Circular RNA circRHOT1 promotes hepatocellular carcinoma progression by initiation of NR2F6 expression

We obtained 100 paired HCC tissues and adjacent normal tissues from the Affiliated Hospital of Guilin Medical University. The relationship between circRHOT1 expression and clinical severity is listed in Additional file 1: Table S1. Samples receiving chemotherapy or radiotherapy before collection were excluded. Written consent approving the usage of HCC tissues in our study were obtained from each patient. This study was approved by the ethnic committee of the Affiliated Hospital of Guilin Medical University.

For primary HCC cell culture, the surgical-isolated HCC tissues were digested with collagenase IV and hyaluronidase for 50 min at 37 °C. Afterwards, the cell suspensions were washed, filtered using a 70 μm cell strainer (BD, USA) and seeded on collagen-coated Petri dishes. Fibroblasts and blood vessels were discarded. The primary cancer cells were cultured in DMEM medium supplemented with 20% FBS, 2 mM glutamine, 1 mM pyruvate, 10 mM HEPES, 100 units/mL penicillin/streptomycin, 0.1 mg/mL gentamicin, and 2 g/L Fungizone as previously reported [13]. The primary cells at passage 4~10 were used for experiments.

circRHOT1, TIP60 or NR2F6 knockout and NR2F6 promoter deletion in HCC cells were established using the CRISPR/Cas9 system as previously described [14, 15]. Briefly, we designed single guide RNAs (sgRNA) using an online tool (http://​crispr.​mit.​edu/​) and cloned them into the lenti-Cas9-eGFP vector (Addgene, 63592). Next, 293 T cells were transduced with the lenti-Cas9-eGFP-sgRNA as well as pVSVg (Addgene, 8454), and psPAX2 (Addgene, 12260) to produce lentivirus. The viruses were concentrated with PEG5000 (Sigma Aldrich) and were used to infect HCC cells. For TIP60 or NR2F6 deletion, GFP⁺ cells were isolated by FACS, and the knockout efficiency was verified by western blotting. For circRHOT1 and NR2F6 promoter deletion, two sgRNAs flanking the target sequence were used at the same time. Next, GFP⁺ cells were isolated by FACS, followed by monoclonalization. Their deletion was verified by DNA sequencing. The sgRNA sequences were listed in Additional file 1: Table S2.

Anti-GAPDH (5174), anti-Cyclin D1 (2978) anti-TIP60 (12058), anti-P21 (2947), and anti-PCNA (13110) were from Cell Signaling Technology. Anti-Flag (F1804) was purchased from Sigma-Aldrich. Anti-NR2F6 (ab137496) was obtained from Abcam.

For in vivo tumor growth, 1 × 10⁶ HCC cells were subcutaneously injected into the flanks of 5-week-old nude mice from HFK Biosciences (n = 6 for each group). The tumor volumes and weights were measured at the indicated time points. Animal assays were approved by the ethnic committee of the Affiliated Hospital of Guilin Medical University.

Cell proliferation was measured using CCK-8 assays as previously reported [16]. Briefly, 1 × 10³ cells were seeded in 96-well plates and were cultured for 24 h, 48 h, 72 h and 96 h, followed by incubation with 10 μl of CCK-8 assay solution in each well for 2 h. The absorbance values at 450 nm were then measured using an enzyme immunoassay analyzer (Thermo Fisher Scientific, Inc., Waltham, MA, USA). This assay was repeated three times.

Cell migration and invasion assays have been described previously [17]. Briefly, 2 × 10³ HCC cells were planted in the upper chamber (Mitrogel (BD) pre-coated for invasion) with 200 μl serum-free medium. The lower chamber was added with complete medium. Forty-eight hours later, cells in the lower chamber were fixed and stained with crystal violet (0.2%). Cell numbers were determined using a microscope. This assay was repeated three times.

TRIzol reagent was used to extract RNA from HCC tissues or cell lines according to the manufacturer’s protocol. Then 1 μg RNA was used to synthesize cDNA, followed by analysis of gene expression on ABI 7300 qPCR system. Relative expression was normalized to ACTB or U6. The analysis of circRHOT1 was carried out using convergent primers (Forward, 5′-ATCACCATTCCAGCTGATGT-3′ and reverse, 5′-TGCTGTCTTTGTCTGTTCTTTC-3′) and divergent primers (Forward, 5′-AGTCGATGGATTCCTCTCAT-3′ and Reverse, 5′-AAGTTGTTCATCACTCTGTT-3′). The sequences of other primer are listed in Additional file 1: Table S3.

ChIP assays were conducted as previously described [15]. Briefly, HCC cells were fixed with 1% formaldehyde at 37 °C for 10 min and lysed. The genome was sonicated to ~ 500 bp fragments. Anti-TIP60 were added into the lysates and incubated overnight at 4 °C, followed by incubation with Protein A Agarose/Salmon Sperm DNA (50% Slurry) beads for 4 h. Finally, precipitants were eluted and analyzed by qPCR. Primer sequences were listed in Additional file 1: Table S4.

For RNA EMSAs, the biotin-labeled circRHOT1 RNA probe was obtained by in vitro transcription using the Biotin RNA Labeling Mix (Roche). The LightShift Chemiluminescent RNA EMSA Kit (Thermo Scientific) was used for the shift assay according to the manufacturers’ instructions. This assay was repeated three times.

CircRHOT1 expression in HCC tissues was measured using biotin-labeled circRHOT1 probes. Paraffinized sections were deparaffinized with xylene and 100% ethanol. The sections were then incubated with biotin-labeled probes for 18 h at 40 °C. The DAB substrate was used for the colorimetric detection of circRHOT1. Finally, the sections were co-stained with hematoxylin, followed by dehydration in graded alcohols and xylene. Biotin-conjugated probes were purchased from Invitrogen. The CircRHOT1 probe sequence was as follow: 5′-GGAGGAACCTGCTGTCTTTGTCTGTTC-3′. This assay was repeated three times.

Hybridization was performed overnight with circRHOT1 probes. Specimens were analyzed using a Nikon inverted fluorescence microscope. The circRHOT1 probe for fluorescence in situ hybridization (FISH) is 5′-GGAGGAACCTGCTGTCTTTG-3′. This assay was repeated three times.

RNAs were isolated from HCC samples using TRIZOL (Invitrogen). CircRHOT1, linear RHOT1 and 18S probes for northern blotting were achieved using the Biotin RNA labeling mix (Roche). The RNA samples were separated by electrophoresis and were transferred to NC membranes, which were then incubated with the hydration buffer containing the probes. Finally, the RNA signal was detected using the Chemiluminescent Nucleic Acid Detection Module (Thermo Scientific). This assay was repeated three times.

RNA pulldown and mass spectrometry were performed as described before [15]. This assay was repeated three times.

This assay was performed according to a previous study [18]. In brief, the sequence of circRHOT1 flanked by two complementary regions (left and right) or only by the left complementary region was inserted into overexpressing vector. Then plasmids were transfected into circRHOT1-depleted 293 T cells and the level of circRHOT1 was measured by northern blotting.

The normalized circRNA expression values in two online-available datasets (GSE94508 and GSE97332) were downloaded from GEO database. Then we first searched differentially expressed circRNAs based on the dataset GSE94508. We selected upregulated circRNAs in tumor tissues according to the dataset GSE94508. Then, we analyzed whether these selected circRNAs were still upregulated in the dataset of GSE97332. Finally, circRNAs that were upregulated in these two datasets were determined. Thus, the most differentially expressed circRNAs between HCC tissues and normal tissues were screened out. Afterwards, we analyzed whether these circRNAs were conserved between human and mouse according to Yang’s cohort, which is directly available in their published article [19]. So circRHOT1 was eventually selected. The heatmaps were determined according to the normalized expression levels of circRNAs.

WT and circRHOT1-deficient HCC cells were lysed with TRIZOL reagent, and total RNA was extracted according to the standard method. Next, RNA sequencing was performed (Shenzhen, BGI Genomics Co., Ltd.), and target genes were analyzed.

To identify essential circRNAs expressed in HCC tissues, we analyzed online-available datasets concerning the circRNA profiles in HCC tissues (GSE94508 and GSE97332) [7, 20]. We found that many circRNAs are overexpressed in HCC tissues compared to peritumor tissues (Fig. 1a). Among them, circRHOT1 is one of the most upregulated and conserved circRNAs according to Yang’s cohort [19]. Firstly, we validated its features as a circRNA by PCR (Additional file 1: Figure S1a) and RNA sequencing (Additional file 1: Figure S1b). Then, the levels of circRHOT1 in 100 paired HCC tissues and peritumor tissues were determined. Results confirmed that circRHOT1 was significantly upregulated in tumor tissues (Fig. 1b). Next, we chose 5 pairs of HCC sample tissues to subject to northern blotting and validated circRHOT1 was markedly overexpressed in tumor tissue (Fig. 1c). Notably, the linear RHOT1 mRNA level was only slightly elevated as determined by online available datasets and qRT-PCR analysis (Additional file 1: Figure S1c). However, the change on RHOT1 protein level was not obvious in HCC tissues (Additional file 1: Figure S1d). In situ hybridization (ISH) and fluorescence in situ hybridization (FISH) using a biotin-labeled specific probe were further carried out to measure circRHOT1 expression. Results indicated that more circRHOT1 was present in HCC tissues (Fig. 1d, e). In addition, circRHOT1 level was increased in HCC cell lines (Additional file 1: Figure S1e). To investigate the relationship between circRHOT1 level and clinical severity, we checked circRHOT1 expression in HCC tissues with different stages by qRT-PCR. As shown in Fig. 1f, circRHOT1 was expressed more highly in stage III tissues than in stage I/II samples. Moreover, the expression of circRHOT1 in advanced HCC (aHCC) tissues was higher than that in early HCC (eHCC) tissues (Fig. 1g). Next, we investigated the correlation of circRHOT1 expression with the prognosis of HCC patients. We found that HCC patients with higher circRHOT1 expression showed a poorer prognosis by Kaplan–Meier survival analysis (Fig. 1h, i). Taken together, our data suggest that circRHOT1 is highly expressed in HCC tissues and correlated with a poor prognosis.

We next explored the role of circRHOT1 in HCC progression. A previous study showed that the flanked complementary elements are essential for the formation of circular RNAs [21]. We identified three complementary elements flanking circRHOT1 (Additional file 1: Figure S2a), and only the deletion of the complementary region #3 in the right flank of circRHOT1 exon 6 abrogated the expression of circRHOT1 as determined by minigene assay (Additional file 1: Figure S2a). Thus, we constructed circRHOT1-deficent HCC cell lines through deleting the region #3 in the right flank of circRHOT1 by CRISPR/Cas9 technology. We confirmed the efficiency of circRHOT1 deletion by PCR (Additional file 1: Figure S2b), Northern blotting (Additional file 1: Figure S2c) and qRT-PCR (Fig. 2a). Notably, circRHOT1 deletion did not affect the expression of RHOT1 mRNA and protein (Additional file 1: Figure S2c, d). Then, we performed CCK-8, colony formation and EdU incorporation assays to assess the effect of circRHOT1 on cell proliferation. As shown, circRHOT1 knockout significantly inhibited cell proliferation, colony formation and EdU incorporation (Fig. 2b-d). Next, we conducted Transwell assays and found that circRHOT1 knockout significantly suppressed the migration and invasion of HCC cells (Fig. 2e, f). Moreover, circRHOT1 deletion dramatically promoted the apoptotic cell percentage (Fig. 2g), which is further confirmed by the detection of Caspase 3/7 Activity (Fig. 2h). To further validate the effects of circRHOT1 on HCC cells, we silenced its expression in HCC cell lines, including Hep3B and Huh7 cells (Additional file 1: Figure S2e). Functional experiments from CCK8, colony formation, EdU incorporation, Transwell and FACS results also demonstrated that circRHOT1 knockdown suppressed proliferation, migration and invasion but promoted apoptosis in HCC cells (Additional file 1: Figure S2f-k). Additionally, we conducted rescue assay by re-expressing circRHOT1 in circRHOT1-deleted HCC cells (Additional file 1: Figure S2l). A series of experiments indicated that restoration of circRHOT1 rescued the effects of circRHOT1 knockout on HCC cell proliferation, migration, invasion and survival (Additional file 1: Figure S2m-q), indicating the essential role of circRHOT1 in HCC.

To evaluate the function of circRHOT1 in vivo, we performed xenograft experiments with WT and circRHOT1-deleted HCC cells. As shown in Fig. 2i, circRHOT1 knockout significantly reduced the tumor weight. Furthermore, xenograft experiments with circRHOT1 low- or high-expressed HCC cells showed that circRHOT1 overexpression promoted tumor growth in vivo (Fig. 2j). Collectively, the above data convincingly demonstrated that circRHOT1 promotes HCC development and progression.

To further investigate the underlying molecular mechanism, we analyzed the differentially expressed genes after circRHOT1 deletion using RNA sequencing. Many genes were downregulated, among which NR2F6 is the most downregulated and possesses high expression value in HCC cells (Fig. 3a). We chose seven downregulated genes based on fold-change and expression level in WT HCC cells to verify the RNA sequencing results by qRT-PCR (Fig. 3b). Notably, these seven genes were also downregulated in Hep3B and Huh7 cells after circRHOT1 silence (Additional file 1: Figure S3a). Western blotting indicated NR2F6 protein level was significantly decreased in circRHOT1-deleted HCC cells (Fig. 3c) and circRHOT1-silenced HCC cell lines (Additional file 1: Figure S3b). Our above data showed that circRHOT1 is mainly distributed in the nucleus (Fig. 1d, e). To determine whether circRHOT1 regulates NR2F6 transcription, we performed chromatin isolation by RNA purification (CHIRP) with specific biotin-labeled circRHOT1 probe and found that circRHOT1 could be enriched on the region of − 1000~ − 800 bp from the transcription start site (TSS) of NR2F6 promoter (Fig. 3d). Moreover, the enrichment of the histone active modification H3K27Ac was reduced after circRHOT1 knockout while that of H3K27me3 was increased (Fig. 3e, f). Consistently, RNA pol II could not bind to the NR2F6 promoter after circRHOT1 deletion (Fig. 3g). Moreover, the promoter of NR2F6 in circRHOT1-deleted HCC cells was more resistant to DNaseI digestion (Fig. 3h), suggesting that circRHOT1 is indispensable for accessibility of the NR2F6 promoter. To further demonstrate the role of the circRHOT1 binding region on NR2F6 promoter, we deleted this region in HCC cells. CircRHOT1 could not enrich on the NR2F6 promoter after this region deletion (Fig. 3i). Additionally, we found that the overexpression of circRHOT1 promoted the mRNA level of NR2F6, while deletion of the binding region abrogated this effect (Fig. 3j). Thus, circRHOT1 was associated with the NR2F6 promoter and promoted its expression.

To investigate whether circRHOT1 cooperated with specific proteins to regulate NR2F6 expression, we performed RNA pulldown using HCC cell lysates, followed by silver staining and mass spectrum identification. We identified TIP60 as a potential interactive protein of circRHOT1 (Fig. 4a). Furthermore, RNA pulldown with TIP60-overexpressing HCC cells showed that biotin-labeled circRHOT1 precipitated Flag-TIP60 (Fig. 4b). Moreover, biotin-labeled circRHOT1 probe could precipitate endogenous TIP60 in WT HCC cell lysates (Fig. 4c). Besides, TIP60-specific antibodies enriched endogenous circRHOT1 in HCC cells (Fig. 4d). According to FISH assay, circRHOT1 was co-localized with TIP60 in HCC cells (Fig. 4e). RNA-EMSA assay also demonstrated that TIP60 directly interacts with circRHOT1 (Fig. 4f).

To further determine whether TIP60 participates in the regulation of NR2F6 expression, we performed ChIP assays with HCC cells. We found that TIP60 was enriched in the same region of the NR2F6 promoter as circRHOT1 (Fig. 5a). Furthermore, TIP60 knockout also inhibited the enrichment of H3K27Ac and RNA pol II on the NR2F6 promoter (Fig. 5b, c). In addition, TIP60 deletion downregulated the mRNA and protein levels of NR2F6 in HCC cells (Fig. 5d, e). Next, we wanted to determine the role of circRHOT1 in TIP60-mediated expression of NR2F6. We found that circRHOT1 knockout abrogated the enrichment of TIP60 on the promoter of NR2F6 (Fig. 5f). DNA FISH also indicated that circRHOT1 knockout abolished the co-localization of TIP60 with the NR2F6 promoter (Fig. 5g). Importantly, circRHOT1 deficiency lost the recruitment of the NuA4 complex to the NR2F6 promoter, whereas the NuA4 complex bound to the NR2F6 promoter in control HCC cells (Fig. 5h). Moreover, double knockout of circRHOT1 and TIP60 further inhibited the mRNA level of NR2F6 in HCC cells (Fig. 5i). We then analyzed the relationships among the expression levels of circRHOT1, TIP60 and NR2F6 in HCC tissues. We found that the protein and mRNA levels of NR2F6 were positively correlated with those of TIP60 in HCC tissues by qRT-PCR and according to the GSE45436 dataset (Fig. 5j, k). Additionally, there was a positive correlation between the expression of NR2F6 and circRHOT1 in HCC tissues (Fig. 5k and Additional file 1: Figure S3c) while no significant expression correlation between circRHOT1 and TIP60 existed (Additional file 1: Figure S3c). Altogether, circRHOT1 is essential for TIP60-mediated NR2F6 expression in HCC tissues.

To further explore the role of NR2F6, we examined its expression in HCC tissues. According to Park’s cohort (GSE36376) [22] and Zhang’s cohort (GSE25097) [23‐25], NR2F6 was upregulated in HCC tissues compared with that in adjacent normal tissues (Fig. 6a, b). Furthermore, we examined the expression of NR2F6 in 100 paired HCC samples and found that NR2F6 was upregulated in tumor tissues (Fig. 6c). Immunohistochemical analysis and western blotting confirmed that NR2F6 level was elevated in HCC tissues (Fig. 6d, e). Moreover, higher expression of NR2F6 in HCC patients suggested a poorer prognosis (Fig. 6f). We then knocked out NR2F6 or restored its protein level in circRHOT1-deleted HCC cells (Fig. 6g). CCK-8 and Transwell assays revealed that knockout of either circRHOT1 or NR2F6 inhibited cell proliferation, migration and invasion, while restoration of NR2F6 in circRHOT1-deleted HCC cells rescued cell proliferation, migration and invasion in vitro (Fig. 6h-j). In vivo xenograft experiments showed that knockout of circRHOT1 or NR2F6 suppressed tumor growth. However, restoration of NR2F6 rescued circRHOT1-deletion-mediated inhibition on tumor growth (Fig. 6k). Thus, our results demonstrated that circRHOT1 inhibits HCC progression via the activation of NR2F6.