N6-methyladenosine regulates ATM expression and downstream signaling

N6-methyladenosine (m6A) is the most abundant modification in eukaryotic mRNAs, which plays an important role in regulating multiple biological processes. ATM is a major protein kinase that regulates the DNA damage response. Here, we identified that ATM is a m6A-modificated gene. METTL3 (a m6A “writer”) and FTO (a m6A “eraser”) oppositely regulated ATM expression and its downstream signaling. Mechanically, m6A “readers” YTHDFs and eIF3A suppressed ATM expression in the post-transcriptional levels. We also revealed the oncogenic potential of METTL3 and YTHDF1 related to ATM modulation. This is the first report that ATM, a master in the DNA damage response, is modified by m6A epigenetic modification, and METTL3 disrupts the ATM stability via m6A modification, thereby affecting the DNA-damage response.


Introduction
N6-methyladenosine (m6A) is the most prevalent and representative chemical modification in eukaryotic mRNAs as well as other non-coding RNAs, which are widely found in yeast, Arabidopsis, Drosophila, mammals, and viruses [1,2]. m6A modification system is composed of "writers", "erasers", and "readers", which allows the m6A modification becoming a dynamic and reversible regulatory process. METTL3 is the major catalytic subunit and was the first identified methyltransferase ("writer"), while METTL14, as another "writer", plays a crucial role in substrate recognition [3,4]. METTL3 and METTL14 proteins can form a stable heterodimeric core complex that promotes m6A deposition [5,6]. m6A "erasers" are demethyltransferases, which erase m6A modification and play reversible regulatory roles, mainly including FTO and ALKBH5 [7][8][9]. m6A "readers" specifically target m6A sites in a methylation-dependent manner, which mainly include YTHDF1-3, eIF3, nuclear protein YTHDC1-2, and other molecules [10][11][12]. Different "readers" bind to the m6A modified sites and function as RNA-binding proteins in many biological processes. YTHDF1 recognizes m6A-modified mRNA that results in decreased translation efficiency [13], whereas YTHDF2 reduces mRNA stability with recruitment of mRNA degradation systems [14]. YTHDF3 cooperates with YTHDF1 to promote protein synthesis and mediates the decay of methylated mRNA with the help from YTHDF2 [15].
ATM is a protein kinase, a major molecule that regulates the DNA damage response by phosphorylating a variety of key proteins in response to DNA double-strand breaks and DNA single-strand Ivyspring International Publisher breaks, thus playing a key role in DNA damage repair [16]. In higher eukaryotes, ATM kinase and ATR kinase are two important kinases governing the DNA damage response. ATM and ATR belong to the PI3K-like kinase (PIKK) family along with the DNA-dependent protein kinase catalytic subunits (DNA-PKcs) encoded by the PRKDC gene [17]. ATM is a component of multiple signaling pathways, including cell growth [18], metabolism [19], protein synthesis, and autophagy [20], which are involved in tumorigenesis.
For instance, ATM-mediated phosphorylation activates and stabilizes the tumor suppressor p53, which triggers cell cycle arrest by upregulating the cell cycle protein-dependent kinase inhibitor p21, resulting in induction of apoptosis [21].
Notably, a stressful m6A modification of intracellular RNA was found to occur following only 2 minutes of UV irradiation. This modification can occur on a variety of poly(A)+ transcripts and is regulated by METTL3 and FTO. UV light induces the formation of cyclobutane pyrimidine dimer adducts on the genome, and repair against this adduct is delayed when cells lack METTL3 catalytic activity [22]. This study demonstrates the importance of m6A modifications in the DNA damage response. However, the presence of m6A modification in ATM mRNA, and its corresponding regulatory mechanisms have not been reported.
Previous studies [23][24][25] including ours [26] have identified the effect of Epstein-Barr Virus (EBV) infection on the level of transcriptomic m6A modifications in host cells. Upon EBV infection, we identified increased levels of m6A modification of ATM mRNA. In the current study we revealed a potential role of m6A modification in ATM expression and the downstream signaling in the DNA damage response. Our study illustrates a novel regulatory mechanism of DNA damage response through m6A modification of the ATM mRNA.

Cell transfection, plasmids and RNA interference
Cells were cultured to a predetermined time. The siRNAs and plasmids were transfected using Lipofectamine 3000 (Invitrogen) according to the manufacturer's instruction. DNA fragments encoding full-length ORF of YTHDF1, YTHDF2, or YTHDF3 were cloned into a Flag-tagged pcDNA3.1 (+) empty vector. shRNA lentiviruses and siRNA (targeting METTL3, YTHDF1, YTHDF2, or YTHDF3) were obtained from GenePharma (Shanghai, China). Lentiviruses were transduced into cells according to the manufacturer's instructions. The siRNA and shRNA sequences are listed in Table S1.

MeRIP-qPCR
According to the manufacturer's protocol, intact poly-A-purified RNA was isolated using Magnetic mRNA Isolation Kit (S1550S, New England Biolabs). With reverse transcription with Oligo(dT) and PCR amplification, mRNA incubated with 5 μg m6A antibody (202003; Synaptic Systems) in IP buffer (PH7.4) containing RNase inhibitor (N2111SV, Promega), 150 mM NaCl, 10 mM Tris-HCl and 0.1%NP-40 for 2 h at 4 °C. Protein A/G magnetic beads (823202, Selleck) were washed, incubated with mixture buffer for 2 h at 4 °C with rotation. m6A RNA was eluted with 6.7 mM m6A sodium salt (Santa Cruz Biotechnology) and precipitated with 100% ethanol. Enrichment of m6A was detected through qRT-PCR. Normal rabbit IgG were designed as negative control in this experiment group.

RNA immunoprecipitation (RIP) assays
Briefly, Cells were transfected with Flag-YTHDF1 or Flag-YTHDF2 or Flag-YTHDF3 plasmids, respectively, for 48 h. Then cells were harvested in RIP lysis buffer including RNase inhibitor (N2111SV, Promega), protease inhibitor and phosphatase inhibitor (b14001; b15001, Selleck). The supernatants were retained and mixed with anti-Flag antibody (F1804, Sigma-Aldrich) or normal mouse anti-IgG (sc-2025; Santa Cruz Biotechnology) for 12 h at 4 °C after centrifugation with 12000 rpm at 4 °C. Protein A/G magnetic beads were added to the mixture and incubated for 4 h at 4 °C with rotation. The beads were then washed with RIP wash buffer. Total RNA was isolated and proved to qRT-PCR analysis.

RNA isolation and reverse transcription (RT)-PCR assay
Total RNAs from cells were isolated using TRIzol reagent (Invitrogen). 2 μg RNA was reverse transcribed into synthesize complementary DNA (cDNA) using RevertAid RT Reverse Transcription Kit (Thermo Scientific). Real-time reversetranscription PCR was carried out by SYBR premix Ex TaqII Kit (Takara). Actin was performed in parallel as a control. Each mRNA expression was quantified by measuring cycle threshold (Ct) values. The relative expressions were calculated using the 2 -ΔΔCt method. The primers for qRT-PCR are listed in Table S2.

Cell cycle assay
SGC7901 cells were transfected with siRNA and incubated until a predetermined time. Wash the cells twice with pre-chilled PBS, collect the cells by centrifugation and discard the supernatant, add 50 μL of Binding Buffer and gently resuspend the cell precipitate. Add 1 mL of 70% pre-cooled ethanol to fix the cells and place the cells in the rotator overnight at 4 °C. Centrifuge at 400 g for 5 min at 4 °C, wash the cells twice with PBS, centrifuge, and discard the PBS. Add staining solution and stain the cells at 37 °C for 30 min under low light and perform flow cytometry assay. The whole procedure should be as light as possible to avoid cell fragmentation.

Co-Immunoprecipitation (Co-IP) assay
Cells were transfected with indicated plasmids and cultured to a predetermined time, cells were lysed on ice with IP lysis solution, cell supernatant was collected and mixed with anti-Flag antibody or anti-IgG antibody (negative control), and rotated overnight at 4 °C. Protein A/G magnetic beads were added, and the mix was incubated for 2 h at 4 °C. The magnetic beads were adsorbed on a magnetic frame, the supernatant was discarded, and Western blot experiments were performed.

Immunofluorescence assay
Cells were transfected with siRNAs for 48 h, and then were fixed in medium containing 4% paraformaldehyde for 1 h and then permeabilized using 0.5% Triton X-100 and blocked using normal goat serum. The primary antibodies (anti-p-ATM, or anti-γH2A.X) were added and incubated at room temperature for 2 h. Secondary antibody (Cy3 conjugated; red) (Beyotime) were added and incubated for 1 h. Stained cells were examined using a fluorescence microscope.

RNA stability assay
Cells were added with actinomycin-D (Act-D, Meilunbio) at 5 μg/ml to measure RNA stability in cells with siRNA or shRNA pre-treatment. Cells were harvested at an indicated time points and extracted RNA for reverse transcription. The mRNA transcript levels were normalized to 0 h of actinomycin-D treatment by RT-qPCR.

CCK8 cell proliferation assay
Cells were transfected with siRNA culturing in 96-well plates (1000 cells per well). Cells were cultured until a predetermined time in the presence or absence of the ATM inhibitor KU55933 (10 μM). 10 μL CCK8 reagents (CK04, Dojindo) were added in 96-well plates incubating for 4 h at 37 °C. Absorbance was obtained at 450 nm by a microplate reader (ThermoFisher).

Statistical analysis
Statistical significance was calculated using Prism (GraphPad Software) and SPSS17. All experiments were performed in triplicate. Data represent the mean ± SD. Statistical differences were assessed with the unpaired Student t-test, and p <0.05 were considered to reflect statistical significance. A one-way analysis of variance test was performed for comparing three or more groups within the same experiment.

Identification of ATM as an m6A-modificated gene
From our previous results [26], we found that the m6A abundance was increased in ATM mRNA transcripts upon EBV infection ( Figure 1A). ATM is a key kinase in DNA damage response, which is responsible for detecting, transmitting and repairing DNA damage response [21]. However, the role of the m6A modification involved in ATM signaling is not yet fully understood. SRAMP (sequence-based RNA adenosine methylation site predictor) [27], a bioinformatical tool to predict the m6A site of mRNAs, has predicted five high confidence m6A sites in ATM mRNA sequences ( Figure 1B). MeRIP-RT-qRCR assay was performed to detect enriched ATM mRNAs after anti-m6A immunoprecipitation. ATM mRNAs were specifically enriched by the anti-m6A antibody in cells ( Figure 1C). These results suggested that ATM gene is m6A modified in cells.

METTL3 and FTO modulate ATM expression
is a general methylation inhibitor, and we found that it can substantially increase the ATM mRNA and protein levels in cells (Figure 2A), suggesting methylation levels contributed to ATM expression regulation. Since METTL3 is the main m6A "writer", we used shRNA to knockdown METTL3 expression in cells, and found that silence of METTL3 resulted in significant increase in ATM expression ( Figure 2B, C). sh-METTL3 delayed the degradation of ATM mRNA ( Figure 2D), indicating that METTL3 inhibited ATM expression and decreased its mRNA stability. FTO is a demethylase for m6A modification, and eliminates m6A modifications on mRNAs [28]. Inhibition of endogenous FTO significantly reduced ATM expression levels ( Figure 2E, G). However, inhibition of FTO did not affect ATM mRNA stability ( Figure  2F). In addition, we examined whether the downstream signaling of ATM were regulated by FTO. Western blot analysis revealed that inhibition of FTO decreased the expression levels of ATM, CHK2, and RBBP8 ( Figure 2G). These results suggested METTL3 ("writer") and FTO ("eraser") could regulate ATM expression and its downstream signaling.

METTL3 promotes cell cycle progression, cell proliferation and involves in the DNA damage response
Previous studies have identified that METTL3 facilitated tumor progression via an m6A-dependent mechanism [29,30]. To further verify the oncogenic role of METTL3, we inhibited METTL3 in gastric cancer cells SGC7901 to detect its effects on cell cycle and cell proliferation. Knockdown of METTL3 resulted in an increase in G2/M phase ( Figure 3A). Cyclin B2 contributes to the G2/M phase transition by activating CDK1 kinase, and cyclin B2 inhibition induces cell cycle arrest [31]. We found that inhibition of endogenous METTL3 significantly reduced the expression levels of CDK1 and cyclin B2, implying that G2/M phase block may be attributed to the downregulation of CDK1 and cyclin B2 ( Figure 3B). Inhibition of METTL3 also decreased tumor cell proliferation by means of CCK8 assay ( Figure 3C). Next, we pre-treated cells with specific ATM inhibitor KU55933 [32,33], and CCK8 assay revealed that KU55933 significantly reduced cell proliferation. In addition, inhibition of METTL3 in combination with KU55933 significantly inhibited cell proliferation, comparing to sh-METTL3 and KU55933 only group ( Figure 3C). These results suggested that simultaneous inhibition of METTL3 and ATM could be used as a therapeutic strategy to inhibit tumor cells proliferation.
Next, we investigated the effect of METTL3 on the ATM downstream signaling. Expression levels of ATM, p-ATM, RBBP8, BRCA1, H2A.X and p53 were significantly increased, whereas the levels of γH2A.X (reflecting DNA damage signal) was significantly decreased after the inhibition of METTL3 expression without etoposide treatment ( Figure 3D, lane 3 vs lane 1). Following the addition of etoposide (to induce a DNA damage response in cells [34]), the expression levels of ATM, p-ATM, RBBP8, BRCA1, p53 and H2A.X were increased ( Figure 3D, lane 2 vs lane 1). Overall, the DNA damage response was stronger in sh-NC group comparing to sh-METTL3 group, as reflected by the more significant increase in p-ATM, RBBP8, BRCA1, and p53 ( Figure 3D, lane 2/1 vs lane 4/3). We also examined the effects of sh-METTL3 on p-ATM (S1981) and γH2A.X levels under the etoposide treatment by immunofluorescence. sh-METTL3 increased p-ATM, whereas decreased γH2A.X levels ( Figure 3E). In other words, when the m6A modification system is compromised (for instance, METTL3 inhibition), the DNA damage response is becoming "weaker", and the DNA damage sites are more abundant in cells.

YTHDF1 or YTHDF2 binds to ATM mRNA and regulates its downstream signaling
The m6A "readers" participate in posttranscriptional regulation by binding directly to the m6A modification sites of target mRNAs. m6A "readers" YTH family proteins are mainly distributed in the cytoplasm, suggesting that their functions may be related to RNA in the cytoplasm and play a role in the translation and degradation of RNA [35][36][37]. To investigate the distribution of ATM protein and m6A reader proteins during the DNA damage, we irradiated SGC7901 gastric cancer cells with ionizing radiation (4Gy) for 30 min, and found that ionizing radiation-induced DNA damage resulted in increased expression of ATM, p-ATM, and H2A.X. YTHDF1, YTHDF2 and YTHDF3 proteins were mainly localized in the cytoplasm. The nuclear localizations of H2A.X, p-ATM, and YTHDF2 were increased upon ionizing radiation ( Figure 4A). We used siRNAs to suppress the expression of endogenous YTHDF1 or YTHDF2, and noticed that the expression levels of ATM and p-ATM proteins were significantly increased. However, inhibition of YTHDF3 decreased the expression levels of ATM and p-ATM proteins ( Figure  4B). Inhibition of YTHDF1 or YTHDF2 did not alter the ATM mRNA expression levels as detected by RT-qPCR, whereas si-YTHDF3 decreased ATM mRNA levels ( Figure 4C).
To explore the mechanism of how m6A modifications regulate ATM expression, we hypothesized that m6A readers YTHDF proteins could bind to ATM mRNA and regulate mRNA translation. We transfected Flag-tagged YTHDF1, YTHDF2, and YTHDF3 expression vectors into cells, and found that ATM mRNA was enriched with Flag-YTHDF1 and Flag-YTHDF2, but not Flag-YTHDF3, suggesting that YTHDF1 and YTHDF2 could bind to ATM mRNA ( Figure 4D). We further investigated whether YTHDF1 or YTHDF2 could affect the expression of downstream molecules of the ATM signaling. Eexpression of YTHDF1 inhibited the expression of ATM, 53BP1, BRCA1, RBBP8, γH2A.X, Chk1, and p-Chk2 ( Figure 4E). Meanwhile, YTHDF2 reduced the expression levels of ATM, 53BP1, BRCA1, RBBP8, γH2A.X, but not the CHK1 and p-CHK2 ( Figure 4F). In addition, immunofluorescence experiments confirmed that inhibition of YTHDF1 and YTHDF2 upregulated p-ATM expressions ( Figure  4G). These results suggested that YTHDF1 and YTHDF2 are involved in the regulation of ATM downstream signaling.  and RT-qPCR (C) were used to assay the indicated molecules. (D) RIP-qPCR. SGC7901 cells were transfected with Flag-tagged YTHDF1, YTHDF2, or YTHDF3 expression plasmids for 48 h. Cell lysates were immuoprecipitated with anti-Flag beads, and then RNAs were isolated from the immunoprecipitates, and used for RT-qPCR to detect ATM mRNA levels. Mean ± SD of three independent experiments. Two-tailed unpaired t test. *p < 0.05, **p < 0.01. (E and F) SGC7901 cells were transfected with the YTHDF1 or YTHDF2 expression vectors for 48 h, respectively, and the cellular proteins were isolated for Western blot assay. (G) SGC7901 cells were transfected with siRNAs (si-YTHDF1 or siYTHDF2) for 48 h and immunofluorescence assay was performed to detect the fluorescence expression of p-ATM (Ser1981). Scale bar: 25 μm. ****p < 0.0001.

YTHDF1 suppresses ATM expression in the post-transcriptional levels
We further explored how YTHDFs influence ATM in the post-transcriptional levels. We used Cycloheximide (CHX, 25 μg/ml) to prevent protein synthesis and examined the stability of ATM protein. Both YTHDF1 and YTHDF2 significantly reduced the stability of ATM protein ( Figure 5A). We co-transfected the Flag-YTHDF1 and HA-Ubiquitin plasmids into cells and found that overexpression of YTHDF1 increased ubiquitination levels of ATM proteins ( Figure 5B). eIF3A is also a m6A "reader", and study has suggested that YTHDF1 regulates the translation process by interacting with eIF3A or eIF3B [38]. Co-IP analysis showed that YTHDF1 is interacting with eIF3A protein (Figure 5C), and we speculate that YTHDF1 binds to ATM mRNA while recruiting eIF3A to regulate ATM mRNA translation. Inhibition of endogenous eIF3A with siRNA increased the expression levels of ATM proteins and is involved in ATM downstream signaling ( Figure 5D, E). Immunofluorescence confirmed that inhibition of eIF3A upregulated p-ATM and γH2A.X levels ( Figure  5F). These results suggested m6A readers YTHDF1/ eIF3A could decrease ATM protein stability. In addition, we also found that inhibition of YTHDF1 decreased the G0/G1 phase and increased S-phase in cancer cells (Supplementary Figure 1A). p21 is known to inhibit the kinase activity of cyclin A/CDK1, 2, leading to cell cycle inhibition into S phase [39,40] Mechanistically, we showed that inhibition of YTHDF1 significantly decreased p21 expression, and increased CDK1, CDK2 expressions, whereas cyclin B2 had no change (Supplementary Figure 1B). Inhibition of YTHDF1 suppressed tumor cell proliferation. In addition, inhibition of YTHDF1 in combination with ATM inhibitor KU55933 significantly inhibited cell proliferation, comparing to si-YTHDF1 and KU55933 only group (Supplementary Figure 1C). These results suggested YTHDF1 promotes cancer cell proliferation and cell cycle.

Discussion
RNA m6A modifications play important roles in a variety of biological processes. Xiang et al. revealed a novel role for RNA m6A modification in promoting cellular resistance to UV damage, and identified a potential new pathway involving METTL3, FTO, m6A RNA modification, and Pol κ in the early UV-induced DNA damage response [22]. However, the regulatory mechanisms of m6A modification on the ATM signaling are still unclear. In the current study, we mainly focused on the m6A modifications of ATM and the impact on ATM downstream signaling.
ATM kinase is a major regulator of the DNA damage response, and the activation of ATM upon DNA damage involves multiple cascade reactions, including acetylation [41], UFMylation [42] and destabilization of linker histone H1.2 [32]. In this study, we confirmed the presence of m6A modifications on ATM mRNA by MeRIP. Inhibition of METTL3 (an m6A "writer") upregulated ATM expression levels and increased ATM mRNA stability. Meanwhile, inhibition of FTO (an m6A "eraser") downregulated ATM expression levels, suggesting that m6A modifications could dynamically regulate ATM expression levels ( Figure 6).
METTL3 is an important component of the m6A methyltransferase complex, and it also acts as an oncogene in the process of tumorigenesis [43]. Luo et al. previously reported that silencing METTL3 significantly suppressed melanoma cell proliferation and colony formation [44]. Wang et al. demonstrated that METTL3 promoted gastric cancer cell proliferation and metastasis [30]. Yang et al. found that METTL3 promotes the progression of gastric cancer via targeting the MYC pathway [45]. Consistent with these results, we showed that METTL3 accelerate cell cycle progression and cell proliferation in gastric cancer cells; and inhibition of METTL3 in combination with specific ATM inhibitor KU55933 further inhibited cancer cell proliferation, comparing to sh-METTL3 and KU55933 only group. In addition, we found that METTL3 also modulated the ATM downstream signaling: when the expression of METTL3 was inhibited, the ATM signaling (the DNA damage response) was less evident, and the DNA damage sites were more abundant, indicating an important role of m6A modification in modulating ATM signaling.
As m6A "readers", YTH family proteins serve as important binding proteins for m6A modifications to regulate RNA metabolism, including splicing, translation, and degradation [37,38]. We used RIP assay to reveal that YTHDF1 and YTHDF2 could bind to ATM mRNA. Zhuang et al. had demonstrated that YTHDF1 regulated the translation of m6A-modified Robo3.1 mRNA. Knockdown or mutation of YTHDF1 resulted in a dramatic reduction of Robo3.1 protein without affecting Robo3.1 mRNA level [46]. Consistent with their study, we found that YTHDF1 or YTHDF2 increased ATM protein degradation but did not affect ATM mRNA expressions. We further confirmed that overexpression of YTHDF1 promoted ATM degradation via ubiquitination. Additionally, YTHDF1 is involved in regulating the mRNA translation by interacting with eIF3A and eIF3B translation initiation factors [15,38]. Our co-IP experiments revealed that YTHDF1 interacted with eIF3A. Silencing of eIF3A could increase ATM protein levels. We speculated that YTHDF1 recruits eIF3A to promoting ATM protein degradation. However, the specific mechanism of how YTHDF1 recruits eIF3A to regulate ATM expression has yet to be determined in the future.
In summary, we have revealed for the first time the presence of m6A modifications in ATM mRNAs. m6A modification regulates ATM mRNA metabolism and ATM downstream signaling, which illustrates the importance of m6A modification-related molecules for being used as therapeutic targets in DNA damage-related diseases.