Sp1 and NF-Y Synergistically Mediate the Effect of Vitamin D3 in the p27Kip1 Gene Promoter That Lacks Vitamin D Response Elements*

Vitamin D3 promotes myeloid leukemic cell lines to differentiate terminally into monocytes/macrophages. It has been reported that overexpression of thecdk inhibitor p27Kip1 results in the differentiation of the myelomonocytic U937 cell line and that this gene is the target of vitamin D3. To identify the sequences required for the positive regulation of p27Kip1transcription by vitamin D3, a 3.6-kilobase 5′-flanking region of the human p27Kip1 gene was examined by transiently transfecting luciferase reporter constructs into U937 cells. The transcriptional activity of this construct was activated by vitamin D3. Deletion and mutational analysis revealed that both a GGGCGG sequence (−545/−539) and a CCAAT sequence (−525/−520) were necessary to induce p27Kip1 gene expression. Importantly, the region containing both of these elements conferred positive responsiveness to vitamin D3 to a heterologous promoter. Gel shift assays showed that Sp1 binds to the GGGCGG sequence and that NF-Y binds to the CCAAT sequence. Consistent with the roles of these transcription factors, treatment with vitamin D3stimulated the DNA binding activities of these factors to each element and induced the change of one NF-Y subunit. We conclude that vitamin D3 stimulates transcription of the p27Kip1 gene by a novel mechanism involving Sp1 and NF-Y, but not the vitamin D receptor, during the early stages of U937 cell differentiation.

Vitamin D 3 promotes myeloid leukemic cell lines to differentiate terminally into monocytes/macrophages. It has been reported that overexpression of the cdk inhibitor p27 Kip1 results in the differentiation of the myelomonocytic U937 cell line and that this gene is the target of vitamin D 3 . To identify the sequences required for the positive regulation of p27 Kip1 transcription by vitamin D 3 , a 3.6-kilobase 5-flanking region of the human p27 Kip1 gene was examined by transiently transfecting luciferase reporter constructs into U937 cells. The transcriptional activity of this construct was activated by vitamin D 3 . Deletion and mutational analysis revealed that both a GGGCGG sequence (؊545/؊539) and a CCAAT sequence (؊525/؊520) were necessary to induce p27 Kip1 gene expression. Importantly, the region containing both of these elements conferred positive responsiveness to vitamin D 3 to a heterologous promoter. Gel shift assays showed that Sp1 binds to the GGGCGG sequence and that NF-Y binds to the CCAAT sequence. Consistent with the roles of these transcription factors, treatment with vitamin D 3 stimulated the DNA binding activities of these factors to each element and induced the change of one NF-Y subunit. We conclude that vitamin D 3 stimulates transcription of the p27 Kip1 gene by a novel mechanism involving Sp1 and NF-Y, but not the vitamin D receptor, during the early stages of U937 cell differentiation.
1,25-Dihydroxyvitamin D 3 is not only a major regulator of mineral homeostasis but also a potent modulator of differentiation in several types of cells including monoblastic cells and osteoblasts (1). Recent studies have revealed that the cdk inhibitors p21 Cip1 and p27 Kip1 act as molecular switches that facilitate the vitamin D 3 -induced differentiation of the U937 myeloid leukemic cell line. These genes are regulated both at the transcriptional level and the post-transcriptional level by vitamin D 3 during the early stages of this process (2). Vitamin D 3 transduces its signal to the nucleus directly, mainly through a regulatable DNA-binding transcription factor, the vitamin D receptor (VDR) 1 (3), and ligand-inducible effects on differenti-ation are initiated through the direct activation of target genes by VDR. In fact, vitamin D 3 induces p21 Cip1 transcription in a VDR-dependent manner through a functional vitamin D response element (VDRE) in its promoter (2). To investigate the regulatory mechanisms of p27 Kip1 by vitamin D 3 , we previously characterized the human p27 Kip1 gene promoter (4). We found no canonical VDRE (5Ј-RGKTCANNNRGKTCA-3Ј) within a 3.6-kilobase 5Ј-flanking region upstream of the translation start site. This implies that p27 Kip1 is regulated by mechanisms that are different from those of other vitamin D 3 target genes, such as p21 Cip1 (2), c-fos (5), and osteocalcin (6). In addition, recent studies have shown the involvement of a VDRindependent mechanism in the regulation of vitamin D 3 -induced cell differentiation. For instance, monocyte differentiation is mediated by vitamin D 3 without requiring binding to VDR (7), and keratinocyte differentiation-related genes are stimulated by vitamin D 3 without the presence of VDRE (8). Clarification of the regulatory mechanisms of p27 Kip1 transcription is crucial for the understanding of the molecular mechanisms of vitamin D 3 action and for the understanding of the early processes during monocyte/macrophage differentiation. Although the post-transcriptional regulation of p27 Kip1 has been studied intensively (9 -11), very little is known about the transcriptional regulation of the p27 Kip1 gene.
In this study, we have analyzed the p27 Kip1 promoter in order to identify elements required for vitamin D 3 -induced up-regulation of transcription, which might reveal the early and novel mechanism of differentiation induction by vitamin D 3 .

Construction of the Luciferase Reporter Plasmid-
The promoter region of the human p27 Kip1 gene (p27PF) or enzyme-generated 5Ј-deletion sequences (p27Apa I, p27Afl II, p27No. 2, p27No. 12, p27No. 1, p27MB-435, and p27Sac II) were subcloned into the XhoI site of the pGL2 Basic vector (Promega) as described previously (4). The plasmids with point mutations in p27PF (p27mSp1-1, p27mSp1-2, and p27mCTF) were generated by in vitro mutagenesis as described previously (12). Furthermore, we generated constructs containing four tandem copies of the specific sequence of the p27 Kip1 promoter fused to a minimal promoter. A double-stranded 52-bp DNA fragment containing a 44-bp sequence corresponding to the Ϫ555/Ϫ512 region of the human p27 Kip1 promoter and linker sites (indicated by lowercase letters below) at both ends was generated from two oligonucleotides. The top strand (5Ј-agggAGCCTCGGCGGGGCGGCTCCCGCCGCCGCAACCAATGG-ATCTCC-3Ј) and the bottom strand (5Ј-ccctGGAGATCCATTGGTTGC-GGCGGCGGGAGCCGCCCCGCCGAGGCT-3Ј) were annealed, ligated, blunted, and subcloned into the SmaI site upstream of the SV40 early promoter in the PicaGene Promoter Vector 2 (Nippon Gene, Tokyo, Japan) in a forward or reverse orientation to generate PGPV2[Ϫ555/ Ϫ512wild] 4 Table I).
Cell Culture and Differentiation Induction-The human myelomono-cytic cell line U937 (a kind gift from Dr. Y. Honma at Saitama Cancer Center Research Institute) was maintained in RPMI 1640 supplemented with 10% (v/v) fetal bovine serum in an atmosphere of 5% CO 2 . Cells were induced toward monocytic differentiation in the presence of 1,25-dihydroxyvitamin D 3 (Wako, Tokyo, Japan). Luciferase Assay-Transfections into U937 cells were performed as described (13). Briefly, cells were electroporated at 250 V, 960 microfarads (Bio-Rad Gene Pulser; Bio-Rad), incubated for 15 min on ice, and then transferred to 20 ml of prewarmed RPMI 1640 containing 10% fetal bovine serum and divided equally into two cultures. They were incubated with either 1 ϫ 10 Ϫ7 M vitamin D 3 or 10 l of equivalent vehicle (ethanol). Cells were harvested 40 h after treatment, and preparation of extracts and the luciferase assay were performed using the Dual-Luciferase Reporter Assay (Promega), according to the manufacturer's instructions. All transfections included a reference sample with pGL2 Basic Vector or PicaGene Promoter Vector 2. For normalization of transfection efficiencies, 2 g of renilla (sea pangy) luciferase expression plasmid (pRL-TK, Promega) was included in the transfections. The experimental reporter luciferase activity was calculated by subtracting the intrinsic activity as measured by samples corresponding to the pGL2 Basic Vector or PicaGene Promoter Vector 2 and then normalized to transfection efficiency as measured by the activity deriving from pRL-TK.
Gel Shift Assay-Nuclear extracts of U937 cells were prepared according to the procedure of Andrews and Faller (14). The cells were treated with either 10 Ϫ7 M vitamin D 3 or vehicle for 36 h before extraction. Gel shift assays were carried out as described by Orita et al. (15). The reaction mixture for the gel shift assay (25-l final volume) contained 20 mM Tris-HCl (pH 8.0), 100 mM KCl, 10% glycerol, 1 g of poly(dI-dC), and 2 g of nuclear extract. After preincubation for 15 min at 23°C, probe DNA (approximately 0.5 ng, 10,000 cpm) was added to the mixture, and the binding reaction was allowed to proceed at 23°C for 20 min. The product was then resolved by electrophoresis on 4% acrylamide:bisacrylamide (29:1), 0.5 ϫ TBE nondenaturing gel at 10 V/cm for 150 min. Competition analyses were performed by mixing the indicated amount of appropriate competitor DNA to the binding reaction prior to addition of nuclear extracts. In supershift experiments, antibodies against Sp1 (sc-59X; Santa Cruz), Sp3 (sc-644X; Santa Cruz), and C/EBP-␣ (sc-7204X; Santa Cruz) were purchased. The antibodies against NF-YA, -YB, and -YC (IgG fraction) were kindly provided by Drs. T. Orita and S. Nagata at Osaka University Medical School. These antibodies were added to the incubation mixture containing nuclear protein before the addition of probe DNA.
Western Blotting-U937 cells treated with either 10 Ϫ7 M vitamin D 3 or vehicle at different time points were harvested. Cells (3 ϫ 10 7 ) were washed in cold phosphate-buffered saline twice and then resuspended in 150 l of lysis buffer (50 mM Tris-HCl (pH 7.9), 150 mM NaCl, 0.1% SDS, 0.5% deoxycholate, 1% Nonidet P-40, 10% glycerol, 0.2 mM phenylmethylsulfonyl fluoride, and 10 g/ml aprotinin). This was subjected to mild sonication and used as whole cell extracts. Extracts (from 1 ϫ 10 6 cells) were subjected to 12% SDS-polyacrylamide gel electrophoresis and blotted onto polyvinylidene difluoride membranes (Millipore). The membranes were incubated with the primary antibodies and then incubated with horseradish peroxidase-conjugated secondary antibody. The immune complexes were visualized using an enhanced chemiluminescence system (Amersham Pharmacia Biotech). The antibodies against NF-YA, -YB, and -YC and the antibody against Sp1 (sc-59) were used at a 1:1,000 and 1:200 dilution in blocking buffer (3% milk powder in phosphate-buffered saline), respectively.
Analysis of NF-YA mRNA-Total RNA was prepared using the TRIZOL TM (Life Technologies, Inc.)/chloroform method from U937 cells treated with either 10 Ϫ7 M vitamin D 3 or vehicle at different time points. Reverse transcriptase PCR was performed using the RNA PCR Kit Ver. 2.1 (Takara, Tokyo, Japan). The forward (F) and reverse (R) primers for NF-YA-short and -long were AATAGTTCGACAGAGCAGATTG (primer F1), CCTCCTGATTGGGTTTCGGAGT (primer F2), and GGGGTTAG-GACACTCGGATGAT (primer R). The forward and reverse primers for glyceraldehyde-3-P-dehydrogenase were ACCACAGTCCATGCCTCA and TCCACCACCCTGTTGCTGTA. 1 g of total RNA was used for reverse transcriptase PCR. PCR was performed for 25 and 20 cycles to detect NF-YA subunits and glyceraldehyde-3-P-dehydrogenase, respectively. PCR products were subjected to electrophoresis in a 3% agarose gel. The primers for NF-YA were designed to detect two different isoforms simultaneously and were located in regions that were conserved in the two isoforms (16).

Vitamin D 3 -responsive Elements in the Human p27 Kip1
Promoter-It has been reported that p27 Kip1 and p21 Cip1 are transcriptionally induced by vitamin D 3 (2). We also observed that p27 Kip1 mRNA was induced after treatment of U937 cells with vitamin D 3 , and it peaked between 24 h and 48 h (approximately 4-fold compared with 0 h), whereas p21 Cip1 mRNA induction was more rapid (data not shown). This suggests that vitamin D 3 regulates transcription of p27 Kip1 in a VDR/VDRE-independent manner unlike the case of p21 Cip1 transcription. To investigate the regulatory mechanisms behind p27 Kip1 gene expression, we first investigated the effect of vitamin D 3 on the transcriptional activity of the promoter of the p27 Kip1 gene. The effect of vitamin D 3 on the wild type p27 Kip1 promoter-lucif- Ϫ555/Ϫ512wild

5Ј-CGCCGCAACCTTTGGATCTCC-3Ј
This study Regulation of p27 Kip1 Transcription by Vitamin D 3 erase fusion plasmid, p27PF, was examined by transient transfection. Following a 40-h exposure to vitamin D 3 , the luciferase activity from p27PF plasmid was increased approximately 3-4fold of that of the vehicle-treated control (Fig. 1). This result was consistent with our observation of the effect of vitamin D 3 on p27 Kip1 expression by Northern analysis, indicating that this 3.6-kilobase promoter fragment was necessary and sufficient for the response of p27 Kip1 gene to vitamin D 3 . Next, we tried to determine whether any particular regions in the 3.6kilobase fragment were responsive to vitamin D 3 . For this purpose, a series of 5Ј-deletion constructs of the p27 Kip1 promoter was examined. Fig. 1 shows that deletion up to position Ϫ549 (relative to the translation start site) did not result in significant changes in the response to vitamin D 3 and that the responsiveness was abolished completely using deletions up to Ϫ511, whereas deletions up to Ϫ311 exhibited some promoter activities in the absence of vitamin D 3 . These results indicated that potential vitamin D 3 regulatory elements appeared to be located between Ϫ549 and Ϫ511.
The region between Ϫ549 and Ϫ511 harbored two Sp1 sites (Ϫ544 and Ϫ534) and a CCAAT box (Ϫ522) which were adjacent to each other and were conserved between the human and mouse p27 Kip1 promoters (4,17). In this study, we termed the two upstream Sp1 sites Sp1-1 and Sp1-2 (Fig. 2). To determine whether the two Sp1 sites and the CCAAT box were involved in activation by vitamin D 3 , a series of mutants of p27PF with mutations in the Sp1-1 site, the Sp1-2 site, or the CCAAT box was constructed and termed p27mSp1-1, p27mSp1-2, and p27mCTF, respectively (Fig. 2). As shown in Fig. 2, the response to vitamin D 3 was abolished using p27mSp1-1 and p27mCTF but not p27mSp1-2. On the other hand, all mutants retained some promoter activities in the absence of vitamin D 3 . Therefore we concluded that at least both the Sp1-1 and the CCAAT box were the vitamin D 3 -responsive elements and that the Sp1-2 site was not involved in the activation by vitamin D 3 .
A  4 . We concluded that the region between Ϫ555 and Ϫ512 relative to the translation start site (Ϫ118 and Ϫ75 relative to the transcription start site 2 ) was sufficient for vitamin D 3 -induced transcription of the p27 Kip1 gene, although we could not exclude the possibility that other sequences in the p27 Kip1 promoter were also required. Furthermore, both the Sp1-1 site and the CCAAT box in this region were vitamin D 3 -responsive elements and were required for vitamin D 3 -induced transcription of the p27 Kip1 gene. min D 3 -responsive Sequence-To identify the nuclear factors binding to the vitamin D 3 -responsive sequence, a set of oligonucleotides spanning Ϫ555 to Ϫ512 was used as a probe for gel shift assays (Ϫ555/Ϫ512wild, see Table I). Nuclear extracts were prepared from U937 cells treated with vitamin D 3 for 36 h. As shown in Fig. 4A, the oligonucleotides Ϫ555/Ϫ512wild yielded a single major retarded band (lane 1), which was competed away by an excess of unlabeled oligonucleotide (lanes 2  and 3). To localize the sequence that binds to nuclear factor(s), a series of oligonucleotides that carried point mutations in the Sp1-1, the Sp1-2, or the CCAAT box (Ϫ555/Ϫ512mSp1-1, Ϫ555/ Ϫ512mSp1-2, and Ϫ555/Ϫ512mCTF, respectively), oligonucleotides carrying canonical wild type or a mutated sequence for the sequence Sp1 site or CCAAT box (Sp1wild, Sp1mt, and NF-Ywild), and oligonucleotides spanning Ϫ534 to Ϫ512 (Ϫ534/Ϫ511) were used as competitors. As shown in lanes 4 -17, the retarded band was not competed out by the addition of Ϫ555/Ϫ512mSp1-1, Sp1mt, NF-Ywild, or Ϫ534/Ϫ512wild, indicating that Sp1 family protein(s) bind to the Sp1-1 site. To elucidate whether the retarded band represents the binding of Sp1 or Sp3, gel shift assays were performed with nuclear extracts that were preincubated with anti-Sp1 or -Sp3 antibody for band supershift experiments (18). As shown in Fig. 4B, in the presence of anti-Sp1 antibody but not anti-Sp3 antibody, the complex was supershifted. We concluded that Sp1 binds to the Sp1-1 site of the vitamin D 3 regulatory region of the p27 Kip1 promoter in U937 cells.

Identification of Nuclear Proteins
We also analyzed the sequence between Ϫ534 and Ϫ512 which carries the CCAAT box but not the Sp1-1 and Sp1-2 sites. A set of oligonucleotides spanning this region was used as a probe for gel shift assays. As shown in Fig. 5A, we observed a single major retarded band, which was competed away by excess unlabeled wild type oligonucleotides (Ϫ534/Ϫ512wild) but not those carrying a mutation in the CCAAT box (Ϫ534/ Ϫ512mCTF) (lanes 1-5). This indicated that nuclear factor(s) bind to the CCAAT box of the p27 Kip1 promoter. So far, it has been reported that several different transcriptional factors including NF-Y, C/EBP, CAT-binding protein, and NF-I are capable of binding to CCAAT sequences (19 -22). To elucidate which transcription factor binds to the CCAAT box in the regulatory region of the p27 Kip1 promoter, we performed competition experiments using unlabeled oligonucleotides carrying the CCAAT sequence that had been reported to bind to NF-Y, C/EBP-␣, or NF-I with a high affinity (NF-Ywild, C/EBP-␣wild, and NF-Iwild, respectively). As shown in Fig. 5A, the retarded band was competed by the addition of NF-Ywild but not NF-Iwild or C/EBP-␣wild. This suggested that NF-Y binds to this CTF site. To confirm this result, we performed supershift assays using anti-NF-Y antibodies. NF-Y is composed of three subunits, NF-YA, NF-YB, and NF-YC. We used anti-NF-YA, -YB, and -YC antibodies (15,23) and anti-C/EBP-␣ antibody for supershift experiments. As shown in Fig. 5B, in the presence of anti-NF-YA, -YB, or -YC antibodies, the complex was supershifted. The addition of anti-C/EBP-␣ did not affect the complex. These results demonstrated that trimeric NF-Y binds to the CCAAT box of the regulatory region of the p27 Kip1 promoter, although we could not observe NF-Y binding to Ϫ555/ Ϫ512wild probe. To verify that NF-Y could bind not only to the CCAAT box in the Ϫ534/Ϫ512wild sequence but also to the site in the Ϫ555/Ϫ512 wild sequence, we again used oligonucleotides spanning Ϫ555 to Ϫ512, with or without mutations in the Sp1-1, Sp1-2, or CTF site as competitors. For the same purpose, we investigated the effects of the addition of anti-NF-YA, -YB, or -YC antibodies on the formation of the complexes of Ϫ555/ Ϫ512wild probe and nuclear proteins by gel shift assays. The addition of anti-NF-YA, -YB, or -YC antibody did not affect the complexes of Ϫ555/Ϫ512wild and nuclear proteins (data not shown). However, as shown in Fig. 5A, NF-Y binding to the probe Ϫ534/Ϫ512wild was competed by Ϫ555/Ϫ512wild, Ϫ555/ Ϫ512mSp1-1, and Ϫ555/Ϫ512mSp1-2, although these oligonucleotides were less effective than Ϫ534/Ϫ512wild, while NF-Y binding to Ϫ534/Ϫ512wild was not affected by Ϫ555/ Ϫ512mCTF. These results indicated that NF-Y binds to the CCAAT box in Ϫ555/Ϫ512wild as well as the site in Ϫ534/ Ϫ512wild. Therefore, we concluded that Sp1 and NF-Y bind to the Sp1-1 site and the CCAAT box, respectively, the regulatory elements that are required for vitamin D 3 -induced transcription of the p27 Kip1 gene.
To confirm further the involvement of NF-Y in the vitamin D 3 -induced transcription of p27 Kip1 , we cotransfected p27PF or p27mCTF with a dominant negative NF-YA mutant expression plasmid (pNF-YA29) (24). As shown in Fig. 6, pNF-YA29 suppressed the vitamin D 3 -induced luciferase activities from p27PF but not p27mCTF in a dose-dependent manner. This result demonstrates directly that NF-Y mediates the up-regulation of p27 Kip1 transcription by vitamin D 3 via the CTF site of its promoter.
Analysis of Sp1 and NF-Y Subunits after Vitamin D 3 Treatment-The results described above suggested strongly that Sp1 and NF-Y are the essential regulators in vitamin D 3 -induced transcription of the p27 Kip1 gene. To investigate the mechanism how Sp1 and NF-Y regulate transcription of the p27 Kip1 gene, we examined whether binding of Sp1 and NF-Y to the p27 Kip1 promoter was altered following vitamin D 3 treatment, using gel shift assays. Nuclear extracts were prepared from U937 cells treated with either vitamin D 3 or vehicle alone for 36 h. As a probe, Ϫ555/Ϫ512wild oligonucleotides or Ϫ534/ Ϫ512wild oligonucleotides were used to detect Sp1 and NF-Y, respectively. As shown in Fig. 7, we observed that Sp1 and NF-Y binding activities increased significantly after vitamin D 3 treatment. Because the DNA binding activity of an unrelated transcription factor, NF-I, was not changed by the treatment of vitamin D 3 , we concluded that the treatment of vitamin D 3 specifically stimulates the binding of Sp1 and NF-Y to the p27 Kip1 promoter. To analyze the protein levels of Sp1 and NF-Y subunits, whole cell extracts were prepared from U937 cells treated with vitamin D 3 or vehicle alone at different time points. The amounts of these proteins were analyzed by Western blotting. As shown in Fig. 8, the level of Sp1 increased slightly from 12 h to 18 h after the vitamin D 3 treatment, and after 36 h the level of Sp1 was reduced and barely detectable. The increase in the level of Sp1 occurred prior to the increase of p27 Kip1 mRNA following the vitamin D 3 treatment (data not shown). Interestingly, the level of Sp1 protein showed little difference between vitamin D 3 and vehicle-treated cells after 36 h, when we could observe a significant increase in the binding of Sp1 to the Sp1-1 site of the human p27 Kip1 promoter (Fig. 7). These results indicate that post-translational regulation of Sp1 such as phosphorylation (25) and glycosylation (26) could contribute to up-regulation of p27 Kip1 transcription by vitamin D 3 in combination with slight induction of Sp1 protein.
On the other hand, as reported previously, we also observed two bands for NF-YA (36 kDa and 39 kDa) which are thought to result from differential splicing (15,16,27) (Fig. 8). To our surprise, the level of the 39-kDa form of NF-YA decreased from 12 h to 48 h after the vitamin D 3 treatment. To confirm that this change resulted from the decrease of long form NF-YA mRNA, we performed reverse transcriptase PCR analysis of NF-YA mRNA using primers F1 and R that were located in the regions that are conserved in the two isoforms. As shown in   8B, we obtained two bands that were derived from two isoforms of NF-YA mRNA judging from restriction enzyme analysis (data not shown). As expected, the long form of NF-YA decreased gradually in vitamin D 3 -treated cells. The same results were obtained with a different combination of PCR primers (primer F2 and R). These results suggested that a decrease of the band for 39-kDa NF-YA protein resulted from the decrease of its mRNA and not protein modification. This decrease in the low molecular weight form of NF-YA was observed prior to the increase of p27 Kip1 mRNA after vitamin D 3 treatment. A significant change was not seen in the level of NF-YB and NF-YC between vitamin D 3 and vehicle-treated cells (Fig. 8A). Taken together, these results suggest that NF-Y acts as a mediator of vitamin D 3 in the regulation of p27 Kip1 transcription via the decrease in the high molecular weight form of NF-YA, although the precise mechanism remains to be solved. DISCUSSION In this study, we analyzed the vitamin D 3 -induced transcription of p27 Kip1 as a model to understand a possible novel pathway of vitamin D 3 action which does not directly involve VDR/VDRE and the mechanism of U937 differentiation. From transient transfection studies, we conclude that Sp1 and NF-Y mediate the vitamin D 3 -induced transcription via elements that are closely located adjacent to each other in the promoter region of p27 Kip1 . Importantly, the 44-bp element in the p27 Kip1 promoter that carries Sp1 and NF-Y binding sites is sufficient for the response to vitamin D 3 , and both elements are required for this response. We believe that this is the first report that explains the molecular mechanism of vitamin D 3 -induced transcription that does not directly require VDR, although vitamin D 3 -dependent transcriptional repression that does not require VDR has been reported previously (28,29).
The next question to be solved was how vitamin D 3 activated the p27 Kip1 gene expression via Sp1 and NF-Y. Binding activities of Sp1 and NF-Y to each element in the p27 Kip1 promoter were stimulated significantly after vitamin D 3 treatment. Furthermore, Western blotting analysis showed that Sp1 increased slightly, and one subunit of NF-Y changed to a low molecular weight form prior to the accumulation of p27 Kip1 mRNA. These findings raise the hypothesis that post-translational modification of Sp1 and differential splicing of NF-YA induced by vitamin D 3 lead to the activation of p27 Kip1 transcription through enhanced binding of these factors to regulatory elements in the p27 Kip1 promoter. It has been shown that Sp1 can mediate responses to several inducers of myeloid differentiation, and several myeloid promoters are dependent on a functional Sp1 site (30 -32). NF-Y also can mediate responses to several inducers of myeloid differentiation and macrophage maturation (15,33,34), However, little is known about the mechanism of how NF-Y mediates these responses. We found that the form of NF-YA changes after differentiation induced by vitamin D 3 . Existence of two isoforms of NF-YA, which result from differential splicing, has been reported in human and mouse (16,27). The functional difference between the two isoforms is not clear. However, there is a possibility that the two forms have different biological activities since two forms are expressed in various types of cells with a strong tissue-specific bias (16,27). Our data indicate that a switch in isoform resulting in the decrease of high molecular weight form of NF-YA protein may be involved in the transcriptional activation of p27 Kip1 by vitamin D 3 .
Another question to be solved is why both Sp1 and NF-Y are required for the vitamin D 3 -induced transcription of p27 Kip1 . The possible mechanism is the involvement of histone acetyltransferases, known as transcriptional cofactors including  6. A dominant-negative NF-YA expression plasmid (pNF-YA29) suppresses the vitamin D 3 responsiveness of the p27 promoter. 9 g of p27PF or p27mCTF was cotransfected into U937 cells with various amounts of expression plasmid for the dominant negative NF-YA (pNF-YA29) with 2 g of pRL-TK. Luciferase activities were analyzed after a 40-h treatment with 10 Ϫ7 M vitamin D 3 compared with p27PF without vitamin D 3 treatment. Data are shown as means (bars, standard deviation) (n ϭ 3). *, p Ͻ 0.01. p300, GCN5, and P/CAF (35). Recently it was shown that NF-Y interacts with p300 in vivo, and NF-Y establishes a pre-set promoter architecture that can facilitate transcription within chromatin by recruiting p300 protein (36). Similarly, NF-Y has been shown to be associated with GCN5 and P/CAF in vitro, and GCN5 activates NF-Y-dependent transcription in vivo (37). We reported previously that histone deacetylase inhibitors activate the p21 Cip1 gene promoter through Sp1 sites that can interact with Sp1 and Sp3 (18,38). Thus, it is possible that NF-Y activates p27 Kip1 transcription through modification of Sp1 sites by recruiting histone acetyltransferases.
In this study, we found a novel pathway to mediate vitamin D 3 action which does not directly involve VDR. We then searched for the genes that carry Sp1 and CCAAT sequence in their promoter regions. We found that both elements exist in the promoter region of VDR, whose transcription is also upregulated by vitamin D 3 (39,40). Moreover, no VDRE exists in the promoter region of the VDR gene. This suggests that the regulatory mechanism we found might be more generally applicable to other genes that are regulated by vitamin D 3 .
In U937 cell differentiation induced by vitamin D 3 , p21 Cip1 acts as a molecular switch to trigger U937 cell differentiation, and its transcription is up-regulated along with p27 Kip1 (2). In this experimental system, transcription of the p21 Cip1 gene is activated during the initial processes by VDR/VDRE. The present study shows that p27 Kip1 is activated at a slightly later stage by Sp1 and NF-Y in a cooperative manner. Such phased regulation of molecular switches by different transcriptional regulatory mechanisms will be important for the commitment event between growth and differentiation during development and cellular differentiation. Panel A, whole cell extracts were prepared from U937 cells at the indicated time points after the treatment of vitamin D 3 . Extracts were subjected to 12% SDSpolyacrylamide gel electrophoresis, followed by Western blotting with anti-Sp1 or with anti-NF-YA, -YB, -YC antibodies. Panel B, total RNAs were prepared from U937 cells at the indicated time after the treatment of vitamin D 3 . Reverse transcriptase PCR was carried out using different combinations of PCR primers. Top panel, primers F1 and R; middle panel, primers F2 and R; bottom panel, primers for glyceraldehyde-3-P-dehydrogenase.