JOURNAL OF BONE AND MINERAL RESEARCH Volume 8. Number 8, 1993 Mary Ann Liebert, lac., Publishers

Role of the Negative Glucocorticoid Regulatory Element in Glucocorticoid Repression of the Human

Osteocalcin Promoter

NlGEL MORRISON and JOHN EISMAN

ABSTRACT

In osteoblast-like cells in culture, the human osteocalcin gene promoter basal activity is repressed by gluco- corticoids, reflecting the repression of serum osteocalcin concentrations noted in syndromes of glucocorti- coid excess. 1,25-Dihydroxyvitamin D, [ 1 ,25-(OH),D3], the active hormonal form of vitamin D, induces the osteocalcin promoter through a vitamin D response element (VDRE), and glucocorticoids also repress the vitamin D-induced promoter. This study addresses the role of a glucocorticoid receptor (CR) binding site overlapping the TATA box of the osteocalcin promoter, which had been proposed as a negative glucocorti- coid response element (nCRE), invoking a steric interference mechanism of glucocorticoid repression. Con- firmation of the role of the nCRE in regulating basal activity was obtained using promoter constructs con- taining a TATA box swap. However, a minor component of repression of 1,25-(OH),D3 induced activity re- mained in the absence of the nCRE. In addition, glucocorticoid repression of the human osteocalcin pro- moter was shown to be cell line specific. This result is not compatible with a simple model of repression and suggests the existence of unidentified cell-specific factors that are involved in the repression event. Repres- sion of the osteocalcin promoter was compatible with a composite model involving both the nGRE site and glucocorticoid regulation of factors that bind the vitamin D response element.

INTRODUCTION

STEOCALCIN IS A BONE MATRIX PROTEIN produced by 0 the osteoblast. The small amount of free osteocalcin in serum is assayed as a clinical indicator of bone turn- over.“) Low serum osteocalcin is a characteristic of gluco- corticoid excess syndromes, which can cause osteoporo- sis.I2) The induction of the human osteocalcin gene by 1.25-dihydroxyvitamin D3 [ 1 ,25-(OH),D3] and repression by glucocorticoids has been reported.”) Glucocorticoids repress the induction of osteocalcin by other agents, such as retinoic acid and basic fibroblast growth factor (unpub- lished data). Therefore, glucocorticoid repression overrides known inductions of the human osteocalcin promoter. In contrast, glucocorticoids can positively regulate other bone-related genes in osteoblast-like cells, in particular alkaline p h ~ s p h a t a s e ‘ ~ . ~ ) and the vitamin D receptor.I6)

A steric hindrance model has been proposed for a num- ber of other genes repressed by glucocorticoid receptor (GR).(’-”) In this model, a strong positive activator of

gene expression is competitively displaced by the GR as a result of mutually incompatible (overlapping) DNA bind- ing sites in the gene promoter. The overall result is a net negative effect on gene transcription. Sequences defined as positive glucocorticoid response elements (GREs), in which GR binds and induces gene transcriptional activity, are similar enough for a consensus sequence to be pro- posed(13); however, the exact sequence requirements for negative glucocorticoid response elements (nGREs) are not fully understood.

A potential GR binding site that overlaps the TATA box of the osteocalcin gene was proposed”’ as an nGRE medi- ating GR repression by a steric hindrance model, as pro- posed for a number of other genes repressed by GR.(’-”’ Steric inhibition of TATA binding factors by GR binding over the TATA box is a reasonable model for an overrid- ing negative effect. Although GR binding to this site was subsequently demonstrated in vitro,‘14) its biologic impor- tance has not been verified.

This study clarifies further the role of the putative osteo-

Garvan Institute of Medical Research, St. Vincent’s Hospital, Sydney, Australia.

969

970 MORRISON AND EISMAN

calcin nGRE. We verify that the nGRE is involved in the glucocorticoid repression of the basal activity of the osteo- calcin promoter. However, the glucocorticoid repression of osteocalcin promoter activity does not occur in breast cancer cell lines, despite adequate receptor number, a find- ing incompatible with a steric hindrance model. In addi- tion, glucocorticoids regulate the abundance of nuclear factors capable of interacting with the osteocalcin vitamin D response element (VDRE). The data are compatible with a composite model of glucocorticoid repression involving cell-specific events at the nGRE site and glucocorticoid regulation of factors that bind the VDRE.

MATERIALS AND METHODS

Osteocalcin promoter constructs driving CAT (chloram- phenicol acetyltransferase) gene expression were previously described.(3) All media and sera were obtained from ICN- Flow Laboratories (Sydney, Australia). The rat osteoblas- tic osteosarcoma cell line ROS 17/2.8 (the gift of S. Rodan, Merck Sharp and Dohme, West Point, PA) was maintained in Ham's F12 medium with 5 % fetal calf serum (FCS), as described.Il5) ORG2058 (16a-ethyl-21-hydroxy- 19-nor-pregn-4-ene-3,20-dione) a synthetic progestin, was obtained from Amersham Australia. 1 ,25-(OH),D, was the gift of M. Uskokovic (Hoffman-LaRoche, Nutley, NJ). Dexamethasone (Dex), a potent synthetic glucocorticoid, was obtained from Sigma Chemical Corp. (Sydney, Aus- tralia). Hormones were used as previously described, with treatment times of 24 h unless otherwise indicated. t 3 )

Changes in cell number were not noted over these treat- ment periods. Hormones were added to cultured cells as concentrates in ethanol, ensuring that the final concentra- tion of vehicle was 0.1 %. Previous dose response of Dex repression(3) indicated maximal repression occurred at M. Constructs were made by standard means using synthe- tic oligonucleotides. Human progesterone receptor expres- sion vector was obtained from Professor P. Chambon (Strasbourg, France). Transfections and hormone treat- ments were done in Dulbecco's modified Eagle's medium (DMEM) with 2% charcoal-stripped FCS (cFCS). Cotrans- fection experiments contained equimolar quantities of CAT construct and expression vector. CAT activity was assessed by the method of Sleigh,(16) as previously de- scribed.") CAT activity data are presented as dpm per h of ''C transferred from acetyl coenzymeA to chlorampheni- col per pg CAT construct in extracts derived from 10" cells. Transfections used 40 pg DNA per 2 x lo6 cells in a 150 cm' flask, as previously described. 0 ) Student's t-test was used to determine significance in transfection experiments, and p values for described effects were typically <0.001. A p value > 0.05 indicated nonsignificance.

Nuclear extracts for DNA gel mobility shift experiments were prepared according to Dignam et a1."') ROS 17/28 cells were treated overnight with hormone [ 1 ,25-(OH),D3 at 10-8 M and D e x at M; addition to cells was simulta- neous] before preparation of nuclear extracts. The probe was a 65 base pair (bp) oligonucleotide duplex, which is defined as the VDRE region,"' labeled by a fill-in reaction

(Klenow fragment, Pharmacia) of Hind111 ends and gel purified. DNA-protein interactions were assessed using the binding conditions of Sagami et aI.(l8) Briefly, 5 pg nuclear extract was incubated on ice in 10 mM HEPES (pH 7.9), 50 mM NaCI, 0.5 mM EDTA, 2 mM MgCI, 10% glycerol, and 1 mM dithiothreitol in the presence of nonspecific competitor DNA [poly(dI), poly(dC) Pharmacia]. Probe ( 5 x lo' dpm) was then added and the incubation continued for 20 minutes on ice. Reactions were analyzed on 5 % polyacrylamide native gels run at 100 V in Tris-borate- EDTA running buffer (45 mM Tris-borate and 1 mM EDTA) at room temperature. Gels were dried and autora- diographed for 16 h using Kodak XAR film without inten- sifying screens. Shorter exposures were used to quantify ratios of band intensities by laser densitometry.

Osteocalcin promoter constructs were previously de- scribed."' The 5' ends of different constructs are pOSCATl , - 344 bp, and pOSCAT2, - 1.3 kilobasepairs (kbp). These constructs have the same 3' end at +31 bp, which encompasses the normal osteocalcin mRNA start site. pOSCAT1-VDRE was previously described and con- tains a synthetic vitamin D response element (-521 to -477 bp of the osteocalcin promoter) inserted at an up- stream Hind111 site in pOSCAT1. All plasmid DNA was prepared by alkaline lysis and cesium chloride ultracentri- fugation essentially according to Maniatis et Con- structs were verified by sequencing using the double- stranded method. Plasmids were grown in Escherichia coli strains RRl and XL-I blue (Stratagene, San Diego).

Human breast cancer cell lines, T47D and MCF-7, were cultured in DMEM with 10% FCS as described by Hall et al. ( l o )

RESULTS

Progesterone receptor (PR) recognizes the same DNA binding sites as GR yet is activated by different l i g a n d ~ . ~ " ) Lack of PR repression would be incompatible with a steric hindrance model involving the TATA box. In ROS 17/28 cells the specific synthetic progestin, ORG2058 [which has low affinity for estrogen, mineralocorticoid, glucocorti- coid, and androgen receptors(11'), had no effect on the osteocalcin promoter (CAT activity derived from pOSCAT2) at low concentrations (not shown) and only a minor repressive effect at high concentrations ( M), at which an interaction with the GR is expected (Fig. 1A). After transfection of PR expression vector, however, ORG2058 was a potent repressor of both basal and 1,25- (OH),D,-induced CAT activity, demonstrating that repres- sion is receptor mediated (Fig. 1B).

The steric hindrance model relies on the presence of a GR binding site overlapping the osteocalcin TATA box. In ROS 17/28 cells, pTKCAT [driven by the thymidine ki- nase (TK) promoter of herpes simplex virus] is resistant to glucocorticoid repression. A construct (pTATA-VDRE) was made, replacing the osteocalcin TATA box region with that of the TK promoter using pOSCAT1-VDRE as a parent as described in Fig. 2A and B. A consensus GRE se-

OSTEOCALCIN nCRE

A . B .

971

+1,2S(OH)2Dg

ai 10% No ORG2058 log IORG2058) M

FIG. 1. Progestins potently repress the osteocalcin promoter in cells transfected with the progesterone receptor. (A) Progestins (ORG2058) have a minor effect on CAT activity derived from pOSCAT2 transfected into ROS 17/2.8 cells in the absence of PR expression vector. ORG2058 (ORG) at lo-" M is compared with potent glucocorticoids cortisol (CO), corticosterone (CS), and triamcinalone acetonide (TA), all at lo-' M. ORG2058 does not repress the basal activity and has only a minor effect on CAT activity induced by 1,25-(OH),D1 (10-8 M). (B) Cotransfection of expres- sion vector for human PR in ROS 17/2.8 cells now permits repression of pOSCAT2-derived CAT activity by ORG2058 (ORG). Both the basal and the l,ZS-(OH),D,-induced (10-O M), activities are repressed. Treatments were for 24 h before harvest and CAT assay. CAT activities quoted are dpm derived from 10" cells per hour per pg CAT construct transfected. Student's [-test p values for repression from control and from 1,25-(OH),D,-induced cells are all <0.001.

quence consists of an imperfect inverted repeat of two 6 bp half-sites separated by a 3 bp gap.(l3.l4) A comparison of the GRE with the osteocalcin TATA box region shows 75% identity with the GRE: a conservation of 5 and 4 bp, respectively. of 6 bp possible in each potential GR binding half-site, giving a total of 9 bp conserved of 12 bp possible. In contrast, the TK TATA box region has only 3 of 12 bp in common with the GRE, with 2 of 6 bp and 1 of 6 bp in each potential response element half-site (Fig. 2B). There- fore, the TK TATA box has little in common with the con- sensus GR binding site. pOSCAT1-VDRE, which contains from a +31 bp to -344 bp of the osteocalcin promoter and a synthetic VDRE, responds five- to sixfold to 1,25- (OH),D,, and Dex represses both the basal and induced ac- tivity. pTATA-VDRE was transfected into ROS 1712.8 and tested for repression of basal and 1,25-(OH),D,-in- duced CAT activity. In contrast to the repression of the parent pOSCATI-VDRE, pTATA-VDRE basal activity was not repressed by Dex (Fig. 2C). However, the 1.25- (OH),D,-induced CAT activity of pTATA-VDRE was re- pressed by Dex.

pTATA-VDRE was cotransfected with PR expression vector, permitting a test of repression by endogenous GR (activated by Dex) and exogenous PR (activated by ORG- 2058). The basal activity of pTATA-VDRE was not re- pressed by Dex or ORG2058 (Fig. 2D). If the binding of GR to the nGRE in the TATA region were the only mecha- nism of repression, pTATA-VDRE would be resistant to the repression of 1,25-(OH),D,-induced activity. However, the response of pTATA-VDRE to 1,25-(OH),Ds was cur- tailed by both Dex and ORG2058 (Fig. 2D). Dose-response experiments confirmed that pTATA-VDRE basal activity is resistant to Dex repression, and this construct retains a

component of repression of the 1 ,25-(oH),D, response (Fig. 2E). These data confirm that the osteocalcin TATA box is necessary for glucocorticoid repression of basal ac- tivity but indicate that the repression of the basal activity and repression of the induced activity are different phe- nomena.

I t was evident that in comparison to pOSCATI-VDRE, pTATA-VDRE basal promoter activity was inherently lower and the magnitude of the I,2S-(OH),D3 response in pTATA-VDRE (approximately twofold) was lower than the five- to sixfold response consistently seen in pOSCAT1-VDRE (Fig. 2B-F and data not shown). The shape of the dose-resonse curve and concentration re- quired for maximal CAT activity ( M) was not appreci- ably altered in pTATA-VDRE (Fig. 2F) from that de- scribed for the entire promoter (pOSCAT2; see Ref. 3). Although pTATA-VDRE contains only a small region of the TK promoter, the magnitude of the 1,25-(OH),D3 re- sponse matched that previously noted for pTKCAT-VDRE (which has 105 bp of TK promoter).(31 These data suggest a role for the osteocalcin TATA box region in interactions with upstream elements to regulate the magnitude of the 1 ,25-(OH)1D, response and basal activity. Although pTATA-VDRE basal activity was reduced, the level was such that repression could have been detected.

Although glucocorticoid repression of pOSCAT2 OC- curred in ROS 17/2.8 cells, it was not detected in human breast cancer lines MCF-7 and T47-D (data not shown). In contrast to the expected repression, treatment with either OGR2058 or Dex resulted in a modest but significant in- crease in CAT activity derived from pOSCAT2. This was true for the basal promoter activity and the 1,25- (OH),D,-induced activity. Similarly, the basal and 1,25-

972 MORRISON AND EISMAN

A .

T A I A I I O X

D .

? l H H ) 1

I hlH> I

CON - 8 - 7 DEX CON .R - 7 DF,h U U ORC.2058 o n ~ z o 5 8

I I

1

; - 101111 i

i i l H l 1-

('Oh - 8 - 7

log IDEXl M

- + I.?5D3

pTATA-VDRE +1,25D3

pOSCATl -VDRE

+ 1.25D3

FIG. 2. Glucocorticoid repression of basal but not induced CAT activity is eliminated by replacing the osteocalcin TATA box region with the TK TATA box. (A) Construction of the TATA box swap. Comparison of the parent con- struct pOSCATI-VDRE and pTATA-VDRE. The swap was made using synthetic oligonucleotides spanning the region between the Sphl and Xhol sites indicated, such that the region covering the GR footprint described in Stromstedt et al.("' was replaced by TK promoter sequences (stippled box). Sphl site is at -67 bp in the osteocalcin promoter; the Xhol site is in the CAT leader sequence. (B) Sequence of the relevant region around the TATA box at the junction of the replacement. Underscored sequence on the left is osteocalcin promoter sequence starting at -48 bp. The TATA box region is indicated by a striped bar. Above the respective TATA boxes of the osteocalcin and TK promoters is a consen- sus GRE sequence. The consensus GRE comprises a nearly perfect palindrome separated by a 3 bp gap. Overlined base pairs in the TATA box regions represent bases conserved between the consensus GRE and the respective TATA box re- gions of the promoters. The osteocalcin TATA box region has 9 of 12 base pairs conserved, but the TK TATA box has only 3 of 12 conserved. (C) Magnitude of induction comparison of pTATA-VDRE and pOSCAT1-VDRE, indicating lack of repression of pTATA-VDRE basal activity by Dex. However, Dex repression of the 1 ,25-(OH)1D,-induced activ- ity of pTATA-VDRE can be detected. Note that the total fold 1,25-(OH)1D, induction has been affected by the TATA box swap. (D). Cotransfection of hPR confirms of lack of repression of pTATA-VDRE basal activity by endogenous GR activated by Dex (lo* M) and transfected PR activated by ORG2058 (lo-8 and lo-' M). In contrast to the basal activ- ity, the 1,25-(OH),D,-induced activity of pTATA-VDRE is repressed by Dex and ORG2058. p Values for repression of the 1,2S-(OH),D3-induced CAT activity are all <O.OOOl. (E) Progressive reduction in l,%(OH),D, induction obtained by increasing Dex concentrations. Basal activity was not repressed. Treatment for 36 h. 1.25-(OH)1D, was at lo-' M. (F) Although the magnitude of induction of pTATA-VDRE is reduced, the dose response retains the sigmoidal shape, apparent EDso concentration (between and M), and maximal concentration M) previously observed for p0SCAT2.t3l

OSTEOCALCIN nCRE 973

(OH),D,-induced CAT activities of pTATA-VDRE were not repressed by Dex in MCF-7 cells.

GR effects on basal and induced activity could be sepa- rated in pTATA-VDRE. arguing for additional hormon- ally modulated effects acting through the VDRE. The pat- tern of nuclear factors binding to the VDRE was assessed under different hormone treatments. These data (Fig. 3) show a complex pattern of binding activities in untreated cells. Surprisingly, no difference in gel shift pattern was detected after treatment with 1 ,25-(OH),D3. Dex treatment resulted in the elimination of two prominent bands and the presence of an additional, fainter band. Combined treat- ment with 1,25-(OH),D3 and Dex resulted in a dramatic re- duction in the abundance of one of the two major binding activities, a return of the bands affected by Dex alone and an elimination of two other low-mobility bands, which were unaffected by Dex-alone treatment.

DISCUSSION

The data support the role of the nGRE binding site over- lapping the TATA box as a component of the GR-medi- ated repression of osteocalcin promoter activity in osteo-

sarcoma cells in culture. However, it is evident that further complexity exists in the GR-mediated repression. If the nGRE was the sole basis of repression, then cell lines with a comparable content of OR or PR should be competent for repression. MCF-7 and T47-D have GR content similar to ROS 17/2.8 cells(a2' and a high PR content under the conditions used. I1O) Repression events should have been detectable, since @SCAT constructs have moderate CAT activity in these cell lines and 1,25-(OH),D3 induction is between three- and fivefold. Despite that these breast can- cer cell lines have GR and PR, repression did not occur, indicating that a simple nGRE mechanism is not strictly valid and suggesting the involvement of yet unidentified cell-specific factors in the repression event.

The TATA box swap construct clearly had lost the GR- mediated repression of the basal promoter activity but re- tained a component of repression of the l.25-(OH),D3-in- duced activity. This construct also had simultaneous changes in the magnitude of the 1,25-(OH),D3 response and the basal activity. These data suggest that the TATA box region is involved in interactions with upstream ele- ments that enhance the magnitude of the 1,25-(OH),D, re- sponse and regulate the basal activity and is involved in cell-specific GR repression. Recently published data dem-

b + Extract d

I I .- - + + D e x - _._

Coldoligo: I - 50x I - + - + 1,25 1 1 2 3 4 5 6 1

FIG. 3. The pattern of nuclear factors binding to the osteocalcin VDRE is affected by hormone treatment. Track 1 shows the free labeled probe; track 2 shows specific competition using control ROS 17/2.8 nuclear extract and a 50-fold molar excess of unlabeled probe. Track 3 shows the pattern of binding using extracts from untreated cells (the same ex- tract was used in track 2). The pattern of binding consists of two extremely intense bands (here overexposed to permit visualization of the weaker bands) and a number of prominent secondary bands, which are well resolved. Cells treated with 1,25-(OH),D3 (track 4) show no discernible change in the pattern of binding. However, track 5 shows that extracts from cells treated with Dex do not display two prominent shifted bands, indicated by the arrowheads on the right. In ad- dition, an additional faint low mobility band is apparent (not indicated, but visible as the upper most band in track 5 ) . Combined treatment of cells with 1,25-(OH),D3 and Dex demonstrates quite a different pattern of binding, with a reap- pearance of the two prominent bands that were affected by Dex alone treatment and the elimination of two low-mobility bands seen in tracks 3 and 4, and 5 (the two lowest mobility bands in tracks 3 and 4). There is also a marked reduction (greater than 50% by densitometry) in one of' the two strongest binding activities (indicated by the open triangle).

974

onstrated increased binding of nuclear factors on the rat osteocalcin promoter TATA box region after ROS 17/2.8 cells were treated with I ,25-(0H)]D3.(l3)

GR can regulate target promoters through a direct inter- action and also by cross-regulation of additional factors that act on the same promoter. Hormone-induced changes in the pattern of protein binding on the VDRE indicate complex combinatorial effects are possible on such a pro- moter. The identity of the binding factors in ROS 17/2.8 cells is currently unknown; however, factors known to bind the osteocalcin VDRE include vitamin D retinoic acid receptors, retinoid X receptor RXR-a (un- published results), and Jun and Fos. (1‘.15) Jun and Fos are nuclear oncogene products that form activator protein 1 (Ap-1), which induces genes through an Ap-1 site.‘26) Studies on the collagenase p r o m ~ t e r ( ~ ’ . ’ ~ ) show that GR can inhibit promoters that are regulated by Ap-1 by a mechanism of mutual inhibition that is independent of DNA binding. Since the osteocalcin promoter contains Ap-1 sites,”) which regulate basal a~tivity,(~‘.’~l GR regu- lation of osteocalcin may include such a mechanism. It is also possible that the GR represses 1 ,25-(OH)]D, induction by reducing the vitamin D receptor content of the cell; however, Dex induction of vitamin D receptor has been re- ported in osteoblast-like cells.(6)

The data are consistent with a model of regulation in which GR binding to the nGRE can cause repression of basal promoter activity. This occurs through a cell-specific mechanism that may involve interactions between factors that bind in the vicinity of the TATA box and upstream constitutive activators, which contribute to the basal activ- ity of the whole promoter. Repression of the 1,25- (OH),D,-induced activity is more complex but evidently has a component of repression generated by GR binding at the nGRE. The gel shift data are consistent with GR-in- duced changes in the pattern of DNA binding activities on the VDRE. A component of repression might be ascribed to GR elimination of DNA binding factors by the mutually inhibitory mechanism proposed for the collagenase pro- moter.(27.2n) Other possibilities include the GR downregu- lation of VDR and other as yet unidentified components of the vitamin D response system, such as coactivators, and accessory proteins, such as RXR-a, which enhances VDR binding to a response element.(”’ These possibilities are under further study.

These data also sugggest that progestins may have a role in the regulation of osteocalcin promoter activity in osteo- blast cells i f the progesterone receptor is present. Although our experiments relied on transfected PR, since the ROS 1712.8 cells did not have an endogenous progestin re- sponse, we note that Eriksen et observed that the PR can be induced by estrogen in primary human bone cells in culture.

In conclusion, this study confirms the role of the nGRE in the regulation of promoter basal activity and demon- strates that GR regulation of 1 ,25-(OH),D3-induced pro- moter activity is cell specific, is not fully explained by a simple nGRE model, and may involve complex changes in DNA binding activities on the VDRE.

MORRISON AND EISMAN

ACKNOWLEDGMENTS

Judith Flanagan and Susan Gillies are thanked for ex- pert assistance. Professor P. Chambon (Strasbourg) is thanked for providing human progesterone expression vec- tors. Sevgi Rodan (Merck Sharpe and Dohme Laborato- ries, West Point, PA) is thanked for providing ROS 17/2.8 cells.

REFERENCES

1. Hauschka PV, Lian JB. Cole DE. Gundberg CM 1989 Osteocalcin and matrix gla protein: Vitamin K dependent proteins of bone. Physiol Rev 69:990-1047.

2. Reid IR, Chapman GE, Fraser TRC, Davies AD, Surus AS, Meyer J , Hyq NL. lbbertson HK 1986 Low serum osteocal- cin levels in glucocorticoid-treated asthmatics. J Clin Endo- crinol Metab 62:379-383.

3. Morrison NA, Shine J , Fragonas JC, McMenemy LM, Ver- kest V, Eisman JA 1989 1,25 Dihydroxyvitamin D responsive element and glucocorticoid repression in the human osteocal- cin gene. Science 254: 1 154- I 159.

4. Majeska RJ, Nair BC. Rodan GA 1985 Glucocorticoid regu- lation of alkaline phosphatase in the osteoblastic osteosar- coma cell line ROS 1712.8. Endocrinology 116:170-179.

5 . Green E, Todd B, Heath D 1990 Mechanism of glucocorti- coid regulation of alkaline phosphatase gene expression in osteoblast-like cells. Eur J Biochem 188:147-153.

6. Chen TL. Cone CM, Morley-Horton E, Feldman D 1982 Glucocorticoid regulation of I .25(OH),-vitamin D, receptors in cultured mouse bone cells. J Biol Chem 257:13564-13569.

7. Akerblom IE, Slater EP, Bearo M, Baxter JD, Mellon PL 1988 Negative regulation by glucocorticoids through interfer- ence with a CAMP responsive enhancer. Science 241:350-353.

8. Sakai DD, Helms S. Carlstedt-Duke J, Gustafsson J-A, Rottman F, Yamamoto KR 1988 Hormone mediated repres- sion: A negative glucocorticoid response element from the bovine prolactin gene. Genes Dev 2:1144-1154.

9. Guertin MH. la Rue H. Bernier D, Wrange 0, Chevrette M, Gingras MC, Belanger L 1988 Enhancer and promoter ele- ments directing activation and glucocorticoid repression of the a-fetoprotein gene in hepatocytes. Mol Cell Biol 8:1398- 1407.

10. Mordaq JC, Linzer DIH 1989 Co-localization of elements re- quired for phorbol ester stimulation and glucocorticoid re- pression of the proliferin gene expression. Genes Dev 3:760- 769.

1. Drouin J, Trifiro MA, Plante RK, Nemer M, Eriksson P. Wrange 0 1989 Glucocorticoid receptor binding to a specific DNA sequence is required for hormone-dependent inhibition of proopiomelanocortin gene transcription. Mol Cell Biol 9:

2 . Oro AE, Hollenberg SM, Evans RM 1988 Transcriptional in- hibition by a glucocorticoid receptor-beta-galactosidase fu- sion protein. Cell 55:1109-1114.

3. Evans RM 1988 The steroid and thyroid hormone receptor superfamily. Science 240:889-895.

4. Stromstedt P-E, Poellinger L, Gustafsson J-A, Carlstedt- Duke J 1991 The glucocorticoid receptor binds to a sequence overlapping the TATA box of the human osteocalcin pro- moter: A potential mechanism for negative regulation. Mol Cell Biol 11:3379-3383.

5305-5314.

OSTEOCALCIR nGRE 975

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

Rodan GA, Rodan SB 1984 In: WA Peck (ed.) Bone and Min- eral Research, Annual II, Elsevier, Amsterdani. pp. 244-285. Sleigh MJ 1986 A nonchromatographic assay for the expres- sion of the chloramphenicol acetyl transferase gene in eu- karyotic cells. Anal Biochem 156:251-254. Dignam JD, Lebovitz RM, Roeder RG 1983 Accurate tran- scription initiation by RNA polymerase I I in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res 11:1475- 1489. Sagami I, Tsai H, Wang M, Tsai B. O'Malley B 1986 Identi- fication of two factors required for transcription of the oval- bumin gene. Mol Cell Biol 6:4259-4267. Maniatis T , Frisch EF, Sambrook J 1981 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Hall R, Lee CSL, Alexander AA, Shine J , Clarke C, Suther- land RL 1990 Steroid hormone receptor gene expression in human breast cancer cells: Inverse relationship between oes- trogen and glucocorticoid receptor messenger RNA levels. Int J Cancer 46:1081-1087. Raynaud J-P, Ojasoo T, VachC V 1989 Unusual steroids use- ful in measuring steroid receptors. In: Thompson EB, Lipp- man ME (eds.) Steroid Receptors and the Management of Cancer. CRC Press, Boca Raton, FL, pp. 216-232. Haussler MR, Manolagas SC, Deftos LJ 1980 Glucocorticoid receptors in clonal osteosarcoma cell lines: A novel system for investigating bone active hormones. Biochem Biophys Res Commun W:373-380. Bortell R. Owen TA, Bidwell JP. Gavazzo P , Breen E, van Wijnen AJ, DeLuca HF, Stein JL, Lian JB, Stein GS 1992 Vitamin D-responsive protein-DNA interactions at multiple promoter regulatory elements that contribute ro the level of rat osteocalcin gene expression. Proc Natl Acad Sci USA 89:

Ozono K. Liao J, Kerner SA, Scott RA, Pike JW 1990 The vitamin D responsive element of the human osteocalcin gene. J Biol Chem M5:21881-21888.

61 19-6123.

25. Owen TA, Bortell R, Yocum SA, Smock SL, Zhang M, Abate C, Shalhoub V, Aronin N, Wright KL, van Wijnen AJ , Stein JL, Curran T. Lian JB, Stein GS 1990 Co-ordinate occupancy of Ap-l sites in the vitamin D responsive and CAAT box elements by Fos-Jun in the osteocalcin gene: Model for phenotype suppression of transcription. Proc Natl Acad Sci USA 81:9990-9999.

26. Lamph WW, Wamsley P , Sassone CP, Verma 1M 1988 In- duction of proto-oncogene Jun/AP-1 by serum and TPA. Nature 3M629-63 1.

27. Jonat C, Rahmsdorf HJ , Park KK, Cato AC, Gebel S , Ponta H, Herrlich P 1990 Antitumor promotion and anti-in- flammation: Down modulation of AP-I (JunIFos) activity by glucocorticoid hormone. Cell 62:1189-1204.

28. Yang-Yen H-F, Chambard J-C, Sun Y-L, Smeal T. Schmidt TJ , Drouin J , Karin M 1990 Transcriptional interference be- tween c-Jun and the glucocorticoid receptor: Mutual inhibi- tion of DNA binding due to direct protein-protein interac- tion. Cell 62:1205-1215.

29. Kliewer SA, Umesono K. Mangelsdorf DJ, Evans RM 1992 Retinoid X receptor interacts with nuclear receptors in reti- noic acid, thyroid hormone and vitamin D signalling. Nature 355:44&449.

30. Eriksen EF, Colvard DS. Berg NJ, Graham ML, Mann KG, Spelsberg TC. Riggs BL 1988 Evidence of estrogen receptors in normal human osteoblast-like cells. Science 3241 :84-96.

Address reprint requests to: Dr. N , Morrison

Garvan Institute of Medical Research St. Vincent's Hospital

Sydney 2010, Australia

Received in original form August 31, 1992; in revised form Janu- ary 21, 1993; accepted January 25, 1993.