice by a specific phosphorylation of C/EBP␣ at Ser 193 by cyclin D3/CDK4 -6 (4) and that a phosphatase PP2A dephosphorylates Ser 193 and reduces the complex in liver tumors (10). Data in this paper show that both GH and ghrelin treatment of old mice reduces cyclin D3-associated kinase activities and activates the PI3K-Akt pathway, leading to accumulation of the catalytic subunit C of PP2A in nuclei. Although we have injected exogenous GH in concentrations close to endogenous levels of GH in young animals, this replacement of GH in old animals does not precisely mimic the episodic endogenous GH profile measured in plasma. A more physiological approach is to treat animals with an agonist of the GH secretagogue receptor (18,29). Our results with ghrelin-injected mice show that physiologically relevant concentrations of GH restore the young liver phenotype in old mice. The restoration of the young liver proliferative molecular phenotype in old mice by FIGURE 7. Ghrelin-mediated release of GH reduces C/EBP␣-Brm complex in livers of old mice. A, expression of the components of C/EBP␣-Brm complex, cyclin D3, and PP2A in livers of ghrelin-treated mice. Nuclear proteins were loaded on the gel and examined by Western blotting with Abs shown on the left. Protein loading was verified by a reprobe of each membrane with ␤-actin. ␤-Actin controls for E2F4 and cyclin D3 are shown. B, ghrelin-mediated release of GH dephosphorylates C/EBP␣ at Ser 193 . C/EBP␣ was immunoprecipitated with polyclonal Abs and cyclin D3, PP2A, and Ser(P) 193 isoform of C/EBP␣ were examined by Western blotting. C, ghrelin-mediated release of GH eliminates the C/EBP␣-Brm complex. C/EBP␣ was immunoprecipitated from livers of young, old, and ghrelin-or GH-treated old mice, and Brm, Rb, E2F4, and C/EBP␣ were examined in IPs by Western blotting with specific Abs. IgG, heavy chains of immunoglobulins detected by Coomassie Blue. D, examination of C/EBP␣-Brm complex by size exclusion chromatography. Protein extracts from control and ghrelininjected mice were fractionated by gel filtration, and the fractions were probed with Abs to C/EBP␣. E, examination of PEPCK transcripts by real time PCR. Amounts of PEPCK mRNA are shown as ratios to glyceraldehyde-3-phosphate dehydrogenase. F, a hypothetical model suggesting the role of C/EBP␣-Brm complex in GH-mediated alterations in expression of genes in livers of old mice. See "Results" for details.