The Transcription of FOXO Genes Is Stimulated by FOXO3 and Repressed by Growth Factors*

FOXO (Forkhead box O) transcription factors induce cell growth arrest and apoptosis, which can be prevented by FOXO phosphorylation by AKT in response to growth factors such as platelet-derived growth factors (PDGF) and insulin-like growth factor I (IGF-I). In addition to this well characterized post-translational modification, we showed that FOXO1, FOXO3, and FOXO4 were also regulated at the transcriptional level. PDGF, fibroblast growth factors (FGF), and IGF-I repressed the expression of FOXO genes in human fibroblasts. This process was sensitive to phosphatidylinositol 3-kinase inhibition by LY294002. FOXO1-specific shRNA decreased FOXO1 mRNA expression and enhanced fibroblast proliferation, mimicking the effects of growth factors. Conversely, ectopic FOXO3 activation blocked the proliferation of fibroblasts and induced the expression of FOXO1, FOXO4, and p27-KIP1. Using luciferase reporter assays and chromatin immunoprecipitations, we identified a conserved FOXO-binding site in the promoter of the FOXO1 gene, which was required for regulation by PDGF, and mediated the up-regulation of FOXO1 by itself and by FOXO3. Altogether, our results suggest that the expression of FOXO1 and FOXO4 genes is stimulated by FOXO3 and possibly by other FOXO factors in a positive feedback loop, which is disrupted by growth factors.

The activity of FOXO proteins is tightly controlled by multiple post-translational modifications (3,4). Growth factors, insulin, and other cell stimuli induce FOXO phosphorylation and inactivation by AKT (also called protein kinase B), a serine/ threonine kinase that is activated via the phosphatidylinositol (PI) 3 3-kinase pathway (1,3). All FOXO proteins are substrates of AKT, which phosphorylates three conserved sites, resulting in the exclusion of FOXOs from the nucleus and in their subsequent ubiquitination and degradation. Phosphorylation by AKT may also regulate FOXO ability to bind to DNA (3). In addition, the mitogen-activated protein kinases ERK and p38, as well as serum-and glucocorticoid-inducible kinase, DIRK1, and IKK␤, also inactivate FOXO1 and/or FOXO3 by direct phosphorylation (1,5,6). By contrast, phosphorylation by c-Jun N-terminal kinase (JNK) kinases upon cell stress activates FOXO4 (7).
In the absence of growth factors, FOXOs reside in the nucleus and up-regulate genes that inhibit the cell cycle (p27 KIP1 and p21 WAF1), promote apoptosis (Fas ligand, Bim, and TRAIL), and decrease oxidative stress (superoxide dismutase and catalase) (1). A number of genes are also repressed by activated FOXOs, including cyclin D1 and D2. The different members of the FOXO family share a common DNA-binding site and regulate overlapping sets of target genes. FOXO transcriptional activity is further regulated by acetylation in the nucleus (3).
In line with their ability to block cell growth and to induce apoptosis, FOXO genes act as tumor suppressors. Deletion of all FOXO1, FOXO3, and FOXO4 alleles in adult mice induces a cancer prone condition, characterized by hemangiomas and thymic lymphomas (8). In human cancers, several chromosomal translocations disrupt FOXO genes, producing hybrid proteins in which the forkhead DNA-binding domain and the AKT phosphorylation sites are lost (9). Pax3-FOXO1 and Pax6-FOXO1 were described in alveolar rhabdomyosarcomas, and the fusion of MLL with FOXO3 or FOXO4 was found in acute myeloid leukemias. In many tumor types, the constitutive activation of PI 3-kinase, AKT, and ERK is expected to inhibit FOXO proteins (9).
Not much is known about the regulation of FOXO mRNA expression. The transcription factor E2F1 was shown to induce FOXO1 and FOXO3 expression (10), and FoxC1 up-regulates FOXO1 in the eye (11). In the present report, we observed that the RNA expression of FOXO1, FOXO3, and FOXO4 was repressed by growth factors such as platelet-derived growth factors (PDGF) and fibroblast growth factors (FGF), which induce cell proliferation. These growth factors play a key role in the development of the embryo and in various human diseases, including cancer (12)(13)(14). PDGFs are dimeric ligands that bind to specific receptor tyrosine kinases, PDGF receptors ␣ and ␤, forming homo-or heterodimers (15). They activate multiple signaling effectors, including PI 3-kinase, AKT, ERK, and phospholipase C␥ (15)(16)(17)(18). These signal transduction pathways regulate the activity of numerous transcription factors, controlling the expression of more than one hundred target genes (19 -22). PDGF was reported to induce the phosphorylation and inactivation of FOXO1 protein in hepatic stellate cells, in fibroblasts, and in vascular smooth muscle cells (23)(24)(25)(26)(27). Here, we suggest that the inactivation of FOXO3 protein by PDGF disrupts a positive feedback mechanism, whereby FOXO3 stimulates the expression of FOXO genes.
Thymidine incorporation assays were performed as described with minor modifications (28). Fibroblasts were seeded in triplicates in a 96-well plate in Quantum 333 (4000 cells/well). After 1 day, the cells were washed and starved in serum-free minimum Eagle's medium (Sigma) for 24 h. Growth factors (25 ng/ml) were then added with [ 3 H]thymidine (0.5 Ci/well; GE Healthcare) for 24 h. Microtiter plates were harvested after a brief incubation with trypsin using a cell harvester (PerkinElmer Life Sciences). The incorporation of [ 3 H]thymidine was quantified using a TopCount instrument (PerkinElmer Life Sciences).
Gene Expression Analysis-Subconfluent AG01518 fibroblasts were incubated for 48 h in minimum Eagle's medium with bovine serum albumin (0.05%). PDGF-BB (25 ng/ml) or basic FGF (10 ng/ml, heparin 50 g/ml) were added 1 or 24 h before harvest. RNA extraction was performed using the RNeasy kit (Qiagen). The RNA quality was first tested in agarose gel. As a further control, the expression of known target genes NR4A1 and SCD was checked by quantitative reverse transcription-PCR (19). Human Affymetrix chips HGU133A2 were hybridized following the manufacturer's instructions.
The results were analyzed using GCOS suite software and normalized using the control condition as a base line to calculate ratios (29). In the MAS5 algorithm, the scale factor using all probe sets was set to 100, and the normalize factor was set to 1. The data were submitted to GEO (accession number GSE14256). We performed two biological replicates. Probe sets that were flagged "absent" across all conditions in the two experiments were discarded. Quantitative PCR analysis was performed as described (19,20,29) using ABgene SYBR greenbased kits (Thermo Scientific), and the oligonucleotides are shown in supplemental Table S1.
Lentivirus-mediated shRNA Silencing-A set of five shRNAspecific clones for human FOXO1 was obtained from Sigma. All of them were packaged for viral production and infection and tested for target knockdown. Two constructs (TRCN0000039578 and TRCN0000039581) were used for further studies. A negative "scrambled" control was obtained from Addgene (catalog number 1864).
The shRNA-expressing and control lentiviral plasmids were transfected with the packaging plasmid pCMV-dr8.2 dvpr and the envelop plasmid pCMV-VSV-G (Addgene catalog numbers 8455 and 8454) (30) into HEK293T cells using FuGENE 6 (Roche Applied Science). Virus-containing supernatants were collected at 36 and 60 h after transfection and used to infect fibroblasts in the presence of 8 g/ml polybrene (Sigma). Then the cells were incubated with standard growth medium containing 2 g/ml puromycin for 3 days prior to cell analysis.
Immunofluorescence Experiments-NIH3T3 fibroblasts (200,000/well) were grown on glass coverslips for 24 h in a 6-well plate. Then cells were co-transfected using Lipofectamine 2000 (Invitrogen) with pEGFP-N1 and pCMV6--FOXO1 (Origene) or pCMV5-FOXO3 (Addgene catalog number 14937) (31) or empty vector. After 24 h, the cells were washed with PBS and starved in Dulbecco's modified Eagle's medium in the presence of bovine serum albumin (0.05%) for 48 h. The cells were stimulated with PDGF-BB (25 ng/ml) for 1 h, washed with cold PBS, and fixed in paraformaldehyde (4% in PBS) for 15 min. The cells were washed and incubated for 1 h at room temperature in PBS containing 5% fetal calf serum and 0.5% saponin. The cells were incubated overnight at 4°C with anti-FOXO1 or anti-FOXO3 antibodies in the presence of 4Ј,6diamidino-2-phenylindole. The cells were then washed and incubated with secondary antibodies coupled to Alexa-Fluor (Molecular Probes). After 2 h, the cells were washed, mounted on slides, and visualized using a fluorescent microscope. Transfected cells were identified by GFP fluorescence.
Luciferase Assays-The FOXO1 promoter was cloned by PCR from the BAC clone RP11-181D10 (BACPAC) as template (see supplemental Table S1 for primer sequences). The promoter was sequenced and subcloned into pGL3-Luciferase, using SmaI and KpnI restriction sites. The 588-bp mutant resulted from digestion with EcoRI (within the FOXO1 promoter) and KpnI (within the plasmid). The 200-bp mutant was generated similarly by digestion with ApaI and KpnI. The 132and 79-bp mutants were amplified by PCR and subcloned into pGL3 (KpnI and HindIII sites). The F mutant was obtained by QuikChange mutagenesis (Stratagene). All of the constructs were sequenced.
The plasmids were linearized by BamHI and KpnI digestion before transfection. Subconfluent mouse NIH3T3 fibroblasts were transfected using Lipofectamine 2000 with 1 g of pGL3 construct and 1.5 g of pEF-␤-galactosidase/well (6-well plates) in the presence of 10% calf serum. After 24 h of incubation, the cells were washed with PBS and starved in medium with 0.05% bovine serum albumin for 24 h and then stimulated with PDGF-BB (25 ng/ml) for 24 h.
Chromatin Immunoprecipitation Assay-Chromatin crosslinking was performed with 1% formaldehyde and was stopped by the addition of glycine to a final concentration of 125 mM. The cells were harvested with cold PBS buffer containing protease inhibitors. Chromatin was fragmented by enzymatic digestion (enzymatic shearing kit; Active Motif). An aliquot was used to check fragmentation on an agarose gel and to quantify the amount of DNA. Digested chromatin (50 g) was first precleared for 1 h with Ultralink protein A/G beads (Pierce). After centrifugation, the supernatants were incubated overnight at 4°C with anti-FOXO1 (H-128; Santa Cruz Biotechnology) or anti-FOXO3 (Cell Signaling Technologies) or anti-STAT5 (Santa Cruz Biotechnology), used as a control (3 g). The immunoprecipitated complexes were bound to protein A/G beads for 3 h at 4°C and washed successively in a radioimmunoprecipitation buffer (10 mM Tris, 1 mM EDTA, 1% Triton X-100, 0.1% SDS, 140 mM NaCl, 0.1% sodium deoxycholate, and 1 mM phenylmethylsulfonyl fluoride, pH 7.4), LiCl buffer (1 mM Tris, 250 mM LiCl, 0.5% Nonidet P40, 0.5% sodium deoxycholate, and 1 mM EDTA, pH 7.4) and Tris-EDTA buffer (1 mM EDTA and 10 mM Tris, pH 7.4). The proteins were then eliminated using proteinase K in the presence of 10% (w/v) SDS overnight at 37°C. After phenol extraction, DNA was precipitated, dissolved in water, and used as template for PCR with the primers shown in supplemental Table S1. PCR products were visualized on 3% agarose gels. Alternatively, quantitative PCR was performed using the same primers in the presence of SYBER-green (ABgene kit) on an iCycler instrument (Bio-Rad). Enrichment of specific promoter regions after immunoprecipitation was expressed as the percentage of the total input chromatin.
Statistical Analysis-Unless otherwise stated, one representative experiment of at least three replicates is shown with standard deviations. Statistical significance was calculated according to Student t test (*, p Ͻ 0.05; **, p Ͻ 0.01).

RESULTS
Growth Factors Repress the Expression of FOXO1, FOXO3, and FOXO4 Genes-In a microarray experiment using Affymetrix Genechips, we analyzed the regulation of gene expression by PDGF and FGF in AG01518 normal human fibroblasts. The detailed microarray results will be described elsewhere. We noticed that the expression of the genes FOXO1, FOXO3, and FOXO4 was decreased 24 h after growth factor stimulation. To confirm this observation, we measured the RNA levels of FOXO genes by quantitative PCR. Fig. 1A shows that the expression of FOXO1 is repressed 3-fold after stimulation of AG01518 fibroblasts with PDGF-BB, FGF-2, or FGF-4. FOXO4 expression was also strongly repressed, whereas FOXO3 was regulated to a lesser extent by PDGF. FOXO6, which was described as a brain-specific isoform, was not expressed in these cells (data not shown). Similar results were obtained in BJ dermal fibroblasts immortalized with telomerase (BJ-hTert; Fig. 1A, right panels). Insulin-like growth factor I also repressed FOXO1 expression in a reproducible manner. In time course experiments, the expression of FOXO1 and FOXO4 was decreased 6 h after the addition of PDGF (Fig. 1B). Shorter treatment periods did not significantly lower FOXO FIGURE 1. FOXO expression is down-regulated by growth factors in human fibroblasts. A, AG01518 human fibroblasts (left panels) were starved for 24 h in serum-free medium and then stimulated for 24 h with the indicated growth factor before RNA extraction. The expression of FOXO genes was monitored by quantitative PCR. The same experiment was performed with BJ-hTert cells stimulated for 16 h (right panels). B, starved AG01518 cells were stimulated with PDGF-BB for the indicated periods of time and analyzed as above. C, starved AG01518 cells were treated with PDGF-BB or control medium for 24 h. The cells were lysed in the presence of 1% Triton X-100. FOXO1 was immunoprecipitated (IP) and analyzed by Western blot with specific anti-FOXO1 antibodies as described (19).
expression (data not shown). Next, the level of FOXO1 protein was visualized by Western blot in AG01518 cells stimulated for 24 h with PDGF-BB (Fig. 1C). The result confirmed the downregulation of FOXO1 in PDGF-treated cells.
PDGF Inactivates FOXO-The activity of FOXO proteins is inhibited by phosphorylation upon growth factor stimulation. We confirmed the phosphorylation of FOXO1, FOXO3, and FOXO4 in AG01518 fibroblasts stimulated with PDGF by Western blot using antibodies directed against phosphoserine 256 of FOXO1 (which cross-react with FOXO4) or antibodies against serine 253 of FOXO3 ( Fig. 2A). This conserved site is phosphorylated by AKT (2, 3), which was also activated in these cells upon PDGF treatment, as judged from blots with the antiphosphoserine 473 antibodies.
It is well established that FOXO phosphorylation induces its translocation from the nucleus to the cytosol. Interestingly, this was reported in vascular smooth muscle cells treated by PDGF (24). To confirm that PDGF affects FOXO localization in fibroblasts, we transfected NIH3T3 cells with FOXO1 or FOXO3 and studied the localization of the proteins by immunofluorescence. Fig. 2B shows that FOXO1 and FOXO3 were excluded from the nucleus upon treatment with PDGF for 1 h.
To evaluate FOXO transcriptional activity in cells treated by growth factors, we analyzed the regulation of known FOXO target genes by PDGF-BB and FGF-2 in AG01518 fibroblasts using our microarray data (supplementary Fig. S1). Based on published work, we identified 58 genes that were reported to be up-regulated by at least one FOXO isoform and were present in our initial microarray data, as well as 36 down-regulated genes. Most of the genes known to be induced by FOXO were repressed by stimulation of fibroblasts with PDGF and FGF-2 for 24 h. Well characterized FOXO target genes, such as p27/ CDKN1B, SOD2, CITED2, GADD45A, CASP8, and BAX, were all repressed by growth factors. Conversely, the majority of the genes known to be repressed by FOXO were induced by growth factors, including cyclin D (CCND1 and CCND2) and BIRC5. To exclude the possibility that this result was obtained by chance, we performed Fisher's statistical test, which gave a p value of 1.6 ϫ 10 Ϫ4 . No significant effect of growth factors was observed after 1 h of treatment (supplementary Fig. S1). Altogether, these results suggested that the expression of FOXO targets was affected by PDGF and FGF-2 in a manner consistent with FOXO inhibition.
FOXOs Play a Key Role in the Proliferation of Fibroblasts-It is well established that FOXO factors induce cell growth arrest and apoptosis (33). To test whether this also applies to human fibroblasts, we used a FOXO3-estrogen receptor fusion protein (FOXO3-A3-ER), which can be activated by 4-hydroxy-tamoxifen (32,34). In this construct, the three AKT phosphorylation sites of FOXO3 are mutated to alanine, preventing the inactivation of the protein by growth factors. FOXO3-A3-ER was expressed in AG01518 using a retroviral vector. The number of viable cells was estimated by measuring ATP levels or by counting cells in the presence of trypan blue (Fig. 3). Activation of FOXO3-A3-ER by 4-hydroxy-tamoxifen blunted the effect of PDGF and FGF-2 on fibroblast proliferation.
Because FOXO proteins are tightly regulated by several posttranslational mechanisms, it was not clear whether a change in FOXO RNA level affects cell growth. To tackle this question, we used the pLKO lentiviral vector, which is designed to encode short inhibitory RNA hairpins (shRNA). Two specific FOXO1 shRNA reproducibly reduced FOXO1 mRNA expression measured by quantitative PCR, whereas a control sequence had no effect (Fig. 4A). When fibroblasts were infected with both constructs together, FOXO1 expression was repressed by 63%. Remarkably, this is comparable with the effect of PDGF on FOXO1 expression. Transduction of these two FOXO1 shRNA alone or in combination increased the basal proliferation of AG01518 fibroblasts, as demonstrated by the thymidine incorporation experiment shown in Fig. 4B. PDGF stimulated fibroblast proliferation more than 2-fold in the presence of control shRNA but did not significantly increase cell growth in the presence of FOXO1 shRNA. In conclusion, these data suggested that decreasing FOXO1 expression is enough to enhance fibroblast cell growth.
The Regulation of FOXO1 by PDGF Is Sensitive to PI 3-Kinase Inhibition-To decipher the mechanism of regulation of FOXO expression by PDGF, we first tested whether de novo protein synthesis was required. As shown in Fig. 5A, PDGF was able to decrease FOXO1 expression in the presence of cycloheximide, a potent protein synthesis inhibitor, suggesting the existence of a direct regulatory mechanism.
We next sought to identify which signal transduction pathway was required in this process, using small inhibitory molecules. The PI 3-kinase inhibitor LY294002 abolished the regulation of FOXO1 by PDGF in AG01518 cells, whereas the mitogen-activated protein kinase pathway inhibitor U0126 had no effect (Fig. 5B). Similar results were obtained in BJ-hTert cells (Fig. 5C). LY294002 also blocked the down-regulation of FOXO4. We had observed previously that LY294002 prevents the activation of AKT by PDGF in AG01518, as expected (19). 4 Because PI 3-kinase is also required for FOXO phosphorylation and inactivation by PDGF (23,27), our data raised the possibility that FOXO proteins stimulate the expression of their own genes, a process that would be disrupted upon AKT activation by growth factors. FOXO1 Transcription Is Regulated by FOXO3-To test this hypothesis, we first analyzed the expression of FOXO genes in AG01518 fibroblasts transduced with FOXO3-A3-ER. The activation of FOXO3-A3-ER by 4-hydroxy-tamoxifen induced the expression of FOXO1 and FOXO4, as shown by quantitative PCR (Fig. 6). The p27-KIP1 gene, a well characterized target of FOXO3, was used as a positive control (32). As shown in Fig. 6, it was regulated to a similar extent as FOXO1 and FOXO4. By contrast, we did not observe any regulation of endogenous FOXO3, which was amplified with probes hybridizing with the 3Ј-untranslated region of FOXO3, absent in the FOXO3-A3-ER construct. The expression of FOXO3-A3-ER was controlled using FOXO3 oligonucleotides that were able to amplify the retroviral construct. Collectively, these data indicated that FOXO3 activation can induce the expression of FOXO1 and FOXO4.
We next analyzed the FOXO1 promoter. We isolated a 1238base pair fragment of the human FOXO1 gene promoter (Fig.  7A), which presents a relatively high level of homology with the mouse sequence. This fragment was cloned in front of a luciferase reporter gene and co-transfected with FOXO3-A3-ER in HEK293T cells. In the presence of 4-hydroxy-tamoxifen, FOXO3-A3-ER enhanced FOXO1 promoter activity (Fig. 7B). We next performed serial 5Ј deletions of the promoter (Fig. 7A) and analyzed the effect of FOXO3 activation. Deletion of 650 nucleotides did not affect the promoter activity or its response to FOXO3 activation (Fig. 7B). By contrast, FOXO3 activation had a much reduced impact on a 200-base pair promoter and did not stimulate the activity of promoters retaining only 132 or 79 base pairs. The in silico analysis of the FOXO1 promoter revealed a putative FOXO-binding site located 370 nucleotides  before the first exon. This sequence is conserved in the mouse and rat genome (see supplemental Fig. S2). Based on the consensus FOXO-binding site, we mutated two key adenines into cytosines (Fig. 7A). A full-length 1238-bp promoter harboring this mutation responded only weakly -but significantly (p Ͻ 0.01) -to FOXO3-A3-ER activation, compared with wild type (Fig. 7B).
To test whether FOXO1 could regulate the expression of its own gene, we transfected HEK293T cells with wild-type FOXO1 and compared it with wild-type FOXO3, in a luciferase assay (Fig. 7C). Transfection of FOXO3 strongly stimulated FOXO1 promoter activity, confirming the experiments performed with FOXO3-A3-ER. FOXO1 stimulated the promoter of its own gene significantly but to a lesser extent compared with FOXO3. Mutation of the FOXO-binding site of the FOXO1 promoter abolished the effect of FOXO1 and decreased the effect of FOXO3.
To assess whether FOXO1 and FOXO3 were binding directly to the FOXO1 promoter, we performed chromatin immunoprecipitation experiments with anti-FOXO antibodies. A 108-bp DNA fragment that included the FOXO consensus site of the promoter was specifically associated with FOXO1 in HEK293T cells (Fig. 7D). This effect was increased by FOXO1 transfection. FOXO3 was also bound to the same promoter fragment. Like in luciferase experiments described above, the highest signal was observed in cells transfected with FOXO3-A3-ER and stimulated with 4-hydroxy-tamoxifen. In conclusion, our data suggest that FOXO1 expression is up-regulated upon binding of FOXO1 and FOXO3 proteins to the FOXO1 gene promoter. Nevertheless, a second regulatory mechanism may contribute to FOXO1 regulation independently of the consensus FOXObinding site that we identified, as suggested by the effect of FOXO3 on the mutant promoter.
PDGF Blunts the Activation of the FOXO1 Promoter by FOXO Factors-To investigate whether this mechanism could account for the regulation of FOXO1 by PDGF, luciferase reporter assays were performed in transfected NIH3T3 fibroblasts. We observed that PDGF significantly repressed the transcriptional activity of the FOXO1 promoter (Fig. 8A). The activity of the mutant FOXO1 promoter, which lacks the consensus FOXO-binding site, was not stimulated by PDGF. Next, we tested whether PDGF affected FOXO recruitment to the FOXO1 promoter by chromatin immunoprecipitation. We observed that FOXO1 and FOXO3 were associated to the FOXO1 promoter in AG01518 fibroblasts (Fig. 8B), in agreement with the results that we obtained in HEK293T cells. Fibroblast stimulation with PDGF decreased the binding of FOXO1 and FOXO3 significantly (Fig. 8B). Altogether, our data suggested that PDGF dampen the activation of FOXO1 promoter by FOXO proteins. As mentioned above, the inactivation of FOXO proteins by growth factors has been ascribed to the phosphorylation of FOXOs and their subsequent translocation from the nucleus to the cytosol (1,35). This well established pathway likely accounts for the inhibition of FOXO proteins by PDGF in the present study (Fig. 9). In addition, Aoki et al. (25) showed that PDGF induces the degradation of FOXO1 protein by proteasomes. We observed that treating fibroblasts with the proteasome inhibitor MG132 blocked the repression of FOXO1 mRNA expression by PDGF (supplementary Fig. S3), suggesting that the degradation of FOXO by proteasomes may play a role in our model. Altogether, these observations are in accordance with the model proposed by Matsuzaki et al. (36), who showed that both cytosolic retention and degradation of FOXO1 are required for its efficient inactivation by insulin. We cannot rule out the involvement of additional layers of regulation, at the level of FOXO DNA binding or transcriptional activity, for instance (35).

DISCUSSION
Our data demonstrate that FOXO1 expression is stimulated by activated FOXO3, based on the following observations: (i) FOXO3-A3-ER activation leads to increased FOXO1 expression and increased FOXO1 promoter activity, (ii) wildtype FOXO3 overexpression also increases FOXO1 promoter activity, (iii) FOXO3 binds to at least one consensus site in FOXO1 promoter, and (iv) inactivation of FOXO3 by phosphorylation in the presence of PDGF correlates with FOXO1 down-regulation. Likewise, FOXO3 may also regulate the expression of FOXO4, because its expression is induced by activated FOXO3-A3-ER and decreased by PDGF. However, no conserved consensus FOXO-binding site could be identified in the promoter of FOXO4. In addition, luciferase experiments and chromatin immunoprecipitations suggested that FOXO1 induces the expression of its own gene. Because all FOXO transcription factors bind to the same promoter elements and share a number of target genes, it is possible that they all induce the expression of FOXO1 and FOXO4, at least to some extent, creating a positive feedback network controlling FOXO gene expression. Such positive feedback regulation has been described for other transcription factors such as SREBP (37). Understanding the regulation of the FOXO3 gene by growth factors such as FGF requires further studies.
In accordance to the decrease in FOXO1 mRNA, we also observed a decrease in FOXO1 protein expression in cells treated with PDGF for 24 h. However, FOXO1 degradation by proteasomes may also contribute the decrease in FOXO1 protein. Because the transcriptional repression of FOXO1 gene also depends on protein degradation (supplemental Fig.  S3), it was not possible to discriminate between these two possibilities.
In line with previous reports on other cell types (25,33,38), we show that activation of FOXO3 blocked the proliferation of human fibroblasts in response to growth factors. Conversely, reducing the expression of the FOXO1 transcript with specific shRNA enhanced fibroblast proliferation. This latter experiment indicated that the level of FOXO1 mRNA expression is an important determinant of cell proliferation, in addition to the well characterized post-translational modifications of FOXO proteins by AKT and other signaling kinases.
The regulation of FOXO mRNA expression may sustain or amplify the anti-proliferative and anti-apoptotic effects that are initiated by the direct and rapid control of FOXO proteins by phosphorylation in the presence of growth factors. Conversely, up-regulation of FOXO genes by activated FOXO3 may also help to restore FOXO expression level upon growth factor withdrawal or stress.
In a microarray study, Zhu et al. (39) suggested that FOXO1 expression was down-regulated in mouse B lymphocytes by several mitogens, such as interleukin-4, lipopolysaccharide, anti-IgM, and anti-CD40. Decreased mRNA expression of FOXO1, FOXO3, and FOXO4 has also been reported in B cells upon activation of the B-cell receptor, which also induced FOXO1 protein phosphorylation (40). The decrease in FOXO1, FOXO3, and FOXO4 expression in these cells was sensitive to PI 3-kinase inhibition, like in our study. However, FOXO1 was not down-regulated in cells deficient in Bruton's kinase, a key mediator of BCR signaling, although it could still be phosphorylated on serine 256, suggesting that the two processes were uncoupled. Thus, a different mechanism may account for FOXO down-regulation in B lymphocytes.
Because the constitutive activation of growth factor receptors and the PI 3-kinase pathway is a frequent event in cancer cells, our data predict that FOXO gene expression may be decreased in tumors, which may contribute to uncontrolled cell growth. Interestingly, a decreased expression of FOXO3 correlates with prostate cancer cell progression toward androgen independence (41). Future studies will have to determine whether the mRNA expression of FOXO genes is reduced in tumor cells. In conclusion, we identified a new mechanism of regulation of FOXOs at the transcriptional level, which modulates fibroblast proliferation. FIGURE 8. PDGF inhibits FOXO binding to the FOXO1 promoter. A, NIH3T3 mouse fibroblasts were transfected with the indicated FOXO1 promoter construct and stimulated with PDGF-BB for 24 h. Luciferase activities were measured as described under "Experimental Procedures." One representative experiment is shown with S.D. B, AG01518 fibroblasts were starved for 48 h and stimulated for 24 h in the presence of PDGF (black bars) or control medium (white bars). Chromatin immunoprecipitations (IP) were performed as described for Fig. 7. The average values of three independent experiments, expressed as percentages of the input DNA, are shown with S.E. n.t., not tested.