Single-stranded DNA Induces Ataxia Telangiectasia Mutant (ATM)/p53-dependent DNA Damage and Apoptotic Signals*

Single-stranded DNA has been speculated to be the initial signal in the DNA damage signaling pathway. We showed that introduction of single-stranded DNA with diverse sequences into mammalian cells induced DNA damage as well as apoptosis signals. Like DNA damaging agents, single-stranded DNA up-regulated p53 and activated the nuclear kinase ataxia telangiectasia mutant (ATM) as evidenced by phosphorylation of histone 2AX, an endogenous ATM substrate. Single-stranded DNA also triggered apoptosis as evidenced by the formation of caspase-dependent chromosomal DNA strand breaks, cytochrome c release, and increase in reactive oxygen species production. Moreover, single-stranded DNA-induced apoptosis was reduced significantly in p53 null cells and in cells treated with ATM small interfering RNA. These results suggest that single-stranded DNA may act upstream of ATM/p53 in DNA damage signaling.

DNA damage plays an important role in neuron degeneration, carcinogenesis, aging, as well as cell killing by chemotherapeutic agents (1)(2)(3). At the cellular level, DNA damage induces complicated cellular responses including checkpoint activation and apoptotic cell death (4,5). Ataxia telangiectasia mutant (ATM) 1 and p53 are two key regulatory molecules in DNA damage signaling and apoptosis. The tumor suppressor p53 is known to mediate G 1 cell cycle arrest and apoptosis (6). p53, which is a transcription activator, is up-regulated during DNA damage and transactivates a number of genes such as gadd45, p21, and bax involved in cell cycle arrest and apoptosis (7,8). While multiple mechanisms, including modification of p53 and Mdm2, are involved in up-regulation of p53 upon DNA damage (9,10), the immediate signal following DNA damage remains unclear.
The nuclear kinase, ATM, and its related kinase ATR are critically involved in integrating the initial DNA damage signals to cell cycle checkpoints and apoptosis (1,2). Activation of ATM/ATR by DNA damage has been widely demonstrated (1,11). For example, ionizing radiation and etoposide have been shown to induce ATM-dependent apoptosis (12), while camptothecin and UV have been shown to induce ATR-dependent apoptosis (13,14). Oxidative stress-induced neuron degeneration has also been shown to depend on ATM (15). It is generally believed that ATM mediates a signal from DNA double-strand breaks, whereas ATR mediates a signal from replication fork arrest (1,16). Activation of the ATM kinase, which is evidenced by phosphorylation of the ATM target proteins, has been demonstrated in cells treated with DNA damage agents, especially those that produce DNA double-strand breaks (17,18). ATM phosphorylates a large number of proteins in response to DNA damage, e.g. Ser-139 of histone 2AX (H2AX) (19), Thr-68 of Chk2 (20), and Ser-15 of p53 (21)(22)(23). While ATM kinase is considered to be one of the earliest regulatory molecules in DNA damage signaling, the initial DNA damage signal that triggers ATM activation remains unclear.
Single-stranded DNA, which activates the RecA protease activity toward LexA, is known to be the initial signal for activating the SOS repair response in bacteria (24). Singlestranded DNA has also been speculated to act as the initial signal for DNA damage responses in eukaryotic cells (16). In the current study, we have investigated the effect of singlestranded oligonucleotides on DNA damage signaling and apoptosis. We showed that transfection with synthetic oligodeoxyribonucleotides (oligos) as short as 5-mer up-regulated p53 and activated ATM. Moreover, single-stranded oligos induced ATM/p53-dependent apoptosis. Our results thus suggest that single-stranded DNA may act as a signal upstream of ATM/p53 in DNA damage and apoptosis signaling.  -AAG AGG  TGG TGG AGG AGG TGG TGG AGG AGG TGG AGG-3Ј), NT27 (5Ј-TTG AAT TCC TAG TTT CCC AGA TAC AGT-3Ј), NT12 (5Ј-TCG GTA  ACG GG-3Ј), NT18 (5Ј-TTA GGG TTA GGG TTA GGG-3Ј), THIONT5 (a phosphorothioate oligo of the sequence 5Ј-CGTTA-3Ј), NT8F (5Ј-GC-CACTGC-3Ј), and NT8R (5Ј-GCAGTGGC-3Ј). Alkaline Single Cell Gel Electrophoresis (SCGE) Assay-Genomic DNA strand breaks were analyzed by alkaline SCGE as described previously (25). Briefly, pelleted cells (10 4 /ml) were resuspended in 0.7% low melting point agarose (SeaKem Gold from BioWhittaker Molecular Application, Rockland, ME) in phosphate-buffered saline (PBS) at 37°C. After transfection, cells (3000/sample) in 150 l of liquefied 0.7% agarose were applied on top of a very thin agarose layer on a frosted glass slide. Cells were then lysed by treatment with an alkaline lysis solution (1% N-lauryl sarcosine, 1% Triton X-100, 2.5 M NaCl, 100 mM EDTA, and 10 mM Tris, pH 10.5) for 1 h, followed by incubation in a DNA-denaturing solution (0.3 M NaOH and 1 mM EDTA, pH 11.5) for 20 min. Slides were subjected to electrophoresis at 1.7 V/cm for 10 min (DNA SubCell TM gel electrophoresis apparatus from Bio-Rad) using the DNA-denaturing solution as the gel electrophoresis buffer. After electrophoresis, DNA was stained by submerging the slides into a DNA staining solution containing 1ϫ SybrGold (Molecular Probes, Eugene, OR). Nuclei in random and non-overlapping fields were visualized using epifluorescent illumination on a Zeiss Axioplan microscope. A comet tail length that was more than twice the diameter of the nuclei was considered as comet-positive. Comet-positive nuclei (damaged DNA) and intact nuclei (undamaged nuclei) were determined by analyzing, random microscopic field, at least 100 nuclei for each experiment. Each experiment was repeated three times. The percent of comet-positive nuclei was calculated and plotted.

Materials-Tissue
FIG. 1. Oligos induce nuclear DNA strand breaks in BALB/c 3T3 cells. BALB/c 3T3 cells were cultured to semiconfluency. Each oligo was complexed with 10 g of Cellfectin and then transfected into BALB/c 3T3 cells as described under "Experimental Procedures." Nuclear DNA strand breaks were monitored after 4 h using an alkaline SCGE assay. A, induction of nuclear DNA strand breaks by different oligos. Different oligos (NT36, NT27, THIONT5, NT18, or NT12, 200 nM each) were transfected into BALB/c 3T3 cells, and nuclear DNA strand breaks were determined by alkaline SCGE assay. Representative nuclei with intact DNA (no comet) and broken strands (comet) are shown. B, dose-dependent increase of nuclear DNA strand breaks induced by oligos. The NT36 oligo was used in this experiment. The concentration of the NT36 oligo for each reaction is indicated at the bottom of the figure in nM. C, time-dependent increase of oligo-induced nuclear DNA strand breaks. The THIONT5 oligo (1 M) was used, and the alkaline SCGE assays were performed after 1, 2, and 3 h of transfection. D, single-stranded, but not double-stranded, oligos induce nuclear DNA strand breaks. NT8F, NT8R, and annealed NT8F:NT8R duplex oligos (1 M each) were transfected into BALB/c 3T3 cells, and nuclear DNA strand breaks were monitored by alkaline SCGE assay. Each experiment was repeated three times showing similar results. Preparation of Cytoplasmic Fraction-BABL/c 3T3 cells were harvested and lysed in a modified buffer containing 100 mM Tris, pH 7.4, 10 mM MgCl 2 , 10 mM CaCl 2 , 1 mM dithiothreitol, 100 M EGTA, 0.5% Nonidet P-40, and 10 g/ml each of leupeptin, pepstatin, and aprotinin (26). The cytoplasmic fraction was separated from the mitochondrial and nuclear fractions by centrifugation at 12,000 ϫ g for 10 min at 4°C. Supernatants were mixed with the Lammelli SDS sample buffer and analyzed by Western blotting.
Measurement of Reactive Oxygen Species (ROS) Production-ROS production was measured using flow cytometry as described previously (27). Briefly, BALB/c 3T3 cells were grown to semiconfluence and transfected with the oligo:Cellfectin (10 g) complex as described above. After 4 h, 5-(6)-carboxy-2Ј,7Ј-dichlorodihydrofluorescein diacetate (DCFH-DA; Molecular Probes) was added to the medium to a final concentration of 20 M. Cultures were then incubated again for 40 min at 37°C to allow accumulation of DCFH. Cells were washed with PBS, trypsinized, and suspended in ice-cold PBS. Cellular fluorescence from a sample of 15,000 cells was analyzed using a Coulter EPICS Profile II flow cytometer (Coulter Electronics, Miami, FL). Fluorescence, excited at 488 nm, was detected using a 525 Ϯ 20 band pass filter. Cellular debris was excluded from analysis by gating based on forward angle light scatter. Histograms were analyzed using EPICS Workstation Software (version 4).
Apoptotic Chromatin Condensation and Caspase Activation Assays-After transfection, cells were harvested and washed twice with 1ϫ PBS buffer. Hoechst 33342 (0.5 M in PBS) and CaspACE (10 M in PBS) were then added to the cells. Cells were then incubated for 20 min at 37°C. Both chromatin condensation and caspase activation were examined by fluorescence microscopy.
3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium Bromide (MTT) Assay-BALB/c 3T3 cells were cultured to semiconfluence. Cells were transfected with indicated oligo (200 nM):Cellfectin (10 g) complexes as described above and incubated for 4 h. Cell viability was evaluated using a modification of the MTT reduction assay (28). Briefly, a sterile solution of 0.1 mg/ml MTT in DMEM was added to each well in a multiwell plate and incubated for 4 h. The MTT color complex was then solubilized in dimethyl sulfoxide (Me 2 SO), the solution transferred to a 96-well plate, and transmission evaluated at 570 nm. Data are expressed as percent of control (no oligo and no Cellfectin).
Western Blotting-Transfected or control cells were lysed in Lammelli SDS sample buffer. Proteins were separated by SDS-PAGE and transferred onto nylon membrane. Western blotting was performed according to the ECL protocol provided by the suppliers (Amersham Biosciences) using specific antibodies.
Knock Down of ATM in BALB/c 3T3 Cells with ATM siRNA-A pair of complementary RNA primers of 21 nucleotides (Dharmacon Research, Inc.) corresponding to the 5Ј-non-coding region of the ATM cDNA (5Ј-AAU GUC UUU GAG UAG UAU GdTdT-3Ј and 5Ј-CAU ACU ACU CAA AGA CAU UdTdT-3Ј) was annealed to form the ATM siRNA (a 19-nucleotide duplex stem with 2-nucleotide overhangs on either side) according to the instructions provided by the Dharmacon Research, Inc. 145 ng of annealed double-stranded ATM siRNA or sense strand of ATM RNA oligo was then complexed with 10 g of Cellfectin.
The RNA:Cellfectin complex was used to transfect BALB/c 3T3 fibroblasts as described above. After 4 h, cells were replenished with fresh DMEM containing 10% bovine serum and further incubated for another 24 h. Cells were then transfected again with indicated oligo:Cellfectin complexes. After 4 -6 h, nuclear DNA damage or cytochrome c release was determined by comet assay or immunocytochemistry, respectively. DNA could induce apoptosis, apoptotic nuclease-mediated DNA cleavage in the chromosomal DNA was monitored by an alkaline SCGE assay. Different single-stranded oligos were complexed with Cellfectin and used to transfect BALB/c 3T3 cells. Transfected cells were then analyzed for apoptotic nucleasemediated DNA strand breaks in genomic DNA using the alkaline SCGE assay. As shown in Fig. 1A, oligos with different sequences and sizes (from 5 to 36 nucleotides) were found to induce genomic DNA strand breaks. The deoxyribonucleoside triphosphates (dNTPs), on the other hand, were unable to induce genomic DNA strand breaks above the background level (Fig. 1A). The THIONT5, which is a 5-mer oligo with phosphorothioate linkages, was found to be as active as other oligos in inducing DNA strand breaks (Fig. 1A). We have also tested different transfection agents, including Oligofectamine, Cellfectin, Lipofectamin, and FuGENE, all of which are effective in mediating the effect of oligos. These results suggest that singlestranded oligos can indeed activate a nuclease (see below for the involvement of an apoptotic nuclease) in treated cells.

Single
The oligo-induced chromosomal DNA cleavage showed dose dependence from 0 to 200 nM (Fig. 1B). A maximum of 60% of cells was shown to exhibit DNA cleavage (Fig. 1A). Above 200 nM, the oligo was unable to increase the level of DNA cleavage. The oligo-induced chromosomal DNA cleavage were found to be time-dependent (Fig. 1C). DNA strand breaks were detectable within 1 h of treatment and found to increase with time (Fig.   1C). The oligo-induced DNA strand breaks were shown to depend on the single-strandedness of the oligo. As shown in Fig.  1D, the double-stranded oligo was much less effective than the single-stranded oligo in inducing DNA strand breaks. However, at higher concentrations of double-stranded oligo, DNA cleavage was also detectable (data not shown).
We have studied a possible involvement of apoptotic nuclease in oligo-induced cleavage in the genomic DNA. Treatment of BALB/c 3T3 cells with the caspase 3 inhibitor Z-DEVD-FMK was found to inhibit oligo-induced cleavage in the genomic DNA ( Fig. 2A), suggesting that a caspase-activated apoptotic nuclease(s) is involved in oligo-induced chromosomal DNA cleavage.
The fact that oligos can induce an apoptotic nuclease in treated cells has prompted us to examine whether oligos can induce (apoptotic) cell death. As shown in Fig. 2B, oligos indeed induced cell death in BALB/c 3T3 cells as evidenced by the MTT assay.
Oligos Induce Other Apoptotic Cell Death Markers-To test whether oligo-induced cell death is due to apoptosis, we have measured the expression of various apoptotic markers in oligotransfected cells. As shown in Fig. 3A, transfection of BALB/c 3T3 cells with different oligos caused release of cytochrome c from the mitochondria into the cytoplasm (Fig. 3A). Transfection of oligos also caused increased production of ROS as evidenced by increased fluorescence of oxidized DCFH (Fig. 3B). In addition, the NT8R oligo was shown to induce both chromatin condensation (Fig. 3C, upper panel) and caspase activation (Fig. 3C, lower panel) in BALB/c 3T3 cells. About equal numbers of cells (50%) were shown to undergo chromatin condensation and caspase activation (Fig. 3D). As a positive control, camptothecin, which is a topoisomerase I-specific inhibitor, was shown to induce chromatin condensation and caspase activation in about equal numbers of cells (80%). In the aggregate, these results strongly suggest that oligos can induce apoptotic cell death. Thus, it seems likely that oligo-induced DNA strand breaks, which are blocked by a caspase 3 inhibitor ( Fig. 2A), are due to activation of an apoptotic nuclease(s).
Oligos Induce Up-regulation of p53 and Activate ATM Kinase-Oligos could mimic a DNA damage signal(s) to trigger apoptosis. One of the most important cellular responses to DNA damage is up-regulation of p53 (29,30). We have thus tested whether oligos up-regulate p53 in BALB/c 3T3 cells. As shown in Fig. 4A, both oligos, NT36 and NT27, up-regulated p53 in BALB/c 3T3 cells.
Activation of the nuclear kinase ATM is another important cellular response to many DNA damaging agents (18,31). H2AX is a target protein of the ATM kinase and is known to be phosphorylated at Ser-139 by ATM (19). Using an antibody specific for the phosphorylated peptide epitope of H2AX, we have measured the amount of phosphorylated H2AX in BALB/c 3T3 cells treated with different oligos. As shown in Fig. 4B, NT36 and NT27 oligos stimulated phosphorylation of H2AX in BALB/c 3T3 cells. The increase in the amount of phosphorylated H2AX was not due to an overall elevation of histone H2A (Fig. 4B, lower panel). Phosphorylation of H2AX at Ser-139 suggests that the ATM kinase is activated by oligos. Our results thus suggest that oligos can mimic a DNA damage signal(s) to both up-regulate p53 and activate ATM kinase.
To study the role of ATM in SIA, ATM siRNA was employed to knock down the expression of ATM in BALB/c 3T3 cells (33,34). As shown in Fig. 6A, ATM siRNA significantly reduced (about 60%) SIA in BALB/c 3T3 cells as monitored by the alkaline SCGE assay. As a control, the corresponding ATM sense strand RNA did not affect SIA (Fig. 6A). Western blot analysis indicated that the ATM protein level was reduced more than 3-fold in ATM siRNA-treated, but not the corresponding ATM sense-strand RNA-treated, BALB/c 3T3 cells (Fig. 6A). We have also monitored the effect of ATM siRNA on cytochrome c release from mitochondria in BALB/c 3T3 cells treated with different oligos. As shown in Fig. 6B, oligo-induced release of cytochrome c from the mitochondria was greatly reduced in BALB/c 3T3 cells treated with ATM siRNA. These results suggest that oligos act upstream of ATM in SIA. Taken together, our results suggest that oligos may mimic a DNA damage signal(s) to induce ATM/p53-dependent apoptosis. DISCUSSION DNA damage is known to trigger a myriad of cellular responses including cell cycle arrest and apoptosis (35,36). While many key regulatory proteins (e.g. ATM and p53) have been identified, the immediate signal following DNA damage remains unclear (37). A number of studies have demonstrated that introduction of fragmented or single-stranded DNA into cells can influence the cell cycle and induce cell death, suggesting a potential role of DNA ends or single-stranded DNA in DNA damage signaling. For example, transfection with ran-domly fragmented herring sperm DNA induces apoptosis (38). Nuclear injection of linearized plasmid DNA, circular DNA with a large gap, or single-stranded circular phagemid induces a p53-dependent G 1 arrest (39). Transfection of a plasmid containing the telomeric sequence was found to activate p53 (40). Surprisingly, a CpG sequence has previously been reported to be required for induction of apoptosis in T lymphocyte cell lines (41).
Our studies have shown that single-stranded oligos are more effective than duplex forms in inducing apoptosis. In our system, all oligos tested induce apoptosis. In addition to the oligos shown in Fig. 1, we have tested over a dozen other oligos, all of which induce apoptosis. We did not observe a CpG sequence requirement. However, some oligos appear to be more effective than others in inducing apoptosis. Oligos containing the human telomeric G-tail sequence or a stretch of Gs are much more effective in inducing apoptosis. 2 Indeed, a recent report has shown that poly(G) motif-containing oligos can induce apoptosis in prostate cancer cells (42). The molecular basis for the observed sequence specificity is unclear and is currently under investigation.
Oligos have been used extensively in antisense research (43). SIA could potentially be explained by an antisense effect. However, two observations argue against the antisense explanation for apoptosis induction. First, we have shown that oligos as short as a 5-mer can induce apoptosis. It is unlikely that a 5-mer antisense oligo can form a stable RNA/DNA hybrid (44). Second, all sequences used in this study induce apoptosis. These sequences are not designed to be antisense oligos. It seems highly unlikely that all these sequences act as antisense oligos.
The oligos could potentially activate apoptosis by acting on cytoplasmic or mitochondrial apoptosis modulators. However, our studies have suggested that oligos act in the nucleus. First, oligos activate the nuclear kinase ATM as evidenced by phosphorylation of the ATM target protein H2AX in the nucleus. Second, like many DNA damaging agents, oligos up-regulate p53. Third, like DNA damage-induced apoptosis, SIA was found to be ATM-and p53-dependent. The similarity between SIA and DNA damage-induced apoptosis suggests that oligos may mimic a DNA damage signal in the nucleus to activate ATM/p53. Activated ATM/p53 signals are transduced to the mitochondria leading to the activation of caspase pathway and apoptotic cell death.
The molecular mechanism by which oligos activate ATM is unclear. One possibility is that oligos directly activate ATM kinase, which then triggers DNA damage responses. Consistent with this hypothesis, single-stranded DNA and linear DNA have been shown to activate replication protein A (RPA) phosphorylation by the ATM immunocomplex (45). However, it is unclear whether the role of single-stranded DNA is for direct ATM kinase activation or for binding to the ATM substrate RPA. Alternatively, oligos may act by titrating single-stranded DNA binding proteins (e.g. RPA) (46,47). Such a titration may cause deprotection of single-stranded regions in the genome (e.g. replication forks), which in turn could trigger DNA damage responses. Regardless of the mechanism, our results suggest that the single-stranded DNA must act upstream of ATM/ p53 in DNA damage signaling. As shown in Fig. 7, a schematic model for the role of single-stranded DNA in DNA damage and apoptotic signaling is presented. In this model, DNA damage is first converted into single-stranded DNA by a nuclease. The single-stranded DNA then activates ATM/p53-dependent DNA damage and apoptotic signaling. A double-strand break in the chromosomal DNA or a telomere end may not activate ATM/ p53-dependent DNA damage and apoptotic signaling unless the end is invaded by exonuclease/helicase or de-protected (e.g. inactivation of telomere protecting proteins) to produce singlestranded DNA. This is in contrast to the DNA-PK signaling pathway in which the ends of a DNA double-strand break may be sufficient to signal (48). Similarly, DNA adducts (e.g. UV and carcinogenic DNA adducts) may not signal through this pathway unless they are processed by nuclease/helicase to expose single-stranded DNA.  6. SIA is ATM-dependent. A, ATM siRNA was used to knock down ATM kinase in BALB/c 3T3 cells as described under "Experimental Procedures." The level of ATM in BALB/c 3T3 cells was determined by immunoblotting using ATM antibodies. Equal loading of protein was determined by immunoblotting using actin antibodies (Santa Cruz Biotechnology). B, treatment of ATM siRNA inhibits oligo-induced nuclear DNA strand breaks. Cells were transfected with either ATM sense RNA or ATM siRNA (145 ng). 24 h later, cells were again transfected with Cellfectin complexed with dNTPs (siRNA), NT36, or THIONT5 oligos. 4 h after transfection, cells were harvested and nuclear DNA strand breaks were monitored by alkaline SCGE assay. The ATM protein level in ATM siRNA-treated cells was determined by immunoblotting with anti-ATM antibodies (A). C, ATM siRNA abolishes oligo-induced cytochrome c release from mitochondria. BALB/c 3T3 cells were treated with ATM siRNA to knock down ATM expression as described in A. 24 h later, cells were transfected with different oligos (NT36 and NT27). 6 h after oligo transfection, cells were fixed and immunostained using cytochrome c antibodies followed by treatment with the FITC-conjugated anti-rabbit IgG secondary antibodies. Oligo-induced cytochrome c release from mitochondria was then examined using a fluorescent microscope. The release of cytochrome c (green fluorescence) is evidenced by transition from punctuated staining in the mitochondria (the micrograph labeled siRNA/NT36) to the smooth distribution of cytochrome c in the cytoplasm and punctated distribution in the nucleus (the micrograph labeled sense RNA/NT36). The red fluorescence is due to the DNA stain (propidium iodide). The number of cells exhibiting cytochrome c release was plotted for different treatment conditions. Transfection of oligos to induce DNA damage/apoptosis signals might provide a very useful system to study the molecular mechanism of apoptotic signal transduction pathways. The potency of oligos in inducing apoptosis also suggests their potential use as anticancer agents. Clearly, more studies are necessary to establish the molecular mechanism of SIA and the potential application of oligos in cancer therapy. It should be noted that oligos have already been successfully used to treat tumors in animals (49). However, the antitumor activity of the oligos has been attributed to the antisense effect (50). It remains to be determined whether SIA may play a role in the antitumor activity of oligos.