The phosphoenolpyruvate carboxykinase gene glucocorticoid response unit: identification of the functional domains of accessory factors HNF3 beta (hepatic nuclear factor-3 beta) and HNF4 and the necessity of proper alignment of their cognate binding sites.

Complete induction of hepatic phosphoenolpyruvate carboxykinase (PEPCK) gene transcription by glucocorticoids requires a complex glucocorticoid response unit (GRU). The GRU is comprised of two glucocorticoid receptor (GR)-binding sites (GR1 and GR2) and four accessory factor-binding sites [AF1, AF2, AF3, and cAMP response element (CRE)] that bind distinct transcription factors. Hepatic nuclear factor 4 (HNF4) and chicken ovalbumin upstream promoter transcription factor (COUP-TF) bind to the AF1 element and account for AF1 activity. Members of the hepatic nuclear factor 3 (HNF3) family bind to the AF2 element and provide AF2 activity. In this report, we show that the functions of AF1 and AF2 are dependent on their positions in the promoter, since they cannot substitute for each other nor can they be exchanged without a reduction in the response to glucocorticoids. We also identified the domains of HNF4 and HNF3 beta that are required for the AF1 and AF2 activities, respectively. The carboxy-terminal transactivation domain of HNF4 (amino acids 128-374) confers most of the AF1 activity, while the carboxyterminal transactivation domain of HNF3 beta (amino acids 361-458) mediates AF2 activity. These domains of HNF4 and HNF3 beta appear to have distinct roles in the response to glucocorticoids, as there are unique structural requirements for each, as judged by the failure of most other classes of transactivation domains to serve as accessory factors. These results suggest that the regulation of the PEPCK gene by glucocorticoids requires specific interactions between GR, accessory factors, and coactivators, and that the transactivation domains of AF1 and AF2 are of fundamental importance in the assembly of this multiprotein complex.

Further characterization of the glucocorticoid response unit in the phosphoenolpyruvate carboxykinase gene. The role of the glucocorticoid receptor-binding sites.

Phosphoenolpyruvate carboxykinase (PEPCK) catalyzes the rate-limiting step of gluconeogenesis. The activity of this enzyme is controlled by several hormones, including glucocorticoids, glucagon, retinoic acid, and insulin, that principally affect the rate of transcription of the PEPCK gene. Glucocorticoids induce PEPCK gene transcription through a complex glucocorticoid response unit that consists of, from 5' to 3', accessory factor elements AF1 and AF2; two noncanonical glucocorticoid receptor-binding sites, GR1 and GR2; a third accessory factor element, AF3; and a cAMP-response element, CRE. A complete glucocorticoid response is dependent on the presence of both GR-binding sites, all three accessory elements, and the CRE. In this study we assess the relative roles of GR1 and GR2 in the context of the glucocorticoid response unit and use a combination of binding and function assays to compare GR1 and GR2 to glucocorticoid response elements (GREs) that conform closely to the consensus sequence. The relative binding affinity of GR follows the order: consensus GRE >> GR1 > GR2. Mutations that disrupt the binding of GR to GR1 result in a major reduction of the glucocorticoid response, whereas similar mutations of GR2 have a much smaller effect. Unlike the simple consensus GRE, neither GR1 nor GR2 mediate a glucocorticoid response through a heterologous promoter. The accessory elements appear to have different functional roles. AF2 is still needed for a maximal glucocorticoid response when GR1 is converted to a high-affinity GR-binding element, but AF1 and AF3 are not required.

Hepatic nuclear factor 3 is an accessory factor required for the stimulation of phosphoenolpyruvate carboxykinase gene transcription by glucocorticoids.

Transcription of the hepatic phosphoenolpyruvate carboxykinase gene is stimulated by glucocorticoids and inhibited by insulin. The glucocorticoid response is mediated by a complex glucocorticoid response unit that consists of two glucocorticoid receptor (GR)-binding sites (GR1 and GR2) and two accessory factor-binding sites (AF1 and AF2). The complete unit is required for the full glucocorticoid response. The dominant insulin effect is mediated in part through an insulin response sequence that is coincident with the AF2 element. Members of the hepatic nuclear factor 3 (HNF3) and CCAAT enhancer binding protein (C/EBP) families bind to the AF2 element; however, there is no correlation between binding of these factors and the ability of the AF2 element to mediate an insulin response. We show here that binding of HNF3 does correlate with the stimulation of the glucocorticoid response by the AF2 element and that C/EBP is apparently not involved in this effect. This requirement for HNF3 is quite specific since the substitution of elements known to enhance the action of the GR in other promoters fails to recapitulate AF2 accessory factor activity. By contrast, an HNF3-binding site from the transthyretin gene is able to substitute for the wild type AF2 sequence and elicit a maximal glucocorticoid response. Based on current and previous observations, the glucocorticoid response unit consists of four DNA elements that bind four different proteins. These are: AF1 (hepatic nuclear factor 4/chicken ovalbumin upstream promoter transcription factor), AF2 (HNF3), GR1 (GR), and GR2 (GR).

Expression cloning of a zinc-finger cyclic AMP-response-element-binding protein.

In response to specific extracellular signals, intracellular cyclic AMP levels increase, leading to a variety of responses including the alteration of transcription of many eukaryotic genes. This transcriptional effect is frequently mediated through the cyclic AMP-response element (CRE) motif T(T/G)ACGTCA. Using an expression screening approach we have cloned a yeast gene, MSN2, that encodes a 78 kDa protein that recognizes this consensus CRE motif. Phosphorylation of the MSN2 protein by the catalytic subunit of protein kinase A stimulates DNA binding in vitro. Two putative Cys2His2-type zinc fingers present in the C-terminal 79 amino acids of the MSN2 protein are sufficient to confer CRE-binding specificity. Therefore, MSN2 represents a novel CRE-binding protein distinct from the multiple previously characterized basic region-leucine zipper repeat CRE-binding proteins.

Glucocorticoid receptor inhibits microtubule assembly in vitro.

The effect of glucocorticoid hormones, purified glucocorticoid receptor (GR) and purified heat shock protein M(r) 90,000 (hsp90) on microtubule (MT) assembly in vitro was tested by a spectrophotometric MT assembly assay and electron microscopy. GR significantly prolonged the nucleation phase, slowed down the assembly rate and reduced the maximal amplitude of MT assembly compared with control. The effects were partially reversed by the addition of glucocorticoid hormone. GR associated with MTs. These results indicate that GR affects MT assembly in vitro, which may be a functional correlate to the structural association of GR with MTs. This implies that factors affecting GR may affect MT assembly in vivo.

Characterization of the proximal promoter and two silencer elements in the CYP2C11 gene expressed in rat liver.

The cytochrome P450 gene CYP2C11, expressed in the liver of male rats, is transcriptionally regulated in a dual fashion by the sexually dimorphic secretion pattern of growth hormone. To enable analysis of transcriptional regulatory DNA elements, rat genomic sequences were cloned. DNase I hypersensitivity analysis of rat liver nuclei revealed the existence of two hypersensitive sites whose presence in the vicinity of the transcription start site correlates to high transcriptional activity of the gene. Deletion mutants of the 5' flank were fused to reporter genes and transiently transfected into HepG2 cells or into primary adult rat hypatocytes. Transfection experiments in combination with DNase I footprinting analysis in vitro led to the identification of two negative regulatory regions spanning nucleotides -1,230 to -1,188 and -409 to -368 and designated (SIL1200) and (SIL400), respectively. When placed in front of the heterologous thymidine kinase promoter, SIL1200 and SIL400 reduced the activity of the chloramphenicol acetyl transferase reporter gene to 13% and 23% of the control value, respectively. No sex-dependent binding of liver nuclear extracts to the two silencers could be detected by in vitro footprinting or gel retardation assays. However, a sex-dependent footprint consistently stronger with male liver nuclear extracts than with female extracts was observed in the -320 to -294 region. A significant level of identity was found between the DNA sequence corresponding to this footprint and that of orphan steroid receptor elements as well as with that of a basal transcription element common to several CYP2C genes. However, the identity of a potential trans-acting factor binding between -320 and -294 or response of this element to growth hormone is as yet unknown. A sex- and GH secretory profile-dependent protein-DNA interaction in vitro was observed in the -107 to -95 region. In spite of the sequence similarity that exists between this region and the consensus binding site for HNF-1, this region does not bind HNF-1 alpha. This element acted as a repressor on the heterologous thymidine kinase promoter. To date, the two silencer elements and possibly also the HNF-1-like element are the only functional elements defined in the CYP2C11 gene, and it is conceivable that induction of the gene involves derepression of the silencer elements.

Functional interference between the ubiquitous and constitutive octamer transcription factor 1 (OTF-1) and the glucocorticoid receptor by direct protein-protein interaction involving the homeo subdomain of OTF-1.

The ubiquitous and constitutive octamer transcription factor OTF-1 (Oct 1) is the target of positive regulation by the potent herpes simplex virus trans-activator VP16, which forms a complex with the homeodomain of OTF-1. Here we present evidence that the glucocorticoid receptor can negatively regulate OTF-1 function by a mechanism that is independent of DNA binding. In vivo-expressed glucocorticoid receptor inhibited in a hormone-dependent manner activation of a minimal promoter construct carrying a functional octamer site. Moreover, expression of the receptor in vivo resulted in hormone-dependent repression of OTF-1-dependent DNA-binding activity in nuclear extract. In vitro, the DNA-binding activity of partially purified OTF-1 was repressed following incubation with purified glucocorticoid receptor. Cross-linking and immunoprecipitation experiments indicated that the functional interference may be due to a strong association between these two proteins in solution. Finally, preliminary evidence indicates that the homeo subdomain of OTF-1 that directs formation of a complex with VP16 may also be critical for interaction with the glucocorticoid receptor. Thus, OTF-1 is a target for both positive and negative regulation by protein-protein interaction. Moreover, the functional interference between OTF-1 and the glucocorticoid receptor represents a novel regulatory mechanism in the cross-coupling of signal transduction pathways of nuclear receptors and constitutive transcription factors.

Inhibition of the specific DNA binding activity of the dioxin receptor by phosphatase treatment.

The dioxin receptor stimulates transcription of the cytochrome P-450IA1 gene in response to dioxin. Exposure of the intracellular dioxin receptor to dioxin leads to a rapid conversion of the receptor from a latent form to a DNA binding species which specifically recognizes dioxin-responsive positive control elements in vitro. In this report, we show that treatment of in vivo or in vitro ligand-activated receptor with potato acid phosphatase significantly reduced or abolished its specific DNA binding activity. This effect was inhibited in the presence of sodium phosphate. In control experiments, the ligand-activated glucocorticoid receptor was not inactivated by phosphatase treatment. Moreover, phosphatase treatment did not induce any detectable degradation of covalently labeled dioxin receptor, arguing against protease contamination as a cause for receptor inactivation. Finally, phosphatase-inactivated dioxin receptor exhibited bona fide levels of ligand binding activity. Taken together, these data suggest that phosphorylation may regulate the DNA binding activity of the ligand-occupied dioxin receptor.

The glucocorticoid receptor binds to a sequence overlapping the TATA box of the human osteocalcin promoter: a potential mechanism for negative regulation.

Expression of the human osteocalcin promoter is negatively regulated by glucocorticoids in vivo. In vitro DNase I and exonuclease III footprinting analysis showed binding of purified glucocorticoid receptor in close proximity to and overlapping with the TATA box of the osteocalcin gene. These results imply competition or interference with binding of the TATA box-binding transcription factor IID as a mechanism of repression of this gene by glucocorticoids. In support of this notion, point mutation analysis of the receptor binding site indicated that flanking nucleotides and not the TATA box motif per se were important for receptor interaction. Moreover, DNA binding competition assays showed specific binding of the receptor only to the TATA box region of the osteocalcin gene and not to the corresponding region of an immunoglobulin heavy-chain promoter.

High level expression of functional full length and truncated glucocorticoid receptor in Chinese hamster ovary cells. Demonstration of ligand-induced down-regulation of expressed receptor mRNA and protein.

Full length human glucocorticoid receptor and truncated receptor derivatives lacking the major amino-terminal trans-activating domain were expressed in stably transfected Chinese hamster ovary (CHO) cells. The receptors were co-expressed together with human metallothionein IIa, and the expression levels were amplified in the presence of increasing concentrations of metal. In amplified cells, both synthesized receptor forms showed the expected molecular weights, as assayed by affinity labeling and immunoblotting. They were expressed at concentrations of about 350,000-520,000 molecules/cell which corresponds to a 10-fold increase in receptor levels as compared to rat liver cells. The hormone (agonist or antagonist) binding properties of the expressed proteins were very similar to those characteristic of authentic glucocorticoid receptors in tissues or cultured cells. Moreover, the expressed proteins specifically recognized a glucocorticoid-response element sequence motif in in vitro protein-DNA binding experiments. The activation of a glucocorticoid-responsive reporter gene by the expressed full length receptor was dramatic (about 75-fold) and strictly ligand-dependent. In contrast, the expressed amino-terminal deletion mutant exhibited considerably weaker functional activity but showed normal hormone-binding properties. Upon exposure to dexamethasone in vivo, the expressed receptor mRNAs and proteins were down-regulated about 2- to 6-fold, indicating that regulatory signals important for autoregulation may be contained within structures corresponding to the ligand and DNA-binding domains. Transcription from the expression vector was not negatively regulated from the hormone, strongly arguing that receptor down-regulation was due to a post-transcriptional mechanism. In conclusion, this expression system should be a useful tool for further structural and functional studies of the receptor, including the biochemistry of its activation from a cryptic to a functional species, and its ligand-dependent autoregulation.

The steroid-binding properties of recombinant glucocorticoid receptor: a putative role for heat shock protein hsp90.

The steroid-binding domain of the human glucocorticoid receptor was expressed in Escherichia coli either as a fusion protein with protein A or under control of the T7 RNA polymerase promoter. The recombinant proteins were found to bind steroids with the normal specificity for a glucocorticoid receptor but with reduced affinity (Kd for triamcinolone acetonide approximately 70 nM). Glycerol gradient analysis of the E. coli lystate containing the recombinant protein indicated no interaction between the glucocorticoid receptor fragment and heat shock proteins. However, synthesis of the corresponding fragments of glucocorticoid receptor in vitro using rabbit reticulocyte lystate resulted in the formation of proteins that bound triamcinolone acetonide with high affinity (Kd 2nM). Glycerol gradient analysis of these proteins, with and without molybdate, indicated that the in vitro synthesised receptor fragments formed complexes with hsp90 as previously shown for the full-length rat glucocorticoid receptor. Radiosequence analysis of the recombinant steroid-binding domain expressed in E. coli and affinity labelled with dexamethasone mesylate identified binding of the steroid to Cys-638 predominantly. However, all cysteine residues within the steroid-binding domain were affinity labelled to a certain degree indicating that the recombinant protein has a structure similar to the native receptor but more open and accessible.

Characterization of a complex glucocorticoid response unit in the phosphoenolpyruvate carboxykinase gene.

The minimal DNA sequence required for glucocorticoid induction of the phosphoenolpyruvate carboxykinase (PEPCK) gene in H4IIE rat hepatoma cells was defined. This novel glucocorticoid response unit (GRU) spans about 110 base pairs (bp) and includes two receptor-binding elements plus two accessory factor-binding elements. Purified glucocorticoid receptor bound to two regions (GR1 and GR2) between -395 and -349 bp relative to the transcription start site. Factors in crude rat liver nuclear extract bound to DNA in the regions -455 to -431 and -420 to -403 bp, which are designated accessory factor 1 (AF1) and accessory factor 2 (AF2) elements, respectively. Gel retardation analysis revealed that at least two proteins bound to AF1 and that they were distinct from the protein(s) that bound to AF2. Various combinations of GR1, GR2, AF1, and AF2 were fused to the chloramphenicol acetyltransferase (CAT) reporter gene and cotransfected with a glucocorticoid receptor expression plasmid (pSVGR1) into H4IIE cells to identify the functional GRU. Neither the glucocorticoid receptor binding region nor the accessory factor binding region alone was sufficient to confer glucocorticoid responsiveness. The two components of the glucocorticoid receptor binding region functioned independently, and each accounted for half of the maximal response, provided the accessory factor elements were present. Similarly, deletion of either AF1 or AF2 diminished glucocorticoid induction of the PEPCK gene to approximately half of the maximum. We propose that the complex PEPCK gene GRU provides the stringent regulation required of this critical enzyme in liver.

Radiosequence analysis of the human progestin receptor charged with [3H]promegestone. A comparison with the glucocorticoid receptor.

Partially purified preparations of the human progestin receptor and the human and rat glucocorticoid receptor proteins were covalently charged with the synthetic progestin, [3H]promegestone, by photoaffinity labeling. After labeling, the denaturated protein was cleaved and the mixture of peptides subjected to radiosequence analysis as previously described for the rat glucocorticoid receptor protein (Carlstedt-Duke, J., Strömstedt, P.-E., Persson, B., Cederlund, E., Gustafsson, J.-A., and Jörnvall, H. (1988) J. Biol. Chem. 263, 6842-6846). The radioactivity labels identified, corresponded to Met-759 and Met-909 after photoaffinity labeling of the human progestin receptor, and Met-622 and Cys-754 after labeling of the rat glucocorticoid receptor. The residues labeled in the glucocorticoid receptor are the same as those previously reported to bind triamcinolone actonide. The corresponding residues were also labeled in the human glucocorticoid receptor. Met-759 of the progestin receptor and Met-622 of the rat glucocorticoid receptor are positioned within a segment with an overall high degree of sequence similarity and are equivalent. However, Met-909 (progestin receptor) and Cys-754 (glucocorticoid receptor) do not occur within equivalent segments of the two proteins. Thus, although the two classes of steroid hormone share a common structure within the A-ring, there are subtle differences in their interaction with the two separate receptor proteins.

Structure-function aspects of the glucocorticoid receptor.

The glucocorticoid receptor belongs to a family of ligand activated nuclear receptors. In addition to steroid hormone receptors, this family also includes receptors for thyroid hormone, retinoic acid and 1,25-dihydroxyvitamin D3 as well as some receptors with as yet unknown ligands, the so called orphan receptors. It is possible that the dioxin receptor is a member of this family as well although this cannot be definitely assessed until the receptor has been cloned. Interestingly, the clofibrate receptor, believed to be involved in clofibrate induction of certain isozymes of cytochrome P-450, has recently been shown to be a member of the steroid receptor supergene family and it is conceivable that other P-450 inducers might act via as yet uncharacterized receptors belonging to the same gene family. Therefore, we should learn significantly about principles of P-450 regulation by considering basic mechanisms of glucocorticoid action; the glucocorticoid receptor probably represents the most extensively studied member of the nuclear receptor family. Our laboratory has been particularly interested in structural aspects of this receptor and has used gene technology to overexpress its different domains, particularly the DNA-binding domain, the structure of which has been resolved using NMR.

High level expression in Escherichia coli of the DNA-binding domain of the glucocorticoid receptor in a functional form utilizing domain-specific cleavage of a fusion protein.

A fragment comprising the DNA-binding domain of the human glucocorticoid receptor has been expressed in a functional form in Escherichia coli as a fusion protein with protein A from Staphylococcus aureus. The DNA-binding domain was purified to apparent homogeneity by affinity chromatography on IgG-Sepharose and DNA-cellulose, a purification scheme which does not involve denaturation of the protein at any step. The DNA-binding domain was separated from the protein A part of the fusion protein by domain-specific enzymatic cleavage with chymotrypsin while immobilized on IgG-Sepharose. The recombinant protein has been characterized by amino acid analysis, NH2- and COOH-terminal sequence analysis, sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and reactivity to iodoacetate and was found to correspond to the primary structure derived from the cDNA sequence. DNase I footprinting showed that the purified recombinant protein bound to the same DNA sequences on the mouse mammary tumor virus long terminal repeat as glucocorticoid receptor purified from rat liver does. About 10 times more recombinant protein, on a molar basis, was needed to obtain the same level of protection. However, the protection of the three different footprints (1.3, 1.4, and 1.5') by the recombinant protein differed greatly from that of the natural receptor, with virtually no protection of footprint 1.4. This indicates cooperative binding of the natural receptor to adjacent footprints, dependent on other regions of the receptor than the DNA-binding domain.

DNA binding of glucocorticoid receptor protein A fusion proteins expressed in E. coli.

In an effort to obtain large quantities of glucocorticoid receptor (GR) protein for functional and structural studies, several truncated versions of the human glucocorticoid receptor (hGR) have been expressed in E. coli as C-terminal fusion proteins with protein A. The amount of expressed protein was between 5 and 25 mg/l in the culture. South-Western blotting was initially used to demonstrate the DNA binding capacity of fusion proteins containing the DNA binding domain of GR. The hybrid proteins were highly enriched in the insoluble fraction after cell lysis. For further purification and characterization the fusion proteins were solubilized in 8 M urea. The concentration of denaturing agent was reduced by dilution and the fusion proteins were allowed to refold. The renatured GR protein A fusion proteins bound to DNA in a nitrocellulose filter binding assay. We also show that it is possible to purify the renatured fusion protein to apparent homogeneity using a single chromatographic step on DNA-cellulose.

A progesterone responsive element maps to the far upstream steroid dependent DNase hypersensitive site of chicken lysozyme chromatin.

We have investigated the influence of the 5'-flanking region of the chicken lysozyme gene on steroid dependent gene expression. By transient transfection of lysozyme-CAT fusion genes into the human breast cancer cell line T-47D, a DNA element was identified which stimulates CAT expression when transfected cells are treated with progesterone. This element is distinct from a second hormone responsive element (HRE) located in the lysozyme promoter region; it activates the lysozyme and the TK promoter, irrespective of orientation and distance, and is therefore referred to as hormone responsive element on its own. The location of this newly discovered HRE between -2250 and -1815 relative to the transcriptional start site, corresponds to the position of a steroid inducible DNase I-hypersensitive site in chromatin of oviduct cells. This observation suggests a physiological role for the upstream element. In vitro DNase I protection experiments revealed six binding sites for both progesterone and glucocorticoid receptors within the sequences of the upstream HRE. The three distal binding sites are not required for hormonal stimulation of the TK promoter, while the three proximal binding sites, which are contiguously arranged, work in a cooperative manner.

Identification of hormone-interacting amino acid residues within the steroid-binding domain of the glucocorticoid receptor in relation to other steroid hormone receptors.

Purified rat liver glucocorticoid receptor was covalently charged with [3H]glucocorticoid by photoaffinity labeling (UV irradiation of [3H]triamcinolone acetonide-glucocorticoid receptor) or affinity labeling (incubation with [3H]dexamethasone mesylate). After labeling, separate samples of the denatured receptor were cleaved with trypsin (directly or after prior succinylation), chymotrypsin, and cyanogen bromide. Labeled residues in the peptides obtained were identified by radiosequence analysis. The peaks of radioactivity corresponded to Met-622 and Cys-754 after photoaffinity labeling with [3H]triamcinolone acetonide and Cys-656 after affinity labeling with [3H]dexamethasone mesylate. The labeled residues are all positioned within hydrophobic segments of the steroid-binding domain. The patterns of hydropathy and secondary structure for the glucocorticoid receptor are highly similar to those for the progestin receptor and similar but less so to those for the estrogen receptor and to those for c-erb A.

Domain structure of the glucocorticoid receptor protein.

The purified rat liver glucocorticoid receptor protein was analyzed by limited proteolysis and amino acid sequence determination. The NH2 terminus appears to be blocked. The steroid-binding domain, defined by a unique tryptic cleavage site, corresponds to the COOH-terminal part of the protein with the domain border in the region of residue 518. The DNA-binding domain, defined by a region with chymotryptic cleavage sites, is immediately adjacent to the steroid-binding domain and reflects another domain border in the region of residues 410-414. The results described at the protein level in this report confirm functional data previously obtained by mutations at the genetic level.