Retraction Note: The lincRNA-ROR/miR-145 axis promotes invasion and metastasis in hepatocellular carcinoma via induction of epithelial-mesenchymal transition by targeting ZEB2

Editor's Note: this Article has been retracted; the Retraction Note is available at https://doi.org/10.1038/s41598-023-32251-x

underlying mechanisms in HCC metastasis. It was proved that linc-ROR acted as a sponge for miR-145 to de-repress the expression of the target gene ZEB2, thereby inducing EMT and promoting HCC metastasis. The findings will extend our understanding of the role of lncRNAs in aggressive and metastatic HCC.

Results
Linc-ROR is upregulated in HCC tissues and related to tumor metastasis and poor prognosis of HCC patients. We performed qRT-PCR to detect the expression of linc-ROR in 20 paired of HCC and normal liver tissues. Results showed that linc-ROR was significantly upregulated in HCC tissues (Fig. 1A), which prompts us to investigate the clinical significance of linc-ROR in HCC development. Next, we extended our linc-ROR quantification assay to a cohort of 88 HCC tissues. Higher linc-ROR expression was observed in HCC tissues with lymph node metastasis (LNM) or vascular invasion than those tissues without LNM or vascular invasion (Fig. 1B). Besides, higher linc-ROR expression was related to advanced tumor node metastasis (TNM) stage (Fig. 1C). Most importantly, HCC patients with recurrence showed significant higher linc-ROR expression, comparing to tumors without recurrence (Fig. 1D). Patients were then distributed into high-linc-ROR and low-linc-ROR groups according to the linc-ROR expression in HCC tissues. The correlations between linc-ROR expression and clinic pathological characteristics were presented in Table 1. High linc-ROR expression was observed to be closely correlated with advanced TNM stage, higher incidence of lymph node metastasis and recurrence of HCC patients (p = 0.001, 0.019 and 0.000, respectively). However, there was no significant correlation between Patients were grouped according to ROR expression in HCC tissues. The median expression level was used as the cutoff. Each experiment was performed at least three times. *p < 0.05, **p < 0.01 and ***p < 0.001.
linc-ROR expression and other clinical pathological characteristics including gender, age, tumor diameter, alcohol intake, family history and liver function of patients. Furthermore, Kaplan-Meier survival analyses indicated that HCC patients with high linc-ROR expression had poorer prognosis than those with low linc-ROR expression, together with a shorter disease-free survival (DFS) and overall survival (OS) (Fig. 1E). These results indicated that linc-ROR closely correlates with the increase of HCC metastatic potential.

Linc-ROR significantly promotes in vitro invasion and in vivo metastasis of HCC cells.
To further determine whether linc-ROR is involved in the molecular etiology of metastasis in HCC, we examined the expression of linc-ROR in HCC cell lines with different metastatic potentials. QRT-PCR results showed that linc-ROR increased progressively from normal human hepatocyte cell line (HH) to low metastatic HCC cell lines (HepG2 and SMMC-7721), and finally to highly metastatic HCC cell lines (HCCLM3 and MHCC97-H) ( Fig. 2A) 16 . To explore the biological significance of linc-ROR in HCC, HepG2 or SMMC-7721 cells were stably transfected with linc-ROR overexpression plasmid or control vector respectively and HCCLM3 or MHCC97-H cells were stably transfected with shROR or shCtrl plasmid, respectively. Satisfactory transfection efficiency was obtained at 48 h post-transfection (Fig. 2B). We chose the most efficient shROR plasmid for the forthcoming experiment. Further, we investigated the roles of linc-ROR in migration and invasion of HCC cells. Wound healing assay showed that linc-ROR upregulation significantly increased the capacity of mobility in HepG2 (or SMMC-7721) cells. In contrast, wound healing speed was reduced after linc-ROR knockdown in HCCLM3 (or MHCC97-H)  Then, to further assess the effects of linc-ROR in the in vivo metastasis, we made orthotopic implantation tumor models to test HCC cell invasive behavior changes. After 8 weeks, it was observed that the incidence and number of both intrahepatic and lung metastasis in the linc-ROR group was significantly increased, in compared with the control group. Likewise, the effect of linc-ROR downregulation on the in vivo metastasis of HCC cells was also determined. The incidence and number of both intrahepatic and lung metastasis in the shROR group was significantly decreased, as compared to the control group (Fig. 2E). Furthermore, the difference was further confirmed by hematoxylin and eosin (HE) staining of liver and lung sections (Fig. 2F). On the whole, these in vitro and in vivo results verified the role of linc-ROR in HCC metastasis.

Linc-ROR induces EMT in HCC cells.
EMT, a biological process in which cancer cells lose their epithelial polarity and undergo transition into a mesenchymal phenotype, plays a vital role in tumor invasion and metastasis 17 . Accumulating evidence indicates that various regulators, including lncRNAs, can motivate EMT to increase cancer migration and invasion 18 . Although there have been some studies on the contribution of lncR-NAs during tumor EMT process, much less is known about the role of linc-ROR during HCC EMT process. Therefore, it is worth investigating whether the EMT phenotype of HCC cells was affected by linc-ROR expression. Western blotting and immunofluorescence indicated that linc-ROR upregulation significantly decreased the expression of epithelial markers (E-cadherin and β-catenin) and increased the expression of mesenchymal markers (N-cadherin and Vimentin) in HepG2 (or SMMC-7721) cells ( Identification of miR-145 as a target of linc-ROR. LncRNAs are proved to work as sponges to recruit miRNAs, resulting in upregulation of target genes 19 . To verify whether linc-ROR has a similar mechanism in HCC cells, we used bioinformatics tool Miranda to predict the potential miRNA binding sites in linc-ROR, together  considering previously reported miRNAs 20 . In turn, we measured the expression levels of these 15 predicted miR-NAs in linc-ROR downregulated HCCLM3 cells (Supplementary Table 1). We then focused on miR-145, which exhibited the greatest change. Previously studies have revealed that linc-ROR could act as a sponge for miR-145 in various cancers, such as colorectal cancer, pancreatic cancer and breast cancer 9,12,21 . Also, miR-145 has been implicated to function as tumor suppressor and inhibit cell invasion, migration and metastasis in HCC 22,23 . We hypothesized that the functions of linc-ROR in HCC cells might be mediated by miR-145. In order to prove the hypothesis, we first examined the expression level of miR-145 in HepG2 and SMMC-7721 cells stably transfected with linc-ROR, finding that miR-145 expression was significantly downregulated (Fig. 4A). Meanwhile, miR-145 expression levels in both cells transfected with shROR were significantly upregulated (Fig. 4B). In a word, the miR-145 expression level correlated negatively with linc-ROR expression level. Additionally, we also found that there was no difference in linc-ROR levels after knockdown of miR-145 or ectopic expression (Fig. 4C,D). To validate the direct binding between linc-ROR and miR-145 at endogenous levels, luciferase reporter assays were constructed, which contain wild-type (WT) or mutated (MUT) miR-145 binding sites. Results showed that miR-145/ mimics reduced the luciferase activity of wild-type (WT) linc-ROR reporter vector but not of empty vector and complete mutant reporter vector (Fig. 4E). These data revealed that miR-145 binds to linc-ROR, not inducing the degradation of linc-ROR. It is well known that miRNAs bind their targets and cause translational repression and RNA degradation in an Ago2-dependent manner 24 . To determine whether linc-ROR was regulated by miR-145 in such a manner, we conducted anti-Ago2 RIP in HepG2 cells transiently overexpressing miR-145. Results showed that endogenous linc-ROR pull-down was specifically enriched in miR-145 transfected cells (Fig. 4F), supporting that miR-145 is a bona fide linc-ROR-targeting miRNA. To serve as a sponge, the abundance of linc-RoR should be comparable to or higher than miR-145. We therefore used quantitative real-time PCR to quantify the exact copy numbers of linc-RoR and miR-145 per cell in normal HH cells and HCCLM3 cells (Fig. 4G). As a result, we found that in HCCLM3 cells, the expression level of miR-145 was only about 30 copies per cell, whereas linc-ROR level was approximately 200 copies per cell. In HH cells, miR-145 was up to about 300 copies per cell, whereas linc-ROR level was no more than 40 copies per cell. This result implied that linc-ROR may be able to function as a sponge for miR-145.

Inhibition of miR-145 mimics the biological functions of linc-ROR upregulation in HCC cells.
To and Fig. S1A). The migratory and invasive effect of linc-ROR overexpression in HepG2 (or SMMC-7721) cells was also partially rescued by co-transfection with miR-14/mimics, whereas the anti-metastasis effect induced by linc-ROR downregulation in HCCLM3 (or MHCC97-H) cells could be partially reversed by miR-145/inhibitors ( Fig. 6B and Fig. S1B). Furthermore, western blotting assays indicated that the effect of linc-ROR overexpression on EMT phenotypes of HCC cells could be partially rescued by miR-145/mimics, and re-expression of miR-145/ inhibitor could induce EMT in HCCLM3/shROR (or MHCC97-H/shROR) cells by downregulating epithelial markers and upregulating mesenchymal markers ( Fig. 6C and Fig. S1C). Meanwhile, immunofluorescence analysis showed that downregulated epithelial markers and upregulated mesenchymal markers induced by linc-ROR overexpression in HepG2 (or SMMC-7721) cells was partially reversed by miR-145/mimics, and vice versa in the transfected HCCLM3 (or MHCC97-H) cells ( Fig. 6D and Fig. S1D). These results showed that linc-ROR influences EMT, invasion and metastasis in HCC cells in vitro, at least partially in a miR-145-dependent manner.
ROR and reverse-transcribed miR-145 cDNA were used as standard templates to formulate standard curves with limit dilution approaches, and then the exact copies of linc-ROR and miR-145 per cell were calculated according to their molecular weight and cell counts. Each experiment was performed at least three times. N.S, not significant. *p < 0.05; **p < 0.01.
Linc-ROR positively regulates the miR-145 target gene ZEB2 in HCC cells. It has been reported that miR-145 represses EMT, tumor migration and invasion by directly targeting the 3′-UTRs of ZEB2 in  tumors 25 . Besides, ZEB2 is reported having the ability to stimulate the initiation of EMT by binding to E-boxes on the E-cadherin promoter, thus repressing its promoter activity 26 . We proposed that linc-ROR and ZEB2 interact with miR-145 by functioning as competing endogenous RNAs (ceRNAs). Here, to confirm such model, we measured the level of ZEB2 in response to different levels of linc-ROR. We observed that in HepG2 and SMMC-7721 cells, the ectopic expression of linc-ROR upregulated ZEB2 at the transcript and protein levels. Co-transfection of linc-ROR overexpression plasmid and miR-145/mimics could significantly reduce ZEB2 expression level (Fig. 7A). In HCCLM3 and MHCC97-H cells, shROR decreased ZEB2 at the transcript and protein levels.
Similarly, miR-145/inhibitors restored the decrease of ZEB2 expression in the HCC cells transfected with shROR (Fig. 7B). Furthermore, we demonstrated that miR-145 expression level in HCC tissues was significantly lower than those in the adjacent normal tissues. Additionally, by linear regression analysis, miR-145 level and linc-ROR level were negatively correlated (Fig. 7C). Meanwhile, the expression of ZEB2 in HCC and corresponding nontumor tissues was also measured by qRT-PCR. It was showed that ZEB2 had high expression in HCC tissues compared with those in the corresponding normal tissues. Besides, ZEB2 and linc-ROR expression levels were positively correlated (Fig. 7D). Collectively, these findings indicated that there is a regulatory signaling pathway in which linc-ROR regulates ZEB2 by competitively sponging up miR-145, inducing increased invasion, metastasis and an EMT phenotype in HCC cells (Fig. 7E).

Discussion
In spite of the encouraging advances in therapies for HCC patients, the invasive potential and distant metastasis make the prognosis remain dismal 27 . Hence, it is of great importance to explain the underlying mechanisms on the development and progression of HCC metastasis. Emerging evidence has demonstrated a key regulatory role of lncRNAs in tumorigenesis and metastasis of multiple cancers, including HCC 4 . For example, HBx-induced lncRNA Unigene561591 has an oncogenic effect on cell migration/invasion and EMT by acting as a ceRNA of miR-140-5p in HCC cells 1 . Another lncRNA, DANCR, could increase stemness features of HCC by derepression of CTNNB1 28 . Linc-ROR was first described in iPSCs and can be regulated by the key pluripotency factors Oct4, Sox2, and Nanog 7 . Most recently, scientists pointed out that linc-ROR could be regarded as a novel biomarker, indicating a poor prognosis in gallbladder cancer and pancreatic cancer patients 13,21 . However, the specific role of linc-ROR in mediating the metastatic process of HCC is not well studied.
In the present study, we first demonstrated that linc-ROR was upregulated in HCC tissues compared with pair-matched noncancerous tissues. Patients with high linc-ROR expression levels tended to have more advanced TNM stage and higher incidence of lymph node metastasis or vascular invasion and tumor recurrence. Moreover, HCC patients with high linc-ROR expression levels had a poorer prognosis than those with low linc-ROR expression. Then, we comprehensively investigated the function of linc-ROR in HCC metastasis by employing gain-of and loss-of-function approaches. Indeed, our data showed that linc-ROR upregulation significantly promoted migration and invasion of HCC cells. Besides, orthotopic implantation tumor models provided additional support for the involvement of linc-ROR in promoting hepatic and lung metastasis in vivo.
The acquisition of invasive capabilities, including degradation of the cell matrix and turnover of cell-cell junctions, is one of the most important steps in the metastatic cascade of tumors 29 . EMT is a biological process where epithelial cells lose their polarity and undergo transition into a mesenchymal phenotype 30 . Accumulating evidence indicates that lncRNAs can trigger EMT to increase cancer migration and invasion. Therefore, we proposed that whether the EMT phenotype of HCC cells was affected by linc-ROR expression. Western blotting and immunofluorescence assays revealed that linc-ROR upregulation could exert obvious effects on EMT phenotypes of HCC cells by decreasing epithelial markers and increasing mesenchymal markers. In contrast, linc-ROR downregulation could cause opposite changes of those markers. These data demonstrated that linc-ROR functions as a metastasis inducer in HCC cells through increasing invasion and promoting EMT.
Growing number of research confirmed that lncRNAs can antagonize miRNA function by competing with miRNAs for binding to shared target mRNAs, and then to silence target mRNAs 31 . For example, Jia el al found that lncRNA H19 acts as a sponge for miR-29a to modulate VASH2 expression in glioma-associated endothelial cells 32 . LncRNA MEG3 enhanced cell proliferation by competitively binding miR-29 in HCC 33 . To clarify whether linc-ROR functions as a sponge for miRNA, bioinformatics was used to analyze the potential miRNA binding sites contained in linc-ROR. Besides, based on previously studies and function analysis of linc-ROR, we chose miR-145 for further study. qRT-PCR results showed that miR-145 expression level was downregulated by linc-ROR in HCC cells. Also, we found that there was no difference in linc-ROR levels after ectopic expression or knockdown of miR-145. Fluorescent reporter assay and RIP analysis showed that linc-ROR was a direct target of miR-145 and could act as a sponge for miR-145.
Previous reports have proved that miR-145 possesses a tumor suppressive role in many cancers, including lung cancer, pancreatic cancer, breast cancer, HCC and colorectal cancer [34][35][36][37][38] . Moreover, miR-145 widely participates in cell proliferation, apoptosis, migration and invasion [39][40][41][42] . In the present study, our findings showed that miR-145 suppression yielded very similar effects to ectopic linc-ROR expression in HCC cells. Besides, rescue experiments showed that linc-ROR influences cell invasion, metastasis and EMT in HCC cells in vitro, partially in a miR-145-dependent manner. Taken together, these data strongly suggested that miR-145 is a bona fide linc-ROR-targeting miRNA. MiR-145 has been reported to inhibit EMT and tumor metastasis by directly targeting 3′-UTRs of ZEB2, which is a key transcription factor of EMT 25 . We considered whether linc-ROR and ZEB2 interact with miR-145 by functioning as ceRNAs. Our study demonstrated that miR-145 could reverse the reduction of ZEB2 caused by linc-ROR knockdown. Besides, linc-ROR expression was negatively correlated with miR-145, but positively associated with ZEB2 in HCC tissues. These results revealed that linc-ROR could function as a ceRNA to regulate the expression of ZEB2 in a miR-145-dependent manner.
In conclusion, we discovered that linc-ROR was highly expressed in HCC tissues and played a key role in regulating HCC metastasis. Linc-ROR competitively binds to miR-145, and subsequently up-regulates the expression  Table 1. The data does not contain any information that can identify patients. All patients were accepted a close follow-up observation for disease recurrence at 1 month intervals during the first 2 postoperative years, and every 3 months after that. Informed consent was obtained from patients, and ethical approval was granted by the Review Board of Hospital Ethics Committee (Jingling Hospital, Nanjing University). All experiments were performed in accordance with the approved guidelines and regulations, including any relevant details.
Cell lines and mice. The cell lines HH, HCCLM3, MHCC97-H, HepG2 and SMMC-7721 were cultured in Dulbecco's modified Eagle's medium (GIBCO-BRL) supplemented with 10% fetal bovine serum (FBS), streptomycin (100 μg/ml), penicillin (100 U/ml). All cells were fostered at 37 °C in an atmosphere containing 5% CO 2 . We purchased all female BALB/C nude mice (4-6 weeks old) from the Animal Core Facility of Nanjing Medical University and housed these mice in laminar flow cabinets under specific pathogen-free conditions. All animal experiments were performed with the approval of the Institutional Committee for Animal Research. The study protocol was also approved by the Committee on the Use of Live Animals in research, Nanjing University (Nanjing, China).

Cell transfection. The linc-ROR overexpression plasmid was a gift from Dr Wang Yue (College of Basic
Medicine, Second Military Medical University, China). The shROR expression plasmid was a gift from Dr Hou P (The Institute of Genetics and Cytology, Northeast Normal University, Changchun, China). Hsa-miR-145 mimic/ negative control mimics and hsa-miR-145 inhibitor/negative control inhibitor were purchased from Applied Biological Materials (ABM, Canada). The cell transfection was performed using Lipofectamine 2000 (Invitrogen, CA, USA) following the manufacturer's instructions. HepG2 or SMMC-7721 cells were transfected with the plasmid linc-ROR to constructed cell lines stably expressing linc-ROR and were screened with Puromycin (2.0 μg/ml) for four weeks. HCCLM3 or MHCC97-H cell lines stably suppressing linc-ROR were constructed by transfection with a lentivirus construct containing the desired vector and screened with Puromycin (2.0 μg/ml) for four weeks.
Wound healing assay. Cell migration capacity was measured using a wound healing assay. Briefly, cells were planted in 6-well plates and cultured to confluence. Once cells were attached completely, they were scraped by a plastic scriber to form a wound in the middle of the plates. The cells were washed, and the medium was replaced with serum-free medium. After incubation for 48 h, we observed the cultures under a microscope. A minimum of five randomly chosen areas was measured and the distance of cell migration to the wound area was determined.
Cell migration and invasion assays. Cell migration and invasion were measured by transwell chamber (8.0 μm pore size, Corning) and Matrigel invasion (Becton Dickinson) according to the manufacturer's instructions. In short, the upper chamber was filled with cells in serum-free media, coating with or without 10 μg/ml Matrigel. Media containing 10% FBS was added to the lower chamber. After 48 h of incubation, cells on the lower chamber membrane were fixed with 4% polyoxymethylene and stained with 0.1% crystal violet. Five predetermined fields were counted under a microscope.

RNA immunoprecipitation (RIP).
RNA immunoprecipitation was performed using thermo fisher RIP kit (Thermo, USA) according to the manufacturer's protocol. The Ago2 antibodies were purchased from Abcam (USA). Normal mouse IgG (Abcam, USA) was applied as the negative control. Purified RNA was subjected to qRT-PCR analysis to demonstrate the presence of the binding targets using respective primers.
Western blotting assay. Total protein lysates were separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and were transferred to polyvinylidene difluoride membranes (Roche). Antibody dilutions of 1:2000 were used for rabbit anti-human E-cadherin, β-catenin, N-cadherin, Vimentin, ZEB2, and 1:3000 for GAPDH. The proteins were detected using an enhanced chemiluminescence system and exposed to x-ray firm. All antibodies were purchased from Abcam (USA).
Immunofluorescence. Immunofluorescence staining assay was performed following the standard protocol. In brief, the transfected cells were fixed with 4% paraformaldehyde and permeabilized in 0.1% Triton X-100. 5% BSA was used to blocked cells for 1 h at room temperature. Finally, cells were incubated with primary antibody at 4 °C overnight, followed by incubation with florescent-dye conjugated secondary antibody (Invitrogen) for 1 h, and then stained with DAPI. Images were taken with a Zeiss photomicroscope (Carl Zeiss, Oberkochen, Germany).
In vivo tumor metastasis assay. For in vivo metastasis study, HCCLM3 cells transfected with shROR (or shCtrl) or HepG2 cells transfected with linc-ROR (Vector) were suspended in 100 μl PBS and injected subcutaneously into the right side of the posterior flank of female BALB/C athymic nude mice at 4 to 6 weeks of age. After 4 weeks, the subcutaneous tumors were resected and diced into 1.0 mm 3 cubes, which were then implanted into the left lobes of the livers of the nude mice (10/group). After 8 weeks, the animals were sacrificed for intrahepatic and lung metastases assessment. All animal studies were conducted according to protocols that were approved by the Animal Care and Use Committee of Jinling Hospital (Nanjing University, Nanjing, China).
Statistical analysis. Experimental data were presented as means ± standard deviation (SD). The SPSS 17.0 software (SPSS Inc., Chicago, IL, USA) was used for all statistical analyses. Quantitative variables for nonparametric analyses were performed using Wilcoxon text for paired and Mann-Whitney U test for unpaired analyses. The disease-free survival (DFS) and over survival (OS) curves were plotted using the Kaplan-Meier method. The correlation was evaluated by Spearman's rank correlation coefficients. Differences were considered significant if p < 0.05. Corresponding significance levels are indicated in the figures.