Methylation-mediated regulation of tumor-suppressor function of the miR-379/656 (C14MC) cluster and its clinical utility in hepatocellular carcinoma

Abstract

Background

Hepatocellular carcinoma is a leading cause of cancer-related mortality. Several microRNAs play key roles in HCC development and progression. Epigenetic processes, such as DNA methylation, might regulate these RNAs during HCC pathogenesis. In this study, we show that the miR-379/656 cluster/C14MC acts as a tumor suppressor cluster and is epigenetically regulated by DNA methylation.

Methods

C14MC miRNA expression was determined in HCC cell lines using the nCounter assay and from the TCGA-LIHC clinical cohort. C14MC putative promoter was identified, characterized using cloning and luciferase assay, and the methylation status of promoter-bound CpGs was determined using bisulfite Sanger sequencing. The expressions of C14MC targets were experimentally validated by transcriptomic sequencing or transfecting mimics, followed by qRT-PCR. Furthermore, the diagnostic and prognostic significance of C14MC and its target interactome in HCC was assessed using clinical data from the TCGA-LIHC cohort.

Results

C14MC was downregulated in HCC cell lines and in TCGA-LIHC. The loss of C14MC tumor suppressor function was directly regulated by the hypermethylation of promoter-CpGs. Reactivating specific C14MC miRNAs, such as miR-299-5p and miR-376c-3p via mimics, abrogated the expression of several target oncogenes, including PARP1, SPP1, RAD21, and CENPA, which regulate critical molecular pathways such as the p53 signaling and NF-κB signaling pathways in HCC. Additionally, overexpressing these miRNAs inhibited HCC cell migration and invasion. Also, C14MC and its target interactome exhibited significant clinical correlation in terms of survival outcomes of HCC patients.

Conclusions

This is the first study to show that C14MC is a methylation-dependent cluster in HCC. Several of these miRNAs and their targets can be used for early HCC diagnosis and prognosis. Thus, targeting C14MC can be useful in HCC management.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-026-16318-2.

Background

Hepatocellular carcinoma (HCC) is the most commonly reported form of primary liver cancer. In 2021, HCC accounted for approximately 529,000 cases and 484,000 mortalities among the existing cancer burden reported globally [1]. While hepatitis B remains the leading factor, steatohepatitis is an emerging contributor to HCC [2]. Although the existing vaccination programs against HBV have reduced the HBV-associated HCC burden, the lack of a clear understanding of critical molecular players in pathogenesis has resulted in poor diagnosis and prevention of the disease at an advanced stage, thereby increasing the incidence and mortality associated with HCC [3]. Thus, it is necessary to understand how various molecular genetic and epigenetic factors operate during HCC pathogenesis and progression to identify reliable markers for the clinical management of this disease.

Noncoding RNAs (ncRNAs) are a broad class of regulatory transcripts that perform critical functions associated with gene expression regulation [4]. These RNAs operate at the transcriptional and posttranscriptional levels through various mechanisms, such as gene silencing, and epigenetic regulation that affect processes like DNA methylation and chromatin structure remodeling, and histone modifications [5, 6]. The most widely studied noncoding RNAs comprise a class of small-ncRNAs called microRNAs (miRNAs) with potential implications in disease pathogenesis, including cancers [7]. Aberrant expression of miRNAs can cause either gain or loss of function effects depending upon the site of expression and the type of cancer [8–10]. Additionally, dysregulated miRNA expression might affect the downstream transcriptome and promote pathways and processes involved in cancer development and progression [11, 12]. We and many others have previously reported that aberrant miRNA expression can promote HCC development [13–16]. Therefore, quantifying the miRNA expression profiles has diagnostic and prognostic potential for HCC. Interestingly, several miRNAs are coexpressed and coregulated in unison as clusters [17–20], and the role of epigenetic processes such as DNA methylation operating at critical regulatory genomic locations such as the promoters and enhancers of these clusters at large, might help understand the intricate regulatory processes that affect cluster expression and their biological processes and downstream pathways mediated by their target transcriptomes [21, 22].

The miR-379/656 cluster, also known as the chromosome-14 miRNA cluster (C14MC), is the second-largest miRNA cluster. Located on the 14q32.31 locus, alongside other imprinted coding genes (DLK1, DIO3, and RTL1), noncoding genes (MEG3 and MEG8), and small nucleolar RNAs, this cluster encodes approximately 50 different miRNAs that are regulated by a single promoter region [23, 24]. The expression of C14MC has been reported mainly in tissues of epithelial origin [25]. Aberrant C14MC expression is associated with multiple diseases and developmental disorders [26]. The abnormal expression of several C14MC members has been implicated as either of oncogenic or tumor suppressor functions in several cancers [23, 27–29]. In addition, studies have shown that promoter-bound alterations or changes in other internal regulators of the C14MC might be the underlying cause of disruption in C14MC expression [30].

Several studies have shown that miRNAs of this cluster are aberrantly expressed in HCC. Members like miR-656, miR-409, and miR-379 are downregulated in HCC. Specifically, miR-656 downregulation is known to promote HCC proliferation via SIRT5 upregulation, and upregulating miR-656 inhibited the invasion and migratory potential of HCC cells [31]. Another miRNA of the cluster, miR-379, which was reported to be underexpressed in HCC, was also shown to be associated with TNM stage and metastasis. Additionally, an ectopic expression of this miRNA suppressed HCC migration, invasion, and EMT through targeting the FAK/AKT signaling [32]. Another study pointed towards enhanced HCC invasion and metastasis due to the tumor suppressor function of miR-409 of C14MC, which resulted in BRF2 overexpression and Wnt/β-catenin activation [33]. However, the expression of the complete cluster in HCC and the role of epigenetic mechanisms such as DNA methylation in regulating the cluster are not understood. In this study, we demonstrated that C14MC acts as a tumor suppressor cluster and that promoter hypermethylation can be a key epigenetic driver of this downregulation. Furthermore, we predicted various oncogene targets which included PARP1, SPP1, CENPA, and RAD21, whose expressions are regulated by C14MC miRNAs. Furthermore, restoring C14MC miRNAs such as miR-299-5p and miR-376c-3p inhibited HCC cell migration and invasion. Additionally, we have also evaluated the diagnostic and prognostic potential of C14MC and its target genes, which can be used as biomarkers for HCC.

Methods

Cell culture and spheroid culture

We procured HCC cell lines from NCCS, Pune, India, through their standard procurement procedure (http://ncmr.nccs.res.in/home). The cell lines (HepG2: MEM, and Huh7: DMEM: F12 Ham’s) were maintained with 10% fetal bovine serum supplementation. For spheroid generation, HepG2 cells were used, and the tumuroids were generated via a modified forced suspension method as described previously [34]. The control RNA samples were procured from TakaraBio, Japan [34].

Promoter identification, cloning, and characterization

We identified the putative promoter of C14MC via the FANTOM5 tool [35]. The promoter region was then cloned and inserted into the pGL3-Basic plasmid between the XhoI and HindIII restriction sites to generate the pB-C14MC construct. The successful cloning of the promoter region into the pGL3-Basic vector was confirmed by double digesting the construct with XhoI and HindIII and performing diagnostic digestion with KpnI restriction enzymes. We then performed artificial methylation experiments using the pB-C14MC construct by incubating it with M.SssI methyltransferase enzyme (Thermo Fisher Scientific, USA). The artificially methylated and unmethylated constructs were then cotransfected with the SV40 plasmid using Fugene HD Reagent (Promega, USA). The reporter activity was then measured using a dual luciferase assay kit (Promega, USA) 48 h post-transfection using a GloMax 20/20 Luminometer.

Identification of partial methylation status of C14MC promoter in HCC cells

The partial methylation status of the identified promoter in HCC cells was assessed by performing bisulfite PCR, followed by sequencing the PCR products by Sanger sequencing. The MethPrimer tool was used to design the bisulfite-specific PCR primers [36]. To determine the methylation status, genomic DNA was isolated from HCC cells and modified using the EZ DNA Lightening Conversion Kit (Zymo Research, USA). The converted DNA was then amplified using the primers FP: GTTTATATTTGGGAATTAGTTATGT and RP: TCAAACACAATATATAAAAAAAATC in a thermocycler (Applied Biosystems, USA). The amplification conditions were as follows: 95 °C for 5 min; 34 cycles at 94 °C for 90 s, 54.5 °C for 3 min, 72 °C for 1 min, and 1 cycle at 72 °C for 5 min. The PCR products were then gel-purified using the E.Z.N.A. Gel Extraction kit (Omega Biotek, Georgia). The sequence trace files were then analysed using the BiQ Analyzer software to calculate the percent methylation at individual CpGs [37].

qRT-PCR

We performed first-strand cDNA synthesis of mRNA via the Verso cDNA Synthesis kit (Thermo Fischer Scientific, USA). For miRNA expression analysis, cDNA conversion was performed using the miRCURY LNA RT kit (Qiagen, Germany). qRT-PCR for miR-299-5p, miR-376c-3p, miR-377-3p, miR-376a-3p, and miR-656-3p was performed using miRCURY LNA SYBR Green PCR kit (Qiagen, Germany). qRT-PCR for PARP1, SPP1, RAD21, and CENPA was performed using the SYBR DyNAmo ColorFlash SYBR Green (Thermo Fischer, USA). Β-Actin and U6 snRNAs were used as internal controls. The experiments were performed using a Rotor-GeneQ system (Qiagen, Germany). The relative fold change in the expression was calculated via the 2^−∆∆CT method. The details of the primer used in this study are provided in Supplementary Table S1.

NanoString nCounter miRNA expression analysis

We performed the nCounter Human v3 miRNA Expression Assay (NS_H_miR_v3B) on an nCounter Analysis System (NanoString Technologies) according to the manufacturer’s guidelines to identify the differentially expressed miRNAs [38]. We manually screened for differentially expressed C14MC miRNAs among the 799 endogenous miRNAs present in the panel. Normalization was performed using geometric means of positive controls and the top 100 highly expressed miRNAs (-1.5 ≤ log₂(FC) ≥ + 1.5, p ≤ 0.05).

Whole transcriptomic analysis

RNA-seq was performed to identify the expression of the C14MC target genes in HCC tumouroids. We performed RNA-seq on an Illumina NovaSeq V1.5 instrument according to the manufacturer’s protocol. Briefly, ribosomal RNA depletion was carried out using Qiagen FastSelect rRNAHMR (Qiagen, Germany), and the mRNA libraries were prepared using the NEB Ultra II directional RNA-Seq Library Kit (NEB, USA). First-strand and second-strand cDNA synthesis were carried out using reverse transcriptase and DNA polymerase I enzymes, respectively, followed by adapter ligation. Prior to sequencing, the fragment sizes were analysed via HS NGS Fragment kit (1–6000 bp) (Agilent, USA) on a Fragment analyser. The mapping of the aligned paired reads was performed against the reference genome (GRCh38) using STAR (v.2.7.11b). The DETGs were identified using DESeq2 [39]. The fold change was calculated using the thresholds (-2 ≤ log₂(FC) ≥ + 2, p ≤ 0.05). Benjamini-Hochberg (BH) method was used for p-value correction (FDR < 0.05). The QC details of transcriptomic analysis are provided in Supplementary Table S2.

TCGA-LIHC cohort analysis

We analysed the expression of C14MC in The Cancer Genome Atlas-Liver Hepatocellular Carcinoma (TCGA-LIHC) as described previously in our study [16]. Grade-wise subclassification of tumor samples was further performed by matching the tumor grade and sample ID from TCGA-LIHC cohort (Supplementary Table S3). Furthermore, the diagnostic potential of C14MC expression was identified by calculating the sensitivity and specificity by plotting receiver operating characteristic (ROC) curves using the sum of expressions of differentially expressed miRNAs. The target genes of C14MC were identified using the MIENTURNET tool by querying the individual miRNAs in the miRTarBase [40]. The prognostic significance of C14MC and its target genes in LIHC was analysed using the DoSurvive tool [41]. Protein-protein interaction (PPIN) network analysis was performed using the STRING web-based tool. Further, the top ten hub gene networks were predicted using the MCC algorithm via the CytoHubba tool [42]. Furthermore, the gene ontology associations in terms of biological processes (BPs), molecular functions (MFs), pathway analyses, and disease associations were identified via the ShinyGO online tool [43]. Furthermore, the CpG methylation status in the TCGA-LIHC cohort and other independent datasets was analysed using the Epigenome-Wide Association Study (EWAS) data hub [44].

Cellular assays

Migration assay

We transfected Huh7 cells with miR-299-5p or miR-376c-3p mimics or All-Stars-Negative Control (Qiagen, Germany) according to the manufacturer’s instructions via the TrasIT-X2 Dynamic Delivery System (Mirus Bio, USA). Before transfection, the cells were subjected to serum starvation for 24 h. A wound was made in the center of the plate using a sterile micropipette tip. Posttransfection, complete medium containing 10% FBS was supplemented, and the cell migration into the wounded region was monitored using an EVOS XL Core imaging system (Thermo Fischer Scientific, USA).

2D- agarose spot invasion assay

We used a modified protocol of the 2D-agarose spot invasion assay from what was described earlier [45]. Briefly, Huh7 cells (3 × 10⁵ cells) were seeded into P35 cell culture dishes containing 0.5% agarose spots with 20% FBS in serum-free DMEM: F12 Ham’s. After 4 h of incubation, the media was removed and the cells were transfected with miRNA mimics (miR-299-5p or miR-376c-3p) or negative controls. The cells that invaded the agarose spots were imaged at 10X magnification and were manually scored from at least three microscopic fields to determine the extent of cell invasion.

Statistical analysis

The statistical significance was computed either by Student’s t-test or one-way ANOVA, and a p-value ≤ 0.05 was considered to indicate statistical significance. All the data are represented as means ± standard error of the means (SEMs). All the experiments were performed in biological replicates.

Results

HCC exhibits miR-379/656 cluster downregulation

The analysis of Nanostring nCounter data revealed the downregulation of different miRNAs belonging to C14MC. We observed that cluster miRNAs-miR-376c-3p and miR-382-5p were downregulated in HepG2 cells (Fig. 1A and B). Other cluster miRNAs-miR-299-5p, miR-376c-3p, miR-656-3p, miR-376a-3p and miR-377-3p were downregulated in Huh7 cells (Fig. 1C, D, E, F and G). We could not detect other C14MC miRNAs that were significantly downregulated in HCC cells. However, the TCGA-LIHC data confirmed the downregulation of approximately 30 C14MC miRNAs in primary HCC tissues compared with normal liver tissues (Fig. 1H). Furthermore, the combined sensitivity and specificity associations revealed an area under the curve (AUROC) of 0.8890 (95% CI, p < 0.0001) (Fig. 1I) for miRs-376c-3p and − 382-5p, and an AUROC of 0.8067 (95% CI, p < 0.0001) for miRs- 299-5p, miR-376c-3p, miR-656-3p, miR-376a-3p and miR-377-3p (Fig. 1J).

Fig. 1.

C14MC expression analysis in HCC. A-G: Bar graphs showing differential downregulation of C14MC member miRNAs in HepG2 and Huh7 HCC cell lines compared with control RNA. H: Heatmap showing differential downregulation of the DLK1-DIO3 gene MEG3 and C14MC miRNAs in primary HCC samples in comparison with adjacent normal liver tissues from the TCGA-LIHC cohort. I: and J: Combined ROC curves for downregulated C14MC observed in HepG2 and Huh7 cell lines plotted using corresponding TCGA-LIHC expression data. The sum of the expression of the respective C14MC miRNAs is highlighted. The error bars represent the means ± SEMs of duplicate experiments. Statistical significance was determined by an unpaired Student’s t-test

The subclassification of TCGA-LIHC samples into grade-wise classification showed that miR-376c-3p was differentially downregulated in both grade 1 and grade 2 samples in comparison with normal samples (Supplementary Fig. 1A). Furthermore, miR-376c-3p was downregulated in advanced grade 3 as well as grade 4 samples (Supplementary Fig. S1A). Furthermore, Receiver operating curve analysis showed that this miRNA could differentiate tumor samples based on grade-wise (G1, G2, G3 and G4) classification from normal samples (Supplementary Fig. 1B). Likewise, we also observed that miR-299-5p as downregulated in tumor grades 1, 2 and 3 in comparison with adjacent normal liver tissues (Supplementary Fig. 1C). However, no significant downregulation was observed in grade 4. Also, miR-299-5p could efficiently differentiate normal and HCC tissues of various grades based on its expression as demonstrated through ROC analyses (Fig. 1D). Grade-wise differential expression of miR-377-3p, miR-376a-3p, miR-656-3p, and miR-382-5p was not found to be statistically significant (Supplementary Fig. S2A, Fig. S2B, Fig S2C, Fig. S2D).

Characterization of C14MC cluster promoter

We predicted the possible promoter region of C14MC relative to its transcription start site (TSS) using the FANTOM5 tool. The annotated promoter, with coordinates (chr14:101,291,001–101,292,434), is located within the MEG3 region. The region presented several promoter-associated markers, including multiple cis-regulatory elements, a DNase I hypersensitivity region, H3K27Ac, and a CpG island (Fig. 2A). The associations of these genomic signatures suggested that the predicted region likely functions as the probable putative promoter of the cluster. The interactive map of the promoter region is shown in Fig. 2B. The successful cloning of the promoter was confirmed by isolating the plasmids from transformed E. coli-DH5α cells and subsequently performing double and diagnostic digestions (Fig. 2C and E). We then confirmed the promoter function of this 1,434 bp region via a dual luciferase assay. We observed a nearly 2-fold increase in luciferase activity in the pB-C14MC vector containing the promoter in comparison with the pGL3-Basic vector (Fig. 2F). We then assessed the effect of DNA methylation on promoter activity by artificially methylating the full-length pB-C14MC construct and measuring luciferase activity. Compared with the unmethylated construct, artificial DNA methylation significantly reduced promoter activity (Fig. 2G). Additionally, we identified different transcription factors (TFs) that might bind to the C14MC promoter region using the TRASFAC (https://genexplain.com). We detected that HNF4α, MEF2A, PBX3, c-Myc, NGFIβ, TBP, SMAD4, KLF4, PRDM1, TBX5, SOX6, TCF-7L2, NF-Y, and CEBPα binding sites within the C14MC promoter (Fig. 2H). DNA methyltransferases, primarily de novo (DNMT3A, DNMT3B, and DNMT3L) and maintenance methyltransferases (DNMT1, DNMT2), are the enzymes that catalyze DNA cytosine methylation [46]. We previously reported that DNMT1, DNMT3A, and DNMT3B are upregulated in HCC (16). These experiments suggest that the C14MC upstream region is likely the promoter region, whose function is regulated by methylation levels.

Fig. 2.

Identification and characterization of the C14MC promoter. A: Map of the putative promoter of C14MC (GRCh38 assembly) showing CAGE reads, CpG islands, and histone acetylation marks. B: Interactive map showing the promoter CpG distribution (hg19 assembly) and regions of the promoter characterized using methylation studies and luciferase assay-based characterization. C: Vector map of the recombinant clone pB-C14MC. D: LB agar plates showing transformed E. coli DH5α colonies transformed with recombinant pB-C14MC. E: Agarose gel electrophoresis results confirming the successful cloning of the full-length promoter via XhoI and HindIII double digestion (1434 bp) and orientation confirmation via KpnI digestion (801 bp). The reference ladder used was a 100 bp ladder. F: The relative fold change in luciferase activity, showing a 2-fold increase in HepG2 cells transfected with pB-C14MC, indicates that promoter activity is associated with the cloned region. G: Artificial methylation experiment results showing a significant reduction in promoter function upon M.SssI methyltransferase treatment and subsequent transfection into HepG2 cells, followed by the DLR assay. H: Possible transcription factors that can bind to the C14MC promoter in both forward and reverse genomic orientations. The dual luciferase assay was performed in biological triplicates, with the error bars representing the means ± SEMs. Statistical significance was determined by an unpaired Student's t-test.

HCC exhibits promoter-specific hypermethylation patterns

We further investigated the associations of the exact methylation status of promoter-bound CpGs of C14MC. Bisulfite PCR and sequencing targeting a 212 bp (-1121 to -909 bp) promoter region were performed to detect the partial promoter methylation status (Fig. 3A and B). We observed that CpG-12 was relatively less methylated (1–10%) in both HepG2 and Huh7 cells, while other CpGs-8, 9, 10, 13, 14, 15, 16, 17, and 18 were hypermethylated (90–100%) in HepG2 cells, and CpGs-10, 11, 14 and 17 were significantly hypermethylated in Huh7 cells (Fig. 3C). In the TCGA-LIHC dataset, the C14MC promoter CpGs-cg14245102, cg26374305, and cg04291079 were hypermethylated compared with those in adjacent normal tissues (Fig. 3D). However, we could not find any stage-wise associations of the methylation status of these three CpGs in LIHC (Fig. 3E). We further validated the hypermethylation status of the three promoter CpGs in an independent cohort from the Genome Expression Omnibus (GEO): GSE54503, comprising paired normal and HCC samples, where CpGs (cg14245102, cg26374305, and cg04291079) were hypermethylated in paired HCC tissues (Fig. 3F). We also observed in another independent cohort (GSE78732), the C14MC promoter-associated CpGs-cg10065153, cg10943497, cg14034270, cg02888166, cg15419911, cg11035687, and cg14121301 were hypermethylated in HCC. Additionally, these CpGs relative to the position of the island, located on the N-shore and the S-shelf, were relatively hypermethylated, whereas the island-bound CpGs had relatively stable methylation levels in HCC tissues compared with normal tissues (Fig. 3F).

Fig. 3.

DNA hypermethylation regulates C14MC expression in HCC cells and tissues. A: Agarose electrophoresis image under UV light showing C14MC bisulfite PCR amplicons of 212 bp from HepG2 and Huh7 cells in biological triplicate. B: Multiple sequence alignment results showing alignment of reads obtained from Sanger sequencing of a 212 bp bis-PCR product against the reference promoter sequence. The CpG alignments within the promoter region are highlighted in orange/ purple, respectively. C: Differential methylation of different CpG sites exhibiting hypermethylation in HCC cells. D: The differential methylation data from the TCGA-LIHC cohort show significant hypermethylation of promoter CpGs-cg14245102, cg26274305, and cg04291079. E: Stagewise methylation analysis of CpGs-cg14245102, cg26274305, and cg04291079 in the TCGA-LIHC cohort, which revealed no significant methylation changes. F: Results validating differential hypermethylation of the cg14245102, cg26274305, and cg04291079 CpGs in GSE5450, which included 10 paired normal and HCC tissues. G: Differential heatmap mapping of different C14MC promoter CpGs to genomic regions, including island, shore, shelf, and sea regions from EWAS datahub. Specific C14MC promoter-bound CpGs exhibiting a median increase in hypermethylation mapped to these regions are highlighted

Identification and validation of C14MC cluster target interactome in HCC cells

We identified the targets of downregulated C14MC candidate miRNAs in HepG2 cells using an in silico approach. Furthermore, these targets were validated in HepG2 cancer spheroids by RNA-Seq. The target prediction pipeline included only the genes that had an inverse correlation with individual C14MC miRNA expression, or only the target genes that were upregulated in HCC were considered for expression profiling. We observed that in HepG2 spheroids, miR-376c-3p targets; ANK1, ATAT1, ELF4, CENPA, and CHEK2 as the target oncogenes (Fig. 4A, B and C). The ontologies associated with these genes included biological processes (BP) such as DNA damage induced protein phosphorylation, NK-T cell proliferation, establishment of cell polarity, cytokinesis, microtubule organization; molecular functions (MF) such as insulin-activated receptor activity, tubulin N-acetyltransferase activity, cytoskeletal anchor activity, G-protein alpha subunit binding, PI3-K binding and protein serine/threonine/tyrosine kinase activity; and pathways: Insulin/IGF-mediated MAPK cascade, p53 signaling, and PDGF signalling pathways (Fig. 4D). Similarly, the miR-382-5p target genes-SERGEF, MTHDF2, SYNJ, UBB, DAD1, TM4SF, ATG10, NOL4L, SLC6A8, RPLP0, YBX1, and TRPV2 (Fig. 4E and F) were enriched in critical MFs such as methenyltetrahydrofolate cyclohydrolase/ dehydrogenase activity, and phosphate ion binding, phosphatase activity; and the key pathways identified were formyltetrahydrofolate biosynthesis, ACE2/ ACE4 signaling, and nicotinic acetylcholine receptor signaling cascades (Fig. 4G).

Fig. 4.

Validation of C14MC target gene expression in HepG2 spheroids. A: Micrographs showing HepG2 spheroids developed via a modified force-suspension method. B and C: Transcriptomic analysis showing the expression of 6 differentially overexpressed gene targets of miR-376c-3p. D: Gene ontology analysis showing significant biological processes, molecular functions, and pathways regulated by the miR-376c-3p target genes. E and F: Transcriptomic analysis showing the expression of 12 differentially overexpressed gene targets of miR-382-5p. G: Gene ontology analysis showing significant molecular functions and pathways associated with these 12 miR-382-5p target genes. The analysis of whole transcriptomic sequencing was performed by RNA isolated from at least two biological replicates of normal and HepG2 spheroid samples. DEGs identified were statistically significant if the fold change was -2≤log2(FC)≥+2, p≤ 0.05, and Benjamini-Hochberg correction (FDR≤ 0.05)

Overexpression of miR-379/656 cluster miRNAs abrogates the expression of target oncogenes in Huh7 cells

The posttranscriptional network associated with C14MC targets in Huh7 cells was identified using an in silico approach and further validated by overexpressing specific miRNAs and measuring the target gene expression by qRT-PCR. We identified SPP1, IGFBP3, RAD21, CENPA, PARP1, LMNB1, SERPINE1, ESR1, CXCL2, and CDT1 as top ten hub genes of candidate C14MC (miR-299-5p, miR-376c-3p, miR-376a-3p, miR-656-3p, and miR-377-3p) (Fig. 5A), which regulate critical gene ontologies such as chromosome organization, protein localization, and cell communication regulation (BP) (Fig. 5B) and p53 signaling, NFκβ signaling, apoptosis, apelin signaling and cellular senescence were the key enriched pathways (Fig. 5C). Moreover, the genes PARP1, RAD21, SPP1 (miR-299-5p targets), and CENPA (miR-376c-3p target) were directly associated with liver inflammation and HCC (Fig. 5D) and were upregulated in HCC tissues than normal in TCGA-LIHC (Fig. 5E, F and H). Hence, we profiled the expression of these targets to determine whether their expression is mitigated upon miRNA overexpression. We observed a significant reduction in the mRNA expression levels of all four oncogenes, namely, PARP1, SPP1, RAD21, and CENPA, upon overexpressing miR-299-5p or miR-376c-33p in Huh7 cells compared with those in cells transfected with the negative control (Fig. 5I, J, K and L), suggesting the potential reversal of target oncogene expression upon restoring the expression of component C14MC miRNAs.

Fig. 5.

Restoring C14MC miRNAs mitigates the expression of target oncogenes in Huh7 cells. A: Hub genes targeted by miR-299-5p, miR-376c-3p, miR-376a-3p, miR-377-3p, and miR-656-3p in Huh7 cells. B: Enrichment analysis of the hub genes showing the top biological processes: kinetochore organization, chromosome organization, and cell death. C: Chord diagram showing key pathways regulated by C14MC hub genes. D: Disease associations showing C14MC hub genes enriched in different diseases. PARP1, RAD21, CENPA, and SPP1 were enriched in liver inflammation and HCC. E: - H: Differential expression of the C14MC target transcripts-PARP1, SPP1, RAD21, and CENPA in the TCGA-LIHC cohort was significantly greater in HCC tissues (blue) than in adjacent normal liver tissues (red). I-K: Bar graphs showing the fold change in expression upon transfection of miRNA mimics to Huh7 cells. Ectopic expression of C14MC miRNA-miR-299-5p reduced the expression of the target genes PARP1, SPP1, and RAD21. L: Ectopic expression of C14MC miRNA-miR-376c-3p reduced the expression of CENPA, compared with that in cells transfected with the negative control. All the qRT-PCR experiments were performed on biological triplicates. Statistical significance was determined by an unpaired Student's t-test (Statistical significance, p≤ 0.05)

Overexpressing miR-299-5p and miR-376c-3p mimics inhibits the migration and invasion of Huh7 cells

We investigated the effects of C14MC expression restoration on cell migration and invasion, which are prominent hallmarks directly linked to HCC progression and metastasis. We observed that Huh7 cells transfected with either miR-299-5p or miR-376c-3p mimics presented reduced cellular migration (Fig. 6A, B and C) compared with that of the cells transfected with the negative control. Similarly, the overexpression of C14MC miRNAs-miR-299-5p and miR-376c-3p, inhibited the invasion of Huh7 cells into soft-agar spots (Fig. 6D, E and F). These findings suggest that restoring C14MC expression might mitigate HCC hallmarks such as migration and invasion.

Fig. 6.

Restoring candidate C14MC miRNA expression inhibits Huh7 migration and invasion. A: Representative micrographs showing inhibition of Huh7 cell migration and reduced wound healing upon transfection with miR-299-5p or miR-376c-3p mimics compared with those of cells transfected with negative control. B: Bar graph showing a reduction in the percentage of Huh7 cells migrating upon mimic transfection compared with the negative control. C: Bar graph showing the percentage of wounds remaining upon mimic transfection compared with the negative control. D: Representative micrographs showing a reduction in the invasion of Huh7 cells into the 2D-agarose spot after transfection with miR-299-5p or miR-376c-3p mimics compared with that of cells transfected with negative control post 24 h of transfection. E: Number of cells invading the 2D-agarose spot when transfected with miRNA mimics in comparison to cells transfected with the negative control post 24 h and 48 h of transfection. F: Area of the 2D-agarose spot invaded when transfected with miRNA mimics in comparison to that of cells transfected with the negative control post 24 h and 48 h of transfection. The cellular assays were performed in triplicate, and the values obtained were calculated for statistical significance (p < 0.05) using the one-way ANOVA test

miR-379/656 cluster and its target network are critical to predicting overall survival in HCC patients

We also assessed the clinical utility of C14MC candidate miRNAs and their targets, namely, miR-299-5p, miR-376c-3p, PARP1, SPP1, RAD21, and CENPA. KM survival analysis revealed that the overall survival of HCC patients was significantly correlated with SPP1 (Fig. 7A) and CENPA expression (Fig. 7B). We observed that patients with lower expression levels of SPP1 [HR = 2.24, 95% CI (1.57–3.19), log-rank p = 5.2e-06] and CENPA [HR = 2.3, 95% CI (1.63–3.26), log-rank p = 1.1e-06] had better median survival. Additionally, we investigated whether other cofounding clinical parameters, such as HCC stage, grade, patient gender, patient race, chemotherapy administration via sorafenib, alcohol consumption, and viral hepatitis (HBV), might affect the survival outcomes (Table 1). Furthermore, multivariate-Cox regression analysis was performed to assess the combined effect of C14MC and the validated gene targets. We observed that HCC stage and CENPA were significantly associated with overall survival (OS), disease-free survival (DFS), and progression-free survival (PFS). Other specific miRNAs or target genes associated with individual survival outcomes included SPP1, which was associated with OS; SPP1 and miR-376c-3p, which were associated with DFS (Fig. 7C, D and E).

Fig. 7.

Clinical utility of C14MC miRNAs and gene targets in HCC. A: and B: KM plots of SPP1 and CENPA showing that these genes are significantly associated with the overall survival of HCC patients. Multivariate Cox regression analyses for miR-299-5p, miR-376c-3p, and their target genes PARP1, SPP1, CENPA, and RAD21 revealed that C: HCC stage and CENPA and SPP1 expressions are significantly associated with overall survival. D: HCC stage, CENPA, SPP1, and miR-376c-3p expression are significantly associated with disease-free survival. E: HCC stage and CENPA expression are significantly associated with progression-free survival. Statistical analysis for KM analysis was significant if log-rank p ≤ 0.05, and for multivariate analysis p ≤ 0.05 (CI: 95%)

Table 1.

List of various factors that are significantly associated with overall survival based on C14MC target gene expression among HCC patients. (Survival analysis was performed using survival data and transcript expression data from the TCGA-LIHC cohort)

	Hazard Ratio (HR), 95% C.I.	Log-rank p value (p < 0.05)
PARP1
Stage
1 & 2	1.31 (0.77–2.22)	0.31
3 & 4	2.51 (1.22–5.71)	0.01
Grade
1	2.34 (0.76–7.14)	0.13
2	1.83 (1.07–3.16)	0.026
3	1.48 (0.78–2.80)	0.23
Gender
Male	1.77 (1.08–2.92)	0.023
Female	0.61 (0.88–2.93)	0.099
Race
White	1.7 (1.03–2.8)	0.035
Asian	1.61 (0.88–2.93)	0.12
Treatment
Sorafenib	3.28 (0.71–15.2)	0.11
Alcohol consumption
Yes	1.97 (1.03–3.8)	0.038
No	0.73 (0.46–1.18)	0.2
Viral Hepatitis
HBV +	1.97 (1.03–3.8)	0.038
HBV -	1.56 (0.8–3.04)	0.19
SPP1
Stage
1 & 2	2.31 (1.41–3.78)	0.00059
3 & 4	2.15 (1.18–3.9)	0.01
Grade
1	2.1 (0.81–5.45)	0.12
2	2.07 (1.21–3.53)	0.0066
3	3.44 (1.88–6.29)	2e-05
Gender
Male	2.66 (1.69–4.17)	1.1e-05
Female	1.99 (1.12–3.52)	0.016
Race
White	1.9 (1.12–3.24)	0.016
Asian	3.3 (1.78–6.11)	5.7e-05
Treatment
Sorafenib	2.07 (0.69–6.21)	0.19
Alcohol consumption
Yes	3.03 (1.38–6.63)	0.0037
No	2.38 (1.27–4.44)	0.0051
Viral Hepatitis
HBV +	3.16 (1.65–6.05)	0.00024
HBV -	2.51 (1.54–4.08)	0.00014
RAD21
Stage
1 & 2	0.82 (0.51–1.33)	0.42
3 & 4	2.35 (1.28–4.34)	0.0047
Grade
1	2.41 (0.92–6.28)	0.065
2	1.28 (0.75–2.2)	0.37
3	1.95 (1.04–3.65)	0.034
Gender
Male	1.59 (0.98–2.57)	0.059
Female	1.21 (0.69–2.1)	0.5
Race
White	1.32 (0.83–2.11)	0.24
Asian	2.35 (1.29–4.28)	0.0038
Treatment
Sorafenib	2.29 (0.71–7.32)	0.15
Alcohol consumption
Yes	1.52 (0.78–2.95)	0.21
No	1.4 (0.84–2.36)	0.2
Viral Hepatitis
HBV +	0.59 (0.3–1.19)	0.14
HBV -	2.23 (1.34–3.7)	0.0016
CENPA
Stage
1 & 2	2.55 (1.57–4.14)	8.5e-07
3 & 4	2.18 (1.22–3.89)	0.0069
Grade
1	4.75 (1.73–13.02)	0.0011
2	2.22 (1.32–3.73)	0.002
3	3.01 (1.61–5.63)	0.00031
Gender
Male	3.94 (2.03–7.66)	1.3e-05
Female	2.71 (1.48–4.96)	0.00078
Race
White	1.8 (1.14–2.84)	0.011
Asian	5.65 (3.02–10.57)	1.3e-09
Treatment
Sorafenib	3.91 × 108 (0-∞)	0.013
Alcohol consumption
Yes	2.95 (1.34–6.46)	0.0048
No	2.65 (1.66–4.24)	2.4e-05
Viral Hepatitis
HBV +	1.84 (0.95–3.54)	0.066
HBV -	2.59 (1.64–4.1)	2.6e-05
Stage
1 & 2 (n=257)
3 & 4 (n=90)
Grade
1 (n=55)
2 (n=177)
3 (n=122)
Gender
Male (n=250)
Female (n=121)
Race
White (n=194)
Asian (n=158)
Treatment
Sorafenib (n=30)
Alcohol consumption
Yes (n=107)
No (n=205)
Viral hepatitis
HBV + (n=153)
HBV – (n=169)

Stage: 1 & 2 (n=257), 3 & 4 (n=90); Grade: 1 (n=55), 2 (n=177), 3 (n=122); Gender: Male (n=250), Female: (n=121); Race: White (n=194), Asian (n=158); Treatment: Sorafenib (n=30); Alcohol consumption: Yes (n=107), No (n=205); Viral hepatitis: HBV + (n=153), HBV – (n=169)

Note: Correlation of factors with statistical significance p < 0.05 are indicated in boldface

Discussion

Hepatocellular carcinoma (HCC) is the most common and lethal form of primary liver cancer (PLC). The current diagnostic HCC markers and treatment options available for HCC include a six-month ultrasound scan coupled with or without serum alpha-fetoprotein (AFP) measurements and sorafenib administration respectively [47–49]. Despite improvements in terms of HCC screening, the overall survival of HCC patients remains poor [50, 51]. Novel epi/transcriptomic markers have great potential in terms of cancer diagnosis and therapy [52]. The C14MC or miR-379/656 cluster is the second-largest polycistronic miRNA cluster in humans and encodes nearly 50 different miRNAs [23]. The aberrant expression of this cluster is implicated in the tumorigenesis of several cancers [28, 53, 54]. However, the functional regulation and role of this cluster in HCC pathophysiology have not been established. The present study aimed to understand the mechanistic regulation of C14MC, especially through epigenetic mechanisms such as DNA methylation, and its functions in HCC using in vitro models of HCC and various clinical datasets. This is the first study to delineate the regulatory mechanism by which DNA methylation affects C14MC expression in HCC. We show that (i) C14MC miRNAs are downregulated in HCC cells and tissues, suggesting a tumor-suppressor role of this cluster in HCC. (ii) Promoter CpG hypermethylation suppresses C14MC in HCC cells and tissues. (iii) C14MC miRNAs such as miR-299-5p and miR-376c-3p regulate critical oncogene expression posttranscriptionally, and miRNA reactivation can abrogate PARP1, SPP1, RAD21, and CENPA oncogene expression. (iv) Reactivation of C14MC miRNAs directly inhibits cancer hallmarks such as cell migration and invasion. (v) Several of these miRNAs and gene targets are critical for predicting HCC patient survival outcomes. In summary, we report that C14MC downregulation is tightly regulated by DNA methylation, which drives the overexpression of several oncogenes targeted by this cluster. Thus, the C14MC methylome and its target transcriptome may be critical for HCC management, and designing novel RNA-based therapeutics targeting this region can help mitigate HCC.

The C14MC or miR-379/656 cluster is frequently dysregulated in various cancers. Depending on the specific cancer type, this cluster is known to have both oncogenic and tumor suppressor functions, which point toward diverse biological roles associated with cancer development and progression. C14MC downregulation is often reported in breast cancer, glioblastoma, oligodendroma, and cervical cancer [21, 27, 28, 54]. In contrast, oncogenic functions associated with the C14MC have been identified in a few cancers [55]. We showed that C14MC members were significantly downregulated in HCC cell lines and tissue samples from the LIHC cohort. These findings suggest that the cluster has a broader tumor suppressor function in HCC.

DNA methylation is perhaps the most significant epigenetic alteration that affects posttranscriptional gene regulation [22]. Hypermethylation at upstream regulatory elements, such as the promoter region, can suppress downstream target genes [30]. Abnormal promoter methylation might result in the development of various pathophysiological conditions beyond cancer [56]. Hypermethylation at C14MC upstream regulatory elements is implicated in carcinoma of the cervix [54], glioblastomas [27], and oligodendromas [28]. However, hypomethylation of the C14MC regulatory locus has been reported in lung adenocarcinoma, temple syndrome, and conditions such as atherosclerosis [56, 57]. We were interested in understanding the promoter methylation of C14MC in HCC cells and tissues. We identified the putative promoter and confirmed its activity by cloning it into a luciferase plasmid and subsequently performed a luciferase assay, which confirmed the promoter function associated with the genomic region. Furthermore, we observed that upon artificial methylation, there was a substantial reduction in luciferase activity, which suggested that promoter function is tightly linked to methylation levels and that methylation can reduce C14MC promoter function. A mutual exclusiveness exists between promoter methylation and transcription factor binding, which can affect gene expression regulation [46]. Through in silico analysis, the different transcription factors that might bind to the C14MC were identified. Notably, the HNF-4α binding site was identified within the C14MC promoter. Interestingly, HNF-4α is downregulated in HCC, and overexpressing HNF-4α can reverse HCC malignancy by regulating the genes associated with the DLK1-DIO3 locus, including the miRNAs belonging to this cluster, like miR-134, which are known to suppress the KRAS oncogene [58]. Furthermore, the data generated from bisulfite-Sanger sequencing and CpG methylome data of patient samples from different patient cohorts, including the TCGA-LIHC cohort, suggested prominent hypermethylation of specific CpG sites within the promoter of C14MC. These results point to an intricate regulatory mechanism mediated by DNA methylation associated with C14MC expression.

miRNAs are critical for regulating the posttranscriptional gene regulatory networks. Aberrant miRNA expression can affect downstream target gene expression [12]. The involvement of such a dysregulated miRNA-mRNA axis has been well studied in different cancers [18]. Notably, C14MC dysregulation has been shown to directly target PDK1, a critical gene in PI3-Akt signaling in carcinoma of the cervix [54]. We identified and validated the expression of posttranscriptional regulatory networks targeted by C14MC miRNAs in specific hub genes of HCC. PARP1 overexpression in HCC is often associated with resistance to sorafenib-induced cell death and radiotherapy [59–61]. RAD21 is a reliable prognostic marker in HCC patients, and its expression is generally associated with poorer differentiation status and tumor size [62–64]. SPP1 overexpression in HCC is known to drive HCC cell proliferation and tumor growth [65], and higher SPP1 expression levels are associated with tumor macrophage infiltration and can determine survival outcomes among HCC patients [66]. We revealed that PARP1, RAD21, and SPP1 are direct targets of miR-299-5p in C14MC, and that overexpressing miR-299-5p suppressed the expression of these oncogenes. Similarly, the expression of CENPA, a prominent cell cycle regulatory gene with established roles in regulating key mitotic processes such as mitotic protein assembly and chromosomal segregation [67, 68], was abrogated in HCC cells upon miR-376c-3p overexpression in our study.

Cancer cells differ from noncancerous cells in terms of their biological abilities [69]. The emergence of altered cellular epigenetic reprogramming during the plasticity of cancerous cells is considered a hallmark of cancer [70]. We investigated the ability of C14MC restoration to inhibit HCC cell proliferation and invasion through cellular assays. Dysregulated cell proliferation may promote HCC advancement and progression [71]. The ectopic expression of C14MC miRNAs-miR-299-5p and miR-376c-3p, in HCC cells inhibited HCC cell migration and invasion. Previously, it was reported that miR-299-5p suppresses the migratory ability and invasiveness of papillary thyroid carcinoma [72] and breast cancer [73]. Furthermore, overexpressing another C14MC member, miR-376c-3p, is known to inhibit cancer cell proliferation, migration, and invasion in oral squamous cell carcinoma (OSCC) [74], malignant gastric cancer [75], and medullary thyroid carcinoma [76] cells. Furthermore, our pathway enrichment of the hub genes revealed that targets of C14MC, such as PARP1 and SERPINE1, are involved in regulating apoptosis and cellular senescence, which are critical to determining the cellular proliferation, migration, and invasion capabilities. Thus, we show that reactivation of C14MC or its component miRNAs may inhibit cell migration and invasion and thereby might contribute to mitigating HCC progression.

Finally, we identified the clinical significance of C14MC in HCC. Several studies have shown that C14MC members can be useful as diagnostic and prognostic markers in various cancers [12]. However, the clinical utility of the region in HCC remains elusive. We have previously shown that individual miRNAs of this cluster can be used as potential diagnostic markers for HCC [16]. In this study, we report the use of two panels of C14MC miRNAs, namely, miR-382-5p and miR-376c-3p, and miR-299-5p, miR-376c-3p, miR-656-3p, miR-376a-3p and miR-377-3p, which can be used as potential diagnostic panels for HCC. Additionally, we have shown that the C14MC family and their gene targets can be employed to predict survival outcomes in HCC patients (OS, DFS, and PFS). Interestingly, we observed that the expression of several of these targets was significantly associated with sex, race, HCC grade, cancer stage, treatment strategy, and other HCC-related risk factors, such as alcohol consumption and the prevalence of viral hepatitis. Taken together, these observations show that C14MC and its target gene expression might not only help predict survival outcomes in HCC patients, but also help design strategies that can help improve the survival of the patients, depending on the clinical parameters of the disease, such as histopathological outcomes and risk factors associated with HCC.

Conclusions

We reported that members of C14MC are downregulated in HCC and that the miRNA cluster acts as a tumor suppressor in HCC. We also showed that the underlying epigenetic cause behind the loss of tumor suppressor function in HCC is promoter hypermethylation. A systems biology approach was used to identify the C14MC transcriptome, which was subsequently validated experimentally. Additionally, the key ontologies and pathways regulated by the transcriptomic network were identified, and the key cancer hallmarks of HCC, including cell migration and invasion upon cluster miRNA activation, were experimentally validated. We also showed that downregulation of the cluster is significantly correlated with disease prognosis and can be useful for developing strategies in clinical settings for HCC management. Although we discuss the role of methylation in C14MC regulation and the functions of some of the miRNAs of C14MC, additional studies targeting the entire C14MC and establishing C14MC-knockout in vitro and in vivo models of HCC can help characterize the actual transformation ability of the cluster and study its biological functions during HCC carcinogenesis.

Abbreviations

HCC

Hepatocellular carcinoma

C14MC

Chromosome 14 miRNA cluster

SPP1

Osteopontin

RAD21

Double-strand break repair protein homolog 21

PARP1

Poly-(ADP)-ribose polymerase 1

CENPA

Centromere protein A

OS

Overall Survival

DFS

Disease-free survival

PFS

Progression-free survival

Authors’ contributions

SHK designed the overall study and performed the experiments related to promoter cloning and validation, Bis-PCR, qRT-PCR, transfection experiments, cellular assays, cohort validation and bioinformatic analyses. SHK wrote the original manuscript and prepared the figures for the manuscript. GR, SPK, AP, PMV, and RM provided critical inputs to data discussions, and provided critical reviews for the manuscript. PKS designed, supervised the entire project, obtained funding support, and contributed to the manuscript with critical reviews.

Funding

Open access funding provided by JSS Academy of Higher Education and Research, Mysore. We acknowledge the financial support for this study from the JSS Academy of Higher Education and Research-Institutional Research Grant (JSSAHER/REG/RES/URG/54/2011-12) provided to Prasanna Kumar Santhekadur. The authors thank the Vision Group on Science and Technology (VGST), Government of Karnataka and Department of Science and Technology - Fund for Improvement of Science and Technology Infrastructure (DST-FIST), Government of India, for the infrastructure support provided to the Center of Excellence in Molecular Biology and Regenerative Medicine, where this work was carried out in part or whole. Shreyas Hulusemane Karunakara thanks the Lady Tata Memorial Trust, Mumbai, for the fellowship and contingency support for his doctoral studies.

Data availability

The data discussed in this study are enclosed in the article or are provided in the supplementary information. The transcriptomic data discussed in the study is deposited in the GEO portal and can be accessed using the submission ID GSE328880.

Declarations

Ethics approval and consent to participate

The study did not include any direct inclusion of animal or human specimens. All the experiments were performed using the established cell line models freely available for research purpose. The source of these cells are disclosed in the manuscript. All the patient data is taken from TCGA public cohort for clinical validation.

Consent for publication

Not applicable.

Competing interests

GR is an employee of Theraues Innovations Pvt Ltd, Bengaluru, and is a part of collaboration for this study. The company or the employees hold no reservations whatsoever in terms of the data generated from the study.