LncRNA SNHG10 Facilitates Hepatocarcinogenesis and Metastasis by Modulating Its Homolog SCARNA13 via a Positive Feedback Loop

Hepatocellular carcinoma (HCC) is the fourth most common incident malignancy and third leading cause of cancer-related death in China (1). Despite continuous progress in clinical detection and treatment strategies, the prognosis of patients with HCC remains poor largely due to late diagnosis, a high postoperative recurrence rate, and metastasis (2). Because the molecular mechanisms underlying the tumorigenesis and metastasis of HCC have not yet been fully elucidated, identifying pivotal cancer-promoting molecules would contribute to the understanding of HCC pathogenesis and identification of potential therapeutic targets.

Long noncoding RNAs (lncRNA) belong to a class of RNA transcripts over 200 nucleotides with no or low protein-coding potential (3). Accumulating studies have shown that lncRNAs participate in multiple biological regulatory processes, such as cell differentiation, apoptosis, immune response, and carcinogenesis (4). Small nucleolar RNAs (snoRNA), predominately found in cell nucleolus, are a subgroup of ncRNAs with 60–300 nucleotides in length and guide the posttranscriptional modification of small RNAs (5). Emerging evidence has demonstrated that abnormally expressed snoRNAs exert significant and comprehensive influences on the carcinogenesis and progression in diverse human malignancies (6).

Recently, a complicated interaction between lncRNAs and small RNAs has been disclosed in which some lncRNAs can produce or regulate small RNAs (7). For instance, lncRNA MIR17HG–derived miR-17∼92 attenuates the TGFβ signaling pathway to promote angiogenesis and tumor cell growth (8). As a subclass of small RNAs, most snoRNAs are encoded in the introns of small nucleolar RNA host genes (SNHG; ref. 6). Specifically, the introns of primary RNA transcripts from SNHGs are processed into snoRNAs in the nucleus, while the exons are spliced into lncRNAs and transported to the cytoplasm, suggesting that there might be underlying correlations between snoRNAs and the homologous lncRNAs transcribed from SNHGs. However, the particular relationships between these lncRNAs and homologous snoRNAs as well as their specific roles in oncogenesis are poorly understood.

In this study, by performing high-throughput RNA-sequencing (RNA-seq), we identified the contributions of the lncRNA small nucleolar RNA host gene 10 (SNHG10) and its homologous snoRNA, SCARNA13, in the development and progression of HCC. Both SNHG10 and SCARNA13 were dramatically associated with the malignant biological behaviors of HCC cells and poor prognosis of patients with HCC. Specifically, SNHG10 modulated the expression of SCARNA13 through the miR-150-5p/RPL4-c-Myb–positive feedback loop and SCARNA13 exerted its oncogenic activity by regulating SOX9 in HCC. Our findings characterized an epigenetic cause of hepatocarcinogenesis and metastasis with diagnostic and therapeutic implications.

The protocols used in this study conformed to the ethical guidelines of the 1975 Declaration of Helsinki and were approved by the Ethical Review Committees of Sichuan University (Chengdu, China).

HCC tissues and adjacent normal tissues were obtained following curative surgical resections from 64 patients with HCC at the West China Hospital (Sichuan University, Chengdu, China). Ethical approval was granted from the Ethical Review Committees of Sichuan University (Chengdu, China), and written informed consent was obtained from all the patients.

SNU-182, Huh-7, Hep3B, SK-Hep1, and SNU-387 cell lines were purchased from the Shanghai Cell Bank Type Culture Collection Committee (CBTCCC, Shanghai, China). HEK293T cell line was purchased from the ATCC. The HCCLM3 cell line was acquired from the State Key Laboratory of Biotherapy, West China Hospital (Chengdu, Sichuan, China). All cell lines were characterized by short-tandem repeat analysis, Mycoplasma testing, isozyme detection, and cell viability determination by third-party biology services (Feiouer Biology Co., Ltd). Cells were cultured at 37°C in a humidified incubator with 5% CO₂ in DMEM (HyClone) supplemented with 10% FBS (PAN). Actinomycin D was obtained from Sigma-Aldrich.

The animal studies were authorized by the Animal Ethic Review Committees of the West China Hospital (Chengdu, Sichuan, China). Twenty male athymic BALB/c nude mice, ages 4–5 weeks, were purchased from Beijing HFK Bioscience and were fed under standard pathogen-free conditions. All surgical procedures were performed with sodium pentobarbital anesthesia. Mice were subcutaneously injected with 200 μL of cell suspension containing 2 × 10⁶ cells in the right flanks. Tumors were allowed to grow for 2 weeks, after which, pieces of the subcutaneous tumors were excised and cut into 1-mm³ sections, followed by upper-abdominal incisions on mice. The right lobe of the liver was exposed, and part of the liver surface was moderately injured with scissors. A tumor section was implanted in the small incision on the liver surface. The liver was returned to the peritoneal cavity; subsequently, the abdominal wall was strictly sutured. The tumors were permitted to grow for 5 weeks, followed by euthanization of the mice. Next, the lung was carefully anatomized to expose the pulmonary metastatic focuses, the obvious ones of which were completely separated from the lung tissues. The largest one was subjected to primary culture to isolate xenografted HCC cells, while the others were made into tissue sections and subjected to hematoxylin and eosin (H&E) staining to determine the pulmonary metastatic focuses by two pathologists from the Department of Pathology, West China Hospital (Chengdu, Sichuan, China). All animal experiments were strictly implemented in compliance with the NIH Guide for the Care and Use of Laboratory Animals.

Locked nucleic acid-FISH (LNA-FISH) was performed to determine the subcellular location of SNHG10 and SCARNA13. LNA fluorescein–labeled probes against 18S rRNA, U6 snRNA, SNHG10, and SCARNA13 were designed and synthesized by RiboBio (RiboBio Biotechnology). FISH was conducted using the Fluorescent in Situ Hybridization Kit (RiboBio Biotechnology), according to the manufacturer's protocol. Fluorescence signals were scanned using the A1R+MP Confocal Laser Microscope System (Nikon).

RNA immunoprecipitation (RIP) assays were implemented using the Magna RIP RNA-binding Protein Immunoprecipitation Kit (Millipore), according to the manufacturer's instructions. Anti-Ago2 antibody and normal IgG (Millipore) were used for immunoprecipitation. The coprecipitated RNAs were purified with phenol:chloroform:isoamyl alcohol and subsequenty analyzed by qPCR to assess the enrichment of SNHG10 and miR-150-5p to Ago2.

Cells were cross-linked with 1% formaldehyde and quenched in glycine solution. Chromatin immunoprecipitation (ChIP) assays were performed using the Pierce Magnetic ChIP Kit (Thermo Fisher Scientific), according to the manufacturer's protocol. Anti–c-Myb antibody and normal IgG (Millipore) were applied for immunoprecipitation. ChIP-enriched DNA samples were quantified by qPCR to determine the c-Myb binding sites (MBS) of the SNHG10 promoter region. The data were shown as relative enrichment normalized to control IgG. The sequences of primers used for ChIP-qPCR are presented in Supplementary Table S1.

Ten oligonucleotide probes corresponding to the SNHG10 transcript were synthesized with biotin tags located at the 3′ end by RiboBio (RiboBio Biotechnology). To eliminate nonspecific signals, all probes were divided into two pools (even and odd probe sets). The probe set targeting LacZ was used as a negative control. Chromatin isolation by RNA purification (ChIRP) assays were conducted using the EZ-Magna ChIRP RNA Interactome Kit (Millipore), according to the manufacturer's protocol. The precipitated RNA was identified and quantified by qPCR to analyze the enrichment of miR-150-5p to SNHG10. The sequences of probes used for ChIRP are listed in Supplementary Table S2.

Cells were lysed in IP lysis/wash buffer. The immune complex was prepared using anti–c-Myb antibody, anti-RPL4 antibody, or normal IgG (Millipore) and was subsequently captured using the Pierce Classic IP Kit (Thermo Fisher Scientific), according to the manufacturer's instructions. The samples were separated on 10% SDS-PAGE gels and analyzed by Western blot.

To measure promoter activity, the cells were cotransfected with pEZX-PL01-SNHG10, and c-Myb siRNAs or control siRNA using Lipofectamine 2000 (Invitrogen). For 3′ UTR luciferase reporter assays, pmirGLO-SNHG10 or pmirGLO-SNHG10-mut(miR-150-5p) was cotransfected with miR-150-5p mimics or miR-NC into HEK293T cells. Likewise, pmirGLO-SNHG10 or pmirGLO-SNHG10-mut(RPL4) was cotransfected with RPL4 siRNAs or control siRNA into HEK293T cells. After 48 hours of transfection, Renilla and firefly luciferase activities were detected by the Synergy Mx Multi-Mode Microplate Reader (BioTek). Firefly luciferase activity was normalized to Renilla luciferase activity and was presented as the relative luciferase activity.

The Gene Expression Omnibus accession numbers for the RNA-seq data of xenografted HCC cells, ChIRP-seq data of SNHG10, and RNA-seq data for SCARNA13 knockdown are GSE120021, GSE119773, and GSE120095, respectively. The accession number for the quantitative proteomics data reported in this article is Integrated Proteome Resources: IPX0001319000.

Each experiment was performed in triplicate, and the data were presented as means ± SEM. All results were representative of three separate experiments.

All statistical analyses were performed using GraphPad Prism 7 Software (GraphPad Software) and SPSS version 17.0 Software (SPSS, Inc.). Regarding comparisons, Student t test, χ² test, the Wilcoxon signed-rank test, and the Mann–Whitney test were conducted as appropriate. Correlations were calculated by Pearson correlation analysis. The median SNHG10 expression was used as a cut-off value for grouping. The low SNHG10 group in each of the 64 patients was defined as a value below the 50th percentile. The high SNHG10 group in each of the 64 patients was defined as a value above the 50th percentile. Likewise, patients were divided into the high and low SCARNA13 groups according to the median SCARNA13 expression in HCC tissues. The survival curves were measured by the Kaplan–Meier method, and the differences were evaluated by the log-rank test. The univariate and multivariate Cox proportional hazards regression models were utilized to assess the independent factors. Statistical significance was indicated by P values less than 0.05 (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Other detailed materials and methods are provided in the Supplementary Materials and Methods section. The sequences of siRNA against specific target are presented in Supplementary Table S3. The information of primary antibodies used is shown in Supplementary Table S4.

To identify ncRNAs involved in HCC metastasis, we initially grafted HCC cells into nude mice and performed a screen of lung metastasis based on orthotopic implanted models in vivo (Fig. 1A). Subsequently, RNA-seq was utilized to compare the expression profiles between xenografted HCC cells and parental HCC cells (Supplementary Fig. S1A) and demonstrated that 101 lncRNAs were differentially expressed by at least 2-fold change (Fig. 1B). Moreover, we found that six snoRNAs were upregulated and two snoRNAs were downregulated in xenografts compared with those in parental HCC cells (Fig. 1C). Moreover, the The Cancer Genome Atlas (TCGA) Liver Hepatocellular Carcinoma (LIHC) data repository was analyzed to further investigate the roles of these candidate lncRNAs and snoRNAs in hepatocarcinogenesis. In total, 3,592 lncRNAs and 113 snoRNAs were significantly dysregulated in HCC tissues compared with adjacent nontumor tissues from the TCGA LIHC dataset (Fig. 1B and C). With the intersection, 24 lncRNAs (including SNHG10, DLGAP1-AS2, TMEM161B-AS1, LINC01004, and NNT-AS1) and 6 snoRNAs (SCARNA13, SNORD6, SNORD100, SNORD3A, SNORD94, and SNORA71C) were identified to possibly play crucial roles in both the tumorigenesis and metastasis of HCC. Intriguingly, among these abnormally expressed genes were SNHG10 and SCARNA13, which are the different products from the same primary RNA transcript. More specifically, SCARNA13 is processed from the introns of the primary RNA transcript from the SNHG10 gene, whereas the exons are spliced into the SNHG10 transcript (Fig. 1D), suggesting that SNHG10 and SCARNA13 might synergistically contribute to the development and progression of HCC. Therefore, we focused on this lncRNA SNHG10 and its homologous SCARNA13.

First, the protein-coding potential of SNHG10 and SCARNA13 was analyzed to confirm whether they belonged to ncRNAs. As expected, the open reading frame (ORF) finder from the National Center for Biotechnology Information (NCBI), Coding Potential Calculator (CPC), and Coding Potential Assessment Tool (CPAT) revealed that neither SNHG10 nor SCARNA13 could encode any protein (Supplementary Fig. S1B–S1E).

Analysis of the TCGA LIHC data repository confirmed that the expression levels of both SNHG10 and SCARNA13 were significantly higher in HCC tissues than in adjacent normal tissues (Supplementary Fig. S1F and S1G), and SCARNA13 expression was statistically correlated with SNHG10 expression (Fig. 1E). qPCR analysis was performed to further validate their expression levels and correlation. Because SNHG10 has two major transcripts of 1,980 nt and 1,341 nt in length (Supplementary Fig. S1H), both should be detected to determine SNHG10 expression. Compared with the long transcript (1,980 nt), the short transcript (1,341 nt) is spliced and lacks an intron. To distinguish these two transcripts, the reverse primer for the long transcript (1,980 nt) was designed in the intron region, while the reverse primer for the short transcript (1,341 nt) was designed across the exon-junction site. The forward primers of these two transcripts were identical. Importantly, PCR and Northern blot analysis identified the 1,341 nt SNHG10, which does not overlap with the whole sequence of the SCARNA13 transcript as the steady and predominant transcript in HCC tissues and HCC cells. In contrast, the 1,980 nt SNHG10 showed negligible expression (Supplementary Fig. S1I and S1J). Hence, the expression of SNHG10 was almost entirely determined by the levels of the 1,341 nt transcript. Although the transcript of 1,980 nt SNHG10 can barely be detected, it overlaps the whole sequence of the SCARNA13 transcript. Consequently, to accurately determine the expression of SCARNA13 by qPCR, a specific stem-loop primer was designed to perform the unique reverse transcription of SCARNA13 (Supplementary Fig. S1K), after which, the product was verified by Sanger sequencing (Supplementary Fig. S1L). Next, we examined the expression levels of SNHG10 and SCARNA13 in 64 pairs of HCC tissues and their adjacent normal tissues from the West China Hospital (Chengdu, Sichuan, China) dataset. As expected, the levels of both SNHG10 and SCARNA13 were elevated in HCC tissues compared with those in adjacent normal tissues (Fig. 1F and G), and there was a statistically positive correlation between them (Fig. 1H).

Furthermore, the relationship between the expression levels of SNHG10 and SCARNA13 and the clinical characteristics were analyzed in 64 HCC tissues (Supplementary Table S5; Supplementary Table S6). Our results showed that high expression of SNHG10 was significantly associated with tumor size (P = 0.024), serum AFP (P = 0.024), microvascular invasion (P = 0.005), Edmondson's grade (P = 0.001), TNM stage (P = 0.011), and BCLC stage (P = 0.002). Likewise, high expression of SCARNA13 was also statistically correlated with tumor size (P = 0.045), serum AFP (P = 0.012), microvascular invasion (P < 0.001), Edmondson's grade (P < 0.001), TNM stage (P = 0.002), and BCLC stage (P < 0.001). More importantly, Kaplan–Meier analysis displayed that high expression of SNHG10 or SCARNA13 was remarkably associated with poor overall survival (Fig. 1I and J). Besides, SCARNA13, rather than SNHG10, could serve as an independent prognostic indicator for overall survival according to univariate and multivariate Cox regression analysis (Supplementary Table S7; Supplementary Table S8).

Because SNHG10 expression was remarkably correlated with SCARNA13 expression, we asked whether there was a regulatory relationship between them. First, the expression of SNHG10 was detected in five different HCC cell lines. Compared with SNU-182, Hep3B, and Huh-7 cells, HCCLM3 and SNU-387 cells displayed relatively high SNHG10 expression (Fig. 1K). Intriguingly, we found that SCARNA13 showed a substantial increase after overexpression of SNHG10 in Huh-7 and Hep3B cells (Supplementary Fig. S2A–S2D), and knockdown of SNHG10 resulted in significantly decreased expression of SCARNA13 in SNU-387 and HCCLM3 cells (Supplementary Fig. S2E–S2H). In contrast, depletion of SCARNA13 did not lead to any change in the expression of SNHG10 (Supplementary Fig. S2I–S2L). These results suggested that SNHG10 might act as the upstream regulator of SCARNA13.

In addition, HCCLM3 cells infected with LV-shSNHG10 exhibited an oval appearance (Supplementary Fig. S2M), whereas Huh-7 and Hep3B cells infected with LV-SNHG10 presented a spindle-shaped appearance (Supplementary Fig. S2N), suggesting that SNHG10 might cause HCC cells to undergo the epithelial–mesenchymal transition (EMT).

To elucidate the oncogenic role of SNHG10 in hepatocarcinogenesis and metastasis, we examined the effects of SNHG10 on cell phenotypes. Depletion of SNHG10 significantly inhibited SNU-387 and HCCLM3 cell cycle and proliferation, and induced apoptosis (Fig. 2A and B; Supplementary Fig. S3A–S3F). Moreover, the silencing of SNHG10 drastically weakened the invasive and migratory abilities of SNU-387 and HCCLM3 cells (Fig. 2C and D; Supplementary Fig. S3G and S3H).

To further investigate the tumorigenic effects of SNHG10 on HCC cells in vivo, SNU-387 and HCCLM3 cells were subcutaneously injected into nude mice. Both the volumes and weights of the tumors in the LV-shSNHG10 group were remarkably lower than those in the LV-shCtrl group (Fig. 2E–G; Supplementary Fig. S4A–S4C), indicating that SNHG10 enhanced the tumorigenicity of the HCC cells in vivo. Furthermore, the promoting effects of SNHG10 on the metastasis of HCC cells were evaluated. We transplanted the indicated SNU-387 and HCCLM3 cells into the livers of nude mice to construct orthotopic-implanted models for liver metastasis assays. The luciferase signal intensities of liver metastatic nodules were significantly declined in the LV-shSNHG10 group compared with those in the LV-shCtrl group (Fig. 2H; Supplementary Fig. S4D), demonstrating that SNHG10 strengthened the intrahepatic metastatic ability of HCC cells. Eventually, SNU-387 and HCCLM3 cells were labeled with firefly luciferase and were directly inoculated into the tail veins of nude mice for lung metastasis assays. Apparently, the luciferase signal intensities of mice in the LV-shSNHG10 group were markedly lower than those in the LV-shCtrl group (Fig. 2I; Supplementary Fig. S4E), suggesting that the lung metastatic potential of HCC cells could be promoted by SNHG10. H&E staining revealed that the metastatic foci in the LV-shSNHG10 group were dramatically decreased in tissue sections of the lungs and livers (Fig. 2J and K; Supplementary Fig. S4F and S4G).

Thereafter, we assessed the influences of overexpressing SNHG10 on the HCC cell phenotype. Upregulation of SNHG10 significantly promoted the cell cycle, proliferation, and apoptotic resistance of Huh-7, Hep3B, and SNU-182 (Supplementary Fig. S4H–S4M; Supplementary Fig. S5A–S5F). Moreover, overexpression of SNHG10 drastically strengthened the invasive and migratory abilities of Huh-7, Hep3B, and SNU-182 cells (Supplementary Fig. S5G–S5L). Regarding in vivo experiments, Huh-7, Hep3B, and SNU-182 cells infected with LV-SNHG10 were significantly associated with higher volumes and weights of the subcutaneous tumors, as well as with greater luciferase signal intensities of liver and lung metastatic nodules (Supplementary Fig. S6A–S6O). Taken together, these observations illuminated that SNHG10 functions as an oncogenic driver in the tumorigenesis and metastasis of HCC.

Because SCARNA13 and SNHG10 were concomitantly upregulated in HCC tissues, we evaluated the influences of SCARNA13 on the malignant phenotype of HCC cells. Similar to SNHG10, knockdown of SCARNA13 tremendously suppressed the cell cycle and proliferation, and induced apoptosis of SNU-387 and HCCLM3 (Fig. 3A–H). Besides, downregulation of SCARNA13 substantially impaired the invasive and migratory capabilities of SNU-387 and HCCLM3 cells (Fig. 3I–L). These results identify SCARNA13 as an oncogenic promoter in HCC.

Generally, the activation of oncogenes mediated by aberrant promoter methylation levels is a key feature of cancer (9). Hence, to ascertain the underlying mechanisms for the elevated expression of SNHG10 and SCARNA13 in HCC, we initially performed methylation analysis of the SNHG10 gene promoter using whole-genome methylation data from the TCGA LIHC dataset. Because SNHG10 and SCARNA13 are processed from the exons and introns of primary RNA transcripts from the SNHG10 gene, respectively, the SCARNA13 gene promoter is theoretically equal to the SNHG10 gene promoter (6). The analysis results showed that neither SNHG10 nor SCARNA13 expression was statistically associated with the methylation levels of the SNHG10 gene promoter (Supplementary Fig. S7A and S7B), revealing that the dysregulation of SNHG10 or SCARNA13 cannot be ascribed to the abnormal methylation levels of the SNHG10 gene promoter.

In addition, the action of transcription factors (TF) in recognizing and dynamically binding to degenerate sequence motifs located at promoters plays a key role in transcription (10). Therefore, we speculated whether certain TFs were responsible for the aberrant expression of SNHG10 and SCARNA13. The intersection of JASPAR (11), PROMO (12), and LASAGNA (13) databases identified 9 TFs that possibly bound to the promoter region of the SNHG10 gene (Fig. 4A). Analysis of the TCGA LIHC dataset demonstrated that c-Myb expression shows the highest correlation with SNHG10 and SCARNA13 expression among the 9 TFs (Fig. 4B and C; Supplementary Fig. S7C–S7P). Therefore, we deduced that c-Myb might cause the elevated SNHG10 and SCARNA13 expression. As expected, knockdown of c-Myb significantly decreased the expression of SNHG10 and SCARNA13 in SNU-387 and HCCLM3 cells (Fig. 4D and E; Supplementary Fig. S7Q–S7S), identifying c-Myb as the upstream regulator of SNHG10 and SCARNA13.

Subsequent bioinformatics analysis predicted five c-Myb binding sites (MBS) in the promoter region of SNHG10 (Fig. 4F and G). Thereafter, ChIP assays confirmed the significantly high enrichment of c-Myb on MBS1 and MBS3 in the promoter region of SNHG10 (Fig. 4H). To further determine the transcriptional activation of c-Myb on the SNHG10 gene promoter, cells were cotransfected with the SNHG10 promoter luciferase reporter (pEZX-PL01-SNHG10) and siRNAs targeting c-Myb. Depletion of c-Myb markedly reduced SNHG10 promoter activity in HEK293T cells (Fig. 4I). Collectively, these data indicated that c-Myb can directly bind to the promoter region of SNHG10, leading to the upregulation of SNHG10 and SCARNA13 in HCC cells.

Theoretically, snoRNAs and the homologous lncRNAs are located in the nucleus and cytoplasm, respectively (6). RNA FISH was implemented to confirm the localization of SNHG10 and SCARNA13, illustrating that SNHG10 is predominantly localized in the cytoplasm, whereas SCARNA13 is preferentially situated in the nucleus (Fig. 5A). Next, we investigated the specific mechanism by which SNHG10 regulated the expression of SCARNA13.

Emerging evidence has confirmed that cytoplasmic lncRNAs can serve as competing endogenous RNAs (ceRNA) to sequester miRNAs, resulting in the release of corresponding miRNA-targeted mRNAs (14, 15). Accordingly, bioinformatics analysis of miRcode (16), lncRNASNP2 (17), and LncBase (18) suggested that miR-150-5p can bind to the 309-315 nt site of SNHG10 (Fig. 5B).

To verify whether SNHG10 and miR-150-5p were involved in the RNA-induced silencing complex (RISC), RIP assays were performed utilizing the anti-Ago2 (the core component of the RISC) antibody. The results showed that both miR-150-5p and SNHG10 are drastically enriched in Ago2 immunoprecipitates compared with those in the IgG pellet in SNU-387 and HCCLM3 cells (Fig. 5C and D), suggesting that SNHG10 physically existed in Ago2-based miRNA-induced repression complex and is associated with miR-150-5p.

More importantly, ChIRP assays were conducted to determine the direct interaction between SNHG10 and miR-150-5p. Ten oligonucleotide probes targeting SNHG10 were divided into an even set and an odd set to increase the specificity of ChIRP assays (19). The data validated the tremendous enrichment of miR-150-5p on SNHG10 in both even and odd probes pools relative to control LacZ probes set in SNU-387 and HCCLM3 cells (Fig. 5E and F). Moreover, transfection of miR-150-5p mimics significantly suppressed the luciferase activity of pmirGLO-SNHG10 that contained full-length SNHG10 at the 3′ UTR of Rluc. In contrast, pmirGLO-SNHG10-mut(miR-150-5p) presented no response to miR-150-5p (Fig. 5G), confirming the sponging function of SNHG10 to miR-150-5p.

Numerous studies have revealed that miR-150-5p interacts with the 3′ untranslated region (UTR) of c-Myb mRNA and overexpression of miR-150-5p downregulates c-Myb mRNA and protein levels (20, 21), identifying c-Myb as a direct target of miR-150-5p. Strikingly, our previous data proved that c-Myb could directly upregulate the expression of SNHG10. Reciprocally, based on the sponging function of SNHG10 to miR-150-5p and the confirmed inhibitory effect of miR-150-5p to c-Myb, we inferred that SNHG10-mediated sequestration of miR-150-5p might be essential for the upregulation of c-Myb. To test this speculation, Huh-7 and Hep3B cells with LV-SNHG10 were transfected with miR-150-5p mimics. The expression of c-Myb was increased upon upregulating SNHG10, whereas miR-150-5p overexpression entirely abolished this effect (Fig. 5H). Conversely, SNU-387 and HCCLM3 cells with LV-shSNHG10 were transfected with miR-150-5p inhibitors. The expression of c-Myb was declined upon SNHG10 knockdown and was thoroughly rescued by the miR-150-5p sponge (Fig. 5I). These findings demonstrated that SNHG10 functions as a sponge for miR-150-5p to decrease its suppressive effect on c-Myb, subsequently enhancing the expression of c-Myb.

Because c-Myb could directly upregulate the expression of SNHG10, it is reasonable to propose that SNHG10 and c-Myb might form a positive feedback loop to sustain the elevated expression of SNHG10 in HCC. In addition, miR-150-5p did not witness any statistical changes after silencing or overexpressing SNHG10 (Supplementary Fig. S8A and S8B), clarifying that SNHG10 could not exert any influences on the expression of miR-150-5p in HCC. Nevertheless, because of the positive feedback loop induced by c-Myb, the expression of SNHG10 experienced a significant decline and growth after transfection with the miR-150-5p mimics and miR-150-5p inhibitors, respectively (Supplementary Fig. S8C–S8J). Together, these data further confirmed the positive feedback loop and ceRNA model involving SNHG10, miR-150-5p, and c-Myb.

On the basis of the aforementioned positive feedback loop, as well as the regulatory effect of c-Myb on SCARNA13, it can be inferred that SNHG10 might affect the expression of SCARNA13 via the positive feedback loop mediated by c-Myb. Accordingly, SCARNA13 would accept the regulation of miR-150-5p and have no influence on miR-150-5p expression. Indeed, the expression of SCARNA13 displayed statistical a decrease and increase after transfection with the miR-150-5p mimics and miR-150-5p inhibitors, respectively (Supplementary Fig. S8K–S8N). On the contrary, the inhibition of SCARNA13 exerted no influence on the expression of miR-150-5p (Supplementary Fig. S8O and S8P). Collectively, these results indicated SCARNA13 as the downstream effector of the positive feedback loop consisting of SNHG10, miR-150-5p, and c-Myb.

To clarify whether c-Myb was indispensable to the modulatory effect of SNHG10 on SCARNA13, rescue experiments were conducted. Specifically, when the upregulation of c-Myb in LV-SNHG10 cells was completely reversed by c-Myb siRNA at the RNA and protein levels, the expression of SCARNA13 was still statistically higher than before (Fig. 6A; Supplementary Fig. S9A and S9B), implying that SNHG10 might modulate SCARNA13 not merely by regulating the expression of c-Myb. Nonetheless, these data could not exclude the possibility that miR-150-5p, rather than SNHG10, influenced SCARNA13 through other mechanisms. Thereafter, we performed further rescue experiments to test this hypothesis. When the expression of c-Myb remained stable at the RNA and protein level in LV-Control cells after cotransfection with the miR-150-5p inhibitors and c-Myb siRNA, the expression of SCARNA13 showed no significant change (Fig. 6A; Supplementary Fig. S9A and S9B), validating that c-Myb completely mediated the regulatory effects of miR-150-5p on SCARNA13 expression. Therefore, it can be seen that the expression level of c-Myb was partially responsible for the upregulation of SCARNA13 caused by overexpressing SNHG10.

Apart from the expression level, the functional activity of c-Myb has been proven to enormously stimulate the downstream genes (22). Recently, lncRNAs have been reported to bind to mRNA and improve the stability of mRNA, resulting in increased protein levels (15). Thus, we speculated that certain mRNAs bound by SNHG10 might regulate the functional activity of c-Myb. To verify the hypothesis, ChIRP-seq was conducted to pull down endogenous mRNAs bound by SNHG10 and sequence the retrieved RNA (Supplementary Fig. S9C). With the intersection of 407 gene transcripts found by ChIRP-seq and 49 protein molecules proven to directly interact with c-Myb in the BioGRID interaction database (23), two genes were screened out, namely ribosomal protein L4 (RPL4) and damage-specific DNA binding protein 1 (DDB1; Fig. 6B). Compared with DDB1, RPL4 displayed significantly higher correlation with SNHG10 in the TCGA LIHC dataset (Fig. 6C and D). More strikingly, a previous study has described that RPL4 interacts with c-Myb and positively regulates the transcriptional activity of c-Myb (24). Moreover, the expression of RPL4 showed significant reductions at the RNA and protein levels after transfection with siRNAs targeting SNHG10 (Supplementary Fig. S9D–S9F). Thus, we deduced that SNHG10 might regulate the functional activity of c-Myb by affecting the stability of RPL4 mRNA in HCC cells. To confirm this speculation, ChIRP-qPCR was initially implemented to verify the direct combination between SNHG10 and RPL4 mRNA in SNU-387 and HCCLM3 cells, indicating that both even and odd probes pools targeting SNHG10 presented significantly higher enrichment of RPL4 mRNA than the control LacZ probes set (Fig. 6E and F). Subsequently, we identified seven regions of highly complementary sequences between SNHG10 and RPL4 mRNA utilizing BLAST (http://blast.ncbi.nlm.nih.gov/; Supplementary Fig. S9G). Thereafter, these seven binding sites were totally mutated in the full-length of SNHG10. SNU-387 and HCCLM3 cells were treated with actinomycin D to interrupt new RNA synthesis and the loss percentage of RPL4 was measured within a 24-hour period. The findings illustrated that the overexpression of SNHG10, rather than that of SNHG10-mut(RPL4), prolongs the half-life of RPL4 mRNA (Fig. 6G and H).

Next, to evaluate the positive regulatory effect of RPL4 on the functional activity of c-Myb in HCC cells, luciferase reporters containing c-Myb recognition elements (MRE) were constructed. Subsequently, the luciferase reporter assays revealed that RPL4 could enhance the activity of MREs (Fig. 6I). Furthermore, coimmunoprecipitation assays were performed to determine the interaction between RPL4 and c-Myb, validating the drastically higher enrichment of RPL4 in the anti–c-Myb group than in the IgG group (Fig. 6J). Reciprocally, the anti-RPL4 group exhibited substantially increased enrichment of c-Myb compared with the IgG group (Fig. 6K). The effectiveness of siRNAs targeting RPL4 at the RNA and protein levels was verified in SNU-387 and HCCLM3 cells (Supplementary Fig. S9H–S9J). Eventually, further rescue experiments demonstrated that the expression of SCARNA13 did not witness any statistical change in LV-SNHG10 cells after cotransfection with c-Myb siRNA and RPL4 siRNA (Fig. 6L). Overall, our data demonstrated that SNHG10 promoted the expression of c-Myb by, on one hand, absorbing miR-150-5p and, on the other hand, by enhancing the transcriptional activity of c-Myb through interacting with RPL4, consequently modulating the expression of SCARNA13.

Eventually, we investigated whether this circuitry could function under liver physiologic conditions using human normal liver epithelial cells THLE-2 and THLE-3. The results revealed that the expression of SCARNA13 showed no statistical change after overexpressing SNHG10 in THLE-2 and THLE-3 cells (Supplementary Fig. S9K–S9N). Because c-Myb was responsible for the upregulation of SCARNA13 caused by overexpressing SNHG10, we subsequently investigated whether the expression of c-Myb could be affected by SNHG10. Western blot analysis illustrated that upregulating SNHG10 exerted no influence on the expression of c-Myb in THLE-2 and THLE-3 cells (Supplementary Fig. S9O). Because SNHG10 functioned as a sponge for miR-150-5p to decrease its suppressive effect on c-Myb, subsequently enhancing the expression of c-Myb in HCC cells, we speculated that SNHG10-mediated sequestration of miR-150-5p might not operate due to the enormously high expression of miR-150-5p in THLE-2 and THLE-3 cells. Indeed, the expression levels of miR-150-5p in THLE-2 and THLE-3 cells were substantially higher than those in HCC cells (Supplementary Fig. S9P), indicating that miR-150-5p is indispensable to the activation of this circuitry in HCC cells.

On the basis of SNHG10 and SCARNA13 coordinately contributing to the malignant phenotype of HCC cells and SNHG10 modulating the expression of SCARNA13, it can be deduced that the tumor-promoting effects of SNHG10 on HCC cells might be mediated by SCARNA13. Accordingly, SCARNA13 ASOs were transfected into LV-SNHG10 cells. Knockdown of SCARNA13 significantly rescued the influences of SNHG10 overexpression on cell proliferation, invasion, and migration (Fig. 7A and B; Supplementary Fig. S10A and S10B). These findings indicated that SCARNA13 mediates the tumor-promoting function of SNHG10.

To gain insights into the molecular mechanism underlying the oncogenic role of SCARNA13 in HCC, RNA-seq analysis was implemented for SCARNA13 knockdown (Supplementary Fig. S10C). Silencing of SCARNA13 downregulated the expression of 188 genes and upregulated 275 genes. Analysis of the KEGG pathway showed that the cell cycle, TGFβ, PI3K-Akt, and p53 signaling pathways were influenced by inhibiting SCARNA13 (Supplementary Fig. S10D). In addition, depletion of SCARNA13 affected cell adhesion, cell proliferation, angiogenesis, and the apoptotic process in Gene Ontology analysis (Supplementary Fig. S10E), confirming the oncogenic activity of SCARNA13 in hepatocarcinogenesis and metastasis. To further identify candidate downstream factors of SCARNA13, quantitative proteomic analysis was conducted using the Tandem mass tag labeling method, revealing 182 proteins that were differentially expressed (Supplementary Fig. S10F). Eleven genes were screened out via taking the intersection of transcriptomics and proteomics (Fig. 7C). Among them was SOX9, whose expression showed the most significant decrease at the protein level after SCARNA13 knockdown. Hence, SOX9 was selected as the candidate downstream protein of SCARNA13. Accordingly, inhibition of SCARNA13 resulted in the remarkably decreased expression of SOX9 at the RNA and protein levels in SNU-387 and HCCLM3 cells (Fig. 7D–F). Numerous studies have illustrated that SOX9 exerts enormous impacts on the cell cycle and EMT of cancer cells (25–27). Therefore, the influences of SCARNA13 knockdown on the expression of molecular markers of cell cycle and EMT were detected. As expected, corresponding changes were observed for the expression of these molecular markers. Particularly, the expression levels of CDK2, CDK3, Cyclin D1, Cyclin D3, ZEB1, Vimentin, N-cadherin, and Fibronectin presented statistically downward trends after suppressing SCARNA13 in SNU-387 and HCCLM3 cells. Conversely, transfection with SCARNA13 ASOs led to significantly increased expression of p21, p27, and E-cadherin in HCC cells (Fig. 7G). Overall, these findings illuminated that the promoting effects of SNHG10 on tumorigenesis and metastasis are mediated by SCARNA13, which promotes the cell cycle and EMT via upregulating SOX9 in HCC cells.

Hepatocarcinogenesis is regarded as a multistage process involving genetic and epigenetic alterations, as well as extrinsic microenvironment factors, that ultimately result in the malignant transformation of hepatocytes (28). Presently, HCC is frequently diagnosed at an advanced stage, because of insufficient progress in the identification of ideal diagnostic biomarkers for HCC, providing patients with limited therapies. More seriously, postoperative recurrence, mainly caused by intrahepatic metastasis, and extrahepatic metastasis, which most commonly occurs in the lungs, primarily explains the poor prognosis of patients with HCC (29). However, there has been little success in the exploitation of effective interventions against HCC metastasis. To determine the responsible regulators for hepatocarcinogenesis and metastasis, we performed RNA-seq based on lung metastasis screening mice models. By taking the intersection of sequencing data and abnormally expressed genes in HCC tissues from the TCGA LIHC dataset, we identified lncRNA SNHG10 and its homologous SCARNA13 as potential oncogenic drivers for hepatocarcinogenesis and metastasis.

Some 5′ terminal oligopyrimidine (5′ TOP) RNA transcripts from SNHGs contain only short, poorly conserved ORFs and are, therefore, considered as lncRNAs from the perspective of the structure and function (6). Recently, several SNHGs have been identified to be the critical factors for carcinogenesis and metastasis (30). For example, SNHG6 suppresses MAT1A expression by activating the miR-1297/FUS pathway to regulate the global DNA methylation levels, thus stimulating the phenotype of hepatoma cells (31). SNHG15 maintains Slug stability in colon cancer cells through interaction with the zinc finger domain of Slug, promoting colon cancer cell migration (32). In this work, we first reported the cancer-promoting role of lncRNA SNHG10 in cancer. Specifically, overexpression of SNHG10 was observed in 64 HCC tissues from the WCH dataset, and was statistically associated with the clinical characteristics and prognosis of patients with HCC. Next, the phenotypic assays illustrated that SNHG10 exerted remarkable facilitating effects on cell proliferation, the cell cycle, apoptosis resistance, invasion, and metastasis in vitro and in vivo, determining the tumorigenic and metastasis-driving functions of SNHG10 in HCC cells.

In addition, our data uncovered several dysregulated snoRNAs in the process of hepatocarcinogenesis and metastasis. As the processed product of primary RNA transcripts from SNHGs, snoRNAs play regulatory roles primarily by modifying rRNAs, acting as the precursors of miRNAs, and affecting RNA splicing variants in the development and progression of cancer (6). In addition, small Cajal body–specific RNAs (scaRNA), located in small membraneless subcompartments in the cell nucleus (Cajal bodies) rather than in the nucleolus, are a subset of snoRNAs (33). However, although scaRNAs structurally resemble snoRNAs, the specific roles of scaRNAs in oncogenesis are not well studied. In this study, we provided the first evidence of scaRNA dysregulation in HCC. Precisely, SCARNA13 identified by our sequencing data and the TCGA LIHC dataset showed elevated expression in 64 HCC tissues from the WCH dataset. In addition, we found that SCARNA13 is an independent prognostic indicator for the overall survival of patients with HCC. SCARNA13 dramatically contributed to the malignant phenotypes of HCC cells, confirming the tumor-promoting function of SCARNA13 in HCC cells.

To our knowledge, regulatory relationships exist between miRNAs and lncRNAs, particularly the transcripts from the host genes of miRNAs (7). For example, lncRNA MIR100HG–derived miR-100 and miR-125b were reported to upregulate MIR100HG by affecting the transcription factor GATA6 (7). Therefore, we speculated there might be similar modulatory correlation between snoRNAs and lncRNAs from SNHGs. Intriguingly, our results demonstrated that SNHG10 exerts significant regulatory effects on the expression of SCARNA13. To be specific, SNHG10 served as a sponge for miR-150-5p to abolish the suppressive effect of miR-150-5p on c-Myb, resulting in the elevated expression of c-Myb. On the other hand, SNHG10 promoted the expression of RPL4 by boosting the stability of RPL4 mRNA, leading to the improvement of functional activity of c-Myb based on the direct interaction between RPL4 and c-Myb. Reciprocally, overexpressed and hyperactivated c-Myb enhanced the expression of SNHG10 and SCARNA13 through directly binding to the promoter region of SNHG10, thereby regulating its promoter activity and forming a positive feedback loop in HCC cells. Consequently, activation of this feedback loop continuously stimulated the expression of SCARNA13 (Fig. 7H). Collectively, SNHG10 modulated its homologous SCARNA13 via a positive feedback loop to facilitate the development and progression of HCC. Our findings identified the responsible lncRNAs and snoRNAs in hepatocarcinogenesis and metastasis, and presented novel implications for the molecular mechanisms of lncRNAs and snoRNAs in cancer research. More strikingly, we propose a complex regulatory relationship between lncRNAs and its homologous snoRNAs.

In conclusion, we have identified a complex circuitry underlying the concomitant upregulation of SNHG10 and its homologous SCARNA13 in HCC. SNHG10 regulates c-Myb by sponging miR-150-5p and interacting with RPL4 mRNA, modulating the expression of SCARNA13 and its downstream effector SOX9 in HCC cells. The investigation of SNHG10 and SCARNA13 provides a luminous comprehension of hepatocarcinogenesis and metastasis that may develop effective strategies for the diagnosis and treatment of HCC.

No potential conflicts of interest were disclosed.

Conception and design: T. Lan, K. Yuan, X. Hao, J. Wang, Y. Zeng, H. Wu

Development of methodology: T. Lan, L. Xu, J. Wang, X. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Lan, L. Xu, H. Liao, X. Hao, H. Liu, J. Li, M. Liao

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Lan, X. Yan, X. Hao

Writing, review, and/or revision of the manuscript: T. Lan, K. Yuan, K. Xie, J. Huang, Y. Zeng

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Liao, H. Wu

Study supervision: K. Yuan, Y. Zeng, H. Wu

We thank Li Chai and Yan Wang from the Core Facility of West China Hospital (Chengdu, Sichuan, China) for technical assistance and Shanghai Lu-Ming Biotech Co., Ltd. (Shanghai, China) for assistance with quantitative proteomic analysis. This study was supported by grants from the Natural Science Foundation of China (81872004, 81800564, 81770615, 81700555, 81672882, and 81502441), National Key Technologies R&D Program (2018YFC1106803), the Science and Technology Support Program of Sichuan Province (2017SZ0003, 2018SZ0115), the Science and Technology Program of Tibet Autonomous Region (XZ201801-GB-02), and the 1.3.5 project for disciplines of excellence, West China Hospital, Sichuan University (ZYJC18008).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.