Screening SNORNAs related to prognosis in ovarian cancer patients
To find which SNORNAs are related to poor prognosis of ovarian cancer, we used TCGA database to analyze the relationship between of SNORNAs expression and overall survival for 379 ovarian cancer patients. Finally, 4 SNORNAs (SNORA2B, SNORD19, SNORD116-4 and SNORD89) associated with poor prognosis of ovarian cancer were screened out by Kaplan–Meier analysis. Low expression of SNORA2B and SNORD19 is related to poor prognosis of ovarian cancer patients (Fig. 1a, c), and high expression of SNORD116-4 and SNORD89 is associated with poor prognosis of ovarian cancer patients (Fig. 1e, g). Next, we analyzed these four SNORNAs resort to SNORic database. Based on the retrieval of SNORic database, we found that SNORD89 was correlated with 178 mRNAs and participated in splicing of 43 mRNAs in ovarian cancer. SNORA2B, on the other hand, was correlated with only two mRNAs and not participated in mRNA splicing in ovarian cancer. SNORD19 is involved in splicing of 2 mRNAs and not correlated with any mRNA. Further, there is no information of SNORD116-4 in ovarian cancer in SNORic database.
TO avoid the influence of a small number of patients who survived for more than 5 years on survival curve differences, we set a 5-year cut-off data for survival to observe the effect of screened snoRNAs on patients’ prognosis. We found that only the SNORD89 dysregulation was significantly correlated with poor prognosis of patients (Fig. 1b, d, f, h). Thus, these results suggest that SNORA2B and SNORD19 may play roles as suppressor genes, while SNORD116-4 and SNORD89 as oncogenes in ovarian cancer.
In order to comprehend the correlation between the oncogenes and poor prognosis of ovarian cancer patients, we further selected SNORD89 and SNORD116-4 for our study. The ovarian cancer patients in TCGA database are mainly in stage III and IV. Therefore, we analyzed the relationship between the expression of SNORD89, SNORD116-4 and prognosis of patients in stage III and IV. The survival curves showed that the expression of SNORD89 was an important prognostic factor in stage IV, and patients with high expression of SNORD89 have more poorly prognosis (Additional file 1: Figure S1a, b). Among the ovarian cancer patients in TCGA, 152 and 24 patients had high expression of SNORD116-4 in stage III and stage IV, 139 and 33 patients had low expression of SNORD116-4 in stage III and stage IV, respectively. The patients in stage III with high expression of SNORD116-4 have poor prognosis, however, patients in stage IV with high expression of SNORD116-4 have better prognosis, SNORD116-4 shows a trend for better survival in stage IV (Additional file 1: Figure S1c). These results suggest that SNORD116-4 has different effects on the prognosis of ovarian cancer in stage III and stage IV.
Moreover, we further confirmed that the prognosis of patients in stage IV was worse than that in stage III (Additional file 1: Figure S1d). Interestingly, this was reversed by the SNORD89 high expression. From the survival curve, we can see that the patients with SNORD89 high expression in stage III have worse prognosis than the patients with SNORD89 low expression in stage IV (Additional file 1: Figure S1e). In addition, we analyzed the effects of SNORD89, SNORD116-4 and stage on the 5-year survival of patients with ovarian cancer. Unlike the OS curves, SNORD116-4 had no significant effect on the 5-year survival of patients with stage III and IV, and, stage has a strong correlation with the 5-year survival of patients (Additional file 2: Figure S2). However, the prognosis of patients with high SNORD89 expression was worse than that of patients with low SNORD89 expression in OS and 5-year survival curves. The findings suggest that SNORD89 might have an important role in the progress of ovarian cancer.
Prognostic factors of ovarian cancer patients
The high expression of SNORD89 and SNORD116-4 predicts poor prognosis in ovarian cancer patients. We next analyzed the correlation between the two SNORNAs expression and the clinicopathologic features of ovarian cancer patients in TCGA database by Chi square test. We found that the expression of SNORD116-4 was not correlated with any clinicopathological parameters. However, the expression of SNORD89 was correlated with age (P = 0.03, Additional file 3: Table S1). We also found that the expression of SNORD89 was higher in elderly patients (P = 0.0202) and progressive disease (P = 0.0486) by unpaired t test (Additional file 4: Figure S3a, b).
In order to observe SNORD89 expression and the clinical factors affect the prognosis of ovarian cancer, the univariate and multivariate COX’s regression analysis were performed (Additional file 3: Tables S2 and S3). Univariate analysis showed that the expression of SNORD89, age, and tumor size was significantly associated with overall survival (OS), and age was also related to progression-free survival (PFS) of ovarian cancer patients. In addition, multivariate COX’s regression analysis revealed that race and venous invasion were independent predictors of OS, and venous invasion was also independent predictors of PFS in patients with ovarian cancer. Thus, SNORD89 expression may be associated with other factors affecting the prognosis of ovarian cancer patients.
SNORD89 is highly expressed in ovarian cancer stem cells
Tumor stem cells cause tumor recurrence by promoting tumor invasion, metastasis, drug resistance, and so on, leading to poor prognosis of patients [23]. It has been reported that SNORNAs have important roles in the cancer progress [12]. We compared the non-coding RNAs (NC RNAs) expression in ovarian epithelial cells (HOSEpiC), ovarian cancer cells (Ovcar-3, OV) and ovarian cancer stem cells (Ovcar-3-S, OS) by gene chips. Finally, 15 SNORNAs with high expression in HOSEpiC and OS were screened out (Fig. 2a, b, Additional file 3: Table S4). Among them, only SNORD89 and SNORD116-4 were associated with the prognosis of ovarian cancer patients, while SNORD116-4 played an opposite role in stage III and stage IV. Therefore, SNORD89 can better predict the prognosis of ovarian cancer patients. As we know, the poor prognosis of tumors is closely related to cancer stem cells. To our surprise, SNORD89 is not only associated with poor prognosis of ovarian cancer, but also highly expressed in OS vs OV and OS vs HOSEpiC, while SNORD116-4 was not highly expressed in OS. Therefore, we decided to conduct a further research on SNORD89.
Next, we measured the expression difference of SNORD89 in HOSEpiC, OV and OS cells by quantitative real-time PCR (qRT-PCR) and reverse transcription PCR (RT-PCR). qRT-PCR results showed that SNORD89 expression was up-regulated 2.38 ± 0.29-fold in OS vs HOSEpiC cells, and 5.93 ± 0.53-fold in OS vs OV cells (Fig. 2c). And the increased expression of SNORD89 was also observed by RT-PCR (Fig. 2d). The data suggest that SNORD89 upregulated in OS versus OV and HOSEpiC.
Effects of SNORD89 interference on the stemness of ovarian cancer cells
We compared the expression of genes associated with stemness in OV and OS by qRT-PCR. The expression levels of CD133, CD44, and Nanog were increased by 3.32 ± 0.48, 7.11 ± 0.90 and 6.78 ± 1.10-fold in OS cells relative to OV cells, respectively (Fig. 3a). Also, RT-PCR and agarose gel electrophoresis showed the higher exprssion of CD44 and Nanog in OS cells (Fig. 3b). Additionally, we observed that the CD133 positive cells were significantly increased in OS than those in OV by flow cytometry analysis (Fig. 3c).
To investigate whether the SNORD89 interference affects the stemness of ovarian cancer cells, we first examined the SNORD89 expression in OV cells transfected with over expression plasmid (OE) of SNORD89 and in OS cells transfected with silent plasmid (shRNA) of SNORD89 for 24, 48, and 72 h, respectively. The qRT-PCR analysis showed that the SNORD89 overexpression in OV and CA cells was the most efficient at 24 h’ transfection (3.81 ± 0.28-fold than NC in OV cells and 20.65 ± 1.195-fold than NC in CA cells, Fig. 3d and Additional file 5: Figure S4a). And the most silence efficiency of three shRNA (shRNA-1, shRNA-2, shRNA-3) was 0.255 ± 0.08, 0.56 ± 0.05 and 0.70 ± 0.04 at 48 h in OS cells (Fig. 3e).
Next, we examined the changes of stemness genes in OE-transfected OV cells at 24 h, and in shRNA-transfected OS cells at 48 h. We found that the expression of CD133, CD44, and Nanog was increased by 4.28 ± 0.48, 4.58 ± 0.28 and 1.90 ± 0.49-fold in OE group than NC group in OV cells by qRT-PCR (Fig. 3f), and increased by 1.80 ± 0.25, 1.40 ± 0.11 and 1.47 ± 0.09-fold in OE group than NC group in CA cells by qRT-PCR (Additional file 5: Figure S4b). RT-PCR and agarose gel electrophoresis also verified the elevated expression of CD44 and Nanog (Fig. 3g). The flow cytometry analysis showed the higher CD133 positive cells in OE group (Fig. 3h). Consistantly, the silence of SNORD89 with shRNA-1 and shRNA-2 at 48 h notably decreased the expression of CD44 and Nanog in OS cells by qRT-PCR (Fig. 3i) and RT-PCR and agarose gel electrophoresis (Fig. 3j). The flow cytometry analysis showed the decreased CD133 in OS cells transfected with SNORD89 silence plasmids (Fig. 3k). These data suggest that SNORD89 can increase the expression of stemness genes in ovarian cancer cells.
Effects of SNORD89 interference on cell proliferation and self-renewal ability of ovarian cancer cells
We analyzed the expression correlation of SNORD89 with other genes using the data from ovarian cancer patients in TCGA, and found that the expression level of SNORD89 was related to some genes of cell cycle checkpoint and cell cycle arrest (Fig. 4a). So, we suspected that the interference of SNORD89 expression may affect cell cycle. And we examined the changes of cell cycle phases in OV and OS cells after interfering the expression of SNORD89. The flow cytometry analysis showed the proportion of S phase was obviously increased in OV cells transfected with SNORD89 OE plasmids (Fig. 4b). Conversely, the proportion of S phase was obviously decreased in OS cells transfected with shRNA-1 and shRNA-2, while the proportion of G2 phase was raised (Fig. 4c). These results indicated that SNORD89 may have an effect on cell proliferation in ovarian cancer.
Then, we checked the cell proliferation ability changes in OV and CA cells of SNORD89 overexpression and OS cells of SNORD89 silence by Cell Counting Kit-8 (CCK-8) assays at 24 h, 48 h, 72 h and 96 h transfection. CCK-8 assays revealed a significant raise cell proliferation in OE cells compared with the NC cells (Fig. 4d and Additional file 5: Figure S4c), and a notable decreased cell proliferation ability in SNORD89 knockdown-OS cells (Fig. 4e). In addition, both plate clone formation assay and soft agar colony formation assay showed that SNORD89 overexpression significantly increased the number of clone formation in OV and CA cells, and the size was larger in OE group (Fig. 4f, g, Additional file 5: Figure S4d, e). Furthermore, colony formation assays indicated that silencing SNORD89 obviously reduced the number of colony formation in OS cells, and the size was smaller in shRNA-1 and shRNA-2 groups (Fig. 4h). These results suggest that SNORD89 may increase cell proliferation and self-renewal ability of ovarian cancer cells.
Effects of SNORD89 interference on cell migration and invasion of ovarian cancer cells
We further detected the effects of SNORD89 interference on invasion and migration ability of ovarian cancer cells by scratch migration assay and cell invasion analysis. The scratch migration assay showed that the overexpression of SNORD89 significantly increased the area wound healed than NC-transfected OV and CA cells at 24 h, 48 h and 72 h (Fig. 5a and Additional file 5: Figure S4f), suggesting SNORD89 can increase the migration ability of ovarian cancer cells. In addition, the cell invasion analysis indicated that SNORD89 overexpression obviously elevated the invasion ability of OV and CA cells (Fig. 5b and Additional file 5: Figure S4g), while SNORD89 knockdown notably reduced the invasion ability of OS cells (Fig. 5c).
The carcinogenicity of SNORD89 may be related to the NOTCH1-MYC highway
Many reports have shown that c-Myc binds with NOTCH1 to promote the development of cancer by acting as a target of NOTCH1 to form a NOTCH1-MYC pathway [24, 25]. We first compared the expression of c-Myc and Notch1 in OV and OS by qRT-PCR, and found that the expression levels of c-Myc and Notch1 were 1.39 ± 0.09 and 2.97 ± 0.37-fold in OS cells than OV cells (Fig. 6a). RT-PCR and agarose gel electrophoresis also verified the increased expression of c-Myc and Notch1 in OS cells (Fig. 6b).
Next, we assessed the effects of SNORD89 interference on the expression of c-Myc and Notch1 in ovarian cancer cells. The Notch1 and c-Myc expression levels were elevated 2.56 ± 0.33 and 1.47 ± 0.05-fold in the OV cells transfected with SNORD89 OE plasmids by qRT-PCR (Fig. 6c), and RT-PCR and agarose gel electrophoresis also showed the consistent results (Fig. 6d). The Notch1 and c-Myc expression levels were elevated 1.75 ± 0.06 and 1.58 ± 0.10-fold in the CA cells transfected with SNORD89 OE plasmids by qRT-PCR (Additional file 5: Figure S4h). Additionally, silencing SNORD89 significantly declined the expression of c-Myc and Notch1 in OS cells (Fig. 6e). The similar results were also verified by RT-PCR and agarose gel electrophoresis (Fig. 6f). We further observed that SNORD89 overexpression obviously increased the protein expression of c-Myc and Notch1 in OV and CA cells (Fig. 6g and Additional file 5: Figure S4i), while SNORD89 knockdown notably decreased the expression of these proteins in OS cells (Fig. 6h).
In our experiments, we found that SNORD89 affected the proliferation, self-renewal ability, invasion and migration of ovarian cancer cells related to stemness phenotype. Notch1 is a well-known pathway associated with stemness phenotype in cancer, and c-Myc can bind with NOTCH1 to promote the development of cancer by acting as a target of NOTCH1 to form a NOTCH1/c-Myc pathway. Thus, the data suggest that SNORD89 might regulate the Notch pathway in ovarian cancer.