A B-myb promoter corepressor site facilitates in vivo occupation of the adjacent E2F site by p107-E2F and p130-E2F complexes

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

It has become well established that E2F transcription factor activity, through its interactions with the retinoblastoma family of pocket proteins, plays a major role in coordinating gene transcription during the mammalian cell cycle. E2F activity is a heterogeneous entity made up of heterodimers from two distantly related protein families, E2F and DP (1,2). Each component of the E2F protein family, E2F-1 through E2F-6, appears to be functional in association with either member of the related DP protein family, DP-1 and DP-2 (3). E2F-1, E2F-2, and E2F-3 heterodimers form complexes preferentially with the retinoblastoma protein Rb, whereas the related pocket proteins p107 and p130 associate preferentially with E2F-4 and E2F-5. This demarcation is not absolute, however; and it has been reported that Rb⅐E2F-4 complexes are the predominant species in the nuclei of actively dividing cells during G 1 (4). E2F-6 lacks the C-terminal domain that, in the other species, has been implicated in transactivation function and interaction with the pocket proteins (5)(6)(7)(8).
The diversity present within the E2F and pocket protein families suggests that different E2F complexes may play distinct roles in cell cycle regulation of gene transcription. In support of this notion, it was found that "knockouts" of specific pocket protein genes by homologous recombination have quite different effects on mouse embryonic development. Whereas Rb Ϫ/Ϫ mice die in utero with defects in the liver, central nervous system, and ocular lens and a profound reduction in definitive erythropoiesis (9), p107 Ϫ/Ϫ and p130 Ϫ/Ϫ mice develop normally (10). Inactivation of both p130 and p107 results, however, in neonatal lethality, with defects in bone and cartilage arising from hyperproliferation of chondrocytes (10). Differences in pocket protein function are also manifest at the level of cell cycle-regulated transcription in mouse embryo fibroblasts (MEFs) 1 derived from mutant animals (11). Whereas elimination of either p107 or p130 had no discernible effect on cell cycle-regulated transcription, knockout of both p107 and p130 genes resulted in deregulated expression of a set of genes (B-myb, cdc2, E2F-1, thymidylate synthase, RRM2, and cyclin A2) distinct from those deregulated in Rb Ϫ/Ϫ MEFs (cyclin E and p107). These phenotypic differences between Rb Ϫ/Ϫ and p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs are unlikely to be due to gene dosage effects, as studies with established knockout fibroblasts showed that deregulation of the B-myb promoter in p107 Ϫ/Ϫ / p130 Ϫ/Ϫ cells could not be recapitulated in either p107 Ϫ/Ϫ / Rb Ϫ/Ϫ or p130 Ϫ/Ϫ /Rb Ϫ/Ϫ cells (12). It is therefore evident that p107 and p130 have a functionally redundant role in E2Fmediated gene regulation, which is distinct from that of Rb.
It is currently unclear how different pocket protein⅐E2F complexes discriminate between the genes they regulate. One possibility is that they demonstrate some sequence specificity for the E2F binding site in the promoters of these genes. Although most in vitro binding studies have failed to show this, repetitive selection of redundant binding sites by the CASTing procedure (13) suggests that different E2F complexes do have inherent preferences for particular E2F sites. An additional possibility is that occupation of the E2F site in vivo is influenced by factors binding to adjacent sites or that the ability of E2F complexes to regulate transcription, whether by repression or activation, is dependent upon interactions with these accessory factors. The contribution of a putative accessory binding site to E2F-dependent transcription is exemplified by B-myb (the MybL2 gene), in which it was found that cell cycle regulation is influenced by a distinct promoter site located immediately down-stream of the E2F site (17,32). Mutations in either the E2F site or the adjacent site (downstream repression site (DRS)) abolish transcriptional repression in G 0 /G 1 . Significantly, in vivo footprinting studies revealed that the B-myb promoter E2F site is occupied in quiescent cells, but becomes unoccupied in mid G 1 preceding the induction of B-myb transcription (14). As E2F abundance actually increases at the G 1 /S boundary, it is possible that the role of the DRS is to stabilize interactions of repressor complexes at the adjacent E2F site specifically in G 0 /early G 1 .
Transcriptional repression is a common mechanism to extinguish expression of periodically regulated genes during cell cycle arrest and quiescence (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28); however, there is evidence that the regulation of certain of these genes is not directly dependent on E2F. Most notably, transcriptional repression of CDC25C in G 0 /early G 1 involves bipartite elements termed CDE/CHR (23,29). The CDC25C CDE/CHR elements are partially homologous to the B-myb E2F/DRS sites and, moreover, have an identical spatial relationship (30); however, E2F binds weakly if at all to CDE/CHR. Compared with E2F/DRS, the CDE/CHR elements confer a subtly different cell cycle kinetics on promoters: both sets of elements permit derepression of transcription in the latter part of G 1 ; however, induction is somewhat prolonged through S/G 2 when regulated by CDE/ CHR (30). To gain further understanding of the relationship between the downstream DRS and CDE elements, we have extensively characterized the B-myb DRS by mutagenesis. We have also studied the role played by B-myb E2F/DRS sites in transcriptional regulation in vivo and conclude that the DRS facilitates interaction of repressive p107⅐E2F and p130⅐E2F complexes with the adjacent E2F site.

EXPERIMENTAL PROCEDURES
Cell Culture and Flow Cytometry-MEFs from control p107 ϩ/Ϫ and gene knockout p107 Ϫ/Ϫ /p130 Ϫ/Ϫ animals were kindly provided by Dr. Nick Dyson (Massachusetts General Hospital Cancer Center, Charlestown, MA) and were expanded and used at passage 4. MEFs and Swiss 3T3 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Established 3T3 cell lines were derived from MEFs by splitting cultures every 3 days according to standard procedures. These established 3T3 and NIH 3T3 cells (obtained from Dr. René Bernards) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% newborn calf serum. Flow cytometry was performed on ethanol-fixed cells stained with propidium iodide as described previously (22).
To insert wild-type, mE2F, and mDRS B-myb promoter sequences into the pAG/EluW luciferase reporter, which contains flanking scaffold/matrix attachment regions from the human interferon-␤ gene to attenuate positional effects in chromosomally integrated plasmids (31), fragments with terminal HindIII sites were generated by PCR with Pfu DNA polymerase using pGL2-(Ϫ536), pGL2-(Ϫ536)mut1, and pGL2-(Ϫ536)mut9 as templates (32). The primers used (5Ј-AGCTAAGCTTC-CAGTCTTTGCTATGTGTGTG and 3Ј-ACGTAAGCTTCGAGCCGCTC-CGGGCCCCAGG) amplified promoter sequences from Ϫ533 to Ϫ88 with respect to the start of the protein coding sequence, and these HindIII fragments replaced the SV40 promoter in pAG/EluW. Plasmids were transfected into NIH 3T3 cells using calcium phosphate coprecipitation, and stable transfectants were selected with 0.5 g/ml G418 using the conditions described to limit copy number (31). Clones carrying approximately five copies of each transgene were selected for chromatin immunoprecipitation (ChIP) analysis.
Pocket Protein Expression Plasmids-An expression plasmid encoding untagged p107, pCMVp107, was kindly provided by Dr. Liang Zhu. pcDNA3 (Invitrogen) expression plasmids encoding Myc epitope-tagged wild-type Rb and non-phosphorylatable Rb (RbNP; in which 14 of 15 cyclin-dependent kinase sites were mutated to alanine) were kindly provided by Dr. Sybille Mittnacht (33). Expression vectors encoding Myc epitope-tagged p107 and p130 proteins were constructed by fusing p107 and p130 coding sequences lacking the initiating methionine codon, generated by a combination of PCR and restriction endonuclease digestion, downstream of the Myc epitope coding sequence in pcDNA3(9E10). A combination of PCR and restriction endonuclease digestion was also used to generate coding sequences specifying fragments of p130. These were cloned downstream of the Myc epitope coding sequence in pcDNA3(9E10): mutant N417 (amino acids 417-1139) contains the A and B pockets, the spacer region, and the entire C-terminal domain; mutant C793 (amino acids 2-793) is C-terminally truncated within the spacer; mutant C1065 (amino acids 2-1065) is C-terminally truncated immediately after the B pocket; and mutant dl622-818 contains a deletion of most of the spacer. PCR-generated sequences were checked by DNA sequencing.
Cell Transfection and Reporter Assays-MEFs and 3T3 derivatives were transfected according to the manufacturer's instruction in 60-mm Petri dishes (10 5 cells) using 20 l of LipofectAMINE (Invitrogen) and 4 g of DNA (3 g of luciferase reporter ϩ 1 g of ␤-galactosidase plasmid). After 5 h, the LipofectAMINE was removed, and cells were allowed to recover in Dulbecco's modified Eagle's medium containing 10% fetal calf serum overnight before replacing the medium with Dulbecco's modified Eagle's medium containing 0.5% fetal calf serum. Cell lysates were made 60 h later, and luciferase and ␤-galactosidase assays were performed as described (15). Where cell extracts were prepared from transfected 3T3 derivatives, cells were transfected in 100-mm dishes using 60 l of LipofectAMINE and a total of 12 g of DNA.
Northern and Western Blots--RNA was prepared from p107 ϩ/Ϫ and p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells using RNAzol B (Biogenesis) as described by the manufacturer and resolved on 1% agarose gels containing 2.2 M formaldehyde. Northern blots were probed with a 32 P-labeled murine B-myb cDNA probe. Western blotting was performed under standard conditions using anti-c-Myc antibody 9E10 (Santa Cruz Biotechnology sc-40) to detect the tagged pocket proteins.
Electrophoretic Mobility Shift Assays-Nuclear and cytoplasmic protein fractions were prepared from MEFs and transfected 3T3 cell lines after lysing in a hypotonic buffer containing 0.4% Nonidet P-40 essentially as described (34). Equivalent amounts of nuclear and cytoplasmic fractions (ϳ10 g) were used in electrophoretic mobility shift assays (EMSAs) using an oligonucleotide probe encompassing the mouse B-myb promoter E2F/DRS sites as described previously (15). Where appropriate, antibodies to p107 (monoclonal antibody SD15; a gift of Dr. Nick Dyson), E2F-4 (polyclonal; a gift of Dr. Eric W.-F. Lam), p130 (polyclonal; a gift of Dr. Antonio Giordano), Rb (monoclonal antibody 21C9; a gift of Dr. Sybille Mittnacht), or c-Myc (monoclonal antibody 9E10) were added to supershift specific complexes. Whole cell extracts were prepared from NIH 3T3 cells stably transfected with the wild-type or mDRS B-myb promoter as described previously (15). These were used in EMSAs as described above; however, the antibodies used in supershifts were the same as those used in ChIP assays (rabbit anti-p107 polyclonal antibody, Santa Cruz Biotechnology sc-318; rabbit anti-p130 polyclonal antibody, Santa Cruz Biotechnology sc-317; and anti-Rb monoclonal antibody 21C9).
In Vivo Footprinting and Chromatin Immunoprecipitation-Footprints were obtained by linker-mediated PCR on DNA alkylated in vivo by addition of 0.2% dimethyl sulfate to cell cultures as described previously (22). The B-myb primers used were as published (14).
ChIP assays were performed using essentially the published method (35). Sonicated chromatin from 10 7 cells treated in vivo with 1% formaldehyde for 10 min at ambient temperature was immunoprecipitated with 2 g of each antibody (control (normal rabbit serum), rabbit anti-p107 polyclonal, and rabbit anti-p130 polyclonal antibodies and anti-Rb monoclonal antibody 21C9) and collected by mixing for 1 h at 4°C with 45 l of protein A-Sepharose 4B beads (50% suspension in TE buffer (20 mM Tris-HCl (pH 8.1) and 1 mM EDTA) preadsorbed with sonicated salmon sperm DNA (10 g/ml). Rabbit anti-mouse IgG (2 g; Sigma M7023) was added to monoclonal immunoprecipitations for 1 h before addition of protein A beads. The beads were collected by brief microcentrifugation, and the supernatant of the control antibody sample was retained for use as an input control. The beads were washed successively for 10 min at 4°C with TSE buffer (1% Triton X-100, 0.1% SDS, 2 mM EDTA, and 20 mM Tris-HCl (pH 8.1)); TSE buffer containing 150 mM NaCl; TSE buffer containing 500 mM NaCl; buffer containing 250 mM LiCl, 1% Nonidet P-40, 1% deoxycholate, 1 mM EDTA, and 10 mM Tris-HCl (pH 8.1); and TE buffer. Immunoprecipitated DNA was eluted by washing the beads three times with 150 l of 1% SDS and 100 mM NaHCO 3 , and then 10 g of salmon sperm DNA was added to the pooled eluates. The cross-links were reversed by heating at 65°C for 4 h, and then the DNA was phenol-extracted twice and ethanol-precipitated. DNA was recovered by centrifugation and dissolved in 350 l of buffer containing 100 mM NaCl, 10 mM Tris-HCl (pH 8.1), and 1 mM EDTA. After reprecipitation with ethanol, the final pellet was dissolved in 30 l of TE buffer. PCR amplification was carried out for 25 or 26 cycles on 2-l amounts using Taq polymerase. Separate reactions were carried out to detect the transgene and the endogenous B-myb gene; transgene reactions used a 5Ј-B-myb primer (CCTTCCGTATGCTCCG-CCC) and a 3Ј-primer to the luciferase gene (TCATAGCCTTATGCAG-TTGCTCTCC), whereas endogenous reactions used 5Ј-and 3Ј-B-myb primers (ATTGAATCCCTAAGGTAGGTGTATCTG and TGGTCGCAC-GTTCCCAG, respectively). PCR products were resolved on 1.5% agarose gels, Southern-blotted, and detected by hybridization with a 5Ј-32 P-labeled oligonucleotide probe (TTCTGTGCGCTCCCTAGGCAG-CTGCAGTTCCTGGGAACG). Hybridization was quantitated by phosphorimaging.

Identification of Functionally Critical DRS Nucleotides-
To characterize the precise sequence requirement for transcriptional repression of the B-myb promoter, each nucleotide within the previously delimited DRS motif (GGAAA, nucleotides Ϫ198 to Ϫ194 relative to the start of the coding sequence) was mutated individually to each of the three alternatives. To facilitate this, AatII and BclI restriction sites were first introduced into the B-myb promoter at sequences immediately flanking the E2F/DRS elements by substituting 2 and 3 nucleotides, respectively (Fig. 1A). Previous studies have shown that the regions mutated do not influence cell cycle-regulated transcription (32). Double-stranded oligonucleotides containing DRS mutations could then be quickly transferred into the modified reporter plasmid, the so-called B-myb promoter cassette, by direct ligation (see "Experimental Procedures"). The functions of the wild-type and mutant DRS elements were then assayed in transient transfection assays of NIH 3T3 cells. Following transfection, the cells were arrested in G 0 by serum deprivation, and cell extracts were made from these cells and from duplicates that were induced to enter S phase by serum stimulation for 16 h. The results of a typical experiment ( Fig.  1B) show that the central 3 nucleotides in the DRS (GGAAA, nucleotides Ϫ197 to Ϫ195) are particularly critical for E2Fmediated transcriptional repression. Thus, mutation of either A residue to any other nucleotide resulted in a significant increase in promoter activity in both G 0 and S phase cells. Mutation of the central G residue to A had no effect upon promoter activity, whereas either C or T at this position resulted in a very significant increase in promoter activity. It is pertinent that the equivalent DRS sequence in the human B-myb promoter is GAAAA, consistent with the evidence that either G or A at nucleotide Ϫ197 is permissive for repression function. In contrast, mutation of the first G residue of the DRS (nucleotide Ϫ198) had only a marginal effect on transcriptional derepression; mutation to A led to a slight loss of repression, whereas mutation to C actually resulted in a slight increase in repression. Mutation of the ultimate A residue (nucleotide Ϫ194) to C or T had little or no effect upon DRS function, whereas substitution by G led to a substantial loss of transcriptional repression (Fig. 1B). Comparison of a DRS consensus sequence derived from this analysis with CHR sequences located downstream of the cyclin A2, cdc2, and CDC25C CDE sites (Fig. 1C) revealed that all three CHRs are actually compatible with the DRS sequence requirements.
It is notable that promoters in which the critical GAA core of the DRS is mutated were substantially induced in S phase (Fig.  1B), suggesting that the E2F site can mediate transcriptional activation in S phase when DRS function is compromised. To test this possibility, the activity of a promoter in which both sites have been mutated was compared with those of reporters containing single E2F or DRS mutations, and this assay confirmed that activation of the mDRS promoter was indeed abolished when the adjacent E2F site was also mutated ( Fig. 2A). As the mDRS promoter was significantly more active in S phase than the wild-type promoter, it appears that transactivating E2F complexes are able to bind to the B-myb E2F site only in the absence of the DRS. It may therefore be suggested that the DRS specifically prevents activating E2F species from binding to the adjacent E2F site while conversely facilitating binding of repressive E2F complexes.
To determine whether the CDC25C CHR could indeed substitute for a functional B-myb DRS and, conversely, whether the B-myb DRS could substitute for the CDC25C CHR, the E2F/DRS sites in the B-myb promoter cassette were replaced FIG. 1. Characterization of nucleotides required for DRS function. A, nucleotides flanking the E2F/DRS sites in the B-myb promoter/ reporter pGL2-(Ϫ536) were mutated as indicated in lowercase to generate AatII and BclI restriction sites. B, double-stranded oligonucleotides containing the DRS mutations indicated in lowercase were introduced into the pGL2-(Ϫ536) cassette by replacement of the AatII/ BclI fragment. The wild-type and mutant promoter/reporters were transiently transfected into NIH 3T3 cells, and luciferase activities were measured in cells arrested in G 0 and in parallel cultures that were restimulated by re-addition of serum (S phase). Flow cytometry indicated that 62% of the cells were measurably in S phase at the time the serum-induced samples were collected. C, shown is a comparison of the mouse B-myb E2F/DRS sequences (underlined) with the bipartite CDE/ CHR sequences (also underlined) in the human cyclin A (CCNA), CDC2, and CDC25C genes. With respect to the major transcription start sites, the sequences presented are B-myb (Ϫ49/Ϫ31), CCNA (Ϫ38/Ϫ20), CDC2 (Ϫ25/Ϫ7), and CDC25C (Ϫ19/Ϫ1). Nucleotide changes permissible for DRS function are indicated above the DRS sequence.
with wild-type CDC25C CDE/CHR or hybrid E2F/CHR and CDE/DRS sites (Fig. 2B). Transfection into NIH 3T3 cells showed that the CDC25C CDE/CHR conferred appropriate cell cycle regulation on the B-myb promoter in that the promoter was repressed in G 0 cells and derepressed in S phase (Fig. 2B). Significantly, the hybrid E2F/CHR element conferred more rigorous regulation on the B-myb promoter than either E2F/ DRS or CDE/CHR, whereas in contrast, substitution by the hybrid CDE/DRS element led to complete deregulation (Fig.  2B). These findings indicate that the DRS and CHR are not absolutely equivalent. Whereas both elements are functional in the context of the E2F site, the CHR also has a specific function in the context of the CDE, which cannot be substituted by the DRS.
To determine whether the relative positions of the B-myb E2F/DRS sites are important for their functional interaction, small iterations of 2 and 4 nucleotides were introduced between these sites in the reporter cassette, and transient transfections assays were performed as before. Displacing the DRS from the E2F site by 2 nucleotides resulted in significant transcriptional derepression in G 0 (Fig. 2C), and this effect was even more pronounced when the sites were displaced by 4 nucleotides. These results indicate that the spatial relationship between the E2F and DRS sites is essential for their cooperation.
The DRS Is Required for p107-mediated Repression of B-myb Transcription-Repression of B-myb transcription is substantially deregulated in quiescent MEFs derived from embryos lacking both the p107 and p130 genes, although it is unaffected in MEFs lacking either of these pocket proteins alone or in MEFs lacking Rb (11). These data suggest that p107⅐E2F and p130⅐E2F complexes have a redundant function in repressing the B-myb promoter during G 0 , which cannot be compensated by Rb⅐E2F. Consistent with this notion, reintroduction of p107 and, to a lesser extent, p130 into p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs by transient transfection results in repression of a cotransfected B-myb promoter (11), and this activity depends on an intact E2F site. The p107 Ϫ/Ϫ /p130 Ϫ/Ϫ cells therefore provide a convenient system in which to investigate further the interactions between pocket protein⅐E2F complexes and the bipartite E2F/ DRS site. To determine first the contribution of the DRS in this system, we performed additional transient transfection experiments in p107 Ϫ/Ϫ /p130 Ϫ/Ϫ and control p107 ϩ/Ϫ MEFs. These studies confirmed that the wild-type B-myb promoter is significantly deregulated in quiescent p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs compared with control cells (Fig. 3A). It is notable, however, that deregulation of the wild-type promoter was incomplete in the knockout cells compared with a mE2F site promoter, suggesting that quiescent p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs actually contain diminished rather than no E2F repressor activity. It is notable also that the DRS mutation had similar effects in both quiescent p107 Ϫ/Ϫ /p130 Ϫ/Ϫ and control MEFs, being somewhat de-

FIG. 2. Characterization of DRS function.
A, the B-myb promoter/ luciferase reporter cassette containing the wild-type (wt) sequence or variants in which the wild-type sequence was replaced by E2F/DRS elements containing mutations in the E2F site (mE2F, CTTGtatG-GAGATAGGAAAG), the DRS (mDRS, CTTGGCGGGAGATAGGcctG), or both (mE2F/mDRS, CTTGtatGGAGATAGGcctG) were transiently transfected into NIH 3T3 cells, and luciferase activities were measured in cells arrested in G 0 and in parallel cultures that were restimulated by re-addition of serum for 16 h (S phase). B, the B-myb promoter/luciferase reporter cassette containing the wild-type sequence (E2F/DRS, CTTGGCGGGAGATAGGAAAG) or variants in which the wild-type sequence was replaced by the CDC25C CDE/CHR sequence (CDE/CHR, GCTGGCGGAAGGTTTGAATG) or combinations of these elements (E2F/CHR, CTTGGCGGGAGATTTGAATG; and CDE/DRS, GCTG-GCGGAAGGTAGGAAAG) were transiently transfected into NIH 3T3 cells, and luciferase activities were compared as described for A. C, the B-myb promoter/luciferase reporter cassette containing the wild-type sequence or variants in which the wild-type sequence was replaced by E2F/DRS elements containing intervening iterations of either ϩ2 or ϩ4 nucleotides (CTTGGCGGGAGAtaTAGGAAAG and CTTGGCGG-GAGAtataTAGGAAAG, respectively) were transiently transfected into NIH 3T3 cells, and luciferase activities were compared as described for A.

FIG. 3. Transcriptional repression of the B-myb promoter by p107⅐E2F complexes requires the DRS.
A, the wild-type (wt) B-myb promoter/reporter pGL2-(Ϫ536) or derivatives containing mutations in either the E2F (mE2F) or DRS (mDRS) site were transfected into primary MEFs derived from control p107 ϩ/Ϫ or knockout p107 Ϫ/Ϫ / p130 Ϫ/Ϫ animals. Luciferase was assayed in transfected cells deprived of serum for 60 h. B, knockout p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs were transfected with the B-myb promoter/reporters described for A together with either the empty pCMV expression vector (control (Con)) or vector encoding p107 or Myc epitope-tagged p130. Luciferase was assayed in transfected cells deprived of serum for 60 h. Data are presented as -fold repression, calculated by dividing the average luciferase value obtained with the control empty vector by the average value obtained for that reporter with the p107 or p130 expression vector. The control for each reporter was assigned a value of unity.
As reported previously (11), we found that reintroducing p107 and p130 by transfection into quiescent p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs resulted in repression of the wild-type B-myb promoter (Fig. 3B). Consistent with the previous study (11), we found that p107 repressed the wild-type B-myb promoter more strongly than p130, although this may simply reflect the relative expression levels of these proteins in these cells. Significantly, the ability of p107 and p130 to repress B-myb promoter activity was dependent upon both an intact E2F site and DRS (Fig. 3B). Stimulation by p107 of both the mE2F site and mDRS promoters observed in this experiment presumably reflects an effect on constitutive activators binding to sites upstream of the cell cycle control elements. Overall, it is clear that the DRS is essential for the imposition of transcriptional repression mediated by p107⅐E2F complexes through the adjacent E2F site.
The DRS Is Required for in Vivo Interactions with p130⅐E2F and p107⅐E2F Complexes-Although it is clear from this study that the DRS is required for B-myb transcriptional repression by p130 and p107, previous in vitro binding assays did not indicate any influence of the DRS on binding of E2F complexes containing these pocket proteins to the B-myb promoter site (32). To explain this conundrum, we set out to determine whether the DRS influences occupation of the adjacent E2F site in vivo. To facilitate this analysis, the wild-type and mutant B-myb promoters were cloned into a luciferase reporter gene that incorporates scaffold/matrix attachment sites to attenuate positional effects on the chromosomally integrated promoter (31). Stably transfected NIH 3T3 cells expressing these transgenes were then established using low DNA inputs to reduce plasmid copy number, and clones containing approximately five copies were selected for analysis. It is evident that the wild-type and mutant promoters were regulated in these clones in a manner similar to that in transient transfections (Fig. 4A). Thus, the wild-type B-myb promoter was activated 8.5-fold as serum-arrested cells entered S phase, whereas in contrast, the mE2F site promoter was completely derepressed in G 0 . Significantly, the mDRS promoter was substantially deregulated in G 0 , but was induced ϳ1.5-fold in S phase (Fig.  4A). EMSAs using extracts prepared from wild-type and mDRS promoter-transfected NIH 3T3 cell lines showed a predominance of p130⅐E2F complexes during G 0 (Fig. 4B). Lesser amounts of p107 complexes were also present, but no Rb complexes were detectable (Fig. 4B).
Previous studies of these cell lines using ChIP assays have shown that the wild-type promoter transgene is bound by E2F-4 complexes in quiescent cells and that this association is abrogated by mutation of the E2F site (37). We adopted a similar approach to study whether the DRS influences occupation of the adjacent E2F site. Protein⅐DNA complexes were formaldehyde-cross-linked in cells stably expressing the wildtype and mDRS promoter/luciferase reporters, and sites bound by pocket protein⅐E2F complexes were immunoprecipitated with control, anti-p107, anti-p130, or anti-Rb antibodies. PCR primer pairs were designed to detect selectively either the B-myb transgene (the 3Ј-primer is homologous to luciferase sequences) or the endogenous B-myb gene. We found that the transfected wild-type B-myb promoter was bound predomi-

FIG. 4. The DRS is required for occupation of the B-myb E2F site with p130 and p107 complexes.
A, stable NIH 3T3 transfectants were established with luciferase reporter genes driven by either the wild-type (wt) B-myb promoter or mutant B-myb promoters with nonfunctional E2F (mE2F) or DRS (mDRS) sites. The reporter genes were in each case flanked by scaffold/matrix attachment regions from the human interferon-␤ gene to attenuate positional effects, and clones were selected with approximately five transgene copies. Luciferase activities were assayed in cells arrested in G 0 by serum deprivation for 60 h and after restimulation with serum for 16 h (S phase). B, EMSA was performed with whole cell extracts prepared from NIH 3T3 cell clones stably transfected with the wild-type and mDRS B-myb promoter/luciferase reporters. Pocket protein⅐E2F complexes were supershifted with antibodies (Ab) against p107, p130, or Rb as indicated. Control reactions (Con) contained preimmune rabbit serum. C, NIH 3T3 cell clones stably transfected with the wild-type and mDRS B-myb promoter/luciferase reporters were arrested in G 0 by serum deprivation, and then protein⅐DNA complexes were formaldehyde-cross-linked in vivo. Chromatin fragments from these cells were subjected to immunoprecipitation (IP) with control serum or with antibodies to p107, p130, or Rb as indicated. After cross-link reversal, the co-immunoprecipitated DNA was amplified by PCR and detected by Southern blotting. Separate primers were used to detect the transgene and the endogenous gene. D, shown is the quantitation by phosphorimaging of the results shown in C for the wild-type promoter-transfected cell transgene (wt/Tr) and endogenous gene (wt/En) and the mDRS transgene (mDRS/Tr) and endogenous gene (mDRS/En). nantly by p130 complexes and, to a lesser extent, by p107 complexes, but little Rb was cross-linked to this site (Fig. 4, C and D). The endogenous promoter in these cells showed a very similar pattern of occupancy. The relative levels of p130 and p107 binding to the B-myb promoter detected in the ChIP assays therefore reflected their levels associated with E2F complexes as detected by EMSAs (Fig. 4B). Significantly, p130 and p107 complexes were only weakly associated with the mDRS promoter, although interactions of these proteins with the endogenous promoter were still detected in these cells (Fig. 4, C  and D). These experiments therefore demonstrate that the DRS has an important role in promoting binding of p130⅐E2F and p107⅐E2F complexes to the adjacent E2F site.
In Vitro Analysis of E2F Complexes in p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs-In contrast to NIH 3T3 cells, quiescent p107 Ϫ/Ϫ / p130 Ϫ/Ϫ MEFs contain abundant Rb⅐E2F complexes, which are readily detectable by EMSAs using an adenovirus E2A promoter probe (11). It is therefore unclear why these species are unable to fully repress the B-myb promoter in p107 Ϫ/Ϫ /p130 Ϫ/Ϫ cells. To begin to address this issue, it was important to determine whether Rb⅐E2F complexes in quiescent p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs are located in the nucleus, as at least certain E2F complexes are found sequestered in the cytoplasm (4). Moreover, it was necessary to confirm that these Rb⅐E2F complexes are able to recognize the distinctive B-myb E2F site. To these ends, nuclear and cytoplasmic extracts were prepared from quiescent p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs, and EMSAs were performed using a B-myb E2F site oligonucleotide. The addition of specific antibodies to the EMSAs revealed that the nuclei of p107 Ϫ/Ϫ / p130 Ϫ/Ϫ MEFs contained Rb⅐E2F complexes, which were able to bind the B-myb E2F site in vitro (Fig. 5B); however, these higher order complexes were absent from the cytoplasmic fractions (Fig. 5A). It is notable that the cytoplasmic species in these extracts was composed exclusively of free E2F-4; however, the anti-E2F-4 antibody failed to recognize the majority of the nuclear Rb complex (Fig. 5B). Although it is possible that the E2F-4 epitope is masked by association with Rb, it is more likely that Rb is complexed predominantly with other E2F species. Control EMSAs using extracts from quiescent p107 ϩ/Ϫ MEFs showed that both cytoplasmic and nuclear fractions contained p130⅐E2F species, whereas Rb⅐E2F complexes were again found exclusively in the nuclear fraction (Fig. 5, C and D).
Transfected Rb Forms E2F Complexes, but Inefficiently Represses the B-myb Promoter-Although the in vitro binding assays described above suggest that Rb complexes in p107 Ϫ/Ϫ / p130 Ϫ/Ϫ MEFs are at least as abundant as the p130⅐E2F complexes, which repressed B-myb transcription in control p107 ϩ/Ϫ cells (Fig. 5B), it remains possible that their levels are insufficient. To address this question, we wished to compare the activities of Rb and p107 when these proteins were reintroduced into p107 Ϫ/Ϫ /p130 Ϫ/Ϫ cells by transfection. As primary MEFs are poorly transfectable, we first established immortalized p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 lines to facilitate the analysis. Northern blotting demonstrated that B-myb was abnormally expressed in the p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells during G 0 compared with control p107 ϩ/Ϫ 3T3 cells (Fig. 6A). Some further increases in B-myb mRNA levels were seen when p107 Ϫ/Ϫ / p130 Ϫ/Ϫ 3T3 cells were induced to enter S phase, similar to findings with primary p107 Ϫ/Ϫ /p130 Ϫ/Ϫ fibroblasts (11), presumably reflecting the increased metabolism of these cells when serum-stimulated. Nonetheless, it is clear that even after establishment of an immortalized line, the dependence upon p107 and p130 for repression of the B-myb promoter in G 0 is maintained (Fig. 6A).
Significantly, transfection of p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells with a p107 expression plasmid resulted in repression of the B-myb promoter (Fig. 6B), whereas in contrast, wild-type Rb and a mutant Rb (RbNP, in which the major cyclin-dependent kinase phosphorylation sites have been eliminated) demonstrated little activity in this assay. Notably, the ability of p107 to repress the B-myb promoter in these cells was largely dependent upon an intact DRS (Fig. 6B).
Cell extracts made from parallel transfections were assayed by EMSAs. Antibody supershifts showed that similar levels of E2F complexes were formed with transfected p107 and Rb (Rb and RbNP were Myc epitope-tagged to distinguish these complexes from endogenous Rb complexes) (Fig. 6C). These experiments therefore indicate that p107 is intrinsically more active than Rb in repressing the B-myb promoter. Moreover, the fact that RbNP is also a weak repressor of B-myb transcription indicates that this differential activity cannot be explained by differences in the sensitivity of p107 and Rb to inactivation by cyclin-dependent kinases.
Repression of B-myb Transcription by p130 Requires Specific Protein Domains-To enable a more systematic comparison between the transcriptional repressive properties of the individual Rb family members on the B-myb promoter, further experiments were performed using Myc epitope-tagged derivatives of all three pocket proteins. By probing Western blots using the common epitope tag, it was evident that Rb was the most highly expressed of these proteins in transfected p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells (Fig. 7B), whereas p107 was relatively poorly expressed (a faint p107 band was detectable just below the background cellular band in Fig. 7B). Additional FIG. 5. Knockout p107 ؊/؊ /p130 ؊/؊ embryos contain abundant nuclear Rb⅐E2F complexes. Cytoplasmic (CYT) and nuclear (NUC) extracts were prepared from knockout p107 Ϫ/Ϫ /p130 Ϫ/Ϫ and control p107 ϩ/Ϫ MEFs that were arrested in G 0 by serum deprivation for 60 h. EMSA was performed with these extracts using a 32 P-labeled oligonucleotide probe containing the B-myb E2F/DRS sites. Control (Con; preimmune rabbit serum), anti-p130, anti-Rb, anti-p107, or anti-E2F-4 antibodies were added to EMSA reactions as indicated. Arrowheads indicate the positions of various E2F complexes and antibody (Ab) supershifts. A, p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEF cytoplasmic extracts; B, p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEF nuclear extracts; C, p107 ϩ/Ϫ MEF cytoplasmic extracts; D, p107 ϩ/Ϫ MEF nuclear extracts. experiments (data not shown) showed that the levels of expression that could be obtained with Myc epitope-tagged p107 were severalfold lower than those obtained with untagged p107 used in previous experiments (Figs. 3 and 6). Despite this, tagged p107 was more active than Rb in repressing the B-myb promoter in quiescent p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells, and notably, p130 had the greatest activity in this assay (Fig. 7A). These findings therefore emphasize that E2F complexes containing Rb are weakly active in transcriptional repression of B-myb in comparison with p107 and p130 complexes.
To define the domains of p130 required for transcriptional repression of B-myb, a number of deletion mutants were employed. These experiments showed that deletion of virtually the entire spacer region (as in mutant dl622-818) had no effect upon the ability of p130 to repress B-myb transcription, whereas in contrast, removal of the B pocket and C-terminal sequences (as in mutant C793) completely abolished activity. The lack of mutant C793 activity can clearly be attributed to its inability to bind E2F. The activity of a less extensively deleted C-terminal mutant (mutant C1065) was impaired to a lesser extent (Fig. 7C), and this defect may be attributed to loss of a potential C-terminal nuclear localization signal or its inability to bind histone deacetylase-1 (36). Most notably, deletion of sequences N-terminal to the pocket domain (as in mutant N417) significantly reduced transcriptional repression (Fig.  7C), although the truncated protein was expressed at much higher levels than wild-type p130 (Fig. 7D). The N-terminal domain is not required for interaction with E2F, and the reduced activity of mutant N417 indicates that this region contributes a novel functional requirement for transcriptional repression. It may be significant that p130 and p107 display significant homology in this N-terminal sequence, but little homology to Rb. The B-myb Promoter E2F Site Is Unbound in p107 Ϫ/Ϫ / p130 Ϫ/Ϫ Cells-A number of explanations could account for the inability of Rb⅐E2F complexes to repress B-myb transcription in p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs: (i) the conformation of the promoter in vivo may preclude strong interactions with Rb⅐E2F complexes while permitting binding of p107⅐E2F and p130⅐E2F complexes; (ii) binding of Rb⅐E2F complexes may be competed by higher levels of free E2F/DP heterodimers resulting from deregulation of E2F gene transcription (11); and (iii) Rb⅐E2F complexes may bind the site, but be unable to repress transcription. To help distinguish between these possibilities, we used the in vivo footprinting technique to study whether complexes are bound to the E2F site in quiescent p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs as would be predicted from Explanations ii and iii. Previous studies in NIH 3T3 cells demonstrated that the B-myb promoter E2F site is occupied in G 0 , but becomes unoccupied in G 1 concurrently with derepression of transcription (14). Using Swiss 3T3 cells initially, we confirmed that the E2F site was partially protected when intact cells were treated with dimethyl sulfate (Fig. 8A), but this degree of protection was lost when cells were stimulated by serum to reenter the cell cycle (16-and 24-h time points). Incomplete protection is a characteristic of this procedure and does not necessarily imply that the site is incompletely occupied. No protection of the DRS was observed, consistent with previous results (14), suggesting that interactions at this site are not made in the major groove. We next compared occupation of the E2F site in serum-starved MEFs derived from p107 Ϫ/Ϫ /p130 Ϫ/Ϫ and control (wild-type) embryos; as a further control, in vivo footprints were also obtained with quiescent Rb Ϫ/Ϫ MEFs. It was apparent that the E2F site was partially protected in both wild-type and Rb Ϫ/Ϫ MEFs (Fig. 8C); however, this site was completely unprotected in the p107 Ϫ/Ϫ / p130 Ϫ/Ϫ MEFs. These findings are therefore inconsistent with occupation of the E2F site by free E2F species or inactive Rb⅐E2F complexes and strongly imply that only p107⅐E2F and p130⅐E2F complexes are able to make stable interactions in vivo. DISCUSSION We have demonstrated here that the sequence requirements for B-myb promoter DRS corepressor function overlap with that of the CHR site, which is part of a bipartite element required for cell cycle regulation of the CDC25C, cyclin A, and cdc2 promoters (38). Notably, the CDE/CHR element was capable of regulating cell cycle transcription in place of the E2F/ DRS sites in the B-myb promoter; however, the DRS was un-FIG. 6. The B-myb promoter can be transcriptionally repressed by transfected p107 (but not Rb) in immortalized p107 ؊/؊ / p130 ؊/؊ 3T3 cells. A, shown is a Northern blot probed for B-myb expression in 3T3 cells derived from control p107 ϩ/Ϫ and knockout p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs. Cells were deprived of serum for 60 h to arrest in G 0 . Flow cytometry showed that both cell lines were quiescent at this time. The cells were then stimulated by re-addition of serum for 18 h (S phase), by which stage Ͼ60% were judged to be in S phase by flow cytometry. B, knockout p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells were transfected in triplicate with the wild-type (wt) B-myb promoter/luciferase reporter pGL2-(Ϫ536) or with variants carrying mutations in the DRS (mDRS) or mutations in both the E2F/DRS sites (mE2F/mDRS) together with a control pCMV plasmid or an expression vector encoding p107 or Myc epitope-tagged Rb and RbNP. Cells were arrested in G 0 by serum deprivation for 60 h, and luciferase activities were assayed in cell extracts. The results are presented as -fold repression of transcription by these pocket proteins, where the average luciferase activity obtained with the control empty expression plasmid transfection for each reporter was given a value of unity. C, EMSA was performed with extracts from knockout p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells transfected with the control pCMV vector (CON) or with an expression vector encoding p107, Rb, or RbNP. To detect E2F complexes containing the ectopically expressed pocket proteins, antibody (Ab) supershifts were done with anti-p107 antibody SD15 or antibody 9E10, which recognizes the Myc epitope tag present on the transfected Rb and RbNP proteins. able to substitute for the CHR in this context (Fig. 2B). This suggests that, although the E2F/DRS and CDE/CHR elements are similar in many respects, there is some homology between the respective elements, and their relative spacing is identical, there are certain functional differences between them. In the case of the cyclin A and cdc2 CDE/CHR elements, where the CDE can clearly double as an E2F site, these bipartite elements may act analogously to the B-myb E2F/DRS elements in binding p107⅐E2F and p130⅐E2F complexes in G 0 and early G 1 . Potentially, the cyclin A and cdc2 elements may then have a dual role, acting analogously to the CDC25C CDE/CHR elements to prolong active transcription into G 2 .
E2F complexed to each of the three pocket proteins can bind the B-myb E2F site in in vitro binding assays, and this is not influenced by the presence of the DRS (32). These assays clearly do not reflect the specificity of the B-myb promoter for binding p107⅐E2F and p130⅐E2F complexes as evidenced in this study by in vivo footprinting (Fig. 8) and repression assays (Figs. 6 and 7). Notably, previous ChIP assays have also led to the conclusion that the B-myb promoter is bound specifically by p107⅐E2F and p130⅐E2F complexes in vivo and that whereas Rb binds to other E2F-regulated promoters, these complexes are absent from B-myb (39). Indeed, this study suggested that E2F-4 complexes exclusively regulate B-myb in cells synchronized by serum starvation. In further distinction to other E2Fregulated promoters, the B-myb promoter is not occupied by E2F complexes during S phase (39,40). Altogether, published evidence and our current findings suggest that B-myb displays a distinctive mode of regulation that utilizes a closely linked DRS corepressor site to selectively bind transcriptionally repressive p130⅐E2F and p107⅐E2F complexes. A role for the DRS in excluding binding of transcriptionally active "free" E2F complexes is also suggested by our finding that mutation of the DRS led to activation of the promoter through the adjacent site in S phase ( Fig. 2A).
Clearly, a full understanding of the way in which DRS and CHR elements co-regulate transcription during the cell cycle requires identification of the factors that bind these sites. We FIG. 7. The p130 N terminus is required for transcriptional repression of the B-myb promoter. A, knockout p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells were transfected in triplicate with the wild-type B-myb promoter/luciferase reporter pGL2-(Ϫ536) together with 1-and 2-g amounts as indicated of a control pcDNA3 vector (CON) or pcDNA3 encoding Myc epitope-tagged Rb, p107, or p130. Cells were arrested in G 0 by serum deprivation for 60 h, and luciferase activities were assayed in cell extracts. The results are presented as -fold repression by these proteins on the reporter, where the average luciferase activity obtained with the control transfection was given a value of unity. B, Western blotting was carried out using antibody 9E10 as the primary antibody, and the results show the relative levels of expression of Rb, p107, and p130 in transfected p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells at inputs of 1 and 2 g of expression vector. Note that p107 appears as a faint band immediately below the major background band obtained in all lanes. The background band is indicated with an asterisk. C, transcriptional repression of the B-myb promoter by p130 requires the E2F binding domain and an N-terminal domain. Knockout p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells were transfected in triplicate with the wild-type B-myb promoter/ luciferase reporter pGL2-(Ϫ536) together with 0.2-, 0.5-, and 1 g-amounts as indicated of a control pcDNA3 vector or pcDNA3 encoding Myc-epitope-tagged p130 and p130 mutants N417 (amino acids 417-1139), C793 (amino acids 2-793), C1065 (amino acids 2-1065), and dl622-818 (deletion of amino acids 622-818 within the spacer). Cells were arrested in G 0 by serum deprivation for 60 h, and luciferase activities were assayed in cell extracts. The results are presented as -fold repression by these proteins on the reporter, where the average luciferase activity obtained with the control transfection was given a value of unity. D, Western blotting was performed using antibody 9E10 as the primary antibody, and the results show the relative levels of expression of wild-type and mutant p130 proteins in transfected p107 Ϫ/Ϫ /p130 Ϫ/Ϫ 3T3 cells at 1-g input of expression vector. Lane 1, wild-type p130; lane 2, mutant N417; lane 3, mutant C793; lane 4, mutant C1065; lane 5, mutant dl622-818. The major background band is indicated with an asterisk. E, shown is a schematic representation of Myc epitope-tagged p130 and p130 deletion mutants. Myc epitope-tagged p130 contains amino acids 2-1139 of wild-type p130, and the amino acid numbers defining the boundaries of the A and B pockets and the intervening spacer are indicated. Sequences retained in the p130 deletion mutants are indicated below.
have been unable to detect proteins binding to the DRS in in vitro binding assays, although in the presence of spermine, a novel lower mobility p130⅐E2F complex was detected in G 0 extracts from various rodent fibroblasts that was absent on probes in which the DRS was mutated. 2 A protein (CHF) that can bind the B-myb DRS has been purified (41); however, evidence that this binding correlates with the sequence requirements for transcriptional corepression is lacking. Evidence from in vivo footprinting that protein contacts are made with the CDC25C CHR in the minor groove suggests that minor groove contacts are also made with the B-myb DRS. The CHR and DRS may therefore act in some respects analogously to high mobility group I(Y) binding sites for assembly of an NF-B complex on the human interferon-␤ gene promoter (42). In this respect, it is of interest that the DRS sequence (GGAAA) is present within the PRDII site, which binds high mobility group I(Y) (42); however, the sequence specificity for high mobility group I(Y) binding within the interferon-␤ site is inconsistent with sequence requirements within the B-myb DRS.
Rb⅐E2F complexes were abundant in p107 Ϫ/Ϫ /p130 Ϫ/Ϫ cells (Fig. 5), yet were unable to repress efficiently the B-myb promoter. Moreover, transfected Rb was unable to substitute efficiently for the loss of p107 and p130 function in these cells (Figs. 6 and 7). The latter deficiency cannot be ascribed to hyperphosphorylation of Rb in the p107 Ϫ/Ϫ /p130 Ϫ/Ϫ cells, as RbNP was similarly inactive in repressing B-myb transcription (Fig. 6). It is also apparent that Rb⅐E2F complexes could be formed in the transfected p107 Ϫ/Ϫ /p130 Ϫ/Ϫ cells (Fig. 6). It can be concluded that Rb⅐E2F complexes have relatively low affinity for the B-myb promoter E2F site in vivo in comparison with p107⅐E2F and p130⅐E2F complexes. This conclusion is consistent with a previous ChIP analysis of E2F target gene specificity in NIH 3T3 cells (39), in which Rb could be detected on the promoters of a number of E2F-regulated genes (dihydrofolate reductase, thymidine kinase, cdc2, and cyclin E), albeit at different stages of the cell cycle, whereas in contrast, only p130 and p107 occupied the B-myb promoter. It is also notable that B-myb was found to have a singular mode of regulation by E2F in that it was the only promoter not to bind E2F complexes in S phase (39). As DRS mutations enabled B-myb transcription to be activated through the adjacent E2F site in S phase ( Fig.  2A), it may be suggested that the DRS promotes binding of transcriptionally repressive p107⅐E2F and p130⅐E2F complexes while preventing binding of transactivating E2F species.
Although corepressor CHR sites have been described in the promoters of several genes that are maximally activated during S/G 2 , such as CDC25C, cyclin A, polo-like kinase, cyclin B2, and survivin (38,(43)(44)(45), sites directly analogous to the B-myb DRS have not been described in other G 1 /S-regulated genes. This again points to the singular nature of B-myb regulation and illustrates how the precise positioning and context of an E2F site can influence cell cycle transcriptional control. It is notable that B-myb transcription was found to be the most strongly deregulated of ϳ20 genes tested in p107 Ϫ/Ϫ /p130 Ϫ/Ϫ MEFs (11), possibly reflecting the absolute dependence of p107⅐E2F and p130⅐E2F complexes, under the influence of the DRS corepressor, for transcriptional repression of B-myb.
In view of the significant role played by transcriptional repression in regulating B-myb mRNA levels, it is surprising that induction of B-myb expression is severely impaired in MEFs derived from E2F-3 Ϫ/Ϫ mouse embryos (46), implying that E2F-3 function is required to transcriptionally induce B-myb at the G 1 /S transition. This conclusion is not consistent with our finding that mutation of the B-myb promoter E2F site resulted in constitutive highly active transcription in stably transfected cells that contained a low copy number of the transgene (Fig.  4A), indicating that in this context, at least the E2F site is not required for activation by E2F-3. Moreover, ChIP data indicate that this promoter is not bound by E2F-3 or other E2F species at the G 1 /S transition (39,40), although binding of E2F-1 and E2F-3 was detected transiently in late G 1 (40). The failure of B-myb transcription to be fully induced following serum induction of quiescent E2F-3 Ϫ/Ϫ MEFs may therefore reflect the persistence of p130⅐E2F and/or p107⅐E2F repressor complexes under these conditions or, alternatively, a role for E2F-3 transactivation through distal promoter sites that are not retained in the B-myb promoter/reporters used. Analysis of B-myb promoter occupancy in E2F-3 Ϫ/Ϫ MEFs using the ChIP assay should help to resolve this issue.
The B-myb promoter has been used in many studies as a model for E2F regulation. Our current findings indicate that occupancy of this E2F site in vivo by transcriptionally repres-2 R. J. Watson, unpublished data.
FIG. 8. The B-myb E2F site is unoccupied in p107 ؊/؊ /p130 ؊/؊ MEFs. A, Swiss 3T3 cells were arrested in G 0 by serum deprivation for 60 h and then restimulated to reenter the cell cycle by addition of serum. Cells taken at 0, 16, and 24 h after restimulation as indicated were methylated in vivo with dimethyl sulfate. Footprints obtained by linker-mediated PCR on DNA extracted from the dimethyl sulfatetreated cells were compared with those of Swiss 3T3 DNA methylated in vitro (IV). The sequence of the (ϩ)-strand shown spans the B-myb promoter E2F/DRS sites in a 5Ј to 3Ј direction, top to bottom. B, shown are the results from flow cytometry of propidium iodide-stained Swiss 3T3 cells used in A at 0, 16, and 24 h after serum stimulation. C, MEFs obtained from wild-type (WT), Rb Ϫ/Ϫ , and p107 Ϫ/Ϫ /p130 Ϫ/Ϫ embryos were arrested in G 0 by serum deprivation for 72 h and then methylated in vivo with dimethyl sulfate. Footprints obtained by linker-mediated PCR on DNA extracted from the dimethyl sulfate-treated cells were compared with those of wild-type MEF DNA methylated in vitro (IV). The sequence of the (ϩ)-strand shown spans the B-myb promoter E2F/ DRS sites and a putative SP-1 site (which was occupied in vivo in all the cells tested) located 5Ј to the regulatory elements. sive complexes is governed by the adjacent DRS corepressor site. To date, we have been unable to identify proteins binding to the DRS, and we can only speculate on the mechanism whereby it acts to promote binding of complexes to the E2F site. Recent studies of the E2F-1 promoter have identified a nucleosome proximal to the E2F binding site (37), which may provide a target for histone acetylases and deacetylases to modulate transcription initiation. Possibly, the DRS is involved in nucleosome positioning, thereby influencing the ability of repressive E2F complexes to interact with the adjacent E2F site. The recognition that this corepressor site plays a prominent role in occupancy of the B-myb promoter by transcriptionally repressive p107⅐E2F and p130⅐E2F complexes will add significantly to future studies using this system as a model for E2F regulation.