Highly Potent Transcriptional Activation by 16-ene Derivatives of 1,25-Dihydroxyvitamin D3 LACK OF MODULATION BY 9-CZS-RETINOIC ACID OF RESPONSE TO 1,8&DIHYDROXYVITAMIN D3 OR ITS DERIVATIVES*

Although several studies have been performed on the biological activities of analogs of 1,25-dihydroxyvitamin DS (1,25-(OH)2 DS) at the whole animal and cellular lev- els, little work has been done to analyze their transcriptional activation properties. A highly inducible 1,25-(OH)2 Ds-responsive promoter composed of three copies of the mouse osteopontin vitamin Ds response element CVDRE3) inserted upstream of a herpes simplex virus thymidine kinase promoter has been constructed, and its transcriptional properties have been analyzed by transient transfection into the monkey kidney cell line COS-7 and the rat osteoblast-like osteosarcoma line ROS 17D.8. We have studied systematically transcriptional activation by a number of 1,25-(OH)2 DS analogs, particularly those substituted at positions 16,23,26, and 27, sites that are targets for metabolism. Strikingly, ex- cept for derivatives that bind the 1,26-(OH)%

useful for analyzing the tissue-specific transcriptional activity of 1,25-(OH)2 DS and its derivatives in any cell type amenable to transient transfection.
Vitamin Ds is a secosteroid whose precursor is synthesized in the skin through the cleavage of the B ring of 7-dehydrocholesterol by n o n e n z~a t i c photolysis and isome~zation (Holick, 1981). The resulting product is hydroxylated in the liver to 25-hydroxyvktamin D3 (Ponchon et al., 1969) and further lahydroxylated in the renal proximal tubule to its most biologically active form la,25-dihydroxyvitamin D3 (1,25-(0Hl2 D3)l (Holick et al., 1971;Lawson et al., 1971;Norman et al., 1971).
Because of its lipophilicity 1,25-(0W)2 Ds is capable of passing through cellular membranes and binding to the vitamin D3 receptor (VDR) present in target cells (Haussler and McCain, 1977;Liao et al., 1990).
The VDR is a member of the nuclear receptor family of transcriptional regulators and is part of a subgroup that includes receptors for thyroid hormone and retinoids. These receptors recognize specific DNA sequences, known as response elements, that are composed of direct repeats with the consensus sequence ~r~G~) T C A .
DNA binding and transcription^ activation studies have shown that the VDR recognizes direct repeats separated by 3 bp. 1,25-(OH)Z D3 response elements (VDREs) conforming to this structure have been identified upstream of osteocalcin genes and the mouse osteopontin (MOP) gene (Kerner et al., 1989;Morrison et al., 1989;Noda et al., 1990;Demay et al., 1992). The MOP VDRE is composed of direct repeats containing the consensus sequence GGTTCA (Noda et dl., 1990).
Several studies have shown that the VDR requires an auxiliary factor for binding to its response element (Sone et al., 1991;Ross et aE., 1992). This factor has been demonstrated to be the nuclear retinoid X receptor, RXR (Yu et al., 1991; The abbreviations used are: 1,25-fOH)z D3, Iru,25-dihydroxyvitamin D,; Pur, purine; bp, base pairfs); CAT, chloramphenicol acetyltransferase; MOP, mouse osteopontin; RA, retinoic acid; RAR, retinoic acid receptor; RARE, RA response element; RXR, retinoid X receptor; RXRE, retinoid X response element; tk, thymidine kinase; VDR, vitamin D3 receptor; VDRE, vitamin D3 response element; 24,25-(0H)2 D3, 24,25- 2971 by Vitamin D3 Derivatives al., 19921, which specifically binds the ligand 9-cis-retinoic acid (9-cis-RA) (Heyman et al., 1992;Levin et al., 1992). The VDR binds to response elements as a heterodimer with RXRs. RXRs have also been shown to homodimerize in uitro in the presence of g-cis-RA and bind selectively to response elements composed of direct repeats separated by 1 bp (Zhang et al., 1992). Transcriptional activation studies have shown that these sequences can act as 9-cis-RAresponse elements (Mangelsdorfet al., 1991;Zhanget al., 1992). Recent results have suggested that 9-cis-RA can also augment the response to 1,25-(OH)2 D3 of a synthetic promoter composed of a MOP VDRE placed upstream of a truncated herpes simplex virus thymidine kinase (tk) promoter (Carlberg et al., 1993). This suggests that 9-cis-RA may modulate transcription of a number of target genes containing different response elements and control a wide range of biological processes.
Given its antiproliferative effects, 1,25-(OH)2 D3 is potentially useful clinically in the control of various cancers and in the treatment of psoriasis. However, because of its role in calcium mobilization, hypercalcemia results from 1,25-(OH)2 D3 treatment. Recent efforts have resulted in the development of a number of synthetic 1,25-(OH)2 D3 analogs displaying low calcemic activity, many of which carry various side chain substitutions (Zhou et aZ., 1990(Zhou et aZ., , 1991Abe et al., 1991;Bouillon et al., 1992;Haq et al., 1993). Although these derivatives differ from 1,25-(OH)2 D3 in their calcemic properties, many have been shown to be highly effective in controlling cellular differentiation and maintain a high affinity for the VDR (Zhou et al., 1991). Certain nonhypercalcemic derivatives activate transcription of a chimeric osteocalcin promoterhacterial chloramphenicol acetyltransferase (CAT) reporter recombinant in stably transfected ROS 17/23 osteoblast-like osteosarcoma cells (Morrison and Eisman, 1991).
Here, we have constructed a highly inducible promoted reporter recombinant for analyzing the transcriptional activity of 1,25-(OH)2 D3 and its analogs. This construct contains three MOP VDREs inserted upstream of the herpes simplex virus tk promoter. We have tested the transcriptional activity of a number of 1,25-(OH)2 D3 analogs with different side chain substitutions in the monkey kidney cell line COS-7 and in ROS 1712.8 cells and have found that some of these derivatives activate transcription more potently than 1,25-(OH)2 D3, in some cases at 100-fold lower concentrations. Peak transcriptional activity varies little among derivatives, suggesting that the various substitutions do not affect the activity of the ligand-bound VDR. Although the MOP VDRE binds VDR/RXR heterodimers in uitro, we do not observe any effect of 9-cis-u on transcriptional induction by 1,25-(OH)2 D3 or its derivatives in COS-7 or ROS 17/2.8 cells, suggesting that 9-cis-RA may not directly modulate the transcriptional activity of VDR/RXR heterodimers. This promoter/reporter system should be useful for analyzing the tissue-specific activity of 1,25-(OH)2 D3 and its derivatives in a wide variety of cell types.

Plasmids
The VDREtkCAT and VDRE3tkCAT recombinants were created by inserting a monomer or directly repeated trimer, respectively, of the sequence 5'-GATCCGTACAAGG?TCACGAGG'ITCACGTC'ITA-3' containing the MOP VDRE, which is flanked by BamHI and BglII ends, into the BamHI site of pBLCAT8+ (Klein-Hitpass et al., 1986) upstream of the thymidine kinase promoter. MREMUTtkCAT and VDRE3-MUTtkCAT recombinants were constructed in an identical manner using the sequence 5'-GATCCGTACAAGGCCCACGAGGCCCACGTCT-TA-3', which contains two base changes per repeat of the VDRE. The M R expression vector (VDRIpSG5) was constructed by inserting a 2.1-kilobase EcoRI fragment containing the entire coding region of the human VDR into the EcoRI site of pSG5 . For expression in Escherichia coli, the VDR cDNA was amplified by polymerase chain reaction using oligonucleotides that introduced a KpnI site into the 5' end of the cDNA replacing the initial methionine codon, and an XhoI site immediately after the 3' end of the coding sequence. The resulting amplified fragment was cloned into the bacterial expression vector pET32  digested with KpnI and XhoI.

Methods
Cell Culture and Dansfections-Both COS-7 and ROS 17/2.8 cells were propagated in 9-cm dishes in Dulbecco's modified Eagle's medium containing 5% fetal bovine serum (Life Technologies, Inc.). Fetal bovine serum was charcoal stripped for use in experiments with 9-cis-RA. For CAT assays, 5 pg of the MOP VDRE3tkCAT or related construct, 5 pg of P-galactosidase expression vector pCH110, and 10 pg of BlueScribe (Stratagene) carrier DNAwere transfected onto 9-cm plates of COS-7 or ROS 17/2.8 at approximately 30% confluence using the calcium phosphate coprecipitation technique (Baneji et al., 1981). 1,25-(OH), D, or one of its derivatives was added at final concentrations indicated in the figures. The medium was changed, and fresh hormone was added 24 h after transfection. Cells were harvested 44-48 h after transfection. For gel retardation assays, cells were transiently transfected as above with 10 pg each of the expression vectors VDWpSG5 and FtXRdpSG5 and harvested 44-48 h after transfection.
Extracts of Mammalian Cells"Transient1y transfected cells were harvested for gel retardation assays by washing cells with 5 ml of ice-cold phosphate-buffered saline followed by scraping the cells in 1 ml of phosphate-buffered saline. Cells were centrifuged at 2,000 rpm for 10 min at 4 "C, and the pellets were resuspended in 100 pl of high salt extraction buffer (25 m~ Tris (pH 7.9),0.3 m~ dithiothreitol, 0.1 EDTA, 420 m~ NaCI, 10% (v/v) glycerol). Cells were lysed by three cycles of freezing at -70 "C (20 min) and thawing at room temperature, then centrifuged at 10,000 rpm for 10 min at 4 "C. Supernatants were stored at -70 "C. Extracts were prepared for CAT assays as described (Webster et al., 1988). ROS 17/2.8 cell nuclear extracts were made by the protocol of Digman et al. (1983).
Expression of the VDR in E. colt-The VDR was expressed in E. coli using the bacteriophage T7 expression system (Studier and Moffatt, 1986). Cultures (500 ml) of BLZlpLysS harboring the pET32 control plasmid or VDRpET32 were grown at 37 "C to A595 of 0.4 and induced with 0.4 m~ isopropyl-1-thio-P-D-galactopyanoside for an additional 2 h. Cells were centrifuged and the pellets resuspended in 5 ml of 25 m~ "ris (pH 7.91, 0.3 m~ dithiothreitol, 0.1 m~ EDTA, 200 m~ NaCl, 10% (v/v) glycerol, and 2.5 pg/ml of the protease inhibitors leupeptin, antipain, and pepstatin. Cells were lysed by sonication, and cleared lysates were recovered aRer centrifugation at 30,000 rpm for 45 min in a Beckman Ti-70 rotor. Cells expressing the VDR produced a polypeptide of 50 kDa visible on SDS-polyacrylamide gels (data not shown). VDR activity was assayed for sequence-specific DNA binding and heterodimerization with human RXRa by gel retardation assay using an oligonucleotide containing the MOP VDRE or a mutant sequence.
Gel Retardation Assays-Bacterial and mammalian cell extracts were added to incubations as indicated in the figures. Samples were incubated for 15 min on ice in 10 pl of 25 m~ Tris-HC1 (pH 8.0), 1 m~ dithiothreitol, 50 m~ KCl, 20% glycerol containing 1 pg of poly (dIdC) and then for a further 20 min at 23 "C after the addition of 50,00@ 100,000 cpm (5-10 fmol) of 32P end-labeled double-stranded oligonucleotide. Samples were loaded on 5% polyacrylamide gels equilibrated in 25 promoter. Top, oligonucleotides containing the MOP VDRE, above (the directly repeated portion is indicated by the arrows), and the mutant sequence below (the altered base pairs are boxed). Bottom, map of the 1,25-(0H), D,-inducible VDRE3tkCAT promoter/reporter plasmid used in this paper. Panel B, the MOP VDRE binds VDR/RXR heterodimers. Gel retardation assays were performed with the MOP VDRE using 0.25 pl of the VDR expressed in bacteria (lanes 1 4 and 6-9) and whole cell extracts of COS-7 cells transfected with a human R X R a expression vector (lanes 2 4 ) or untransfected COS-7 cells (lanes 7-10). The complex formed in the presence of the VDR and RXRa is indicated by the arrowhead. The complex indicated by the asterisk is formed with bacterial extracts not expressing the VDR and is therefore considered to be nonspecific (data not shown). Note that specific DNA binding by endogenous VDRs in COS-7 cells (see below) was not seen with whole cell extracts.
CAT Assays-Prior to CAT assays, quantities of extracts were normalized for transfection efficiency by assaying P-galactosidase activity (Tora et al., 1989). CAT assays were performed as described (Tora et al., 1989).
Previous studies with natural and synthetic promoters responsive to nuclear receptors have shown that hormone response elements often act synergistically to mediate transcriptional activation (Jantzen et al., 1987;Schmid et al., 1989;Martinez and Wahli, 1989;Ponglikitmongkol et al., 1990;Mader and White, 1993). Therefore, a synthetic promoter containing multiple copies of a strong VDRE was constructed. The MOP VDRE is composed of a direct repeat of PurG'JTCA motifs separated by 3 base pairs ( 1990;Umesono et al., 1991). To create the recombinant VDRE3tkCAT (Fig. lA), three copies of the MOP VDRE oligonucleotide were inserted immediately upstream of the herpes simplex virus tk promoter in the vector pBLCAT8+ (Klein-Hitpass et al., 1986). We also constructed derivatives containing either one copy of the MOP VDRE (VDREtkCAT), or one or three copies of mutated MOP VDRE (VDREMUTtkCAT and VDRE3MUTtkCAT, respectively) containing two base changes per direct repeat ( Gel retardation assays demonstrate that the MOP VDRE provides a high affinity binding site for heterodimers of the 1,25-(OH)2 D3 receptor (VDR) and RXRa. Aretarded complex is formed which is dependent on the presence of both a bacterial extract containing the VDR and an extract of the monkey kidney cell line COS-7 transfected with an R X R a expression vector (Fig. lB, lanes 2 4 ) . This complex comigrates with the one formed by extracts of COS-7 cells cotransfected with VDR and RXR expression vectors (data not shown, but see Fig. 7A). A similar complex is not observed with VDRor RXRa-containing extracts alone ( Fig. 1 B , lanes 1 and 5 , respectively). This complex is not formed on a MOP oligonucleotide containing point mutations (Fig. IA, top;Noda et al., 1990) that disrupt VDR binding (data not shown, but see Fig. 7A). A much weaker complex of similar mobility, which is not seen with COS-7 cell extracts alone, is observed if the VDR is incubated with an extract of untransfected cells, indicating the presence of endogenous RXR(s) in COS-7 cells (Fig. lB, lanes 6-10). This result is consistent with evidence provided by Northern blots that R X R s are present in COS cells . Induction of the VDRE3tk Promoter in COS-7 and ROS 171 2.8 Cells-The VDRE3tkCAT promoter is responsive to 1,25-(OH)2 D3 in transiently transfected COS-7 cells (Fig. 2). In cells transfected with a VDR expression vector along with reporter plasmids, VDREtkCAT and VDRE3tkCAT were both stimulated by 50 nM 1,25-(OHl2 D3, whereas the parent vector pBLCAT8+ and derivatives VDREMUTtkCAT and VDRE3-MUTtkCAT were not responsive (Fig. 2). CAT activity in cells transfected with VDREtkCAT and VDRE3tkCAT and a VDR expression vector was inducible 10-and 30-fold, respectively, over background by 50 m 1,25-(OH)2 D3 (Fig. 2). At least 4-fold more CAT activity was induced by 1,25-(OHI2 D3 in VDRE3tkCAT-containing cells than in cells transfected with the VDREtkCAT derivative. We have also found that the VDRE3tkCAT promoter was readily inducible in COS-7 cells by 1,25-(0H)2 D3 in the absence of cotransfected VDR expression vector (see Figs. 5, 6, and 8). Taken together, these results indicate that the observed transcriptional stimulation is dependent on ligand-activated VDRs present in COS-7 cells and, moreover, that synthetic promoters containing MOP VDRE(s) are highly inducible in these cells. The activity ofVDRE3tkCAT was also tested in the rat osteoblast-like osteosarcoma cell line ROS 17/2.8 (Fig. 3). The promoter was induced 5-fold by 50 rm 1,25-(OH)z D3 in these cells. The nonhypercalcemic 1,25-(OH)2 D3 analog MC903 induced CAT activity to a similar extent (Fig.  3, A and B ) . This observation is consistent with the results of Morrison and Eisman (1991) who observed similar levels of induction by 1,25-(OH)z D3 and MC903 of a CAT gene under control of the human osteocalcin promoter in stably transfected ROS 17/2.8 cells.  (Zhou et al., 1990). We have used the VDRE3tkCAT promotedreporter system to test transcriptional induction by a number of 1,25-(OH)2-16-ene derivatives. COS-7 cells were chosen for these analyses because of the relatively low background and high inducibility of VDRE3tkCAT observed in this line (see Fig. 2). These experiments were performed in the absence of a VDR expression vector. Transcriptional induction was determined over a range from to loF6 M ligand M for 1,25-(OHl2 D3). A concentration curve was performed for 1,25-(0H)2 D3 for each independent transfection experiment so that transcriptional induction by a given ligand could be normalized to that observed with M 1,25-(OH)z D3. All of the 1,25-(OH)z-16-ene derivatives tested stimulated CAT activity at concentrations at least 10-fold lower than 1,25-(OH)z D3 (Fig. 5 and Table I). Half-maximal stimulation of CAT activity was observed with 10-20 x m 1,25-(OH)z D3 (Fig. 5a), whereas similar levels of induction occurred with 10-20-fold lower concentrations of 1,25-(OH)z-16-ene D3 (Fig. 5b) quired for half-maximal induction to 10-lo M (Fig. 5, c and d ) .
Finally, note that derivatives tested in Figs. 5 and 6 which have low affinity for the VDR are generally weak inducers of CAT activity. However, for those analogs that have moderate to high affinity for the VDR, there is no strict correlation between the affinity for the receptor and the potency of induction of CAT activity (see Table I for a summary of the results of Figs. 5 and 6). 9-cis-Retinoic Acid, the Cognate Ligand for R X R s , Does Not Stimulate the Response of the VDRE3tk Promoter to 1,25-Dihydroxyvitamin 0 3 or 16-ene Derivatives-Previous results have provided evidence that the MOP VDRE binds heterodimers of the VDR expressed in bacteria and RXRs expressed in COS-7 (OH), D3. cells in vitro (see Fig. 1). A similar protein-DNA complex is detected in nuclear extracts of ROS 17/2.8 cells (Fig. 7A, lane  1 ). This complex is not formed on a mutant of the MOP VDRE (Fig. 7A, compare lanes 1 and 3), its formation is stimulated by the addition of the VDR expressed in bacteria (Fig. 7A, lane 2 ), and it comigrates with the VDFURXR.DNAcomplex observed in Fig. 1 (Fig. 7A, lane 5). These data strongly suggest that the retarded complex corresponds to a heterodimer of the VDR and RXR(s) expressed in ROS 17/23 cells. Northern analysis of poly(A)+ RNA from ROS 17/23 cells specifically detects the presence of bands corresponding to RXRa and RXRp, but not RXRy, mRNAs (Fig. 7B), as well as that of the VDR (data not shown). The sizes of the RXR bands observed here correspond closely to those detected by Northern analysis of rat tissues (Yu et al., 1991;Mangelsdorfet al., 1992). These results support the evidence that RXR(s) are present in the retarded complexes formed by the VDR on the MOP VDRE (Fig. 7A).
Given the above DNA binding data and Northern blots, we analyzed the potential modulatory effects of the cognate ligand for the RXR receptors, g-cis-RA, on the response of the VDRE3tk promoter to 1,25-(OH)2 D3 and some of its analogs. Recent studies have suggested that 1,25-(OH)2 D3 and 94s-RA can stimulate transcription synergistically from a synthetic promoter containing a MOP VDRE (Carlberg et al., 1993). It was therefore of interest to determine if the VDR bound to different 1,25-(OH)2 D3 analogs responds differently to 9 4 s -RA. COS-7 and ROS 17/2.8 cells were treated with 50 I" 9 4 s -R A , a concentration that is sufficient for a specific response by RXRs (Heyman et al., 1992;Levin et al., 1992;Mangelsdorf et al., 1992). 50 I" 94s-RA alone had no effect on VDRE3tkCAT promoter activity in COS-7 or ROS 17/2.8 cells (Fig. 8, A and B , first and second lanes; and C and D, first and third lanes). In other experiments, the addition of 1 p~ 9-cis-RA had no effect on the VDRE3tkCAT promoter (data not shown). Strikingly, in both COS-7 or ROS 17/2.8 cells the addition of 50 I" 9-cis-RA had no significant effect on the response to 1,25-(OH)2 D3 of the VDRE3tkCAT promoter (Fig. 8, A and B, first, third, and fourth lanes; and C and D, first, second, and fourth lanes; see also Fig. 9). Similar results were obtained in ROS 17/2.8 cells using the VDREtkCAT promoter which contains only one VDRE (data not shown). The addition of 9-cis-RA had no effect on the response of VDRE3tkCAT to MC903 in ROS 17/23 cells (Fig. 8C,   lanes 5 and 6; Fig. 8D, fifih and sixth lanes). The action of 50 I" 94s-RA was tested further on the response to 1,25-(OH)2 D3 derivatives 1,25-(OH)2-16-ene-23-yne D3, 1,25-(OH)2-16-ene-23-yne-26,27-hexafluoro D3, and la-fluoro-25-(OH)-16-ene-23yne-26,27-hexafluoro D3 in ROS 17/23 cells (Fig. 9). Consistent with the results of Both the compound number and the nomenclature used in the text are shown for each derivative. cells with 1 p~ instead of 50 n~ 9-cis-RA and have obtained essentially identical results (data not shown).
The activities of 9-cis-RA and of RXRs in COS-7 and ROS 17/2.8 cells were tested using a promoter containing a single RXR response element composed of a direct repeat with 1-bp interrepeat spacing. This element has been shown to bind RXR homodimers in the presence of 9-cis-RA (Zhang et ul., 1992). The promoter is stimulated 3-fold by 50 n~ 9-cis-RA in transiently transfected COS-7 cells (Fig. 8E, lunes 1 and 3; Fig. 8F) and 2-fold in ROS 1712.8 cells (Fig. 8E, lunes 4 and 5; Fig. 8F). All-trans-RA is inactive on the FKRE-containing promoter in COS-7 cells at 50 nM (Fig. 8E, lunes 1 and 2 1. Strong responses were observed in both cell lines to 1 p~ 9-cis-RA using RXREcontaining promoters as well as a n analogous recombinant containing the response element from the RARP gene, FL4REP (data not shown). The results of Figs. 7 and 8 provide strong evidence for the presence of active RXR(s) in both COS-7 and ROS 17/2.8 cells and show that the 9-cis-RA is active and will stimulate a promoter containing a single RXR response ele-ment while not affecting an analogous promoter containing three VDREs.
Taken together the results of Figs. 1 and 7-9 suggest that although RXR(s) are active in both COS-7 and ROS 17/23 cells and apparently participate in binding of the VDR to the MOP VDRE, their cognate ligand 9-cis-RA does not contribute to the transcriptional response of promoters containing VDREs to 1,25-(OH)2 D3 or any of its derivatives.

DISCUSSION
A synthetic 1,25-(OH)2 D3-responsive expression vector has been constructed to analyze the transcriptional response to 1,25-(OH)2 D3 and a number of its derivatives. Expression is controlled by consensus VDREs and a truncated herpes simplex virus tk promoter, rendering the vector selectively responsive to 1,25-(OH)2 D3 and its derivatives. Vectors of this type have been used to study the transcriptional activity of a number of other nuclear receptors (Kumar et ul., 1987;Yu et ul., 1991). Transcription was analyzed in transiently transfected COS-7 and ROS 17/2.8 cells. In both cell lines a similar response was observed to 1,25-(OH)2 D3 and its biologically active derivative MC903 (Figs. 2 and 3; and data not shown). These results are consistent with previous studies using an ROS 17/ 2.8 cell line stably transfected with a CAT gene under control of a 1,25-(OH)2 D3-responsive osteocalcin promoter (Morrison and Eisman, 1991). This suggests that the use of a synthetic promoter and transient transfection is an efficient method for rapidly determining the transcriptional activity of 1,25-(OH)2 D3 and its derivatives in any cell line in which DNA can be introduced by transfection or a similar technique. It should be noted, however, that a strong response of the VDRE3tk promoter to ligand is obtained through the presence of three VDREs. The VDRE3tk promoter is not as complex as certain 1,25-(OH)2 D3-responsive promoters such as the osteocalcin promoter which contains a single response element over 400 bp from the site of transcriptional initiation and binding sites for several other transcription factors that can synergize with the VDRE of the osteocalcin promoter and contribute to the overall response to 1,25-(OH)2 D3 (Lian et al., 1989;Terpening et al., 1991;Stein and Lian, 1993). These factors should be taken into account in comparing the results obtained with simple synthetic and more complex physiological promoters.
The VDRE3tk promoter has been used to test the transcriptional response to 1,25-(OH)2 D3 and 10 of its derivatives, many of them potentially clinically useful. A number of observations can be made based on the results of Figs. 5 and 6 and Table I. First, the peak transcriptional activation observed did not vary significantly among 1,25-(OH)2 D3 and any of its active derivatives, indicating that the transcriptional activity of the ligandbound VDR was similar in each case. In other words, none of the compounds tested gave rise to a superactivated receptor. Second, the introduction of multiple bonds to 1,25-(OH)2 compounds at positions 16 and/or 23 invariably gave rise to derivatives that stimulated CAT activity at 10-to more than 100-fold lower concentrations than 1,25-(OH)2 D3. These results are consistent with numerous observations that many of the derivatives tested here as well as related compounds are more potent inhibitors of cellular replication than 1,25-(OH)2 D3 (Tanaka et al., 1984;Eisman et al., 1986;Zhou et al., 1990Zhou et al., , 1991. Derivatives with low affinities for the VDR were weak activators. However, for those derivatives with a moderate to high affinity for the receptor there was no strict correlation between the affinity of a given derivative for the VDR and the concentration at which half-maximal CAT activity was observed ( Figs. 5 and 6). This indicates that parameters other than affinity for the receptor also control the transcriptional activity of a given derivative.
There are several possible explanations for the observations that 1,25-(OH)2 D3 derivatives activate transcription highly potently. The high affinity of 1,25-(OH)2-16-ene D3 for the VDR (Table I) suggests that introduction of a double bond at position 16 apparently fLvs the side chain in a conformation that is similar to that bound by the receptor. In addition, recent studies have shown that both 1,25-(OH)2 D3 and 16-ene derivatives are metabolized to -24-oxo compounds. The 16-ene-24-oxo metabolite is stable in cells, whereas 1,25-(OH)z-24-oxo D3 is rapidly degraded to calcitroic acid. Interestingly, although calcitroic acid is inactive, 1,25-(OH)z-16-ene-24-oxo D3 is more potent than its parent compound in stimulating differentiation of the HL60 promyelocytic leukemia cells.2 Some derivatives may be metabolized to as yet unidentified compounds that have an unusually high affinity for the VDR or high stability. This could render some compounds selectively active depending on the metabolic products formed in a given cell type. Further substitution of 16-ene derivatives at positions 23 and 26 or 27 generates compounds that are more resistant to inactivation by metabolism. Introduction of multiple bonds at position 23 or fluorination at positions 26 and 27 provides resistance to inactivation by hydroxylation (Ikekawa, 1983). A 1,25-(OH)2 D3 derivative hexafluorinated at positions 26 and 27 is more potent and has longer lasting biological effects than its parent compound (Tanaka et al., 1984). It is also possible that derivatives are present in cells at higher concentrations because they are bound with less affinity than 1,25-(OH)2 D3 by component(s) of serum. We note in this regard that the curves of activation of CAT activity obtained with 1,25-(OH)2 D3 are not significantly affected if experiments are performed in 2% or 10% serum (data not shown). Finally the potency of a given derivative could be dependent on the specific components of vitamin D3-dependent transcription complexes which may vary among different normal cell types and between normal and transformed cells.
Derivatives lacking a la-hydroxyl group are generally inefficent activators (Fig. 6). The synthetic derivative 25-(OH)-16,23E-diene and 24,25-(OH)2 D3 both stimulated CAT activity equivalent to half-maximal activity observed with 1,25-(OH)2 D3 at 200 n~ or higher. Although 25-(OH)-16,23E-diene was a  ["H11,, D3 binding were taken from Uskokovic et al. (1991). EC,, values were taken from the plots of Figs. 5 and 6 as the concentration Transcriptional activation values followed by an asterisk represent those obtained from curves that did not reach a peak. Assays of competition of 1,25-(OW2 D3 or derivative giving half-maximal stimulation of CAT activity. weak activator, its 1,25-dihydroxy counterpart was almost 100fold more potent than 1,25-(OH)2 D3 (Fig. 5). This raises the possibility that derivatives lacking a la-hydroxyl group may be selectively active in tissues expressing both the VDR and high levels of the la-hydroxylase enzyme which would convert a 25-(OH) compound into the more active 1,25-(OH)2-derivative. Substitution of the la-hydroxyl group of a derivative with a fluoro group gave rise to a potent activator in the case studied. The derivative la-fluoro-25-hydroxy-26,27-hexafluoro-16-ene-23-yne stimulated CAT activity 10-fold more potently than 1,25-(OH)2 D3, although it was a less potent activator than its 1,25-dihydroxy counterpart (Fig. 6). In contrast, lp-fluor0-25-(OH)~-l6-ene-23-yne was a very weak activator, consistent with its low affinity for the VDR (Fig. 6 and Table I   and evidence was provided that RXRS present in both COS-7 and ROS 17/2.8 cells heterodimerize with the VDR (Figs. 1  and 7). Transcriptional responses were observed in both cell lines with a promoter containing a single RXRE to both 50 ~1 1 and 1 Scis-RA, the cognate ligand for RXRs ( Fig. 8 and data not shown). Taken together these results demonstrate the presence of functional RXR(s) in these cells which can form heterodimers with the VDR on the MOP VDRE. Several experiments were performed in both cell lines to test the effect of 94s-RA alone and in combination with 1,25-(OH)2 D 3 on transcription from the VDRE3tkCAT recombinant. We failed to observe a contribution of 9-cis-RA to the hormone response of the VDRE3 promoter, although a slight inhibitory effect of 94s-FtA on the response to 1,25-(OH)2 D3 was observed in some experiments in both cell lines ( Fig. 8 and data not  shown). Similar experiments in ROS 17/2.8 cells provided no evidence for modulation by 94s-RA of the response to several active 1,25-(OH)2 D3 derivatives (Fig. 9). In addition, increas-ing the concentration of 94s-RA from 50 n~ to 1 p~ had no significant effect on the response to 9-cis-RA (data not shown). These data indicate that under the conditions used here 9 4 s -RA does not contribute significantly to the transcriptional response to 1,25-(OH)2 D3.
Recent studies have shown that 94s-RA activated a synthetic promoter containing the MOP VDRE both in the absence and presence of 1,25-(OHI2 D3 in transiently transfected Drosophila SL-3 cells, which lack endogenous RXRs (Carlberg et al., 1993). The activation of a promoter containing a VDRE by 94s-RA alone, apparently through a heterodimer, suggests that under certain conditions this ligand may activate transcription from promoters containing a wide variety elements that bind RXR-containing heterodimers, including those responsive to all-trans-RA and thyroid hormone.
These observations are in apparent contrast to the results presented here. There are, however, several differences between the two studies. Our experiments were performed using endogenous levels of both VDRs and RXRs present in COS-7 and ROS 1712.8 cells, which are very likely to be lower than those obtained in transiently transfected cells. It cannot be argued that the levels of RXRs are too low to observe a response to 9-cis" in the cells tested here. Our evidence and that of others suggest that the VDR binds to the MOP VDRE and activates as a heterodimer with RXRs. This indicates that RXR levels are sufficient to support a strong transcriptional response to 1,25-(OH)2 Ds. We have also observed activation of the VDRE3tkCAT promoter by 94s-RAalone in the presence of a transiently transfected RXR expression vector in both COS-7 and ROS 17/23 cells (data not shown), suggesting that high concentrations of RXR are required for 9-cis-"dependent activation of promoters containing VDREs. RXR homodimers bound in vitro, albeit with low affinity, to a response element composed of a direct repeat of PurGTTCA motifs separated by 3 bp . It is possible, then, that the response of a promoter containing a VDRE to 9-cis-RA could be caused, at least in part, by the action of ligand-activated RXR homodimers. Carlberg et al. (1993) used a promoter containing a single VDRE, whereas VDRE3tkCAT contains three elements. However, we have performed experiments with 94s-RA in ROS 17/23 cells with VDREtkCAT which contains only one element and have obtained results similar to those presented in Fig. 8 (data not shown). Finally, there may exist intracellular factors known as coactivators or adapters (Lewin, 1990) which specifically link VDR/RXR heterodimers activated by 9-cis-RA to the transcription apparatus which are present in SL-3 cells but absent in the cell lines used here, raising the possibility that the response of VDR/RXR heterodimers to 94s-RA may be cell-specific.
Our results suggest that 1,25-(OH)2 D3 and 94s-RA activate transcription by separate pathways in ROS 1712.8 and COS-7 cells. According to this hypothesis, promoters containing response elements like the MOP VDRE are bound selectively by heterodimers of the VDR and RXR(s) which are activated by 1,25-(OH)2 D3, but not significantly affected by 9-cis-RA. Promoters containing RXREs would be bound by RXR homodimers and activated by 9-cis-RA. It remains to be seen how generally this scheme can be applied to the wide variety of both normal and transformed cell types expressing the VDR.