lncTUG1/miR-144-3p affect the radiosensitivity of esophageal squamous cell carcinoma by competitively regulating c-MET

A total of 50 paired tumor and adjacent normal tissues were retrospectively collected from 50 patients with ESCC. All of the patients had primary, nondistant metastatic ESCC and had undergone complete surgical resection (esophagectomy) at the Cancer Hospital of the Chinese Academy of Medical Sciences (CAMS) between December 2014 and December 2018 after providing informed written consent and agreement. None of the patients received chemo- or radiotherapy prior to surgery. According to the National Comprehensive Cancer Network esophageal cancer guidelines, the normal tissues were at least 5 cm away from the primary lesions. All samples were stored at − 80 °C before further processing. This study was approved by the Medical Ethics Committee of the Cancer Hospital of the CAMS. The clinical characteristics of the patients are shown in Table 1.

Radiation sensitive and resistant samples were retrieved from Gene Expression Omnibus (GEO) repository (GSE61816 and GSE61772). Probes were annotated by the platform information stored in GEO. For gene with multiple probes, the expression value was calculated by averaging the expression values of its probes. To make data from different dastset comparable, the ComBat algorithm implemented in R package sva were used to adjust the batch effects and the batch were set as the different GEO series. R package limma was used to identify the differential expressed genes (DEGs). The design model were generated by “model.matrix(~ 0+ Resistance/Sensitive)”.

Human esophageal epithelial cells (Het-1A) and ESCC cell lines (TE-13, KYSE140, EC9706, and KYSE30) were purchased from the Cell Bank of Type Culture Collection of Chinese Academy of Sciences (Shanghai, China) and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (Gibco, USA) in a 37 °C incubator with 5% CO₂.

Total RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA). cDNAs were synthesized with a reverse transcription kit (Invitrogen). qRT-PCR analysis was performed with SYBR Premix Ex Taq II (TaKaRa, Dalian, China). For mRNA and miRNA, GAPDH and U6 were used as internal controls, respectively. The primers are shown in Table 2.

After reaching 40–50% confluence, cells were transfected with a small interfering RNA (siRNA) targeting TUG1 (si-TUG1), a miR-144-3p mimic, a miR-144-3p inhibitor, si-MET, LV-TUG1 and a nonspecific control (Invitrogen, Shanghai, China) by using Lipofectamine 3000 (Invitrogen, USA).

Luciferase reporter gene vectors (pRL-TK, Promega) containing wild type (WT) or mutant (Mut) lncTUG1 and the 3′-UTR of WT or Mut MET were transfected into HEK293T cells. The miR-144-3p mimic, miR-144-3p inhibitor or negative control (NC) was cotransfected with reporter plasmids for 48 h. Relative luciferase activity was determined using a Dual-Luciferase Reporter Assay System (Promega).

A total of 5000 cells were seeded into a 96-well plate for 24 h, and then cells were exposed to 2 Gy radiation (once). After radiotherapy, cell viability was evaluated by the MTT assay at 0, 24, 48, 72 and 96 h. A range of radiation doses (0, 2, 4, 6 and 8 Gy) was applied in a dose-dependent experiment.

Five hundred cells were seeded into a 6-well plate with or without 2 Gy radiation. After two weeks, the cells were fixed and stained with 0.1% crystal violet solution. The numbers of colonies were counted under an inverted microscope.

EC9706 and KYSE30 cells were harvested at 48 h posttransfection. An Annexin V-FITC/PI Apoptosis Detection Kit (Sigma-Aldrich, St. Louis, MO, USA) was utilized to detect cell apoptosis according to the manufacturer’s instructions, and the percentage of apoptotic cells was calculated using a Beckman Coulter FACS flow cytometer (Beckman Coulter).

The cells were lysed in RIPA buffer (Sigma-Aldrich). After centrifugation, the protein was extracted, and the concentration was quantified using a BCA assay (Pierce, Rockford, IL, USA). Then, protein samples were separated by 10% SDS-PAGE and transferred onto polyvinylidene fluoride (PVDF) membranes (Amersham Pharmacia, Little Chalfont, UK). The primary antibodies used were anti-c-MET (1:1000, Thermo Fisher Scientific), anti-EGFR (1:2500, Invitrogen), anti-t-AKT (1:2000, Cell Signaling), anti-p-AKT (1:500, Invitrogen), and anti-GAPDH (1:1000, Invitrogen), and a secondary horseradish peroxidase (HRP)-conjugated antibody (Invitrogen) was used. GAPDH was chosen as the internal loading control.

A Magna RIP™ RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA) was used for RIP experiments according to the manufacturer’s instructions. The TUG1 level was detected by qRT-PCR.

Twenty male BALB/c nude mice (age, 6 weeks; sex, male; weight, 20 g) were obtained by the Cancer Hospital of the CAMS and maintained in a pathogen-free animal facility at 24 °C with access to distilled food and water. A total of 3 × 10⁶ transfected (LV-NC or LV-TUG1) KYSE30 cells were subcutaneously injected into six-week-old male nude mice (n = 5 per group). The mice were given radiation (2 Gy) for 5 consecutive days when the tumors reached an average volume of approximately 100 mm³. Tumor volume was measured every three days according to the following formula: volume = 1/2 × length × width². All animal procedures were performed following approval from the Animal Care and Use Committee of the Cancer Hospital of the CAMS.

All tissues were cut into 4-μm sections. The sections were incubated with an anti-Ki67 antibody (1:200, Abcam, Cambridge, UK), MET antibody (1:200, GeneTex, GTX50668) and p-AKT antibody (1:200, GeneTex, GTX128414) at 4 °C overnight. Then, biotinylated secondary antibodies were incubated for 1 h at room temperature and visualized with diaminobenzidine substrate (Sigma-Aldrich, St. Louis, MO, USA). Immunohistochemistry (IHC) images were taken using an Olympus microscope.

To identify candidate genes that are associated with ESCC radioresistance, we performed a bioinformatics analysis using published expression data (Fig. 1a). Briefly, two data series consisting of two esophageal cancer cells and their derived radioresistant cell lines were obtained from the Gene Expression Omnibus (GEO) database (i.e., GSE61620, and GSE61772). Differential expression analysis was then performed between radioresistant and radiosensitive cell lines under different irradiation conditions using the normalized microarray data, which identified 341 genes that were significantly upregulated and 594 genes that were significantly downregulated in the radiosensitive cell lines compared with the radioresistant cell lines (P < 0.05; Fig. 1b). As shown in Fig. 1c and d, lncRNA-TUG1 was one of the most upregulated molecules, suggesting that it might play a role in the development of radiotherapy resistance in ESCC. To investigate the biological function of lncRNA-TUG1, genes whose expression levels were tightly correlated (absolute Pearson’s correlation coefficient value > 0.9) with that of the molecule in the cell lines were selected as the input for Metascape pathway analysis [26]. These genes were significantly enriched in meaningful cancer-related processes or pathways, such as the ‘Cell Cycle’ and ‘Transcriptional Regulation by TP53’ (Fig. 1e). To facilitate illustration, Circos was used to visualize the genes related to lncRNA-TUG1 expression in the GO0044772 term (Fig. 1f). To further explore the potential mechanism of lncTUG1 in radioresistance, RAID v2.0 was used to identify molecules that interact with lncTUG1 [27]. Indeed, we observed that hsa-miR-144-3p and hsa-miR-145-5p achieved the highest confidence scores among all types of interactors. The target prediction information is shown in Fig. 1g. Below, we focus on the relationship between lncTUG1 and hsa-miR-144-3p.

Based on the above results, we first examined the expression levels of lncTUG1 and its potential interacting miRNA, miR-144-3p, in ESCC and paired normal tissues. As shown in Fig. 2a-d, lncTUG1 was highly expressed in tumor tissues, while miR-144-3p was weakly expressed. In line with this result, the expression of lncTUG1 was increased in ESCC cell lines compared with normal esophageal cell lines (Fig. 2e). Moreover, decreased miR-144-3p expression was also observed in all the analyzed ESCC cell lines (Fig. 2f). Pearson’s correlation analysis confirmed that the expression of lncTUG1 was inversely correlated with that of miR-144-3p in both tissues and cell lines (Fig. 2g).

To investigate the functional role of lncTUG1 in tumorigenesis, we silenced lncTUG1 expression in EC9706 and KYSE30 cells with an siRNA. As shown in Fig. 3a, si-TUG1 was successfully transfected into the cell lines, and endogenous lncTUG1 was significantly suppressed. Then, the relative cellular abilities of proliferation, migration and invasion were examined. si-TUG1 retarded the growth of EC9706 and KYSE30 cells according to the MTT assay (Fig. 3b and c), and the results of the colony formation assay were largely consistent (Fig. 3d). Moreover, the low level of lncTUG1 expression led to a downward trend in both migration and invasion (Fig. 3e and f). In contrast, lncTUG1 knockdown increased the quantity of apoptotic cells (Fig. 3g). Taken together, these results indicate that lncTUG1 is a potential oncogenic factor that influences the progression of ESCC.

Because the analysis was performed in the radiotherapy samples, the mechanism by which lncTUG1 affects the radiosensitivity of ESCC cells was the point of interest. The level of lncTUG1 was examined in a time- and dose-dependent manner in EC9706 and KYSE30 cells. Both the dose and time affected the expression level of lncTUG1 (*P < 0.05, Fig. 4a and b). More importantly, si-TUG1 combined with 2 Gy radiation showed increased radiation sensitivity in ESCC cells. The MTT assay indicated that this combined treatment had significant inhibitory effects on cell proliferation (*P < 0.05, Fig. 4c). EC9706 and KYSE30 cell colonies were inhibited dramatically by lncTUG1 knockdown plus 2 Gy radiation (Fig. 4d). Moreover, this combined treatment further induced the apoptosis of EC9706 and KYSE30 cells (*P < 0.05, Fig. 4e). Besides these results, in Fig. 4f, we conducted the colony formation assay of EC9706 and KYSE30 cells exposed to the different radiotherapy doses (0, 2, 4, 6, 8 Gy). As shown in Fig. 4g, the cellular survival curves of EC9706 and KYSE30 cells indicated the knockdown of lncTUG1 indeed enhance the radiosensitivity, and the relative relative radiosensitizaion effects data were shown in Table 3. All these results indicate that lncTUG1 is involved in ESCC radiotherapy and affects radiosensitivity.

miR-144-3p was chosen as a target based on the previous prediction. The luciferase system indicated that only the miR-144-3p mimic decreased the luciferase activity of WT-TUG1 but had no effect on Mut-TUG1 (Fig. 5a). The lncTUG1 level was affected by the miR-144-3p level (Fig. 5b). Moreover, the RIP assay further confirmed that lncTUG1 was significantly augmented by Ago2 but not IgG (Fig. 5c). To further identify the target of hsa-miR-144-3p, the 3′-UTR of MET with the potential binding site was examined (Fig. 5d). The luciferase reporter system showed that the luciferase activity of the 3′-UTR of only WT MET was decreased (Fig. 5e). Both the protein and mRNA levels of MET were affected by the miR-144-3p level (Fig. 5f and g). Moreover, the miR-144-3p inhibitor reversed the si-TUG1 effect on the protein level of MET (Fig. 5h). Based on these results, we conclude that the lncTUG1/ miR-144-3p/MET axis indeed exists.

According to the above results, lncTUG1 affects ESCC progression via the miR-144-3p/MET axis. We further determined whether lncTUG1 affects radiosensitivity through miR-144-3p and MET. As shown in Fig. 6a and b, colony formation and apoptosis assays confirmed that the miR-144-3p inhibitor restored the effect of lncTUG1 knockdown on radiotherapy. Moreover, MET knockdown decreased the level of EGFR and lowered the phosphorylation level of AKT (Fig. 6c). It is possible that the p-AKT level is the key factor in ESCC radiotherapy.

Finally, we aimed to explore the effect of lncTUG1 on the radiosensitivity of ESCC tumor tissue. We subcutaneously injected transfected (LV-NC or LV-TUG1) KYSE30 cells into BALB/c nude mice to establish an in vivo model. The findings indicated that lncTUG1 knockdown could enhance the effect of radiotherapy on ESCC in vivo (Fig. 7a, b and c). All of these results exhibited the smallest tumor volume, the slowest tumor growth and the lightest tumor weight when the LV-TUG1 KYSE30 underwent 2 Gy radiation in this xenograft model. Meanwhile we also detected the downstream target expression. The tumor level of Ki67 was also reduced dramatically in the sh-TUG1 plus 2 Gy group (Fig. 7d). The tumor level of MET and p-AKT had the same trend that the lowest level of MET and p-AKT in the sh-TUG1 plus 2 Gy group (Fig. 7e and f).