Report

DNA Repair Pathway Stimulated by the Forkhead Transcription Factor FOXO3a Through the Gadd45 Protein

See allHide authors and affiliations

Science  19 Apr 2002:
Vol. 296, Issue 5567, pp. 530-534
DOI: 10.1126/science.1068712

Abstract

The signaling pathway from phosphoinositide 3-kinase to the protein kinase Akt controls organismal life-span in invertebrates and cell survival and proliferation in mammals by inhibiting the activity of members of the FOXO family of transcription factors. We show that mammalian FOXO3a also functions at the G2 to M checkpoint in the cell cycle and triggers the repair of damaged DNA. By gene array analysis, FOXO3a was found to modulate the expression of several genes that regulate the cellular response to stress at the G2-M checkpoint. The growth arrest and DNA damage response gene Gadd45a appeared to be a direct target of FOXO3a that mediates part of FOXO3a's effects on DNA repair. These findings indicate that in mammals FOXO3a regulates the resistance of cells to stress by inducing DNA repair and thereby may also affect organismal life-span.

The binding of growth factors to specific receptor tyrosine kinases activates the phosphoinositide 3-kinase (PI3K) and the serine-threonine kinase Akt (also called protein kinase B or PKB). Akt promotes cell survival and proliferation in part by directly phosphorylating and inhibiting members of the FOXO subfamily of forkhead transcription factors (1–3). In the nematode Caenorhabditis elegans, null mutants of the PI3K-Akt pathway lead to the activation of the worm FOXO transcription factor DAF-16, resulting in either an extension of adult life-span or, during development, an entrance into the long-lived larval stage termed dauer (4, 5). In both cases, the PI3K-Akt pathway mutants develop a resistance to stress that may account for the longevity phenotype observed (4, 6). One possibility is that in the mutant background, the activation of FOXO transcription factors may mediate the resistance to stress because DAF-16's activity is required for the transcriptional up-regulation of cytosolic catalase and superoxide dismutase, scavenger proteins that protect against oxidative damage (7, 8). In mammals, the role of the forkhead transcription factors in the response to stress is, as yet, unclear.

One mechanism by which cells protect themselves against oxidative stress is by repairing the damage to their DNA and proteins that occurs upon exposure to environmental stress (9). Furthermore, this capacity to repair DNA damage is closely correlated with an increased longevity in mammals (9). Therefore, we considered the possibility that the transcription factor FOXO3a, a mammalian forkhead family member that is similar in sequence to DAF-16, induces a program of gene expression that allows for the repair of DNA damage caused by oxidative stress. Under conditions of DNA damage, cycling cells arrest at one of two critical cell cycle checkpoints (G1-S or G2-M phase of the cell cycle, before and after DNA replication) allowing the cells time to repair their DNA (10).

Overexpression of FOXO3a induces a potent G1 arrest (3), but FOXO3a's role in G2-M progression has not been explored. Consistent with a possible role for FOXO3a in the G2-M phase of the cell cycle, FOXO3a was localized in the nucleus in cells passing through the G2 phase of the cell cycle [Web fig. 1 (11)]. To assess the role of FOXO3a in G2-M progression, we generated a line of Rat-1 fibroblasts that express a fusion protein consisting of FOXO3a fused at its COOH terminus to the ligand-binding domain of the estrogen receptor (TM-ER) (11). In TM-ER, the FOXO3a moiety was mutated such that the key regulatory sites of phosphorylation, Thr32, Ser253, and Ser315, were converted to alanine so that Akt could no longer phosphorylate these sites. The ER portion of this fusion protein was mutated so that it specifically responded to 4-hydroxytamoxifen (4OHT) but not to estrogen (12), thus rendering TM-ER refractory to activation by endogenous estrogen. In TM-ER, the ER domain functions as a molecular switch that rapidly turns on the activity of the FOXO3a protein when TM-ER expressing cells are exposed to 4OHT. Using antibodies that recognize the TM-ER protein, we found that in the absence of 4OHT, the TM-ER fusion protein was sequestered in the cytoplasm (Fig. 1A) (13). However, 1 hour after treatment of cells with 4OHT (1 μM), the TM-ER protein was found almost exclusively in the nucleus (Fig. 1A). We tested the ability of TM-ER to drive FOXO-dependent transcription in response to 4OHT. Activated TM-ER enhanced transcription of a luciferase reporter gene containing three canonical FOXO binding sites within its promoter [forkhead response element (FHRE) luciferase] (Fig. 1B). 4OHT did not induce the reporter gene in cells that did not express TM-ER (Fig. 1B), indicating that the transactivation of the reporter was not a nonspecific effect of 4OHT treatment. 4OHT also did not induce FHRE luciferase expression in cells that expressed a mutant form of TM-ER that has a deletion in the DNA binding domain (TMΔDB-ER) (11) (Fig. 1B).

Figure 1

Inducible activation of FOXO3a. (A) Rat1 cells expressing the inducible form of FOXO3a (TM-ER) were incubated in the presence (+) or absence (–) of 4OHT (1 μM) for 1 hour. The subcellular localization of TM-ER was monitored by immunofluorescence using the antibody to HA. (B) Control Rat 1 cells (CTL) and Rat 1 cells expressing TM-ER or TMΔDB-ER were transfected with the FHRE-luciferase reporter gene. One day after transfection, cells were incubated in the presence (+) or absence (–) of 1 μM 4OHT for 6 hours, and luciferase assays were performed. The graph represents the mean and SEM of three independent experiments conducted in duplicates.

To examine the role of FOXO3a during progression from G2 to M phase, we exposed Rat-1 cells stably expressing TM-ER to aphidicolin to arrest the cells in early S phase (11). The aphidicolin was then washed out to release the cells from the S phase block while FOXO3a activity was simultaneously induced by the addition of 4OHT. Fluorescence-activated cell sorting (FACS) analyses of the Rat-1 cells stably expressing the TM-ER at 7 hours after stimulation with 4OHT revealed a G2-M delay (Fig. 2A). 4OHT alone did not affect the progression of control Rat-1 fibroblasts through the G2 and M phases of the cell cycle (Fig. 2B). Importantly, FACS analyses of Rat-1 clones stably expressing TMΔDB-ER showed that treatment of cells with 4OHT no longer elicited a G2-M delay, indicating that FOXO3a induces G2-M delay by binding to DNA (Fig. 2B). The results obtained by FACS analyses were corroborated by immunocytochemistry experiments with the use of an antibody that specifically stains cells that are in the G2 and M phases of the cell cycle [Web fig. 2, A and B (11)].

Figure 2

FOXO3a induces a delay in G2 to M progression. (A) Rat1 cells expressing TM-ER or TMΔDB-ER were released from an S phase block at various times in the presence or absence of 4OHT (500 nM). Fixed cells were stained with propidium iodide, and FACS analysis was performed. The percentage of cells in each phase of the cell cycle was calculated with CellQuest software (Becton Dickinson, San Jose, CA). The experiment presented is representative of four independent experiments. (B) Control Rat1 cells or Rat1 cells expressing TM-ER or TMΔDB-ER were released from an S phase block at 9 hours in the presence or absence of 4OHT (500 nM). FACS analysis was performed as described in (A). The data presented corresponds to the mean of two independent experiments. (C) Rat1 cells were released from an S phase block at various times in the presence or absence of LY (25μM). FACS analysis was performed as described in (A). The experiment presented is representative of two independent experiments.

To verify that the G2 delay observed in TM-ER–expressing clones also occurs in cells expressing endogenous FOXO3a, we treated Rat1 cells with the PI3K inhibitor LY294002 (LY), which promotes the translocation of endogenous FOXO3a to the nucleus (1, 14). We found that treatment of cells with LY also caused a delay in the passage of the cells though G2 (Fig. 2C), further suggesting a role for FOXO3a in regulating G2-M progression. It is possible that the delay in the progression through the G2 phase induced by FOXO3a activation may allow time for the repair of damaged DNA before progression into M phase.

To determine if FOXO3a induces the repair of damaged DNA, we assessed FOXO3a's ability to reactivate an ultraviolet (UV)-damaged reporter gene in a host cell reactivation assay (15). We cotransfected a cytomegalovirus (CMV)-luciferase reporter plasmid that had been damaged before transfection by 5000 J/m2 of UV irradiation together with various FOXO3a constructs and an undamaged CMV-renilla luciferase reporter control plasmid (11). The damaged reporter, unless repaired, does not display any transcriptional activity. FOXO3a does not substantially activate the CMV promoter (14). Any small transcriptional effect of FOXO3a on the activation of the repaired CMV promoter is offset by normalization to the cotransfected undamaged CMV-renilla luciferase reporter gene that serves as an internal control. Expression of an active FOXO3a in both the RKO human colorectal cancer cell line and wild-type mouse embryonic fibroblasts (MEFs) promoted reactivation of the cotransfected UV-damaged CMV luciferase reporter (Fig. 3, A and B). The reactivation of the luciferase reporter gene by the constitutively active FOXO3a in both cell lines was significantly greater than the amount of reporter gene activation detected in the absence of FOXO3a (P < 0.001, t test). Moreover, a mutant of FOXO3a that has a deletion in FOXO3a DNA binding domain (TMΔDB) (11) was less effective at inducing the repair of the CMV promoter (P < 0.001 versus TM, t test) (Fig. 3A), indicating that FOXO3a induces DNA repair by binding to DNA, possibly to sites within the promoters of genes involved in the DNA repair process.

Figure 3

FOXO3a-induced repair of damaged DNA. (A) RKO cells were cotransfected (Effectene, Qiagen, Valencia, CA) with 0.4 μg of empty vector or constructs encoding TM or TMΔDB, together with a UV-damaged CMV-luciferase reporter construct (0.5 μg) and an undamaged CMV-renilla luciferase reporter construct (0.1 μg). Twenty-four hours after transfection, luciferase assays were performed. The data presented represent the mean and SEM of five independent experiments conducted in duplicates. Asterisk indicates statistical significance (P < 0.001, t test). (B) WT MEFs were transfected as described in (A). Twenty-four hours after transfection, luciferase assays were performed. The data presented represent the mean and SEM of 10 independent experiments conducted in duplicates. (P < 0.001, t test). (C) RKO cells were cotransfected with empty vector (CTL) or constructs encoding TM or TMΔDB together with UV-damaged pBluescript and an undamaged pGL3-Basic vector. Seventy-two hours after transfection, total DNA was isolated and quantitative PCR was performed. The data presented represent the mean and SEM of six independent experiments conducted in triplicates.

To confirm that FOXO3a induces the repair of damaged DNA, we used a quantitative polymerase chain reaction (PCR)-based assay to detect the presence of damaged DNA templates that had been repaired (11). We cotransfected a DNA template damaged by 5000J/m2 of UV irradiation together with various FOXO3a constructs and an undamaged control DNA template. Repair of the template was assessed by measuring its amplification by PCR and was quantified by real-time fluorescence detection. As with the promoter reactivation experiments, TM but not TMΔDB induced repair of the damaged template (Fig. 3C).

We used high-density cDNA microarrays to identify FOXO3a-regulated genes that might mediate the G2-M delay and DNA repair response. We isolated messenger RNA (mRNA) from TM-ER–expressing Rat 1 cells at 0, 1, 3, and 8 hours after 4OHT addition and hybridized the mRNA to DNA microarrays containing 12,000 cDNAs and expressed sequence tags (ESTs) (16,17). Transcriptional profiles were compared, and a clustering algorithm was applied to identify genes regulated after FOXO3a activation. From this clustering analysis, we identified several genes involved in apoptosis and the regulation of the cell cycle (Table 1), consistent with the previously known functions ascribed to FOXO3a (1,3). FOXO3a also increased expression of a group of genes that takes part in the cellular response to stress. Expression of these genes was not increased in control Rat1 cells treated with 4OHT. Several of the genes expressed in response to activation of TM-ER also showed increased expression when cells were treated with the PI3K inhibitor LY (Table 1). Of these genes, Gadd45aappeared to be a good candidate for mediating the effects of FOXO3a on G2-M progression and DNA repair. All three members of the Gadd45 family, Gadd45a, Gadd45β, and Gadd45γ, are expressed in response to stress stimuli that induce DNA damage (18–20). The GADD45A promoter contains three FOXO transcription factor binding sites (21). Of these sites, one is in a region that is conserved across species, suggesting that its function is critical. Lastly, Gadd45a is important for DNA repair and cell cycle arrest at the G2-M checkpoint (22–24).

Table 1

Identification of FOXO3a target genes by transcriptional profiling.

View this table:

To verify that FOXO3a regulates Gadd45a expression, we isolated mRNA at hour 0, 1 hour, and 3 hours after exposure of TM-ER–expressing cells to 4OHT. By real-time quantitative reverse transcriptase–PCR (RT-PCR), we found that expression of the Gadd45a mRNA was induced upon exposure of TM-ER–expressing cells to 4OHT (Fig. 4A) (11). Western analysis with a Gadd45 antibody revealed that the abundance of Gadd45a protein increased in three independent TM-ER–expressing clones in response to 4OHT treatment (Fig. 4B) (13). In contrast, 4OHT did not increase abundance of Gadd45a mRNA or protein in cells that did not express the TM-ER protein (Fig. 4, A and B), nor did 4OHT cause accumulation of the Gadd45a protein in cells that expressed TMΔDB-ER (Fig. 4B).

Figure 4

FOXO3a-induced expression of Gadd45a mRNA and protein. (A) Rat1 cells either expressing empty vector (solid bars) or TM-ER (open bars) were stimulated with 4OHT (1 μM) for 0, 1, and 3 hours. Real-time quantitative RT-PCR was performed with the use of primers to Gadd45a cDNA. (B) (Left) Rat1 cells expressing TM-ER were stimulated with 4OHT (1 μM) for 0, 1, 3, or 8 hours. Western blot experiments were performed with the antibody to Gadd45a (top), the antibody to HA to detect TM-ER (middle) and an antibody to a control protein (MKK1) to monitor equal loading (bottom). (Center) A clone of Rat1 cells expressing TM-ER and the control Rat1 cells were incubated in the presence (+) or absence (–) of 4OHT (1 μM) for 8 hours. Western blot experiments were conducted as described above. (Right) Clones of Rat 1 cells expressing TM-ER or TMΔDB-ER were incubated in the presence (+) or absence (–) of 4OHT (1 μM) for 8 hours. Western blot experiments were conducted as described above.

To determine if the induction of Gadd45a transcription is mediated by the direct binding of FOXO3a to the promoter of theGadd45a gene, we examined the effects of FOXO3a on the activity of the human GADD45A promoter. An analysis of the human GADD45A promoter revealed the presence of one conserved consensus FHRE, ATAAACAA, at 505 nucleotides 5′ of the transcriptional start site, and two suboptimal FHREs, TAAAACAA and TTGTTTGG, located 377 and 803 nucleotides 5′ of the transcriptional start site, respectively. To assess the ability of FOXO3a to regulateGADD45A transcription, we cotransfected CCL39 fibroblast cells with various FOXO3a constructs together with a reporter gene in which the GADD45A promoter drives the expression of a luciferase gene (13). We found that the constitutively active mutant of FOXO3a (TM) but not a mutant in which the transactivation domain was deleted (TMΔCt) effectively inducedGADD45A promoter driven luciferase activity (Fig. 5A). A mutant of FOXO3a that does not bind DNA (TMΔDB) did not activate GADD45A promoter expression (Fig. 5A).

Figure 5

Expression from the GADD45A promoter induced by FOXO3a through direct binding to two FHREs. (A) CCL39 cells were cotransfected with empty vector (CTL) or constructs encoding TM, TM-ΔCt, or TM-ΔDB together with theGADD45A-luciferase reporter plasmid. Thirty-six hours after transfection, luciferase assays were performed. (B) CCL39 cells were cotransfected with CTL or with a construct encoding TM together with the wild-type GADD45A luciferase reporter construct (P-WT, open bars) or the double mutant GADD45Aluciferase reporter construct (P-2/3m, hatched bars). Twenty-four hours after transfection, luciferase assays were performed. Asterisk indicates statistical difference from P-WT cotransfected with TM (P < 0.05, t-test). (C) CCL39 cells were transfected with either P-WT or P-2/3m. The cells were then either treated (solid bars) or not treated (open bars) with 30 J/m2 of UV irradiation for 6 hours, and luciferase assays were performed (P < 0.005, t test).

GADD45A promoters in which we replaced the core consensus sequence AACA of FHRE-2 within the GADD45Apromoter with GGGG and also replaced the AC within the core consensus sequence of FHRE-3 with GG (P-2/3m) showed diminished activity in response to FOXO3a (P < 0.05, t test) (Fig. 5B). Thus, direct DNA binding of FOXO3a to FHRE-2 and FHRE-3 of theGADD45A promoter appears to be important for the transactivation of the GADD45A promoter.

To determine if FOXO3a promotes UV expression of Gadd45a in cells exposed to UV irradiation, we transfected CCL39 fibroblasts with a luciferase reporter containing either the wild-type GADD45Apromoter or the double mutant GADD45A promoter construct (P-2/3m) and exposed the cells to 30J/m2 of UV irradiation. The UV responsiveness of the GADD45A promoter was diminished when the FOXO3a binding sites were mutated (P <0.005,t test) (Fig. 5C), suggesting that FOXO3a mediates the UV induction of the GADD45A promoter.

We used Gadd45a-deficient fibroblasts to determine whether FOXO3a's ability to induce a G2 delay and DNA repair requires the activity of Gadd45a. We assessed the G2 progression of wild-type (WT) orGadd45a −/− fibroblasts by FACS analysis in the absence or presence of the PI3K inhibitor LY, which activates endogenous FOXO3a. The LY-induced G2 delay was somewhat diminished in the Gadd45a −/− fibroblasts compared with that in WT (Fig. 6A), suggesting that Gadd45a may be partly required for the G2delay induced by the activation of endogenous FOXO3a. We note that in the absence of LY, the Gadd45a −/− fibroblasts exhibit a slowed progression through the cell cycle compared with the WT fibroblasts (Fig. 6A), presumably due to the accumulation of cells in S phase, as had been observed previously (22). The effect of FOXO3a TM on reactivation of the damaged CMV reporter gene was diminished in the Gadd45a −/− MEFs compared with the WT MEFs (P < 0.05, two-samplet test) (Fig. 6B), indicating that Gadd45a mediates, at least in part, FOXO3a-dependent DNA repair.

Figure 6

FOXO3a-induced G2-M delay and DNA repair is mediated by Gadd45a. (A) Transformed WT MEFs or Gadd45a −/− fibroblasts cells were released from an S phase block at hour 0 or 9 hours in the presence or absence of LY (25μM). FACS analysis was performed as described inFig. 2C. The experiment presented is representative of two independent experiments. (B) Gadd45a −/− MEFs (solid bars) and WT MEFs (open bars) were transfected as described in Fig. 5A. Twenty-four hours after transfection, luciferase assays were performed. The data presented represent the mean and SEM of six independent experiments. Asterisk indicates statistical difference from Gadd45a −/− MEF transfected with TM (P < 0.05, two-sample ttest).

These findings define a role for FOXO3a in inducing a delay in the G2-M phase in response to stress stimuli during which FOXO3a contributes to the repair of damaged DNA by a Gadd45a-dependent mechanism. FOXO3a can also induce cell cycle arrest in G1(3) and can enhance the exit from M phase (25), suggesting that FOXO3a has multiple roles in cell cycle control. FOXO3a's ability to induce G1 arrest, apoptosis, and, as we show here, G2-M delay and DNA repair, makes it an attractive candidate as a tumor suppressor in a manner reminiscent of the tumor suppressor protein p53. Loss of FOXO3a function may lead to a decreased ability to induce cell cycle arrest at the critical G1-S and G2-M checkpoints, leading to tumor development.

Our findings suggest that an important role for the forkhead transcription factors in stress-resistance and the aging process may be conserved in mammals. An organism's ability to respond to stress, in particular the capacity of cells to repair damage to DNA, correlates with an increased longevity (26). We propose that under low stress conditions, FOXO3a may promote DNA repair, whereas under higher levels of stress, FOXO3a may induce a program of cell death. This graded response to stress stimuli would protect cells from damage but also facilitate the removal of heavily damaged cells, resulting in an overall increase in the life-span of the organism.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: michael.greenberg{at}tch.harvard.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article