Hypermethylation of the Promoter of miR-338-5p Mediates Aberrant Expression of ETS-1 and Is Correlated With Disease Severity Of Astrocytoma Patients

The pro-oncogene ETS-1 (E26 transformation-specific sequence 1) is a key regulator of the proliferation and invasion of cancer cells. The present work examined the correlation of the aberrant expression of ETS-1 with histological or clinical classification of astrocytoma: grade I (pilocytic astrocytoma), grade II (diffuse astrocytoma), grade III (anaplastic astrocytoma), and grade IV (glioblastoma multiforme). MicroRNA, miR-338-5p, was predicted by an online tool (miRDB) to potentially target the 3’ untranslated region of ETS-1; this was confirmed by multi-assays, including western blot experiments or the point mutation of the targeting sites of miR-338-5p in ETS-1’s 3’untralation region (3’UTR). The expression of miR-338-5p was negatively associated with that of ETS-1 in astrocytoma, and deficiency of miR-338-5p would mediate aberrant expression of ETS-1 in astrocytoma. Mechanistically, hypermethylation of miR-338-5p by DNA methyltransferase 1 (DNMT1) resulted in repression of miR-338-5p expression and the aberrant expression of ETS-1. Knockdown or deactivation of DNMT1 decreased the methylation rate of the miR-338-5p promoter, increased the expression of miR-338-5p, and repressed the expression of ETS-1 in astrocytoma cell lines U251 and U87. These results indicate that hypermethylation of the miR-338-5p promoter by DNMT1 mediates the aberrant expression of ETS-1 related to disease severity of patients with astrocytoma.


INTRODUCTION
Astrocytoma, which originates from astrocytes in the brain, is the most common and fatal type of brain tumor (glioma) (1)(2)(3)(4)(5). There are four histological or clinical grades of the disease: grade I (pilocytic astrocytoma) (6), grade II (diffuse astrocytoma) (7), grade III (anaplastic astrocytoma) (8), and grade IV (glioblastoma multiforme, GBM) (9). More severe astrocytoma (i.e., grades III and IV) (10)(11)(12) is associated with worse prognosis and more aggressive disease. Accumulating evidence indicates that grade I astrocytomas can be considered to be benign and are associated with long-term survival. Patients with grade II astrocytoma also have a survival of approximately 7-8 years (13)(14)(15). However, patients with grade III astrocytoma have an average/median 2-3 year survival, and the median survival of patients with GBM (grade IV astrocytoma) is only 9-14 months (13)(14)(15). Detection of protein factors related to the occurrence and progression of the disease in clinical specimens of different grades is helpful for diagnosis, and can also help to clarify the molecular mechanisms underlying disease progression.
The transcription factor ETS-1 (E26 transformation-specific sequence 1) is an important regulator of malignant tumor cell proliferation that also has central roles in metastasis and invasion (16)(17)(18). Some evidence indicates that ETS-1 is closely related to the progression of a variety of tumors, including breast cancer and liver cancer, and its high expression is positively correlated with the degree of tumor malignancy (19,20). MicroRNAs (miRs) are a type of small non-coding RNA (generally 20-25 nucleotides in length) transcribed and synthesized by RNA polymerase II (21,22). In cells, miRs can act as tumor suppressors by directly acting on the 3' untranslated regions (3'UTRs) of pro-oncogenes to inhibit their expression (23)(24)(25). The deficiency of miRs as tumor suppressors, represented by miR-34a, are considered an important mechanism of cell carcinogenesis (26)(27)(28). Promoter methylation is considered to be a promising mechanism for the deficiency of miRs as tumor suppressors (28)(29)(30). In the present work, the expression of ETS-1 and miR-338-5p was detected in a series of clinical specimens. T h e r e s u l t s r e v e a l e d t h a t t h e D N M T 1 -m e d i a t e d hypermethylation of the miR-338-5p promoter region could participate in the loss of miR-338-5p expression and the high expression of ETS1 associated with disease severity in astrocytoma patients.

Clinical Specimens and Cell Lines
The collection and use of clinical specimens and the protocol of the present work were all reviewed and approved by the Second Affiliated Hospital of Dalian Medical University. Written consent was obtained from all patients. The pathological grades of tumor tissues were confirmed according to the World Health Organization histological classification. Thirty specimens each of grade of astrocytoma (grades I-IV) were included in the present work. All the experiments using human-related materials were performed according to the principles described in the Declaration of Helsinki (1986).

The Small Molecular Inhibitors Used in the Presence Work
The small molecular inhibitors used in the presence work, Decitabine (the Cat. No.: S1200), Azacitidine (the Cat. No.: S1782) or Trichostatin A (TSA, the Cat. No.: 1045) was purchased from the Selleck Corporation, Houston, Texas, USA. The pure powders of these drugs (purity greater than 98.7%) used in this study was dissolved by the dimethyl sulfoxide (DMSO) and diluted by the DMEM without FBS (17,31,32). The solution of the agents was filtered using a sterile 0.22mm pore filter membrane.

Prediction of miRs Potentially
Targeting EST-1, and The Cancer Genome Atlas (TCGA) The miRs potentially targeting ETS-1 were predicted using the online tool miRDB, and those with high scores were selected as candidates. The expression levels of miRs in para-cancerous nontumor tissues were obtained from the GEPIA database, TCGA, and the GTEx project using a standard processing pipeline (36,37). The sequences of the miRs or the ETS-1 with its 3'UTR were searched in NCBI to support the prediction results from online tool miRDB and the interaction between miR-338-5p with ETS-1 was further confirmed by using multi-assays. The endogenous expression levels of miR-338-5p, DNMt-1, ETS-1, and related factors in astrocytoma specimens were examined by quantitative PCR (qPCR) according to the methods described by Chi et al., Yang et al.,. The CT values of the target genes (such as ETS-1 or hsa-pre-miR-338-5p, etc. and the internal reference b-Actin) obtained by qPCR detection are calculated, and the expression levels of these target genes are finally calculated (ie folds to b-Actin). Thereafter, the expression level of the target gene in any specimen is converted to "1" for conversion, and finally the relative expression level of the specific gene in all specimens is obtained. The following primers were used: (1) has-pre-miR-338, stem-loop primer 5'-GTCGTATC CAGTGCGTGT CGTGGAGTCGGCAATTGCACTGGATA CGACCACTCAG-3', forward primer 5'-CCTGGTGCTGAGT GGTCG-3', reverse primer 5'-CAGTGCGTGTCGTGGAGT-3'; (2) ETS1, forward primer 5'-GAGTCAACCCAGCCTA TCCAGA-3, reverse primer 5'-GAGCGTCTGATAGGACTC TGTG

Western Blotting
Two astrocytoma cell lines, U251 and U87, were purchased from the National Infrastructure of Cell Line Resources of Chinese Government, Chinese Academy of Medical Sciences/Peking Union Medical University, Beijing, China. Antibodies against the ETS1 (cat. no. ab220361) and DNMT-1 (cat. no. ab188453) were purchased from Abcam Corporation (Cambridge, UK). The antibody against the intracellular domain of Notch protein was a gift from Dr. Yingshi Zhang of Shenyang Pharmaceutical University, Shenyang City, Liaoning Province, China. The cells were cultured and protein samples were extracted and analyzed using standard western blotting methods (41)(42)(43). GAPDH (cat. no. ab8245) was used as the loading control. Images of protein bands were quantitatively analyzed using Image J software (National Institutes of Health, Bethesda, MD, USA) as described in a previous publication (28). Use Image J software to quantitatively analyze images of images from Western blot, the specific method is: Use Image J software to mark the bands of proteins, and determine the amount of protein expression level based on the total intensity of bands (the area of the bands and grayscale-value intensity). The expression level of ETS-1 or DNMT-1 was corrected by the loading control (GAPDH) (folds of GAPDH). Take the control group as "1" to measure the relative protein expression values of each group.

Cell Culture and Transwell Experiments
U87 or U251 cells were cultured in Dulbecco's modified Eagle medium (DMEM) containing 10% fetal bovine serum (FBS) and transfected with plasmids. Then, cells were harvested for in vitro invasion (transwell) experiments (38). Extracellular matrix (Sigma-Aldrich, Merck & Co., Inc., Kenilworth, New Jersey, USA; diluted 1:10 with DMEM without FBS) was pre-added to the chambers of the transwell plates. Then, the cells were harvested and prepared as single-cell suspension via DMEM; and the cell suspensions of U87 or U251 cells were added to the inner chambers (20000 cells per chamber). Next, the transwell plates were incubated at 37°C in 5% CO 2 for 12 h. The transwell plates were mixed with absolute ethanol and stained with purplecrystal (0.5%, w/v). Photographs of invading cells were taken and quantitatively analyzed with Image J (USA National Institutes of health [NIH], Bethesda, Maryland, U.S.) (28). Use Image J software to quantitatively analyze Invaded cells, the specific method is: Use Image J software to mark the cells, and determine the amount of Invaded cells based on the total number of pixels in the photo and the number of circled Invaded cells. Take the amount of invaded cells in the control group as "1" to determine the relative number of Invaded cells in each group.

Subcutaneous Tumor Model
The animal experiments included in the present work were performed in accordance with the UK Animals (Scientific Procedures) Act 1986 and the associated guidelines. U87 or the U251 cells were cultured and transfected with vectors according to the manufacturer's instruction (the Vigene Corporation, Jinan City, Shandong Province, China). Then, the cells were harvested and cell suspensions were injected subcutaneously (near the veins located in the groin on the inside of the lower extremities) into the nude mice (44). At the injection, mice were reared for 4-8 weeks; then, the mice were sacrificed and tumor tissues were collected for calculation of tumor volumes and weights. Tumor volumes were calculated using the following formula: tumor width × tumor width × tumor length/ 2. Tumor weights (g) were determined using a high-precision electronic balance.

Statistical Analysis
Statistical analysis was performed by SPSS statistical software (software version 9.0, the IBM Corporation, Armonk, NY, the USA). A P value < 0.05 was considered as the significant difference between two groups.

Endogenous Expression of ETS-1 Is Associated With Disease Severity of Astrocytoma
First, the endogenous expression of ETS-1-related factors in astrocytoma was examined. As shown in Figure 1A, the expression level of ETS-1 in clinical specimens gradually increased from grade I to grade IV astrocytoma, with the highest expression level in grade IV. MMPs (matrix metalloproteinases), an invasion-related downstream gene of ETS-1, showed a similar expression trend to that of ETS-1 in astrocytoma ( Figures 1B, C). These results show that the aberrant activation of the ETS-1 pathway is correlated with the severity of astrocytoma.
The expression levels of miR-338-5p-5p in astrocytoma clinical specimens are shown in Figure 1G. The expression level of miR-338-5p-5p (detected as has-pre-miR-338-5p) in clinical specimens gradually decreased from grade I to grade IV, with the highest expression level in grade I ( Figure 1H). Moreover, the expression level of miR-338-5p-5p was negatively correlated with that of ETS-1 ( Figure 1I) and invasion-related genes downstream of ETS-1, MMP1 ( Figure 1J) and MMP9 ( Figure 1K). These results indicate that deficiency of miR-338-5p-5p would promote the aberrant activation of ETS-1 pathway.
Hypermethylation of miR-338-5p-5p Mediated by DNMT-1 Is Associated With miR-338-5p-5p Deficiency and Aberrant Activation of the ETS-1 Pathway Next, the methylation rate of the miR-338-5p-5p promoter region was examined as a potential mechanism of miR-338-5p-5p deficiency. As shown in Figure 2A, there is a CpG site in the promoter region of the selected miR-338-5p-5p. The methylation rates of the miR-338-5p-5p promoter in clinical specimens gradually increased from grade I to grade IV, with the highest level in grade IV ( Figure 2B). The expression of miR-338-5p-5p was negatively correlated with the methylation rate of its promoter region ( Figure 2C). Given that the DNA methylation of tumor suppressors is mediated by DNMT1, the expression of DNMT-1 and its association with miR-338-5p-5p were examined. As shown in Figure 2D, the expression level of DNMT1 in clinical specimens gradually increased with severity, with the highest level of DNMT1 found in grade IV samples. Moreover, the DNMT-1 level was positively correlated with the methylation rate of the miR-338-5p promoter region ( Figure 2E) and negatively correlated with expression levels of miR-338-5p in clinical specimens ( Figure 2F). These results indicate that DNMT-1 contributes to the deficiency of miR-338-5p-5p via enhancing its promoter's methylation.

Knockdown of DNMT1 Enhanced the Expression of miR-338-5p-5p and Repressed the Expression of ETS1 in Astrocytoma Cells
To examine the effects of DNMT1 knockdown on the miR-338-5p-5p/ETS-1 axis and confirm the specificity of miR-338-5p-5p's targeting of ETS-1, qPCR and western blot assays were performed. As shown in Figures 3A, B, knockdown of DNMT1 inhibited the expression of DNMT1 and ETS-1 in U87 and U251 cells. Treatment with decitabine, an antagonist of DNMT1, inhibited the expression of ETS-1 but not that of DNMT1 ( Figures 3A, B). Moreover, treatment with decitabine or a small interfering RNA (siRNA) targeting DNMT1 (siDNMT1) decreased the methylation rate of the miR-338-5p-5p promoter and enhanced the expression of miR-338-5p-5p in U251 cells (Figures 3A, B).
The specificity of miR-338-5p-5p toward ETS-1 was also confirmed. As shown in Figure 3, overexpression of miR-338-5p-5p did not affect the expression of DNMT1 or the methylation rate of the miR-338-5p-5p promoter, but it decreased the expression of ETS-1 in U87 ( Figure 3A) and U251 cells ( Figure 3B). Transfection of ETS-1 with a mutated miR-338-5p-5p targeting site (ETS-1 Mut ) almost completely blocked the effects of miR-338-5p-5p or decitabine treatment (Figures 3A, B). Figures 3A, B show the western blotting results, and histograms (mean ± SD) of the methylation rates of the miR-338-5p-5p promoter and the expression levels of miR-338-5p-5p and invasion-related genes downstream of ETS-1 (MMP1, MMP3, MMP9, and uPA). These results indicate that knockdown of DNMT1 or inhibition of its activity repressed the activation of the ETS-1 pathway via enhancing the expression of miR-338-5p-5p, which targets the 3'UTR of ETS1 in astrocytoma cells.

Inhibition of DNMT1 Activity Inhibited Invasion of Astrocytoma Cells In Vitro
In addition to the above experiments, which demonstrated the effects of DNMT1 on the miR-338-5p-5p/ETS-1 axis, the transwell assays were performed to examine the effects of DNMT1 on the invasion of astrocytoma cells. As shown as Supplemental Figure 1A, treatment with decitabine inhibited the invasion of U87 cells in vitro, whereas transfection of ETS-1 Mut almost completely blocked the effects of decitabine on U87 cells. Similar results were obtained in U251 cells (Supplemental Figure 1B). Therefore, inhibition of DNMT1 activity inhibited the invasion of astrocytoma cells in vitro.

Inhibition of DNMT1 Activity Inhibited Subcutaneous Growth of Astrocytoma Cells
Next, the effects of DNMT1 knockdown on in vivo growth of astrocytoma cells was examined using a nude mice model. The cells were transfected with vectors and then harvested for the subcutaneous tumor model (Figures 4 and 5). As shown in  Transfection of siDNMT1 or miR-338-5p inhibited the subcutaneous growth of U87 cells (Figures 4A-C). Transfection of ETS-1 Mut almost completely blocked the inhibitory effect of siDNMT1 or miR-338-5p on the subcutaneous growth of U251 cells ( Figures 4A-C). The specificity of the effects of the DNMT1/ miR-338-5p-5p/ETS1 axis on the subcutaneous growth of U87 cells was confirmed by the expression levels of factors related to the DNMT1/miR-338-5p-5p/ETS1 axis and the methylation rates of the miR-338-5p-5p promoter ( Figures 4D-L). Similar results were obtained in U251 cells ( Figure 5). Therefore, inhibition of DNMT1 activity inhibited the subcutaneous growth of astrocytoma cells.

The Effect of miR-338-5p's Promoter Methylation by Other Inhibitors to Discard Other Mechanisms
The above experiments mainly focused on the usage of Decitabine, the other inhibitors, Azacitidine or TSA, were also used to discard other mechanisms. As shown in Supplemental Figure 2, Decitabine or Azacitidine inhibited the methylation of miR-338-5p promoter in U251 or U87 cells in a dose dependent manner, and Decitabine is better than Azacitidine on the methylation of miR-338-5p promoter (Supplemental Figure 2). Treatment of TSA did not affect the methylation of miR-338-5p promoter (Supplemental Figure 2). These results further confirmed the specificity of DNMT-1 on miR-338-5p.

DISCUSSION
Accumulating data indicate that prognosis and therapeutic indications vary significantly among different grades of astrocytoma (1)(2)(3)(4)(5). In the present work, aberrant activation of the ETS-1 pathway was identified in high-grade astrocytomas compared with low-grade astrocytomas. Ren et al. in 2016 showed that some extracellular matrix (ECM)-related proteins, including MMPs, were upregulated in high-grade astrocytomas (13). Our results were consistent with those of Ren et al. and would reveal the potential mechanism underlying the high expression of MMPs, that is, the high expression of ETS-1 in astrocytoma tissue may eventually participate in the high expression of MMPs. The highly aggressive characteristics of malignant tumor cells are closely related to disease progression; the main mechanism of this process is that malignant tumor cells destroy ECM through MMPs and so on to reshape the basic structure of tissues, thereby promoting both the growth of tumor cells and their transfer to other locations (44)(45)(46)(47)(48). As a transcription factor, ETS-1 can mediate the transcription of MMPs, etc. The results of this study indicate that the high expression of ETS-1 in astrocytoma and other tissues represents a possible mechanism of disease progression (49)(50)(51). Owing to the intercranial position of astrocytoma tumors, it was difficult to obtain many adjacent non-tumor tissues. Instead, we used data of glioma non-tumor tissues from the TCGA database. Nevertheless, our results confirmed the expression levels of ETS-1 in tissue specimens of different astrocytoma grades, reflecting the role of ETS-1 in astrocytoma. ETS-1 is an ideal target for astrocytoma Our results firstly revealed the relationship between ETS-1 expression and disease progression in clinical specimens. After that, the possible mechanism of the high expression of ETS-1 in astrocytoma was initially explored, and a series of research techniques (Transwell, nude mouse subcutaneous tumor formation experiment) were used to detect the invasion of U87 and U251 cells in vitro by ETS-1 and its upstream pathways and the influence of these factors was also revealed in tumor formation of nude mice. Furthermore, this study investigated the mechanism underlying the high expression of ETS-1 in astrocytoma tissue. Loss of miR-338-5p-5p expression is a possible cause of ETS-1 overexpression ( Figure 6). Methylation of the promoter region of miR-338-5p-5p can also lead to low levels of miR-338-5p-5p in astrocytoma. The miRNA is an important type of small molecule non-coding RNA, which can not only be used as an anti-tumor treatment strategy (for example, in this study, the entire sequence of hsa-pre-miR-338-5p was prepared as rotavirus/lentivirus), but also the deficiency of miRs would be the possible mechanisms for the aberrant expression of its target gene (some pro-oncogene) in malignant tumor tissues (26,(52)(53)(54). In terms of molecular mechanism, miRNA is used in the 3'UTR region of its target gene through a sequence-specific mechanism (55)(56)(57)(58)(59). For this reason, it is necessary to first predict based on a sequence-specific method, and then conduct confirmation of miRs on a specific target gene. In this study, miRDB was first used for prediction, and it was found that there is a miR-338-5p site in the 3'UTR region of ETS-1. The precursor molecule of miR-338 (has-pre-miR-338) was further prepared into lentiviral particles and transfected into U251 and U87 cells, and it was found that miR-338-5p can inhibit the expression of ETS-1. At the same time construct the ETS-1 full-length expression vector containing 3'UTR and the ETS-1 expression vector containing the mutant miR-338-5p site (at this time, the ATATTGT sequence in ETS-1 3'UTR is mutated to CGCGGTG). Overexpression of miR-338-5p can inhibit the expression of wild-type ETS-1 in U251 and U87 cells, but cannot affect the expression of ETS-1 mutants. These all confirmed the effect of miR-338-5p on ETS-1 Since the methylation of the promoter region is an important mechanism for the loss of specific gene expression, the methylation of the miR-338-5p promoter region was tested for this reason (10,60,61). The results of this study found that the methylation level of the promoter region of miR-338-5p increased as the stage of astrocytoma increased. At the same time, the methylation rate of miR-338-5p promoter region is also negatively correlated with the expression of ETS-1. To analyze the possible mechanism of miR-338-5p promoter region, the high expression of DNMT-1 in astrocytoma is the possible mechanism of miR-338-5p promoter region hypermethylation. DNMT1 mediates the methylation of the promoter region of miR-338-5p-5p, and the effect of DNMT1 on miR-338-5p-5p has been confirmed in clinical specimens. The use of a siRNA or antagonist against DNMT1 in U251 and U87 cells was shown to decrease the methylation level of the promoter region of miR-338-5p-5p, upregulate miR-338-5p expression, and finally downregulate ETS-1 expression. Among the four types of DNA methyltransferases, DNMT1 not only participates in the normal methylation process but also induces the silencing of tumor suppressors via hypermethylation of their promoter regions (62,63). The main function of DNMT3 (DNMT3a and DNMT3b) is de novo methylation in the early stage of embryonic development, whereas DNMT2 (also known as tRNA aspartic acid methyltransferase 1) is classified as an RNA methylase (64)(65)(66). Therefore, our results mainly focus on DNMT1.
In the present work, the methylation rate of the miR-338-5p-5p promoter was examined using a BSP-NGS method, in contrast to the MSP method traditionally used in methylation research (28). MSP uses sulfite treatment followed by PCR; the PCR product is cloned into a T-vector to transform Escherichia coli, which are plated to form single-cell colonies, and single colonies are picked for first-generation sequencing analysis. The main limitation of the MSP strategy is that it uses only one pair of primers to amplify one amplicon, which is then transformed to form a single bacterial colony, representing one cell. Thus, it can only be used to qualitatively determine whether there is a methylation modification at a given CpG site. The BSP-NGS methods used in the present work can overcome the shortcomings of the MSP method: after sulfite treatment and high-fidelity PCR, the PCR product (amplicon) is directly subjected to high-throughput sequencing, and the DNA CpG positions of all cells in the astrocytoma tissue are detected. The sequencing results directly reflect the methylation rate of the selected CpG site. The Ion 318 ™ Chip v2 BC chip used in this study can sequence 96 samples at a time. It can not only provide new research methods for methylation-related research, but also provides new ideas for the epigenetic-related research (67)(68)(69)(70).
Based on the factors that ETS-1 is not only an important regulator of human cancerous cells' proliferation, but also enhancing the invasive ability of cells, the transwell assays were performed (16,18,71). The results were shown as the images of the invaded cells or the quantitative results of the images. The basic principle of Transwell is to use 24-well plates as the outer chambers and the inner chambers (72). The bottom of the inner chamber is a thin film with evenly distributed 8mm diameter holes (45). First, 10% concentration of ECM gel (the ECM gel produced by Sigma, diluted with serum-free DMEM in a ratio of 1:10) is added to the bottom of the inner chambers and a thin film is formed to simulate the basement membrane of the tissue (45). After that, the cells are prepared as a single cell suspension (the cell suspension is prepared in DMEM without serum at this time) and then added to the inner chamber (45). The outer chambers (that is, the wells of the 24-well plates) are added with DMEM containing 20% FBS, and then the inner chambers are added to the 24-well cell plates (45). At this time, the cells in the inner chambers will be driven and attracted by the FBS in the outer chambers (this is a standard chemotaxis), migrate from the inner chamber to the outer chamber, and finally distribute on the outside of the bottom of the inner chambers. Since there is a layer of ECM gel inside the bottom of the inner chambers to mimic the tissue basement membrane, the cells need to destroy the ECM gel before they can migrate from the inside of the inner chambers to the outside (45). This process perfectly mimics the invasion of tumor cells. In this study, through the Transwell experiment, the use of siRNA of DNMT-1 or hsa-pre-miR-338 can significantly inhibit the in vitro invasion of U251 or U87 cells; transfection of ETS-1 mutants can inhibit the inhibitory roles of miR-338 or siDNMT on U251 or U87 cells' in vitro invasion.
Decitabine is a common and typical DNMT-1 inhibitor, which has a good inhibitory activity on the activity of DNMT-1 (73)(74)(75). While using Decitabine, this study also used inhibitors of other mechanisms. Azacitidine (5-Azacytidine, 5-AzaC, Ladakamycin, AZA, 5-Aza, CC-486) is a nucleoside analog of cytidine, which inhibits DNA methylation by binding to DNA methyltransferase in competition with cytidine (75)(76)(77)(78). Our results show that Azacitidine can also down-regulate the promoter methylation rate of miR-338-5p, but the activity of Decitabine is better than that of Azacitidine (that is, the inhibitory activity of Decitabine is higher than that of Azacitidine at the same concentration). This may due to the fact that Decitabine directly inhibits DNMT-1, while Azacitidine needs to compete with cytidine to bind to DNMT-1. Trichostatin A (TSA) is an HDAC inhibitor (79)(80)(81), which has nothing to do with DNA methylation. TSA treatment of U251 or U87 cells does not affect the methylation rate of the miR-338-5p promoter region in the cells. This result further confirms the specificity of DNMT-1/miR-338-5p/ETS-1 axis.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics Committee of the Second Affiliated Hospital of Dalian Medical University. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Animal Ethics Committee of the Second Affiliated Hospital of Dalian Medical University.

AUTHOR CONTRIBUTIONS
JW, LM, and DW: concept, design, statistics, data collection, manuscript writing, final approval. CH: design, statistics, data collection. JY: concept, data collection. JW and JY: statistics, manuscript writing. LT: statistics, data collection. LM and LT: statistics, data collection. DW: concept, design, statistics, data collection, manuscript writing, final approval. All authors contributed to the article and approved the submitted version.

SUPPLEMENTARY MATERIAL
The The results were shown as histogram of miR-338-5p promoter's methylation rates (mean ± SD). The statistical analysis was performed by student T-test without two-tails. P < 0.05.