LncRNA Sox2OT-V7 promotes doxorubicin-induced autophagy and chemoresistance in osteosarcoma via tumor-suppressive miR-142/miR-22

Doxorubicin (Dox) is one of the most commonly used chemotherapeutic drugs for osteosarcoma (OS) treatment. In the present study, we attempted to investigate the mechanism by which Sox2OT-V7 dysregulation affects Dox chemoresistance to provide a novel experimental basis for developing neoadjuvant therapy. Sox2OT-V7 expression is upregulated in OS tissues, particularly in chemoresistant OS tissues, and in OS cell lines compared to controls. Dox treatment induces autophagy and Sox2OT-V7 expression in U2OS cells, and Dox-induced autophagy is partially attenuated by Sox2OT-V7 silencing. Knocking down Sox2OT-V7 or blocking autophagy in Dox-resistant U2OS/Dox cells resensitizes the cells to Dox treatment in vitro. Moreover, Sox2OT-V7 directly targets miR-142/miR-22 to inhibit their expression, and the effect of Sox2OT-V7 silencing on U2OS cell autophagy and U2OS/Dox cell sensitivity to Dox can be reversed by miR-142/miR-22 inhibition. Sox2OT-V7 silencing enhances the suppressive effects of Dox on U2OS/Dox cell-derived tumor growth in vivo, while miR-22 inhibition or miR-142 inhibition reverses the effects of Sox2OT-V7 silencing on Dox-induced suppression on tumor growth. Finally, miR-142 directly targets ULK1, ATG4A, and ATG5, while miR-22 directly targets ULK1 to inhibit the expression of the target gene; The Sox2OT-V7/miR-142/miR-22 axis modulates autophagy in OS cells by regulating ULK1, ATG4A, and ATG5.


INTRODUCTION
Osteosarcoma (OS) is one of the most common primary bone tumors in children and adolescents, OS originates from the malignant transformation of mesenchymal cells and has a high mortality rate [1]. Doxorubicin (Dox) is one of the first line chemotherapeutic drugs for OS [2]. However, the use of Dox at a low-dose not only reduces its effectiveness but also leads to drug resistance, while dose increase would cause severe cardiotoxicity [3], leading to limits on clinical application. Therefore, there is an urgent need to investigate the mechanism of Dox chemoresistance and to develop neoadjuvant therapy.
Dox treatment can induce pro-survival autophagy in cells [4][5][6]. Autophagy is the primary pathway involved in the degradation of proteins and organelles, cellular remodeling, and survival during nutrient starvation [7]. Autophagy is a critical regulator of cellular AGING homeostasis, promoting the controlled degradation of cytoplasmic material both at steady-state and during nutrient deprivation [8]. Studies have shown the critical roles of autophagy in OS chemoresistance to Dox [9,10]. In our previous study, we also revealed that Epigallocatechin gallate (EGCG) reduced doxorubicininduced pro-survival autophagy by decreasing Sox2OT (SOX2 overlapping transcript) lncRNA transcript variant 7 (Sox2OT-V7) to improve the growth inhibition of Dox [11]. Since lncRNA Sox2OT-V7 is overexpressed in OS tissues and cell lines, investigating the specific function and mechanism of Sox2OT-V7 in autophagy and the resistance of OS cells to Dox may provide new directions for combating OS chemoresistance.
LncRNAs can exert their biological functions by serving as sponges for miRNAs to counteract miRNAmediated suppression of miRNA downstream targets [12]. To investigate the molecular mechanism of Sox2OT-V7 in Dox-induced autophagy in OS, we searched TCGA online data for candidate miRNAs related to Sox2OT-V7 and autophagy. As shown in Supplementary Table 2, some miRNA, including miR-22 and miR-142, were negatively correlated with Sox2OT-V7 expression in OS (Supplementary Figure  1). miR-22 regulates 5-FU sensitivity by inhibiting autophagy and promoting apoptosis in colorectal cancer cells [13], while miR-142 targets ATG5/ATG16L1 to sensitize hepatocellular carcinoma cells to sorafenib by inactivating autophagy [14]. Moreover, online tools indicate that Sox2OT-V7 may target miR-142 and miR-22. Thus, we hypothesize that Sox2OT-V7 may exert its effect on OS by regulating miR-142 and miR- 22. In the present study, the expression of Sox2OT-V7 in OS tissue samples and cell lines was examined. The effect of Sox2OT-V7 on autophagy-related markers and autophagy flux was evaluated. The effects of Sox2OT-V7 silencing on Dox-resistant U2OS/Dox cell viability and U2OS/Dox cell-derived tumor growth were also examined. Next, the predicted binding between Sox2OT-V7 and miR-142/miR-22, between miR-142/miR-22 and ULK1, ATG4A, and ATG5 was validated. In addition, the dynamic effect of Sox2OT-V7 and miR-142/miR-22 on Dox suppression on U2OS/Dox cell viability, and the dynamic effect of Sox2OT-V7, miR-142/miR-22, ULK1, ATG4A, and ATG5 on autophagy-related genes and autophagy flux was evaluated. Finally, the expression of these factors and their correlations in tissue samples were determined. In summary, we provide a novel mechanism by which Sox2OT-V7 modulates autophagy in OS cells, therefore affects chemoresistance to Doxbased therapy from the perspective of lncRNA-miRNA-mRNA regulation.

Expression of Sox2OT-V7 in tissue samples and cell lines and its correlation with autophagy markers
Since we have revealed that Sox2OT-V7 contributes to the chemoresistance of OS to doxorubicin via autophagy [11], in the present study, the expression of Sox2OT-V7 was first examined in tissue samples and cell lines. Consistent with our previous research, Sox2OT-V7 expression was significantly upregulated in OS cell lines, compared to a normal osteoblast cell line, hFOB ( Figure  1A), and expression was most upregulated in U2OS cells. Similarly, Sox2OT-V7 expression was upregulated in OS tissue samples, compared to noncancerous tissue samples ( Figure 1B) and was higher in chemo-resistant tissues than in chemosensitive tissues ( Figure 1C).
As for autophagy markers, the mRNA level of Beclin 1 was higher in chemoresistant tissues than in chemosensitive tissues ( Figure 1D). IHC staining showed that the Beclin 1 and LC3II levels were increased in chemoresistant tissues ( Figure 1F). Moreover, Sox2OT-V7 expression was positively correlated with Beclin 1 ( Figure 1E). These data indicate that Sox2OT-V7 is overexpressed in OS, and positively correlated with autophagy markers.

Dox treatment induces autophagy and Sox2OT-V7 expression in U2OS cells
Next, we examined whether Dox induces autophagy and Sox2OT-V7 expression in U2OS cells. U2OS cells were treated with Dox (1, 2.5, and 5 µM) or rapamycin (2 µM), a representative autophagic agonist, and examined for autophagy markers, autophagy flux, and Sox2OT-V7 expression. As shown by immunoblotting, Dox significantly increased LC3II and Beclin 1 protein levels in a dose-dependent manner ( Figure 2A). The inducible effect of 5 µM Dox on autophagy was close to that of 2 µM rapamycin ( Figure 2A). Simultaneously, the protein levels of autophagy substrate p62 were significantly decreased by 2.5 and 5 µM Dox and 2 µM rapamycin (Figure 2A), indicating that autophagy was enhanced by 2.5 and 5 µM Dox treatment and 2 µM rapamycin.
U2OS cells with stable eGFP-LC3 expression were used to confirm Dox-induced autophagy using rapamycin treatment as a positive control. Figure 2B shows that the autophagy in U2OS cells was increased by Dox and rapamycin treatment. As shown in Figure 2C, green puncta were increased after exposure to Dox in U2OS cells with stable eGFP-LC3 expression; this result was similar to that of rapamycin. Moreover, Sox2OT-V7 expression was induced by Dox treatment in a concentration-dependent manner ( Figure 2D), further AGING indicating that Dox treatment induced autophagy and Sox2OT-V7 expression in U2OS cells.
Autophagy might be one mechanism by which neoplastic cells maintain survival upon chemotherapy treatment. Under such conditions, blocking autophagy could trigger apoptosis, therefore enhancing the curative effects of these chemotherapies and reducing chemoresistance [15,16]. Here, we constructed a Doxresistant U2OS cell line, U2OS/Dox, by treating AGING original U2OS cells with increasing concentrations of Dox (data not shown). Next, we treated U2OS/Dox cells with a series of concentrations of Dox (0.001, 0.01, 0.1, 1, 10, and 100 µM) in the presence or absence of the autophagy inhibitor 3-MA and examined cell viability to investigate whether the autophagy inhibitor would affect the effects of Dox on U2OS cells. As shown in Figure  2E, 3-MA treatment significantly (**P<0.01) enhanced the suppressive effects of Dox on U2OS cell viability, as manifested as the IC50 values reduced from 4.326 to 2.069 µM. These data suggest that autophagy inhibitors might resensitize U2OS cells to Dox treatment. To investigate the specific function of Sox2OT-V7 on Dox-induced autophagy in OS cells, we achieved Sox2OT-V7 silencing by infecting cells with Lsh-Sox2OT-V7, and confirmed the efficiency by qPCR (Supplementary Figure 1). Next, Sox2OT-V7 silenced U2OS cells were treated with Dox and examined for autophagy-related factors. As shown by immunoblotting, Dox-induced LC3II and Beclin 1 protein levels were both significantly decreased, while Dox-suppressed p62 protein was partially rescued by Sox2OT-V7 silencing ( Figure 3A-3D). Consistent results were observed in OS cells with stable eGFP-LC3 expression, in which the green puncta were reduced by Sox2OT-V7 silencing ( Figure 3E).
To further confirm the function of Sox2OT-V7 on Doxinduced autophagy in U2OS cells, we cotreated Sox2OT-V7 silenced U2OS cells with Dox and 3-MA and then examined autophagy-related factors. As revealed by the protein levels of LC3II, Beclin 1 and p62, Dox-induced autophagy in U2OS cells could be partially attenuated by Sox2OT-V7 silencing or 3-MA treatment alone and markedly suppressed by a combination of Sox2OT-V7 silencing and 3-MA treatment ( Figure 3F-3I). These data indicate that Sox2OT-V7 silencing serves as an autophagy inhibitor in U2OS cells.

LncRNA Sox2OT-V7 silencing promotes the sensitivity of OS cells to Dox
After confirming that Sox2OT-V7 silencing could block autophagy in U2OS cells, we continued to investigate whether lncRNA Sox2OT-V7 silencing could resensitize OS cells to Dox treatment. As shown in Figure 4A, under Dox treatment, the IC50 value of the original U2OS was 1.238 µM, while that of U2OS/Dox was upregulated to 4.449 µM, indicating that U2OS/Dox cells are resistant to Dox treatment. In U2OS/Dox cells, the expression of lncRNA Sox2OT-V7 was significantly upregulated, compared to that in the original U2OS cells ( Figure 4B). After generating lncRNA Sox2OT-V7 silencing in U2OS/Dox cells, the suppressive effects of Dox on U2OS/Dox cell viability were significantly (**P<0.01) enhanced, as manifested by a reduced IC50 value: 4.380 µM for non-transfected U2OS/Dox cells, 4.250 for Lsh-NC-transfected U2OS/Dox cells, and 1.533 for Lsh-Sox2OT-V7-transfected U2OS/Dox cells ( Figure 4C). Moreover, the apoptosis percentage of U2OS/Dox cells was significantly lower than that of the original U2OS cells ( Figure 4D). After silencing lncRNA Sox2OT-V7 in U2OS/Dox cells, the apoptosis rate was increased compared to that in the non-transfected cells and Lsh-NC-transfected cells ( Figure 4E).

MiR-142 and miR-22 are inhibited both by Dox treatment and negatively correlated with Sox2OT-V7
As we have mentioned, miR-142 and miR-22 were selected for further experiments because of their correlations with Sox2OT-V7 and autophagy [13,14]. miR-142 and miR-22 expression were both significantly lower in chemo-resistant OS tissues than in normal tissues ( Figure 5A-5B). Consistent with data in the TCGA (Supplementary Figure 2), both miR-142 and miR-22, were negatively correlated with Sox2OT-V7 expression in tissue samples ( Figure 5C, 5D). In OS cells, Dox treatment significantly inhibited the expression of miR-142 and miR-22 ( Figure 5E).
To investigate the molecular function of miR-142 and miR-22, we achieved miR-142 and miR-22 expression by transfection of miR-142 or miR-22 mimics or inhibitor, as confirmed by qPCR ( Figure 5F, 5G). In Sox2OT-V7-silenced U2OS cells, both miR-142 and miR-22 expression were significantly upregulated ( Figure 5H). Next, a luciferase reporter assay was performed to validate the predicted binding of Sox2OT-V7, miR-142, and miR-22. Wild-type and mutant-type Sox2OT-V7 reporter vectors were constructed ( Figure  5I) and cotransfected into HEK293 cells with miR-142 or miR-22 mimics or inhibitor and examined for luciferase activity. As shown in Figure 5J, the luciferase activity of wild-type vectors was significantly suppressed by miR-142 or miR-22 overexpression but enhanced by miR-142 or miR-22 inhibition. After mutating the predicted miR-142 or miR-22 binding sites, the alterations in the luciferase activity were abolished.
To further confirm the binding, RIP assays were performed. As shown in Figure 5K, Sox2OT-V7, miR-142, and miR-22 were associated with AGO2 in HEK293 cells. Sox2OT-V7, miR-142, and miR-22 levels were dramatically higher in the RNA derived from precipitated AGO2 protein than in that derived from IgG ( Figure 5K). We also performed a RIP assay in HEK293 cells transfected with control miRNA (miR-NC) or miR-142 or miR-22 mimics followed by real-time PCR to detect the association of Sox2OT-V7 with AGO2; the results shown in Figure 5L and 5M confirmed the interaction between Sox2OT-V7 and miR-142 and between Sox2OT-V7 and miR-22.
Next, the effect of miR-142 and miR-22 on the abovedescribed factors was investigated under Dox treatment. The protein levels of ULK1, ATG4A, and ATG5 were markedly decreased by miR-142 overexpression but increased by Dox treatment, and Dox-induced ULK1, ATG4A, and ATG5 upregulation was partially reversed by miR-142 overexpression ( Figure 7G, 7H). Similarly, ULK1 protein was significantly decreased by miR-22 overexpression but increased by Dox treatment, and Dox-induced ULK1 upregulation was partially reversed by miR-22 overexpression ( Figure 7I, 7J). These data indicate that miR-142 targets ULK1, ATG4A, and ATG5, and miR-22 targets ULK1 to regulate the expression of these critical autophagy-related factors upon Dox treatment.

The Sox2OT-V7/miR-142/miR-22 axis modulates autophagy in U2OS cells by regulating autophagyrelated genes
As we have mentioned, ULK1, ATG4A, and ATG5 are key autophagy-related factors [17,18,20]. Next, we investigated whether the Sox2OT-V7/miR-142/miR-22 axis could modulate autophagy in U2OS cells via these autophagy-related genes. U2OS cells were cotransfected with Lsh-Sox2OT-V7 and miR-142 inhibitor under Dox treatment, and the protein levels of ULK1, ATG4A, and ATG5 was examined. As shown in Figure 8A and 8B, the protein levels of ULK1, ATG4A, and ATG5 were significantly decreased by Sox2OT-V7 silencing but increased by miR-142 inhibition under Dox treatment, the effect of Sox2OT-V7 silencing was partially reversed by miR-142 inhibition. Next, U2OS cells were c-transfected with Lsh-Sox2OT-V7 and miR-22 inhibitor under Dox treatment and these factors were examined. Similarly, the protein levels of these autophagy-related genes were significantly decreased by Sox2OT-V7 silencing but increased by miR-22 inhibition under Dox treatment, and the effect of Sox2OT-V7 silencing was also partially reversed by miR-22 inhibition ( Figure 8C, 8D).
To further confirm these findings, we evaluated the effect of ULK1, ATG4A, and ATG5 on the autophagy markers LC3, Beclin 1, and p62, and performed IF staining to monitor autophagy flux upon Dox treatment. The silencing of ULK1, ATG4A, and ATG5 was achieved by transfection of si-ULK1, si-ATG4A, and si-ATG5, as confirmed by immunoblotting (Supplementary Figure 4). As shown in Figure 8E-8H, Dox-induced LC3II and Beclin 1 upregulation was partially suppressed, while Dox-suppressed downregulation of p62 could be partially rescued by the silencing of ULK1, ATG4A, or ATG5. For IF observation, the green puncta were reduced by the silencing of ULK1, ATG4A, or ATG5 ( Figure 8I), indicating that Dox-induced autophagy in U2OS cells was suppressed by the silencing of ULK1, ATG4A, or ATG5. Since the Sox2OT-V7/miR-142/miR-22 axis regulated the expression of ULK1, ATG4A, and ATG5 in U2OS cells, it can be concluded that the Sox2OT-V7/miR-142/miR-22 axis modulates autophagy in U2OS cells by regulating the autophagy-related genes ULK1, ATG4A, and ATG5.

DISCUSSION
In the present study, we revealed that Sox2OT-V7 expression is upregulated in OS tissues, particularly in chemoresistant OS tissues, and in OS cell lines compared to control. Dox treatment simultaneously induces autophagy and Sox2OT-V7 expression in U2OS cells, while Dox-induced autophagy can be partially attenuated by Sox2OT-V7 silencing. Moreover, Sox2OT-V7 directly targets miR-142 and miR-22 to inhibit their expression, and the effect of Sox2OT-V7 silencing on U2OS cell autophagy can be partially reversed by miR-142 or miR-22 inhibition. Finally, miR-142 directly targets ULK1, ATG4A, and ATG5, while miR-22 directly targets ULK1 to inhibit the expression AGING of the target genes; the Sox2OT-V7/miR-142/miR-22 axis modulates autophagy in U2OS cells by regulating ULK1, ATG4A, and ATG5 ( Figure 10).
Autophagy is a self-digestion and recycling mechanism used by cells to survive harsh environmental conditions. Autophagy and/or autophagy dysregulation are speculated to play various, sometimes conflicting roles in the development and progression of neoplasia [21,22]. The chemotherapeutic agent Dox induces autophagy in human osteosarcoma cell lines [23]. Autophagy contributes to chemoresistance in canine appendicular osteosarcoma and adding an autophagy inhibitor to the standard of care has the potential to improve outcome [15]. In the present study, we constructed a Dox-resistant U2OS cell subtype, U2OS/Dox, and consistently observed that autophagy inhibitor 3-MA could resensitize U2OS/Dox to Dox treatment.
The dysregulation of Sox2OT-V7 has been widely reported in cancers, including esophageal squamous cell carcinoma [24], hepatocellular carcinoma [25], gastric cancer [26], cholangiocarcinoma [27], and OS [11,28]. In the present study, we also observed significantly upregulated Sox2OT-V7 expression in OS cell lines and tissue samples, particularly chemoresistant tissue samples. In our previous study, we demonstrated that Dox-induced pro-survival autophagy was suppressed by decreasing Sox2OT-V7; therefore, the growth inhibition of Dox was improved [11]. Zhao D et al found that Dox treatment induced increased expression levels of cleaved caspase-3 and LC3, but reduced levels of p62 in osteosarcoma cells, and inhibition of autophagy notably enhanced the effects of Dox [23]. In the present study, we also found that Dox induced the ratio of LC3II/I and Beclin 1 protein while decreasing p62 protein, therefore causing autophagy flux. The upregulation of Sox2OT-V7 was also induced by Dox treatment. More importantly, the effect of Dox treatment on autophagy markers and autophagy flux was significantly attenuated by Sox2OT-V7 silencing, indicating that Sox2OT-V7 silencing may improve OS chemoresistance by acting on LC3, a core factor in the process of autophagy [29]. To further investigate the speculation that Sox2OT-V7 silencing may improve OS chemoresistance by modulating autophagy, in vitro and in vivo experiments were performed. As expected, knocking down Sox2OT-V7 resensitize U2OS/Dox cells to Dox treatment in vitro, similar to treatment with the autophagy inhibitor 3-MA. AGING Accumulating evidence suggests that lncRNAs act as miRNA sponges and inhibit miRNA functions [12]. Recently, lncRNA-associated ceRNA networks were revealed to play an essential role in many types of human cancer, including OS [30]. For example, lncRNA TUG1 promotes cell proliferation and suppresses apoptosis in osteosarcoma by regulating the miR-212-3p/FOXA1 axis [31]. LncRNA HOXA-AS2 promotes migration and invasion by acting as a ceRNA of miR-520c-3p in OS [32]. Here, we searched the TCGA database for miRNAs negatively correlated with Sox2OT-V7 and found that miR-142 and miR-22 may be direct downstream targets of Sox2OT-V7. In chemoresistant tissue samples, miR-142 and miR-22 expression were both significantly downregulated. Both miR-142 and miR-22 are regarded as inhibitory miRNAs in autophagy. miR-22 regulates 5-FU sensitivity by inhibiting autophagy and promoting apoptosis in colorectal cancer cells [13]. By targeting HMGB1, miR-22 inhibits HMGB1-mediated autophagy in OS cells, thereby acting as a tumor suppressor [33,34]. miR-142-3p overexpression increases the chemosensitivity of nonsmall-cell lung cancer by inhibiting HMGB1-mediated autophagy [35]. MiR-142-3p inhibits autophagy by targeting ATG16L1 in thymic-derived regulatory T cells [14,19]. As predicted by online tools and validated by luciferase reporter assays, miR-142 directly targets ULK1, ATG4A, and ATG5, and miR-22 directly targets ULK1.

CONCLUSIONS
lncRNA Sox2OT-V7 promotes Dox-induced autophagy and chemoresistance in OS via tumor-suppressive miR-142/miR-22 and downstream core Atg genes ( Figure 10). Upon further in vivo validation, these findings may provide new directions for combating OS chemoresistance to Dox-based therapies.

Clinical samples and chemoresistance evaluation
A total of 32 paired OS and nontumorous tissue samples were obtained from patients who received the same chemotherapy regimen before surgery and underwent complete resection surgery at The Second Xiangya Hospital with the approval of the Ethics Committee of The Second Xiangya Hospital, and written informed consent was obtained from all the patients. All the resected specimens were stored at -80 °C. According to the Huvos scoring system [45], the patients were classified as good responders (non-chemo-resistant) and poor responders (chemo-resistant).

Quantitative RT-PCR
Total RNA was extracted from cells using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA) following the protocols, and then the RNA was reverse transcribed using the PrimeScript RT Master Mix Perfect Real-Time kit (TaKaRa, Dalian, China) to obtain the cDNA. A real-time PCR assay was performed under the reaction conditions described in our previous study [11] using cDNA as the template. After the reaction, the data were subjected to statistical analysis. Relative gene expression was calculated using the 2 -ΔΔCT method, and GAPDH (for mRNA) or U6 (for miRNA) served as an internal control. The primer sequence was listed in Supplementary Table 1.

Immunohistochemical (IHC) analysis
Immunohistochemistry (IHC) was performed according to the indirect immunoperoxidase method. In brief, following deparaffinization, hydration and blockage of endogenous peroxidase, the specimens were incubated for 20 min with 10% nonfat milk in PBS to block specific sites and then individually incubated at 4°C overnight with the following primary antibodies: anti-LC3 (1:2000, ab48394, Abcam), anti-Beclin 1 antibody (1:400, ab207612, Abcam), and anti-Ki67 antibody (ab15580, Abcam). After rinsing, the slides were washed, and the sections were incubated with biotinylated goat anti-rabbit secondary antibody for 30 min at room temperature, and then incubated using the Vectastain ABC-AP kit (Vector Laboratories, Burlingame, CA, USA) for 30 min. Finally, the sections were washed and incubated with DAB substrate for 2-8 min.

Autophagy detection
Autophagy levels were detected using immunofluorescence staining and transmission electron microscopy. U2OS cells were transduced with eGFR-LC3 lentivirus (Hanbio, China) for 24 h. Then, transfected cells were treated with DOX or rapamycin (Sigma) for 24 h. Cells were then fixed with 4% PFA for 30 min and washed with PBS 3 times. The immunofluorescent images were obtained by using a confocal fluorescence microscope (BX50; Olympus, Japan) and representative images are presented. Transmission electron microscopy detection of autophagic vacuoles was performed as previously described [46].

In vivo tumor xenograft
U2OS/Dox (Dox-resistant) cells (1 × 10 7 ) transduced with lentivirus (Lsh-NC or Lsh-Sox2OT-V7) were subcutaneously injected into the nude mice (SLAC Experimental Animal Center, Changsha, China). Two weeks after tumor formation, mice were received intraperitoneal injection. There were four experimental groups: control, Lsh-Sox2OT-V7, Lsh-Sox2OT-V7+miR-22 inhibitor and Lsh-Sox2OT-V7+miR-142 inhibitor. The mice in all groups were intraperitoneally injected with Dox (2 mg/kg, once a week for 4 weeks). Tumor volume was measured every 3 days after the formation of the xenografted tumor. Forty-two days after the inoculation of U2OS/Dox cells, the mice were sacrificed by euthanasia. All experimental procedures in the current study were approved by the Central South University Animal Care Committee.

RNA immunoprecipitation (RIP)
RIP was performed using Magna RIP RNA-Binding Protein Immunoprecipitation Kit (17-700, Millipore) following the protocols. RNA for in vitro experiments was transcribed using T7 High YieldRNA Synthesis Kit (E2040S, NEB) according to the manufacturer's instructions. Sox2OT-V7, miR-142, and miR-22 levels in the immunoprecipitates were measured by qRT-PCR.

Statistical analysis
Data was expressed as the means ± SD of at least three independent experiments and statistically analyzed by one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test or independent sample t-test using the SPSS Statistics 17.0 software. The level of significance was assessed by thresholds of P < 0.05 and P < 0.01.

AUTHOR CONTRIBUTIONS
Yang Yuan operated the experiments and wrote the manuscript. Jie Wen conducted the cellular and molecular experiments. Ding Chen, Weihong Zhu, Zhengxiao Ouyang and Wanchun Wang have contributed to clinical sample collection and data analysis. Kewei Zhu conducted the experiments and revised the manuscript. All authors have read and approved the final manuscript.

CONFLICTS OF INTEREST
Authors declare no conflicts of interest.