E2F3 Is a Mediator of DNA Damage-Induced Apoptosis

Abstract

The E2F transcription factors have emerged as critical apoptotic effectors. Herein we report that the E2F family member E2F3a can be induced by DNA damage through transcriptional and posttranslational mechanisms. We demonstrate that the posttranslational induction of human E2F3a is dependent on the checkpoint kinases. Moreover, we show that human E2F3a is a substrate for the checkpoint kinases (chk kinases) and that mutation of the chk phosphorylation site eliminates the DNA damage inducibility of the protein. Furthermore, we demonstrate that E2F1 and E2F2 are transcriptionally induced by DNA damage in an E2f3-dependent manner. Finally, using both in vitro and in vivo approaches, we establish that E2f3 is required for DNA damage-induced apoptosis. Thus, our data reveal the novel ability of E2f3 to function as a master regulator of the DNA damage response.

The orderly progression through the cell cycle is governed by the coordinated activity of the E2F transcription factors (16). The E2Fs are a family of DNA binding proteins whose activity is regulated through their interaction with the retinoblastoma family (pRb, p107, and p130) (6). The E2Fs regulate a cohort of cell cycle regulatory genes, and they thereby coordinate the transitions through the different phases of the cell division cycle. The 8 E2F proteins (E2F1 to E2F8) are subdivided into two classes based on their transcriptional regulatory activities: the transactivating members (E2F1 to E2F3) and the transrepressing members (E2F4 to E2F8) (6). In general, the transactivating members drive cell cycle progression by inducing the expression of proliferation-associated genes, whereas the transrepressing members impede cell growth by repressing these genes.

In addition to regulating cell growth, the E2F transcription factors can promote apoptosis through the activation of death-inducing genes, such as p73, caspases, Apaf1 and Bcl-2 homology region 3 (BH3)-only proteins (9, 14, 24, 25, 27, 36). Ectopic E2f1 expression can induce both p53-dependent and p53-independent apoptosis (16). Although initial studies had suggested that only E2F1 could induce apoptosis, other studies have demonstrated that E2F2 and E2F3a also possess proapoptotic functions (7, 17, 31, 38). Indeed, the ectopic expression of E2F3a has been shown to promote apoptosis both in vitro and in vivo (5, 17, 31, 38). Interestingly, apoptosis induced by ectopic E2F3 expression can occur in a p53-independent manner (17, 31). However, the absence of E2f1 eliminates the proapoptotic function of E2F3a (17, 31). Thus, one interpretation of these studies is that among the activating E2Fs, only E2F1 has a specific proapoptotic function and that the induction of apoptosis by other E2Fs may be a consequence of deregulated E2F1 activity.

Analysis of Rb null embryos has revealed that E2Fs can exhibit proapoptotic activities in a physiological setting (13). Loss of Rb results in extensive apoptosis in developing embryos (15, 21). Genetic deletion of either E2f1 or E2f3 suppresses cell death in the lens of Rb null embryos (39).

Treatment of cells with DNA damaging agents can induce E2F1 through posttranslational stabilization of the protein (3, 10, 11, 23, 29). The induction of E2F1 by DNA damage appears to be a specific property of E2F1, since it has previously been reported that the other E2F transcription factors are not similarly regulated (18, 32). The posttranslational stabilization of E2F1 appears to be dependent on its phosphorylation by two different DNA damage-responsive kinase families, the ATM (ataxia-telangiectasia mutated)/ATR (ataxia-telangiectasia and Rad3-related) kinases and chk kinases (checkpoint kinase) (18, 34, 35). Recently, small interfering RNA (siRNA)-mediated knockdown of either chk1 or chk2 was shown to prevent the induction of E2F1 protein in response to DNA damage (37). In addition, it was demonstrated that siRNA-mediated depletion of E2F1 prevented the induction of the proapoptotic E2F target gene, p73, and apoptosis (37). Thus, these studies reveal that in response to DNA damage, E2F1 can be induced through a posttranslational mechanism and that it plays a role in genotoxic stress-induced cell death.

Recently, it was reported that E2F7 and E2F8 can also be induced by subapoptotic doses of DNA damage (30). However, in contrast to E2F1, E2F7 and E2F8 cooperatively suppress chemotherapy-induced apoptosis, potentially by repressing the expression and activity of E2F1 (30).

During an analysis of E2F expression in cells treated with chemotherapy, we observed that some DNA damaging agents can induce the other members of the activating E2F subclass, E2f2 and E2f3a. We demonstrate that E2f3a is induced in a transcriptional and posttranslational manner by DNA damage. In addition, we show that DNA damage can transcriptionally induce both E2f1 and E2f2 in an E2F3a-dependent manner and we demonstrate that E2f3 is required for DNA damage-induced apoptosis both in vitro and in vivo.

MATERIALS AND METHODS

Cell culture.

The cell lines Saos2, U2OS, HCT116, H1299, and MDAH041 were cultured in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS). Immortalized mouse embryonic fibroblasts (MEFs) derived from wild-type, E2f1 null, E2f1 E2f2 double null, and E2f3 null mice were cultured in DMEM supplemented with 10% FBS. For retroviral infections, the Ecopack (Clontech) retrovirus packaging cell line was transfected with either an empty vector (pBabe) or myc-tagged E2f3a-expressing vector. Two days later, the viral medium was collected and filtered through a 0.45-μm filter, supplemented with 4 μg/ml Polybrene, and applied overnight to the MEF cultures. The medium was changed the following day, and drug selection (puromycin [1 μg/ml]) was performed for 3 days. The surviving cells were allowed to expand and immediately used for experiments.

Western blot analysis.

To harvest cells, the plates were rinsed with ice-cold phosphate-buffered saline (PBS), and the cells were lysed in PBS supplemented with 0.5% sodium dodecyl sulfate (SDS), 0.5% NP-40, 0.1% sodium deoxycholate, 1 mM EDTA, Roche protease inhibitor cocktail, 1 mM dithiothreitol (DTT), and phenylmethylsulfonyl fluoride (PMSF). The lysates were sonicated and clarified by centrifugation. The protein concentration was determined using the BCA method (Pierce), and equal amounts of protein were separated using SDS-polyacrylamide gel electrophoresis (PAGE). Western blot analysis was performed using chk1 (Cell Signaling; 1:1,000), chk2 (Cell Signaling; 1:1,000), E2F1 (Upstate; 1:1,000), E2F2 (TFE-25; 1:200; Santa Cruz), E2F3 (C-18; 1:1,000; Santa Cruz), E2F3 (N-20; 1:400 and 1:1,000; Santa Cruz), E2F3 (PG-14, 1:400; Santa Cruz), E2F4 (C-20; 1:200; Santa Cruz), E2F5 (C-20 and E-19; 1:200; Santa Cruz), and actin (Sigma; 1:10,000). All primary antibody incubations (except actin) were performed overnight in blocking buffer (PBS containing Tween 20 [PBST] or Tris-buffered saline supplemented with Tween 20 [TBST] containing 5% milk). To analyze protein stability, cycloheximide was added to drug-treated cultures for the indicated times, after which the cells were processed as described above. Protein levels were quantitated from Western blot data using the NIH ImageJ software. All experiments were independently repeated at least three times. The phosphoserine 124 antibody was generated by 21st Century Biochemicals using the peptide PPALGRGG[pS]GG, with the phosphorylated serine shown in brackets. Sera from immunized mice were initially exhaustively immunodepleted with a nonphosphopeptide affinity column and then subsequently immunopurified with a phosphopeptide affinity column.

Northern blot analysis.

Total RNA was isolated using Tri reagent. Equal amounts of total RNA (10 μg) were processed for Northern blot analysis using standard methods. The E2F3 probe was generated by PCR using the full-length E2f3a cDNA as a template. The hybridized blot was extensively washed and exposed to film.

RT-PCR analysis.

Total RNA (1 μg) was used for reverse transcription-PCR (RT-PCR). PCR was performed with the following primers: E2F1 FW (FW for forward) (ATGTTTTCCTGTGCCCTGAG) and RV (RV for reverse) (ATCTGTGGTGAGGGATGAGG), E2F2 FW (GGCCAAGAACAACATCCAGT) and RV (TGTCCTCAGTCAGGTGCTTG), and E2f3a FW (AAGAGCAGGAGCGAGAGATG) and RV (GAGGTGGTGGAAGTGTTCGT). The cycling conditions were 3 min at 95°C, followed by 27 cycles, with 1 cycle consisting of 30 s at 95°C, 30 s at 55°C, and 30 s at 72°C. PCR products were resolved on a 2% agarose gel, and digital images were captured using an Eagle Eye II apparatus (Stratagene). Analysis of transactivation-competent p73 (TA-p73) expression was performed with two different TaqMan probes Hs01056228_m1 and Hs01065727_m1 in an Applied Biosystems Prism 7000 sequence detection system (Sealy Center for Cancer Cell Biology Real-Time PCR Core Facility). Data analysis was performed using Microsoft Excel as described by the manufacturer (Applied Biosystems). The amount of target was normalized to endogenous 28S rRNA content. PCR analysis was performed at least three times on independently generated experimental samples.

siRNA transfection.

siRNA was transfected using Lipofectamine (Invitrogen) as previously described (22). The siRNAs used were as follows: control (CATGTCATGTGTCACATCTC), E2F1 (GCTCGACTCCTCGCAGATC), and E2f#2 (AAACTGTTAACCGAGGATTCA). The previously described E2F3 siRNA (referred to herein as E2F3#1) was also used as indicated in the figure legend (20). To knock down the chk kinases, we used chk1 and chk2 siRNAs as has previously been described by others (37).

Expression vectors.

E2F3 constructs were created using the human E2f3a cDNA as a template for PCR. The myc tag was introduced into the N terminus of the E2F cDNA by PCR. The myc-tagged version was then used to generate the S124A mutant with the QuikChange kit (Stratagene) with the following oligonucleotides 5′GGACGCGGCGGCGCCGGCGGCGGCGGC and GCCGCCGCCGCCGGCGCCGCCGCGTCC (targeted bases are shown in boldface type). The constructs were sequenced in both directions. The mouse myc-tagged E2f3a expression vector has been previously described (33).

Kinase assays.

The amino terminus of E2f3a was amplified from either a wild-type or S124A mutant construct and cloned in frame downstream of a glutathione S-transferase (GST) tag in the PGEX-2T vector. Recombinant protein expression was induced with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG). Bacterial cell pellets were resuspended in PBS containing 5 mM DTT, sonicated, clarified by centrifugation, and subsequently used for batch purification of recombinant protein using glutathione Sepharose 4 (Pharmacia). The purified protein was then quantitated after SDS-PAGE and Coomassie blue staining by comparing the band intensities to those of bovine serum albumin (BSA) standards. Approximately 1 μg of recombinant protein was used in an in vitro kinase assay with 100 ng of recombinant chk1 or chk2 (Cell Signaling). The reaction buffer contained 50 mM Tris-HCl, 1 mM DTT, 10 mM MgCl₂, 50 μM cold ATP, 100 μM sodium orthovanadate, and 10 μCi [γ-³²P]ATP in a 20-μl volume. The reaction was performed at 30°C for 20 to 30 min, after which loading buffer was added to the reaction mix, followed by denaturing and separation by SDS-PAGE. The gel was dried and exposed to film. The results were confirmed in three independent experiments.

Chromatin immunoprecipitation.

Saos2 cells were harvested and processed following the Upstate chromatin immunoprecipitation (CHIP) protocol, except that the protein G agarose was from KPL Laboratories. The antibodies used were control rabbit IgG (Santa Cruz), E2F3 (C-18; Santa Cruz), and E2F4 (C-20; Santa Cruz). The PCR was performed using primers specific to the E2F1 promoter: forward (GATGGCTGGGCTGTGGAA) and reverse (GCCACTTTTACGCGCCAA). Similar results were obtained in two independent experiments.

In vivo apoptosis assay.

Pregnant females at either 10.5 or 13.5 days postcoitus (dpc) were irradiated with 5 Gy gamma radiation and then sacrificed 5 h later to collect embryos. The embryos were fixed in phosphate-buffered formalin overnight, then processed, embedded, and serially sectioned. Terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) analysis was performed using the TdT FragEL kit from Oncogene Research as previously described (19). At least 4 embryos from different litters were assessed for each genotype, and a minimum of 300 cells were counted from each embryo.

RESULTS

E2F3a is induced by DNA damage.

We assessed expression of the E2F family of transcription factors in cells treated with different DNA damaging agents (doxorubicin, etoposide, and neocarzinostatin). As previously reported by several laboratories, E2F1 was induced by all three drugs (Fig. 1a). Interestingly, we observed that E2F3a was also induced by doxorubicin and etoposide, whereas E2F3b protein levels declined in response to these drugs (Fig. 1a). The radiomimetic drug neocarzinostatin did not reproducibly alter E2F3a levels, while it reduced E2F3b expression. In contrast, none of the chemotherapeutic treatments altered E2F4 and E2F5 expression (Fig. 1a).

FIG. 1.

DNA damage induces E2F3a. (a) Saos2 cells were left untreated (Unt) (control) or treated with doxorubicin (DX) (300 ng/ml), etoposide (ET) (6 μg/ml), or neocarzinostatin (NCZ) (300 ng/ml) overnight. E2F expression was determined by Western blot analysis. (b) U2OS cells were treated overnight with doxorubicin (DX) or etoposide (ET) at the indicated doses. (c and d) Western blot analysis was performed to detect E2F expression. HCT116 (c) or MDAH041 (d) cells were treated with doxorubicin overnight and then harvested for Western blot analysis. The polyclonal antibody, C-18, was used to detect E2F3 expression in panels a to d. (e) Saos2 cells were transfected with different amounts of siRNA (control [CT], E2f3#1, and E2f3#2 siRNA) as indicated. Two days later, whole-cell extracts were used for Western blot analysis. Note that the N-20 antibody fails to detect any changes, whereas the C-18 antibody detects the dose-dependent siRNA-mediated knockdown of E2F3. Two different lots of the N-20 antibody failed to detect E2F3a by Western blot analysis. (f) Saos2 cells were transfected overnight with control or E2f3 siRNA. The following day, the transfected cultures were divided into three plates, allowed to recover overnight, and then treated with doxorubicin as indicated. Western blot analysis was performed with a monoclonal E2F3 antibody (PG-14) and actin.

The observation that E2F3a was inducible by DNA damage in Saos2 cells prompted us to examine whether other cell types also exhibited this response. The DNA damage induction of E2F3a is not cell type specific, since similar results were obtained with osteosarcoma U2OS cells (Fig. 1b), HCT116 colorectal cells (Fig. 1c), and MDAH041 skin fibroblasts cells (Fig. 1d). These data indicate that E2F3a, like E2F1, can be induced by DNA damage.

These data differ from the results reported in two previous studies (18, 32). We noted that in those studies, the E2F3 N-20 antibody had been used to examine E2F3 expression. In an attempt to reconcile the differences between these results, we assessed the ability of the E2F3 N-20 antibody to detect E2F3a by Western blot analysis on lysates from cells in which E2F3 had been knocked down with two different E2F3 siRNAs. The N-20 antibody detected a band that was not affected by either E2F3 siRNA (Fig. 1e). In contrast, the C-18 antibody detected a band that was reduced in a dose-dependent manner by both E2F3 siRNAs (Fig. 1e). To further validate our results, we detected E2F3 expression with a monoclonal antibody, PG-14. Western blot analysis with this monoclonal antibody revealed that E2F3a was induced by DNA damage (Fig. 1f). The specificity of this antibody is demonstrated by the fact that it does not detect E2F3 in cells transfected with E2f3 siRNA (Fig. 1f). We conclude that E2F3a is inducible by DNA damage.

DNA damage enhances E2F3a protein stability.

Having determined that E2F3a was induced by genotoxic stress, we wanted to determine whether this was occurring through a transcriptional or posttranslational mechanism. The E2f3a and E2f3b mRNAs arise from the use of alternative promoters that produce mRNA species of different sizes (1). To determine whether changes in the mRNA levels correlated with the observed changes in protein levels, we performed Northern blot analysis to detect both E2f3 species. Both the E2f3a and E2f3b mRNA transcripts were attenuated within 6 h after drug treatment (Fig. 2a). Thus, the decrease in E2F3b protein observed in Fig. 1 may involve a decrease in the E2f3b mRNA level after drug treatment. The regulation of E2f3a was distinct from that of E2f3b in that it was biphasic. Initially, there was a decrease in E2f3a mRNA within 9 h after treatment and a pronounced increase at 24 h posttreatment. The expression pattern of E2f3a mRNA differed from the protein levels which were induced early on (3 h) after chemotherapy treatment (Fig. 2b). Since the increase in E2F3a protein did not correlate with a concurrent increase in mRNA levels, we interpreted this to indicate that there is an initial posttranscriptional mechanism that regulates E2F3a protein accumulation in response to DNA damage. To directly test this possibility, we assessed whether exogenous E2f3a could be induced by doxorubicin. We transfected Saos2 cells with either a control vector or a myc-tagged human E2f3a (driven by a promoter that is not responsive to DNA damage) and then treated the cells with doxorubicin. Western blot analysis with an anti-myc tag antibody revealed that the exogenously expressed E2F3a was induced by DNA damage (Fig. 2c). Therefore, we conclude that E2F3a can be induced by both transcriptional and posttranslational mechanisms in response to DNA damage.

FIG. 2.

DNA damage induces E2F3a in a transcriptional and posttranslational manner. Saos2 cells were treated with doxorubicin (DX) (300 ng/ml) and harvested at the indicated times (in hours). (a and b) Northern (a) or Western (b) blot analysis was performed to detect E2F3 expression. EtBr, ethidium bromide. (c) Saos2 cells were transfected with a myc-tagged E2F3 expression vector or a control vector. The next day, the cells were treated with doxorubicin and then harvested the following day for Western blot analysis with an anti-myc tag antibody. (d) Cycloheximide (CHX) was added to doxorubucin-treated Saos2 cultures, and then the cells were harvested at the indicated times (in minutes). (e) Densitometric analysis was performed using the NIH ImageJ software to determine the stability of E2F3a protein in control and doxorubicin-treated cells. Similar results were obtained in three independent experiments.

We next determined whether the increase in E2f3a was due to an extended half-life of the protein in drug-treated cells. Saos2 cells were treated with doxorubicin overnight, and then cycloheximide was added to block new protein synthesis. Comparison of the E2F3a levels in untreated and doxorubicin-treated cells revealed that E2F3a exhibited an enhanced stability in the latter (Fig. 2d and e). Indeed, the half-life of E2F3a in untreated cells was 60 min, whereas it was >90 min in doxorubicin-treated cells. Taken together, these data suggest that the posttranslational induction of E2F3a by doxorubicin is a result of increased protein stability.

E2F3a is induced by DNA damage in a chk1-dependent manner.

DNA damaging agents can induce the accumulation of E2F1 through ATM and chk kinase-dependent mechanisms (18, 35, 37). Since ATM has been reported to lack the ability to phosphorylate E2F3 (18), we focused on the chk kinases to determine whether they mediated the induction of E2F3a after DNA damage. We first assessed whether the siRNA knockdown of the chk kinases impacted E2F3a induction by genotoxic stress. siRNA knockdown of chk1 prevented the induction of E2F3a by doxorubicin (Fig. 3a). Similar results were observed with siRNA knockdown of chk2 (data not shown). Taken together, these data implicate the chk kinases in the control of E2F3a accumulation in response to DNA damage.

FIG. 3.

chk-dependent phosphorylation of E2F3a promotes its posttranslational induction in response to DNA damage. (a) Saos2 cells were transfected with either a control or chk1 siRNA. The next day, the cells were split into different dishes, and 24 h later, they were treated overnight with doxorubicin. The cells were harvested, and Western blot analysis was performed with the indicated antibodies. (b) Schematic representations of E2F3a and E2F3b proteins. Note that the putative chk1 site is present only in the N terminus of E2F3a. (c) Saos2 cells were transfected with an empty vector, a vector expressing wild-type (WT) E2F3, or the E2F3-S124A mutant (carrying a serine to alanine mutation at residue 124) and a green fluorescent protein (GFP) expression vector. The next day, the cells were treated overnight with doxorubicin and then harvested for Western blot analysis. Western blot detection of GFP was used as a control for transfection and loading. To calculate the fold induction, E2F3a levels were normalized to the GFP levels and expressed relative to their respective untreated sample. Note that the S124A mutant is not induced in response to DNA damage. (d) A GST E2F3a fusion protein carrying either the wild-type or S124A mutant N terminus was used as a substrate for an in vitro kinase assay with recombinant chk1 and chk2. The results of autoradiography of the phosphorylated substrates and a Coomassie blue-stained gel demonstrating that equivalent amounts of the E2F3 substrates were used in the kinase reactions. (e) An in vitro kinase assay was performed as described above except that no radioactive label was added to the reaction mixture. Western blot analysis was performed with an anti-phospho-S124 antibody (anti-P-S124) or anti-E2F3 antibody. (f) Saos2 cells were treated with doxorubicin and harvested for Western blot analysis with the indicated antibodies. (g) Saos2 cells were transfected with a control or chk1 siRNA as in panel a, and then treated with doxorubicin. Western blot analysis was performed with the indicated antibodies.

Phosphorylation of serine 124 is required for the posttranslational induction of E2F3a in response to DNA damage.

To determine whether the chk kinases were directly regulating E2F3a, we initially tested whether they could phosphorylate E2F3a. Since E2F3a, and not E2F3b, was induced in drug-treated cells, we speculated that the additional 132 amino acid residues in the N terminus of E2F3a might harbor a chk phosphorylation site. Indeed, examination of the amino acid sequence of the E2F3a N terminus revealed a potential chk phosphorylation motif at serine 124 that is similar to the chk consensus sequence (Fig. 3b) (12). To assess whether serine 124 is important for the posttranslation induction of E2F3a, we mutated this serine to an alanine and then tested what effect the mutation (S124A) had on E2F3a accumulation in response to chemotherapy. We observed that the S124A mutant was no longer induced by doxorubicin (Fig. 3c). Next, to determine whether E2F3a is a direct target of the chk kinases, we performed an in vitro kinase assay using the amino terminus of E2F3a as a substrate. We observed that both chk1 and chk2 were able to phosphorylate a GST-E2F3a fusion protein (Fig. 3d). A GST-E2F3a mutant in which serine 124 was mutated to alanine (S124A) was not phosphorylated by either kinase, indicating that the phosphorylation occurs at this site (Fig. 3d). These results are in line with the notion that chk1 controls the induction of the E2F3a protein by chemotherapy.

We wanted to determine whether phosphorylation of E2F3a occurs in vivo, and thus we generated a phospho-specific antibody that detects only phosphorylated serine 124. We first assessed the specificity of the antibody. We performed an in vitro “cold” kinase assay and then subsequently used Western blot analysis to determine whether the antibody could distinguish between the phosphorylated and nonphosphorylated forms of GST-E2F3a. We observed that the phospho-S124 antibody detected only the phosphorylated form of GST-E2F3a (Fig. 3e). Next we determined whether this modification occurs in intact cells and whether it is chk1 dependent. Thus, we transfected Saos2 cells with a control or chk1 siRNA and then treated them with doxorubicin. Western blot analysis with the phospho-S124 antibody revealed that doxorubicin could induce E2F3a phosphorylation on serine 124 (Fig. 3f). In addition, chk1 knockdown eliminated the DNA damage-induced phosphorylation at this residue (Fig. 3g). Taken together, these results suggest that E2F3a is directly phosphorylated by chk kinases and that the phosphorylation of serine 124 is required for the posttranslational induction of E2F3a protein by chemotherapy. Importantly, this chk phosphorylation site is present only in E2F3a and not E2F3b, which provides an additional mechanism for the observed difference in regulation of the protein levels of these E2Fs in chemotherapy-treated cells.

E2F3 is required for the induction of E2F1 by DNA damage.

The E2F transcription factors are able to regulate each other's expression (17, 28). Thus, we sought to determine whether the induction of the activating E2Fs in response to DNA damage was interdependent. We transfected Saos2 cells with a control, E2F1 or E2f3 siRNA and then treated the cells with doxorubicin to induce E2F expression. As expected, we observed that both E2F1 and E2F3a were induced by DNA damage (Fig. 4a). Surprisingly, we found that E2F2 was also induced (Fig. 4a). We confirmed that this was indeed E2F2 using several different E2F2 antibodies (data not shown). Thus, all three activating E2Fs are inducible by DNA damage. Importantly, we observed that E2F1 knockdown not only reduced E2F1 protein levels but also eliminated the induction of E2F2 (Fig. 4a). Additionally, E2F3a levels were lower in cells transfected with E2f1 siRNA, although the protein was still induced by DNA damage. E2F3 knockdown greatly reduced the levels of E2F3 as well as both E2F1 and E2F2 (Fig. 4a). These data suggested that a hierarchy might exist in which E2F3 controls the expression of E2F1, and in turn, E2F1 controls E2F2. Importantly, E2F3 knockdown effectively reduces the DNA damage induction of all three activating E2Fs.

FIG. 4.

E2F3 is required for the induction of E2F1 by doxorubicin. (a) Saos2 cells were transfected with a control, E2f1, or E2f3 siRNA. The following day the cells were replated into parallel cultures and then treated with doxorubicin as indicated. Western blot analysis was performed on whole-cell extracts to detect E2F1, E2F2, E2F3, and actin. (b) Saos2 cells were transfected with a control siRNA or an siRNA targeting the 3′ untranslated region (E2f3 UTR) of E2f3. The cells were replated into parallel plates and treated as indicated with doxorubicin. Western blot analysis was performed with the indicated antibodies. Note that both E2f3 siRNAs (from panels a and b) downregulate E2F1 expression in Saos2 cells. (c) U2OS osteosarcoma cells were transfected with either a control or E2f3 UTR siRNA. The following day, the cells were replated into parallel plates and subsequently treated with doxorubicin as indicated. Western blot analysis was performed with the indicated antibodies. (d) Normal Wi38 lung fibroblasts were transfected twice with control or E2f3 siRNA. The cells were then replated into parallel cultures and subsequently treated with doxorubicin. The cells were harvested the next day and processed for Western blotting with the indicated antibodies. A monoclonal antibody (PG-37) was used to detect E2F3.

Although BLAST analysis of the E2f3 siRNA sequence failed to identify the E2f1 mRNA as a potential target, we wanted to eliminate the possibility that we were observing off-target effects of the siRNA, and thus, we employed another siRNA that was directed against the 3′ untranslated region (3′ UTR) of E2f3. In contrast to the coding region of the E2F transcription factors, the sequences of their untranslated regions are highly divergent. Saos2 cells were transfected with either a control or E2f3 UTR siRNA, then 48 h later they were harvested, and Western blot analysis was performed to assess E2F expression. Again, we observed that the siRNA directed against the UTR of E2f3 reduced E2F3 and E2F1 in both unstressed and stressed conditions (Fig. 4b). Thus, since different E2f3 siRNAs are able to downregulate E2F1 expression, it is unlikely that this is an off-target effect of the siRNAs.

Our data thus far suggested that E2F3 could control E2F1 expression, although these observations had been made in Saos2 cells which lack both pRB and p53. Thus, we sought to determine whether E2F3 could regulate E2F1 in cells which have these pathways intact. We selected the U2OS osteosarcoma cell line because it is wild type for both Rb and p53. Transfection of the U2OS cells with the E2f3 UTR siRNA caused a reduction in the E2F3a protein levels, although to a lesser extent than in the Saos2 cells. We observed that E2F1 basal levels were essentially not affected by the E2f3 UTR siRNA, although we did not achieve a complete reduction of E2F3 (Fig. 4c). However, E2F1 was clearly not induced by doxorubicin in cells transfected with the E2f3 UTR siRNA (Fig. 4c). These data suggest that E2F3 controls the induction of E2F1 in response to DNA damage.

The E2F family has been shown to regulate the expression of chk1. Since chk1 has been shown to regulate the posttranslational induction of E2F1 in response to DNA damage, the possibility existed that E2F3 was regulating E2F1 accumulation indirectly by transcriptional regulation of chk1. However, we observed no difference in chk1 protein levels between control and E2f3 siRNA-transfected cells (Fig. 4c). Thus, it does not appear that E2F3 regulates E2F1 induction through an indirect mechanism involving chk1.

We next examined whether E2F3 could regulate E2F1 expression in normal, nonimmortalized cells. We transfected the Wi38 human fibroblasts with a control or E2f3 siRNA and then challenged them with chemotherapy. Doxorubicin treatment induced the expression of both E2F1 and E2F3a (Fig. 4d). Importantly, E2F3 knockdown also reduced the basal expression of E2F1 and strongly reduced its induction in response to DNA damage (Fig. 4d). We conclude from these experiments that E2F3 can control the accumulation of E2F1 in response to DNA damage in normal cells.

The activating E2F transcription factors are transcriptionally induced in response to certain DNA damaging agents.

DNA damaging agents can induce E2F1 through posttranslational mechanisms (16). Given that we observed an increase in E2f3a mRNA levels after doxorubicin treatment, we sought to determine whether there were similar changes in the level of E2f1 mRNA that correlated with its induction by genotoxic stress. The Saos2 cell line was treated with three different drugs (doxorubicin, etoposide, and cisplatin) for 24 h and then the cells were harvested for RT-PCR analysis of E2F expression. Both doxorubicin and etoposide induced expression of the E2f1 mRNA, whereas cisplatin did not (Fig. 5a). The observation that cisplatin treatment does not modulate E2f1 mRNA levels is consistent with what has previously been shown for this drug (18). These data raised the possibility that E2f1 induction by some DNA damaging agents can occur through transcriptional and posttranslational mechanisms.

FIG. 5.

The activating E2F transcription factors are transcriptionally induced by some DNA damaging agents. (a) Saos2 cells were treated overnight with doxorubicin (DX) (300 ng/ml), etoposide (ET) (6 μg/ml), or cisplatin (CIS) (6 μg/ml). Total RNA was isolated and used for RT-PCR analysis of E2f1 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). (b) Saos2 cells were treated overnight with the indicated doses of doxorubicin and total RNA was isolated and used for RT-PCR analysis to detect E2f1, E2f2, E2f3a, or GAPDH. (c) Normal IMR90 cells were treated overnight with the indicated doses of doxorubicin and then processed for RT-PCR analysis to detect E2f1 and GAPDH. (d) Normal IMR90 cells were treated with doxorubicin and then processed for real-time RT-PCR analysis of E2f1 and E2f3. (e) Control or E2f3 siRNA was transfected into Saos2 cells, and on the following day, the cells were replated, allowed to recover overnight, and then treated with doxorubicin. The next day, the cells were harvested and processed for RT-PCR analysis of E2f1, E2f2, and GAPDH. (f) Saos2 cells were treated overnight with doxorubicin. The cells were fixed and processed for chromatin immunoprecipitation (IP) analysis to detect the association of E2F3 with the E2f1 promoter.

To further explore the involvement of a transcriptional mechanism in the induction of E2f1 by DNA damage, we determined whether the mRNA was induced in a dose-dependent manner by doxorubicin. RT-PCR analysis revealed that E2f1 mRNA levels correlated with the increasing doses of doxorubicin (Fig. 5b). Similarly, E2f2 was induced in a dose-dependent manner (Fig. 5b). In contrast, E2f3a was less responsive and was induced only at the higher dose of doxorubicin (Fig. 5b). Notably, E2f1 mRNA induction by doxorubicin was not limited to cancer cells, since we observed the same in normal human IMR-90 fibroblasts (Fig. 5c). Similar results were observed using real-time RT-PCR analysis of E2f1 and E2f3a (Fig. 5d). Thus, all three activating E2Fs (E2f1, E2f2, and E2f3a) can be transcriptionally induced by DNA damage.

We assessed whether E2f3 was contributing to the transcriptional induction of the other activating E2F transcription factors. We knocked down E2F3 and then assessed the impact on E2f1 and E2f2 expression. RT-PCR analysis revealed that E2F3 knockdown virtually eliminated the transcriptional upregulation of E2f1 and E2f2 in drug-treated cells (Fig. 5e). These results suggest that the transcriptional induction of the activating E2Fs (i.e., E2f1 and E2f2) in response to chemotherapy occurs in an E2F3-dependent manner.

To determine whether E2F3 was directly regulating E2F1 expression, we examined whether it was present on the E2F1 promoter using chromatin immunoprecipitation (CHIP). E2F3 was associated with the E2f1 promoter both in untreated and doxorubicin-treated Saos2 cells (Fig. 5f). Since the C-18 polyclonal antibody recognizes both E2F3a and E2F3b, we were unable to determine whether there were any relative differences in the association of these two isoforms with the E2f1 promoter in untreated cells. However, since treatment with doxorubicin strongly reduces the level of the E2F3b protein (Fig. 1a), it is likely that any binding activity observed in doxorubicin-treated cells is due to E2F3a. In contrast, although we observed that E2F4 also associates with the E2f1 promoter in untreated cells, this association was lost after doxorubicin treatment. These data support a direct role for E2F3a in the regulation of E2f1 expression in response to DNA damage.

E2f3-deficient MEFs exhibit a defective induction of E2f1 in response to DNA damage.

The experiments described above relied on the use of siRNA to selectively test the role of E2f3 in the response to DNA damage. We wanted to use an alternative approach to test the possibility that there is a hierarchal mechanism for E2F induction by DNA damage, and thus we used mouse embryonic fibroblasts (MEFs) from genetically modified mice. MEFs derived from wild-type or E2f3 null mice were treated overnight with different doses of doxorubicin. In the wild-type MEFs, doxorubicin induced both E2F1 and E2F3a (Fig. 6a). In contrast, in E2f3 null MEFs, E2F1 was only modestly induced (Fig. 6a). We further examined whether the defect in E2f1 induction was at the transcriptional level by using real-time RT-PCR to analyze E2f1 mRNA levels in MEFs treated with doxorubicin. We included two other E2f3 null cell lines to ensure that the results we observed were not due to a nonspecific defect in one of the E2f3 lines. We observed that E2f1 induction by DNA damage was lost in all three E2f3-deficient lines (Fig. 6b).

FIG. 6.

Transcriptional induction of E2f1 in response to DNA damage is E2f3 dependent. (a) Mouse embryonic fibroblasts from wild-type or E2f3 null mice were treated overnight with doxorubicin. Western blot analysis was performed to detect the expression of E2F1, E2F3a, and actin. E2F3a was detected with the monoclonal antibody PG-14. (b) MEFs treated as described above for panel a were harvested, and total RNA was used to perform real-time RT-PCR analysis of E2f1 expression. Note the absence of E2f1 induction in three different E2f3 knockout (KO) MEF cell lines. (c) E2f3 knockout cells were infected with a retrovirus carrying either an empty vector (pBabe) or myc-tagged E2f3 expression vector (myc-E2f3a). After selection, the cells were treated with doxorubicin, and Western blot analysis was performed as indicated. (d) E2f3 knockout cells treated in parallel as described above for panel c and then processed for RT-PCR analysis of E2f1 and GAPDH expression.

If indeed the loss of E2f3 was compromising the induction of E2f1 by DNA damage, then it would follow that the ectopic expression of E2f3 should be sufficient to rescue this defect. To that end, we infected the E2f3 null MEFs with an empty vector or a vector expressing myc-tagged mouse E2f3a. After selection, the cells were treated with doxorubicin, and then Western blotting and RT-PCR analysis was performed. We found that the ectopic expression of E2F3a was sufficient to restore the DNA damage inducibility of E2F1 both at the protein and mRNA level (Fig. 6c and d). Taken together, we conclude that E2F3a directly controls the induction of E2F1 by DNA damage.

These results support the notion that a hierarchy exists in the mechanism by which E2F transcription factors are induced by DNA damage. However, an alternative explanation is that the DNA damage induction of the activating E2Fs is interdependent. If indeed the E2Fs are induced through an interdependent mechanism, then one would predict that a deficiency in any single member of the activating E2Fs should compromise the induction of the other two. Thus, one could predict from this model that a deficiency in either E2f1 or E2f2 should eliminate the induction of E2F3a by chemotherapy. Therefore, we tested this model by examining the induction of E2f3a in MEFs deficient for E2f1 (E2f1 knockout [KO]) or doubly deficient for E2f1 and E2f2 (E2f1/2 KO). Western blot analysis of wild-type, E2f1 KO, and E2f1/2 KO MEFs treated with doxorubicin revealed that E2F3a induction was not affected by the deficiency of the other activating E2Fs (Fig. 7a). Moreover, real-time RT-PCR analysis revealed that a deficiency in either E2f1 or E2f1 and E2f2 had no effect on the transcriptional induction of E2f3a (Fig. 7b). Taken together, these data support the hierarchal mechanism for the induction of the activating E2Fs by DNA damage and suggest that E2F3 is upstream of the other activating E2Fs.

FIG. 7.

E2f1 is not required for E2F3a induction by DNA damage. (a) Mouse embryonic fibroblasts derived from wild-type, E2f1 null (E2f1 KO) or E2f1 and E2f2 double null (E2f1/2 KO) mice were treated overnight with doxorubicin and processed for Western blot analysis of E2F3 expression with the monoclonal antibody PG-14 and actin. Note that neither deficiency in E2f1 or E2f2 affects the induction of E2f3a in response to DNA damage. (b) MEFs of the indicated genotypes were treated as described above for panel a and then processed for real-time RT-PCR to detect E2f3a expression. Similar results were obtained in another independent experiment. DKO, double knockout.

DNA damage-induced apoptosis requires E2F3.

The induction of apoptosis by different DNA damaging agents has been shown to involve E2F1 (18, 37). Given our findings that E2F3 was upregulated by DNA damaging agents, we wanted to determine whether it also played a role in drug-induced apoptosis. To test this possibility, we compared the ability of E2f1 and E2f3 knockdown to protect cells from DNA damage-induced cell death. Consistent with previous reports, siRNA-mediated knockdown of E2F1 reduced the extent of cell death induced by doxorubicin, etoposide, and neocarzinostatin (Fig. 8a) (37). Interestingly, E2F3 knockdown more potently suppressed the induction of cell death by these chemotherapeutic agents (Fig. 8a). Similar results were obtained in U2OS and H1299 lung adenocarcinoma cells treated with doxorubicin or etoposide (Fig. 8b and data not shown).

FIG. 8.

E2F3 mediates chemotherapy-induced apoptosis. (a) Saos2 cells were transfected with a control, E2f1, or E2f3 siRNA. Two days later, the cells were treated with etoposide (6 μg/ml), doxorubicin (300 ng/ml), or neocarzinostatin (300 ng/ml) for 48 h. The cells were fixed in culture, and nuclei were stained with 4′,6′-diamidino-2-phenylindole (DAPI) to permit the identification of apoptotic (pyknotic) nuclei. (b) U2OS cells were transfected with a control or E2f3 siRNA and then replated the next day as described above for panel a. Doxorubicin (500 ng/ml) or etoposide (6 μg/ml) was added to the cultures, and the cells were harvested the following day. The apoptotic cells were assessed as described above for panel a. A minimum of 300 cells were counted in each condition, and the experiment was repeated at least three times. Similar results were obtained with two different E2f3 siRNAs. (c) Saos2 cells were transfected with the indicated siRNAs (control, E2f1, and E2f3) and harvested 48 h later for analysis of TA-p73 expression by real-time RT-PCR. (d) Saos2 cells were transfected overnight with the indicated siRNAs (control, E2f1, and E2f3), replated the following day into two plates, and allowed to recover overnight. The cells were then either left untreated (UNT) or treated with doxorubicin (DX) overnight and then harvested for real-time RT-PCR analysis of TA-p73 expression. Note that either E2F1 or E2F3 knockdown suppresses the induction of p73 by doxorubicin.

DNA damage has been reported to induce the transcriptional activation of the p53 family member p73 in an E2F1-dependent manner (2, 37). Therefore, we analyzed whether E2F3 knockdown was affecting the induction of p73 by DNA damage. Real-time RT-PCR analysis revealed that E2F3 was also involved in the transcriptional regulation of p73, since siRNA knockdown of E2F3 partially reduced the basal expression of p73 and virtually eliminated its induction by doxorubicin (Fig. 8c and d). These results were remarkably more pronounced than the results obtained from cells in which E2F1 was knocked down (Fig. 8c and d). Thus, E2F3 deficiency compromises the DNA damage activation of two proapoptotic transcription factors, E2F1 and p73. Furthermore, these results provide insight into the mechanism by which E2F3 deficiency abrogates DNA damage-induced apoptosis. Taken together, our results suggest that E2F3 is a critical component of the apoptotic process initiated by genotoxic stress.

E2f3 is required for DNA damage-induced apoptosis in the developing CNS.

The data presented thus far suggest that E2F3 is responsive to DNA damage and that it may be required for the induction of genotoxic stress-induced apoptosis. Thus, far we have used in vitro models to assess the role of E2f3 in the response to DNA damage and we wanted to test whether the function that we were ascribing to E2f3 is biologically relevant. Therefore, we used a well-established in vivo model in which pregnant female mice are irradiated and then the embryos are harvested to determine the apoptotic response in the developing central nervous system (CNS) (19). We observed that irradiation caused a significant induction of TUNEL-positive staining (approximately 25%) in the developing CNS of E2f3 heterozygous embryos, whereas E2f3 null embryos had much reduced (approximately 5%) levels of apoptosis (Fig. 9). Taken together, our data indicate that E2F3 is a critical component of the DNA damage response.

FIG. 9.

E2f3 deficiency abrogates irradiation-induced apoptosis in the developing CNS. Pregnant E2f3 heterozygous females were irradiated with 5 Gy and sacrificed 5 h later, and the embryos were processed as described in Materials and Methods. (a) Representative pictures of TUNEL staining in the developing CNSs of embryos irradiated at 13.5 dpc. Shown are pictures of an E2f3 heterozygous (HET) and knockout (KO) embryo. The black arrows point to examples of apoptotic nuclei. (b) The extent of apoptosis after irradiation was assessed in heterozygous and knockout mice. At least 300 cells were counted per field, and several fields were assessed for each embryo. A minimum of four embryos were assessed for each genotype.

DISCUSSION

In this report we demonstrate that E2F3a can be induced by DNA damage. Our results differ from other studies which reported that E2F3 was not responsive to genotoxic stress. Although we do not know the reason for the discrepancy between the results, we speculate that at least some of the differences are due to the use of the E2F3 N-20 antibody, which in our experience did not recognize endogenous E2f3a by Western blotting (Fig. 1e). However, similar to what has been reported previously by others, neocarzinostatin did not induce E2F3 (Fig. 1a) (18). Thus, the possibility exists that only certain types of DNA damage induce the E2F3a protein. Importantly, there is precedence for this type of DNA damage-specific induction of a protein. The p33Ing2 protein is induced by some but not all DNA damaging agents (26). Nonetheless, since our analysis involved the use of a variety of techniques, several different antibodies and a variety of cell lines (Saos2, U2OS, MDA041, HCT-116, and MEFs), we propose that E2F3a is indeed responsive to genotoxic stress.

The induction of E2f3a by DNA damage occurs in a transcriptional and posttranslational manner. Interestingly, the E2f3 gene encodes two isoforms, E2F3a and E2F3b, which are under the control of different promoters. The E2F3 isoforms are identical except that the E2F3a protein has an additional 132 amino acids in its N terminus. In our study, we observed that doxorubicin initially reduced both E2f3a and E2f3b mRNA levels. During this same time period, the E2F3a protein was induced, whereas the E2F3b protein was not. Given that the proteins differ in their N terminus, we inferred that the longer N terminus of E2F3a contains a motif that permits the protein to be induced by DNA damage. Indeed we identified a chk consensus site (surrounding serine 124) in E2F3a's amino terminus, and we demonstrated that mutation of this serine to an alanine eliminates the DNA damage inducibility of the mutant protein. We further demonstrated that E2F3a can be phosphorylated both in vitro and in vivo by the chk kinases and that siRNA knockdown of chk1 abrogates induction of the E2F3a protein by DNA damage. Thus, our results illustrate that the accumulation of human E2F3a protein in response to DNA damage is under the control of the chk kinases.

Interestingly, serine 124 is not conserved in mouse E2F3a. Thus, the induction of E2F3a in the mouse appears to be primarily at the level of transcription. This interpretation is supported by the fact that unlike its human homologue, ectopically expressed mouse E2F3a is not induced by DNA damage (Fig. 6c). The fact that mouse E2F3a can stimulate the expression of E2f1 mRNA in response to DNA damage and can induce apoptosis in response to irradiation in vivo suggests that serine 124 in the human protein affects only its accumulation and not its function. It is intriguing that neither serine 124 in human E2f3a nor serine 364 in human E2f1 are conserved in the mouse. We can only speculate that these chk motifs in the human E2fs constitute a mechanism for a rapid increase in protein levels.

We observed in both human and mouse cells that DNA damage upregulated E2f3a mRNA levels. Similarly, E2f1 and E2f2 were transcriptionally induced by DNA damage. Thus, it appears that upregulation of all three activating E2F transcription factors is part of the DNA damage response. We observed that E2F1 knockdown eliminated the induction of E2F2 in human cells, suggesting that E2F1 is upstream of E2F2. Furthermore, E2F3 knockdown in human cells or E2f3 genetic deficiency in MEFs eliminated the induction of E2f1 mRNA, thus indicating that the transcriptional induction of E2f1 in response to DNA damage is E2f3 dependent. Therefore, one interpretation of our results is that in response to DNA damage there is a hierarchal regulation of activating E2f expression. Since the transcriptional induction of E2F1 (and by extension E2F2) in response to DNA damage requires E2F3, this suggests that E2F3 is the master regulator of this genotoxic stress-induced transcriptional program. The fact that E2f3 deficiency in vivo reduces the extent of gamma irradiation-induced apoptosis supports the notion that E2f3 is a key player in the DNA damage response.

Our study highlights the existence of a transcriptional mechanism for the induction of E2f1 in response to DNA damage. A recent study has also reported that E2f1 can be transcriptionally induced by a variety of DNA damaging agents in an ATM-dependent manner (4). Taken together, our results suggest that the induction of E2f1 by DNA damage can involve both transcriptional and posttranslational mechanisms.

Further studies are required to determine whether E2F3's sole function in the DNA damage response is to activate E2f1 or whether it is also regulates the induction of other proapoptotic genes. A previous study reported that E2F3a could not induce apoptosis in cells deficient for E2f1, whereas E2F1 did not require E2f3 for the induction of cell death (17). If the induction of apoptosis is occurring through the transcriptional activation of proapoptotic genes, this would suggest that only E2F1 has the capacity to activate this cohort of E2F targets. An intriguing finding is that the knockdown of either E2F1 or E2F3 eliminates the transcriptional induction of the proapoptotic p73 gene (Fig. 8d). In cells in which E2F3 is knocked down, we observed that E2F1 levels are also reduced, and thus, the lack of p73 induction in response to chemotherapy might be due to a reduction in total activating E2F activity. Likewise, the knockdown of E2F1 resulted in a modest decrease in E2F3, which coincided with a lack in p73 induction by chemotherapy. These data appear to support the model proposed by Helin's group in which total E2F1 activity is the determinant of whether cells will undergo apoptosis (17). Alternatively, the activator E2Fs (at physiologically expressed levels) might function cooperatively to activate gene expression, and thus, a reduction in E2F1 or E2F3 may be sufficient to prevent the activation of p73 in response to chemotherapy. In support of this possibility is the fact that we have observed that both E2F1 and E2F3 associate with the p73 promoter (data not shown). A similar cooperative mechanism has been reported for the activation of proapoptotic genes by the p53 family members (8). In that study, cells deficient for p63 or p73 fail to activate proapoptotic genes even though they contain a wild-type p53. This transcriptional dysfunction (i.e., lack of proapoptotic gene activation) of p53 correlates with its inability to interact with the target gene's promoters. Thus, the activator E2Fs, like the p53 family, when not overexpressed, might only be able to activate proapoptotic genes in a cooperative and codependent mechanism. At present, a similar transcriptionally cooperative mechanism for the E2Fs is speculative, and further studies are required to determine whether under physiological levels, both E2F1 and E2F3 are required for the apoptotic program to be initiated.