Malignant glioma constitutes one of the fatal primary brain tumors in adults. Such poor prognosis calls for a better understanding of cancer-related signaling pathways of this disease. Here we elucidate a MYC-miRNA-MXI1 feedback loop that regulates proliferation and tumorigenesis in glioma. MYC suppressed MXI1 expression via microRNA-155 (miR-155) and the microRNA-23a∼27a∼24-2 cluster (miR-23a cluster), whereas MXI1, in turn, inhibited MYC expression by binding to its promoter. Overexpression of miR-155 and the miR-23a cluster promoted tumorigenesis in U87 glioma cells. Furthermore, fat mass and obesity-associated protein (FTO), an N6-methyladenosine (m6A) RNA demethylase, regulated the loop by targeting MYC. The ethyl ester form of meclofenamic acid (MA2) inhibited FTO and enhanced the effect of the chemotherapy drug temozolomide on suppressing proliferation of glioma cells and negatively regulated the loop. These data collectively highlight a key regulatory circuit in glioma and provide potential targets for clinical treatment.
These findings elucidate a novel feedback loop that regulates proliferation in glioma and can be targeted via inhibition of FTO to enhance the efficacy of temozolomide.
Malignant glioma is a highly devastating cancer (1), which constitutes the most common primary intracranial tumor (2), accounting for 74.6% of all malignant brain tumors and 24.7% of all primary brain tumors (www.agta.org). Despite numerous studies regarding its pathogenesis, treatment of malignant glioma continues to be challenging and patient prognosis remains poor, primarily owing to its high proliferative potential, infiltrative growth, and high recurrence rate (3). In the clinical treatment of malignant glioma, a combined approach including maximum safe surgical resection followed by radiation with concomitant and adjuvant temozolomide (TMZ) is generally used (4). But none of which can eradicate the tumor cells. The application of TMZ had increased median overall survival in a retrospective cohort study (5). However, the median survival time of newly diagnosed patients remains less than 15 months (6). Thus, a more thorough study of the molecular mechanisms of glioma oncogenesis and proliferation is required.
The MYC transcription factor (TF) is a well-studied proto-oncogene. MYC serves as a master regulator of cell proliferation, the upregulation of which has been demonstrated in a large range of cancers including lung cancer (7), prostate cancer (8), as well as glioma (9). It also regulates cell growth, apoptosis, and inflammation. As a front-line drug for the treatment of glioblastoma multiforme (GBM), TMZ could downregulate MYC expression by activating TAP63 (10, 11). In the physiologic pH, TMZ spontaneously converts to 5-(3-methyltriazen-1-yl)imidazole-4-carboximide (MTIC), a reactive methylation agent, which then transfers the methyl group to DNA (11). The methylated DNA induced mismatch and DNA damage, leading to the activation of TAP63, which directly repress MYC transcription by binding to the promoter (10). The MAX interactor 1 (MXI1) protein specifically competes with MYC for the MAX protein by binding to its MYC-MAX recognition site, forming a MXI1-MAX heterodimer. Thus, MXI1 is considered as a tumor suppressor gene, as it inhibits the function of MYC by preventing the formation of the MYC-MAX heterodimer (12). Accordingly, MXI1 allelic loss was identified in approximately 54% of melanoma cases and occurred at an even higher level in recurrent or metastatic tumors (13). Moreover, a higher MXI1 protein level is associated with a better prognosis in breast cancer (14). A study has also shown that MXI1 inhibits human U87 glioma cell line proliferation by repressing the cyclin B1 gene (15). Together, these observations suggest MXI1 as a critical protein for further studies in glioma.
m6A modification, the most abundant modification in eukaryotic mRNA (16), is involved in various aspects of mRNA metabolism such as alternative splicing and decay (17). An abnormal level of m6A modification contributes to various diseases including cancers (18, 19). Fat mass and obesity-associated protein (FTO) was identified as an N6-methyladenosine (m6A) demethylase, which decreases the global levels of m6A in RNAs, especially mRNAs (20). The FTO gene has also been found to express widely in adult and fetal tissues (21). Several studies with clinical samples have implicated that FTO as an m6A eraser plays a crucial role in tumorigenesis. For example, a high level of FTO is shown to promote the gastric and cervical cancer development, while its expression varies with the subtype of breast cancer. Moreover, Rui and colleagues reveals that FTO regulates MYC transcription and its downstream pathway in leukemia (22), and Qi and colleagues indicates that knockdown of FTO inhibits glioblastoma stem cell (GSC) growth and self-renewal (23). Furthermore, the ethyl ester form of meclofenamic acid MA2, which competes with FTO binding for m6A-containing nucleic acid, could suppress the activity of FTO in HeLa cells and elevate the m6A modification level in RNA (24). Therefore, we wonder whether FTO is a therapeutic target in glioma and affects its development.
miRNAs comprise small (approximately 22 nt), endogenous RNA molecules that act as gene expression regulators in both animals and plants (25). Notably, the dysregulation of numerous miRNAs is related to the carcinogenesis or metastasis of various types of cancers. For example, glioma cell proliferation has been found to be associated with several miRNAs including miR-124a, miR-499a, miR-128a, and miR-199a-3p (26), suggesting that miRNA interference is very common in the molecular mechanism of glioma cell proliferation. In addition, miR-155-5p is upregulated in cancers in B cells, breast, bladder, and lung (27), whereas miR-24-3p, a member of the microRNA-23a∼27a∼24-2 cluster (miR-23a cluster), inhibits metastasis in osteosarcoma cells. Moreover, it participates in the miR-24-3p/p130Cas axis, which modulates migration and invasion in different cancers (28).
Our previous studies have found that the miR-23a cluster and miR-155 downregulate MXI1 and thus promote cell proliferation in several glioma cell lines (29, 30). However, whether MXI1 interacts with other molecules or is affected by TMZ or MA2 in glioma remains unknown. Here, we further studied the regulatory relationship among miR-155, miR-23a cluster, MXI1, and MYC, as well as how these miRNAs affect glioma cell proliferation and tumorigenesis. On the basis of our results, we postulated a MYC-miR-155/23a cluster-MXI1 positive feedback circuit in glioma. Also, FTO could regulate MYC, therefore affecting the feedback circuit. Moreover, MA2 as the inhibitor of FTO can enhance the efficiency of TMZ in suppressing the proliferation rates of glioma cells by targeting MYC and negatively regulate the loop.
Materials and Methods
To overexpress human MYC, MXI1, and FTO, cDNAs lacking their native 3′UTRs were cloned downstream of the CMV promoter in the lentiviral expression vector pCDH-CMV-MCS-EF1-copGFP (pCDH; System Biosciences). Construction of the luciferase reporter containing the MXI1 3′UTR fused to the Renilla luciferase reporter gene in the psiCHECK2 vector (Promega), termed MXI1-3′UTR, was described in a previous publication (30). The construction and expression of human miR-155 and the miR-23a cluster were described in our previous publication (30, 31). The luciferase reporter vectors containing promoter regions of the miR-23a cluster and miR-155 fused to the firefly luciferase reporter gene at the 5′-end were cloned into the pGL3-basic vector (Promega). The shRNAs for FTO, MYC, and MXI1 were cloned into the pLenti-hU6BX (gift from Zhigang He's lab at Harvard Medical School, Boston, MA). DNA sequencing was performed to confirm all of the constructs. The primers used are listed in Supplementary Table S1.
Cell lines and cell culture
The HEK293T (subsequently referred to as 293T) cell line and the human glioma cell lines U87, U251, and A172 were purchased from the ATCC. All the cells were cultured in DMEM (Invitrogen) supplemented with 10% FBS (Invitrogen) and penicillin/streptomycin (Invitrogen; 100 U/mL) in an incubator at 37°C with 5% CO2. All these cell lines were characterized by finger printing analysis. All cell lines used were passaged less than 3 months, at which point, a previously frozen vial was thawed. Mycoplasma was tested every 2 months by 4′,6-diamidino-2-phenylindole (DAPI) staining, and no Mycoplasma contaminated cells had been detected during the experiments.
Glioma cells treated with chemical compounds
TMZ (Selleckchem) and MA2 (BSZH Scientific Inc.) were dissolved in DMSO to 0.1 g/mL and then diluted to final concentration by culture medium. All of compounds were freshly prepared for experiment use. Glioma cells were grown for 24 hours before the addition of chemical compounds in each tests, and the medium was changed every 48 hours with the addition.
All of the RNA oligonucleotides including siRNAs, miRNA mimics, and inhibitors were purchased from GenePharma. The RNA sequences are listed in Supplementary Table S2. The control RNA contained random sequences without known interactions with other mRNA molecules in cells. A total of 100 nmol/L or indicated amounts of siRNAs, miRNA mimics, or inhibitors were used to perform cell transfections using the X-tremeGENE siRNA Transfection Reagent (Roche) according to the manufacturer's instructions.
Dual luciferase reporter assay
The dual luciferase reporter assay was performed as described previously (30). Briefly, 50 ng luciferase reporter vector containing a predicted target site and 150 ng miRNA expression vector or 6 ng miRNA mimic were transfected into 1.5 × 104 293T cells in 96-well plates using FuGENE HD (Roche) or polyethylenimine (Sigma) according to the manufacturer's instructions. After 48 hours, the luciferase activity was measured using the dual luciferase reporter assay system (Promega) following the manufacturer's instructions and Renilla luciferase activities were normalized to firefly luciferase activities.
Western blot analysis and chromatin immunoprecipitation assay
These procedures were performed as described previously (31). The separated proteins by gel electrophoresis were transferred to a polyvinylidene fluoride membrane (Millipore). The membrane was incubated overnight at 4°C with a primary rat polyclonal antibody against human MYC (1:5,000; Santa Cruz Biotechnology), rabbit polyclonal antibody against human FTO (1:1,000; Abcam), MXI1 (1:3,000; Santa Cruz Biotechnology Inc. or AB clonal Technology), or mouse mAb against human GAPDH (1:3,000; Abcam). Chromatin immunoprecipitation (ChIP) was performed following the protocol of ChIP Assay Kit (Millipore). Immunoprecipitations were performed using 2 μg each of above MXI1 antibody, and anti-mouse IgG (Millipore) overnight at 4°C with rotation, followed by PCR to confirm the enrichment of binding DNA fragments.
RNA extraction and qRT-PCR
Total RNA of the cell lines was extracted using TRizol (Invitrogen). For reverse transcription (RT), the PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa) was used following instructions from the manufacturer. A specific RT primer was used for each miRNA. One microliter of each primer (10 μmol/L) was mixed together for the RT reaction. The expression of primary and mature miR-23a-3p, -24-3p, -27a-3p, -155-5p, and the 19 mature miRNAs listed in Supplementary Table S3, as well as FTO, MYC, and MXI1 mRNAs in the cell lines was quantified using the SYBR Premix Ex Taq II (TliRNaseH Plus) Kit (TaKaRa). Thermo cycling conditions were as follows: preincubation at 95°C for 2 minutes, followed by 40 cycles of denaturation at 95°C for 15 seconds, annealing and extension at 60°C for 40 seconds. U6 snRNA was used as an endogenous control for mature miRNAs and GAPDH was used for primary miRNAs and mRNAs. qPCR was performed using the LightCycler 480 Real-Time PCR system (Roche). The data were analyzed using LightCycler 480 Software Version 1.5. All primers used are listed in Supplementary Table S1.
Cell proliferation assays
Glioma cell proliferation was measured by 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay and 5-ethynyl-2-deoxyuridine (EdU) incorporation assay. For the MTT assay, A172, U251, and U87 cells were seeded on a 96-well plate 1 day before transfection or treatment. At 72 hours after transfection or indicated time after drug treatment, MTT reagent (5 mg/mL) was added directly to the medium, and the plate was incubated in a 37°C, 5% CO2 incubator for 4 hours. The supernatant was then removed and 100 μL of DMSO was added to each well and thoroughly mixed for 10 minutes. The spectrometric absorbance of the samples at 490 nm was measured using a microplate reader (BioTek). The absorbance value was normalized to vector control or DMSO control. All experiments were performed in triplicate. For the EdU incorporation assay, an EdU Assay Kit (Ribobio) was used according to the manufacturer's instructions. The cells were subsequently stained with 100 μL of DAPI for 30 minutes and visualized under an ECLIPSE Ti-U fluorescence microscope (Nikon). The total number of cells (blue cells) and EdU-positive cells (red cells) was counted using NIS-Elements BR 3.0 software (Nikon). The EdU incorporation rate was expressed as the ratio of the number of EdU-positive cells to the total number of DAPI-positive cells. All experiments were performed in triplicate.
Lentivirus production and infection
293T (8 × 105) cells were seeded on a 6-cm culture dish and were then cotransfected with 1.8 μg of the packaging plasmid psPAX2, 0.6 μg of the envelope expressing plasmid pMD2.G, and 2.5 μg of the gene expression vector or empty vector pCDH using the transfection reagent polyethylenimine (Invitrogen) according to the manufacturer's instructions one day later. Viral supernatants were harvested and stored in a −80°C refrigerator 48 hours after transfection. To improve the infection efficiency, 10 μg/mL polybrene (Sigma) was added prior to infection.
Glioma xenograft model
We first established lentiviral-mediated miR-155, miR-23a cluster, or empty vector stably infected U87 cell lines for further use. The virus was prepared and U87 cells were infected as described previously (see lentivirus production and infection). At 3 days after infection, GFP positive cells were collected by fluorescence activating cell sorting for miR-155, or puromycin was added to kill uninfected cells for miR-23a cluster sorting. These cells were plated on a 10-cm culture dish and were cultured for another 11 days to obtain a sufficient cell number for injection. For each miRNA, a total of 16 female nude mice aged between 6 and 8 weeks were randomly distributed into two groups, each containing eight mice. Previously prepared U87 cells (5 × 106 cells, 60 μL) stably infected with miRNA were injected subcutaneously in the right flank of the mice, whereas control U87 cells infected with empty vector were injected in the left flank. The length and width of tumors were measured using a caliper every two days for statistical analysis. Tumor volume was calculated as 0.5 × width2 × length. After 22 days, all the mice were sacrificed by cervical dislocation and the tumors were harvested. Tumor tissues were immersed in 10% neutral buffered formalin overnight for the IHC study. For hematoxylin and eosin (H&E) staining, deparaffinized tissue sections were stained with Mayer Hematoxylin and Eosin solution. The nude mice were purchased from and bred and housed in Sun Yat-sen University Laboratory Animal Center. The use of nude mice was approved by the Animal Ethics Committee of Sun Yat-sen University. All animal protocols were in accordance with guidelines of Animal Ethics Committee of Sun Yat-sen University.
Sections (4 μm) of formalin-fixed, paraffin-embedded mouse xenograft tissues were generated using a rotary microtome (Leica) and labeled with anti-Ki-67 (Dako) primary antibody. 3,3-Diaminobenzidine tetrahydrochloride was used to visualize the staining reaction, and Mayer hematoxylin was used for subsequent counterstaining. Total numbers of cells and Ki-67–positive cells were counted from three independent 50 × fields and the percentage of Ki-67–positive cells was calculated and averaged.
For miRNA library screening analysis, a modified SSMD was used as reported previously (32).
For each experiment, triplicate repeats were performed. A double-tail Student t test was used as the significance testing method.
MYC downregulates MXI1 in glioma cell lines via its 3′UTR
Given that MXI1 is downregulated in glioma, we hypothesized that upregulated miRNAs may inhibit MXI1 translation by binding to its 3′UTR (30) and that some TFs may in turn be responsible for the upregulated miRNAs, eventually leading to lower MXI1 protein level. Therefore, we first screened for miRNAs that target the MXI1 3′UTR by a large-scale screening method using a human miRNA expression library containing over 600 pre-miRNAs that were previously developed in our laboratory. We cotransfected a luciferase reporter vector containing the full-length 3′UTR of the MXI1 gene (MXI1-3′UTR) along with each of the miRNA expression vectors from the library and found that 42 pre-miRNAs regulated luciferase activities via the MXI1 3′UTR using strictly standardized mean difference (SSMD) analysis (SSMD < −2, Fig. 1A and B; Supplementary Table S4). Among these, we then chose 28 pre-miRNAs with overall stronger inhibiting efficiency for further analysis, which contained 52 mature miRNAs (log2 fold change < −0.5; Supplementary Tables S4 and S5). To select mature miRNAs for further investigation, we chose those that had been reported to be upregulated in glioma through an intensive search of the recently published literature, which identified a total of 366 such miRNAs, as summarized in Supplementary Table S6 (33–37). Overall, 19 mature miRNAs (belonging to 17 pre-miRNAs) were both upregulated in glioma and identified as possibly targeting the MXI1 3′UTR (Fig. 1C; Supplementary Table S3).
To clarify which TFs might regulate the expression of these 19 miRNAs we were interested in, we used approximately 2 kb sequences upstream of each pre-miRNA as the promoter to predict probable TFs through TFBIND software (38). In particular, MYC appeared to represent a promising candidate because of its established role in accelerating cell proliferation. Thus, we hypothesized that MYC might regulate the expression of miRNAs we were interested in. To test this, we first tested whether MYC regulated the MXI1 3′UTR using a dual luciferase reporter assay. The cotransfection of the MXI1 3′UTR luciferase vector into 293T cells along with a MYC-overexpressing vector showed that MYC significantly lowered the luciferase activity. In contrast, the depletion of MYC by transfecting the siRNAs si-MYC-1 or si-MYC-2 into 293T cells enhanced the luciferase activity (Supplementary Fig. S1A). Consistent with these findings, further investigation in U87 glioma cells confirmed that lentiviral-mediated MYC overexpression suppressed luciferase expression by targeting the MXI1 3′UTR (Fig. 1D), suggesting that MYC may regulate MXI1 expression via its 3′UTR.
Next, we tested whether MYC regulated endogenous MXI1 expression in glioma cell lines. In U87 cells, overexpression of MYC by lentivirus infection lowered MXI1 protein (Fig. 1E) and mRNA (Fig. 1F) level. These results were further validated in U251 and A172 glioma cells using Western blot analysis (Supplementary Fig. S1B). Conversely, depletion of MYC by transfecting the siRNAs si-MYC-1 or si-MYC-2 into U87 cells led to a significant increase in MXI1 protein level (Supplementary Fig. S1C). The results were further confirmed in 293T cells (Supplementary Figs. S2A and S2B). Summarily, these results indicated that MYC could repress MXI1 expression in various glioma cell lines as well as in 293T cells.
MYC regulates miR-155 and the miR-23a cluster at the transcriptional level
Given that miRNAs usually play regulatory roles by targeting gene 3′UTR, we next ascertained the miRNAs by which MYC could regulate MXI1 3′UTR in glioma. Toward this purpose, we delivered the MYC gene into U87 cells with lentivirus, and determined the levels of the 19 miRNAs described above (Fig. 1C) using quantitative reverse transcription-PCR (qRT-PCR). The results indicated that overexpression of MYC significantly increased the levels of nine of these miRNAs in U87 cells including miR-155-5p, and two miRNAs (miR-24-3p, and -27a-3p) from miR-23a cluster (fold change > 1.5, P < 0.05, Fig. 2A). An analysis of the Spearman rank correlation coefficients (Rho value) between 662 human miRNAs and MYC mRNA levels across pan-cancer based on 9,125 samples in The Cancer Genome Atlas indicated that MYC mRNA levels correlated with the expression levels of 88 human miRNAs, which includes miRNA-155-5p, miRNA-24-3p, and miRNA-27a-3p (39), and does not include any other of the above nine miRNAs (Fig. 2B). Therefore, we focused on these three miRNAs to gain insights into the miRNAs, which may mediate the regulation of MYC on MXI1 in both glioma cell lines and clinical tissues.
To further confirm the regulation of MYC on miR-155 and the miR-23a cluster, we examined the expression levels of primary transcript miRNAs (pri-miRs) in glioma cells. As expected, the normalized expression levels of pri-miR-155, pri-miR-23a, pri-miR-24, and pri-miR-27a all increased significantly in MYC-overexpressing U87 cells (Fig. 2C). This is further confirmed by the result that MYC knockdown by siRNAs decreased pri-miR-23a, pri-miR-24, pri-miR-27a, and pri-miR-155 levels (Supplementary Fig. S2C). In addition, we constructed luciferase vectors containing different segments of the miR-23a cluster and miR-155 promoters. The transcription start site (TSS) of the miR-23a cluster gene was indicated as +1 (40). However, we were unable to identify any studies regarding the TSS of miR-155; therefore, the position of the first nucleotide of pre-miR-155 was designated as +1. For the miR-23a cluster, −912 to +49 bp (miR-23a-p-1k) and −603 to +49 bp (miR-23a-p-0.5k) were cloned. For miR-155, −965 to −14 bp (miR-155-p-1k) and −491 to −14 bp (miR-155-p-0.5k) was cloned (Supplementary Figs. S3A and S3B). No change was observed in the relative luciferase ratio with miR-23a-p-0.5k, whereas miR-23a-p-1k markedly promoted the luciferase level (Fig. 2D), suggesting that MYC accelerated miR-23a cluster expression as a TF, and that the interaction site was located in the region from −912 to −603 bp of the miR-23a cluster promoter. However, both miR-155-p-1k and miR-155-p-0.5k promoted luciferase expression, suggesting a potential interaction site with MYC located from the −491 to −14 bp region of the miR-155 promoter (Fig. 2E). To further validate that MYC indeed regulate MXI1 expression through miR-155-5p, miR-24-3p, and miR-27a-3p, we sought to test whether inhibiting these miRNAs would rescue decreased MXI1 expression mediated by overexpression of MYC. We treated the MYC-overexpressing U87 cells with either equally mixed inhibitors of miR-155-5p, miR-24-3p, and miR-27a-3p, or control. As expected, MYC-overexpressing cells treated with miRNA inhibitors showed increased MXI1 level as compared with MYC-overexpressing cells without inhibitors (Fig. 2F). This result further validated our MYC-miRNAs-MXI1 model.
MXI1 downregulates MYC by targeting its promoter
Having confirmed that MYC downregulated MXI1 in glioma cells, we then interrogated whether MXI1 serves as a regulator of MYC in turn. In 293T cells, Western blot showed that transient transfection with a MXI1 expression vector markedly decreased MYC mRNA and protein levels (Supplementary Fig. S4A). Accordingly, lentiviral-mediated MXI1 overexpression in U87, U251, and A172 cells suppressed MYC expression at the mRNA and protein levels (Supplementary Fig. S4B). Consistently, shRNAs targeting MXI1 increased MYC mRNA and protein levels in U87, U251, and A172 cells (Supplementary Fig. S4C). Dual luciferase assay of the wild-type or putative MXI1-binding-site-mutated MYC promoter and ChIP assay (Supplementary Figs. S4D and S4E) indicated that MXI1 downregulated MYC expression levels by effectively targeting its promoter. Taken together, these results elucidate a MYC-miR-155/23a cluster-MXI1 feedback loop in glioma cells, in which MXI1 is negatively regulated by MYC via miR-155 and miR-23a cluster, whereas MXI1, in turn, impaired MYC expression by targeting the MYC promoter.
MXI1 impairs the effect of miR-155 and MYC on proliferation in glioma cells
Next, we addressed how this feedback loop impacted proliferation in glioma cells. As previous studies linked MYC expression with glioma malignant transformation (41) whereas MYC inhibition reduces proliferation (9), we first validated its role in accelerating proliferation in glioma cells. A representative graph from an EdU incorporation and DAPI staining assay is shown in Supplementary Fig. S5A. Proliferating cells were stained red by EdU, whereas all nuclei were stained blue by DAPI. MYC gene delivery into both U251 and U87 cells led to a significantly higher EdU-positive to DAPI-positive cell ratio (Supplementary Figs. S5A and S5C). In addition, in the MTT assay, significantly higher relative absorbance was observed after overexpression of MYC in U251 and U87 cells (Supplementary Figs. S5B and S5D). Collectively, these findings demonstrated that MYC accelerated the proliferation rate in glioma cells.
To further validate that this effect was mediated via MXI1, we first conducted a rescue experiment by cotransfecting MXI1 and different dosages of a miR-155 mimic (50 or 100 nmol/L) into U87, U251, and A172 cells followed by the MTT assay to measure cell proliferation. When 50 nmol/L miR-155 mimic was cotransfected, no significant change in the absorbance value was observed compared with the control group (cotransfection of pCDH and the mimic control), which suggested that MXI1 completely rescued the effect of 50 nmol/L miR-155 mimic in promoting proliferation. In comparison, cotransfection with 100 nmol/L miR-155 mimic showed a markedly higher absorbance value, suggesting a higher proliferation rate and that the same dosage of MXI1 only partially rescued the effect of the miR-155 mimic in promoting proliferation (Supplementary Fig. S5E). Consistently, silencing MXI1 rescued the effect of miR-155 inhibitor in inhibiting proliferation (Supplementary Fig. S5F).
Next, we overexpressed MYC and MXI1 individually or in combination with lentivirus in U87 and 293T cells and measured cell proliferation using the EdU incorporation assay. As expected, overexpression of MYC markedly promoted the proliferation rate, whereas overexpression of MXI1 alone inhibited proliferation (Supplementary Fig. S5G). When MYC was coexpressed with MXI1, no obvious change was observed compared with the control group (Supplementary Fig. S5G), indicating that MXI1 rescued the higher proliferation rate caused by MYC.
miR-155 and the miR-23a cluster promote proliferation and tumorigenicity in U87 cells
To evaluate how miR-155 and the miR-23a cluster affected tumorigenicity, U87 cells infected with lentivirus overexpressing miR-155 or the miR-23a cluster (with empty vector as a control) were subcutaneously injected into nude mouse. U87 cells overexpressing miR-155 and the miR-23a cluster grew into much larger xenografts (Fig. 3A and B; Supplementary Figs. S6A and S6B). After measuring the tumor volume every 2 days from 8 to 22 days after injection, it was obvious from the resultant growth curves that miR-155 and the miR-23a cluster greatly enhanced the tumor size, with miR-155 promoting oncogenesis in U87 cells to a much greater degree than the miR-23a cluster. On day 22, the tumor volume of miR-155-overexpressing cells was almost twice of that of miR-23a cluster-overexpressing cells (Fig. 3A and B, bottom). The overexpression of miR-155 and miR-23a cluster was confirmed by the measurement of the primary and mature transcripts of miR-155, miR-27a, and miR-24 in the tumors from each group with qRT-PCR (Supplementary Fig. S6C).
Representative portions of the xenografts were then processed for HE staining (Fig. 3C, left). Xenografts with overexpression of the miR-23a cluster or miR-155 showed typical tumor morphology, although the morphology was most notable in the miR-155-overexpressing group. In addition, we IHC stained tumors from a miR-23a cluster-overexpressing- (No. 2) and miR-155-overexpressing- (No. 8) nude mouse for Ki-67, a proliferation marker. A greater number of cells were stained for Ki-67 in the miR-23a cluster- and miR-155-overexpressing groups than in the controls, as indicated by both microscopy observation and statistical analysis (Fig. 3C, right). Western blot analysis and qRT-PCR performed on the tumors from each groups indicated that a decrease in MXI1 and an increase in MYC protein and mRNA levels due to the overexpression of miR-23a cluster or miR-155 (Fig. 3D; Supplementary Fig. S6C), suggesting that the feedback loop also exists in the xenografts.
FTO regulates the expression of MYC, miR-155, and miR-23a cluster, and proliferation of glioma cells
FTO, the first identified RNA demethylase, has been reported to increase the stability and the translation efficiency of MYC through decreasing its level of m6A mRNA modification, play a crucial role in the leukemia progression (22, 42), Therefore, we wonder if FTO also regulated MYC and its feedback circuit in glioma cells. We knocked down FTO with two distinct shRNAs in A172, U251, and U87 glioma cell lines, the efficiency of which was confirmed by qRT-PCR and Western blot analysis (Fig. 4A and B). In the three glioma cells, FTO knockdown decreased and increased respectively the expression levels of MYC and MXI1 (Fig. 4A and B; Supplementary Fig. S7A). To further investigate whether the decrease in the expression of FTO also affects the feedback circuit, we measure the expression levels of both the primary and mature transcripts of miR-155-5p, miR-24-3p, and miR-27a-3p. The qRT-PCR results showed that FTO knockdown significantly decreased the levels of the primary and mature transcripts of the three miRNAs in A172, U251, and U87 glioma cells (Fig. 4C; Supplementary Fig. S7B).
To further confirm the regulatory effect of FTO on the circuit, MA2, a highly selective inhibitor of FTO (24) was used. When A172, U251, and U87 glioma cells treated with 200 μmol/L MA2 for 2 days, a notable decrease in both the mRNA and protein levels of MYC was detected, while we saw no difference in FTO expression (Fig. 4D and E). The expression of the primary and mature transcripts of the three miRNAs downstream of MYC also decreased significantly after the MA2 treatment (Fig. 4F; Supplementary Fig. S7C). Conversely, ectopically expression of FTO in A172, U251, and U87 glioma cells promoted the expression of both MYC and the miRNAs downstream of MYC (Fig. 4G and H; Supplementary Fig. S7D). In addition, the increased expression of the miRNAs downstream of MYC were inhibited by silencing MYC while ectopically expression of FTO (Supplementary Fig. S7E). Summarily, our data indicated that FTO regulated the feedback loop in glioma cells by targeting MYC transcripts.
To confirm the effects of FTO on glioma proliferation, we treated the A172, U251, and U87 glioma cells with 0, 20, 100, 200, and 400 μmol/L of MA2 for 1 to 5 days. MTT assay showed that the relative absorbance of all cells had decreased after MA2 treatment, among which, A172 was the most sensitive one (Fig. 5A). As a negative control, the treatment with 400 μmol/L sucrose, an irrelevant compound to MA2, did not affect proliferation (Supplementary Fig. S8A). We then performed the EdU assay on A172, U251, and U87 glioma cells after treated them with 0, 20, 100, 200, and 400 μmol/L of MA2 for 2 days for further validation. The data showed that MA2 treatment caused a significant decrease in EdU positive cells in a dosage-dependent manner in the three glioma cell lines (Fig. 5B). Meanwhile, the MTT and EdU assays also showed that FTO knockdown in glioma cells by lentivirus expressing two distinct FTO shRNAs (shFTO-1 and shFTO-2) could suppress cell proliferation (Fig. 5C; Supplementary Fig. S8B). This effect was further confirmed by the overexpression of FTO (FTO-OE) in A172, U251, and U87 glioma cells (Fig. 5D). Thus, we concluded that FTO could promote glioma proliferation.
MA2 promotes the effect of TMZ on decreasing the viability of glioma cells
To investigate whether FTO inhibition could increase the efficiency of TMZ in suppressing cell proliferation of glioma, we treated A172, U251, and U87 glioma cells with TMZ and MA2 separately or in combination for 1 to 5 days. Then the MTT and EdU assays were conducted to evaluate cell proliferation. The combination of MA2 and TMZ suppressed cell proliferation far more notably than single treatment (Fig. 6A and B). Consistently, similar results were obtained in A172 and U251 glioma cells when TMZ treatment was combined with FTO knockdown (Fig. 6C and D; Supplementary Figs. S9A–S9C). Summarily, our results showed a powerful enhancing effect of FTO inhibition on the suppression of cell proliferation in combination with TMZ, implying that the potential of MA2 in the treatment of glioma.
To reveal the molecular mechanism underlying the above enhancing effect of FTO inhibition, we detected the expression levels of MYC in A172, U251, and U87 glioma cells treated with TMZ and MA2 separately or in combination. As shown by Western blot analysis and qRT-PCR, the protein and mRNA levels of MYC in TMZ- or MA2-treated groups decreased compared with the vehicle control. Such down-regulatory effect was even more evident in the combination treatment group (Fig. 7A and B). Then we performed qRT-PCR to examine the levels of primary and mature transcripts of miR-24-3p, miR-27a-3p, and miR-155-5p. As expected, the miRNA expression at both primary and mature levels trended downward with the MYC decline (Fig. 7C; Supplementary Figs. S10A and S10B). Collectively, we proved that FTO inhibition could suppress glioma proliferation through the feedback loop by downregulating the expression level of MYC.
The current poor prognosis of glioma, the most common brain tumor (2) suggests that a more thorough study of the molecular mechanisms regarding glioma tumorigenesis, proliferation, and chemoresistance is urgently needed. In this study, we proposed a MYC-miR-155/23a cluster-MXI1 feedback circuit in glioma (Fig. 7D), wherein the MYC-regulated MXI1 protein inhibits MYC expression in turn by binding to the MYC promoter (Supplementary Fig. S4). Specifically, we found that MYC regulated MXI1 via miR-155 and the miR-23a cluster by targeting its 3′UTR (Figs. 1 and 2). Moreover, we found that FTO regulated the feedback circuit via MYC and confirmed that FTO is involved in MYC-mediated promotion of miR-155 and miR-23a cluster, which enhances the malignant properties in glioma cells (Figs. 4 and 5). MA2 could enhance the growth-inhibition effect of TMZ by targeting MYC and downstream pathways (Figs. 6 and 7).
Several feedback circuit regulatory mechanisms have been previously identified in cancers. Typical examples include the MiT/TFE-RagD-mTORC1-MiT/TFE feedback circuit (43), PKM2/NF-κB/miR-148a/152 regulatory circuit (44), and Her2-let-7-β2-AR circuit in breast cancer (45), all of which are related to cancer growth or prognosis. We postulated that the feedback circuit might show a more robust balance between important genes, while simultaneously being highly sensitive, as the dysregulation of each gene may impact two or more others. This distinctive feature renders these genes as key regulators of cell proliferation and cancer progression.
MYC is an established oncogene in various cancers including glioma (41) and MYC protein has been shown to compete with MXI1 for MAX, forming a MYC-MAX heterodimer and functioning as a TF (12). Although MYC and MXI1 share a bHLH-Zip motif (12), little is known regarding the regulatory interactions between these two genes at a translational level. We found that MYC could inhibit MXI1 expression by targeting its 3′UTR via miRNAs (Fig. 1). In particular, the inhibition of MXI1 upon overexpression of MYC was a result of MYC-mediated up-regulation of miR-155 and the miR-23a cluster, which then bound to the MXI1 3′UTR and suppressed its expression (Figs. 1 and 2). These results support that, in contrast to MYC, MXI1 functions as a tumor inhibitor. Thus, our findings present a novel aspect of the interactions between MYC and MXI1, two master regulators of cell proliferation.
Recently, accumulated studies have shown that FTO plays a vital role in cell proliferation and metastasis (19). Exploring the molecular mechanism of FTO function in glioma may provide a new avenue for treating GBM. In this study, we demonstrated that knockdown or inhibition of FTO declined cell growth and the transcripts of MYC. Given the crucial role of MYC in glioma, it is possible that the diminishment of its expression could be therapeutic. It has been previously shown that TMZ-induced TAP63 suppresses MYC promoter activity (10, 46), and FTO can increase the transcription level of MYC by improving its stability through an m6A-dependent mechanism (47). The present study showed that the combination of TMZ and FTO inhibitor MA2 markedly decreased MYC mRNA level and cell viability in glioma cells compared with their single treatment. Our work has validated FTO as a new target, which might support the development of novel therapies against GBM.
In this study, the proteins that were found to regulate miR-155 and the miR-23a cluster outside of the feedback loop may play a critical role in controlling the balance of the whole loop, thus impacting the tumorigenesis of glioma cells. Specifically, the MYC-miR-155/23a cluster-MXI1 loop is a positive feedback circuit. For example, the overexpression of MYC would lead to a higher level of inhibition of MXI1, suggesting that the level of MYC may increase even higher as a consequence of less inhibition from MXI1. Therefore, we hypothesized that glioma malignancy could be attributed in part to the loss of balance between MYC and MXI1. In turn, this suggests the existence of some key factors that disrupt the initial balance. Notably, our current findings indicate that FTO may constitute such regulators. The significance of FTO as a human m6A demethylase in the regulation of MYC in epigenetic processes and its promoting effect on GSC progression have been discussed by previous studies (23). Our study further our understanding of the function of FTO by proving its regulatory effect on the MYC-miR-155/23a cluster-MXI1 loop. According to our understanding, this is also the first work revealing the regulatory effect of FTO on a miRNA cluster included feedback loop before we submitted the paper.
Our large-scale screening using a miRNA library suggested the existence of a more complicated regulatory network among miRNAs, MYC, and MXI1 than had been previously recognized. Many miRNAs that showed a strong interaction with the MXI1 3′UTR have been found to be concurrent regulators of MYC, although how the three molecules interact with each other and their potential role(s) in cancer have not been clearly understood. For example, previous studies reported that miR-486, which appeared at the top of our screening list (Supplementary Table S4), suppresses hepatocellular carcinoma cell viability and proliferation by inhibiting insulin-like growth factor, which includes MYC among its downstream mediators (48). Further study is therefore indicated to investigate whether miR-486, MXI1, and MYC interact with each other and thus modulate proliferation in glioma. Our work also suggests that the combinational application of FTO inhibitors and TMZ could be more effective in inhibiting the progression of glioma cancer cells. To better evaluate such effect, further studies would be necessary in other cancers, and more experiments might need to be applied to patients for clinical proofs.
In summary, our data collectively highlight a key regulatory MYC-miR-155/23a cluster-MXI1 feedback circuit in glioma, in which MYC negatively regulates MXI1 expression via miRNAs and is itself inhibited through MXI1 binding to its promoter, with the feedback circuit regulated as a whole by FTO-mediated control of MYC expression. Furthermore, MA2, the inhibitor of FTO, can enhance the efficiency of TMZ in killing the glioma cells by targeting MYC and affect the loop. Thus, these components provided novel potential targets for clinical treatment in glioma and the prediction of prognosis.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
L. Xiao: Investigation, methodology and writing-original draft. X. Li: Investigation and methodology. Z. Mu: Investigation, methodology, writing-original draft, writing-review and editing. J. Zhou: Investigation. P. Zhou: Investigation and methodology. C. Xie: Resources and investigation. S. Jiang: Conceptualization, supervision, writing-review and editing.
S. Jiang was supported by grants from the National Natural Science Foundation of China (Grant nos. 81272773 and 81572567) and the Guangdong Basic and Applied Basic Research Foundation (Grant no. 2020A1515010197). The authors thank all members of the Jiang Laboratory, especially Xueling Peng and Renxing Xue for support and advice. They also thank Yong Zhao, Zhou Songyang, Weihua Xu, and Wenqing Zhang Laboratories for advice and technical support, and State Key Laboratory of Biocontrol (SYSU) for microscopy equipment and support.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.