In vivo analysis of the model tyrosine aminotransferase gene reveals multiple sequential steps in glucocorticoid receptor action.

We are studying the mechanisms of transcriptional activation by nuclear receptors and we focus our studies on the glucocorticoid regulation of the model tyrosine aminotransferase gene. Rather than using in vitro biochemical approaches, we determine the actual events occurring in the cells. Our experimental approaches include genomic footprinting, chromatin immunoprecipitation, in situ hybridization and transgenic mice. Our results show that the glucocorticoid receptor uses a dynamic multistep mechanism to recruit successively accessory DNA binding proteins that assist in the activation process. Chromatin is first remodelled, DNA is then demethylated, and the synthesis of an accessory factor is induced. Efficient transcription induction is finally achieved upon the formation of a 'stable' multiprotein complex interacting with the regulatory element. We discuss: the relative contribution of histone acetyltransferases and ATP-dependent remodelling machines to the chromatin remodelling event; the nature of the remodelled state; the contribution of regulated DNA demethylation to gene memory during development; the mechanisms of regulated DNA demethylation; the dynamics of protein recruitment at regulatory elements; the control of the frequency of transcription pulses and the control levels of the cell-type specificity of the glucocorticoid response.

Local DNA demethylation in vertebrates: how could it be performed and targeted?

In vertebrates, cytosine methylation is an epigenetic DNA modification that participates in genome stability and gene repression. Methylation patterns are either maintained throughout cell division, or modified by global or local de novo methylation and demethylation. Site-specific demethylation is a rather elusive process that occurs mainly in parallel to gene activation during development. In light of our studies of the glucocorticoid-dependent DNA demethylation of the tyrosine aminotransferase gene, we discuss the potential biochemical mechanisms allowing DNA demethylation and its targeting to specific sequences by transcription factors as well as possible links to DNA replication and chromatin remodelling.

Glucocorticoid-induced DNA demethylation and gene memory during development.

Glucocorticoid hormones were found to regulate DNA demethylation within a key enhancer of the rat liver-specific tyrosine aminotransferase (Tat) gene. Genomic footprinting analysis shows that the glucocorticoid receptor uses local DNA demethylation as one of several steps to recruit transcription factors in hepatoma cells. Demethylation occurs within 2-3 days following rapid (< 1 h) chromatin remodeling and recruitment of a first transcription factor, HNF-3. Upon demethylation, two additional transcription factors are recruited when chromatin is remodeled. In contrast to chromatin remodeling, the demethylation is stable following hormone withdrawal. As a stronger subsequent glucocorticoid response is observed, demethylation appears to provide memory of the first stimulation. During development, this demethylation occurs before birth, at a stage where the Tat gene is not yet inducible, and it could thus prepare the enhancer for subsequent stimulation by hypoglycemia at birth. In vitro cultures of fetal hepatocytes recapitulate the regulation analyzed in hepatoma cells. There fore, demethylation appears to contribute to the fine-tuning of the enhancer and to the memorization of a regulatory event during development.

Identification of 5-methylcytosine in complex genomes.

Cytosine methylation is attracting new attention for regulatory roles in gene expression and there is an increasing interest in detecting, at a single-base resolution, any 5-methylcytosine in genes from complex genomes. Differential base modification by chemicals followed by PCR-based genomic sequencing procedures can provide the resolution, sensitivity, and specificity required for such a goal. The various methods available are not devoid of artifacts but if used carefully and in combination, very reliable information can be obtained. We compare the methods using bisulfite and conventional PCR with those using either hydrazine or potassium permanganate and ligation-mediated PCR and provide a step-by-step description of the corresponding procedures.

Phorbol esters down-regulate alpha-fetoprotein gene expression without affecting growth in fetal hepatocytes in primary culture.

The effects of phorbol esters (phorbol-12,13-dibutyrate, PDB) on alpha-fetoprotein expression and cell growth were assayed by using fetal hepatocytes in primary culture. PDB acts synergistically with epidermal growth factor (EGF) to specifically decrease alpha-fetoprotein (AFP) mRNA levels, without affecting the expression of other genes of the same family, such as albumin and Vitamin D-binding protein (DBP). This effect is PDB-dose dependent, maximal effects being at 10 ng/ml. The implication of protein kinase C (PKC) in this effect seems clear since bisindolylmaleimide (BIS), a specific PKC inhibitor, completely blocks the PDB effect on AFP expression. Nuclear run-on experiments show that the decrease in AFP mRNA levels is mainly due to an inhibition in the transcription rate of the gene. Determination of PKC activities shows that fetal hepatocytes contain mainly Ca(2+)-independent isoenzymes, which patterns of activation was not modified by EGF plus PDB treatment with respect to PDB treatment. We have found that MAPK and JNK activities, c-jun and c-fos mRNA levels and AP-1 binding activity are notably increased when cells are incubated with both EGF and PDB, PDB does not stimulate growth of fetal hepatocytes, measured either as [3H]-thymidine incorporation into DNA or by cell cycle analysis using flow cytometry. All these results suggest that activation of PKC may affect liver gene expression rather than cell growth in fetal hepatocytes.

Chicken ovalbumin upstream promoter-transcription factor, hepatocyte nuclear factor 3, and CCAAT/enhancer binding protein control the far-upstream enhancer of the rat alpha-fetoprotein gene.

We have further characterized the most distal of the three alpha-fetoprotein (AFP) enhancers required for expression of the AFP gene in fetal hepatocytes and yolk sac endodermal cells. Almost total rat AFP enhancer 3 (E3) activity is driven by a 160-bp fragment at -6 kb containing three target regions for nuclear proteins that cooperate to stimulate transcription from the AFP and the thymidine kinase promoters in HepG2 hepatoma cells. Region 1, recently shown to be crucial for correct function of the enhancer in liver of transgenic mice, is recognized by two sets of transcription factors that bind to partly overlapping sites, 1a and 1b, in a noncooperative and nonexclusive manner. Site 1a contains a motif, AGGTCA, which is recognized by chicken ovalbumin upstream promoter transcription factors (COUP-TFs), but not by hepatocyte nuclear factor 4. Hepatocyte nuclear factor 3 (HNF3) and CCAAT/enhancer binding protein (C/EBP), which bind to regions 2 and 3, respectively, are likely responsible for the liver-specific E3 action. They play a key role by acting in synergy. The participation of nuclear receptors such as COUP-TFs, with C/EBP and HNF3, in the tight control of the distal AFP enhancer is a new, and perhaps key, step toward understanding the regulation and function of this enhancer, which may remain active throughout development.

Hepatoma cell-specific expression of a retrovirally transferred gene is achieved by alpha-fetoprotein but not insulinlike growth factor II regulatory sequences.

To target gene expression to malignant hepatic cells, we have constructed recombinant retroviral vectors containing a reporter gene encoding nuclear beta-galactosidase (nls-LacZ) under transcriptional control of regulatory sequences from the rat alpha-fetoprotein (AFP) or human insulinlike growth factor II (IGFII) genes. The AFP and IGFII P3 promoters activate transcription during fetal development and are often reactivated in hepatocellular carcinoma (HCC). Infection of several cultured cell types with the retroviral vector containing the IGFII P3 sequence resulted in expression of the reporter gene in all cell lines tested, including those that do not produce IGFII. In contrast, selective expression was achieved by vectors containing the AFP transcriptional regulatory sequence. Nuclear beta-galactosidase activity was detectable in cells from lines that produce AFP, and not in cells that do not express the AFP gene. In most infected cell lines, retroviral RNA synthesis from the 5' LTR was inhibited, and deletion of the retroviral LTR enhancer did not change expression from either the IGFII P3-nls-LacZ or the AFP-nls-LacZ cassettes. After treatment of cells with 12-O-tetradecanoylphorbol-13-acetate and epidermal growth factor (EGF), the decrease in concentrations of endogenous AFP messenger RNA (mRNA) and nls-LacZ mRNA transcribed from the transferred AFP regulatory sequence were similar. In the context of an integrated provirus, the AFP transcriptional regulatory sequence is therefore subject to similar regulatory control as that of the endogenous gene. These data show that the AFP sequence, and not the IGFII P3 promoter we used, is suitable for targeting gene expression to malignant hepatic cells.

[Several transcription factors participate in the functioning of the alpha-fetoprotein gene promoter]. / Plusieurs facteurs de transcription participent au fonctionnement du promoteur du gène alpha-foetoprotéine.

The oncodevelopmentally regulated alpha-fetoprotein (AFP) gene offers a very good model system to better understand the molecular mechanisms which dictate the specificity of gene expression in liver and control its tight modulation in the course of development and carcinogenesis. Transcription factors of the CCAAT/enhance-binding protein (C/EBP), hepatocyte nuclear factor-1 (HNF-1), and nuclear factor-1 (NF-1) families can bind in vitro to the promoter of the rat AFP gene, which makes the expression of the AFP gene specific to the liver. We have evaluated the influence of some of these factors on the activity of the AFP promoter by transfection of HepG2 hepatoma cells with the appropriate expression vector plus a CAT plasmid under the control of the AFP promoter. A similar plasmid bearing the rat albumin promoter was used as a control. C/EBP alpha, and C/EBP beta acted as transactivators on the AFP promoter, while LIP, a truncated form of C/EBP beta, was a potent negative regulator of the promoter. Interestingly, HNF-1 beta was found to be more potent than HNF-1 alpha in activating the AFP promoter in the HepG2 cells. This effect was highly promoter and cell specific since it did not occur with the rat albumin promoter or in Chinese hamster ovary cells. HNF-1 beta, which is produced earlier than HNF-1 alpha during liver development, would thus have the greater influence on the AFP promoter in early development. Our results pointed to a key role that NF1 might play in the functioning of the AFP promoter. Indeed, overexpression of NF1 induced a specific decrease in the activity of the AFP promoter. Competition between NF1 and HNF-1 for binding to their overlapping binding sites on the AFP promoter would be critical for modulating its activity.

The c-jun proto-oncogene down-regulates the rat alpha-fetoprotein promoter in HepG2 hepatoma cells without binding to DNA.

The effects of a phorbol ester (TPA) and of members of the Jun and Fos oncoprotein family on the activity of the rat alpha-fetoprotein (AFP) promoter were checked by using transient expression experiments in HepG2 hepatoma cells. TPA blocked the activity of the rat AFP promoter in a dose-dependent manner. Overexpression of c-Jun specifically repressed the rat AFP promoter but not the albumin promoter. JunB and JunD were poorer inhibitors. c-Fos expression did not potentiate the negative effect of Jun. The Jun-induced repression does not require binding of c-Jun to the AFP promoter. DNase 1 footprinting experiments did not display any high affinity binding site for Jun on the AFP promoter. Integrity of the c-Jun DNA binding domain is not required for the c-Jun protein to block the AFP promoter. The N-terminal part of Jun, which contains the activating domain, is responsible for the repression as shown by using Jun-Gal4 chimera. Jun likely exerts its negative control on the AFP promoter via protein-protein interactions with a not yet identified trans-activating factor within the -134 to +6 region or with a component of the general machinery of transcription. Jun proteins can thus be key intermediates in regulatory cascades which result in the differential modulation of the AFP and albumin gene expression in the course of liver development and carcinogenesis.

Poly(ADP-ribose) catabolism in mammalian cells.

Poly(ADP-ribose) catabolism is a complex situation involving many proteins and DNA. We have developed an in vitro turnover system where poly(ADP-ribose) metabolism is monitored in presence of different relative amounts of two principal enzymes poly(ADP-ribose) transferase and poly(ADP-ribose) glycohydrolase along with other proteins and DNA. Our current results reviewed here show that the quality of polymer, i.e. chain length and complexity, as well as preference for the nuclear substrate varies depending upon the availability of poly(ADP-ribose) glycohydrolase. These results are interpreted in the light of the recent data implicating poly(ADP-ribose) metabolism in DNA-repair.

Functional analysis of developmentally regulated chromatin-hypersensitive domains carrying the alpha 1-fetoprotein gene promoter and the albumin/alpha 1-fetoprotein intergenic enhancer.

During liver development, the tandem alpha 1-fetoprotein (AFP)/albumin locus is triggered at the AFP end and then asymmetrically enhanced; this is followed by autonomous repression of the AFP-encoding gene. To understand this regulation better, we characterized the two early developmental stage-specific DNase I-hypersensitive (DH) sites so far identified in rat liver AFP/albumin chromatin: an intergenic DH-enhancer site and the AFP DH-promoter site. Mutation-transfection analyses circumscribed the DH-enhancer domain to a 200-bp DNA segment stringently conserved among species. Targeted mutations, DNA-protein-binding assays, and coexpression experiments pinpointed C/EBP as the major activatory component of the intergenic enhancer. Structure-function relationships at the AFP DH-promoter site defined a discrete glucocorticoid-regulated domain activated cooperatively by HNF1 and a highly specific AFP transcription factor, FTF, which binds to a steroid receptor recognition motif. The HNF1/FTF/DNA complex is deactivated by glucocorticoid receptors or by the ubiquitous factor NF1, which eliminates HNF1 by competition at an overlapping, high-affinity binding site. We propose that the HNF1-NF1 site might serve as a developmental switch to direct autonomous AFP gene repression in late liver development. We also conclude that the intergenic enhancer is driven by C/EBP alpha primarily to fulfill albumin gene activation functions at early developmental stages. Factor FTF seems to be the key regulator of AFP gene-specific functions in carcinoembryonic states.

Poly(ADP-ribosyl)ation of chromatin in an in-vitro poly(ADP-ribose)-turnover system.

This paper describes the effect of an in-vitro poly(ADP-ribose) turnover system on the poly(ADP-ribosyl)ation of chromatin. Both poly(ADP-ribose)polymerase and poly(ADP-ribose)glycohydrolase were highly purified and used in 4 different turnover systems: non-turnover, slow, medium and fast turnover. These turnover systems were designed to reflect possible turnover conditions in intact cells. The major protein acceptors for poly(ADP-ribose) are histones and the polymerase itself, a process referred to as automodification. The level of poly(ADP-ribose) modification of polymerase, histone H1 and core histones has been measured. The size of the polymer for each of the 3 groups of acceptor proteins has been determined by gel electrophoresis. After many turnover cycles at medium and fast turnover, the histones (H1 and core) become the main poly(ADP-ribose) acceptor proteins. The rate at which steady-state polymer levels are reached and the total accumulation of polymer in a given turnover system are both inversely proportional to the amount of glycohydrolase present. Furthermore, increasing amounts of glycohydrolase in the turnover systems reduces average polymer size. The polymer synthesized in the medium and fast turnover systems is degraded by glycohydrolase in a biphasic fashion and in these systems the half-life of polymer agreed with results found in intact cells. Our results show that the relative levels of polymerase and glycohydrolase activities can regulate the proportional poly(ADP-ribose) distribution on chromatin-associated acceptor proteins during steady-state turnover conditions. The patterns of modification of polymerase and histones under turnover conditions agree with in vivo observations.

Molecular cloning of two C/EBP-related proteins that bind to the promoter and the enhancer of the alpha 1-fetoprotein gene. Further analysis of C/EBP beta and C/EBP gamma.

In an attempt to identify proteins that may regulate alpha 1-fetoprotein (AFP) gene expression, we screened a cDNA expression library from neonatal rat liver with two essential cis-elements of the AFP promoter and enhancer. We isolated two cDNAs which were found to correspond to leucine zipper proteins of the CC-AAT/enhancer binding protein (C/EBP) family: C/EBP beta and C/EBP gamma. The three related proteins C/EBP alpha, beta and gamma bind with indistinguishable specificity to multiple DNA sites in the promoter and the enhancer of the AFP gene. In addition, C/EBP beta and C/EBP gamma readily heterodimerize with each other as well as with C/EBP alpha. The mRNAs coding for C/EBP beta and C/EBP gamma are expressed in a wider variety of rat tissues than C/EBP alpha mRNA, including yolk sac and fetal liver. The steady-state levels of C/EBP alpha, beta and gamma mRNAs increase during liver development, in parallel with their respective gene transcriptional rates. The high levels of C/EBP beta and gamma mRNAs in rat yolk sac and fetal liver, where C/EBP alpha is poorly expressed, suggest that C/EBP beta and/or gamma could be preponderant or early regulators of the AFP gene in these tissues.

An affinity matrix for the purification of poly(ADP-ribose) glycohydrolase.

The preparation of quantities of poly(ADP-ribose) glycohydrolase sufficient for detailed structural and enzymatic characterizations has been difficult due to the very low tissue content of the enzyme and its lability in late stages of purification. To date, the only purification of this enzyme to apparent homogeneity has involved a procedure requiring 6 column chromatographic steps. Described here is the preparation of an affinity matrix which consists of ADP-ribose polymers bound to dihydroxyboronyl sepharose. An application is described for the purification of poly(ADP-ribose) glycohydrolase from calf thymus in which a single rapid affinity step was used to replace 3 column chromatographic steps yielding enzyme of greater than 90% purity with a 3 fold increase in yield. This matrix should also prove useful for other studies of ADP-ribose polymer metabolism and related clinical conditions.

Cytoplasmic poly(ADP-ribose) polymerase and poly(ADP-ribose) glycohydrolase in AEV-transformed chicken erythroblasts.

Poly(ADP-ribose) polymerase and poly(ADP-ribose) glycohydrolase activities were both investigated in chicken erythroblasts transformed by Avian Erythroblastosis Virus. Respectively 21% and 58% of these activities were found to be present in the post-mitochondrial supernatant (PMS). Fractionation of the PMS on sucrose gradients and poly(A+) mRNA detection by hybridization to [3H] poly(U) show that cytoplasmic poly(ADP-ribose) polymerase is exclusively localized in free mRNP. The glycohydrolase activity sedimented mostly in the 6 S region but 1/3 of the activity was in the free mRNP zone. Seven poly(ADP-ribose) protein acceptors were identified in the PMS in the Mr 21,000-120,000 range. The Mr 120,000 protein corresponds to automodified poly(ADP-ribose) polymerase. A Mr 21,000 protein acceptor is abundant in PMS and a Mr 34,000 is exclusively associated with ribosomes and ribosomal subunits. The existence of both poly(ADP-ribose) polymerase and glycohydrolase activities in free mRNP argues in favour of a role of poly(ADP-ribosylation) in mRNP metabolism. A possible involvement of this post translational modification in the mechanisms of repression-derepression of mRNA is discussed.

Characterization of the poly(ADP-ribose) polymerase associated with free cytoplasmic mRNA-protein particles.

Poly(ADP-ribose) polymerase associated with free cytoplasmic messenger ribonucleoprotein particles (mRNP) has been characterized in mouse plasmacytoma. This cytoplasmic enzyme undergoes auto-ADP-ribosylation and has a similar molecular weight and common antigenic sites with the chromatin bound poly(ADP-ribose) polymerase in spite of its DNA independency. The free mRNP poly(ADP-ribose) polymerase is released from the particle only by high saline concentrations (0.7 M KCl) and the dissociated enzyme expresses a higher activity. The treatment of free mRNP by RNase A stimulates the poly(ADP-ribose) polymerase activity. Partial destruction of mRNP by high saline concentration or mRNA digestion unmasks new protein sites for ADP-ribosylation. In view of the changes that occur in the free mRNP structure to permit mRNA translation, a possible role of poly(ADP-ribosylation) as an important post-synthetic modification of some of the mRNP proteins is discussed.

Adenosine diphosphate ribosyltransferase and protein acceptors associated with cytoplasmic free messenger ribonucleoprotein particles.

ADP-ribosyltransferase activity has been characterized in free messenger ribonucleoprotein particles (mRNP) from mouse plasmacytoma cells. This enzymatic activity appears to be associated with the free mRNP and not due to nuclear contamination. The enzyme activity is not stimulated by added DNA or histone H1 and represents 34 per cent of the total cellular ADP-ribosyltransferase activity while the DNA contamination in free mRNP is less than 4 per cent of the total cellular DNA. Moreover, the ADP-ribosyltransferase specific activity per mg of DNA is about 75-fold higher in free mRNP than in the nuclei. During CsCl gradient centrifugation of the cytoplasmic fraction, the ADP-ribosylated material separates out at a buoyant density similar to that of free mRNP. This ADP-ribosyltransferase activity is inhibited by thymidine, nicotinamide and 3-aminobenzamide, while it is highly stimulated by exogenous pancreatic RNase. The in vitro synthesized acid insoluble material is rendered partly soluble by treatment by a proteolytic enzyme or by snake venom phosphodiesterase resulting in phosphoribosyl-AMP formation: the pancreatic RNase does not solubilize this material. Several ADP-ribosylated proteins are detected by lithium dodecylsulfate gel electrophoresis. Such an ADP-ribosyltransferase activity has also been detected in free mRNP from rat liver. It is suggested that this ADP-ribosylation of specific free mRNP proteins may be associated with free mRNP structure and/or with some chemical covalent type of modification rendering mRNA available for translation.