FOXO Transcription Factors Cooperate with δEF1 to Activate Growth Suppressive Genes in B Lymphocytes

Primary cells and cell lines

BALB/c mice were purchased from The Jackson Laboratory and bred in our colony. Animals were housed and studied in accordance with protocols approved by the Institutional Animal Care and Use Committee. Splenic B cells were purified, as previously described (15).

A20 (from American Type Culture Collection) and M12 (from M. Schlissel, University of California, Berkeley, CA) B lymphoma cells were maintained in RPMI 1640 medium supplemented with 10% FCS (HyClone), 10 mM HEPES, 2 mM l-glutamine, 50 μM 2-ME (Sigma-Aldrich), and antibiotics in a humidified incubator at 37°C in 5% CO₂. During experiments, the cultures were maintained in the exponential phase of growth. All supplements not indicated were obtained from Mediatech.

Cos7 cells were obtained from C. Hughes (University of California, Irvine, CA). The 293T cells were obtained from American Type Culture Collection. Both cell lines were maintained in DMEM supplemented with 10% FBS and 2 mM glutamine.

Reporter and expression plasmids

A 1464-bp fragment corresponding to the Rbl2 5′ flanking sequence −1413 to +51 was amplified by PCR from the mouse genomic DNA using the primer set CGC GGA TCC TGC TCC CGC AAC ACA TGT GAG and CGC GGA TCC GCG GCT CTT CAG ATG CGC TCA GG with a BamHI site at both of the primer ends. A 2200-bp fragment corresponding to the Ccng2 5′ flanking sequence −1554 to +646 was amplified by PCR from the mouse genomic DNA using the primer set GGG GTA CCC CAG TCT TCT TGT TCG GCT CC and GGG GTA CCC CTG TCA GTC CCT CAG CTG CG with a KpnI site at both of the primer ends. Both PCR fragments were gel purified and cloned into the pDrive cloning vector (Qiagen). The Rbl2 promoter fragment was excised with BamHI and subcloned into the BglII site of pGL2.Basic (Promega). The Ccng2 promoter fragment was excised with KpnI and subcloned into the same site of pGL2.Basic. Sequence and orientation were checked both by restriction enzyme digestion and sequencing (University of California Davis sequencing). A series of pGL2.Rbl2 promoter and pGL2.Ccng2 truncation mutants, designed to remove potential FOXO and δEF1 DNA binding sites, were created by PCR using a series of nested 5′ and 3′ oligonucleotide primers. Those promoters designed to mutate the putative binding site FOXO and δEF1 DNA binding sites were created by oligonucleotide-mediated mutagenesis using the QuickChange site-directed mutagenesis kit (Stratagene) and subcloned into the MCS site of pGL2.Basic. The DNA sequence of each promoter mutant was confirmed before use in transient transfection assays.

pGL2.TKmin promoter reporter was provided by M. Waterman (University of California, Irvine, CA). pGL2.PU.1 was provided by E. Rothenberg (California Institute of Technology, Pasadena, CA). The pMIT empty vector, pMIT.FOXO1, pMIT.FOXO1.A3, pMIT.FOXO3a, and pMIT.FOXO3a-A3 vectors used for expression of FOXO genes were previously described (15). pECE.FOXO3a-A3-ER was provided by A. Brunet (Stanford University, Stanford, CA). To subclone into the pMIT vector, pECE.FOXO3a-A3-ER was digested with BglII (partial) and NotI to obtain the 5′ 2-kb fragment, and inserted into the same sites of the pMIT vector. pECE.FOXO3a-A3-ER was then digested with NotI to obtain the 3′ 1-kb fragment and ligated with the 2-kb fragment in pMIT at NotI site. The sequence and orientation were checked both by restriction enzyme digestions and sequencing.

pcDNA3.1/mδEF1 was a gift from Y. Higashi (Osaka University, Osaka, Japan).

pCI.CycG2 vector encoding the cyclin G₂ cDNA was obtained from E. Conner (National Cancer Institute, Bethesda, MD). pCI.CycG2 was digested by XhoI and NotI, and the insert was cloned into same sites of the pMSCV.IRES.hCD4 (pMIC) vector. pcDNA3.Rbl2/p130 encoding the human Rbl2/p130 cDNA was a gift from A. Giordano (Temple University, Philadelphia, PA) and was digested with BamHI and NotI and cloned into the BglII and NotI sites of pMIT.

Transient transfection experiments and luciferase assays

Transient transfection experiments in A20 or M12 cells were performed by electroporation (Bio-Rad Genepulser). Exponentially growing cells were washed with RPMI 1640 medium without serum and resuspended in RPMI 1640 medium at 2 × 10⁷ cells/ml. A total of 500 μl of cell suspensions was mixed with the promoter-reporters and various expression vectors and incubated for 10 min at room temperature. Cells and DNA mixture were then transferred to 4-mm Genepulser cuvettes (Invitrogen Life Technologies) and electroporated at 300 V and 960 μF. The total amount of DNA per transfection was made equal by the addition of an appropriate amount of the pGEM3Z vector (Promega) or pMIT or pcDNA3.1 (Invitrogen Life Technologies) empty vectors. After electroporation, cells were immediately transferred to tissue culture flask with 10 ml B cell medium. Forty hours posttransfection, the cells were lysed and luciferase activity was measured using a commercial luciferase assay kit (Promega). Luciferase activity was normalized based on protein concentration, as assayed by BCA Protein Assay Reagent (Pierce).

EMSA

Cos7 cells were transfected with 10 μg of pECE.Foxo3a-A3, pcDNA3.FOXO1, pcDNA3.FOXO1H215R, pcDNA3.FOXO1.A3, or pcDNA3.1 empty vector by calcium phosphate precipitation. At 48 h posttransfection, cells were washed twice with PBS, and 300 μl of lysis buffer (10 mM Tris-HCl (pH 7.4), 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 5 μM ZnCl₂, 100 μM sodium orthovanadate, 1% Triton X-100, and protease inhibitors (Complete, protease inhibitor mixture tablet; Roche)) was added to each dish. Cells were harvested by scraping and transferred into microcentrifuge tubes, vortexed for 30 s, and centrifuged at the maximum setting for 20 min at 4°C. Supernatants were aliquoted to microcentrifuge tubes and stored at −80°C. The 293T cells were transfected by pcDNA3.δEF1myc or pcDNA3.1, as described for Cos7 cells. EMSA was conducted using Cos7 or 293T cell extracts, as described (22). A total of 100 ng of a double-stranded oligodeoxynucleotide, containing various FOXO or δEF1 binding sites, was labeled with [α-³²P]dCTP (3000 mCi/mmol; PerkinElmer) by fill-in reaction using the Klenow fragment of DNA polymerase I. Unincorporated nucleotides were removed from the labeling reactions using the Nucleotide Removal Kit (Qiagen). For supershift experiments, nuclear extracts were preincubated with 1 μg of anti-FOXO3a (Upstate Biotechnology) and anti-human c-Myc epitope 9E10 (Santa Cruz Biotechnology) Abs for 30 min at 4°C. Competition experiments were performed with 1:100 molar ratios of labeled to nonlabeled oligonucleotides. The oligonucleotides used for labeling were as follows: Rbl2-foxo-a (5′-CTA GTT TAT TTT GTT TTT GTT TG-3′), Rbl2-foxo-b (5′-AGC TCG TTT TTT GCT TTC TTC TTC-3′), Ccng2-foxo-a (5′-AGA AAG TAA AAC AAA CAA A CA AAA CCA AAAC-3′), Ccng2-foxo-b (5′-CTG AAA ACC AAA AGC AAA CAG TAC AAC-3′), Ccng2-foxo-c (5′-CTA GAC AGC GAA AAC AAA ACA AAT CGG-3′), Ccng2-foxo-d (5′-ATC GCA GAC TCA AAAC AAA AAC AAG G-3′), Ccng2-δEF1-a (5′-GAC CTT TAA AAG GTG TGA GGC GT-3′), and Rbl2-δEF1-c (5′-CGC ACT CTC CTC CCT CAG GTG GCT CAG-3′). The sequence of mutant oligos used for cold competitions are: Rbl2-foxo-a-mut (5′-CTA GTT TCT TTT CTT TTT CTT TC-3′), Ccng2-foxo-a-mut (5′-AGA AAG TAA AAG AAA GAA A GA AAA GCA AAAC-3′), and Ccng2-foxo-d-mut (5′-ATC GCA GAC TCA AAAG AAA AAG AAG G-3′).

Chromatin immunoprecipitation (ChIP) assay

ChIP assay was performed using the anti-acetylated histone H3 Ab-based ChIP assay kit (Upstate Biotechnology) with minor modifications. In brief, 37% formaldehyde solution was added directly to purified primary B cells at a final concentration of 1% and cells were incubated for 10 min at 37°C. Cells were harvested and washed twice with ice-cold PBS containing protease inhibitors (Roche) and lysed with SDS lysis buffer for 10 min on ice. The lysate was sonicated to an average length of 500-1000 bp and clarified by centrifuging at 16,000 × g for 10 min at 4°C. The samples were then precleared with salmon-sperm DNA-protein A-Ag beads (Upstate Biotechnology) for 2 h. After a brief spin, the supernatant fraction was collected and diluted 10-fold with ChIP dilution buffer and split into five equal parts for immunoprecipitation with either normal rabbit IgG (Santa Cruz Biotechnology; SC-2027), anti-FOXO1 (Cell Signaling Technology), anti-FOXO3a (Upstate Biotechnology), anti-δEF1 (anti-ZEB1; Santa Cruz Biotechnology), or no Ab. Immunoprecipitation was done by overnight rotation at 4°C. Beads were harvested, washed, and eluted per the manufacturers’ instructions. Cross-links were reversed by heating at 65°C for 4 h. DNA was purified by phenol-chloroform extraction and recovered by ethanol precipitation. Precipitated DNA was analyzed by PCR amplification for 25 cycles (94°C for 1 min, 55°C for 1 min, and 72°C for 1 min), separated on 1.5% Agarose gel, transferred to Nylon membranes, and hybridized to each probe (same as the ChIP PCR fragments). The primer sets used for PCR are as follows: Rbl2 forward, 5′-GTG TGA TTG AGG GTT GAG GGT TGA G-3′; Rbl2 reverse, 5′-GCC TGC TGG GAA ATG TAG TCT CG-3′; Ccng2 forward, 5′-CAG AGG ACG CAA CAG CCA GTC GG-3′; Ccng2 reverse, 5′-CCG CAG CCT CCG GAG CTG AGC-3′. Probes were amplified by PCR from the cloned promoters using the same primer sets. The PCR products were gel purified using the Qiagen gel extraction kit and radiolabeled by using RadPrime DNA Labeling System (Invitrogen Life Technologies). Hybridization was performed at 42°C for 24 h, according to the protocol provided by BD Clontech. Filters were exposed to a phosphor screen and visualized by using a phosphor imager system. For the ChIP control, PCR primers for Carbonic anhydrase I (CA-I) promoter were provided by T. Bender (University of Virginia, Charlottesville, VA). The CA-I ChIP DNA was PCR amplified for 35 cycles (94 °C for 1 min, 58°C for 45 s, 72°C for 45 s, with a 7-min final extension at 72°C), and PCR samples were then resolved on 2.5% agarose gels and visualized by ethidium bromide staining.

Northern blot and Q-PCR analysis

A20 cells were transfected by electroporation, as described above. Five batches of 10⁷ cells received 10 μg of pMIT-FOXO3a.A3-ER and 10 μg of either empty vector (pMIT) or pMIT-δEF1. Forty-eight hours after transfection, cells were harvested, pooled, and resuspended in 40 μl of MACS buffer per 10⁷ total cells. Cells were first stained with anti-Thy-1.1-biotin (eBioscience) and then incubated with anti-biotin microbeads (Miltenyi Biotec), and purified via positive selection on MACS columns per the manufacturers’ instructions. Purity of Thy-1.1-positive cells was greater than 70% via FACS analysis. Cells were cultured for an additional 16 h in the absence or presence of 400 nM 4-hydroxytamoxifen (Sigma-Aldrich). Total cellular RNA was extracted by using TRIzol reagent (Invitrogen Life Technologies). Fractionation of RNA and transfer to nitrocellulose membranes were conducted, as previously described (23). The following [α-³²P]dCTP probes were used: a NotI and XhoI fragment encoding the whole mouse Ccng2 cDNA cut from the pCI.CycG2, a 578-bp mouse Rbl2 cDNA fragment amplified from A20 cells by RT-PCR (forward primer, 5′-CCA GTG ATG AGG TCA AAC AGC ACC-3′; reverse primer, 5′-GCA TTT GCT GCC ACA CTG AGT CAG-3′), and a 1.2-kb PstI fragment of chicken GAPDH cDNA fragment (23). Hybridization conditions were conducted, as described for the ChIP assays. Filters were exposed to a phosphor screen, and results were quantified using ImageQuant software (Amersham). cDNA was synthesized, and real-time quantitative PCR using SyBr Green was performed, as described, using primers specific for p130, cyclin G₂, and β-actin cDNA (15).

Retroviral infection and cell cycle analysis of primary B cell blasts were performed, as described (15). Cell cycle analysis of transfected A20 cells was done similarly. In both cases, FlowJo software was used to calculate cell cycle distribution of cell populations gated by expression of the IRES-linked surface marker gene product.

Cyclin G₂ suppresses B cell proliferation, and the Ccng2 promoter contains IRS motifs

Previously, we reported that retroviral transduction of FOXO proteins into activated splenic B cell blasts induces partial cell cycle arrest or delay (15). We also noted that mRNA levels of Ccng2 and Rbl2 are diminished following BCR cross-linking of splenic B cells, in a manner partially dependent on PI3K (15). These observations suggested that Ccng2 and Rbl2 might be FOXO target genes in B cells. Both cyclin G₂ and p130/Rb2 can suppress cell proliferation when overexpressed in certain cell types (24, 25). To determine whether expression of cyclin G₂ or p130 alone has any effect on cell cycle progression in this system, we generated retroviruses containing cDNAs for these proteins. Retroviral transduction of cyclin G₂, but not p130, caused an increase in the fraction of B cells in G₀-G₁ phase of the cell cycle, with a concominant decrease in S phase (Fig. 1⇓). The effect was comparable to that achieved by expression of wild-type FOXO1 or FOXO3a. These findings are consistent with the possibility that induction of cyclin G₂ by FOXO expression contributes to cell cycle arrest in activated mature B cells. As observed previously (15), A3 mutants of FOXO proteins in which all Akt phosphorylation sites have been mutated were more effective at inducing cell cycle arrest/delay, with FOXO1.A3 also causing a significant decrease in G₂-M phase.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Cyclin G₂ induces cell cycle arrest or delay in primary B cell blasts. LPS-stimulated splenic B cells were infected with MSCV-based retroviruses expressing cyclin G₂, p130, FOXO1, FOXO1.A3, FOXO3a, FOXo3a.A3, or their empty vector controls (pMIC or pMIT). Cell cycle analysis was performed 48 h later using propidium iodide, gating on marker-positive (hCD4 or Thy1.1) cells. The data represent the mean ± SEM of three independent experiments. A t test was done to determine statistical differences of cell cycle stages relative to empty vector control (pMIC or pMIT). ∗, p < 0.05; ∗∗, p < 0.005.

FOXO consensus (IRS) elements have been found in the murine Ccng2 promoter (21) and in the first intron of the human Rbl2 gene (11). We found two motifs with close similarity to IRS elements in the murine Rbl2 promoter. The locations of these motifs and the four IRS-like elements in the murine Ccng2 regulatory region are shown in Fig. 2⇓. In the remainder of this study, we focused primarily on the promoter of Ccng2, although most observations were reproduced with the Rbl2 promoter.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Locations of consensus binding sites (underlined) for FOXO and δEF1 transcription factors in the regulatory regions of the Ccng2 and Rbl2 genes. The start sites of transcription (obtained from the Ensembl Mouse Genome Server) are shown as nucleotide no. 1. Bold type, primers used to clone the promoters; italic type, primers used to clone del5 fragment.

FOXO proteins enhance transcription of Ccng2 and Rbl2 promoters

We tested the ability of FOXO proteins to activate transcription from these promoters in reporter assays. Fragments of the Ccng2 and Rbl2 promoters were cloned upstream of the luciferase gene in pGL2.Basic. Transient transfections were conducted in A20 B lymphoma cells. Both promoter-reporters were more basally active in A20 cells compared with the pGL2.Basic vector or the promoters cloned in the reverse orientation (Fig. 3⇓, A and B). Cotransfection of FOXO1 or FOXO3a enhanced this activation in a dose-dependent manner, ranging from 1.5- to 10-fold over promoter-reporter alone (Fig. 3⇓, A and B).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Transfection of A20 cells with FOXO constructs augments transcription of reporter genes driven by the Ccng2 or Rbl2 promoters. A and B, A20 cells were electroporated with 5 μg of pGL2.Basic, pGL2.Ccng2⁽⁻⁾ or pGL2.Ccng2⁽⁺⁾ (A) and pGL2.Rbl2⁽⁻⁾ or pGL2.Rbl2⁽⁺⁾ (B) alone or with FOXO expression vectors as indicated. The empty vector pMIT was added to make the total amount of DNA equal in all transfections. The bars indicate the average luciferase activity normalized to protein concentration. Luciferase activity of the reporter gene in the absence of exogenous transactivator was designated as one. All luciferase data shown represent the average of three independent experiments. Error bars indicate SD of the mean.

Akt-mediated phosphorylation of FOXO proteins at three conserved residues leads to cytoplasmic sequestration and inactivation of these transcription factors (2, 3, 27). Consistent with this paradigm, cotransfection of wild-type Akt suppressed FOXO1-mediated Ccng2 promoter activation, whereas a kinase-dead Akt enhanced the effect of FOXO expression (Fig. 4⇓). Akt expression did not oppose the effect of the FOXO1.A3 mutant in which all three phosphorylation sites are mutated to alanines.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Akt opposes promoter activation by wild-type FOXO. cDNAs for wild-type Akt or kinase-dead (K179M) were cotransfected with FOXO1 or FOXO1.A3 as indicated, and analyzed for luciferase activity as described in Fig. 3. One of three independent experiments is shown (bars represent mean ± range of duplicate luciferase determinations). The average (mean ± SD; n = 3) of the ratio of fold changes for FOXO1+Akt vs FOXO1 alone was 0.5 ± 0.06. The average (mean ± SD; n = 3) of the ratio of fold changes for FOXO1.A3+Akt vs FOXO1.A3 alone was 2.1 ± 1.5.

δEF1 synergizes with FOXO proteins to activate Ccng2 and Rbl2 promoters

Previous studies have shown that FOXO proteins cooperate with other transcription factors and coactivators to regulate promoter function (20, 28). Further analysis of the promoter sequences of Ccng2 and Rbl2 using the Matinspector program revealed several sites that precisely match the consensus for binding (AGGTG) by the transcription factor δEF1 (other names include ZEB1, AREB6, MEB1, BZP, and zfhep) (29, 30) (Fig. 2⇑). This zinc finger protein can be either a repressor or activator of transcription in different systems (29, 31). δEF1 has been shown to function in T cell development and to repress IL-2 transcription in T cells (32, 33), but its function in B cells has not been investigated. Transfection of A20 cells with δEF1 alone augmented Ccng2 and Rbl2 promoter activity 10- to 20-fold (Fig. 5⇓A). Cotransfection of δEF1 and FOXO1 or FOXO3a resulted in synergistic activation of the promoters (Fig. 5⇓A). Supporting the idea that cooperation between FOXO proteins and δEF1 is a general phenomenon in mature B cells, we observed similar effects in M12 B lymphoma cells (Fig. 5⇓B). Importantly, the synergistic transactivation of the Ccng2 and Rbl2 promoters by FOXO1 or FOXO3a and δEF1 was promoter specific, as FOXO, δEF1, or the combination did not activate control constructs, including the pGL2.TKmin, pGL2.PU.1, pGL2.basic, Rbl2, or Ccng2 promoters in reverse orientation (Fig. 5⇓C and data not shown). This demonstrates that the increased transcriptional activity from the Ccng2 and Rbl2 promoters was not due to a general increase in transcription resulting from forced expression of FOXO proteins and/or δEF1. Together, these findings indicate that FOXO proteins and δEF1 cooperate functionally to specifically augment transcription from the Ccng2 and Rbl2 promoters in B cells.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

δEF1 activates the Ccng2 and Rbl2 promoters and synergizes with FOXO factors. A, A20 cells were cotransfected with the indicated reporter constructs and FOXO or/and δEF1 expression vectors, and analyzed for luciferase activity as described in Fig. 3. B, Luciferase reporter assays in M12 cells, a murine IgG⁺ mature B cell lymphoma. C, Control promoters pGL2.TKmin and pGL2.PU.1 are not activated by FOXO and/or δEF1 in A20 cells. A, The mean ± SE of three independent experiments. B and C, Data that are representative of three independent experiments with similar results (error bars represent range of duplicate measurements).

FOXO and δEF1 factors bind to Ccng2 and Rbl2 promoters

We performed EMSA experiments to determine whether the consensus-binding elements for FOXO and δEF1 could bind these factors in vitro. Extracts of Cos7 cells transfected with FOXO constructs were used as a source of abundant FOXO protein. These extracts formed specific binding complexes with the four IRS elements of the Ccng2 promoter, and with two IRS elements in the Rbl2 promoter (Fig. 6⇓, A and B, and data not shown), as compared with the extracts of cells transfected by empty vector constructs. The specificity of these complexes was confirmed by incubating with Ab against FOXO3a, which caused supershifts in the Ccng2 probes and inhibited the specific binding of the Rbl2 probe. The specificity of the FOXO3a-DNA complex was also shown by competition by a 100-fold molar excess of unlabeled probe, but not by oligomers that have mutated FOXO binding sites. Extracts of 293T cells transfected with δEF1 also formed specific complexes with consensus-binding elements in both the Ccng2 and Rbl2 promoters (Fig. 6⇓C). Similarly, this specific complex could be supershifted by incubating with 9E10 Ab directed against the Myc epitope tag, and inhibited by excess unlabeled probes, but not affected by mouse preimmune serum or irrelevant oligos.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

FOXO and δEF1 bind to isolated promoter elements in EMSA assays, and to the endogenous promoters in ChIP assays. A, FOXO3a binds to four IRS motifs in the Ccng2 promoter. Specificity controls are shown for two of these sites. Cos7 cell extracts transfected by FOXO3.A3 or empty expression vector (−) were incubated with ³²P-end-labeled oligonucleotide probes. Normal rabbit IgG (lanes 3 and 9) or Abs against FOXO3a (lanes 4 and 10) were included in the binding reactions. Specificity of the FOXO3a-DNA complex was also shown by competition of a 100-fold molar excess of unlabeled probes (lanes 5 and 11), but not by DNA oligos with mutated IRS motifs (lanes 6 and 12). A representative result of three independent experiments is shown. B, FOXO3a binds to two IRS motifs in the Rbl2 promoter. Specificity controls are shown for one of these sites. C, δEF1 binds to the consensus δEF1 motifs from Ccng2 and Rbl2 promoter region. Specificity controls are shown for both of these sites. 293T lysates transfected with δEF1 or empty vector were used in the reactions. 9E10 Ab directed against the δEF1-Myc tag supershifts the specific δEF1 complex (lanes 3 and 9). D, FOXO proteins and δEF1 are able to bind the Ccng2 and Rbl2 promoters in vivo. Chromatin proteins were cross-linked to DNA by formaldehyde in primary mouse B cells and immunoprecipitated by specific Abs (to FOXO1, FOXO3a, and δEF1), nonimmune IgG control or no Ab (H₂O). PCR in the exponential phase were detected by Southern blotting for Ccng2 and Rbl2 promoter amplicons, and by ethidium bromide staining for the CAI promoter amplicon.

To determine whether these transcription factors are bound to the Ccng2 promoter in resting splenic B cells, we conducted ChIP assays. Abs specific for FOXO1, FOXO3a, and δEF1 all immunoprecipitated fragments of genomic DNA that could be amplified by primers specific to the Ccng2 promoter, whereas a control IgG did not (Fig. 6⇑D). In the case of the Rbl2 promoter, FOXO3a and δEF1 were detected by ChIP, but FOXO1 was not (Fig. 6⇑D). Immunoprecipitation of the same chromatin preparations with Abs to FOXO and δEF1 revealed no binding to the carbonic anhydrase I promoter, even after prolonged amplification (Fig. 6⇑D). Together, these findings establish that FOXO and δEF1 can bind to isolated consensus elements of the Ccng2 and Rbl2 promoters in vitro, and are bound to the endogenous promoters in resting B cells.

FOXO-dependent promoter induction and cooperation with δEF1 do not require binding to IRS elements

To determine whether these FOXO binding sites are essential for activation of the Ccng2 promoter, we made a series of truncation mutants and also used site-directed mutagenesis to specifically mutate the consensus FOXO binding sites. Mutation of individual IRS elements did not diminish promoter activation by FOXO cotransfection (data not shown). FOXO1 transfection also augmented promoter activity of a 770-bp fragment (Del5) in which only one IRS element is preserved (Fig. 7⇓A). Furthermore, mutation of this single IRS sequence within Del5 did not abolish FOXO1-dependent activation. For each mutant construct, synergy with δEF1 was at least partially maintained. Similar results were obtained using FOXO3a ± δEF1 (data not shown). This raised the possibility that binding to IRS elements was not essential to the function of FOXO transcription factors in this system.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Promoter activation by FOXO proteins does not require binding to consensus IRS elements. A, Ccng2 promoter without consensus IRS sites can still be activated by FOXO1 and δEF1. Reporter assays in A20 cells were done using the full-length promoter, a deletion retaining only the FOXO-c site (Del5), or the Del5 mutation in which the FOXO-c site has been mutated (Del5F) as shown above the graph. F = FOXO binding site; E = δEF1 binding site. The full-length and mutant promoters had comparable basal luciferase values. B, EMSA assays were conducted using Cos7 extracts transfected by pcDNA3 empty vector (EV), FOXO1 wild type, the FOXO1(H215R) mutant that cannot bind IRS elements, and FOXO1.A3. C, FOXO1(H215R) mutant can activate the Ccng2 and Rbl2 promoter in a similar pattern as wild-type FOXO1. A20 cells were cotransfected by Ccng2 or Rbl2 full-length promoter reporter with wild-type FOXO1 or FOXO1(H215R) and/or δEF1 as indicated.

To test this further, we used a mutant of FOXO1 (H215R) that eliminates binding to classical IRS elements (19). The mutation affects a critical contact residue between the winged-helix domain and the IRS sequence. As expected, extracts of Cos7 cells transfected with this mutant failed to form IRS-binding complexes, as assessed by EMSA (Fig. 7⇑B). In reporter assays in A20 cells, the FOXO1(H215R) mutant was as potent or more than wild-type FOXO1 (Fig. 7⇑C). Similar results were observed in M12 B lymphoma cells (data not shown). These results support the model that FOXO proteins can cooperate with δEF1 to activate transcription in a manner that does not require direct interaction of FOXO proteins with IRS elements. However, the H215R mutant only affects one binding surface of the transcription factor, and, therefore, the mutant protein might still bind to DNA via other interactions at nonconsensus elements (19).

δEF1 augments expression of endogenous Ccng2 and Rbl2 and enhances FOXO-dependent cell cycle arrest

To assess the effect of FOXO and δEF1 on endogenous gene expression, we isolated total RNA from A20 cells transfected with these factors. We used a regulatable FOXO variant (FOXO3a.A3-ER) that is expressed in an inactive form until the addition of the estrogen analog 4-hydroxytamoxifen (4-OHT) (7). RNA was prepared 64 h after transfection, with some samples receiving 4-OHT during the last 16 h. Northern blot analysis indicated that either FOXO3a.A3-ER or δEF1 alone augmented expression of endogenous cyclin G₂ mRNA by 2- to 3-fold (Fig. 8⇓, A and B). p130/Rb2 mRNA could not be detected by Northern blot under any condition (data not shown), so we used a more sensitive real-time quantitative PCR assay. The results showed that FOXO3a.A3-ER or δEF1 alone enhanced levels of endogenous p130/Rb2 mRNA, similar to the effect on endogenous cyclin G₂ (Fig. 8⇓, C and D). Induction of endogenous cyclin G₂ has been observed in FOXO-transfected fibroblasts (21), and endogenous p130/Rb2 mRNA is induced by FOXO expression in a colon carcinoma cell line (11), but the effects of δEF1 on these genes have not been reported. Cotransfection of A20 cells with FOXO3a.A3-ER and δEF1 resulted in a greater induction of cyclin G₂ and p130/Rb2 mRNA than that observed with either factor alone (Fig. 8⇓, A–D).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8.

A–D, FOXO3a and δEF1 up-regulate endogenous cyclin G₂ and p130 mRNA in B cells. A20 cells were transfected with pMIT-FOXO3a.A3-ER and either empty vector pMIT (lanes 1 and 2) or pMIT-δEF1 (lanes 3 and 4) as indicated. Forty-eight hours later, Thy1.1-positive cells were purified by MACS column and cultured for an additional 16 h in the absence or presence of 4-OHT as indicated. Total RNA was extracted for Northern blot analysis with radiolabeled cyclin G₂ and GADPH cDNA probes. Filters were exposed to phosphor screen (A) and results were quantified (B) using ImageQuant software (Amersham). Quantitation of p130 (C) and cyclin G₂ (D) real-time PCR data from analysis of RNA extracted from A20 cells transfected as above. Expression relative to β-actin is shown. Numbers on y-axes represent fold increase. The results are representative of two experiments. E, δEF1 augments FOXO-mediated cell cycle arrest in A20 cells. Constructs for FOXO1.A3 and δEF1 were transfected into A20 cells by electroporation, together or with empty vector (EV) controls. Forty-eight hours after transfection, cell cycle analysis was done by FACS, gating on live cells expressing the IRES-linked Thy1.1 marker gene. The results are representative of four experiments using FOXO1.A3 and three using FOXO3a.A3. Compared with cells transfected with empty vectors, the average (mean ± SD; n = 4) percentage of increases in G₀-G₁ phase of cells expressing FOXO1.A3 and FOXO1.A3+δEF1 were 15 ± 3.5 and 20.5 ± 1.9, respectively (p = 0.022 for FOXO1.A3+δEF1 vs FOXO1.A3 alone). The average (mean ± SD; n = 3) percentage of increases in G₀-G₁ phase of cells expressing FOXO3a.A3 and FOXO3a.A3+δEF1 were 7 ± 1 and 12.3 ± 1.5, respectively (p = 0.057 for FOXO3a.A3+δEF1 vs FOXO3a.A3 alone).

To further evaluate the functional significance of the FOXO-δEF1 cooperation in B cells, we assessed whether cell cycle arrest by FOXO expression could be enhanced by δEF1. In A20 cells, transfection of FOXO1.A3 in retroviral vectors induced an increase in the percentage of cells in G₀-G₁ phase (Fig. 8⇑E) similar to the effect observed in retrovirally transduced B cell blasts (Fig. 1⇑). Transfection of δEF1 alone did not affect cell cycle in A20 cells, but consistently enhanced the effects of FOXO1.A3 cotransfection (Fig. 8⇑E). In these experiments, empty vectors were cotransfected in the appropriate samples to ensure equivalent amounts and similar composition of exogenous DNA, and to allow flow cytometric gating on transfected cells for cell cycle analysis. The increased G₀-G₁ phase caused by the combination of FOXO1.A3 + δEF1 compared with FOXO1.A3 alone was statistically significant over multiple experiments (see legend to Fig. 8⇑E). Transfection of FOXO3a.A3 ± δEF1 produced a similar trend of increased G₀-G₁ phase, although the effects were smaller (Fig. 8⇑E, legend).