SNHG16/miR‐605‐3p/TRAF6/NF‐κB feedback loop regulates hepatocellular carcinoma metastasis

Abstract The mechanism by which miR‐605‐3p regulates hepatocellular carcinoma (HCC) metastasis has not been clarified. In this study, we found that miR‐605‐3p was down‐regulated in HCC and that low miR‐605‐3p expression was associated with tumour thrombus and tumour satellites. HCC patients with low miR‐605‐3p expression showed shorter overall survival and disease‐free survival after surgery. Overexpression of miR‐605‐3p inhibited epithelial‐mesenchymal transition and metastasis of HCC through NF‐κB signalling by directly inhibiting expression of TRAF6, while silencing of miR‐605‐3p had the opposite effect. We also found that SNHG16 directly bound to miR‐605‐3p as a competing endogenous RNA. Mechanistically, high expression of SNHG16 promoted binding to miR‐605‐3p and inhibited its activity, which led to up‐regulation of TRAF6 and sustained activation of the NF‐κB pathway, which in turn promoted epithelial‐mesenchymal transition and metastasis of HCC. TRAF6 increased SNHG16 promoter activity by activating NF‐κB, thereby promoting the transcriptional expression of SNHG16 and forming a positive feedback loop that aggravated HCC malignancy. Our findings reveal a mechanism for the sustained activation of the SNHG16/miR‐605‐3p/TRAF6/NF‐κB feedback loop in HCC and provide a potential target for a new HCC treatment strategy.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related deaths worldwide, with nearly half of all cases occurring in China. 1 The recurrence and metastasis rates of HCC after surgery remain high, and the prognosis of HCC patients is poor. 2 Epithelialmesenchymal transition (EMT) is an important cause of HCC metastasis. 3 However, its underlying mechanism remains unclear.
Therefore, exploring the molecular mechanism of EMT in HCC is not only critical for the further understanding of metastasis in HCC, but may also provide clues towards identifying early diagnostic markers of HCC and new therapeutic targets.
Non-coding RNAs are a class of RNAs that lack protein-coding ability and are involved in regulating several cellular processes, including malignant biological activity in cancers. 4 These molecules are classified as short non-coding RNAs or long non-coding RNAs (lncRNAs) according to their length. MicroRNAs (miRNAs) are a class of short non-coding RNAs of 21-23 bp in length that mediate RNAinduced silencing complex (RISC) formation at the 3′-untranslated region (UTR) of target genes, which results in target gene degradation or translational inhibition. Previous studies have shown that miRNAs are involved in the regulation of cellular processes such as organ formation, fat metabolism, cell proliferation and apoptosis. 5 Furthermore, studies have shown that abnormal miRNA regulation may be a critical factor leading to HCC metastasis. 6 LncRNAs are non-coding RNAs of more than 200 bp in length that regulate chromatin modification, transcription, miRNAs, mRNA stability, protein function and other important regulatory functions. 7,8 LncRNAs exert important functions as critical regulators of various cell processes, and their abnormal expression may play an important role in cancer metastasis. 9 Competitive endogenous RNAs (ceRNAs) represent a novel gene expression regulatory mechanism for lncRNA-miRNA-mRNA interactions. 10 Recent studies showed that ceRNAs are involved in regulating various cellular activities such as HCC metastasis. 11 LncRNAs act as miRNA sponges and competitively bind to common miRNAs through the miRNA response element, which attenuates the miRNA-mediated inhibition of target genes and increases target gene expression. The target gene in turn regulates the progression of cancer metastasis. miRNAs are the core RNA molecules in the ceRNA hypothesis. 10 miR-605-3p is a recently discovered miRNA with tumour suppressor functions. [12][13][14] Overexpression of miR-605-3p inhibits the migration and invasion of bladder cancer and glioma cells. 13,14 However, its expression and function in HCC have not been reported.
In this study, we revealed a tumour-suppressive function of miR-605-3p in human HCC for the first time. Mechanistically, miR-605-3p targets TRAF6, a signal transducer in the NF-κB signalling pathway that has a crucial role in activating NF-κB signalling to suppress TRAF6 expression and thereby repress NF-κB signalling. We further found that SNHG16 is overexpressed in HCC cells and can directly bind to and affect miR-605-3p function, which leads to up-regulation of TRAF6 and continuous NF-κB activation in HCC. In turn, TRAF6 up-regulates SNHG16 expression via NF-κB/p65.

| Cell lines, cell culture and reagents
HCCLM3, MHCC97L and MHCC-97H cell lines were gifts from the Liver Cancer Institute, ZhongShan Hospital. A normal hepatocyte cell line (L02) and two HCC cell lines (Hep3B and HepG2) were purchased from GeneChem. All cells were cultured in Dulbecco's modified Eagle's medium containing 10% foetal bovine serum with 100 U/mL penicillin and 100 mg/mL streptomycin in a humidified incubator at 37°C containing 5% CO 2 . SN-50 was purchased from MedChem Express.

| Online bioinformatics analysis
Putative miR-605-3p target genes were predicted by TargetScan and RELA (NF-κB/p65), SNHG16 expression and the prognostic significance of SNHG16 expression in the liver hepatocellular carcinoma data set of TCGA database.

| Chromatin immunoprecipitation (ChIP) assays
Cells were cross-linked with 1% formaldehyde and quenched in glycine solution. ChIP assays were performed using a Pierce Magnetic ChIP Kit (Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer's protocol. Anti-p65 antibody and normal IgG (MultiSciences) were used for immunoprecipitation. ChIP-enriched DNA samples were analysed by qRT-PCR to quantify the putative p65-binding sites in the SNHG16 promoter region. The data are shown as relative enrichment normalized to control IgG. Primer sequences for ChIP assays were as follows: forward, 5′-CCTGGTAAGTGCTATGAAGT-3′; reverse, 5′-TCTATCCCTGCAAACATAGT-3′.  Renilla luciferase activity and presented as relative luciferase activity.

| miR-605-3p is down-regulated in HCC and is associated with poor prognosis
To investigate the role of miR-605-3p in HCC, we first detected the expression levels of miR-605-3p in 78 HCC and paired adjacent normal tissues. The results showed that miR-605-3p expression was down-regulated in HCC tissues ( Figure 1A). Analysis of the clinicopathological characteristics showed that low miR-605-3p expression was positively associated with tumour thrombus and tumour satellites in the 78 HCC patients (Table 1). qRT-PCR

| miR-605-3p suppresses HCC cell metastasis in vitro and in vivo
qRT-PCR analyses showed that miR-605-3p expression was highest in HepG2 cells and lowest in HCCLM3 cells among the five HCC cell lines examined in this study ( Figure S1A). Therefore, HCCLM3 and HepG2 cells were chosen for subsequent experiments. We confirmed the efficiency of overexpression or silencing of miR-605-3p in both HCC cell lines ( Figure S1B). Wound healing and Matrigel invasion assays showed that overexpression of miR-605-3p specifically suppressed HCCLM3 cell migration and invasion (Figure 2A, B), while miR-605-3p silencing led to increased HepG2 cell migration and invasion ( Figure 2C, D).
We further investigated whether miR-605-3p expression regulated the metastatic ability of HCC cells in vivo. We found that the number and size of metastatic colonies on the lung surface of mice were largely decreased in the HCCLM3/miR-605-3p group ( Figure 2E).
We also detected metastatic colonies in the HepG2/anti-miR-605-3p group, while no metastatic colonies were found in the HepG2/anti-miR-NC group ( Figure 2F). These results indicated that miR-605-3p inhibited the metastatic ability of HCC cells in vitro and in vivo.

| miR-605-3p inhibits EMT in HCC cells
EMT is a critical process involved in cancer metastasis. 15
Among the identified genes, TRAF6 was identified as a gene of interest, as previous studies showed that TRAF6 regulates the activation of NF-κB signalling and affects NF-κB-mediated EMT in carcinogenesis and cancer development. 17

| Reciprocal negative regulation between miR-605-3p and SNHG16
Accumulating evidence has revealed that lncRNAs can function as ceRNAs for miRNAs. To determine whether any lncRNAs regulate TRAF6 expression through miR-605-3p, we searched for potential lncRNAs that bind to miR-605-3p using an online bioinformatic tool (Starbase). The small nucleolar RNA host gene (SNHG) family caught our attention. Among the predicted SNHG family members, SNHG16 expression was highest in HCC tissues and predominantly located in the cytoplasm of HCC cells ( Figure S2A; Figure 5A, B). SNHG16 expression was up-regulated in the 78 HCC tissues compared with adjacent non-tumour tissues ( Figure 5C, D) as well as in HCC cell lines ( Figure S2B) and in the liver HCC data set of TCGA database ( Figure S2C). SNHG16 expression was also negatively correlated with miR-605-3p expression in the 78 HCC tissues ( Figure 5E, F). High SNHG16 expression was not only positively associated with tumour diameter, tumour thrombus, presence of an envelope and tumour satellites in the 78 HCC patients (Table S1), but also served as an independent prognostic indicator for both OS and DFS ( Figure S3A-D; Table S2).
To validate miR-605-3p as a target for SNHG16, dual-luciferase reporter assays were performed ( Figure 5G). The results showed that cotransfection of WT-SNHG16 and miR-605-3p in HCCLM3 cells ( Figure 5H) and MHCC-97H cells ( Figure 5I) decreased luciferase activity compared with the control group. However, the luciferase activity in the MUT-SNHG16 group was not affected.
We further examined whether To examine whether SNHG16 could sponge miR-605-3p expression in an RISC-dependent manner, RNA immunoprecipitation assays were conducted. SNHG16 and miR-605-3p were more abundant in the AGO2 pellet compared with the IgG pellet ( Figure 5N, 5O).
These data indicated that SNHG16 may act as a ceRNA by sponging miR-605-3p.

| Down-regulation of SNHG16 suppresses HCC metastasis, EMT and NF-κB activation by interacting with miR-605-3p in vitro and in vivo
We found that SNHG16 knockdown inhibited EMT and NF-κB signalling activity in HCC cells ( Figure S4A nude mice. We found that SNHG16 silencing decreased the number and size of metastatic colonies from HCCLM3 cells ( Figure 6A) and MHCC-97H cells ( Figure 6B) on the lung surface of mice. However, the effect of SNHG16 silencing was partly abolished by miR-605-3p silencing.
A subcutaneously implanted tumour model in nude mice showed that SNHG16 silencing decreased both the weight and growth rate of mice compared with mice injected with control HCCLM3 or MHCC-97H cells ( Figure S6A-E). The effects of SNHG16 silencing in HCCLM3 and MHCC-97H cells were partly reversed by miR-605-3p silencing. We examined E-cadherin, vimentin, TRAF6 and p65 expression in the subcutaneous tumours by IHC and found that SNHG16 silencing increased E-cadherin expression, decreased vimentin expression, decreased TRAF6 expression and decreased p65 nuclear signals and that these effects were partly abolished by miR-605-3p silencing ( Figure 6C, D).

| SNHG16 and NF-κB form a positive feedback loop
We further examined whether any of the identified factors act upstream of SNHG16 in HCC. We analysed the promoter sequence of SNHG16 using the PROMO, JASPAR and LASAGNA databases and found that RELA (NF-κB/p65) was identified as a potential binding factor to the SNHG16 gene promoter in all three databases ( Figure 7A). We also found a positive correlation between the expression of SNHG16 and NF-κB/p65 in HCC using the GEPIA database ( Figure 7B; R = .26, P < .001). These findings suggest that the promoter region of SNHG16 may contain a binding motif for NF-κB/p65 ( Figure 7C, D). SNHG16 expression was up-regulated in HCCLM3 and MHCC-97H cells in response to ectopic expression of NF-κB/p65 compared with the empty vector group ( Figure 7E).
As NF-κB is a downstream target of TRAF6, we examined whether TRAF6 overexpression significantly elevated SNHG16 expression ( Figure 7F). However, treatment with the NF-κB inhibitor SN-50 largely abolished the effect of TRAF6 on SNHG16 expression ( Figure 7G). We next examined whether NF-κB/p65 interacted with promoter region of SNHG16 via the predicted binding site.
ChIP assays demonstrated that NF-κB/p65 binds the SNHG16 promoter ( Figure 7H). Dual-luciferase reporter assays showed that ectopic expression of NK-κB/p65 enhanced luciferase activity driven by the WT-SNHG16 promoter ( Figure 7I). When the NF-κBbinding sequence in the SNHG16 promoter was mutated, luciferase expression was significantly decreased. These results indicate that NF-κB/p65 interacted with the SNHG16 promoter via the predicted binding site.

| D ISCUSS I ON
In the present study, we revealed that the putative tumour suppressor miR-605-3p was frequently silenced in HCC cell lines and tis- The ceRNA hypothesis proposes that numerous non-coding RNAs may function as molecular sponges for miRNAs and thus functionally liberate RNA transcripts that are targeted by these miRNAs.
Through bioinformatics analysis and PCR validation, SNHG16 was determined to likely function as a ceRNA with miR-605-3p and affects the expression of TRAF6. SNHG16 has been identified as an oncogene in many cancers. [25][26][27] Up-regulation of SNHG16 predicts poor prognosis and induces sorafenib resistance in HCC. 28  One of the most interesting findings in this study was that NF-κB can also regulate the expression of SNHG16. While SNHG16 has been identified as an oncogene in many cancers including HCC, [28][29][30][31][32][33][34] the biological functions of SNHG16 and its underlying mechanisms in HCC are not fully understood. Positive feedback regulation is common in the regulation of many biological functions, especially sustained activation of cancer-promoting signalling pathways. 35 Multiple studies have shown that abnormal activation of the NF-κB pathway occurs in various cancers. 36,37 However, the mechanism for its continued activation in HCC remains unclear. We found that SNHG16 acts as a ceRNA to compete for miR-605-3p, thereby pro- reversed the effects of LPS. 38,39 Furthermore, LPS is known as an NF-κB pathway activator. 40 We speculated whether NF-κB, as a transcription factor, could promote SNHG16 expression by binding to its promoter region. Thus, we analysed the promoter sequence of SNHG16 using the PROMO algorithm and JASPAR. The results revealed the presence of a putative binding site for NF-κB within the SNHG16 promoter region. Moreover, we found a positive correlation between the expression of SNHG16 and p65 in HCC in the GEPIA database. Taken together with the ChIP and dual-luciferase reporter assays, these results demonstrated that activation of the NF-κB signalling pathway promotes SNHG16 expression.
In summary, the present study demonstrates the tumour-suppressive role of miR-605-3p in HCC metastasis for the first time and indicates the importance of the interactions between SNHG16, to improve the treatment and survival of HCC patients ( Figure 7J).

ACK N OWLED G EM ENTS
We thank the Liver Cancer Institute of ZhongShan Hospital

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used and/or analysed during the current study are available from the corresponding author on reasonable request.