The Rat Glucocorticoid Receptor Mutant K461A Differentiates between Two Different Mechanisms of Transrepression*

The glucocorticoid receptor (GR) can both activate and repress transcription of target genes by interaction with specific genomic response elements, glucocorticoid response elements (GREs). Activation of transcription is usually the result of the direct interaction between GR and the GRE, whereas GR-mediated transcription repression is either the result of the indirect action of GR, mediated by a response element as a result of protein·protein interaction or by an occlusion mechanism in which GR displaces a general or regulatory transcription factor. A specific mutation of rat GR, K461A, has previously been described to transform the indirect protein·protein interaction-dependent transrepressive effect of GR into an activating function (Starr, D. B., Matsui, W., Thomas, J. R., and Yamamoto, K. R. (1996) Genes Dev. 10, 1271–1283). In HOS D4 and COS7 cells, this mutation was shown to transform the transrepressive effect of wild-type GR, acting on reporter constructs containing the composite GRE from the proliferin gene (plfG) or the negative tethering GRE from the collagenase A promoter (colA), into an activating function. In contrast, the K461A mutation had no effect on the transrepressive effect of GR on the human osteocalcin gene in which repression apparently occurs through the binding of GR to a negative GRE that overlaps the TATA box. The transrepressive function, typically 40% of the basal level in the absence of hormone, required only the isolated DNA-binding domain of wild type or mutant GR and was independent of the nature of transactivation domain. Thus, mutation of rat GR at position 461 differentiates between transrepressive functions of GR dependent on GR·DNA interaction (repression by occlusion) and GR·protein interaction (active repression).

Negative regulation of gene transcription by glucocorticoids and other ligands for nuclear receptor proteins appears to be carried out either by mechanisms involving interference with the DNA binding of upstream or general transcription factors or, alternatively, by repression mechanisms independent of DNA binding. In the first case, transcription repression involves a competition between transcriptionally active and inactive proteins for common or overlapping DNA-binding sites. Thus, transcription repression is achieved when the inactive factor displaces the more active one. For the latter mode, transcriptional repression is achieved by an as yet unidentified mechanism that is postulated to involve a physical interaction between the two proteins in a DNA-independent manner and, presumably, the formation of a transcriptionally inactive complex.
The signal transduction pathway of glucocorticoid hormones provides a number of well described examples for which possible mechanisms involved in negative gene regulation have been postulated. Signal transduction is mediated by an intracellular receptor protein, that, like many other members of the steroid receptor superfamily, functions as a ligand-activated nuclear transcriptional regulator (1,2). The classical mode of gene regulation by glucocorticoids, which accounts for most cases of positive gene regulation, is known to be mediated by interaction of the ligand-activated GR 1 with positive control elements (glucocorticoid response elements; GREs) which are present in single or multiple copies upstream of or within target genes (3).
Although not completely understood, the mechanism by which GR activates transcription in response to glucocorticoids is fairly simple in comparison to the variety of mechanisms employed by nuclear receptor proteins for the negative modulation of gene transcription (4 -6). Despite recent advances, it has not been possible to formulate a simple, all inclusive model accounting for negative gene regulation mediated by GR (7,8), although several physiologically relevant models have been suggested for the small number of genes studied mechanistically in a detailed manner. It has become apparent that simple DNA binding by the receptor may not be sufficient or, in some cases, even required for ligand-dependent repression for the majority of examples described for glucocorticoid-mediated transcription repression, which are related to their anti-inflammatory effects (9 -11). One of the best characterized examples of this mode of repression is the interaction between the transcriptional activator AP1 and the activated GR. Under certain conditions, interference between the GR and the AP1 signal transduction pathways appears to occur on composite response elements that have the potential to bind both the receptor and the AP1 complex, whereas in other cases, the repression mechanism requires only an AP1⅐DNA interaction (18,19). An ex-* This work was supported in part by grants from the Swedish Medical Research Council (2819), Åke Wibergs Foundation, and the National Board for Industrial and Technical Development (93-03522), and by grants from the National Science Foundation and the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ ample of the first case is seen within the promoter region of the proliferin gene. A 25-bp composite GRE, termed plfG, is responsible for mediating the negative GR effect. GR binds to plfG in the absence of AP1, but regulates transcription only in the presence of AP1, activating if AP1 consists of c-Jun homodimers, and repressing if AP1 is comprised of c-Jun⅐c-Fos heterodimers (19,20). An example of the alternative AP1-dependent mechanism is the glucocorticoid-dependent repression of the collagenase gene. This effect requires AP1 binding to its specific recognition element in the upstream promoter region of the collagenase gene, whereas direct GR⅐DNA contact does not seem to be a prerequisite for efficient ligand-dependent repression of the activated gene expression level. Although the GR does not bind to the AP1 site, a functional GR DBD is required for repression of AP1 activity. A direct protein⅐protein interaction between GR and AP1 has been demonstrated (18,21).
There are some cases, however, where DNA binding by GR is both necessary and sufficient for transcription repression. For example, at the pro-opiomelanocortin gene, repression occurs without assistance or interference from other sequence-specific transcription factors (12,13). Detailed analysis of the interaction between GR and the GRE indicated that three moieties of the receptor molecule form a unique complex with the proopiomelanocortin GRE, in contrast to a positive regulated GRE which interacts with the receptor protein as a homodimer (14). Examples of repression due to transcriptional interference with other regulatory proteins at a response element have been postulated in several cases including the bovine prolactin and the c-fos genes (15)(16)(17). A final example of a gene negatively regulated by glucocorticoids through an interference mechanism is the human bone-specific gene osteocalcin (22)(23)(24). The overlap of a competitive nGRE with the basal TATA box element suggested that the hormone-activated GR can function as a negative regulator on osteocalcin gene activity by competing with a specific TFIID-induced complex at the DNA binding level and that this binding is mutually exclusive (25,26).
GR action is strongly determined by its context within the unique architecture and requirements of each gene promoter (27,28). As demonstrated by the above described examples of glucocorticoid-dependent transcriptional regulation, GREs can be classified into at least three independent subclasses: simple GREs capable of interacting with the hormone-induced receptor without the assistance of other sequence-specific regulators, resulting in transactivation or, in more specialized cases, repression; composite GREs having the ability to interact with the receptor protein and additional factors resulting in either transactivation or transrepression; and cases in which direct GR⅐DNA interaction is not required for GR-mediated gene regulation, called tethering GREs (18,21). Starr et al. (29) defined a single amino acid change (K461A) within the rat GR capable of distinguishing between simple GREs and composite and tethering GREs. The receptor, containing a mutation within the DBD at position 461 (human GR 442), provides a powerful tool for comparing and defining mechanisms involved in transmission of negative regulation by GR.
To further characterize the mechanism proposed for the repression of the human osteocalcin gene by glucocorticoids, we compared the action of this specific mutated receptor, chimeric proteins containing the GR DBD as well as the isolated GR DBD on both the osteocalcin promoter and the two well defined AP1-dependent systems described above, colA and plfG. In this report we present evidence that the mechanism involved in the negative transcriptional effect mediated by glucocorticoids on the human osteocalcin promoter is strictly dependent on binding of GR to a composite functional GRE (competitive nGRE). DNA binding and cotransfection experiments suggest that the hormone activated GR and the specific GR mutant K461A repress osteocalcin gene activity in a similar fashion, primarily mediated by competitive binding to a dual binding site that disrupts an alternative protein⅐DNA contact.
Cell Culture-HOS D4 osteosarcoma cells were cultured at 37°C in a humidified atmosphere with 5% CO 2 in Eagle's medium buffered with bicarbonate and supplemented with 5% fetal calf serum, penicillin (100 IU/ml), and streptomycin (0.1 mg/ml). COS7 cells were cultured in Dulbecco's modified Eagle's medium supplemented as described above.
Transient Transfection Assays and Luciferase Assay-Cells were seeded in 6-cm plates 24 h before a transfection experiment and transfected at 50 -60% confluence using the calcium phosphate coprecipitation technique. The precipitate contained 5 g of supercoiled luciferase reporter plasmid DNA and varying amounts (0 -2 g) of different expression plasmids. The overall amount of DNA was kept constant by the addition of parent expression vector. After 12-14-h exposure to the calcium phosphate precipitate, medium was refreshed and cells treated for 24 h with 20 nM dexamethasone. Transfected cells were subsequently harvested for luciferase assay by scraping the cells into 1 ml of phosphate-buffered saline, cenrtrifuging for 10 min in a microcentrifuge, and resuspending in 50 l of lysis buffer (25 mM Tris acetate, pH 7.8, 2 mM dithiothreitol, 1.5 mM EDTA, 10% glycerol, and 1% Triton X-100). Luciferase activity was monitored according to the GenGlow luciferase assay kit (Bio Orbit) using an Anthos Lucy 1 luminometer. The results are expressed as light units measured. All experiments were performed in triplicate on three separate occasions.
Protein Expression-Extracts from COS7 cells, transiently transfected with 15 g of GR expression vector/15-cm cell culture plate, were prepared by homogenizing the cell pellets with a Dounce homogenizer in 500 l of 10 mM sodium phosphate, pH 7.4, 1 mM EDTA, 0.5 mM dithiothreitol, 10% glycerol, 400 mM KCl, and centrifugation at 100,000 ϫ g for 1 h. The supernatant was aliqoted and stored at Ϫ70°C.
DNA-binding Assays-GR binding activity was monitored by an electrophoretic gel mobility shift assay. A 32 P-labeled, double-stranded oligonucleotide spanning the GRE sequence and TATA box of the human osteocalcin promoter (Ϫ41/Ϫ9) or mutated versions of this DNA stretch were used as specific probe (wt: AGCCCAGAGGGTATAAA-CAGTGCTGGAGG, mutant: AGCCCAGAGGGTgTAAACAGTGCTG-GAGG). The recombinant GR was incubated for 10 min on ice in a buffer containing 0.5 g of poly(dI⅐dC), 60 mM KCl, 10 mM Hepes, pH 7.9, 0.1 mM EDTA, 10% glycerol, 5 mM dithiothreitol. Competing oligonucleotides were incubated with the binding reactions for 10 min prior to addition of the 32 P-labeled probe. After adding the specific DNA probe the mixture was incubated for 20 min at room temperature. The protein-DNA complexes were resolved on 5% native polyacrylamide gels.

RESULTS
GR can antagonize the function of c-Jun and c-Fos, which are both components of the phorbol ester-activated transcription factor AP1 (8). We employed two well defined examples for repression of AP1-activated gene activity by GR to further characterize the mechanism of glucocorticoid mediated repression of the human osteocalcin gene in comparison. In all three examples of gene promoter regions used in this study, the negative modulation of target gene activity seems to be medi-ated by interference of hormone activated GR protein with positive acting sequence-specific transcription factors. The experimental basis for this comparative study was based on a recently published GR mutant having the ability to activate genes independent of the class of GRE used (29). In this mutant, an amino acid change from Lys to Ala at position 461 within the DBD of the rat GR resulted in a phenotype characterized by hormone-dependent transactivation at simple GREs as well as at composite and tethering GREs. As reference systems in the present study, we used either a promoter construct containing three copies of a 25-bp DNA sequence (composite GRE) from the proliferin promoter (plfG) (19) known to mediate a negative GR effect or an nGRE (tethering GRE) from the the collagenase promoter here denoted as colA (31). In contrast to the wt GR, which represses transcription from these two reporters, the GR mutant K461A activates transcription (25,26,29).
We were interested in studying the effect of this constitutively positive acting receptor variant on the osteocalcin promoter-controlled transcription rate where the interdigitation of a GR binding site and the TATA box suggest that repression of osteocalcin gene expression is mediated by interference of the GR with the basal transcription machinery. To this end, we constructed a reporter plasmid containing a fragment of the human osteocalcin promoter spanning nucleotides Ϫ344/ϩ31 driving the firefly luciferase gene. This reporter construct was introduced into the human osteosarcoma cell line HOS D4 or, alternatively, into COS7 cells. The endogenous GR in these cell lines are expressed at very low levels and are hardly detectable by ligand binding or immunochemical assays (32). Strikingly, the cotransfection of HOS D4 cells with either wt GR or the mutated receptor (K461A) together with the above described osteocalcin reporter construct resulted, after induction with the synthetic glucocorticoid dexamethasone, in a clear repression of the luciferase activity to 50% (Fig. 1A). To exclude the influence of possible cell-specific effects on GR function, we repeated the experiments in the non-bone cell line COS7 and found again that the receptor variant K461A behaved in a similar fashion as the wt GR. Both hormone-activated GR variants had the capacity to repress basal osteocalcin gene activity (Fig. 1B).
The original characterization of the K461A mutant and its transformation from transrepression to transactivation of composite and tethering GREs was carried out in F9 mouse embryonic carcinoma cells and CV-1 cells (29). To exclude cellspecific differences in the function of the K461A mutant, the function of the composite plfG GRE and the tethering colA GRE was tested in COS7 cells. In contrast to the results obtained with the osteocalcin gene, the expression of the K461A GR mutant activated the gene activity in both reference promoters in COS7 cells (Fig. 2). In the presence of dexamethasone, wt GR repressed luciferase activity to 60% from a plfG GRE, whereas the K461A mutant induced luciferase activity about 25-fold ( Fig. 2A). The effect on the tethering colA element was similar but to a much lesser degree (Fig. 2B) with dexamethasonedependent repression of luciferase activity to about 60% with wt GR and dexamethasone-dependent stimulation of luciferase activity to about 160% with the K461A mutant.
We have previously shown that GR binds specifically to the negative response element thought to be responsible for the transrepressive effect on the human osteocalcin promoter (25,26). The K461A mutant binds to the Ϫ41/Ϫ9 fragment of the human osteocalcin promoter, containing the nGRE, in a manner similar to that of wt GR (Fig. 3A, lanes 3 and 4). The GR-specific band with wt GR is competed for specifically by an unlabeled oligonucleotide containing a standard GRE sequence (tyrosine aminotransferase GRE; TAT) (Fig. 3A, lanes 5 and 6). The use of a vitamin D-responsive element did not affect GR binding (Fig. 3A, lane 7). Mutation of the nGRE sequence diminishes the GR-specific complex (Fig. 3A, lane 10). Competitive DNA binding was estimated by titrating increasing amounts of the unlabeled probe into the binding reactions containing either wt GR or the K461A mutant. Measurement of the relative binding of the radiolabeled probe showed that there was no major difference in DNA-binding affinity between wt GR and the K461A mutant (Fig. 3B).
To further support the hypothesis that competitive GR binding is responsible for repression of the osteocalcin gene, we compared the effect of a GR chimera, containing the GR DBD K461A fused to the activation domain of the viral transcrip-  2. Glucocorticoid-dependent effects on AP1-dependent transactivation. COS7 cells were transfected with wt GR or K461A mutant GR together with a luciferase reporter gene containing A, three tandem copies of the proliferin GRE (plfG) or B, a single consensus AP1 site from the collagenase A gene (ColA), upstream of the Drosophila alcohol dehydrogenase minimal promoter (Ϫ33/ϩ53), and incubated with 20 nM dexamethasone. Luciferase activity was assayed in cell extracts from one 6-cm plate and related to the activity in cells transfected with wt GR exposed to dexamethasone. Shown is the mean Ϯ S.D. of three separate experiments, each carried out with three independent triplicate analyses.
tional activator VP16, on osteocalcin, pflG and colA controlled reporter gene activity. As shown in Fig. 4, expression of the GR-VP16 chimera containing the mutant K461A DBD increased the luciferase activity measured in the case of both reference promoters used in this study (Fig. 4, B and C). However, the repression mediated by this constitutively active chimeric transcription factor on osteocalcin gene transcription was still comparable to the repressive effect mediated by the wt GR-VP16 protein (Fig. 4A). The magnitude of transactivation/ transrepression with the GR K461A-VP16 chimera (Fig. 4A) was identical to the dexamethasone-dependent activity of the full-length GR variants on osteocalcin promoter activity (Fig.  1). Thus, the repression of the osteocalcin promoter by GR is dependent on the DNA binding function and is independent of the transactivation domain associated.
Finally, the function of the isolated DBD of the two GR variants was tested with regard to their transrepressive effect. Expression of the isolated DBD of either wt GR or the mutant K461A repressed the osteocalcin promoter to an equal degree, with virtually identical activity to the full-length variants (Fig.  5A). In contrast, the isolated wt DBD had no effect on the plfG reporter activity (Fig. 5B). The isolated DBD of the K461A GR mutant demonstrated a weak stimulatory activity on the plfG element (Fig. 5B). However, this was considerably reduced compared with the dexamethasone-dependent activity of the full-length K461A mutant ( Fig. 2A). DISCUSSION The repression of gene transcription is of particular interest since the mechanism underlying these repressive effects are less well understood than those governing activation. It has previously been shown that administration of glucocorticoids leads to a transcriptional repression of several target genes. These transrepressive effects of glucocorticoids are related to their clinically important anti-inflammatory effects (e.g. repression of collagenase A) or relevant side effects of pharmacological usage of glucocorticoids such as steroid-dependent osteoporosis (e.g. repression of osteocalcin) (23,31,33). In contrast to the unifying model proposed for gene activation by GR, a simple unique model has not been formulated to account FIG. 3. DNA binding of wt GR and K461A mutant GR. COS7 cells were transfected with wt GR or K461A mutant GR and cell extracts prepared. A, aliquots of the soluble fraction were incubated with 32 Plabeled oligonucleotides corresponding to nucleotides Ϫ41/Ϫ9 of the osteocalcin promoter containing the nGRE or a nonfunctional mutant thereof (GREmut) and analyzed by gel mobility shift assay. The upper band (marked GR) was competed for by an 10 -50ϫ molar excess unlabeled oligonucleotide containing a standard GRE (TAT). A mutation in one of the GRE half-site sequences diminishes the GR specific complex (GREmut). Extract of cells transfected with parent expression vector (C) was used as control. B, relative binding affinity was estimated by titrating excess unlabeled oligonucleotide (5-20-fold molar excess) into the binding reaction and measuring the relative binding of 32 P-labeled oligonucleotide by densitometric scanning of the GR⅐DNA band following electrophoresis and autoradiography. The results of the densitometric analysis are expressed as [GR⅐DNA spec ]/[DNA free ] in proportion to specific binding activity in the absence of competitor oligonucleotide, from three independent experiments (mean Ϯ S.D.).

FIG. 4. Constitutive activity of GR-VP16 chimeric proteins.
COS7 cells were transfected with parent expression vector (Ctr) or expression vectors for chimeric receptors consisting of GR DBD (wt or K461A mutant) fused to the transactivation domain of VP16 together with a luciferase reporter gene. The reporter gene was A, driven by nucleotides Ϫ344/ϩ31 of the human osteocalcin gene (OC) or driven by Ϫ33/ϩ53 of the Drosophila alcohol dehydrogenase promoter with upstream elements consisting of B, three tandem copies of the proliferin GRE (plfG) or C, a single consensus AP1 site from the collagenase A gene (ColA). Luciferase activity was assayed in cell extracts from one 6-cm plate and related to the activity in cells transfected with wt GR exposed to dexamethasone. Shown is the mean Ϯ S.D. of three separate experiments, each carried out with three independent triplicate analyses.
for receptor-dependent gene repression.
In the present study we further characterized the mechanism underlying the negative glucocorticoid-dependent regulation of the human osteocalcin gene in comparison with two well described reference systems for repression committed by GR on phorbol ester-activated gene transcription, the mouse proliferin gene, and the collagenase A gene. We took advantage of a recently described rat GR variant, GR K461A, that is able to distinguish between at least three functional subclasses of glucocorticoid responsive elements: simple GREs binding the activated GR molecule in a homodimeric fashion, composite GREs in which the exertion of receptor action requires the binding of additional sequence specific transcription factors to a common binding site, and tethering GREs in which GR mediates the transcriptional rate of target genes by interfering with stimulatory transcriptional activators already bound to DNA (Fig. 6) (27). The K461A mutation results in a reduced transactivating activity on simple GREs that corresponds to a reduced binding affinity for the GRE sequence (29). On tethering or composite GREs, the K461A mutation results in a switch from glucocorticoid-dependent transrepression, in the presence of both c-Jun and c-Fos, to transactivation. Identical results were obtained in this study, using COS7 cells or HOS D4 osteosarcoma cells. Thus, the effect of the K461A mutation on AP1-dependent GR transrepression is not cell-specific. However, the magnitude of AP1-dependent GR-dependent transactivation of the colA element induced by the K461A mutation was considerably decreased as compared with that obtained previously with either F9 or CV-1 cells. The magnitude of induction induced by K461A on the plfG element was considerably larger and of the same order of magnitude seen previously in F9 and CV-1 cells.
The osteocalcin gene is an osteoblast-specific gene expressed in late stages of differentiation (34). Although the exact function of osteocalcin remains unclear, osteocalcin production is related to bone density and increased osteoblast activity (35). Exposure to glucocorticoids results in the reduction of osteocalcin mRNA to about 50%. Analysis of the promoter region of the human osteocalcin gene identified one specific binding site for GR which completely overlapped the TATA box (26). Transient expression of reporter genes driven by constructs containing the minimal osteocalcin promoter resulted in a glucocorticoiddependent reduction in reporter gene activity to 50% or less (23,25). Mutations of the promoter that eliminated GR binding obliterated the glucocorticoid-dependent repression of reporter gene activity. GR homodimer and TBP bind competitively for overlapping DNA elements in vitro (25). Based on these results we proposed a mechanism for glucocorticoid-dependent repression of the osteocalcin gene in which binding of GR to a negative GRE (competitive nGRE, Fig. 6) reduces the availability of the promoter for the basal transcriptional complex. Glucocorticoid-dependent transrepression of target genes by competitive binding of GR and transcriptional activators to overlapping DNA elements has been proposed for the regulation of the type 1 vasoactive intestinal polypeptide receptor gene as well as for the prolactin gene (15,17,36).
In contrast to the switch of AP1-dependent GR transrepression to transactivation by the K461A mutation, no effect at all was seen on the GR-dependent transrepression of the osteocalcin promoter. Both wt GR and the K461A mutant induced repression of the osteocalcin promoter to about 40% of basal activity. Both GR variants bound equally well to the osteocalcin nGRE (Fig. 3B). This is in contrast to the decreased binding and function on a simple positive GRE and transactivation with the mutant as described previously (29). These results confirm the dependence of osteocalcin repression by glucocorticoids on GR⅐DNA interaction rather than by protein⅐protein interaction.
Detailed molecular studies have demonstrated the functional requirement of the GR DBD in the modulation of gene expression by GR. The DBD has been shown to be necessary for both the transactivation and the transrepression functions of the receptor (20,28). The isolated DBD, either wt or the K461A mutant, were sufficient to induce transrepression of the osteocalcin promoter, which further strengthens the hypothesis of a direct competition in binding to overlapping DNA sequences between GR and TBP. In contrast, the isolated DBD was virtually inactive on the plfG element, even with the K461A mutation. Fusion of the DBD K461A to a heterologous transactivation domain from VP16 restored the function of the protein on the AP1-dependent elements, plfG and colA, resulting in constitutive transactivation with the K461A mutation. In contrast, the transrepression of the osteocalcin promoter remained unchanged, indicating that the glucocorticoid-dependent repression of the osteocalcin promoter is independent of the transactivation domain associated.
In conclusion, mutation of a single amino acid located at the FIG. 5. Constitutive activity of the isolated GR DBD. COS7 cells were transfected with parent expression vector (Ctr) or expression vectors for GR DBD (wt or K461A mutant) together with a luciferase reporter gene. The reporter gene was A, driven by nucleotides Ϫ344/ ϩ31 of the human osteocalcin gene (OC) or B, driven by Ϫ33/ϩ53 of the Drosophila alcohol dehydrogenase promoter with upstream elements consisting of three tandem copies of the proliferin GRE (plfG). Luciferase activity was assayed in cell extracts from one 6-cm plate and related to the activity in cells transfected with wt GR exposed to dexamethasone. The figure shows the mean Ϯ S.D. of three separate experiments, each carried out with three independent triplicate analyses. DNA-binding surface of rat GR, K461A, results in a receptor variant that can differentiate between two different mechanisms of glucocorticoid-dependent transrepression. The mutation switches AP1-dependent transrepression, involving protein⅐protein interaction, to transactivation, whereas DNAdependent transrepression is unaffected. In the crystal structure of the rat GR DBD bound to a simple positive GRE (37,38), the lysine side chain at position 461 makes a specific base contact with the DNA sequence. Loss of this contact following the mutation K461A would be expected to result in decreased affinity for the GRE and thereby decreased transactivation, which is what was previously reported for this mutation. However, the exact contacts between GR and the osteocalcin nGRE have not been demonstrated. The K461A mutation does not result in any loss in transrepression function, which would indicate that this residue does not play as active a role in binding to the osteocalcin nGRE as it does in binding to a classical positive GRE. Another explanation may be that Lys-461 plays an active role in DNA sequence-dependent conformational change of GR required for transactivation. In the osteocalcin nGRE, GR appears to exert its role by competing away TBP from the promoter, thereby reducing the transcriptional rate of the gene. Thus no further change or subsequent step in GR action would be required for the regulation of this gene.