ATM Knockdown renders Glioma resistant to Temozolomide by Inhibiting IFIT1 Expression


 Background: Temozolomide (TMZ) chemotherapy has been a standard of care in treating malignant glioma. Although TMZ chemotherapy can extend patient’s survival, resistance to TMZ is observed in most cases. The drug resistance is reported to be mainly mediated by O6- methyl guanine-DNA methyltransferase (MGMT) expression.Methods: By means of molecular biology, cell biology, construction of nude mouse xenograft tumor model and analysis of clinical specimens, a new molecular mechanism for the regulation of TMZ chemotherapy resistance of glioma was proposed.Result: The present study indicates a pathway of TMZ resistance in glioma via suppressing ataxia-telangiectasia mutated (ATM) gene. Whole genome expression profile of glioma cells and tissue samples revealed a positive correlation between the ATM and IF1T1, the decreased expression of which might be the underlying cause of the ATM knockdown induced TMZ resistance.Conclusions: Our study revealed that ATM silencing induced TMZ chemotherapy resistance of glioma in vitro and in vivo by inhibite IFIT1 expression, the p-ATM, IFIT1 and MGMT expression acts as a prognostic marker of glioma chemotherapy. And combination of IFIT1 and MGMT seems to be a more significant molecular marker to predict the prognosis in glioma patients.


Introduction
Malignant glioma is a progressive brain tumor with high morbidity and mortality. In recent times, a multidisciplinary approach of treating gliomas using surgery and radiotherapy coupled with chemotherapy, predominantly with temozolomide (TMZ) [1,2], has been a used to improve the survival and quality of life. Although TMZ chemotherapy can extend patient's survival, resistance to TMZ is observed in most cases, which leads to poor prognosis of glioma [3,4]. The drug resistance is reported to be mainly mediated by O6-methyl guanine DNA methyltransferase (MGMT), generated by TMZ [5]. The activation of MGMT depends on the occurrence of DNA damage response (DDR), where mutation of ataxia-telangiectasia mutated (ATM) gene is the key regulator of the process [6]. The double-stranded DNA damage caused by cell cycle arrest, apoptosis and DNA damage repair play an important role in signal transduction pathways [7,8]. ATM is a mutant gene in ataxia-telangiectasia (AT), located on human chromosome 11q22 ~ 23 [9] with a total length of about 150 kb, encoding sequence of about 12 kb [10], and is involved in regulating cell cycle, repairing damaged DNA and maintaining telomere length. ATM is found to promote cell proliferation through protein kinase B (PKB/Akt) pathway [11,12].
Understanding several mechanism of TMZ resistance and strategies to overcome it to enhance the effect of TMZ for a successful treatment e cacy of glioma is of utmost importance. Thus, in view of this hypothesis, the present study aims to explore the molecular mechanism of glioma TMZ chemotherapy resistance induced by ATM silencing. This was carried out using human genome-wide screening, chips and HCS system on glioma U251 cells. We found that ATM knockdown increase the resistance of TMZ, by inhibiting the expression of IFIT1. ATM promote cell apoptosis by inducing the expression of IFIT1, improve the sensitivity of glioma TMZ chemotherapy, and improve the clinical prognosis of glioma patients.

Materials And Methods
Glioma tissues, Cell culture and stable cell line construction. This study protocol was reviewed and approved by the hospital ethics committee, Sanbo Brain Hospital of Capital Medical University. Post consent, 70 glioblastoma tissues, from the patients who underwent surgical resection of glioblastoma, were obtained from the department of clinical surgery. Each sample was cut into half, with one half stored in 4% paraformaldehyde and the other half was frozen in liquid nitrogen and stored at − 80 ℃ till further use. U251, U87, LN229, U373 and 293T cell lines were purchased from Shanghai cell institute. HEB, HepG2, Hela, and MCF-7 cell lines were maintained in our laboratory. All cells were cultured in RPMI1640 or DMEM, supplemented with 10% FBS, 100U/mL penicillin, and 100 mg/mL streptomycin, and incubated at 37 ℃ under 5% CO 2 . Further, we constructed U251 and U87 stable ATM knockdown cell lines, through lentivirus (LV-siATM-KD1 and LV-siATM-KD2) infection; the cells are hereby referred to as siATM-KD1 and siATM-KD2, respectively, with control cells referred to as ATM-NC.
qRT-PCR. Total RNA from glioma cells and tissues were extracted by TRIzol reagent kit (TaKaRa, Japan), eluted by RNase free ddH 2 O and stored at − 80 ℃. The reverse transcription was then performed with TAKARA retrovirus kit (TAKARA, Japan) as follows: 37 ℃ for 1 h and 85 ℃ for 15 s. SYBR-green PCR Master Mix (TaKaRa, Japan) and primers (Table 1) were used for qRT-PCR, carried on bio-rad iQ5 system (bio-rad, USA), ATM or IFIT1 expression was quanti ed using the 2 −∆∆Cq method [13].

Result
High expression of ATM in human glioma cell lines and tissues. Expression of ATM was determined in HEB, the human normal glial cell line, and U87, U251, LN229 and U373, the human glioma cell lines, by qRT-PCR (Fig. 1A) and western blot (Fig. 1B). All glioma cell lines expressed signi cantly higher levels of ATM as compared to HEB (P < 0.001) ( Fig. 1A&B). Further, the expression of ATM in glioma tissue and its adjacent normal tissue was also detected by qRT-PCR and western blot. Following the trend, the mRNA level of ATM was found signi cantly increased in glioma tissue as compared to its adjacent tissue (P < 0.001) (Fig. 1C&D).
ATM silencing induced TMZ chemotherapy resistance of glioma in vitro. U251-LV-ATM-KD and U87-LV-ATM-KD stable cell lines were constructed by lentivirus infection and the e ciency was veri ed by measuring the expression of ATM mRNA and protein level (Fig. 2A&B). These cells were used to investigate the role of ATM in TMZ chemoresistance. The 50% inhibitory concentration (IC 50 ) of TMZ for U87 and U251 cells, as determined by MTT assay, was found to be 53.8 µmol/ml and 113.9 µmol/ml for U87 and U251 cells, respectively (Fig. 2C&D). Further, we detected the effect of ATM knockdown on the resistance of U87 and U251 cells to TMZ chemotherapy. As shown in Fig. 2E, after ATM knockdown, the resistance of U87 and U251 cells to TMZ chemotherapy was signi cantly decreased, thus signi cantly reducing the percentage of cellular apoptosis in response to TMZ treatment (p < 0.01) ( Table 2). Cell cycle analysis of ATM knockdown U87 and U251 cells in response to TMZ (IC50) chemotherapy was done by owcytometry. Results revealed a signi cant decrease in percentage of G1 phase cells (P < 0.01) and increase in S phase cells (P < 0.05) in both the cell lines. Besides, a signi cantly higher percentage of U251 cells were found in G2/M phase as compared to U87 cells (P < 0.05) ( Table 3 and Fig. 2F). ATM knockdown signi cantly inhibited the apoptosis of U87 and U251 cells in response to TMZ (IC50) chemotherapy (p < 0.01) (Fig. 2G). The above results suggests that the ATM knockdown in glioma cells may promote their DNA synthesis and cell division, while inhibiting apoptosis, in response to TMZ chemotherapy. ATM silencing induced TMZ chemotherapy resistance of glioma, in vivo. A subcutaneous xenograft tumor model of ATM knockdown glioma cells was constructed in nude mice, to further study the molecular mechanism in vivo. ATM knockdown could signi cantly inhibit the tumor growth rate in nude mice, as compared to control group ( Fig. 3A-C). Further, in response to TMZ chemotherapy, 3 days after stopping the treatment (the 32nd day after injecting cells), an increase in tumor growth was observed. Tumor volume in the control group decreased to the lowest level on the 10th day (the 39th day after injecting cells) after stopping the chemotherapy, but started to increase from the 13th day (the 42nd day after injecting cells). However, nude mice in ATM knockdown group showed little response to TMZ chemotherapy, demonstrated by continued proliferation of the tumor reaching a signi cantly increase in volume on 10th and 13th day (the 39th day and 42nd day after injecting cells) after stopping TMZ chemotherapy, as compared to the control group (p < 0.05), (Fig. 3D).
Screening for ATM knockdown-induced TMZ chemotherapy-resistant genes in glioma. In order to further explore the molecular mechanism of ATM knockdown induced glioma resistance to TMZ chemotherapy, we used human genomewide oligonucleotide microarray to detect the changes of gene expression pro le after ATM knockout in U251 cells.
Further, using the High Content Screening Cellomics system, the role of differentially expressed genes in U251 cells in response to TMZ chemotherapy and the target genes related to the chemotherapy-resistance caused by ATM knockdown, especially those capable of inducing apoptosis and improving chemotherapy-sensitivity, were screened out.
As shown in Fig. 4A, B, C, D, and E, the quality of RNA samples used for gene chip detection and the coincidence degree of each signal distribution curve of the chip data was good, indicating the reliability of the experiment. The interchip correlation coe cients in each group were very close to 1.0, indicating that the data had good repeatability and reliability, and could be used for the screening of downstream differentially expressed genes. The screening criteria for genes with signi cant differences must be in line with |FC| >1.5, and P < 0.05, and according to this standard, 324 upregulated genes and 402 down-regulated genes were screened out. The two factors of p value and FC value obtained by T test analysis were used to draw the volcano map that highlighted the signi cant difference between the two groups of sample data. In the volcano map, the y-coordinate shows the -lg p value calculated by T-test, and the x-coordinate shows the FC value after log2 conversion. In the upper left red region, down-regulated genes were expressed, while in the upper right red region, up-regulated genes were expressed (Fig. 4F). Cluster analysis was performed on all differentially expressed genes (Fig. 4G ). In the gure, there was a gene in each row and a specimen in each column. Different colors represented the expression levels of different genes; red represents up-regulation and green represents down-regulation of gene expression, while black represents no difference in gene expression. The expression pro les of the three samples in the U251-NC and U251ATM-KD groups were highly consistent, and the differences between the samples in the U251-NC and U251ATM-KD groups could be clearly distinguished by the color gradient change, indicating the high reliability of these differentially expressed genes screened. We also conducted pathway, molecular function, biological process and cellular component GO analysis of differentially expressed genes after ATM knockdown, sort by p value. Pathway analysis showed the possibility of high correlation of MAPK and tumor-related pathways with this study (Fig. 4H, Table 4). Molecular function GO analysis showed that ATM is mainly related to the function of DNA binding, transcription-related factor activity, transcription factor binding, receptor binding, G protein and small G protein regulatory active molecules (Fig. 4I, Table 5). Biological process GO analysis showed that ATM is mainly involved in biological processes, including intracellular signal transduction, tissue and organ development, negative regulation of cellular biological processes, and gene transcription (Fig. 4J, Table 6). Cellular component GO analysis showed that ATM knockdown induced differentially expressed genes were composed of cellular components, including cytoplasmic proteins and membrane related proteins (Fig. 4K, Table 7).    Screening and identi cation of genes related to ATM knockdown induced TMZ chemotherapy resistance of glioma. After ATM knockdown, differentially expressed genes were screened to enter the downstream test according to the following criteria: differentially expressed new genes with p < 0.05 and |FC|>1.5, whose functions were not fully de ned, remove multiple transmembrane protein genes, and remove genes with unknown functions. For the 402 down-regulated genes, 19 genes were preliminarily screened by the screening criteria, and the relative expressions of 19 candidate genes in U251 cells were detected by qRT-PCR (Fig. 5A). In order to identify the role of 19 candidate genes in glioma TMZ chemotherapy, RNA interference of multiple targets was performed on each of them, and stable lentivirus knockdown cell lines were constructed. CDC23, RAC1 and CD164, which were less than 70% of the uorescence e ciency after lentivirus infection, were not considered further and a total of 16 genes were nally detected downstream. High content screening Cellomics system (HCS) was used to detect the effect of candidate gene RNA interference on TMZ treated U251 cell, the cell proliferation, chemotherapy proliferation rate and chemotherapy inhibition rate after 5 days of chemotherapy (Table 8); U251-NC was used as negative control and U251-ATM-KD-1 was used as positive control.
Among the 16 candidate genes, 14 had lower chemotherapy inhibition rate than the negative control (40.05%). Among and NME4 respectively, the proliferation of U251 cells in TMZ (IC50) were all higher than those in U251-NC group. Of these, IFIT1 showed the highest response, which was equivalent to the positive control U251-ATM-KD-1. Therefore, in order to reveal the molecular mechanism of ATM knockdown induced resistance to TMZ chemotherapy of glioma, we choose IFIT1 for in-depth study. revealed an increased expression in all the tested glioma cell lines, as compared to HEB cell line ( Fig. 6A and B). Then, with lentivirus interference, we constructed U251-IFIT1-KD and U87-IFIT1-KD stable cell lines ( Fig. 6C and D). Both regular and knockdown versions of U251 cells were treated with TMZ for 5 days and then assessed for changes in Cell proliferation, clone formation, cell cycle and apoptosis. MTT assay was used to detect the changes in cell proliferation, and as shown in Fig. 6E and F, the proliferation of U251-IFIT1-KD cells was higher than that of the control U251-NC cells.
On the 5th day of TMZ treatment, the inhibition rate of chemotherapy on U251-IFIT1-KD group (15.3 ± 3.2 %) was signi cantly lower than that of the negative control group (29.3 ± 6.6 %), (p < 0.05). Meanwhile, a similar suppression in the inhibition rate of TMZ chemotherapy was also seen in U87 cells (Fig. 6G, P < 0.05). Further, the number of clonal formation of U251-IFIT1-KD + TMZ (IC50) and U87-IFIT1-KD + TMZ (IC50) group was signi cantly less than their respective control cells, (Fig. 6H, p < 0.05). FACS method was used to detect cell cycle stage and the results showed that when IFIT1 was knockdown, cells at G1 phase and S phase increased signi cantly (p < 0.001), while proportionally decreasing cells at G2/M phase. This indicates that the IFIT1 knockdown can promote DNA synthesis and cellular proliferation while inducing resistance to TMZ chemotherapy in U251 and U87 cells (Fig. 6I and Table 9). Annexin v-apc single staining apoptosis analysis showed that the rate of apoptosis of U251/U87-IFIT1-KD + TMZ (IC50) group (6.61 ± 0.28%) was signi cantly lower than that of the control group (7.43 ± 0.41%) (Fig. 6J, p < 0.001). The above results showed that IFIT1 knockdown can signi cantly promote the proliferation of glioma cells, while inhibiting their apoptosis, thus imparting resistance in glioma cells to TMZ chemotherapy. Thus, IFIT1 may serve as a molecular marker indicating the sensitivity of glioma to TMZ chemotherapy. Expression of p-ATM, IFIT1 and MGMT in glioblastoma tissues. Immunohistochemical staining was used to evaluate the expression of p-ATM, IFIT1 and MGMT in tissues obtained from 70 cases of glioblastoma. p-ATM was expressed in the nucleus and/or cytoplasm, and the median expression was 0 % (range 0-30%), including 55 cases with low expression (Fig. 7A) and 15 cases with high expression (Fig. 7B), IFIT1 was primarily expressed in the cytoplasm, with a median expression of 24% (range 0-65%), including 13 cases with low expression (Fig. 7C) and 57 cases with high expression (Fig. 7D). Further, with 55 cases of low expression and 15 cases of high expression, the median expression of MGMT was found to be 10% (range 0-50%). In addition, the expression of IFIT1 was observed in endothelial cells of certain specimens (Fig. 7E), tumor tissue necrosis area (Fig. 7F), edematous area (Fig. 7G), and around vascular endothelial cells (Fig. 7H). Spearman's analysis revealed a positive correlation between p-ATM and IFIT1 expression (r = 0.249, p < 0.05), and a negative correlation between IFIT1 and MGMT expression (r=-0.288, p < 0.05) ( Table 10). The relationship between the clinical characteristics and expression of p-ATM, IFIT1 and MGMT. Mann-whitney test found no correlation between p-ATM expression and the clinical characteristics, including gender, age and KPS (Table 11). In order to analyze the relationship between the prognosis and expression of p-ATM, we used progression free survival (PFS) and overall survival (OS) as evaluation indicators, by following up 70 patients for a period of  weeks, median follow-up time was 45 weeks. During this period, tumor progression was observed in 64 cases (91.4%) and death in 43 cases (61.4%). Univariate analysis indicated (Table 12) no correlation of ATM expression with neither PFS (Fig. 8A) nor OS (Fig. 8B). Further, we veri ed the expression of p-AKT and IFIT1 in glioblastoma tissues and their relationship with prognosis. Results found that the IFIT1 was widely expressed in glioblastoma, and the extent of its expression was positively correlated with the expression of p-ATM, and negatively correlated with the expression of MGMT. Increased expression of IFIT1 generally indicates a better prognosis, while the combined analysis of IFIT1 and MGMT is more accurate in predicting the prognosis.

Discussion
Temozolomide (TMZ) is currently the rst-line chemotherapy drug for malignant glioma, which can signi cantly improve the survival period of patients [14,15]. However, resistance to TMZ is a major issue that can impair the prognosis of the patient. Previous studies have shown that MGMT plays an important role in TMZ resistance [16,17].
Activation of MGMT depends on DDR initiation, and ATM is a key regulator of DDR [6]. It is known that the activation of ATM leads to cell cycle arrest and repair of damaged DNA. Mohanty S et al. found that the chemotherapeutic resistance of non-small cell lung cancer H460 cells, to alkylating agent chemotherapy, signi cantly increased after down-regulation of ATM [18]. Zhou Y et al. found that inhibiting ATM increased the resistance of colorectal cancer cells to oxaliplatin chemotherapy [19]. Dixit D et al. found that high ROS can induce apoptosis in glioma cells by activating ATM and its downstream gene YAP1 [9], but no further research on its association with chemotherapy resistance was conducted. Nadkarni A et al. found that ATM knockdown in U87 and U251 cells inhibited their susceptibility to TMZ-induced apoptosis, thus demonstrating their increased resistance to TMZ (IC50) chemotherapy [21]. Therefore, we infer that ATM can mediate TMZ induced apoptosis in glioma cells, inhibition of which can result in resistance to TMZ chemotherapy.
In line with the above inference, we observed a similar result in our in vitro and in vivo experiments. ATM knockdown cells demonstrated continued cellular proliferation and decreased apoptosis to TMZ chemotherapy in in vitro studies. In vivo, in response to TMZ chemotherapy, the tumor growth was signi cantly inhibited and the tumor volume decreased to the minimum on the 10th day after stopping chemotherapy in the control group. However, in the ATM knockdown group the tumor growth gradually increased, and reached a signi cantly higher volume than control group on the same 10th day after the cessation of chemotherapy (p < 0.05). This indicates that the suppression of ATM in glioma cells can induce resistance to TMZ chemotherapy.
ATM plays an important role in promoting cell proliferation. It is found that the expression of ATM is up-regulated in the proliferating cells [22,12]. Studies have shown that ATM can activate the Akt pathway [22,12], and PI3K/Akt pathway is a known initial factor of tumorigenesis that can further promote the proliferation and survival of tumor cells [23]. ATM is also shown to promote the migration and invasion of glioma cells through Akt, extracellular regulatory protein kinase and other pathways [24][25][26].
ATM is one of the central regulator of the DNA damage response pathway that can initiate phosphorylation of key proteins leading to cell cycle arrest, DNA repair or apoptosis [7,8]. Increased expression of ATM can not only promote cellular apoptosis, but can also activate cascade of genes that contributes to cell cycle arrest [6,21,25,27,28].  [20]. Other studies have shown that ATM can activate p53 directly or through Chk2 [7,29,30], and then activate p21 to inhibit the activation of Cdc2 [7,31], leading to cell cycle arrest at G2/M phase and chemotherapy resistance against alkylating agents. Eich M et al. [4] found that ATM knockdown can improve the sensitivity of glioblastoma cell LN229 to TMZ chemotherapy. Some studies have also found that ATM inhibitors can further enhance the chemotherapy-sensitivity of TMZ-sensitive glioma cell lines [20], and can reverse their drug resistance to TMZ by weakening the autophagy [27].  [32], and can also promote the expression of ISGs by activating IFN regulatory factors (IRFs) [33,34]. IFIT1 also has the function of promoting apoptosis [34,35]. Human IFIT family includes IFIT1/ISG56, IFIT2/ISG54, IFIT3/ISG60 and IFIT5/ ISG58, all of which are located on chromosome 10 [37]. Studies have shown that activated ATM can promote the expression of IFN that can further promote the expression of ISGs through IRF1 and IRF7 [33,34]. IFIT family genes are involved in biological processes related to translation initiation, virus replication, double-stranded RNA signal transduction, cell migration and proliferation [37][38][39]. The current studies on IFIT1 largely focus on its antiviral aspects [37,40,41], with few studies studying its role in cancer. In the present study, the expression of IFIT1 in glioma U251 cells was veri ed at the RNA and protein levels. On down-regulation of IFIT1 in U251 glioma cells, an inhibition in cellular apoptosis, increased DNA synthesis, accelerated cell proliferation and increased clone formation of U251 cells was observed. Further, resistance of U251 cells to TMZ chemotherapy was also observed, suggesting IFIT1 as a possible molecular marker indicating the sensitivity of glioma to TMZ chemotherapy.
The endogenous expression of IFN in glioblastoma cells can be enhanced by various factors, including ATM [32][33][34]42]. IFN can induce the expression of IFIT1 [43], and also can improve tumor sensitivity to chemotherapy by inhibiting the expression of MGMT [44][45][46]. Our study found that the expression of IFIT1 in glioblastoma was negatively correlated with MGMT expression, which we speculated to be related to IFN activity, and needs to be studied further.
The current understanding on the role of ATM in glioma chemotherapy has a problem that we can't predict whether activated ATM in glioma after TMZ chemotherapy induces chemoresistance through DNA repair pathway, or improves chemo-sensitivity by inducing apoptosis. With such double-edged characteristics of ATM, it is particularly important to screen the downstream genes induced by ATM to better our understanding of the possible outcome. Clinically, the resistance of glioma to TMZ chemotherapy can be effectively reversed by regulating these genes. Our study starts with the key gene ATM at DNA damage checkpoint, and found that ATM could promote apoptosis by inducing the expression of IFIT1, and improved the sensitivity of TMZ chemotherapy of glioma, thus improving the prognosis of glioma patients.

Conclusion
Our study investigated the role of ATM in TMZ chemotherapy-resistant glioma, in vitro and in vivo. Post ATM knockdown, the changes in the whole genome expression pro le of glioma U251 cells revealed that suppressing ATM could promote resistance to TMZ in glioma by inhibiting IFIT1 expression. This elucidates the underlying mechanism of ATM-regulation mediated development of resistance to TMZ chemotherapy in glioma. Further, we found that p-ATM, IFIT1 and MGMT expression are closely related to the prognosis of glioma patients; especially increased expression of IFIT1 indicates a better prognosis and combination of IFIT1 and MGMT is more accurate in predicting the prognosis.
Thus, ATM, IFIT1 and MGMT can serve as molecular markers to predict glioma's sensitivity to TMZ chemotherapy and patient prognosis.

Availability of data and materials
All authors ensure that our data does not contain any of the following: (a) Infringes or makes unauthorized use of the Intellectual Property Rights or any other right of any person; (b) Is defamatory, derogatory, discriminatory, or violates any rights of privacy; (c) Breaches any applicable law or regulation; (d) Contains a virus, malware, or other potentially harmful component, information, or instructions; (e) Is indecent, obscene, or offensive.
Ethics approval and consent to participate The present study was approved by the Ethics Committee of The Sanbo Brain Hospital of Capital Medical University.
The parents of the patients provided written informed consent.

Patient consent for publication
The pathological specimens and relevant information of the patients used in this study have been asked the patient for consent, and provided written informed consent.