INTRODUCTION

The steroid hormone 1,25-(OH)₂D₃ ¹ is a key regulator of mineral homeostasis in mammals. Its main function is to promote intestinal absorption of mineral ions and to mobilize calcium from bone (1, 2). In bone, 1,25-(OH)₂D₃ favors osteoclast and osteoblast differentiation (1, 2, 3, 4). Although the overall function of 1,25-(OH)₂D₃ in bone is to promote resorption, it is the osteoblasts and not the osteoclasts that contain vitamin D receptors (VDRs). As a consequence, the expression of several genes in osteoblasts is regulated by 1,25-(OH)₂D₃. For instance, the type I collagen genes (5) and the bone sialoprotein genes (6) are down-regulated by the hormone, while the osteopontin (7) and osteocalcin (8, 9, 10, 11, 12) genes are up-regulated in human and rat.

The role of 1,25-(OH)₂D₃ in the regulation of synthesis of osteocalcin, a noncollagenous protein of the bone extracellular matrix that regulates bone formation (13), has been extensively studied in human and rat. 1,25-(OH)₂D₃ increases the synthesis of this protein by rat osteosarcoma cells and increases the accumulation of its mRNA in rat osteoblastic cells (8). It also induces a rapid increase in Osteocalcin expression in the whole animal early after 1,25-(OH)₂D₃ treatment (14). The promoters of the human and rat osteocalcin genes were the first promoter elements in which VDREs were identified and characterized (10, 11, 12). The VDRE is typically made up of two direct repeats separated by a 3-bp spacer. This finding contributed to the definition of the "3, 4, 5 rule" for DNA binding of various nuclear receptors (15). The hormone-VDR complex binds to this element as an heterodimer with the retinoid X receptor (16). Notably, the VDRE is located in most cases within the first kilobase (kb) of 5-flanking sequence of the target gene (17). In spite of this regulatory role, the precise physiological function of 1,25-(OH)₂D₃ during osteoblast differentiation in vivo remains unclear. The fact that normal bone mineralization can be achieved in human in the absence of a functional VDR, provided that adequate serum calcium and phosphorus levels are maintained (18), suggests that the main function of the hormone in bone is to mobilize calcium through resorption.

We recently cloned the two mouse osteocalcin genes, OG1 and OG2 (19), and used the OG2 promoter to study the mechanisms controlling the osteoblast-specific expression of this gene (20, 21). In the course of this work, we were surprised to find a total absence of up-regulation of this promoter activity by 1,25-(OH)₂D₃, even though a VDRE-like sequence was present. Thus, we decided to systematically study this aspect of the regulation of expression of the mouse osteocalcin genes in vivo and in vitro. Here we show that the expression of the mouse osteocalcin genes, unlike their rat and human homologues, is down-regulated by 1,25-(OH)₂D₃ and that an indirect mechanism is involved in this down-regulation.

EXPERIMENTAL PROCEDURES

Primary osteoblasts were isolated from calvariae of 2-day-old mice as described previously (20, 22), with the following modifications. Calvariae were sequentially digested for 20, 40, and 90 min at 37 °C in -modified minimum essential medium (Life Technologies, Inc.) containing 0.1 mg/ml collagenase P (Boehringer Mannheim) and 0.25% trypsin (Life Technologies, Inc.). Cells of the first two digests were discarded, whereas cells released from the third digestion were plated in -modified minimum essential medium and 10% fetal bovine serum. After 2 days, this medium was replaced by mineralization medium (-modified minimum essential medium and 10% fetal bovine serum containing 5 mM -glycerophosphate and 0.1 mg/ml ascorbic acid), which was changed every 2 days. After 10 days of culture in mineralization medium, the cells were treated with 1,25-(OH)₂D₃ dissolved in ethanol or with vehicle and harvested at specific intervals before RNA extraction. ROS 17/2.8 rat osteosarcoma cells (23) were maintained in Dulbecco's modified Eagle's medium, Ham's F-12 medium, and 10% fetal bovine serum; MC3T3-E1 mouse cells (24) were maintained in -modified minimum essential medium and 10% fetal bovine serum. Both cell lines were treated with 1,25-(OH)₂D₃ or with vehicle for the appropriate period of time. RNA was prepared using RNAzOL^TM (Cinna/Biotecx Laboratory) following the manufacturer's instructions.

All inserts were cloned in the p4luc promoterless luciferase expression vector (25), upstream of the luciferase gene (luc). A SalI site was introduced at position +13 of OG2 by PCR amplification of the sequence extending from positions 657 to +13, using as primers the oligonucleotides 5-CCAAGACCTGGCCCAG-3 for the 5-side and 5-TGGTCGACTTGTCTGT-3 for the 3-side. This initial fragment was used to generate p657OG2-luc (20). p1316OG2-luc was created by inserting the EcoRV-NcoI fragment of the OG2 promoter (positions 1316 to 657) at the 5-end of p657OG2-luc (20). p147OG2-luc was obtained by an NcoI-PvuII (positions 343 to 147) deletion of p657OG2-luc (20). p2900OG2-luc was created by inserting at the 1035 PstI site in p1316OG2-luc a 1.9-kb PstI-BamHI fragment of the OG2 promoter. p147OG2mut-luc was generated by cloning in p4luc a PCR fragment resulting from the amplification of p147OG2-luc, using as primers the oligonucleotides 5-GATCCGCTGCAATCACCAAGAACAGCA-3 for the 5-side and 5-TGGTCGACTTGTCTGT-3 for the 3-side. The existence of the 2-bp substitution mutation was verified by DNA sequencing. To generate pOG1-luc constructs, a SalI site was first introduced at position +13 of OG1 by PCR amplification of the OG1 promoter sequence extending from positions 343 to +13, using as primers the BglII site-containing oligonucleotide 5-CCGAATTCAGATCTCT-3 for the 5-side and the SalI site-containing oligonucleotide 5-TGGTCGACTTGTCTGT-3 for the 3-side. A PstI-BglII fragment of the OG1 promoter (positions 1035 to 343) was cloned upstream of the initial BglII-SalI fragment to generate p1049OG1-luc. p3300OG1-luc was created by cloning a PstI-EcoRI fragment extending 2.3 kb upstream of the PstI site (position 1035) to this latter construction. PCR product sequences were all verified by DNA sequencing. The rat osteocalcin-luc construct was created by insertion of 1.7 kb of the HindIII-BamHI rat Osteocalcin promoter fragment (a gift from Dr. M. Demay), blunt-ended, into the SmaI site of p4luc. pROP-luc contains 1.7 kb of the rat Osteopontin promoter fused to luciferase (26); pMOP-luc contains 1 kb (positions 910 to +90) of mouse Osteopontin promoter fused to luciferase (a gift from Dr. A. Ridall). Orientation and conformity of the inserts cloned in p4luc were confirmed by restriction analysis.

Nine-week-old mice (28 g) were injected intraperitoneally with 24 ng of 1,25-(OH)₂D₃ or with vehicle as described previously (27) and killed 24 h later. Calvariae were dissected out and cleaned of soft tissues, and RNA was prepared by extraction with guanidine isothiocyanate, followed by cesium chloride density gradient ultracentrifugation (28).

Northern blot analysis was performed as described previously (28). Denatured RNA samples were loaded on 1% formaldehyde-agarose gels and transferred onto Hybond N⁺ nylon filters (Amersham Corp.). The filters were then hybridized with either the Osteocalcin (19) or Osteopontin (29) probes labeled by random priming. After autoradiography and signal quantification using a PhosphorImager, filters were stripped and reprobed with a mouse -actin probe that was used as a control of RNA loading. Reverse transcription was conducted on 5 µg of DNase-treated total RNA from 10-day-old primary osteoblast cultures treated with 1,25-(OH)₂D₃ (10⁷ M) or vehicle 60 h before harvest. cDNA synthesis was performed by oligo(dT) priming using avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim) under conditions suggested by the supplier. PCR amplification was conducted on one-tenth of the reverse transcription reaction using as 5-primer 5-CAAGTCCCACACAGCAGCTT-3 (positions +8 to +27) and as 3-primer 5-AAAGCCGAGCTGCCAGAGTT-3 (positions +359 to +378) to generate a 370-bp PCR product for osteocalcin cDNAs (19). The primers 5-GTTGAGAGATCATCTCCACC-3 and 5-AGCGATGATGAACCAGGTTA-3 were used to generate a 320-bp PCR product corresponding to exon 2 of the hypoxanthine-guanine phosphoribosyltransferase cDNA. For the analysis of reverse transcription-PCR products, aliquots of the PCR mixtures were separated on a 1% agarose gel, and the gel was dried. Two synthetic oligonucleotides were used as probes to detect mRNA specifically transcribed from either OG1 (5-AGGACCATCTTTCTGCTC-3, positions +35 to +52) or OG2 (5-AGGACCCTCTCTCTGCTC-3, positions +35 to +52) (19). Hybridizations were carried out at 42 °C directly on the gel as described previously (19, 30).

DNA transfections of ROS 17/2.8 and MC3T3-E1 cells were performed by calcium phosphate precipitation (31). Cells were treated with 1,25-(OH)₂D₃ or vehicle for 64 h before harvest as described previously (11). Luciferase activities were assayed by using a Monolight 2010 luminometer (Analytical Luminescence Laboratory) and D-luciferin substrate (Analytical Luminescence Laboratory) in 100 mM Tris-HCl (pH 7.8), 5 mM ATP, 15 mM MgSO₄, and 1 mM dithiothreitol. -Galactosidase activities in each lysate, measured by a colorimetric enzyme assay using resorufin -D-galactopyranoside (Boehringer Mannheim) as a substrate, were used to normalize transfection efficiency between experiments.

Nuclear extracts from ROS 17/2.8 cells and primary osteoblasts were prepared as described (20). Cells were treated with 1,25-(OH)₂D₃, retinoic acid, BMP7 (one orphogenetic rotein ), -estradiol, or vehicle 48 h before harvest. For the electrophoretic mobility shift assay, double-stranded mouse VDRE-like, rat VDRE, OSE2 (steoblast-pecific lement ), and mOCE1 oligonucleotides were labeled and purified as described previously (20). Five fmol of labeled double-stranded oligonucleotides were incubated in binding buffer (50 mM KCl, 12 mM HEPES (pH 7.9), 1 mM EDTA, 1 mM dithiothreitol, and 12% glycerol) with 8 µg of nuclear extracts and 2 µg of poly(dI-dC) for 15 min at room temperature. For competition experiments, the indicated amount of unlabeled double-stranded oligonucleotide was added to the binding reaction with the other components. The samples were loaded on a 5% polyacrylamide gel in 0.5 × Tris borate/EDTA (44.5 mM Tris-HCl, 44.5 mM boric acid, and 4 mM EDTA). DNA binding and electrophoresis conditions for OSE2 and mOCE1 were as described previously (20).

RESULTS

In the promoter of one rat osteocalcin gene, a functional VDRE is located between positions 455 and 441 (10, 11). A similar sequence, but containing a 2-bp substitution, is located at approximately the same place (positions 465 to 451) in the promoter elements of OG1 and OG2. To determine whether this was a functional VDRE, we performed DNA transfection experiments in ROS 17/2.8 rat osteosarcoma cells using four different constructs. The p1049OG1-luc and p657OG2-luc promoter-luciferase chimeric genes both contain the VDRE-like sequence; p147OG2-luc, which lacks any recognizable VDRE, served as a negative control, and the 1.7-kb rat Osteopontin promoter-luciferase chimeric gene (pROP-luc), which contains two well characterized functional VDREs, served as a positive control (26). ROS 17/2.8 cells were transfected with these vectors and treated with or without 1,25-(OH)₂D₃ (10⁷ or 10⁸ M) for 64 h. In agreement with previous findings (32), the expression of OG1-luc and OG2-luc chimeric genes was never up-regulated by treatment of transfected cells with 1,25-(OH)₂D₃ (Fig. 1). Instead, their expression was always reduced by this treatment, whereas the expression of pROP-luc, the positive control vector, was always up-regulated (Fig. 1A).

8

To demonstrate that the absence of up-regulation of the OG1 and OG2 promoter activities was not due to the choice of a rat cell line for studying the activity of mouse promoters, we performed two additional experiments. First, we tested whether the mouse Osteopontin promoter activity was increased by 1,25-(OH)₂D₃ treatment in ROS 17/2.8 cells. As shown in Fig. 1A, 1,25-(OH)₂D₃ treatment of ROS 17/2.8 cells transfected with the pMOP-luc construct, which contains 1 kb of the mouse Osteopontin promoter, increased the activity of this promoter 3.2 times. Second, the regulation of the OG1 and OG2 promoter activities by 1,25-(OH)₂D₃ was tested in a mouse osteoblastic cell line (MC3T3-E1). The expression of p147OG2-luc was always reduced by treatment of these cells with 1,25-(OH)₂D₃, while the activity of pMOP-luc exhibited at least 10 times the control activity following the same treatment. These results indicate that the absence of up-regulation of OG1 and OG2 promoter-luciferase constructs by 1,25-(OH)₂D₃ is not due to the absence of a mouse specific cofactor in ROS 17/2.8 cells.

The results of these experiments also implied that the 2-bp substitution in the VDRE-like sequence of OG1 and OG2 promoters abolished or greatly diminished the binding of the hormone-VDR complex to this site. To test this hypothesis, we performed an electrophoretic mobility shift assay using as probes an oligonucleotide encompassing the VDRE of the rat Osteocalcin promoter (rat VDRE) and an oligonucleotide containing the VDRE-like element present in OG1 and OG2 promoters (mouse VDRE-like) (Fig. 2A). As a source of proteins, we used nuclear extracts of ROS 17/2.8 cells, either untreated or treated for 48 h with retinoic acid (10⁷ M), -estradiol (10⁸ M), or 1,25-(OH)₂D₃ (10⁸ M). Incubation of untreated or retinoic acid- or -estradiol-treated ROS 17/2.8 nuclear extracts with rat VDRE or mouse VDRE-like probes did not generate a protein-DNA complex (Fig. 2B, lanes 1-3 and 5). In contrast, when we used nuclear extracts of 1,25-(OH)₂D₃-treated ROS 17/2.8 cells, there was a marked increase in the amount of binding to the labeled rat VDRE oligonucleotide, but not to the labeled mouse VDRE-like oligonucleotide (Fig. 2B, compare lanes 1 and 4 and lanes 5 and 6). This result suggests that the mouse VDRE-like sequence has a much lower affinity for the 1,25-(OH)₂D₃-VDR complex than does the rat VDRE. This was confirmed by DNA competition experiments in the electrophoretic mobility shift assay. The binding of hormone-treated ROS 17/2.8 nuclear extracts to the labeled rat VDRE oligonucleotide was substantially inhibited by excess molar amounts of the unlabeled rat VDRE oligonucleotide, but not by the same molar amounts of the unlabeled mouse VDRE-like oligonucleotide (Fig. 2C).

Sequence 455 to 441 in OG1 and OG2 promoters is not a high affinity VDRE.

8

7

8

Thus, taken together, the DNA transfection experiments, the DNA binding studies, and the DNA competition experiments demonstrate that the 2-bp substitution in the VDRE-like sequence of OG1 and OG2 promoters nearly abolished the binding of the hormone-VDR complex to this sequence and explain the absence of up-regulation of OG1 and OG2 promoter-luciferase activities by 1,25-(OH)₂D₃ treatment.

We then analyzed the regulation of expression of the endogenous osteocalcin genes by 1,25-(OH)₂D₃. We first used mouse primary osteoblast cultures as an experimental system. These cells were cultured in the absence or presence of 1,25-(OH)₂D₃ (10⁷, 10⁸, or 10⁹ M) for 36, 48, or 60 h and assayed for Osteocalcin expression by Northern blot analysis. Surprisingly, osteocalcin mRNA was readily detectable in untreated cells, but was nearly undetectable following 1,25-(OH)₂D₃ treatment of these cultures. This inhibition of Osteocalcin expression was concentration-dependent and occurred at every time point analyzed (Fig. 3A). In contrast, as previously shown (33), Osteocalcin expression in ROS 17/2.8 cells was barely detectable before 1,25-(OH)₂D₃ treatment, but was up-regulated up to 7-fold by the hormonal treatment (Fig. 3B). To demonstrate that this down-regulation of the mouse Osteocalcin expression was not due to our culture conditions, we monitored the expression of Osteopontin, a gene whose expression is known to be increased by 1,25-(OH)₂D₃ treatment in mouse (34). Consistent with our DNA transfection experiments, Osteopontin expression was always up-regulated by 1,25-(OH)₂D₃ treatment of these cultures (Fig. 3A). -Actin expression was used as an internal control in these experiments since it is not affected by 1,25-(OH)₂D₃ treatment (27). To show that the down-regulation of Osteocalcin in 1,25-(OH)₂D₃-treated primary osteoblasts was an accurate reflection of the regulation occurring in vivo, we treated 9-week-old mice with a single dose of 1,25-(OH)₂D₃ (24 ng), isolated RNA from calvariae 24 h later, and measured Osteocalcin expression. In the entire animal, 1,25-(OH)₂D₃ also led to a marked decrease (6-fold) in osteocalcin mRNA accumulation (Fig. 3C). Taken together, our results show that in mouse, 1,25-(OH)₂D₃ inhibits Osteocalcin expression.

9

7

8

To determine whether both osteocalcin genes were down-regulated by 1,25-(OH)₂D₃ treatment, we used an allele-specific oligonucleotide hybridization of the reverse transcription-PCR product assay (19). RNA from untreated or 1,25-(OH)₂D₃-treated mouse primary osteoblasts was used as template. Aliquots of each PCR mixture were electrophoresed on an agarose gel along with aliquots of plasmids containing either OG1 or OG2. The gel was hybridized with a labeled "OG1" oligonucleotide that recognizes only transcripts originating from OG1 (19). OG1 expression was virtually abolished by 1,25-(OH)₂D₃ (10⁷ M) treatment of these cells (Fig. 4). This gel was stripped and rehybridized with the "OG2" oligonucleotide designed to visualize cDNAs originating only from OG2 (19). OG2 expression was also virtually abolished by 1,25-(OH)₂D₃ treatment of these cultures (Fig. 4). Thus, the expression of the two mouse osteocalcin genes is down-regulated by 1,25-(OH)₂D₃ in vivo.

7

Given this inhibition of OG1 and OG2 expression by 1,25-(OH)₂D₃, we then searched for VDREs located farther upstream in the promoters of these genes. First, we sequenced 2.9 kb of the OG2 promoter and 3.3 kb of the OG1 promoter, but no VDRE consensus sequence (15) could be identified (data not shown). Next, we generated promoter-luciferase constructs containing the largest OG2 promoter fragment isolated (2.9 kb) or 3.3 kb of the OG1 promoter. Each of these constructs was transfected into ROS 17/2.8 or MC3T3-E1 cells, which were subsequently treated with 1,25-(OH)₂D₃ or vehicle. We used three different positive controls in this experiment: pROP-luc, pROC-luc, and pMOP-luc. The activity of the three positive control constructs in treated cells was always several times that in untreated cells. The activity of the OG1 and OG2 promoter-luciferase constructs was constantly reduced by approximately half by the same treatment in both cell lines (Fig. 5).

8

The moderate effect observed in DNA transfection experiments suggested that 1,25-(OH)₂D₃ might act indirectly on Osteocalcin transcription, possibly by inhibiting the expression of a gene that encodes a factor controlling Osteocalcin expression. Although no such gene has been identified yet, osteoblast-specific nuclear proteins binding to osteoblast-specific cis-acting elements have been characterized (20, 21, 35). To determine whether 1,25-(OH)₂D₃ modulated binding of these factors, we monitored the amount of binding to two well characterized cis-acting elements in nuclear extracts of untreated and 1,25-(OH)₂D₃-treated mouse primary osteoblasts. OSE2 is an osteoblast-specific cis-acting element that binds OSF2 (osteoblast-pecific actor ), an osteoblast-specific nuclear protein. OSE2 controls the osteoblast-specific expression of OG2 (20, 21). mOCE1 binds a ubiquitously expressed protein and does not affect OG2 promoter activity significantly (20). When we performed an electrophoretic mobility shift assay using as a probe a labeled OSE2 oligonucleotide, OSF2 binding was nearly abolished in 1,25-(OH)₂D₃-treated nuclear extracts (Fig. 6A, lane 3). Two lines of evidence indicate that this effect was specific. First, it could not be obtained by using nuclear extracts of cells treated with another differentiation agent of the osteoblasts such as BMP7 (Fig. 6A, lane 2); second, the binding of nuclear proteins to mOCE1 was unaffected by treatment of the cells with 1,25-(OH)₂D₃ (Fig. 6B, lane 3). This latter result demonstrated the integrity of the nuclear extracts. Likewise, 1,25-(OH)₂D₃ treatment of primary osteoblasts did not affect the binding of nuclear proteins to OSE1 (20), another osteoblast-specific cis-acting element (data not shown). The specific inhibition of binding of OSF2 to OSE2 in 1,25-(OH)₂D₃-treated nuclear extracts suggests that the regulation of Osteocalcin expression by 1,25-(OH)₂D₃ occurs through an indirect mechanism: the hormone either down-regulates OSF2 expression or induces post-translational modification, preventing binding of OSF2 to DNA.

7

7

As shown in Fig. 1, treatment of ROS 17/2.8 or MC3T3-E1 cells with 1,25-(OH)₂D₃ down-regulated the activity of the 147-bp OG2 promoter fragment that contains the originally described OSE2 sequence (20). To demonstrate the importance of this sequence in the inhibition of OG2 promoter activity, we introduced in this OG2 promoter fragment a 2-bp substitution mutation in OSE2 that abolishes OSF2 binding (20). We then performed DNA transfection experiments using constructs containing either the wild-type or mutated promoters fused to luciferase in ROS 17/2.8 cells or MC3T3-E1 cells. In agreement with our DNA binding data, the activity of the mutated promoter was not down-regulated by 1,25-(OH)₂D₃ treatment of these cells, but rather slightly up-regulated (Fig. 7).

8

DISCUSSION

We have analyzed the regulation of the mouse osteocalcin genes by 1,25-(OH)₂D₃ and showed that their expression is down-regulated by this hormone. Our results also indicate that this effect is mediated, at least in part, through an indirect mechanism inasmuch as 1,25-(OH)₂D₃ treatment of osteoblasts abolished OSF2 binding to OSE2, a potent osteoblast-specific cis-activator of transcription (20, 21).

One of the roles of 1,25-(OH)₂D₃ in bone is to promote osteoblast differentiation, although the overall physiological consequences of this differentiating action are not clearly understood (1). 1,25-(OH)₂D₃ regulates the expression of several genes expressed in osteoblasts, among them Osteocalcin (8, 9, 10, 11, 12). This regulation of expression occurs at the transcriptional level, wherein the hormone-receptor complex binds to a VDRE located in the promoter of the rat and human osteocalcin genes (9, 10, 11). However, a growing body of evidence indicates that once the osteoblasts begin to express Osteocalcin, the effect of 1,25-(OH)₂D₃ on gene expression can vary with the time at which the hormonal treatment is initiated. For instance, when 1,25-(OH)₂D₃ treatment of rat calvaria cultures was initiated during the proliferative stage of this culture, there was no increase in Osteocalcin expression (36). These data suggest that the hormone may have multiple and complex effects in vivo on osteoblast gene expression varying with the stage of differentiation of the cells.

Several experimental arguments suggest that the inhibitory effect of 1,25-(OH)₂D₃ on mouse Osteocalcin expression was not due to the stage of differentiation of the osteoblasts under our culture conditions. First, we did not begin 1,25-(OH)₂D₃ treatment before the osteoblasts reached the mineralizing stage. Second, the expression of Osteopontin, a gene expressed earlier than Osteocalcin during osteoblast differentiation and which was used in these experiments as a positive control (7), was always up-regulated in our cultured osteoblasts. Third and most important, the same inhibition of Osteocalcin expression occurred when we performed this treatment in the entire animal.

What is the mechanism by which 1,25-(OH)₂D₃ inhibits mouse Osteocalcin expression? Based on our results, we propose that the hormone achieves this function through an indirect mechanism. 1,25-(OH)₂D₃ treatment of mouse primary osteoblasts selectively abolished the binding of OSF2, an osteoblast-specific transcription factor that binds to and controls the activity of OG1 and OG2 promoters (20, 21).² Our working hypothesis is that 1,25-(OH)₂D₃ treatment of primary osteoblasts inhibits transcription of the gene coding for OSF2. Such an indirect mechanism may also explain why we observed only a 2-fold effect when we tested the effect of 1,25-(OH)₂D₃ treatment on OG1 and OG2 promoter activity in DNA transfection experiments. Indeed, it is known that other cis-acting elements are involved in the control of Osteocalcin expression (20, 35), and the binding of nuclear factors to these elements was not affected by the hormonal treatment. The fact that the inhibitory effect of 1,25-(OH)₂D₃ treatment was greater on longer promoter fragments suggests that other OSE2-related sequences could be present farther upstream in OG1 and OG2 promoters. We (21) and others (37) recently presented evidence that OSF2 may be a member of the PEBP2 family of transcription factors. The eventual identification of a cDNA for OSF2 will allow a direct demonstration of the proposed mechanism. Broess et al. (38) reported recently that 1,25-(OH)₂D₃ inhibits Osteocalcin expression in differentiated osteoblasts from chicken; it will be important to determine whether 1,25-(OH)₂D₃ also inhibits OSF2 binding to the promoter of the chicken osteocalcin gene.

This hypothesis does not exclude the possibility that 1,25-(OH)₂D₃ affects Osteocalcin expression directly. The hormone-VDR complex may bind to an as yet unidentified VDRE and, through its interaction with specific cofactors, inhibit its transcription. This mechanism has been shown recently to explain the ligand-independent repression of transcription by the thyroid hormone receptor (39). Although most of the VDREs identified to date are located within 1 kb of 5-flanking sequence of the target gene (17), we cannot exclude the possibility that the VDRE is located farther upstream or downstream in the Osteocalcin locus, as has been demonstrated for the regulation of the HoxA cluster by retinoic acid (40). Experiments in progress are designed to test these two hypotheses, which are not mutually exclusive.

A more general question raised by our studies is, what is the importance of the regulation of Osteocalcin expression by 1,25-(OH)₂D₃ during osteoblast differentiation in vivo? Since this hormone differently regulates the expression of the same gene in closely related species such as rat and mouse, its role is not likely to be essential for the establishment of the osteoblastic phenotype. Indeed, we have observed in osteocalcin-deficient mice that the osteoblasts do differentiate normally and can lay down an extracellular matrix and mineralize it (13).