Patients of EOC were enrolled from the liaoning cancer hospital between March 2010 and November 2015. The including criteria were as follow: patients were with definite pathological diagnosis; And the exclusion criteria: patients were received with chemotherapy or radiotherapy before surgery. Serum sample were collected from EOC patients and healthy control. The tumor and paired non-tumor tissues were also collected after lesion excision with 30 mins and stored in liquid nitrogen, then transferred to -80 ${}^{\circ }$C refrigerator. And the characteristics of cases were thoroughly noted.

The EOC cell lines of ES-2, HO-8910, OVCAR-3, A2780 and human normal ovarian epithelial (HNOE) were from American Type Culture Collection (ATCC). The test of mycoplasma contamination was performed to make sure the cell lines were mycoplasma-free. All the cells were cultured at 37 ${}^{\circ }$C in a humidified incubator with 5% CO₂.

The expression levels of RNA were calculated by the qRT-PCR system. Total RNA was extracted by TRIzol reagent (Invitrogen), and 1 $\mu$g of total RNA was reverse transcribed using the PrimeScriptP RT Reagent Kit (Perfect Real-Time; Takara). SNHG22 cDNA, shRNA interference vector lentiviruses were purchased from GeneChem Corporation (Shanghai, China). SP1 siRNAs were purchased from Oligobio Company (Beijing, China). The primers were designed in Supplementary Tables 1,2. EOC cells were transfected with plasmids in the presence of Lipofectamine 3000 (Invitrogen). After 48 h of transfection, cells were gathered for further use in following experiments. The gene expression quantity was calculated using the 2${}^{-\mathrm{\Delta }\mathrm{\Delta }Ct}$ method.

The levels of glucose and lactate were calculated with a Glucose Colorimetric Assay Kit (BioVision, CA) and a Lactate Assay Kit (BioVision, CA) in line with the instructions of manufacturer.

ChIP was performed with the protocol of SimpleChIP®Enzymatic Chromatin IP Kit (Magnetic Beads) (#9003). Cells are fixed with formaldehyde and lysed, and chromatin is fragmented by partial digestion with Micrococcal Nuclease to obtain chromatin fragments of 1 to 5 nucleosomes. Chromatin immunoprecipitations are performed using ChIP-validated antibodies and ChIP-Grade Protein G Magnetic Beads. Anti-SP1 were performed in ChIP assay, and goat anti-rabbit IgG used as a negative control. After reversal of protein-DNA cross-links, the DNA was purified using DNA purification spin columns and was quantified by real-time qPCR analysis.

The SNHG22 promoter after PCR amplification was inserted into the downstream of the firefly luciferase gene in pGL3-Basic vector (Promega, USA) to construct pGL3-SNHG22 promoter reporter vector. Cells were co-transfected with a reporter construct (pGL3-basic plasmid or pGL3-SNHG22 plasmid) and shRNA or negative control. Dual Luciferase Reporter Assay System (Promega) was used to measure the luciferase activity 48 h after transfection. The relative luciferase activity was contrasted the Renilla luciferase and firefly luciferase activity.

All the data were showed as the mean $\pm$ standard deviation, at least three independent experiments. Data were compared using the chi-square ($\chi$² test), Student’s t-test or one-way analysis of variance (ANOVA). ROC curve was applied to evaluate the predictive ability of EOC diagnosis. Univariate and multivariate analysis were performed to ascertain independent factors for the diagnosis of EOC. Kaplan-Meier method was applied to measure the overall survival (OS). SPSS 16.0 (SPSS Inc., Chicago, IL, USA) was used to conduct statistical analyses, and differences were ensured when P-value was 0.05.

To investigate the expression status of SNHG22, we examined its expression using qRT-PCR in a cohort of 60 paired and non-tumor tissues of EOC. The level of SNHG22 was predominantly elevated in EOC tissues compared with the paired non-tumor tissues (Fig. 1A,B, P 0.01). Then, we divided the samples into SNHG22 low-expression group and SNHG22 high-expression group on the basis of the median SNHG22 expression. Besides, the clinicopathologic characteristics were demonstrated in Supplementary Table 3. Next, the EOC cell lines of ES-2, HO-8910, OVCAR-3, A2780 and HNOE were tested for SNHG22 expression. As expected, the SNHG22 level was significantly increased in ES-2, HO-8910, OVCAR-3 and A2780 cells, compared with HNOE cells (Fig. 1C). These results indicate that SNHG22 expression is upregulated in the tissues and cell lines of EOC.

Fig. 1.

SNHG22 is upregulated in EOC tissues, and cell lines.
(A, B) qRT-PCR analysis was used to evaluate SNHG22 expression in 60 paired tumor and paired adjacent non-tumor tissues.
The expression profiles of SNHG22 in ES-2, HO-8910, OVCAR-3, A2780 and human normal ovarian epithelial (HNOE) was detected with qRT-PCR.
*P 0.05, **P 0.01, ***P 0.001.

Our study also found that SNHG22 level in serum was remarkably upregulated in serum of EOC cases than healthy controls (Fig. 2A). Moreover, the AUC of ROC curve was 0.824 (95% CI 0.747–0.900), with the sensitivity 83.8% and specificity 70.9%, respectively (Fig. 2B). As a whole, our data show that SNHG22 is ectopic expressed in the circulation of EOC patients.

Fig. 2.

SNHG22 is upregulated in EOC serum.
qRT-PCR analysis was used to evaluate serum SNHG22 expression levels in EOC patients and healthy controls.
ROC curve analysis of the diagnostic performance of serum SNHG22.
Kaplan-Meier survival curves revealed a relationship between SNHG22 expression and overall survival of patients with EOC.
Comparison of the expression level of SNHG22 in serum among healthy controls, early stage RCC patients, and late stage RCC patients.
Expression levels of SNHG22 in serum decreased after removal of primary tumors in patients with EOC were quantified by quantitative RT-PCR (n = 24).
*P 0.05, **P 0.01, ***P 0.001.
Signaling pathways of downregulated genes after SNHG22 knockdown in A2780 cells detected by microarray analysis.

Kaplan-Meier survival curves illustrated that high SNHG22 expression patients had shorter overall survival time than low SNHG22 expression (Fig. 2C, P 0.01). Next, we calculated the predictive ability of SNHG22 in serum. The SNHG22 expression was delivered both in stage I–II (early stage) and stage III–IV (advanced stage) of EOC (Fig. 2D). In addition, we analyzed the association between SNHG22 expression and the clinicopathological characteristics of EOC cases (Supplementary Table 4). Intriguingly, SNHG22 expression was linked with vessel invasion (P 0.05) and TNM stage (P 0.01), but not with age (P = 0.68), gender (P = 0.56), tumor size (P = 0.06), tumor invasion depth (P = 0.23), lymph node metastasis (P = 0.08), or lymph node metastasis (P = 0.81). Furthermore, we evaluated the prognostic ability of SNHG22 expression level in EOC. As shown in Supplementary Table 5, univariate analyses showed highly expressed SNHG22 was associated with a dramatic risk of death compared to low SNHG22 expression (P 0.01). Besides, multivariate analysis demonstrated that SNHG22 level could be an independent prognostic factor. So our findings denoted that SNHG22 expression in serum is related to the prognosis of EOC. Additionally, we tested the SNHG22 expression in EOC patients (n = 20) serum before and after the lesion of the primary cancer. Expertly, the SNHG22 expression was considerably weakened after the surgery than before (Fig. 2E; P 0.01).

To investigate the biological function of SNHG22, we compared A2780 cells transfected with either sh-SNHG22 or a negative control. The pathway analysis showed that SNHG22 depletion had obvious effects on genes, which were significantly related to metabolic pathways (Fig. 2F). Accumulating evidence showed that cancer cells conduct glycolysis at a rapid speed than normal tissues [14]. And malignant tumor might be originated from abnominal energy metabolism. Consistent with previous reports, our data elucidated that glucose starvation could induce SNHG22 elevation and modulate the glycolytic metabolism in EOC cell. Glucose starvation obviously increased SNHG22 level in ES-2 cells (Fig. 3A). To explore the function of SNHG22 in glycolysis, we inhibited SNHG22 expression in A2780 cells and upregulated SNHG22 expression in ES-2 cells (Fig. 3B). Consequently, we calculated the extracellular acidification rate (ECAR) [15] to investigate whether SNHG22 could effectively influence glycolytic metabolism in EOC. Subsequently, sh-SNHG22 dramatically impaired ECAR levels in A2780 cells (Fig. 3C), compared to control group. We also explored the level of extracellular lactic acid, which is common used as a glycolysis metabolite. As a result, the production of Lactic acid was greatly alleviated in A2780 cells with SNHG22 abalation (Fig. 3D). Conversely, pCDNA-SNHG22 predominantly elevated ECAR (Fig. 3E), and lactic acid production (Fig. 3F) in ES-2 cells.

Fig. 3.

SNHG22 regulates glycolysis in cancer cells.
(A) qRT-PCR analysis was used to evaluate SNHG22 expression levels in ES-2 cells of glucose starvation.
(B) Relative expression levels of SNHG22 in transfected cells.
(C) The change of ECAR level was detected in A2780 cells transfected with sh-control or sh-SNHG22.
(D) The relative lactic acid level was measured in A2780 cells transfected with sh-control or sh-SNHG22.
(E) The change of ECAR level was detected in ES-2 cells transfected with vector or pCDNA-SNHG22.
(F) The relative lactic acid level was measured in ES-2 cells transfected with vector or pCDNA-SNHG22.
*P 0.05, **P 0.01, ***P 0.001.

We next explored the upstream regulation mechanism of SNHG22. We assumed that SP1 contributed to the upregulation of SNHG22 by binding to the promoter, according to the JASPAR and UCSC website analysis. To verify this speculation, we first suppressed the SP1 expression using siRNA (Fig. 4A) and found SNHG22 with diminished level (Fig. 4B). Second, luciferase reporter assay denoted that the SP1 depletion could weaken the SNHG22 promoter activity, suggesting that SP1 modified SNHG22 expression at the transcriptional level (Fig. 4C). Third, we explored the enrichment of the DNA motif of SP1 and the promoter region of SNHG22. In detail, we divided the promoter region into four sections on the basis of possible binding sites (Fig. 4D). And ChIP analysis showed that the enrichment of SP1 binding was obvious in the P1 promoter region (-500~1 nt) (Fig. 4E). In addition, we assembled pGL3-SNHG22 promoter full region and P1 deleted region and carried out ChIP analysis again (Fig. 4F). Intriguingly, we found that P1 section (-500~1 nt) was necessary for SP1 binding, for there is nearly no enrichment in P1 deleted group (Fig. 4G). Finally, the correlation between SP1 expression in serum and clinicopathologic characteristics of EOC patients in Supplementary Table 6.

Fig. 4.

SP1 regulates SNHG22 expression by serving as a transcription activator.
(A) The SP1 interference efficiency using 2 different siRNAs and negative control detected by qRT-PCR.
(B) RT-qPCR analysis of SNHG22 expression after SP1 knockdown in A2780.
(C) Luciferase reporter assay was performed using vector-containing sequences of the SNHG22 promoter region.
(D) SNHG22 promoter was divided into four sectional fragments.
(E) ChIP analysis was performed to detect the affinity of SP1 with SNHG22 promoter.
(F) The full LINC00958 promoter region and P1 deleted region was constructed.
(G) ChIP analysis was performed to determine the relative enrichment of SNHG22 promoter.
*P 0.05, **P 0.01, ***P 0.001.