Long non-coding RNAs HERH-1 and HERH-4 facilitate cyclin A2 expression and accelerate cell cycle progression in advanced hepatocellular carcinoma

Twelve pairs of human HCC tissue samples, including HCC primary and recurrent tissues from patients undergoing liver transplantation (Table S1), were obtained from the Biological Sample Resource Sharing Center (BSRSC) of the Tianjin First Central Hospital with the patients’ informed consent. After surgical resection or biopsy, the tissue samples were flash-frozen in liquid nitrogen and then stored at − 80 °C until use. Two pairs of the tissues were applied in microarray analysis, and the other ten pairs were used in the following quantitative reverse transcription PCR (qRT-PCR) for validation of the dysregulated lncRNAs. This study was performed in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the Tianjin First Central Hospital.

An immortalized human benign hepatocyte cell line HL-7702, and four human HCC cell lines HepG2, QGY-7703, SMMC-7721 and Huh-7 were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China) with confirmed identities of these cell lines. These cell lines were maintained in DMEM (Solarbio, Beijing, China; for HL-7702, HepG2 and Huh-7) or RPMI-1640 medium (Solarbio; for QGY-7703 and SMMC-7721) supplemented with 10% fetal bovine serum (FBS; Biowest, Nuaillé, France) at 37 °C in a humidified chamber supplemented with 5% CO₂. Transfection of plasmids or oligonucleotides was performed using Lipofectamine 2000 (ThermoFisher, Waltham, MA, USA).

Complementary DNA (cDNA) labeled with Cy3-dCTP (for primary HCC) or Cy5-dCTP (for recurrent HCC) was produced by Eberwine’s linear RNA amplification method and subsequent enzymatic reaction using a cRNA Amplification and Labeling Kit (CapitalBio, Beijing, China) [22]. The labeled cDNA was hybridized with CapitalBio Technology Human LncRNA Array V4 containing probes inspecting about 41,000 human lncRNAs and approximately 34,000 human mRNAs. Microarray data were analyzed using the GeneSpring software V13.0 (Agilent, Santa Clara, CA, USA). Genes with an absolute fold change value ≥2 and a Benjamini-Hochberg corrected P- value ≤0.05 were treated as differentially expressed genes. Hierarchical clustering analysis was performed using Cluster 3.0 software (Stanford University, CA, USA).

For quantification of lncRNAs and protein-coding genes, 5 μg of long RNA was reverse transcribed into cDNA using oligo dT (for mRNAs) or Random 6 (for lncRNAs) primers using a PrimeScript II 1st Strand cDNA Synthesis Kit (TaKaRa, Otsu, Shiga, Japan). The cDNA was then used for amplification of the target RNA. β-actin was used as an endogenous control. Quantification of miRNAs and endogenous control U6 snRNA was performed using the stem-loop RT-PCR method [23]. All the qRT-PCR tests were carried out in two independent experiments with each PCR reaction performed in duplicate. The sequence of all primers and oligonucleotides used in this study are provided in Table S2.

The lncRNA length was evaluated by Northern blot assay using a NorthernMax Kit (ThermoFisher), following the manufacturer’s instructions. The blot images were captured using a ChemiDoc XRS+ imaging system (Bio-Rad, Hercules, CA, USA).

The subcellular localization of lncRNA was evaluated by FISH assay using a Ribo Fluorescent In Situ Hybridization Kit (Ribobio, Guangzhou, China). Briefly, HCC cells were planted into 24-well plate at 8 × 10⁴ (HepG2) or 4 × 10⁴ (QGY-7703) per well. The transfected HCC cells were fixed in 4% paraformaldehyde and incubated in 0.5% Triton X-100 for higher membrane permeability. After prehybridization, the cells were incubated with Ribo lncRNA FISH Probe Mix (Ribobio) at 37 °C overnight. After stringency washing and DAPI staining, fluorescence was observed under an IX71 fluorescence microscope (Olympus, Shinjuku, Japan).

The protein-coding potential was assessed using the ORFfinder (https://​www.​ncbi.​nlm.​nih.​gov/​orffinder/​) and CPAT (http://​lilab.​research.​bcm.​edu/​cpat/​index.​php) databases. The concordance of HCC gene profiles from microarray data with defined gene sets was analyzed using the Gene Set Enrichment Analysis (GSEA) database (https://​www.​gsea-msigdb.​org/​gsea/​index.​jsp) [24]. Potential TFs or miRNAs binding with lncRNA were predicted using the RegRNA 2.0 database (http://​regrna2.​mbc.​nctu.​edu.​tw/​detection.​html) [25]. The gene promoter region was predicted using PROMO (http://​alggen.​lsi.​upc.​es/​cgi-bin/​promo_​v3/​promo/​promoinit.​cgi?​dirDB=​TF_​8.​3) [26], and potential TFs binding with the gene promoter were predicted using PROMO and GeneCards (https://​www.​genecards.​org/​) [27]. The DNA-binding preferences of TFs were analyzed by JASPAR (http://​jaspar.​genereg.​net/​) [28]. Potential targets of miRNA were predicted using the TargetScanHuman database (http://​www.​targetscan.​org/​) [29].

Overexpression of lncRNAs in HCC cells was achieved by the eukaryotic expression plasmid pcDNA3.1(+) (ThermoFisher). The exon fragments were amplified by PCR using human genomic DNA as a template. The fragments were then cloned into the pcDNA3.1(+) plasmid (Fig. S1). Suppression of endogenous lncRNAs was achieved by transfecting synthesized siRNA into HCC cells.

A double-strand RNA fragment was synthesized to serve as mimic of the miR-29b/c. A 2′-O-methyl modified single-strand RNA fragment that was inversely complementary to mature miR-29b/c (antisense oligonucleotide, ASO) was synthesized to serve as miR-29b/c inhibitor. These oligonucleotides were induced into HCC cells to artificially change miR-29b/c levels.

Overexpression of protein-coding genes was also achieved using the pcDNA3.1(+) vector. The full-length coding sequence was amplified by PCR using a human cDNA library as a template. The fragment was then cloned into the pcDNA3.1(+) plasmid.

HCC cells were planted into 24-well plate at 8 × 10⁴ (HepG2) or 4 × 10⁴ (QGY-7703) per well and transfected on the next day. At 24 h after transfection, the cells were dissociated and planted into 96-well plate at 1 × 10⁴ (HepG2) or 5 × 10³ (QGY-7703) per well with three duplicates for each group. All the CCK-8 tests were carried out in three independent experiments. HCC cell viability was detected using the CCK-8 reagent (Dojindo, Tokyo, Japan). The absorbance at 450 nm (A₄₅₀) was measured using an EnSpire Multimode Plate Reader (PerkinElmer, Waltham, MA, USA).

The HCC cells were planted and treated similarly as the CCK-8 assay. At 24 h and 48 h after planting, the number of cells in each well of the 96-well plate was accurately counted. The doubling time was calculated using the formula: doubling time = Δt(lg2/(lgNt-lgN0)), in which Δt is the interval between the cell counting (24 h in this experiment), and N0 and Nt are the cell numbers at 24 h and 48 h after planting, respectively. All the cell doubling time tests were carried out in three independent experiments with each group in triplicate.

As a kind of thymidine analog, EdU incorporates into genome DNA in the S phase during DNA synthesis. The quantity of the incorporated EdU reflects cell proliferation activity. EdU staining of HCC cells was performed using the Cell-Light EdU Apollo488 In Vitro Flow Cytometry Kit or Cell-Light EdU Apollo488 In Vitro Kit (Ribobio). For the flow cytometry analysis, HCC cells were planted into 6-well plate at 3 × 10⁵ (HepG2) or 1.5 × 10⁵ (QGY-7703) per well. The cells were transfected, stained and analyzed using an Accuri C6 Plus flow cytometer (BD, San Jose, CA, USA). For imaging, the HCC cells were planted and treated similarly as the CCK-8 assay. The cells in 96-well plate were stained and captured under an IX71 fluorescence microscope (Olympus). Fluorescence intensity was quantified using ImageJ software. The EdU cytometry and imaging tests were performed in two independent experiments with each group in duplicate.

HCC cells were planted into 6-well plate at 3 × 10⁵ (HepG2) or 1.5 × 10⁵ (QGY-7703) per well. Approximately 10⁶ transfected HCC cells were fixed in 70% ethanol for at least 2 h. After washing with PBS, cells were stained in 1 mL PI staining solution containing 10 μg/mL of PI, 100 μg/mL of RNase A, and 0.1% Triton X-100 dissolved in PBS, for 30 min in the dark. Cell cycle distribution was analyzed by flow cytometry using an Accuri C6 Plus flow cytometer (BD). The cell cycle tests were performed in two independent experiments with each group in duplicate.

The predicted CCNA2 promoter region was amplified by PCR and cloned into the pGL3/Enhancer vector (Promega, Madison, WI, USA). In addition, a CCNA2 promoter with deleted CREB1 binding sites was also amplified and cloned into the reporter vector. HCC cells were planted into 24-well plate at 8 × 10⁴ (HepG2) or 4 × 10⁴ (QGY-7703) per well. The reporter plasmids were transfected into HCC cells and the promoter efficiency was evaluated by measuring the luciferase activity using a Luciferase Assay System (Promega). Chemiluminescence was measured using an EnSpire Multimode Plate Reader (PerkinElmer). The gene promoter efficiency tests were performed in three independent experiments with each group in triplicate.

The interaction between TF and gene promoter was confirmed by ChIP assay using ab500 ChIP Kit (Abcam, Cambridge, UK) according to the manufacturer’s instructions. Approximately 1 × 10⁶ HCC cells were collected for each ChIP assay. The TF conjugated DNA fragments were purified and the target DNA was identified using quantitative PCR.

The interaction between lncRNA and TF was detected by RNA pull-down assay using a Pierce Magnetic RNA-Protein Pull-Down Kit (ThermoFisher) according to the manufacturer’s instructions. Approximately 6 × 10⁶ HCC cells were collected for each RNA pull-down assay. The potential RNA-binding proteins were eluted and purified for the further Western blot analysis.

Protein samples were resolved on an SDS denaturing polyacrylamide gel and transferred onto a nitrocellulose membrane (Boster, Wuhan, China). The membrane was incubated with the primary antibody overnight at 4 °C. The membrane was then washed and incubated with horseradish peroxidase (HRP)-conjugated secondary antibody. After chemiluminescence, the bands were captured using a ChemiDoc XRS+ imaging system (Bio-Rad). The band intensity was quantified using AlphaView SA software V3.4.0 (ProteinSimple, San Jose, CA, USA).

RNA targets of RNA-binding proteins were identified using a Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA, USA), following the manufacturer’s instructions. Approximately 2 × 10⁷ HCC cells were collected for each RIP assay. RNA in the RBP immunoprecipitation was purified and the target RNA was analyzed via RT-PCR and agarose gel electrophoresis.

HCC cells were planted into 24-well plate at 8 × 10⁴ (HepG2) or 4 × 10⁴ (QGY-7703) per well. After 24 h, the cells were transfected with a GFP reporter vector along with associated plasmids or oligonucleotides. At 48 h after transfection, the cells were lysed with RIPA lysis buffer (Solarbio) and GFP intensity was measured using an EnSpire Multimode Plate Reader (PerkinElmer). The fluorescent reporter assays were performed in three individual experiments with each group in triplicate.

All numerical values were recorded as mean ± standard deviation (SD). The hypothesis test for significance between two groups utilized the Student’s t test. For three or more groups, one-way analysis of variance (ANOVA) was applied, followed by the Student-Newman-Keuls q test for comparing two of these groups. Statistical significance was set at P ≤ 0.05. Data processing and figure drawing were performed using a GraphPad Prism 7 software (GraphPad Software, La Jolla, CA, USA).

The HCC recurrent tissues after LT were treated as typical clinical models of the advanced HCC. To assess the role of lncRNAs for advanced HCC, we investigated a patient sample set of 12 matching HCC primary and recurrent tissues by microarray and following qRT-PCR validation. The microarray analysis from two pairs of HCC tissues revealed that 716 lncRNAs and 2090 mRNAs were upregulated, and 997 lncRNAs and 958 mRNAs were downregulated in HCC recurrent tissues compared with the associated primary tumor tissues (Fig. S2; GEO accession: GSE102759). The cluster analysis indicated that the four HCC tissues could be grouped into two clusters as HCC recurrent and primary tissues (Fig. 1a). We selected the top five mostly dysregulated lncRNAs that have been noted in the Ensembl GRCh37 database (Table S3), and two of these lncRNAs showed good concordance with the microarray data by qRT-PCR in the other ten pairs of HCC tissues. We chose these two lncRNAs for further study, and named them as highly expressed lncRNAs in recurrent HCC (HERH)-1 and HERH-4 (Fig. 1b, c; Fig. S3; Table S4). Bioinformatics analysis indicated that HERH-1/4 lacked protein-coding potential (Table S5). HERH-1/4 was significantly induced in all the four human HCC cell lines compared to the immortalized, but benign hepatocyte cell line HL-7702, while HepG2 and QGY-7703 cells exhibited highest levels of HERH-1/4 (Fig. 1d). These two cell lines were applied in the following mechanistic studies.

Northern blot assays confirmed the existence of the two lncRNAs in HCC cells, and the length was in accordance with the prediction by the Ensembl GRCh37 database (Fig. 1e). More importantly, we found that HERH-1 was mainly located in the nucleus of HCC cells. HERH-4 appeared to be located in both the nucleus and the cytoplasm according to the FISH images. However, qRT-PCR assay confirmed that HERH-4 was mainly located in the cytoplasm of HCC cells (Fig. 1f, g). This indicated their possible functional patterns in the advanced HCC cells.

In order to determine the phenotypes that exhibited dominant alterations in the advanced HCC cells, we applied the GSEA database to analyze the mRNA profile (Table S6). Compared with the primary HCC, the mRNA expression pattern of the HCC recurrent tissues showed more similarity with that of the cells in which the RB gene was inhibited (Fig. 1h, Table S7). As a tumor suppressor, RB is a negative regulator of the cell cycle [30]. We presumed that the advanced HCC cells probably displayed an accelerated cell cycle progression.

We used pcDNA3 eukaryotic expression vectors (Fig. S1) and small interfering RNAs (siRNAs) to achieve gain-of-function and loss-of-function of the two lncRNAs, respectively, in HCC cell lines (Fig. S4). Both HERH-1 and HERH-4 enhanced the viability and shortened the doubling time of HepG2 and QGY-7703 cells (Fig. 2a, b). MKI67 is a marker for general cell cycle activity and is expressed through G1–M phase. PCNA is expressed in cells undergoing DNA replication during S phase or DNA repair processes. The levels of these two proliferation marker genes also indicated positive regulation of these two lncRNAs on the proliferation of HCC cells (Fig. 2c). Genome DNA replication is a key process during cell proliferation. EdU cell proliferation assays suggested that HERH-1/4 levels were positively associated with the DNA replication rate (Fig. 2d, e). Furthermore, HERH-1/4 also affected the HCC cell cycle distribution. These two lncRNAs reduced HCC cells in the G1 phase, and increased those in the S phase, indicating a promoted G1-S transition. The number of cells in the G2/M phase was not affected by the lncRNAs (Fig. 2f). These data indicated that induced HERH-1/4 expression is associated with enhanced cell cycle and S phase activity.

Next, we tended to find out the functional protein-coding genes that were regulated by the lncRNAs HERH-1/4. It was presumed that the significant genes were probably involved in cell cycle process, and might be dysregulated in the advanced HCC cells. First, we searched the GSEA database and found that the gene set CELL_CYCLE_GO_0007049 records 315 genes annotated to the cell cycle (Table S8). Then, by combining our microarray data with this gene set, we found that 76 of the 315 cell cycle-associated genes were dysregulated in the microarray, in which 68 genes were upregulated and 8 genes were downregulated (Fig. 3a, Table S9). This method effectively reduced the number of the candidate genes. Among them, we chose the top five genes that showed the most obvious range of variation in the microarray, and detected their levels in HCC clinical tissues and cell lines. These genes had high probability to be regulated by HERH-1/4. As a result, cyclin A2 (CCNA2) levels exhibited a positive correlation with HERH-1/4 levels in the HCC recurrent versus primary tissues, and was found to be co-regulated in HCC cell lines (Fig. 3b, c). None of the other genes showed statistically significant correlation with the two lncRNAs (Fig. S5). These data indicated that CCNA2 was a potential target of HERH-1/4. Therefore, we investigated a potential regulation of CCNA2 by HERH-1/4.

Given that HERH-1 is mainly located in the nucleus, we considered that HERH-1 may regulate CCNA2 at the transcriptional level by interacting with transcription factors (TFs). According to RegRNA, 26 TFs may bind with HERH-1. PROMO and GeneCards predicted 44 and 7 TFs that potentially interact with the CCNA2 promoter, respectively (Table S10). CREB1 was the only TF that presented in all the three lists (Fig. 3d). In this study, we focused on the interaction of CREB1 with HERH-1 and the CCNA2 promoter.

There are three potential binding motifs within the CCNA2 promoter. In order to investigate the activity of the CCNA2 promoter, we introduced a luciferase reporter pGL3/Enhancer, in which the objective promoter sequence can be cloned at upstream of the luciferase coding region. A naive CCNA2 promoter (Table S11) or a mutated CCNA2 promoter lacking the three TF-binding sites for CREB1 were cloned into the pGL3/Enhancer plasmid (Fig. 3e). As a positive control, luciferase was expressed under control of the SV40 promoter (pGL3-Control; Fig. 3e, f). The naive CCNA2 promoter boosted luciferase expression in HCC cells. In a next step, we co-expressed CREB1 in these cells (Fig. S6) and found a further increase of the luciferase activity. Importantly, when the three potential binding sites of CREB1 were deleted, luciferase expression was no longer strengthened by CREB1 (Fig. 3f).

Next, we verified the interaction between CREB1, CCNA2 promoter and HERH-1 in HCC cells. ChIP assay suggested that the CREB1 protein was directly bound to the CCNA2 promoter region (Fig. 3g). The interaction between HERH-1 and CREB1 was verified by RNA pull-down (Fig. 3h) and RIP (Fig. 3i) assays. These results suggest that lncRNA HERH-1 binds with CREB1 and may facilitate the interaction of CREB1 with the CCNA2 promoter region.

In HCC cells, ectopic expression of CREB1 led to increased CCNA2 expression. This depended on the existence of HERH-1 because CREB1 failed to influence CCNA2 if endogenous HERH-1 was knocked down (Fig. 4a, b). As predicted, CREB1 enhanced cell viability (Fig. 4c), shortened the cell doubling time (Fig. 4d), accelerated genome DNA replication (Fig. 4e, f), and promoted the transition of cells from the G1 phase to the S phase (Fig. 4g) in HCC cell lines. In addition, when endogenous HERH-1 was blocked, the oncogenic effects of the CREB1 protein on HCC cells were no longer detectable (Fig. 4c–g). These results further demonstrated that HERH-1 facilitated CCNA2 expression by assisting CREB1 in CCNA2 transcription activity.

Given that lncRNA HERH-4 is located in the cytoplasm, we explored its mechanism of HCC proliferation at the post-transcriptional level. According to the RegRNA 2.0 database, we found five miRNAs that possibly bound to HERH-4, and miR-29c-3p (miR-29c) had the highest score (Table S12). To search for the potential ceRNA of HERH-4, we used the TargetScan database to predict miR-29-3p targets, and used the GSEA database to identify the cell cycle-associated genes. In addition, we focused on the upregulated protein-coding genes in the microarray because a high HERH-4 level reasonably led to an increase in its ceRNA [31]. As a result, three genes, including CCNA2, were present in all these three gene sets (Fig. 5a; Table S13). The miR-29b is an analog of miR-29c, and they share the same seed region. The levels of miR-29b/c were artificially altered by introducing mimics or ASOs into HCC cells (Fig. S7). miR-29b/c suppressed HERH-4 level by interacting with the predicted miRNA response element (MRE; Fig. 5b). Endogenous HERH-4 was negatively regulated by miR-29b/c (Fig. 5c). Similarly, CCNA2 expression was directly suppressed by miR-29b/c (Fig. 5d, e). These data confirmed that both HERH-4 and CCNA2 mRNA were direct targets of miR-29b/c.

In order to validate that HERH-4 positively regulated CCNA2 expression via absorbing miR-29b/c, we overexpressed the miR-29b/c response elements within HERH-4 in HCC cells. We found the GFP reporter intensity to be increased with CCNA2 wild-type MRE, while further induction of miR-29b/c reduced the GFP intensity to basal level. The positive regulation of HERH-4 MRE on CCNA2 was undetectable when the sequence of either HERH-4 MRE or CCNA2 MRE was mutated (Fig. 5f). Endogenous CCNA2 expression was also regulated by HERH-4 MRE and miR-29b/c following the same model (Fig. 5g).

An optimal ceRNA cross-talk occurs at a near-equimolar equilibrium of all ceRNAs and miRNAs within a network [32]. Absolute quantification indicated that the four molecules involved in the ceRNA network had comparable levels in HCC cells (Fig. 5h). The RIP experiment demonstrated that all the four molecules were recruited to AGO2 protein (Fig. 5i), a key factor in the RNA-induced silencing complex in which miRNAs exert their function. These data further supported the ceRNA network between HERH-4 and CCNA2.

In order to analyze the effects of the HERH-4-miR-29b/c-CCNA2 ceRNA network on HCC progression, we overexpressed the 22 nt length of the miR-29b/c binding sequence within HERH-4, and found a higher viability (Fig. 6a), a shorter doubling time (Fig. 6b), and a rapider genome DNA replication (Fig. 6c, d) in HCC cell lines. The mutated MRE sequence failed to promote HCC cell proliferation (Fig. 6a–d). Importantly, if we introduced miR-29b/c mimics following overexpression of the wild-type HERH-4 MRE, the proliferation activity of the HCC cells restored (Fig. 6a–d). The wild-type rather than the mutated HERH-4 MRE reduced HCC cells in the G1 phase and increased those in the S phase, and miR-29b/c restored these effects (Fig. 6e). These data support the hypothesis that HERH-4 acts as a ceRNA to facilitate CCNA2 expression and promote HCC proliferation and cell cycle progression.