Linc-ROR promotes esophageal squamous cell carcinoma progression through the derepression of SOX9

ESCC cell lines Eca109, EC9706, KYSE150, and TE-1 and embryonic kidney cell line 293 T were purchased from the Cell Bank of Type Culture Collection of Chinese Academy of Sciences (Shanghai Institute of Biochemistry and Cell Biology). All cells were maintained at 37 °C in a humidified incubator under 5% CO₂ atmosphere.

Fourteen pairs of primary ESCC and adjacent non-tumor esophageal tissues were obtained from patients undergoing esophagectomy without prior radiotherapy or chemotherapy at Kashgar Hospital from 2013 to 2015. This study was approved by the Ethics Committee of Shihezi University School of Medicine, and all participants were enrolled with written informed consent. Clinical characteristics of all patients are summarized in Additional file 1 Table S1.

Eca109 cells were seeded at 1.5 × 10³ cells/well in six-well ultra-low cluster plates (Corning) and cultured in DMEM/F12 serum-free medium (Gibco) supplemented with 2% B27 (Invitrogen), 20 ng/mL EGF (PeproTech), and 20 ng/mL bFGF (PeproTech) for 10 days. Subsequently, tumorspheres with diameter larger than 75 μm were counted.

siRNA against human linc-ROR (coding region sense was 5′-GGAGAGGAAGCCUGAGAGUdTdT-3′, antisense was 5′-ACUCUCAGGCUUCCUCUCCdTdT-3′) and SOX9 coding region sense was 5′-CGCUCACAGUACGACUACAdTdT-3′, antisense was 5′-UGUAGUCGUACUGUGAGCGdTdT-3′), miRNA mimics, antisense miRNA inhibitors, and negative scramble control RNA oligos were synthesized by GenePharma. RNA transfection was performed at a final concentration of 50 nM, using Lipofectamine 2000 Transfection Reagent (Invitrogen) according to the manufacturer’s instructions.

Total RNA was isolated from cultured cells or human samples using Total RNA Kit I (Omega Bio-tek) or miRNeasy FFPE Kit (Qiagen) according to the manufacturer’s protocols, respectively. cDNA was synthesized using reverse transcriptase, after which quantitative real-time PCR (qRT-PCR) was performed using Fast SYBR qPCR mixture (CWBIO) with specific primers on 7500 Fast Real-Time PCR System (Applied Biosystems). GAPDH or U6 was used as an internal control. Primer sequences are listed in Additional file 1 Table S2.

Cell proliferation was measured with Cell Counting Kit-8 (CCK-8) (Dojindo). Different groups of cells were plated at 3 × 10³ cells/well in 96-well plates. At 0, 24, 48, and 72 h post-plating, CCK-8 solution was added to each well, and the absorbance at 450 nm (OD₄₅₀) was measured after incubation for 40 min at 37 °C.

After 24 h transfection with various RNA oligos, cells were seeded in six-well plates and cultured for 10 days. After fixation with methanol for 15 min, the colonies were stained with 0.1% crystal violet in 20% methanol and counted.

Migration and invasion assays were carried out using a Transwell chamber (Corning). Cells with serum-free medium were seeded in the top chamber without or with an insert coated with Matrigel (BD Biosciences) for migration assay and invasion assay, respectively. The lower chamber was filled with medium with 20% FBS as chemoattractant. After incubation for 24 h, the cells that have migrated or invaded through the membrane and the cells on the lower surface of the member were fixed, stained, and counted.

293 T cells were seeded at 3 × 10⁴ cells/well in 24-well plates and allowed to settle for 24 h. Subsequently, cells were transfected with various RNA oligos together with pmiR-REPORT-SOX9 3′-untranslated region (3′-UTR) reporter plasmid and a Renilla luciferase vector. Forty-eight hours after transfection, relative luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega) and normalized against Renilla luciferase activity.

RNA-binding protein immunoprecipitation (RIP) experiments were performed using 10 μg Ago2 antibody (ab5072; Abcam) and the Magna RIP Kit (Millipore) in accordance to the manufacturer’s instructions. Co-precipitated RNAs were then analyzed using qRT-PCR analysis.

Equal amounts of cell lysates were electrophoretically separated and transferred to a PVDF membrane. After blocking with 5% skimmed milk, the membrane was incubated with primary antibodies against SOX9 (AB5535, 1:2000; Millipore), E-cadherin (sc-8426, 1:100; Santa Cruz), vimentin (sc-6260, 1:100; Santa Cruz), β-actin (sc-7963, 1:100; Santa Cruz), and appropriate secondary antibodies. The signals were detected using enhanced chemiluminescence (GE Healthcare).

All animal experiments were approved by the Research Ethics Committee of the First Affiliated Hospital of Shihezi University School of Medicine. Five-week-old female athymic BALB/C nude mice were purchased from Beijing Vital River Laboratory Animal Technology. ESCC xenografts were established in the flanks by subcutaneously inoculating 1 × 10⁶ EC9706 cells. Mice bearing established xenografts were treated with 5 nM of cholesterol-conjugated linc-ROR siRNA (Chol-silinc-ROR) or negative control with 100 μL PBS by locally injecting into the tumor mass every 3 days for a total of six times. Three days after the last injection, the mice were sacrificed, and the tumors were excised and weighed. Tumor volumes were calculated according to the formula: length × width²/2. Subsequently, tumors were fixed and paraffin embedded. Sections of 4 μm were cut and subjected to hematoxylin–eosin or immunohistochemistry staining [31] using anti-SOX9 (AB5535, 1:600; Millipore), anti-CD44 (EP44, 1:50; ZSGB-BIO), and anti-vimentin (OTI5D7, 1:600; ZSGB-BIO) antibodies. Negative controls were prepared by replacing the primary antibodies with PBS.

To determine whether linc-ROR expression correlates with SOX9 expression in ESCC, their gene expression levels were determined in 14 paired ESCC and adjacent esophageal tissue samples (Fig. 1a). Interestingly, a significant positive correlation was observed between the relative expression of linc-ROR and SOX9 in ESCC samples when compared to their matched non-tumor tissues (r = 0.562, P = 0.036; Fig. 1b). Subsequently, we examined the expression levels of linc-ROR and SOX9 in a panel of ECSS cell lines (Eca109, EC9706, KYSE150, and TE-1). Cell lines (EC9706 and KYSE150) with higher expression of linc-ROR also exhibited higher levels of SOX9 (Fig. 1c). Accordingly, we found that knockdown of endogenous linc-ROR by siRNA resulted in a significant reduction of SOX9 expression in EC9706 cells (Fig. 1d). Consistent with these observations, the expression of linc-ROR, as well as the known stem/progenitor cell marker SOX9 [32], were higher when Eca109 cells were grown as tumorspheres (stem like) in three-dimensional suspension cultures than when grown as two-dimensional adherent (differentiated) cultures (Fig. 1e). Collectively, these results indicated a positive correlation between linc-ROR and SOX9 expression in ESCC.

The observation that linc-ROR and SOX9 are coordinately expressed in ESCC tissues led to the hypothesis that linc-ROR might promote stem cell-like properties and cancer progression through the regulation of pluripotency transcription factor SOX9. To determine this possibility, we first transfected EC9706 cells with siRNA to efficiently deplete the endogenously expressed linc-ROR. The growth curves detected by CCK8 showed that linc-ROR repression significantly decreased cell proliferation (Fig. 2a). Linc-ROR downregulation also substantially repressed colony-forming capacity as indicated by the formation of fewer and smaller colonies (Fig. 2b). As enhanced cancer cell motility and consequent invasion and metastasis have been associated with the gain of CSC properties and EMT [33], we investigated the effect of linc-ROR inhibition on these phenotypes. Knockdown of linc-ROR in EC9706 cells resulted in a significant attenuation of their migration ability and invasive potential compared to control groups using a Transwell system (Fig. 2c). Accordingly, the expression of epithelial cell marker E-cadherin was robustly increased in silinc-ROR-transfected cells, whereas the expression of mesenchymal marker vimentin was significantly reduced (Fig. 2d), indicating that linc-ROR can promote EMT in ESCC cells and hence facilitates their mobility. Chemoresistance is one of the major characteristics of CSC [34]. We found that linc-ROR repression resulted in the sensitization of EC9706 cells to cisplatin, which is one of the most frequently used chemotherapeutic drug for esophageal cancer [6] (Fig. 2e). With non-adherent sphere formation assay, the tumorsphere-forming capacity was significantly decreased following inhibition of linc-ROR in Eca109 cells compared to scramble control (Fig. 2f). Furthermore, the combination of linc-ROR knockdown and cisplatin treatment synergistically reduced the number and the size of tumorspheres compared to that of cisplatin treatment alone. The expression of stemness-associated genes, including CD44, KLF4, NANOG, OCT4, and SOX2 was also markedly decreased by linc-ROR knockdown in concert with the reduced sphere-forming ability (Fig. 2g). These results suggest that linc-ROR promotes the acquisition of CSC-like properties in ESCC.

Increasing evidence suggests that linc-ROR participates in the regulatory network of ceRNA [19, 35]. We hypothesized that endogenous linc-RoR also function as a ceRNA of SOX9 in ESCC. We used a target prediction tool based on RNA22 [36] to search for miRNAs that target the full-length transcripts of linc-RoR and bioinformatics tools miRWalk 2.0 [37] and miRanda [38] to search for miRNAs that target the 3′-UTR of SOX9 mRNA. Twelve miRNAs whose predicted binding sites were shared by linc-RoR and SOX9 emerged (Additional file 1 Table S3). Mimics of these miRNAs were transfected into EC9706 cells; miR-15b, miR-33a, miR-129, miR-145, and miR-206 exhibited the most significant inhibitory effect on the expression of linc-ROR and SOX9 (Fig. 3a). To assess the potential relationship between linc-ROR, SOX9, and these miRNAs, we transfected Eca109 cells with the inhibitor of miR-145; the inhibitor mixture of miR-15b, miR-33a, and miR-129; and inhibitor cocktail of these five miRNAs, respectively. Inhibition of these five miRNAs resulted in the most remarkable upregulation of linc-ROR and SOX9 expression (Fig. 3b). Concomitantly, expression levels of stemness-associated genes were also decreased by miR-145 whereas the cocktail of miRNA inhibitors promoted their expression (Fig. 3c). Thus, these five miRNAs were pursued as candidates for studies in detail. Their predicted binding sites on linc-ROR transcript and SOX9 3′-UTR are shown in Fig. 3d and e. Moreover, these miRNAs exhibited decreased levels in the ESCC samples where linc-ROR expression was upregulated compared to their matched adjacent tissues (Additional file 2 Figure S1). Previous studies have demonstrated that miRNAs are present in the form of miRNA ribonucleoprotein complexes that contain Ago2, a key component of RNA-induced silencing complexes [39]. To validate the direct binding ability of these predicted miRNAs to linc-ROR and SOX9, RIP analysis was performed on EC9706 cells that were treated with the inhibitor cocktail of all candidate miRNAs or scramble control using Ago2 antibody. Linc-ROR and SOX9 were present in the Ago2 immunoprecipitates detected by qRT-PCR, and their levels were drastically reduced in Ago2 complexes purified from cells treated with miRNA inhibitor mixture (Fig. 3f), indicating that linc-RoR and SOX9 were recruited to the Ago2-containing–RNA-induced silencing complexes and functionally interacted with miRNAs. For further confirmation, we constructed a luciferase reporter containing Sox9 3′-UTR and co-transfected that with linc-ROR siRNA, cocktail of miRNA inhibitors, or miR-145 mimics into 293 T cells. Results of dual-luciferase reporter assay showed that linc-ROR repression or miR-145 overexpression significantly reduced the luciferase activities of the SOX9 3′-UTR reporter, whereas the treatment with the inhibitor cocktail of all candidate miRNAs increased the luciferase activity (Fig. 3g, bottom). The protein levels of SOX9 were also significantly decreased after knockdown of linc-ROR or overexpression of miR-145 (Fig. 3g, top). Collectively, these data suggest that linc-RoR and SOX9 share multiple regulatory miRNAs, and these candidate miRNAs negatively regulate linc-RoR and SOX9 directly.

To gain insight into the functional relevance of these candidate miRNAs, we examined the impacts of these miRNAs on cell proliferation, cell motility, and chemoresistance by antagonizing endogenous miR-15b, miR-33a, miR-129, miR-145, and miR-206 using specific antagomirs. CCK8 proliferation assay revealed that treatment with inhibitor cocktail of all candidate miRNAs promoted EC9706 cell proliferation, whereas co-transfection of miRNA inhibitor mixture and linc-ROR siRNA showed that the inhibition of these miRNAs could abrogate the suppressed cell proliferation caused by linc-ROR downregulation (Fig. 4a). Similarly, treatment with miRNA inhibitor cocktail promoted colony formation and partially abolished the inhibitory effect of linc-ROR knockdown (Fig. 4b). The migration ability (Fig. 4c) and invasive potential (Fig. 4d) of EC9706 cells were also increased following depletion of candidate miRNAs. Inhibition of these miRNAs in silinc-ROR-treated cells partially counteracted with linc-ROR silencing and promoted cell migration and invasion. Moreover, the increased chemosensitivity to cisplatin through knockdown of linc-ROR in EC9706 cells was abrogated upon inhibition of these candidate miRNAs (Fig. 4e). We further evaluated whether overexpression of candidate miRNA could potentiate the antitumor effects of linc-ROR repression. As expected, miR-145 functioned as tumor suppressor and reduced cell proliferation, colony formation, migration and invasion, and chemoresistance (Additional file 3 Figure S2). Combining silinc-ROR and miR-145 mimics synergistically inhibited colony formation and cell mobility in EC9706 cells compared to those of silinc-ROR or miR-145 mimics alone. Taken together, these observations suggest that these candidate miRNAs have potential antitumor effects through diminishing the function of linc-ROR on ESCC progression.

To determine whether linc-ROR/miRNA axis-modulated proliferation, migration and invasion capacities, and chemoresistance were mediated by the target gene SOX9, we knocked down SOX9 expression using siRNA and investigated its effect on these phenotypes. Knockdown of SOX9 in EC9706 cells resulted in significantly reduced cell proliferation; cocktail of miRNA inhibitors-promoted cell proliferation was abrogated by additional repression of Sox9 (Fig. 5a). Similarly, silencing of SOX9 markedly reduced the enhancing effects of miRNA inhibitors on colony formation (Fig. 5b), migration and invasion capacities (Fig. 5c and d), and chemoresistance (Fig. 5e). These results collectively suggest that linc-ROR/miRNA axis-mediated acquisition of CSC properties is at least partly through SOX9 in ESCC.

To provide the proof for the concept of the translational relevance of our findings, we performed in vivo therapeutic intervention studies using cholesterol-conjugated linc-ROR siRNA in established ESCC xenograft models. Mice bearing EC9706-derived xenografts were intratumorally injected with cholesterol-conjugated linc-ROR siRNA every 3 days for six times and sacrificed 3 days later (Fig. 6a). This treatment effectively inhibited tumor growth as measured by tumor volume and tumor weight compared to the control group (Fig. 6b–d). Furthermore, the level of SOX9 as well as CSC marker CD44 and mesenchymal marker vimentin in tumor xenografts was dramatically diminished by the treatment (Fig. 6e). These observations indicate that inhibition of linc-ROR expression decreases SOX9 activity in ESCC and as a result attenuates tumor growth.