A role for additional positive- and negative-acting promoter elements

Our recent work has demonstrated a role for cooperative promoter interactions involving E2Fs and other transcription factors, including TFE3 and YY1, as a mechanism to insure specificity of function (Schlisio et al, 2002; Giangrande et al, 2003, 2004). An inspection of cdc2 promoter sequences, together with past work analyzing the effects of mutation of various elements, suggests a role for at least three other sequences—a Myb element upstream of the distal E2F site, CCAAT elements situated between the E2F elements, and a CHR element adjacent to the proximal E2F element (Figure 3A). Previous work has shown Myb proteins to function as positive regulators of several cell cycle genes, including human cdc2 (Ku et al, 1993) and the Drosophila cyclin B1 gene (Okada et al, 2002). Likewise, the CCAAT elements are binding sites for the positive-acting NF-Y transcription factor (Maity and de Crombrugghe, 1998). They have been shown to be important for promoter activity of many cell cycle genes including cdc2, cycA2, and E2F1 (Zwicker et al, 1995a, 1995b; van Ginkel et al, 1997; Zwicker and Muller, 1997; Liu et al, 1998b). A protein(s) binding to the CHR element has not been identified, but the element has been shown to function as a cell cycle regulator in cooperation with E2F control (Liu et al, 1996).

Open in a separate window

Figure 3

Multiple elements control human cdc2 promoter activity. (A) Schematic of human cdc2 wild-type (WT) and mutant promoters that contain mutations in the MYB binding site (MYBm), the distal CCAAT element (dCCAATm), the CHR element (CHRm), or the proximal E2F site and the CHR element (pE2Fm+CHRm) respectively. The arrows indicate the G1/S transition in the cell cycle. (B) Mutation of the MYB binding site abolishes human cdc2 promoter activity. Left panel: REF52 cells were synchronized by blocking with HU. Right panel: REF52 cells were synchronized by serum starvation. The arrows indicate the G1/S transition. (C) Mutation of the distal CCAAT element dramatically reduces human cdc2 promoter activity. Left panel: REF52 cells were synchronized by blocking with HU. Right panel: REF52 cells were synchronized by serum starvation. The arrows indicate the G1/S transition. (D) Mutation of the CHR element increases human cdc2 promoter activity in quiescence. REF52 cells containing transiently transfected reporter constructs were serum starved for 2 days, and then harvested for luciferase activities.

As shown in Figure 3B, mutation of the Myb element abolished activation of the cdc2 promoter, either after release from an HU block at G1/S or following stimulation of cell growth by serum addition. Likewise, mutation of the distal CCAAT element also abolished promoter activation, equivalent to the mutation of the distal E2F element (Figure 3C). In contrast, mutation of the CHR element led to an increase of promoter activity, similar to the mutation of the proximal E2F site (Figure 3D). This result is the same as previously shown by Zwicker et al (1995b), who referred to the E2F repressive site as a CDE element, which has also been shown to bind E2F4 (Tommasi and Pfeifer, 1995). Taken together, these results suggest a role for multiple elements, both E2F and others, that act in both a positive and negative manner to control the activity of the cdc2 promoter.

Interaction of activator and repressor E2Fs with G2/M-regulated promoters

Given the apparent direct roles for E2F activity in the control of cdc2 and cyclin B1 transcription, both positive and negative, we have investigated the interaction of E2F proteins with the endogenous promoters. Because of availability of human promoter sequence, we made use of T98G cells to examine E2F interaction with these promoters. The expression pattern of these G2/M genes as well as several G1/S genes in the T98G cells was equivalent to that observed in the REF52 cells (Figure 4A, and data not shown). As shown in Figure 4B, both activator and repressor E2Fs were found to interact with the cdc2 and cyclin B1 promoters. Assays for E2F4 revealed an interaction with each of the promoters in the quiescent cell sample, similar to several recent reports (Takahashi et al, 2000; Schlisio et al, 2002). This included both the G2/M (cdc2, cyclin B1, cyclin A2)- and G1/S(PCNA)-regulated genes. This interaction was reduced or absent in the G1/S-blocked cells and remained low as cells progressed through S phase and mitosis (6–9 h time point). Interestingly, the E2F4 interaction then reappeared as the cells came back to the next G1/S (18 h time point).

Open in a separate window

Figure 4

Interaction of E2F proteins with G2-regulated promoters. (A) Analysis of cell cycle gene expression in T98G cells. Left panel: FACS analysis of T98G cells that were synchronized by HU block and released into fresh 10% FBS containing DMEM for the indicated times. Additional sample was prepared from serum-starved cells (Q: quiescence). Right panel: Northern analysis of cdc2, cyclin B1, cyclin A2, and PCNA messages. (B) Both activator and repressor E2Fs bind to endogenous G2-regulated promoters. T98G cells at quiescence (Q), HU block (0 h), or release from HU block (6, 9, and 18 h) were harvested for ChIP analysis using the indicated anti-E2F antibodies. Immunoprecipitated chromatin was analyzed by semiquantitative PCR using primers in the cdc2, cyclin B1, cyclin A2, PCNA, and β-actin promoters. PCNA, an E2F-regulated G1/S gene, is used here as a positive control, while β-actin, a non-E2F-regulated gene, is used as a negative control. Input material represents 0.05% of the total chromatin material used for a given ChIP reaction.

The pattern of interaction of the activator E2Fs was essentially opposite to that for E2F4. In particular, none of these proteins was bound to the promoters in quiescent cells, and then each could be detected on the promoters at G1/S (0 h time point). Each persisted through S phase (6 h sample) and then were absent as cells progressed through G2 and mitosis (9 h time point). Each was then found to interact with the promoters as cells came back to the next G1/S (18 h time point). Taken together, the chromatin immunoprecipitation (ChIP) results provide evidence for interaction of both positive- and negative-acting E2Fs with the G2-regulated promoters. But, the kinetics of these interactions do not suggest a basis for distinguishing control at G1/S versus G2/M.

Activator and repressor E2Fs interact with distinct promoter elements

In light of the clear distinction in the functional role of the individual E2F elements in the promoters, we investigated the relationship between this function and the nature of the E2F proteins that interact with these sites. To do so, we have made use of ChIP assays that specifically measure protein interactions with transfected plasmid sequences rather than interaction with the endogenous promoter sequences (Giangrande et al, 2004). In this way, we can measure the effect of mutation of a particular element on the chromatin interaction profile. We utilized the cdc2 promoter for these assays and measured chromatin interaction in either quiescent REF52 cells that were serum stimulated or G1/S–arrested cells that were released into the cell cycle. The promoter mutants used in the assays, which include alterations in either the proximal or distal E2F elements or both, are shown in Figure 5A.

Open in a separate window

Figure 5

Distinct interaction of E2F proteins with positive- and negative-acting E2F elements. (A) Schematic of wild-type (WT) and mutant human cdc2 promoters that contain mutations in the distal E2F site (dE2Fm), the proximal E2F site (pE2Fm), and both E2F sites (dbE2Fm). (B) Distinct interaction of E2F proteins with positive- and negative-acting E2F elements. REF52 cells containing transiently transfected reporter constructs were harvested at either HU block (0 h) or 6 h after release (6 h). Reporter plasmids containing chromatin material isolated as described in Materials and methods were immunoprecipitated using the indicated anti-E2F antibodies, and then analyzed by semiquantitative PCR using primers specific to transfected plasmids. Control reaction using the purified normal rabbit serum (NR) (Pierce) is shown. Primers specific to cotransfected plasmids harboring either β-gal (shown here) or renilla (data not shown) gene were used as negative controls. Input chromatin represents 0.1% of the total chromatin material in a given ChIP reaction.

Samples of REF52 cells transfected with wild-type and mutant human cdc2 promoter constructs were assayed for chromatin interaction of E2F family members. The results are shown in Figure 5B. As seen in the previous assays of the endogenous cdc2 promoter, E2F4 as well as E2F1–3 was bound to the promoter in the G1/S-arrested cells (0 h) (Figure 5B). As was the case for the endogenous assays, the interaction of E2F4 declined as cells were released into the cell cycle (6 h). Assays of the mutant promoters revealed that the activator E2Fs (E2F1–3) bound to the distal E2F site, while the repressor E2F4 bound to the proximal site. The interaction of E2F4 also coincided with an interaction of either p130 or p107 (data not shown). Strikingly, the E2F4 interaction persisted on the promoter containing the distal site mutation that blocked binding of E2F1–3 to the promoters. These results thus demonstrate that the activator and repressor E2Fs bind to distinct elements within the cdc2 promoter that coincide with function (activator or repressor elements). Moreover, the results suggest that the binding of the activator E2Fs displaces the repressor E2F complexes.

Roles for cooperative interactions involving activator and repressor E2Fs

The fact that the activator E2Fs and the repressor E2F bind to distinct promoter elements implies specificity in site recognition. Yet, the primary sequence suggests little potential for this difference based solely on the E2F interaction. Our recent work has demonstrated a role for cooperative interactions with other transcription factors as a basis for such E2F specificity (Schlisio et al, 2002; Giangrande et al, 2003, 2004). The data presented in Figure 3 demonstrate the role of additional elements in the control of cdc2 promoter activity; to investigate the function of these sequences, particularly in relation to the E2F interactions, we again performed ChIP assays using mutant promoter constructs.

Previous work has shown that the CCAAT element is a binding site for the NF-Y transcription factor (Maity and de Crombrugghe, 1998). As shown in Figure 6A, ChIP assays demonstrate an interaction of two of the NF-Y subunits (NF-YA and NF-YB) with each of the promoters. In contrast to the dynamic nature of the E2F interactions, the interaction of the NF-Y proteins is constant through the cell cycle, consistent with genomic footprint data showing that the CCAAT elements are occupied throughout cell cycle (Tommasi and Pfeifer, 1995). We next used plasmid-based ChIP assays to evaluate the role of the CCAAT element. As shown in Figure 3C, mutation of the distal CCAAT element abolishes promoter activation, equivalent to the mutation of the distal E2F element. The mutation of the CCAAT element also abolished the interaction of NF-YA with the cdc2 promoter, either at 0 or 6 h following release from the HU arrest (Figure 6B). Importantly, this mutation also abolishes the interaction of E2F3 with the cdc2 promoter, suggesting a cooperative binding of E2F3 to the promoter dependent on NF-Y.

Open in a separate window

Figure 6

A role for CCAAT and CHR elements in promoting the interaction of activator and repressor E2Fs. (A) NF-Y protein interacts with the endogenous cdc2, cyclin B1, cyclin A2, and PCNA promoters. Albumin promoter with no NF-Y binding was used as a negative control. T98G cells at quiescence (Q), HU block (0 h), or released from HU block (6, 9, and 18 h) were harvested for ChIP analysis using antibodies against NF-Y A and B subunits. (B) The binding of NF-Y to the distal CCAAT element is necessary for the binding of activator E2F3 to the distal E2F site. REF52 cells transiently transfected with wild-type (WT), the distal E2F site mutant (dE2Fm), and the distal CCAAT element mutant (dCCAATm) human cdc2 promoter reporter constructs were harvested at either HU block (0 h) or 6 h after release (6 h). The luciferase activities for these reporter constructs are shown in Figure 3C. The presence of the cdc2 promoter or control sequence (β-gal) from the transfected constructs in immunoprecipitates using anti-E2F3, anti-NF-YA, or control antibodies (NR) was detected by PCR as described in Figure 5. (C) An intact CHR element is necessary for E2F4 binding to the proximal E2F site. REF52 cells transiently transfected with the indicated reporter constructs were made quiescent by serum deprivation for 2 days. The luciferase activities for these constructs are shown in Figure 3D. Plasmid-based ChIP reaction was carried out using anti-E2F4 and control antibody (NR) and the samples were analyzed as described in Figure 5.

Finally, we also examined the effects of mutation of the CHR element, particularly with respect to the E2F4 interaction with the cdc2 promoter. Although the identity of the CHR binding transcription factor(s) has not been established, it does appear that the CHR element is critical for allowing the interaction of E2F4 since the mutation of the CHR element abolished the E2F interaction, equivalent to the effect of mutation of the proximal E2F element (Figure 6C).

Reporter plasmids and mutagenesis

pGL2hcdc2-wt, pGL2hcdc2-pE2Fm, and pGL2hcdc2-dCCAATm were generated by excising the human cdc2 promoters from pCAThcdc2-wt, pCAThcdc2-E2F4m, and pCAThcdc2-dCCAATm (Yun et al, 1999) into the pGL2-basic firefly luciferase reporter plasmid (Promega) respectively with HindIII and SalI. The HindIII site was filled in and then the promoter fragments were ligated to pGL2-basic cleaved with SmaI and XhoI. All the other mutations were generated by site-directed mutagenesis using the GeneEditor kit (Promega) following the manufacturer's instructions. Oligos used to create the distal E2F site mutation, the CHR mutation, the CHR and the proximal E2F site double mutation, and the B-MYB site mutation were ON11 (5′-CCTCTTTCTTTCGTAATCTAGCCACCC-3′), ON41 (5′-TTAGCGCGGTGAGTCGACAACTGCTCGCACTTGG-3′), ON42 (5′-CCTTTAATATTGTGAGTCGACAACTGCTCGCACTTGG CTTC-3′), and ON86 (5′-GCTGACTAGAGCTCGTAGGACGACACTCTCC-3′), respectively. pGL2upcB (Hwang et al, 1995) was kindly provided by Dr RJ Muschel (University of Pennsylvania). Oligos used to create the distal E2F site mutation and the proximal E2F site mutation were as follows: ON28 (5′-TGCGACCGGCAGCAACCAATGGGAAGGGAGTG-3′) and ON34 (5′-TAGGCTGGCTCTTCTCGTTGTGCTGCGGCGGAA-3′).

Luciferase analysis

REF52 cells were transfected using Superfect (Qiagen) following the manufacturer's instruction. The next day, transfected cells were split 1 to 3 into 24-well culture dishes, and starved in 0.1% FBS containing DMEM for 24 h. Cells were then released from starvation and blocked at the G1/S transition by the addition of DMEM containing 10% FBS and 2 mM HU for 18 h. To release from HU, cells were washed twice with DMEM, and then grown in fresh DMEM containing 10% FBS. At desired time points, cells were harvested and activities of firefly, Renilla luciferases, or β-galactosidase were analyzed following the manufacturer's instructions (Promega). pRL-Renilla or pCMV-βgal was used as an internal control for normalization.

RNA preparation and blotting

REF52 cells or T98G cells were brought to quiescence by serum deprivation for 2 days. Cells were then blocked at the G1/S transition by the addition of DMEM containing 10% FBS and 2 mM HU for 18 h, and then released into the cell cycle by the addition of fresh DMEM containing 10% FBS. At desired time points, cells were harvest for total RNA isolation using Trizol Reagent (Invitrogen). Messenger RNA was then purified using the PolyATract mRNA Isolation System IV following the manufacturer's instructions (Promega). Approximately 2 μg of mRNA for each time point was fractionated by electrophoresis in a 1.0% agarose gel under denaturing conditions, transferred to the GeneScreen membrane (NEN Life Science), and probed with ³²P-labeled probes as described previously (DeGregori et al, 1995). Mouse cDNA fragments used as probes were made from image clones purchased from Resgenetics Inc.

Chromatin immunoprecipitation

T98G cells were starved in 0.1% FBS containing serum for 2 days, and then grown for 18 h in DMEM containing 10% FBS and 1 mM HU. Cells were then washed twice with DMEM, and released from HU blocking by the addition of 10% FBS medium. At each time point, cells were crosslinked in 1% formaldehyde for 15 min at room temperature. ChIP experiments were carried out as described previously (Takahashi et al, 2000) with the following modifications. Briefly, after washing with Buffer 2, isolated nuclei were lysed in 1 × RIPA buffer (50 mM HEPES, pH 7.5, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 1 mM DTT) plus a proteinase inhibitor cocktail (1 μg/ml leupeptin, 1 mM AEBSF, 1 μg/ml pepstatin, 1 μg/ml aprotinin) at 4°C for 30 min. The chromatin material was pelleted by centrifugation at 1000 g for 10 min, resuspended in Buffer 3 and then sonicated to generate DNA fragments with an average size of 1 kb. A 2 μg portion of antibody was added to approximately 600 μg of chromatin material and collected with 20 μl of protein A/G agarose beads (Oncogen) that had been preblocked with 1% BSA and 0.1 μg/μl of poly(dIdC) (Roche). Immunoprecipitated chromatin was washed six times with 1 × RIPA buffer. The samples were then de-crosslinked at 65°C in 1 × TE and 0.5% SDS, phenol extracted, ethanol precipitated, and resuspended in 100 μl 1 × TE followed by polymerase chain reaction (PCR) in the presence of [³²P]dCTP. PCR products were fractionated on 8% native polyacrylamide gels and visualized by autoradiography. Antibodies used in ChIP experiments were as follows: anti-E2F1 (sc-251), anti-E2F2 (sc-633), anti-E2F3 (sc-879), anti-E2F4 (sc-1082X), anti-NF-YA (CBF-B, sc-10779), anti-NF-YB (CBF-A, sc-13045), anti-B-Myb (a mixture of N19, sc-729 and H115, sc-13028), and anti-C-Myb (a mixture of c-2, SC-8412 and H-144, SC-7874) (Santa Cruz Biotechnology). The sequences of the primers used to amplify promoters were as follows: cdc2 (5′-GCTTGCGCTCGCAC TCAGTTGGCG-3′, 5′-CAGATCCCTGACCTCCAGTCC-3′), cyclin A2 (5′-CTGCTC AGTTTCCTTTGGTTTACC-3′, 5′-CAAAGACGCCCAGAGATGCAG-3′), cyclin B1 (5′-CGATCGCCCTGGAAACGCATTC-3′, 5′-CCAGCAGAAACCAACAGCCGTTC-3′), PCNA (5′-CTTCCTCCAATGTATGCTCTAGG-3′, 5′-AGACAACGACCACTCT GCTACG-3′), p68 (5′-GATGAACCAAGGGCACAACAGGCAG-3′, 5′-GCGTGTGG GCGTCTCTACGCACGC-3′), DHFR (5′-GGCCTCGCCTGCACAAATAG GG-3′, 5′-GGGCAGAAATCAGCAACTGGGC-3′), β-actin (5′-CAAGGCGGCCAACGCCAAA ACTCT-3′, 5′-GCCATAAAAGGCAACTTTCGGAACG-3′), and albumin (5′-TGGGG TTGACAGAAGAGAAAAGC-3′, 5′-TACATTGACAAGGTCTTGTGGAG-3′).

In C-Myb ChIP experiments, immunocomplexes were eluted from protein A/G beads in 1 × TE and 1% SDS buffer by heating at 65°C for 10 min after the final wash. In all, 50% of each sample were fractioned in 8.5% SDS–PAGE gel and subjected to Western blot using the mouse monoclonal anti-C-Myb antibody (SC-8412; Santa Cruz). The other 50% of each sample were de-crosslinked overnight and processed for PCR analysis.

Reporter chromatin immunoprecipitation

REF52 cells at 70% confluence were transfected with reporter plasmids using Superfect following the manufacturer's instructions (Qiagen). Briefly, for a 150 mm tissue culture dish, 30 μg reporter plasmid, 5 μg pRL-Renilla, and 5 μg of pCMV-βgal were mixed with 60 μl Superfect reagent and 700 μl OPTI-DMEM medium. After incubation at room temperature for 20 min, the Superfect-DNA mix was then added to each dish along with 7 ml DMEM containing 10% FBS. After approximately 6 h incubation, cells were washed with 1 × PBS and then recovered overnight in DMEM containing 10% FBS. Transfected cells were split 1 to 3 in DMEM containing 0.1% FBS to render them quiescent. They were either harvested at quiescence, or at different time points after HU blocking as described in ChIP assay (see above). The material used for reporter ChIP is the soluble material extracted with 1 × RIPA buffer after the Buffer 2 wash. The endogenous chromatin material is all in the pellet. Antibodies used in reporter ChIP experiments are described above. The sequences of primers used in this assay were as follows: cdc2 (5′-GCTTGCGCTCGCACTCAGTTGGCG-3′, a pGL2-basic-specific primer GLprimer2) (Promega), Renilla (5′-GGAAACGGATGATAACTGGTC-3′, 5′-TGCCCA TACCAATAAGGTCTGG-3′), and β-gal (5′-ACTGGCAGATGCACGGTTACGATG-3′, 5′-CACATCTGAACTTCAGCCTCCAG-3′).

We thank Dr Deug Y Shin and Dr RJ Muschel for generously providing us the promoters of human cdc2 and cyclin B1 genes. We also thank Kaye Culler for help with the preparation of the manuscript. JRN is an Investigator of the Howard Hughes Medical Institute. WZ was supported by the Howard Hughes Medical Institute.