INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

Steroid hormones, which include glucocorticoids and mineralocorticoids from the adrenal cortex and estrogens, progestins, and androgens from the gonads, regulate a vast array of physiological processes that are essential for development, differentiation, growth, metabolism, homeostasis, behavior, and reproduction in vertebrate organisms. In the classical model of steroid hormone action (1), the steroid ligands bind to specific intracellular protein receptors in target cells. The hormone-receptor complexes interact with particular short nucleotide sequences in the chromosomal DNA and modulate transcription of nearby genes. This model, however, cannot account fully for the complex tissue-specific and gene-specific actions of hormones. Transcriptional induction by steroids is influenced by many factors, such as local chromatin structure (2), stages of the cell cycle (3), cellular morphology or differentiation state (4, 5), specific hormone ligand (6), and physiological state (7). Differential hormone responsiveness depends in part on the availability of other transcriptional regulatory proteins including coactivators and corepressors, which do not themselves bind to DNA (8), and accessory factors, which are DNA-binding proteins (9). For glucocorticoid-regulated genes, several accessory factors have been identified (2, 10-17), but the mechanisms by which these proteins potentiate hormonal activation of transcription are not known.

To understand the role of accessory DNA-binding proteins in determining responsiveness to a steroid hormone signal, we are investigating glucocorticoid induction of fibrinogen gene expression in the liver. Fibrinogen is the precursor of fibrin, the major structural protein of a blood clot, and its synthesis is regulated by adrenal steroids both in basal homeostasis and following physiological stresses such as infections, inflammation, surgery, burns, etc. (18, 19). Using primary liver cells from the frog Xenopus laevis as a model system (20), we have demonstrated that glucocorticoids stimulate transcription of the three separate genes coding for the fibrinogen subunits, termed A, B, and  (21).

Identification of the specific DNA sequences that mediate steroid regulation of the Xenopus fibrinogen genes was accomplished by linking the 5'-flanking DNA of these genes to the firefly luciferase reporter gene and transfecting the DNA into primary Xenopus hepatocytes. Hormonal activation of the B gene occurs through a single element (22) with a close match to the consensus glucocorticoid response element (GRE),¹ GGTACAnnnTGTTCT (23). Full stimulation of the  gene, on the other hand, requires three closely spaced weak binding sites for the glucocorticoid receptor (GR) between nucleotides 168 and 135 relative to the start site of transcription (24). These sites have homology only to the downstream portion of the consensus GRE and therefore are referred to as half-GREs. In addition, steroid responsiveness is affected by bases within the sequence AAGAGTTAA at positions 177 to 169, immediately adjacent to the 5'-most half-GRE (24). This tract is unrelated to the conventional GRE and does not match other known transcription factor-binding sites. A protein present in Xenopus liver nuclei binds to this DNA sequence in vitro. Experiments with DNA containing blocks of mutations within and around bases 177 to 169 showed that the DNA-protein interaction correlates with hormonal activation of transcription. Thus, we named the nuclear protein Xenopus glucocorticoid receptor accessory factor (XGRAF). Compared with other known GR accessory factors, the location of the XGRAF-binding site is unusual since it replaces the sequence that would normally constitute the upstream half of a GRE.

In this work, we describe the detailed mutational analysis of the 9-bp region to which XGRAF binds. This investigation located the 5'- and 3'-boundaries of the recognition sequence and revealed which nucleotides are most critical for DNA binding in vitro and for glucocorticoid-stimulated transcription of transfected DNA in Xenopus primary hepatocytes.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Introduction of Point Mutations into Reporter Gene Constructs-- All possible single-point mutations in the potential XGRAF-binding site in the  gene upstream region were obtained by the following general strategy. 1) The  gene sequence from positions 187 to +41 was synthesized by the polymerase chain reaction (PCR) using a downstream primer with wild-type sequence and an upstream primer consisting of a mixed population of oligonucleotides with mutations in the XGRAF-binding region from positions 177 to 169. 2) The PCR products were inserted into a luciferase reporter vector, and the DNA was cloned by transformation into bacteria. 3) The nucleotide sequence of the  DNA in individual clones was determined to ascertain which nucleotide(s) of the XGRAF region had been mutated.

Gel Shift Assays-- Both the probe and the competitors for the gel shift assays contained the  sequence extending from positions 187 to 115 and were generated by PCR using Pfu polymerase. In addition, the molecules included 19 bases of vector sequence upstream of position 187 and a MfeI restriction enzyme site downstream of position 115. Wild-type pLL187 was used as the template for making the probe, and the point mutation constructs described above were the templates for the competitors. The PCR products were purified through 2% low-melting-temperature agarose gels (26), and the probe was 5'-end-labeled (28).

Data Analysis for Gel Shift Assays-- The ability of XGRAF to bind  DNA with mutations in the putative XGRAF-binding region was determined by competition gel shift assays, which contained a constant amount of radioactively labeled wild-type  DNA and either no competitor or various concentrations of DNA with a single-point mutation. An example of the gel shift assay with mutant A¹⁷¹ as the competitor is shown in Fig. 1A. Radioactivity in the shifted DNA·XGRAF complex in each lane was quantitated from the phosphorimaging scan. The total amount of XGRAF in the complex with radioactive DNA in the absence of any competitor (Fig. 1A, lane 2) was defined as 1.0, and the amount of XGRAF in the complex with radioactive DNA in the presence of a 100-fold molar excess of wild-type competitor was defined as zero (Fig. 1A, lane 1). The quantity of XGRAF remaining in the complex with radioactive DNA in the presence of mutated competitor (Fig. 1A, lanes 3-17) was expressed as a fraction of 1.0. The difference between the total and the fraction still bound to the radioactive probe equaled the fraction of XGRAF bound to the competitor DNA.

View larger version (39K): [in this window] [in a new window]   Fig. 1.   Quantitation of binding of XGRAF to  DNA with a single-point mutation at position 171. A, gel shift assay. The binding of XGRAF in Xenopus liver nuclear extract to a DNA fragment spanning positions 187 to 115 of the -fibrinogen subunit gene upstream region was analyzed. The reactions contained the wild-type (WT) DNA sequence as the radioactive probe and the following competitors: lane 1, 100-fold molar excess of wild-type DNA; lane 2, no competitor; lanes 3-17, three replicates (labeled sets A, B, and C) of  DNA from positions 187 to 115 with the single-point mutation (Mut) A171 at 50-, 100-, 200-, 300-, and 500-fold molar excesses. B, Scatchard analysis. The data from A were plotted according to Equation 3 (see "Experimental Procedures"). Bound XGRAF on the x axis corresponds to [DNA·XGRAF], and Bound XGRAF/Competitor DNA on the y axis corresponds to [DNA·XGRAF]/[DNA]t in Equation 3. Bound XGRAF is expressed as a fraction of total XGRAF, and competitor DNA is expressed as the -fold excess over the radioactive probe.

(Eq. 1)

(Eq. 2)

(Eq. 3)

Methylation Interference Footprinting-- DNA probes containing  gene upstream sequence from positions 232 to 6, ³²P-labeled on the sense strand, and from positions 232 to 115, ³²P-labeled on the antisense strand, were produced by PCR with 0.1 µM primers and Taq enzyme (24). The PCR products were purified through a native 6% polyacrylamide gel (26) followed by organic extraction. The end-labeled DNA fragments (~1 × 10⁶ cpm/pmol) were partially methylated at the guanine moieties by incubation for 2-3 min with dimethyl sulfate without carrier (30); in some cases underwent organic extraction; were precipitated twice with ethanol; and were dissolved in 10 mM Tris-HCl and 0.1 mM EDTA (pH 8.0). For preparative binding, the reaction volumes described above for the gel shift assay were scaled up 15-30-fold, with ~300,000 cpm of DNA probe, and electrophoresis was carried out at 250 V for 2.5-3.5 h at 4 °C. Wet gels were exposed to XAR film at 4 °C overnight; portions of the gel containing bound and unbound DNA were excised; and DNA was eluted by shaking overnight once or twice at 37 °C in 0.2 M NaCl, 10 mM Tris-HCl (pH 7.5), and 0.1 mM EDTA. The eluted DNA was purified over a NACS column (Life Technologies, Inc.) according to the manufacturer's instructions, and 8-16 µg of carrier yeast RNA was added. After ethanol precipitation, piperidine cleavage was performed (31), and piperidine was removed (24). The entire recovered bound DNA fraction and an approximately equal amount of radioactivity of the unbound DNA were analyzed by gel electrophoresis as described previously for methylation protection footprinting (24). The gels were exposed to XAR film at 80 °C with one Lightning Plus intensifying screen.

Transfection and Assays for Reporter Gene Activity-- Isolation of hepatocytes from three adult female X. laevis frogs and transfection were as described previously (24), except that 10 µg of control plasmid pCMV-gal (32) was used, and electroporation was conducted at a setting of 130 V. The cells were divided equally and incubated with or without 10⁷ M dexamethasone and 10⁹ M triiodothyronine in 12-well Primaria plates (Falcon) with 1 × 10⁶ cells in 4 ml of medium. After 44-48 h, cell extracts were prepared, and the activities of -galactosidase (using the Galactolight method) and of luciferase were assayed as described (24).

Data Analysis for Transfection Assays-- Luciferase values were normalized to the -galactosidase activity in each transfection. -Fold hormonal induction was calculated by dividing the amount of luciferase activity/-galactosidase activity in hormone-treated cells by the amount of luciferase activity/-galactosidase activity in untreated cells. The data for each mutant are expressed relative to wild-type DNA as follows. XGRAF activity with wild-type DNA is defined as F_WT  F_D, where F_WT represents the -fold increase in transcription in response to hormone treatment for the wild-type construct pLL187 and F_D represents the -fold increase in transcription in response to hormone treatment for the mutation D construct, which has multiple mutations in the XGRAF site. Mutation D (¹⁷⁵ACT¹⁷³) abolished the XGRAF-binding site, so the remaining -fold response is due solely to the half-GREs (24). For each single-point mutant, the percent of wild-type XGRAF activity was calculated as ((F_M  F_D)/(F_WT  F_D)) × 100, where F_M is the -fold increase in transcription in response to hormone treatment for each mutant. In a total of eight experiments, the average F_WT was 5.3 ± 0.6 (mean ± S.E.), and the average F_D was 2.2 ± 0.2 (mean ± S.E.). Each mutant was tested in three independent experiments.

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Identification of Specific Guanosines Involved in XGRAF Binding-- Previously, we mapped the DNA elements between 177 and 135 bp upstream of the transcription start site that are important for glucocorticoid regulation of the Xenopus -fibrinogen subunit gene (24). Three sites designated half-GRE1 (positions 168 to 163), half-GRE2 (positions 156 to 151), and half-GRE3 (positions 140 to 135) in Fig. 2C have a close match to the downstream half of the consensus GRE, bind to the DNA-binding domain of GR in vitro, and contribute to steroid-induced transcription in primary hepatocytes. Half-GRE1 is the most critical of the three GREs for hormonal induction. In addition, the tract from positions 177 to 169 is necessary for full hormone responsiveness even though it does not match the consensus GRE. The nuclear protein XGRAF binds to this region of the DNA (Fig. 2C).

View larger version (56K): [in this window] [in a new window]   Fig. 2.   Identification of guanosines in the -fibrinogen subunit gene upstream regulatory region that contact XGRAF. A, sense strand methylation interference. The assay used  DNA from positions 232 to 6 synthesized by PCR with an end-labeled sense strand primer. Lane G+A, DNA sequencing ladder; lane B, protein-bound DNA; lane F, free DNA. Numbers indicate genomic sequence positions. Arrowheads mark the bands with reduced intensity in the protein-bound DNA fraction compared with free DNA (lane B versus lane F). B, antisense strand methylation interference. The conditions were the same as described in A, except that the probe was  DNA from positions 232 to 115, 32P-labeled on the antisense strand. C, the nucleotide sequence of the  upstream region from positions 187 to 130. Arrowheads at positions 184, 175, and 173 indicate those guanosines at which methylation inhibited binding of XGRAF. The squares above the sequence indicate 10-bp intervals. All the elements that contribute to glucocorticoid responsiveness of the  gene are indicated below the sequence: the XGRAF-binding region between positions 177 and 169 and the three half-GREs (positions 168 to 163, 156 to 151, and 140 to 135).

175

173

184

Effect of Point Mutations on XGRAF Binding Ability-- The relative contribution of each nucleotide to the specific interaction between XGRAF and the  DNA was assessed by saturation mutagenesis of the site from positions 177 to 169, generating all 27 single-point mutants in the 9-bp sequence. The effect of the mutations on DNA binding in vitro was analyzed by the gel shift assay, using the mutated DNA sequences as competitors for binding of XGRAF to radioactively labeled wild-type DNA, as described under "Experimental Procedures." Binding ability is expressed as C₅₀, the -fold excess of competitor required to displace half of XGRAF from the wild-type probe. For each mutant, the C₅₀ value was calculated in three independent experiments, and the results are presented as the mean ± S.E. of the three determinations (Fig. 3 and Table I). The wild-type DNA had a C₅₀ value of 1.7-fold excess.

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Relative ability of single-point mutants of the  DNA from positions 177 to 169 to bind XGRAF. Using the gel shift assay and Scatchard analysis described under "Experimental Procedures" and exemplified in Fig. 1, the ability of each mutant to bind to XGRAF was quantitated. The wild-type (WT) nucleotides at positions 177 to 169 and their corresponding single-point mutations (mut) are indicated along the y axis. C50 on the x axis represents the -fold excess of mutant DNA required as competitor to displace half of XGRAF from the radioactively labeled wild-type DNA probe. The data are shown as the mean ± S.E. of values determined in three independent experiments. For each mutant, three independently produced preparations of competitor DNA and at least two different batches of nuclear extract were used (except for mutant A172, for which the data were derived from two experiments with two different preparations of competitor and one nuclear extract).

View this table: [in this window] [in a new window]   Table I Relative ability of the 27 single-point mutants at positions 177 to 169 of the  gene to bind XGRAF, to enhance glucocorticoid induction of transcription, and to regulate basal transcription

175

171

173

169

Effect of Point Mutations on Glucocorticoid Responsiveness-- We also examined the effects of the point mutations in the XGRAF-binding site on glucocorticoid induction of transcription. The mutated  gene upstream region was inserted into a luciferase reporter vector, and the constructs were transfected into primary Xenopus hepatocytes. Transfected cells were divided for plus or minus glucocorticoid treatment for 2 days, and lysates were analyzed for luciferase activity. The total -fold increase in luciferase levels in hormone-treated cells reflects the role of not only the XGRAF site, but also the three half-GREs in the upstream regulatory region of the  gene (Fig. 2C). As described under "Experimental Procedures," the effect of the mutations only on the XGRAF contribution to the hormonal stimulation was assessed by comparing each single-point mutant with wild-type DNA, representing 100% induction, and with a triple mutant in the XGRAF-binding site, which eliminated XGRAF binding and therefore represented 0% activity of XGRAF in the induction. The triple mutant, which changed ¹⁷⁵GAG¹⁷³ to ¹⁷⁵ACT¹⁷³, has been shown previously to reduce glucocorticoid responsiveness through effects on the XGRAF-binding site rather than the GRE (24). Retention of at least 60% of XGRAF activity was considered normal function, whereas activity below 47% was defined as impaired.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Relative ability of single-point mutants of the  DNA from positions 177 to 169 to enhance glucocorticoid induction of transcription. Transfection of Xenopus primary hepatocytes with constructs containing mutations in the XGRAF-binding region of the  DNA and maintenance of the cells with or without hormone treatment are described under "Experimental Procedures." The wild-type (WT) nucleotides at positions 177 to 169 and their corresponding single-point mutations (mut) are indicated along the y axis. The data are reported as XGRAF activity, which is defined as the portion of the hormonal induction attributable to XGRAF (rather than GR) for each mutant compared with that of wild-type DNA (see "Experimental Procedures"). The values are the mean ± S.E. of three independent experiments. The value obtained in each experiment was the average of usually triplicate, and in a few cases duplicate, measurements.

172

171

Effect of Point Mutations on Basal Transcription-- The level of transcription of the single-point mutants in the absence of hormone treatment ranged from 48 to 142% of that of wild-type DNA (Table I). Each data point for basal expression was obtained from an independent transfection event, whereas changes due to hormone treatment were assessed on a single cell population that was divided after the transfection.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

Correlation between XGRAF Binding to DNA in Vitro and Stimulation of Transcription-- In Table I, the ability of DNA with single-point mutations at positions 177 to 169 to bind to XGRAF in the gel shift assay is classified as strong (+) if half-maximal competition was achieved with <7-fold molar excess of mutated competitor over wild-type probe or as weak () if 20-fold or greater excess competitor was needed. Similarly, in Table I, the ability of DNA with each single-point mutation to enhance glucocorticoid-induced transcription in transfected primary hepatocytes is labeled as high (+) if at least 60% of wild-type XGRAF activity was retained or low () if 46% or less of wild-type XGRAF activity was observed. With only four exceptions, which will be discussed below, high activity in the transfection assay correlated with strong binding ability in the gel shift assay, whereas low functional activity corresponded with weak binding of XGRAF to the mutated DNA in vitro. This correlation can be seen clearly in a plot of XGRAF activity as a function of C₅₀ (Fig. 5). The excellent agreement between binding and function firmly supports the conclusion that XGRAF, defined by its interaction with DNA in vitro, is the same protein that plays a role in hormonal stimulation of gene transcription in vivo.

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Correlation between the ability of the single-point mutants in region 177 to 169 to amplify hormone-induced transcription and to bind XGRAF. The portion of the glucocorticoid response attributable to XGRAF, expressed as XGRAF activity with mutant DNA as a percentage of that with wild-type DNA, is plotted as a function of the ability of the DNA to bind XGRAF in vitro, expressed as C50. See "Experimental Procedures" and "Discussion" for details.

173

173

177

163

173

172

171

169

169

169

169

169

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Correlation between the ability of the single-point mutants in region 177 to 169 to stimulate basal transcription and to bind XGRAF. Basal transcription, the luciferase activity/-galactosidase activity in untreated cells transfected with mutant DNA expressed as a percentage of that with wild-type DNA, is plotted as a function of the ability of the DNA to bind XGRAF in vitro, expressed as C50. See "Experimental Procedures" and "Discussion" for details.

Consensus Sequence for the XGRAF Recognition Site-- Based on the physical and functional data presented here, a consensus sequence can be derived for the XGRAF-binding site that reflects the most favorable nucleotide at each position. Bases 177 and 176 are no longer considered part of the recognition sequence since no substitutions at these positions affected XGRAF binding or activity. For nucleotides 175 to 169, the consensus sequence is BNGTTAA (B = C, G, or T; N = A, C, G, or T). Even with this more well defined site, we found no striking matches to recognition sequences in the transcription factor site data base TRANSFAC 3.2 (34) using the TESS searching program (36).² Thus, XGRAF appears to have a novel sequence specificity for binding to DNA.

Models for Interaction of XGRAF and GR with Contiguous or Overlapping Binding Sites-- Previously, we presented four possible models for the interaction of XGRAF and GR with DNA at closely juxtaposed sites (24). Model 1 depicted simultaneous binding of XGRAF to its site at nucleotides 175 to 169 and GR binding as a dimer at nucleotides 177 to 163, which would constitute a full-length GRE. To bind these sites concurrently, XGRAF must contact the DNA in the minor groove since GR is known to occupy the major groove (35). However, the methylation interference experiment (Fig. 2) established that XGRAF also binds in the major groove since modification of the N-7 positions of guanines, which are accessible only in the major groove, interfered with binding. Therefore, Model 1 is not a likely mechanism for interaction of GR and XGRAF with their respective sites.