Hypoxia-induced lncRNA MRVI1-AS1 accelerates hepatocellular carcinoma progression by recruiting RNA-binding protein CELF2 to stabilize SKA1 mRNA

HCC tissue samples and adjacent non-tumor tissue samples, which were histopathologically confirmed, were collected from 72 patients who underwent surgery in the First Affiliated Hospital of Xi’an Jiaotong University from Jan. 2012 to Jan. 2014. All of the patients did not receive chemotherapy or radiotherapy before surgery. All of the samples were stored at −80℃. Our study got approval from the Ethics Committees of the First Affiliated Hospital of Xi’an Jiaotong University and written informed consent was obtained from all patients. The clinical parameters of HCC patients were shown in Table 1.

The human normal liver cell line (LO2) and five HCC cell lines (Hep3B, Huh7, SK-HEP-1, HepG2, and MHCC-97H) were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). All of the cells were maintained in an incubator (37℃, 5% CO₂) and cultured in DMEM (Gibco, Grand Island, NY, USA) supplemented with 10% FBS (Gibco, Grand Island, NY, USA) and 1% penicillin-streptomycin (Invitrogen, CA, USA). All cell lines that we used in this study were tested and authenticated by DNA sequencing using the AmpF/STR method (Applied Biosystems) and tested for the absence of mycoplasma contamination (MycoAlert) and the latest date tested is 30 October 2022.

The full-length cDNA of MRVI1-AS1 was cloned into the pcDNA3.1 vector (GenePharma, Shanghai, China) to construct the MRVI1-AS1-overexpressed plasmid, and shRNA that specifically targeted HIF-1α, MRVI1-AS1, or SKA1 was cloned into the pLKO.1 vector (GenePharma). For lentiviral vector transduction, cells were seeded onto plate wells and infected with a l entiviral construct containing different vectors supplemented with 5 mg/ml polybrene (Gene-Pharma Co., Suzhou, China). Then the cells were selected with 5 mg/ml puromycin to create stable cell subclones, and all of the experimental operations were based on the product specifications.

TRIzol reagent (Invitrogen, Carlsbad, CA) was used to isolate total RNA from tissue samples and cell lines based on the product manual. Then cDNA was obtained after reverse transcription. RT-qPCR was performed with SYBR Green Master Mixture (Takara, Dalian, China). GAPDH was used as the control. Relative gene expression levels were calculated using the 2^−ΔΔCt method. Primers for MRVI1-AS1: Forward: 5’-GCCCTGGTATTCCTTGAACA-3’, Reverse: 5’-TCAGTCCAGGAAGAGGT-3’. Primers for SKA1: Forward: 5’-CCTGAACCCGTAAAGAAGCCT-3’, Reverse: 5’-TCATGTACGAAGGAACACCATTG-3’. Primers for GAPDH: Forward: 5’-GGAGCGAGATCCCTCCAAAAT-3’, Reverse: 5’-GGCTGTTGTCATACTTCTCATGG-3’.

After being transfected with plasmids for 48 h, the cells were seeded into transwell chambers (8 µm pore size, Corning, USA) containing 200 µl medium with 1% FBS. The lower chambers were added with 800 µl medium containing 10% FBS. For detection of invasion ability, transwell chambers were pre-coated with Matrigel. Twenty-four hours later, cells passed through the membrane were stained with crystal violet (0.1%) and counted.

Transfected cells were seeded into 6-well plates to form cell monolayers. When cell confluency reached to 80%, a 200-µl tip was used to scratch the cell layers. After being gently washed, cells were cultured with serum-free medium for 24 h. A microscope (IX71, Olympus, Tokyo, Japan) was used to image (magnification: 200×) the wounded gaps at 0 and 24 h after being created.

For MTT assay, transfected cells were plated into 96-well plates (2000 cells/well). Then at 0, 24, 48, and 72 h after seeding, MTT (10 µL/well, Sigma, USA) was added to each well and incubated for 4 h at 37℃. Then, DMSO (100 µL/well, Sigma, USA) was used to dissolve the crystals. Absorbance was measured at 490 nm by a microplate reader (Bio-Rad, Richmond, CA). For EdU assay, Cell-Light™ EdU Apollo®567 In Vitro Imaging Kit (RiboBio Co., Ltd., Guangzhou, China) was used. Briefly, transfected HCC cells (1 × 10⁠⁵) were cultured in 96-well plates. Cells were incubated with EdU labeling medium at a moderate concentration for 2 h. Then, the cells were fixed with 4% paraformaldehyde, glycine, and 0.5% TritonX-100 in PBS. Next, cells were stained with 100 µL Apollo dye solution for 30 min at room temperature. The cells were subsequently stained using Hoechst and incubated for 30 min. The photos were taken on a microscope. The percentage of EdU-positive cells was calculated using ImageJ software.

To detect the effects of MRVI1-AS1 on luciferase activity of SKA1 promoter, full-length SKA1 promoter was cloned into pGL3 plasmid (pGL3-SKA1). pGL3 or pGL3-SKA1 with pRL-TK was transfected into MRVI1-AS1 overexpressing or MRVI1-AS1 knockdown HCC cells. After 48 h, the luciferase activities were measured using a dual-luciferase reporter gene assay system (Promega). The relative ratio of firefly luciferase activity to Renilla luciferase activity was measured.

The separation of nuclear and cytosolic fractions was performed using the PARIS Kit (Life Technologies, Carlsbad, CA) according to the manufacturer’s instructions. Then, the subcellular localization of MRVI1-AS1 was detected by RT-qPCR. The GAPDH and U6 transcripts were used as an internal reference of cytoplasmic and nuclear RNA, respectively.

RNA pull-down assay was performed using RNA-Protein Pull-Down Kit (Thermo Scientific) according to the manufacturer’s instructions. Briefly, biotin-labeled RNAs were in vitro transcribed, treated with RNase-free DNase I, and purified. Cell lysates were prepared using lysis buffer. Then, 1 mg cell lysates were mixed with 50 pmol of biotin-labeled RNAs. The washed streptavidin agarose beads were added to each binding reaction and further incubated at room temperature for 1 h. Beads were washed and boiled in sodium dodecyl sulfate buffer. The MRVI1-AS1-pull-down or antisense-MRVI1-AS1-pull-down protein samples were subjected to western blot with CELF2 antibody CELF2 (#NBP2-16035, Novus, USA). The antisense RNA of MRVI1-AS1 was taken as a negative control in RNA pull-down assay.

RIP assay was performed using the EZ-Magna RIP kit (Millipore, Billerica, MA) following the manufacturer’s protocol. HCC cells at 70–80% confluence were scraped off and then lysed in complete RIP lysis buffer. A total of 100 µl of whole cell extract was incubated with RIP buffer containing magnetic beads conjugated with antibodies against CELF2 (#ab156877, Abcam, USA) or control IgG (#ab172730, Abcam, USA) for 6 h at 4°C. The beads were then washed with washing buffer, and the complexes were incubated with 0.1% SDS/0.5 mg/ml Proteinase K (30 min at 55°C) to remove proteins. The immunoprecipitated RNAs were then extracted, and the RNA concentration and quality were determined by NanoDrop spectrophotometer (Thermo Scientific). Finally, immunoprecipitated RNA was analyzed by RT-qPCR.

Total proteins were isolated from cells with RIPA buffer (Beyotime, Hangzhou, China). Ten percent SDS-PAGE gels separated protein, then transferred to PVDF membranes (Millipore, Billerica, MA, USA). After being blocked by 5% nonfat milk for 2 h, antibodies for HIF-1α (1:1000, # ab228649, Abcam, USA), SKA1 (1:1000, #ab91550, Abcam, USA), CELF2 (1:1000, #NBP2-16035, Novus, USA),and β-actin (1:1000, # ab8226, Abcam, USA) were used to incubate membranes at room temperature overnight. Then, the membranes were incubated by the HRP-conjugated secondary antibodies. The blots were detected using an enhanced chemiluminescence reagent (Millipore, Billerica, MA, USA).

Global mRNA expression was analyzed by the PrimeView Human Gene Expression Array (Affymetrix). Total RNA was converted into cRNA and labeled with biotin using MessageAmp Premier RNA Amplification Kit (#1792, Ambion) according to the manufacturer’s instructions. The fragmented cRNAs were hybridized on the gene chip, and then the chip was washed and stained following the manufacturer’s standard protocol. The fluorescent signal was scanned by GeneChip Scanner 3000 (Affymetrix) and converted into digital data (CEL) using Affymetrix GeneChip Command Console (AGCC) software. The resulting data were preprocessed using Robust Multi-array Average (RMA) algorithm. The fold change (FC) of gene expression in shMRVI1-AS1 cells was calculated relative to shNTC cells. A gene was defined as differentially expressed if its log2|FC| > 0.5.

Hep3B and MHCC-97H cells were incubated at 20% or 1% O₂ for 16 h, cross-linked in 3.7% formaldehyde for 15 min, quenched in 0.125 M glycine for 5 min, and lysed with SDS lysis buffer. Chromatin was sheared by sonication, and lysates were precleared with salmon sperm DNA/protein A agarose slurry (Millipore) for 1 h and incubated with antibody against HIF-1α (# ab228649, Abcam, USA) or IgG (#ab97051, Abcam, USA) in the presence of protein salt, high-salt, and LiCl buffers; DNA was 426 eluted in 1% SDS with 0.1 M NaHCO3, and cross-links were reversed by addition of 0.2 M NaCl. DNA was purified by phenol–chloroform extraction and ethanol 427 precipitation and analyzed by qPCR. Primers are as below: MRVI1-AS1-HRE-1-Forward: 5’-AGACGGGCGTCAATAGAATG-3’, MRVI1-AS1-HRE-1-Reverse: 5’-TTGCTAGCTGCTCCAGGACT-3’. MRVI1-AS1-HRE-2-Forward: 5’-TTAGCCGGGTCTCAAGGTAG-3’, MRVI1-AS1-HRE-2-Reverse: 5’-GGCTGGACACCCAAATAAGA-3’.

Nude mice (BALB/c, female, 4 weeks old) were adopted for the establishment of the intravenous transplantation tumor model and the subcutaneous xenograft tumor model. In the intravenous transplantation tumor model, the mice were inoculated with MHCC-97H subclones at a density of 2 × 10⁵ cells/100 µL through the tail vein. Five weeks after cell injection, the mice were euthanized, and the formation of metastatic lung nodes was observed and evaluated. In the subcutaneous xenograft tumor, MHCC-97H subclones cells (2 × 10⁶/200 µL) were subcutaneously injected into the right flank of mice. Then, the tumor growth was measured every week, and calipers were used to measure tumor length (L) and width (W), and tumor volume (V) was calculated as V = L × W² × 0.524. Four weeks after cell injection, the mice were euthanized, then the tumor nodules were resected, and the tumor weight was measured. Part of the tumor nodule was stored at −80℃ for the detection of RT-qPCR, and the rest was fixed in 4% formaldehyde solution for immunohistochemical staining of Ki-67 (#ab238020, Abcam, USA). The protocols for the above mice experiments were approved by the Institutional Animal Ethical Committee of the Xi’an Jiaotong University.

Graphpad Prism 8.0 (San Diago, CA, USA) and SPSS 20.0 (SPSS, Inc., Chicago, IL, USA) were applied to analyze the data. All of the data are presented as mean ± S.D. Statistical methods in this study included Student’s t-test, one-way ANOVA, Chi-square test, Kaplan–Meier method, log-rank test, and Pearson’s correlation coefficient analysis. The difference with P < 0.05 was considered to be statistically significant.

MicroArray or RNA-Seq was performed to analyze the abnormally expressed lncRNAs in HCC tissues and adjacent non-tumor (NT) tissues. The result indicated that MRVI1-AS1 was the one with the highest fold change increase among the upregulated lncRNAs in HCC (Fig. 1A). Furthermore, the data form RT-qPCR suggested MRVI1-AS1 was dramatically upregulated in HCC tissues, compared to that in non-tumor tissues (Fig. 1)B. Additionally, the analysis of TCGA data from GEPIA platform consistently found the higher expression of MRVI1-AS1 in HCC (Fig. 1C), and RT-qPCR results in HCC cell lines revealed that MRVI1-AS1 expressions in all of the five HCC cell lines (Hep3B, Huh7, SK-HEP-1, HepG2, and MHCC-97H) were dramatically higher than that in the human normal liver cell line (LO2) (Fig. 1D).

Then, we explored the correlation between the clinical significance and MRVI1-AS1 expression in HCC by sorting the 72 patients into low and high MRVI1-AS1 group on the basis of the median expression of MRVI1-AS1 in HCC tissues. Intriguingly, MRVI1-AS1 was closely related to tumor size, venous infiltration, and TNM stage (Table 1). Additionally, HCC patients with higher MRVI1-AS1 expression had both worse 5-year overall survival (OS) (Fig. 1E) and disease-free survival (DFS) (Fig. 1F). Thus, the above findings suggest that MRVI1-AS1 may promote HCC progression and development.

Next, to further determine the role of MRVI1-AS1 in HCC, we firstly attempted to investigate whether MRVI1-AS1 was capable of promoting HCC cells invasion and migration. The MHCC-97H subclones stably expressing MRVI1-AS1 shRNAs or control, and Hep3B subclones stably expressing pcDNA/MRVI1-AS1 or control were produced by lentivirus transfection. Then, the knockdown and overexpression efficiencies were validated by RT-qPCR (Fig. 2A). Subsequently, transwell migration and invasion assays were performed. Results manifested that both migration and invasion abilities of MHCC-97H cells were repressed in MRVI1-AS1-knockdown subclones of MHCC-97H (Fig. 2B), while these two abilities were markedly enhanced by ectopic expression of MRVI1-AS1 in Hep3B cells (Fig. 2C). Consistently, the similar results were found in wound healing assay (Fig. 2D, E). Next, we attempted to explore the function of MRVI1-AS1 in HCC cells growth. MTT assay results manifested that MRVI1-AS1 shRNAs weakened MHCC-97H cells viability (Fig. 2F). In contrast, upregulated MRVI1-AS1 promoted Hep3B cells viability (Fig. 2G). Consistently, in EdU assay, the proportions of EdU-positive cells were much lower in MRVI1-AS1-knockdown subclones compared to the control group (Fig. 2H), while the proportion was much higher in MRVI1-AS1-overexpressing subclone (Fig. 2I). Furthermore, we established intravenous transplantation tumor model and subcutaneous xenograft tumor model to respectively examine whether MRVI1-AS1 promoted HCC cells metastasis and growth in vivo. The hematoxylin-eosin (H&E) staining results in lung tissues indicated that tumor nodules were less likely to form or grow bigger in lung tissue of MRVI1-AS1-knockdown mouse group (Fig. 2)J. Additionally, the tumor grew slower and the final tumor weight was obviously lighter in the mice with MRVI1-AS1 silencing compared to the control group (Fig. 2K-M). Expression of MRVI1-AS1 in NTC and MRVI1-AS1-knockdown tumors was validated by RT-qPCR (Fig. 2N). And HCC nodule tissues in MRVI1-AS1 knockdown group showed a weaker Ki-67 staining compared to NTC group (Fig. 2O). Collectively, these findings demonstrate that MRVI1-AS1 facilitates metastasis and growth of HCC cells.

Next, we sought to explore the potential mechanism of MRVI1-AS1 in HCC cells. The mRNA expression difference profile in MRVI1-AS1-knockdown MHCC-97H cells was examined by RNA-seq to identify the downstream targets. And SKA1 in particular caught our attention due to its remarkable expression fold change upon MRVI1-AS1 knockdown (Fig. 3A). Moreover, deletion of MRVI1-AS1 suppressed both mRNA and protein levels of SKA1 in MHCC-97H cells (Fig. 3B). In contrast, overexpression of MRVI1-AS1 increased both mRNA and protein levels of SKA1 in Hep3B cells (Fig. 3C). RT-qPCR analysis determined a higher expression of SKA1 in HCC tissues (Fig. 3D), and TCGA data analysis revealed the consistent result (Fig. 3E). In addition, in HCC tissues, MRVI1-AS1 expression was positively correlated with SKA1 expression (Fig. 3F), and data from UALCAN showed that high SKA1 expression had a close relationship with worse prognosis of HCC patients (Fig. 3G).

Subsequently, we tried to discover the underlying mechanism about how MRVI1-AS1 affected SKA1 expression. Firstly, we assumed that MRVI1-AS1 influenced SKA1 transcription in HCC cells, and the luciferase reporter containing SKA1 promoter was constructed. Unfortunately, the expression of MRVI1-AS1 had no influence on the luciferase activity of SKA1 promoter (Fig. 3H). Nevertheless, we determined that MRVI1-AS1 mainly localized in HCC cells cytoplasm (Fig. 3I), which suggested that MRVI1-AS1 might be associated with the stability of SKA1 mRNA. Furthermore, actinomycin D assay was performed in MRVI1-AS1-related HCC subclones, then isolated RNA was subjected to RT-qPCR analysis. As expected, the half-life of SKA1 mRNA was dramatically shortened in the situation of deletion of MRVI1-AS1 (Fig. 3J), while the half-life of SKA1 mRNA was prolonged by overexpressed MRVI1-AS1 (Fig. 3K). Collectively, these data suggest that MRVI1-AS1 increases SKA1 expression through strengthening the stability of SKA1 mRNA.

It has been reported that mRNA could be stabilized by RNA-binding proteins, which are recruited by lncRNAs [22‐24]. Thus, we made the assumption that MRVI1-AS1 could enhance SKA1 mRNA stability mediated by a certain kind of RNA-binding protein. Firstly, StarBase (http://​starbase.​sysu.​edu.​cn) and RPISeq (http://​pridb.​gdcb.​iastate.​edu/​RPISeq) were employed to predict the potential RNA-binding protein, which bond to both MRVI1-AS1 and SKA1 mRNA. The data showed that CELF2 was the potential RNA-binding protein for both MRVI1-AS1 and SKA1 mRNA (Fig. 4A, B). The binding of MRVI1-AS1 with CELF2 protein was verified by western blot analysis following the RNA pull-down assay both in MHCC-97H and Hep3B cells (Fig. 4C). Data of RIP assay indicated that both MRVI1-AS1 and SKA1 mRNA were enriched by the antibody against CELF2 (Fig. 4D-F). Furthermore, the interaction between CELF2 and SKA1 mRNA in MHCC-97H cells was impaired in the absence of MRVI1-AS1 (Fig. 4E), while the interaction was strengthened in Hep3B cells with overexpression of MRVI1-AS1 (Fig. 4F). In addition, western blot data revealed that SKA1 mRNA expression, but not CELF2, was increased by pcDNA/MRVI1-AS1, and the induction was abrogated by CELF2 shRNA (Fig. 4G, H). Taken together, we conclude that MRVI1-AS1 increases SKA1 expression by recruiting RNA-binding protein CELF2 to stabilize SKA1 mRNA.

Next, we attempted to validate that SKA1 mediated the influences of MRVI1-AS1 on HCC cells. As expected, MRVI1-AS1-knockdown inhibited SKA1 expression, and the repression was blocked by SKA1 overexpressing in MHCC-97H cells (Fig. 5A). On the other hand, SKA1 expression was increased by overexpressed MRVI1-AS1, and the induction was abrogated by SKA1 shRNA in Hep3B cells (Fig. 5B). Moreover, data from transwell assays and wound healing assay collectively manifested that SKA1 overexpression significantly reversed the suppression of MHCC-97H cells migration and invasion by MRVI1-AS1 silencing (Fig. 5C, D). SKA1 silencing offset the positive effects of MRVI1-AS1 overexpression on the migrated ability and invasive ability of Hep3B cells (Fig. 5E, F). In addition, the promoting effect of pcDNA/MRVI1-AS1 on Hep3B cells viability was abrogated by SKA1 silencing, as confirmed by MTT assay (Fig. 5G). In contrast, MRVI1-AS1 shRNA inhibited MHCC-97H cells viability, but SKA1 overexpression reversed the inhibitory effects on (Fig. 5H). Similarly, in EdU assay, the inhibitory effect of MRVI1-AS1 shRNA on MHCC-97H cell proliferation was determined, which then was rescued by SKA1 overexpression (Fig. 5I). SKA1 silencing offset the promoting effects of pcDNA/MRVI1-AS1 on Hep3B cell viability (Fig.5J). In brief, these findings suggest that MRVI1-AS1 promotes HCC cells metastasis and growth under the mediation of SKA1.

In cancer, as a transcriptional trigger for numerous genes, including lncRNAs, hypoxic microenvironment addresses more and more attention to researchers [26, 27]. Here, we attempted to explore whether MRVI1-AS1 could be regulated by hypoxia. We exposed Hep3B and MHCC-97H cells to normoxia (20% O₂) or hypoxia (1% O₂) for 24 h and isolated RNA from these cells. RT-qPCR analysis data indicated that MRVI1-AS1 expressions in both Hep3B and MHCC-97H cells were significantly facilitated by hypoxia (Fig. 6A). Then, in order to determine whether HIF-1α mediated the induction of MRVI1-AS1 by hypoxia, the candidate hypoxia-response element (HRE) in the promoter region of MRVI1-AS1 gene was discovered by JASPAR database. Data indicated that 9 candidate HRE sites were predicted in the promoter region of MRVI1-AS1 gene (Fig. 6B, Supplemental Figure (1)), suggesting that MRVI1-AS1 could be a HIF-1 target gene. After the construction of HIF-1α-knockdown subclones, western blot was employed to verify the knockdown efficiency (Fig. 6C). In shNTC subclones of Hep3B and MHCC-97H cells, hypoxia increased MRVI1-AS1 levels, and the inductions were counteracted by HIF-1α silencing (Fig. 6D, E). Furthermore, ChIP assays were conducted in Hep3B and MHCC-97H cells, which had been exposed to 20% or 1% O₂ for 16 h. The two HRE sites located 0.3 kb 5’ and 0.7 kb 5’ to the transcription start site (TSS) were identified (Fig. 6F-I), suggesting that HIF-1α and HRE sites were essential for MRVI1-AS1 transcription under hypoxia. Taken together, we demonstrate that hypoxia induces MRVI1-AS1 expression through HIF-1-depedent manner in HCC cells.

In order to investigate whether MRVI1-AS1/SKA1 pathway mediated HCC progression induced by hypoxia, we conducted a series of rescue experiments. Rescue experiments of transwell assays revealed that hypoxia dramatically promoted MHCC-97H and Hep3B cells migration and invasion, while MRVI1-AS1-knockdown or SKA1-knockdown counteracted the promoting effects of hypoxia on HCC cells migration and invasion (Fig. 7A, B). Consistently, wound healing assay indicated that MRVI1-AS1-knockdown or SKA1-knockdown offset the promoting effects of hypoxia on MHCC-97H and Hep3B cells mobility (Fig. 7C, D). In addition, MTT assay (Fig. 7E, F) and Edu assay (Fig.7G, H) indicated that MHCC-97H and Hep3B cells proliferation were accelerated by hypoxia, while MRVI1-AS1-knockdown or SKA1-knockdown abrogated the promoting effects. Thus, we demonstrated that hypoxia promotes HCC progression through MRVI1-AS1/SKA1 pathway.