RFX1 Mediates the Serum-induced Immediate Early Response of Id2 Gene Expression*

Id2, a negative regulator of basic helix-loop-helix transcription factors, is involved in regulating cell differentiation and proliferation. To obtain insight into the role of Id2 in cell cycle control, we investigated the mechanisms underlying the immediate early response of Id2 expression to serum stimulation in NIH3T3 cells. Luciferase reporter analysis with deletion and point mutants demonstrated the serum response element of Id2 (Id2-SRE) to be a consensus binding site for RFX1 (regulatory factor for X-box 1) present 3.0 kb upstream of the transcription initiation site of Id2. Gel shift and chromatin immunoprecipitation assays confirmed the binding of RFX1 to Id2-SRE in vitro and in vivo, respectively. In both assays, RFX1 binding was observed not only in serum-stimulated cells, but also in serum-starved cells. Knockdown of RFX1 by RNA interference disturbed the immediate early response of Id2 expression in cells and abrogated the Id2-SRE-mediated induction of luciferase activity by serum. These alterations were rescued by the introduction of RNA interference-resistant RFX1 into cells. On the other hand, in the Id2-SRE-mediated reporter assay, RFX1 with an N-terminal deletion abrogated the serum response, whereas RFX1 with a C-terminal deletion enhanced the reporter activity in serum-starved cells. Furthermore, HDAC1 was recruited to Id2-SRE in serum-starved cells. These results demonstrate that RFX1 mediates the immediate early response of the Id2 gene by serum stimulation and suggest that the function of RFX1 is regulated intramolecularly in its suppression in growth-arrested cells. Our results unveil a novel transcriptional control of immediate early gene expression.

Id2, a negative regulator of basic helix-loop-helix transcription factors, is involved in regulating cell differentiation and proliferation. To obtain insight into the role of Id2 in cell cycle control, we investigated the mechanisms underlying the immediate early response of Id2 expression to serum stimulation in NIH3T3 cells. Luciferase reporter analysis with deletion and point mutants demonstrated the serum response element of Id2 (Id2-SRE) to be a consensus binding site for RFX1 (regulatory factor for X-box 1) present 3.0 kb upstream of the transcription initiation site of Id2. Gel shift and chromatin immunoprecipitation assays confirmed the binding of RFX1 to Id2-SRE in vitro and in vivo, respectively. In both assays, RFX1 binding was observed not only in serum-stimulated cells, but also in serum-starved cells. Knockdown of RFX1 by RNA interference disturbed the immediate early response of Id2 expression in cells and abrogated the Id2-SRE-mediated induction of luciferase activity by serum. These alterations were rescued by the introduction of RNA interference-resistant RFX1 into cells. On the other hand, in the Id2-SRE-mediated reporter assay, RFX1 with an N-terminal deletion abrogated the serum response, whereas RFX1 with a C-terminal deletion enhanced the reporter activity in serum-starved cells. Furthermore, HDAC1 was recruited to Id2-SRE in serum-starved cells. These results demonstrate that RFX1 mediates the immediate early response of the Id2 gene by serum stimulation and suggest that the function of RFX1 is regulated intramolecularly in its suppression in growth-arrested cells. Our results unveil a novel transcriptional control of immediate early gene expression.
Id (inhibitor of DNA binding/differentiation) proteins, bearing a helix-loop-helix motif as their functional domain but lacking a DNA-binding domain, are negative regulators of basic helix-loop-helix transcription factors (1,2). Four members (Id1-Id4) constitute the protein family in mammals and play important roles in controlling cell fate determination, differentiation, and proliferation of various cell types in vitro and in vivo (1,2). In accord with their cell cycle stimulatory activity, Id proteins are considered to be required for the G 1 /S transition in the cell cycle (3,4), and Id2-deficient mice exhibit a severe proliferation defect of mammary epithelial cells during pregnancy (5). In addition, overexpression of Id genes and/or proteins has been reported in various human malignancies (1,2). In growth-arrested fibroblasts, in which Id mRNAs are barely detectable, mitogenic stimulation rapidly induces expression of Id genes (3,6,7). It has been reported that Id2 and Id3 are categorized into a group to which c-myc belongs in terms of the kinetics of mRNA expression in human fibroblasts induced by serum stimulation (8). In fact, Id3 was identified as an immediate early gene by differential screening (6). However, the exact mechanisms underlying Id gene induction upon mitogenic stimulation remain unclear, although Egr-1 (early growth response 1), an immediate early gene, has been shown to regulate the expression of Id1 (9) and Id3 (10).
Immediate early genes are those rapidly induced by growth stimulation without de novo protein synthesis (11,12). Because many of them encode transcriptional regulators or secreted signaling molecules, they are thought to be involved in gene activation cascades required for the progression of the cell cycle (13). Elucidation of the induction pathways of immediate early genes can advance our understandings of how cell proliferation is controlled and how tumors develop. Immediate early genes are subclassified into genes with fast and slow kinetics of induction after mitogenic stimulation (12). Examples of immediate early genes with fast and slow kinetics are c-fos and c-myc, respectively: c-fos mRNA exhibits its peak of expression at ϳ0.5-1 h after growth stimulation and returns to the base line within 2 h (14), whereas c-myc expression peaks at ϳ1-3 h after stimulation (15). Various transcription factors are known to be critical for inducing the expression of immediate early genes after mitogenic stimulation. Examples include the serum response factor and ternary complex factors such as Elk-1 (16). Although the induction mechanisms of several immediate early genes have been extensively studied, numerous immediate early genes have been reported, and the mechanisms of their transcriptional regulation largely remain to be determined.
Here we report that the rapid induction of Id2 gene expression in murine fibroblasts by serum stimulation is mediated by RFX1 (regulatory factor for X-box 1). RFX1 is a transcription factor involved in the induction of gene expression of viruses such as human hepatitis B virus and cytomegalovirus (17) and the suppression of cellular genes such as c-myc (18) and the proliferating cell nuclear antigen (19). Our work reveals a novel molecular mechanism underlying the regulation of immediate early gene expression.
Cell Culture and Transfection-NIH3T3, HeLa, and TIG3 (human diploid fibroblasts; Japanese Collection of Research Bioresources, Tokyo, Japan) cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Sigma) supplemented with 10% (v/v) fetal bovine serum (FBS; MP Biomedicals, Eschwege, Germany), 100 g/ml streptomycin, and 100 units/ml penicillin at 37°C in a 5% CO 2 humidified atmosphere. For synchronization of the cell cycle, cells were serumstarved by culture in DMEM containing 0.2% FBS for 48 h. Cycloheximide was purchased from Sigma. More than 85% of the serum-starved cells were confirmed to be in G 0 or G 1 phase by flow cytometry. To stimulate cells with serum, the low serum medium was changed to DMEM containing 20% FBS, and cells were maintained in this medium until used. DNA transfection was carried out using TransIT reagent (Mirus Bio Corp., Madison, WI) according to the manufacturer's instructions.
Flow Cytometry-Cells were collected by trypsinization, fixed with 70% ethanol at Ϫ20°C for 1 h, and incubated with 25 g/ml propidium iodide (Sigma) and 100 g/ml RNase in phosphate-buffered saline supplemented with 0.1% bovine serum albumin for 20 min. The cells were then analyzed using a FACSCalibur (BD Biosciences) with ModFit LT3 software.
Northern Blot Analysis-Total RNA was extracted from cultured cells using an RNeasy kit (Qiagen Inc., Valencia, CA). Five micrograms of total RNA were separated by electrophoresis on a formaldehyde-containing 1.0% agarose gel, transferred onto a nylon filter, and cross-linked by UV irradiation. Radioactive DNA probes for Id2, c-myc, and Rfx1 and 36B4 rRNA were prepared by random primer labeling of the respective cDNAs with [␣-32 P]dCTP. Hybridization and washing were performed according to standard procedures. Hybridized filters were exposed to a BAS-2000 imaging plate (Fujifilm, Tokyo, Japan) to quantify the radioactivity. 36B4 rRNA served as a loading control.
Luciferase Reporter Assay-NIH3T3 cells were plated in a 24-well plate at a density of 1 ϫ 10 5 cells/well. After 24 h, cells were transfected with reporter plasmids for 4 h, and serum starvation was started by changing the medium to 0.2% FBScontaining DMEM. After serum starvation for 48 h, cells were stimulated with serum by changing the low serum medium to 20% FBS-containing DMEM. Before and 4 h after serum stimulation, cell lysates were prepared, and their luciferase activities were measured as described previously (26) using the Dual-Luciferase reporter assay system (Promega Corp.) according to the manufacturer's instructions with a Luminescencer PSN (Atto Corp., Tokyo, Japan). The Renilla luciferase plasmid pRL-TK (Promega Corp.) was cotransfected as an internal control. The magnitude of the serum response was determined by calculating the -fold induction of luciferase activity before and 4 h after serum stimulation. Assays were independently carried out at least three times in triplicate. The coefficients of variation (defined as the ratio of the S.D. to the mean) of luciferase activities in arrested cells were Ͻ0.3, indicating that the luciferase activities in arrested cells were accurately and reproducibly measured.
Data Analysis-Data are expressed as the means Ϯ S.E. Differences between groups were analyzed using either Student's t test or one-way analysis of variance, followed by the Student-Newman-Keuls test. p Ͻ 0.05 was considered statistically significant.
Electrophoretic Mobility Shift Assay (EMSA)-Preparation of nuclear extracts of NIH3T3 cells and EMSA were done as described previously (24). The protein content was determined by the Bradford method using bovine serum albumin as a standard. Five micrograms of nuclear extracts were incubated with 20000 cpm of [␣-32 P]dCTP-labeled oligonucleotides in 20 l of 20 mM HEPES (pH 7.9), 50 mM KCl, 0.5 mM EDTA, and 1 mM dithiothreitol (21) at room temperature for 30 min and then separated on a 6.0% polyacrylamide gel. After drying, the gel was exposed to x-ray film or a BAS-2000 imaging plate. The sequences of the oligonucleotides containing the consensus RFX1-binding site were as follows: wild-type Id2, 5Ј-GTGACGGCGCGGTTGCTATGGCAG-CCG-3Ј; mutant Id2, 5Ј-GTGACGGCGCGGTTGCT-GGTGCAGCCG-3Ј; and major histocompatibility complex (MHC) class II HLA-DR (20), 5Ј-GATCCTTCCCCTAG-CAACAGATA-3Ј. A 200-fold excess of unlabeled oligonucleotide was added to the reaction as an unlabeled competitor. For the supershift assay, 2 g of anti-RFX1 (I-19), anti-RFX2 (C-15), or anti-RFX3 (T-17) antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or anti-RFX5 antibody (193 or 194; Rockland Immunochemicals, Inc., Gilbertsville, PA) were added to the reaction mixture. In vitro translation of the human RFX1 protein was done using pSG5-HARFX1 (20) and the TNT protein expression system (Promega Corp.), and 1% of the product was used for the binding reaction.
Western Blot Analysis and Immunoprecipitation-Total cell extracts were prepared by incubating cells in lysis buffer (25 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1.0% Triton X-100, 1 mM EDTA, 10% glycerol, 1 mM dithiothreitol, and protease inhibitor mixture) at 4°C for 30 min and centrifuging at 17,000 ϫ g for 20 min at 4°C as described (27). Fifty micrograms of proteins were separated by 8 or 15% polyacrylamide gel electrophoresis. The resolved proteins were transferred onto nitrocellulose membranes (GE Healthcare). The membranes were blotted with anti-RFX1 or anti-␤-actin (Sigma) antibody. Protein blots were visualized using the ECL Plus chemiluminescence system (GE Healthcare) on an LAS-3000 image analyzer (Fujifilm). For immunoprecipitation, 500 g of whole cell lysates were incubated with anti-HDAC1 (histone deacetylase-1) antibody (H-51; Santa Cruz Biotechnology, Inc.) or anti-RFX1 antibody for 1 h at 4°C and then incubated with protein A-or G-Sepharose (GE Healthcare) at 4°C for 3 h with rotation. The precipitated immunocomplex was released by boiling in 2ϫ SDS electrophoresis sample buffer and subjected to Western blot analysis as described above.
Chromatin Immunoprecipitation (ChIP) Assay-The ChIP assay was done as described previously (24,28). In brief, NIH3T3 cells were cross-linked with 1.0% formaldehyde in the medium for 2 min at 37°C, rinsed twice with ice-cold phosphate-buffered saline, resuspended in 1.0 ml of SDS lysis buffer (28), and incubated on ice for 10 min. Cell lysates were then sonicated with a Bioruptor (Cosmo Bio Co., Ltd., Tokyo, Japan), and the sonicated lysates were precleared as described (24,28). Immunoprecipitation was performed with 5 g of anti-RFX1, anti-HDAC1, or anti-CAAT/enhancer-binding protein-␤ (C/EBP␤) polyclonal (C-19; Santa Cruz Biotechnology, Inc.) antibody at 4°C for 2 h. Normal rabbit IgG (Sigma) served as a negative control. Immunocomplexes were then mixed with 30 l of protein A-Sepharose at 4°C for 2 h and washed sequentially with low salt buffer, high salt buffer, LiCl wash buffer (28), and TE (Tris/EDTA) buffer. Immunocomplexes were eluted with elution buffer, and reversal of cross-linking and purification of DNA were done as described (28). Precipitated DNA was resuspended in 100 l of TE buffer for input DNA and 20 l of TE buffer for ChIP DNA. The DNA fragments obtained were analyzed by PCR using the following primer pair: Id2, 5Ј-CTTTTGCTTTGG-CAAGGTCT-3Ј or 5Ј-CTTGCA-GGCATTGATCAGC-3Ј (forward) and 5Ј-CCCTGCAGCCTTGTC-CTC-3Ј (reverse). The ChIP assay of C/EBP␤ was performed as described previously (24) and used as a positive as well as a negative control. Quantitative real-time PCR for ChIP was performed using SYBR Premix Ex Taq (TaKaRa, Ohtsu, Japan) on an ABI PRISM 7000 sequence detection PCR machine (Applied Biosystems, Foster City, CA).

Id2 Is an Immediate Early
Response Gene-We examined Id2 gene expression in a murine fibroblast cell line (NIH3T3) during cell cycle progression. To obtain a synchronized cell population, NIH3T3 cells were serum-starved with 0.2% FBS-containing DMEM for 48 h and then stimulated with 20% FBS-containing DMEM. RNA was prepared before and after the stimulation and subjected to Northern blot analysis. Id2 mRNA was undetectable in a growth-arrested state and was induced as early as 30 min after serum stimulation (Fig. 1A) as reported previously (7). Id2 mRNA was induced in a biphasic pattern with peaks at 4 and 14 -16 h after serum stimulation, similar to Id2 expression at the mRNA level in human primary cultured fibroblasts (8) and in human diploid fibroblast TIG3 cells (3). To determine whether serum-induced Id2 expression depends on de novo protein synthesis, we treated NIH3T3 cells with cycloheximide, an inhibitor of protein synthesis, and analyzed Id2 mRNA expression after serum stimulation by Northern blotting. As shown in Fig. 1B, cycloheximide did not interfere with seruminduced Id2 mRNA expression and rather augmented it at 30 min after serum stimulation. Similar results were obtained with TIG3 cells (data not shown). These results indicate that Id2 is one of the immediate early genes that are rapidly induced in response to growth stimulation without de novo protein synthesis.
Induction of Id2 Gene Expression by Serum Is Dependent on RFX1-To explore the molecular basis of the immediate early response of Id2 to serum stimulation, we tried to identify the genomic region that confers the serum response using a luciferase reporter assay. To this end, the upstream genomic fragment of the mouse Id2 gene was subcloned into the pGL4.12 reporter plasmid, which encodes a degradationprone luciferase protein, enabling us to trace a rapid change in reporter activity. After transfection of the reporter plasmid pGL4-Id2-1, into NIH3T3 cells, the cells were serumdeprived with 0.2% FBS-containing DMEM for 48 h and then stimulated with 20% FBS-containing DMEM for 4 h. The extent of serum-induced transcription was evaluated by calculating the -fold induction of the luciferase activity before and after serum stimulation (luciferase activity at 4 h/luciferase activity at 0 h). As shown in Fig. 2A, the 4.6-kb DNA fragment upstream of the Id2 gene exhibited an ability to induce 5.5-fold induction of the luciferase activity, suggesting that this region contains an element responsive to serum. We then constructed a series of deletion mutants of this genomic region and analyzed their reporter activities. As a result, a 142-bp region flanked by PstI sites, located 3.0 kb upstream of the transcription initiation site of mouse Id2, was found to be sufficient to induce the serum response (pGL4-Id2-8 in Fig. 2A), although a minor contribution of the proximal 2350-bp region was also noted. The 142-bp serum-responsive region contains potential binding sites for a basic helix-loop-helix factor, cAMP response elementbinding protein, cut-like homeodomain protein, NF-E2 (nuclear factor erythroid 2), RFX1 and p53 and is highly conserved between humans and mice (Fig. 2B). To specify which transcription factor is involved in the serum response of Id2, we introduced site-directed mutations into each potential binding site in the region and measured the resultant reporter activities before and after serum stimulation. As shown in Fig. 2C, the reporter bearing a mutation in the potential binding site for RFX1 displayed virtually no serum response, in contrast to the lack of a drastic alteration in the serum response with the mutants of the other consensus binding sites, although a slight reduction was observed in the mutants of E-box 2 and of the potential binding sites of NF-E2 and p53. In accordance with this, deletion of the RFX1 site led to loss of the serum response of the 142-bp region (Fig. 2C). Thus, the luciferase reporter assay identified the serum response element of Id2 (Id2-SRE) to be the potential binding site for RFX1.
RFX1 Binds Id2-SRE in Vitro and in Vivo-We next performed EMSA to investigate whether RFX1 binds directly to Id2-SRE using an oligonucleotide containing Id2-SRE. Lysates of serum-starved NIH3T3 cells formed two specific DNA-protein complexes, a major slower migrating complex and a minor faster migrating complex, whereas only the slower migrating complex was detected with the lysates prepared from asynchronous and serum-stimulated cells (Fig.  3A, lanes 1-3). Both complexes were abolished by the addition of excess unlabeled competitors derived from Id2-SRE or from one of the MHC class II promoters (Fig. 3A, lanes  4 -6 and 13-15) (20) and were supershifted in the presence of anti-RFX1 antibody (lanes 7-9). In addition, the oligonucleotide bearing a mutation in the RFX1-binding site failed to form such DNA-protein complexes (Fig. 3A, lanes  10 -12). Because it is known that, in addition to binding DNA as a monomer, RFX1 binds DNA as a homodimer or heterodimers with other RFX family members (29), particularly with RFX2 and RFX3, we examined whether other RFX family members are involved in the complexes detected by EMSA. As shown in Fig. 3B (lanes 1-12), no supershift of the DNA-protein complexes was observed in EMSA with anti-RFX2, anti-RFX3, or anti-RFX5 antibody, although these antibodies could properly recognize the respective antigens in NIH3T3 cells transfected with cDNAs for the RFX isoforms and in HeLa cells (data not shown). On the other hand, reverse transcription-PCR failed to detect expression of Rfx4  SEPTEMBER 7, 2007 • VOLUME 282 • NUMBER 36

JOURNAL OF BIOLOGICAL CHEMISTRY 26171
in NIH3T3 cells (data not shown). RFX1 isoforms other than RFX1 therefore seem not to be involved in complexes formed with Id2-SRE. Furthermore, in vitro translation products of the human RFX1 cDNA in rabbit reticulocyte lysate formed DNA-protein complexes in EMSA, similar to those in NIH3T3 cell lysates (Fig. 3B). These results suggest that the RFX1 homodimer binds Id2-SRE, irrespective of cell conditions, and that the RFX1 monomer appears to be related to cell conditions associated with serum starvation.
To confirm the in vivo binding of RFX1 to endogenous Id2-SRE, we carried out ChIP assay with asynchronous, serum-starved, and serum-stimulated NIH3T3 cells. With chromatin fractions of serum-stimulated and asynchronous cells, anti-RFX1 antibody enriched chromatin-associated DNA containing Id2-SRE (Fig. 3C,  lanes 1 and 3), similar to the result obtained for C/EBP␤ (lane 10), which is a direct upstream factor of Id2 (24), whereas normal IgG did not (lanes 4 -6). The binding of RFX1 to the region containing Id2-SRE was also observed with serum-starved cells (Fig. 3C, lane  2). These results indicate that RFX1 resides on genomic Id2-SRE irrespective of whether cells are growth-arrested or proliferating.
Reduction of RFX1 Leads to the Impaired Immediate Early Response of the Endogenous Id2 Gene-We examined the role of RFX1 in serum-induced Id2 gene expression by utilizing RNA interference-mediated gene suppression. For this, two siRNAs (RFX1-siRNA1 and RFX1-siRNA2) were designed according to a previous report (25), and the respective short hairpin RNA vectors were constructed based on pSUPER.puro. Their efficacies were confirmed at the RNA and protein levels by transiently expressing them in NIH3T3 cells. As shown in Fig. 4A, RFX1-siRNA1 was effective in reducing RFX1 to one-fifth the level in untreated cells, but RFX1-siRNA2 was not. We next established NIH3T3 cells harboring the respective short hairpin RNA vectors by selection with puromycin and investigated the effect of reduced RFX1 on endogenous Id2 mRNA expression by performing Northern blotting before and after serum stimulation.
In cells with RFX1-siRNA1, the induction of Id2 mRNA at 0.5-12 h after serum stimulation was suppressed to be less than half compared with that observed in cells expressing non-effective RFX1-siRNA2, but the subsequent expression was not altered (Fig. 4B). In contrast, c-myc mRNA became detectable even before serum stimulation, and its expression was enhanced by a maximum of 2-fold until 4 h after serum stimulation (Fig. 4B), consistent with the report that RFX1 negatively regulates c-myc expression (18). These results demonstrate that Id2 mRNA expression is regulated positively by RFX1 during serum-induced re-initiation of growth.
Perturbed Id2 Expression in RFX1 Knockdown Cells Is Rescued by siRNA-resistant RFX1-To exclude the possibility that the phenotypes observed in RFX1 knockdown cells were derived from cross-suppression of other genes, we generated a plasmid (pSG5-HARFX1-Rs) encoding an RFX1 cDNA that was resistant to RFX1-siRNA1, but preserved the amino acid sequence of the wild-type RFX1 protein by sitedirected mutagenesis as shown in Fig. 5A. After confirming that siRNA-resistant RFX1 was able to restore the level of RFX1 protein in cells expressing RFX1-siRNA1 (Fig.  5B), we performed a reporter assay and Northern blotting to examine its effect. In the reporter assay, the serum-induced luciferase activity in cells with RFX1-siRNA1 showed 2-fold reduction compared with that in parental NIH3T3 cells, whereas the reporter activity before serum stimulation seemed to be  4 -6), RFX3 (lanes 7-9), RFX5 (lanes 10 -12), or RFX1 (lanes 13  and 14) was included in the reaction mixture. In vitro translation products of RFX1 were used in lanes 13 and 14 instead of nuclear extracts of NIH3T3 cells. C, ChIP assay for RFX1. The chromatins prepared from asynchronous cells, serum-starved cells, and cells at 4 h after serum stimulation were precipitated with anti-RFX1 antibody (lanes 1-3 and 7-9), normal rabbit IgG (lanes 4 -6), or anti-CEBP/␤ antibody (lane 10), and DNA purified from them was amplified with a primer set for Id2-SRE (lanes 1-6) or for the CEBP/␤ response element of Id2 (lanes 7-10). ChIP for CEBP/␤ with anti-RFX1 antibody (lanes 7-9) and anti-CEBP/␤ antibody (lane 10) served as negative and positive controls, respectively. DNA prepared from chromatin before ChIP was amplified by PCR with the respective primer sets, and 10% of the products are shown as input. puro-RFX1-1 (RFX1-siRNA1), or pSUPER.puro-RFX1-2 (RFX1-siRNA2) and selected with puromycin for 2 days. RNA and protein were then prepared and subjected to Northern and Western blot analyses, respectively, as described under "Experimental Procedures." B, Northern blot analysis of Id2 and c-myc mRNA expression in NIH3T3 cells during serum stimulation. NIH3T3 cells expressing RFX1-siRNA1 or RFX1-siRNA2 were established as described under "Experimental Procedures" and serum-starved for 48 h. After stimulation with 20% FBS, RNA was purified at the indicated time points and subjected to Northern blotting. Cells expressing noneffective RFX1-siRNA2 served as a control. 36B4 rRNA was used as a loading control. The relative mRNA expression levels of Id2 and c-myc normalized by 36B4 rRNA expression are shown at the bottom. Open and closed bars indicate the results with RFX1-siRNA1 and RFX1-siRNA2, respectively. Asterisks indicate that alterations of mRNA expression levels caused by siRNA were statistically significant (*, p Ͻ 0.05; **, p Ͻ 0.01).
slightly elevated in RFX1-reduced cells (Fig. 5C). In this experimental setting, coexpression of siRNA-resistant RFX1 restored the reporter activities before and after serum stimulation, whereas wild-type RFX1 had no apparent effect (Fig. 5C). In Northern blot analysis, Id2 and c-myc mRNA levels were reduced and enhanced, respectively, by RFX1-siRNA1 at 4 h after serum stimulation and were restored to the levels in parental NIH3T3 cells by the presence of siRNA-resistant RFX1 (Fig. 5D). These results demonstrate that RFX1-siRNAmediated alteration of gene expression is due to the decreased RFX1 level.
N-and C-terminal Regions of RFX1 Differentially Regulate Id2 mRNA Expression-As shown by the EMSA and ChIP analyses, in addition to its involvement in serum-induced immediate transactivation of Id2, RFX1 appears to play some role in the suppression of Id2 gene expression in growtharrested cells. In fact, RFX1 has been reported to be a binary transcription factor: a transactivator for MHC class II (30,31) and a repressor for c-myc (18), for example. Related to this, it has been demonstrated that the activation and repression domains are located in the N-and C-terminal regions of RFX1, respectively (32). To examine how RFX1 regulates transcription of the Id2 gene, we introduced wild-type RFX1 or its deletion mutants (20), together with pGL4-Id2-142 bearing Id2-SRE, into NIH3T3 cells and examined their effect on the reporter activity before and after serum stimulation. In cells before serum stimulation, the reporter activ-ities of RFX1⌬N and RFX1⌬C were substantially altered to onefourth and 6-fold that of wild-type RFX1, respectively (Fig. 6A). On the other hand, serum stimulation further induced the reporter activity in the presence of RFX1⌬C, although the extent of induction was weak, whereas serum-induced reporter activity was substantially suppressed in the presence of RFX1⌬N (Fig. 6A). As shown in Fig. 6B, the expression levels and subcellular localizations of wildtype and mutant RFX1 proteins were similar in immunocytochemical analyses, although RFX1⌬C showed cytosolic distribution in a fraction of the transfectants. In addition, cotransfection experiments demonstrated that wildtype RFX1 was localized in the nucleus irrespective of the presence of the RFX1 deletion mutants (data not shown). Therefore, the results with RFX1⌬N demonstrate that RFX1 is involved in serum-induced Id2 expression through its N-terminal domain, whereas those with RFX1⌬C suggest that the C-terminal domain of RFX1 is required to suppress Id2 expression during serum starvation.
HDAC1 Interacts with RFX1 and Is Recruited to Id2-SRE during Serum Starvation-Although a coactivator of RFX1 is not known, RFX1 has been reported to interact with HDCA1 and to suppress the expression of collagen ␣2(I) (33). To examine whether HDAC1 is also involved in the suppression of Id2 mRNA expression during serum deprivation, we first performed immunoprecipitation and Western blot analyses with cell lysates prepared from NIH3T3 cells before and 4 h after serum stimulation. As shown in Fig. 7A, the physical interaction between HDAC1 and RFX1 was detected in the lysate of serum-starved cells, whereas the interaction was much weaker in the lysate prepared from serum-stimulated cells. Next, we carried out quantitative PCR after ChIP assay to evaluate the recruitment of HDAC1 to endogenous Id2-SRE. At 24 h after serum starvation, anti-HDAC1 antibody enriched chromatin-associated DNA containing Id2-SRE to an extent similar to that observed at 48 h after serum starvation (Fig. 7B). The enrichment was reduced as early as 30 min after serum stimulation, and the situation continued afterward (Fig.  7B). The enrichment of chromatin-associated DNA was not detected with normal goat IgG or with a primer set for the C/EBP␤-binding site in the Id2 promoter (Fig. 7B). These results demonstrate that HDAC1 is recruited to Id2-SRE during serum starvation and suggest that there is an inverse correlation between the recruitment of HDAC1 to Id2-SRE and Id2 mRNA expression.

DISCUSSION
We have demonstrated here that Id2, a negative regulator of basic helix-loop-helix factors, is an immediate early gene and that its rapid induction in growth-arrested NIH3T3 cells in response to serum stimulation is dependent on RFX1. This is the first indication that RFX1 plays a role in the induction of immediate early gene expression. Our results reveal that RFX1 regulates Id2 gene expression in two man-ners: cell context-dependent and -independent manners. Cell context-dependent regulation is observed in the difference before and after serum stimulation. It is associated with a functional difference in the N-and C-terminal domains of RFX1, and the N-terminal domain of RFX1 is essential for the induction of Id2 gene expression. Furthermore, the suppression of Id2 gene expression is associated with the recruitment of HDAC1 to Id2-SRE. On the other hand, comparison of the effect of RFX1 on Id2 expression with that on c-myc expression revealed cell context-independent regulation. RFX1 differently regulates gene expression, depending on its target genes, even under the same cell conditions. RFX1 was first identified as a factor that binds the X-box present in the regulatory element of the MHC class II genes (30,31) and was subsequently found to be identical to a factor that binds cis-regulatory elements present in the enhancers of several viruses, including human hepatitis B virus (17). It is characterized by a winged helix-type DNA-binding domain with a unique binding property (34) and constitutes the RFX family together with four structurally related factors (RFX2-RFX5) in mammals (22,(35)(36)(37). RFX1 displays a binary function as a transcription factor: it activates expression of the interleukin-5 receptor-␣ gene (21) and the neuronal glutamate transporter type 3 gene (38), whereas it suppresses expression of several genes such as c-myc (18), proliferating cell nuclear antigen (19), collagen ␣(I) (33,39), and RFX1 (40). Analysis of the structurefunction relationship of RFX1 has revealed that the activation and repression domains are present in the N-and C-terminal domains of RFX1, respectively (32).
The expression level and subcellular localization of the RFX1 protein are not changed in NIH3T3 cells before and after serum stimulation, although it was reported that RFX1 translocates to the nucleus upon activation of protein kinase C in HL-60 leukemia cells (41). In addition, irrespective of the cell context, ChIP assay revealed the loading of RFX1 on the 5Ј-flanking sequence of the Id2 gene, and gel shift assay demonstrated that the RFX1 homodimer is a major complex formed with Id2-SRE. These results suggest that RFX1 is always loaded on Id2-SRE and ready to induce Id2 gene expression via its N-terminal domain in response to serum stimulation. This mode of action is in accordance with the properties of immediate early genes that are induced rapidly without de novo protein synthesis upon stimulation. It has been postulated that RFX1 behaves as a nearly inactive transcription factor at certain times because of mutual neutralization by the activation and repression domains of RFX1 and that relief of its neutralization allows RFX1 to serve as an activator or a repressor (32). On the basis of this notion, we speculate that serum stimulation leads to the functional shift in RFX1 from being a "neutralized" transcription factor to being a transactivator, which explains why C-terminal deletion of RFX1 caused a substantial enhancement of Id2-SREmediated reporter activity in growth-arrested cells, whereas reduction of RFX1 by siRNA did not result in an apparent increase in Id2-SRE-mediated reporter activity or in endogenous Id2 mRNA expression. It is natural to think that serum-induced alteration of cell conditions leads to the modification of RFX1 and then triggers the activation of A, NIH3T3 cells were transfected with the reporter plasmid pGL4-Id2-8, together with pSG5 vector (Vector), pSG5-HARFX1 (RFX1 WT), pSG5-HARFX1⌬N (RFX1⌬N), or pSG5-HARFX1⌬C (RFX1⌬C), and luciferase (LUC) activity was measured before and 4 h after serum stimulation. Open and closed bars indicate relative luciferase activities before and 4 h after serum stimulation, respectively. The results are presented as the means Ϯ S.E. and are representative of at least three independent experiments with triplicate assays. Asterisks indicate a significant difference in reporter activity (*, p Ͻ0.05; **, p Ͻ0.01; ***, p Ͻ0.001). B, shown are the results of immunocytochemical analyses of wild-type RFX1 and deletion mutants. NIH3T3 cells were transfected with pSG5-HARFX1, pSG5-HARFX1⌬N, or pSG5-HARFX1⌬C and starved-serum for 48 h as described for A. Transfectants were fixed with 4% formaldehyde-containing phosphate-buffered saline before or 4 h after serum stimulation, and immunocytochemical analyses were performed with mouse anti-hemagglutinin (␣-HA) monoclonal antibody and Alexa Fluor 488-conjugated goat anti-mouse IgG. 4Ј,6-Diamidino-2-phenylindole dihydrochloride (DAPI) was used to visualize the nuclei. The subcellular localizations of the respective RFX1 proteins were examined and are tabulated at the bottom. N, nuclear localization; C, cytosolic localization. Open and closed bars indicate the results before and 4 h after serum stimulation, respectively.