This article is mentioned in:

Introduction

RNA modifications have emerged as a new layer of epigenetic regulation and are anticipated to further deepen our understanding of molecular complexity and to optimize therapeutic inventions for patients. Currently, more than 170 RNA modifications, including N6-methyladenosine (m6A), N1-methyladenosine (m1A), and 5-methylcytidine (m5C), have been identified in nearly all classes of coding and non-coding RNAs (1). Of these modifications, m6A has been comprehensively characterized; it regulates gene expression and various cellular processes, including RNA folding, stability, splicing, transport, and translation. Accumulating evidence suggests that RNA modifications mediate various biological processes, and aberrant RNA modification has been linked to numerous human diseases such as cancer and cognitive dysfunctions (2-5).

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer and the fourth leading cause of cancer-related deaths worldwide (6). Similar to the evolution of other cancers, hepatocarcinogenesis is a multistep process involving genetic and epigenetic alterations (7-12). In terms of RNA modulations in HCC, only a few prevalent mRNA regulators have been examined. For example, m6A writer methyltransferase-like 3 (METTL3) has been reported as an unfavorable prognostic factor in HCC, and its overexpression contributes to HCC progression by reducing cytokine signaling 2 (SOCS2) messenger RNA (mRNA) stability in an m6A-dependent manner (13). Intriguingly, METTL3 has been recruited to the transcriptional start sites, and the promoter-bound METTL3 induces co-transcriptionally modified m6A within the coding region of the transcript (14). Another m6A writer member, METTL14, is associated with HCC metastasis by regulating the processing of miR-126 (15). Incidentally, miR-126 is downregulated in various cancers and is known to play a role in carcinogenesis in various cancers, including in HCC (16). The m6A reader protein YTHDF2 has been shown to suppress cancer cell proliferation by destabilizing epidermal growth factor receptor (EGFR) mRNA in HCC (17).

To date, only a few studies have been conducted on METTL6, a tRNA methylation enzyme, with limited knowledge about its role and mechanism of action in cancer. One study showed that METTL6 was upregulated in highly proliferative luminal breast cancer (18) and another study revealed that knockout of METTL6 altered cisplatin sensitivity in lung cancer cells (19). Very recently, Ignatova et al reported that loss of METTL6 impaired the pluripotency of mouse stem cells, and that this molecular was a vital regulator of HCC cell proliferation (20). Another recently published paper established a prognostic risk score based on five metabolism-related genes, including that of METTL6, indicating high-risk patients had a poor overall survival (OS) compared to low-risk patients with HCC (21).

Although METTL6 has been identified in various human cancers, the characterization of this tRNA modification and its mechanisms of action in cancer remain unclear. Hence, the aim of this study was to investigate the functional roles of METTL6 in cancer development, and to study its potential mechanisms of action. In the present study, we demonstrated that tumorigenicity in HCC was due to the upregulation of METTL6. Furthermore, the depletion of METTL6 mitigated HCC cell proliferation, migration, invasion, and attachment ability, possibly through the regulation of cell adhesion-related genes post-transcriptionally. In addition, METTL6 was found to be localized in the cytosol. Our findings may provide insights into the functional roles of tRNA and post-transcriptional regulation in cancer, and may lead to the development of a therapeutic strategy for HCC.

Materials and methods

Bioinformatics analysis

We retrieved multi-omics data from RNA-sequencing (RNA-seq), DNA promoter methylation, DNA copy number, and clinical information from The Cancer Genome Atlas (TCGA) using cBioPortal (https://www.cbio-portal.org) corresponding to a total of 372 HCC samples. The differential expression analysis between HCC and non-HCC tissues and across four different tumor stages, as well as OS and disease-free survival (DFS) analyses were performed using the Gene Expression Profiling Interactive Analysis (GEPIA) webserver (http://gepia.cancer-pku.cn/). In addition, GraphPad Prism 8 software (GraphPad Software, Inc.) was used to analyze the association between DNA copy number alterations (CNAs) and mRNA expression. A one-way ANOVA test was performed to compare mRNA expression levels between groups of different CNAs. Statistical significance was set at P<0.05.

Plasmid DNA constructs

The lentiviral packaging plasmids pMD2.G and psPAX2 were obtained from Addgene (#12259 and #12260, Addgene, USA, respectively). To generate inducible Cas9 nuclease-expressing cell lines, we purchased the Edit-R inducible lentiviral plasmid (#CAS11229, Dharmacon, UK). Five individual sgRNAs targeting the METTL6 gene used in this study were as follows (5′→3′ orientation): sgRNA1: 5′-GGA GCT AAG ATC ATG TAG AG-3′, sgRNA2: 5′-ATA TGA TAC AGA AAG ATG CA-3′, sgRNA3: 5′-GTT TCA TAG GTA TTA AAA CC-3′, sgRNA4: 5′-ATA ACA ACA TCC ACA GAC TC-3′ and sgRNA5: 5′-ACA GCA GAA ATT GGA ACA AG-3′. sgRNAs were designed and cloned into the pLKO.1-puro U6 sgRNA BfuAI large stuffer plasmid (#52628, Addgene). All plasmids were verified using Sanger sequencing.

Cell lines and culture conditions

The human HCC cell lines SNU-423 and SNU-475 were purchased from the American Type Culture Collection (ATCC), and tested and authenticated using DNA profiling for polymorphic short tandem repeat (STR) markers analyzed by BEX Co. Ltd. using GenePrint 10 systems (Promega Corp.) before starting the project (Table SI). The human liver cancer cell lines HepG2, Huh-7, and Li-7 were purchased from RIKEN Bioresource Research Center (RIKEN BRC, Japan), and tested and authenticated using DNA profiling for polymorphic short tandem repeat (STR) markers analyzed by Applied Biosystems using GeneMapper ID software before starting the project (Table SI). SNU-423, SNU-475, Huh-7, and Li-7 cells were cultured as monolayers in RPMI-1640 supplemented with 10% FBS and antibiotics, and HepG2 cells were cultured as monolayers in MEM supplemented with 10% FBS, 0.1 mM NEAA and antibiotics. All cells were maintained at 37°C in humid air with 5% CO2. In the present study, SNU423 and SNU475 cells were mainly used as HCC cell lines for various experiments. The reason for this is as follows. i) HepG2 was originally considered a hepatocellular carcinoma cell line, but it has been shown to be a hepatoblastoma (22). In addition, HepG2 cells were established from a younger boy (15 years of age). Therefore, we decided not to use this line in further studies. ii) As for Huh-7 cells, previous literature has indicated that Huh-7 cells may produce adhesion-related factors and growth factors autonomously (23). Therefore, we considered Huh-7 cells as inappropriate because we examined the role of adhesion-related genes with and without METTL6 expression. iii) Regarding Li-7 cells, a previous study showed that of five hepatocellular carcinoma cells, only Li-7 cells showed a population change after two months of culture (24). Then, we decided not to use this line in further studies. iv) SNU-423 is an adult hepatocellular carcinoma that was generated from a 40-year-old man. SNU-475 is another adult hepatocellular carcinoma cell line generated from a 43-year-old man. In addition, we previously constructed a knockout system for both SNU-423 and SNU475 cells using CRISPR/Cas9, and confirmed that the target genes were properly knocked out (25). Hence, we decided to use these two cell lines mainly in this study.

CRISPR/Cas9 gene editing system

A lentivirus transduction system was used to generate CRISPR/Cas9 knockout cells. To produce lentiviruses, viral vectors and packaging plasmids were co-transfected into 293T cells using Lipofectamine 3000 (#L3000-008, Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. After 48 h, the cell culture medium containing lentiviruses was collected and filtered through a 0.45-µm filter. Target cell lines were plated in 6-well plates and cultured with lentivirus-containing medium for 3 days, which was carried out in the absence of polybrene. Stable cell clones were selected using blasticidin S (10 µg/ml) (#029-18701, FUJIFILM Wako Pure Chemical Corporation, Japan). For inducible CRISPR/Cas9 knockout experiments, conditional Cas9 expression cells further underwent lentivirus transduction of conditional sgRNA expression plasmid and selection in the presence of both blasticidin S (10 µg/ml) and puromycin (2 µg/ml) (#ant-pr-1, InvivoGen). Knockout of METTL6 was induced by adding doxycycline (Dox) (1 µg/ml) (#D9891, Sigma-Aldrich; Merck KGaA) for 48 h.

Western blot analysis

SNU-423 and SNU-475 cells were lysed with CelLytic™ Reagent (#C2978, Sigma-Aldrich; Merck KGaA) using standard methods. Protein lysates (15 µg) were separated on a 10% SDS-PAGE gel and transferred to nitrocellulose membranes. The membranes were blocked with 5% skimmed non-fat milk for 2 h at 25°C, and then the membranes were incubated with rabbit polyclonal anti-METTL6 primary antibody (#HPA035166, Sigma-Aldrich; Merck KGaA; dilution used in WB: 1:2,000); rabbit polyclonal anti-METTL2 primary antibody (#PA5-113304, Invitrogen/Thermo Fisher Scientific, Inc.; dilution used in WB: 1:2,000); rabbit poly-clonal anti-METTL8 primary antibody (#ab122273, Abcam; dilution used in WB: 1:2,000); rabbit polyclonal anti-ITGA1 primary antibody (#22146-1-AP, Proteintech; dilution used in WB: 1:2,000); chicken polyclonal anti-SPON1 primary antibody (#ab14271, Abcam; dilution used in WB: 1:1,000); goat polyclonal anti-CLDN14 primary antibody (#ab19035, Abcam; dilution used in WB: 1:1,000); and mouse monoclonal anti-α-tubulin (DM1A, EMD Millipore; dilution used in WB: 1:1,000) antibodies at 4°C overnight. After primary antibody incubation, the membranes were incubated with an HRP-linked donkey anti-rabbit secondary antibody (#NA934V, GE Healthcare, USA; dilution used in WB: 1:5,000) and mouse anti-goat IgG-HRP (#sc-2354, Santa Cruz Biotechnology, Inc.; dilution used in WB: 1:5,000) at 25°C for 1 h. The signal was detected using the ECL system (#RPN2236, GE Healthcare, USA), and NIH ImageJ software (version 1.53) (National Institutes of Health) was used for quantification (26).

Colony formation assay

Cells were seeded in 6-well plates at a density of 1,000 cells/well and cultured at 37°C. The culture medium with or without Dox was replaced every 3 days. After 14 days, cells were fixed with 1% formaldehyde (#252549, Sigma-Aldrich; Merck KGaA) and 1% methanol (#137-01823, FUJIFILM Wako Pure Chemical Corporation), stained with 0.05% crystal violet (#V5265, Sigma-Aldrich; Merck KGaA) for 20 min, and then washed three times with Milli-Q water. Colony numbers were quantified using the NIH ImageJ software (26).

Cell proliferation assay

Cell viability was measured using the Cell Counting Kit-8 (CCK-8) (#343-07623, Dojindo Laboratory, Japan). Cells were seeded in 96-well plates at a density of 1.0×103 cells/well and cultured for 24, 48, 72 and 96 h. Then, 10 µl of CCK-8 solution was added and the plates were incubated at 37°C for 2 h. The absorbance of each well was measured at 450 nm using a microplate reader (Multiskan FC, Thermo Fisher Scientific, Inc.).

Cell scratch assay

Cells were seeded in 6-well plates at a density of 3×105 cells/well and incubated for 24 h. When the cellular confluence reached 90%, a 200-µl pipette tip was used to create wounds in the monolayer cells. The wells were then rinsed with PBS three times to remove any detached cells, and fresh serum-free medium was added. The wells were then placed in an incubator, and visualized using a phase-contrast microscope (Olympus CKX53, Japan) at 0 and 48 h. The percentage of wound area was calculated using the ImageJ software.

Cell invasion assay

For the cell invasion assay, a 96-well Boyden chamber coated with basement membrane extract (BME) was used (#ab235697, Abcam). Prior to the assay, the cells were starved for 24 h in serum-free media. Then, cells were suspended in serum-free medium and seeded into the upper chamber (8.0-µm pore size) at a density of 50,000 cells per well. Further, 200 µl of RPMI medium containing 10% serum was added to the lower chamber. The cells were allowed to invade for 24 h at 37°C. A standard curve was constructed for each cell type. After incubation, cells were washed carefully, and those that did not invade the lower side of the chamber were removed from the top side. Then, 100 µl of the cell invasion dye and cell dissociation solution were added to each well in the lower chamber. The chamber was incubated at 37°C for 1 h, and the bottom wells were read at excitation and emission wavelengths of 530/590 nm using a Synergy H1 Hybrid Multi-Mode Reader (BioTrek).

Cell adhesion assay

Cells were seeded in 6-well plates at a density of 0.5×105 cells/well (SNU-423) and 1.0×105 cells/well (SNU-475). The cells were then incubated for 15, 30, 45, and 60 min. Next, the wells were washed with PBS three times to remove any detached cells, and images were obtained using a phase-contrast microscope (Olympus CKX53, Japan) at 15, 30, 45, and 60 min. Cell numbers were calculated using ImageJ software.

RNA-seq and data analysis

Total RNA was extracted using the QIAzol lysis reagent (#79306, Qiagen). Extracted RNA was treated with DNase I (#043-31261, FUJIFILM Wako Pure Chemical Corporation) at 37°C for 30 min, followed by acid phenol:chloroform extraction (#AM9722, Thermo Fisher Scientific, Inc.). Purified RNA was re-suspended in Takara's buffer for mRNA amplification using 5′ template switching PCR with Takara's SMART-Seq v4 Ultra Low Input RNA Kit (both from Takara). The amplified cDNA was fragmented and appended with dual-indexed barcodes using Illumina Nextera XT DNA Library Prep Kits (Illumina, Inc.). Libraries were validated by electrophoresis, pooled, and sequenced on an Illumina NovaSeq 6000 (150 base pairs, paired ends) (Illumina, Inc.). RNA-seq data were mapped to the hg38 version of the human genome using the DRAGEN Bio-IT Platform (v3.6.3) (Illumina, Inc.). Raw counts were converted into transcripts per million reads. Genes with q≤0.05 and |log2_ratio|≥1 were identified as differentially expressed genes (DEGs) after the knockout of METTL6 in SNU-423_KO2, SNU-423_KO5, SNU-475_KO3, and SNU-475_KO4. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of the DEGs were performed using Metascape (27) with P<0.05, indicating statistically significant enrichment. Reads from all sequencing experiments were deposited into the DNA Data Bank of Japan (DDBJ) with the accession number DRA012940.

Chromatin immunoprecipitation sequencing and data analysis

Chromatin immunoprecipitation sequencing (ChIP-seq) analyses were performed using the SimpleChIP Enzymatic Chromatin IP Kit (#9003, Cell Signaling Technology, Inc.) with minor modifications. Briefly, cells cultured in 10 cm2 dishes were fixed in 1% formaldehyde for 10 min, and fixation was quenched with the addition of glycine to 125 mM for an additional 5 min. Cells were harvested by scraping from the plates and stored at −80°C. During the solubilization of chromatin, fixed nuclei were sonicated with an E220 focused ultrasonicator (Covaris, USA). Immunoprecipitation antibodies were rabbit polyclonal anti-H3K27ac (#ab4729, Abcam) and rabbit polyclonal anti-CTCF (#3418s, Cell Signaling Technology, Inc.). Approximately 5 µg of chromatin was incubated with the indicated antibody overnight at 4°C on a rotator. Then, FG Beads HM Protein G (#TAB8848N3173, Tamagawa Seiki, Japan) was added to the solution and washed with buffer, as described previously (28). Immunoprecipitated chromatin was eluted and reverse-crosslinked according to the manufacturer's instructions (#9003, Cell Signaling Technology, Inc.). Immunoprecipitated DNA was purified using a QIAquick PCR Purification Kit (#28106, Qiagen). DNA libraries were prepared using the QIAseq Ultralow Input Library Kit (#180492, Qiagen) for Illumina. Library quality was checked using a TapeStation 4200 instrument (Agilent Technologies, Inc.). DNA libraries were sequenced using the NovaSeq 6000 system (Illumina). Sequencing reads from ChIP-seq experiments were mapped to the hg38 version of the human genome using Bowtie (v2.2.9) and parameters (local) (29). Duplicate reads were removed using the Samtools (v1.3.1). The normalized ChIP-seq signals were visualized using the Integrative Genomics Viewer (IGV; v2.3.91) (30). Reads from all sequencing experiments were deposited into the DNA Data Bank of Japan (DDBJ) with the accession number DRA012940.

Immunofluorescence analysis

Cells were seeded at a density of 1×104 cells/well (SNU-423_KO2) with and without Dox treatment in 2-well chamber culture slides (#154852, Lab-Tek™ II CC2™ Chamber Slide System, Thermo Fisher Scientific, Inc.). After 4 days, the cells were rinsed with ice-cold PBS and fixed with 4% paraformaldehyde (#163-20145, Paraformaldehyde Phosphate Buffer Solution, FUJIFILM Wako Pure Chemical Corporation) for 15 min at 25°C. Cells were then blocked with 3% BSA (#A-9647, bovine serum albumin, Sigma-Aldrich; Merck KGaA) and 0.3% Triton X-100 [#160-24751, Polyoxyethylene(10) Octylphenyl Ether, FUJIFILM Wako Pure Chemical Corporation] for 60 min. The cells were subjected to immunofluorescence staining with rabbit polyclonal METTL6 primary antibody (#HPA035166, Sigma-Aldrich; Merck KGaA; dilution used: 1:200) overnight at 4°C. Next, the cells were washed with cold PBS three times for 5 min each time, and incubated with donkey anti-rabbit IgG (H+L) secondary antibody Alexa Fluor 594 (#A21207, Thermo Fisher Scientific, Inc., dilution used: 1:500) at 25°C for 1 h. The cells were washed with cold PBS and mounted using mounting medium containing DAPI (#H-1200, Vector Laboratories). The cells were examined using fluorescence microscopy (Celldiscover 7, ZEISS) at 20× with 5×5 tiling.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software, Inc.). The cut-off value for METTL6 mRNA expression was determined at the median level. Differences in METTL6 mRNA expression levels between HCC samples and control tissues, as well as expression levels at different stages, were calculated using one-way analysis of variance (ANOVA). Survival curves were obtained using Kaplan-Meier curves and log-rank tests. The association of METTL6 mRNA expression levels with DNA CNAs was assessed using one-way ANOVA. All quantitative results, including relative expression analysis by western blot analysis, cell colony assay, cell proliferation assay, cell scratch assay, and cell attachment assay, are presented as means ± standard deviation (SD) of at least three separate experiments. A two-tailed unpaired t-test was used to compare the mean values between the two groups. Results were considered significant when the P-value was *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001, or NS=non-significant (as indicated with these symbols/initials in the figures).

Results

METTL6 is elevated in HCC tissues and is associated with poor prognosis

First, we established a list of 60 known RNA regulators, including methyltransferases, demethylases, and acetyltransferases. We then analyzed their mRNA expression levels in HCC tumor samples and adjacent non-tumor tissues and examined whether they might be correlated with HCC patient survival rates based on the Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) (31) and OncoLnc platforms based on TCGA independently (32). To assess the novelty of these regulators, we conducted a literature survey of HCC-related studies for each candidate (Table SII and Fig. S1). Notably, these results suggest that METTL6, which was recently identified as a methylcytidine (m3C) methyltransferase on tRNA (33) might be a promising novel target among all candidates. mRNA expression of METTL6 was significantly upregulated in HCC tumor tissues compared to that in adjacent non-tumor tissues (Fig. 1A). Meanwhile, a gradual increase in the mRNA level of METTL6 was observed from tumor stages I to III, while a slight decrease was observed in stage IV (Fig. 1B). This suggests that METTL6 is highly expressed in the early stages of cancer and involved in the development of cancer, and that there is no significant correlation with the degree of cancer progression. Kaplan-Meier analysis revealed that patients with higher METTL6 expression levels had significantly poorer OS and DFS (Fig. 1C and D), suggesting that METTL6 could be a potential prognostic indicator for patients with HCC. In addition, we analyzed the TCGA_LIHC (liver HCC) dataset, which contains CNAs and DNA methylation levels of METTL6. The METTL6 copy number was positively correlated with its mRNA levels (Fig. 1E), indicating that the CNAs of METTL6 might be involved in HCC through alterations in METTL6 gene expression levels.

Figure 1 Bioinformatics analysis of METTL6 expression in HCC samples using GEPIA. (A) METTL6 mRNA expression levels in TCGA_LIHC tumor tissues (n=369) and paired TCGA non-tumor tissues (n=50) using GEPIA. (B) METTL6 mRNA expression in different stages of HCC (F value=4.75; Pr [>F]=0.00295). (C) Overall survival rate and (D) disease-free survival rate in patients with HCC with high and low METTL6 mRNA levels. The group cut-off value is based on the median. (E) METTL6 copy number alterations (CNAs) and METTL6 mRNA levels (P<1.0e-04, ANOVA test). All data are presented as means±standard deviation. *P<0.05. METTL6, methyltransferase-like 6; GEPIA, Gene Expression Profiling Interactive Analysis; HCC, hepatocellular carcinoma; TCGA, The Cancer Genome Atlas.

Knockout of METTL6 inhibits HCC colony formation, cell proliferation, and cell migration in vitro

To investigate the biological functions of METTL6 in HCC and to assess the effect of METTL6 expression, we first assessed its protein expression levels across five different HCC cell lines (Fig. S2A). We established stable METTL6 knockout models in SNU-423 and SNU-475 cells using the CRISPR/Cas9 gene editing system, with four independent single-guided RNAs targeting the promoter regions of METTL6. Knockout of METTL6 was confirmed using western blot analysis (Fig. 2A). No noticeable changes were observed in the protein expression levels of METTL2 and METTL8, which are the other m3C methyltransferases that exhibit a functional similarity and are closely related to METTL6, as analyzed by a family tree diagram (33,34). Our bioinformatics analysis indicated that METTL6 was associated with prognosis; therefore, we first performed multiple cancer phenotypic assays using established knockout cell lines. A colony formation assay was performed to detect the effect of METTL6 expression on the clonogenicity of HCC cells. As anticipated, the METTL6 knockout group had significantly fewer and smaller colonies than the METTL6 expressing groups in both SNU-423 and SNU-475 cell lines, indicating that METTL6 expressing cells had the ability to form colonies (Fig. 2B).

Figure 2 Loss of METTL6 inhibits HCC cell colony formation and proliferation in vitro. METTL6 knockout SNU-423 cells and SNU-475 cells were generated by using the CRISPR/Cas9 system with five different guide RNAs and were selected by treatment with doxycycline (Dox+). (A) Western blot analysis of METTL6 protein expression level in SNU-423 and SNU-475 cells with (Dox+) and without (Dox-) Dox treatment. (B) Colony formation assays were conducted to determine the effect of METTL6 on colony formation in the infected HCC cells. Representative images are shown and quantified by ImageJ. (C) Cell proliferation assays are performed to determine the cell viability of infected HCC cells by using the CCK-8 kit. All data are presented as mean ± standard deviation of the least three independent experiments. *P<0.05; **P<0.01; ***P<0.001; NS, non-significant; METTL6, methyltransferase-like 6; HCC, hepatocellular carcinoma.

Next, the role of METTL6 in HCC cells was analyzed using the CCK-8 to examine cell proliferation. We found that knockout of METTL6 remarkably inhibited HCC cell viability, suggesting that METTL6 played a role in cell survival (Fig. 2C). Additionally, we investigated the effect of METTL6 knockout on HCC cell migration. Cell scratch assays showed that HCC cell migration was significantly suppressed at 48 h post-scratch in the METTL6 knockout cells (Fig. 3). Collectively, these results indicate that specific tRNA modifications regulated by METTL6 enhance HCC cell colony formation, proliferation, and migration abilities.

Figure 3 Cell scratch assays determine cell migration of the infected HCC cells. Images are acquired at the start of the experiment (0 h) and 48 h later (48 h) using ×4 magnification. Representative images at 0 and 48 h are shown and quantified using ImageJ software. Scale bar, 200 µm. Yellow lines indicate the areas without migrating cells. All data are presented as means ± standard deviation of at least three independent experiments. *P<0.05; **P<0.01; ***P<0.001; NS, non-significant; HCC, hepatocellular carcinoma. Knockout of METTL6 was induced by adding doxycycline (Dox+).

Loss of METTL6 decreases cell adhesion-related gene expression levels

To investigate the molecular mechanisms of METTL6 action in HCC, we performed RNA-seq on four METTL6 stable knockout cells (SNU-423_KO2, SNU-423_KO5, SNU-475_KO3, and SNU-475_KO4). A heat map and dendrogram summary of the RNA-seq data are shown in Fig. 4A. In total, we identified 1,365 and 1,309 DEGs in SNU-423 and SNU-475 cells, respectively (Fig. 4B). GO analysis using Metascape highlighted that these DEGs were involved in cell-cell adhesion via plasma-membrane adhesion molecules, extracellular matrix, cell migration and invasion, cell apoptotic process, xenobiotic metabolism, and other metabolic processes (Fig. 4C). This indicates that METTL6 may play an important role in cancer cell biology. In addition, there was an overlap of 32 upregulated and 51 downregulated genes between SNU-423 and SNU-475 cells in the different knockout groups (Fig. 4B). Among the downregulated genes, we selected a group of cell adhesion proteins for further investigation, including integrin α-1/β-1 (ITGA1), claudin-14 molecules (CLDN14), and spondin1 (SPON1). ITGA1 mediates the distinct interactions between cells and those in the surrounding extracellular matrix (35); CLDN14 forms tight junctions and interacts with other CAMs (36) while SPON1 is another cell adhesion protein that maintains the extracellular matrix (37). In addition, these cell adhesion genes have been associated with tumor progression and metastasis in HCC (38-40). Immunoblotting revealed that the knockout of METTL6 resulted in a decrease in ITGA1 in SNU-423 cell lines and CLDN14 in SNU-423 and SNU-475 cell lines (Fig. S2B). Unfortunately, SPON1 protein expression was not detected in either SNU-423 or SNU-475. RNA-seq results showed that the average TPM value of ITGA1 in SNU-475 cell lines was 0.351 while it was 1.61 in SNU-423, which could be one of the reasons why the protein expression was relatively weak in SNU-475 cell lines with non-significant decrease after the knockout of METTL6. In addition, the average TPM value of SPON1 among SNU-423 and SNU-475 cell lines was 0.13, which indicated that gene expression levels of SPON1 were the lowest compared with other proteins. This might explain why SPON1 expression levels were not detected in these cell lines. Our ChIP-seq analysis demonstrated that enhancer activity was not significantly different between control cells and METTL6 KO3 in SNU-475 cells (Fig. S2C). The chromatin binding activity of the transcription factor CTCF also showed a similar binding affinity between the control cells and METTL6 KO3 cell lines (Fig. S2C). These results suggest that METTL6 does not contribute to the transcriptional activity but stabilizes the mRNA expression of cell adhesion genes such as ITGA1 and CLDN14 and increases their protein expression, probably via post-transcriptional tRNA modifications mediated by METTL6.

Figure 4 RNA-seq analysis using METTL6 knockout in SNU-423 and SNU-475 cells. (A) Heat Map of RNA-seq results in SNU-423_KO2, SNU-423_KO5, SNU-475_KO3 and SNU-475_KO4 samples. (B) Venn diagram of differentially expressed genes (DEGs) in SNU-423_KO2, SNU-423_KO5, SNU-475_KO3 and SNU-475_KO4 cells. (C) Gene Ontology and Pathway Enrichment Analysis of consistently downregulated genes in two different METTL6 knockouts within SNU-423 cells and SNU-475 cells. METTL6, methyltransferase-like 6.

To determine whether METTL6 depletion attenuated cell adhesion ability and cell invasiveness by altering the expression levels of adhesion molecules, we performed a cell adhesion assay, which is often used to evaluate the metastatic ability of cancer cells. The assay results demonstrated that knockout of METTL6 dramatically decreased the number of attached cells after 60 min in SNU-423 and 45 min in SNU-475 cells (Fig. 5A). Furthermore, we conducted a cell invasion assay and found that the cell invasion ability was significantly suppressed in METTL6 KO cells (Fig. 5B) compared to that in control cells. Taken together, RNA-seq analysis, cell adhesion assay and cell invasion assay results reveal that METTL6 exerts oncogenic effects by altering the expression of CAMs.

Figure 5 Loss of METTL6 inhibits HCC cell adhesion and cell invasion in vitro. (A) Cell adhesion assays were performed to determine the cell adhesion ability of the infected HCC cells. Images are captured at 15, 30, 45, and 60 min after the treated cells were loaded in the culture dishes using ×4 magnification. Representative images are shown and quantified using ImageJ software. Scale bar, 200 µm. (B) Cell invasion assays were performed to determine the cell invasiveness of the infected HCC cells. The invasion rate was determined by counting the cells that invaded to the lower chamber compared to the cells added to the top chamber (50,000 cells/well). All data are presented as means ± standard deviation of at least three independent experiments. **P<0.01; ***P<0.001. HCC, hepatocellular carcinoma. METTL6, methyltransferase-like 6; HCC, hepatocellular carcinoma. Knockout of METTL6 was induced by adding doxycycline (Dox+).

METTL6 is localized in the cytoplasm

Our ChIP-seq results suggested that METTL6 was directly associated with post-transcriptional regulation; however, some mRNA expression levels were upregulated and downregulated in METTL6 KO cells. Therefore, we investigated the subcellular localization of METTL6. For this, we conducted immunofluorescence assays and observed an obvious signal reduction in the cytoplasm after the knockout of METTL6 in SNU-423 and SNU-475 cell lines, indicating that METTL6 was mainly localized in the cytosol (Fig. 6). To the best of our knowledge, this is the first study to confirm that METTL6 is localized in the cytoplasm. In line with our findings, previous studies have shown that METTL6 catalyzes the formation of m3C at C32 (33,41), and another study claimed that m3C32 modification must occur in the cytoplasm (42).

Figure 6 Immunofluorescence experiments demonstrate that METTL6 protein is located in the cytoplasm. Images were captured using ×20 magnification. Representative images are shown, which were captured using Celldiscover. Scale bar, 50 µm. METTL6, methyltransferase-like 6. Knockout of METTL6 was induced by adding doxycycline (Dox+).

Discussion

The field of post-transcriptional regulation of gene expression at the RNA level, and dynamic and reversible modifications of almost all forms of coding and non-coding RNAs, including tRNA, have attracted increasing attention. tRNA is an essential component of protein synthesis. In cancer, deregulation of tRNA can elevate oncogenic protein translation and fulfill the energy demand required for cancerous growth. A recent large-scale analysis revealed an overall overexpression of tRNA levels and amplification of tRNA modification enzymes across 31 cancer types, which indicated the dynamic role of tRNA and tRNA modification proteins in cancer initiation and progression (43). Despite the fact that tRNA modifications are involved in various cancers, the detailed characterization of this process and mechanisms of such regulation are still lacking. This could be attributed to the difficulty of tRNA quantification due to its abundance, and the intricate interactions between tRNA and tumor metabolism (44). METTL6 is a tRNA methylation enzyme that is present in many cancer types; however, we have limited knowledge about its functions and mechanisms of action in cancer. Earlier, the enzymatic activity of METTL6 was identified by Xu et al (33). The knockout of m3C tRNA-modified enzyme METTL2 showed a 35% reduction, whereas knockout of METTL6 accounted for 12% reduction in the total RNA compared to that in the wild-type (WT) tissues. Moreover, METTL6 is a homolog of Trm141 in S. pombe, targeting tRNASer; therefore, they identified a novel target of tRNA and found that m3C32 of tRNASer(AGA) and tRNASer(GCU) were modified by METTL6.

RNA modifications of mRNA are spatiotemporally regulated. For example, METTL14 binds to histone H3 trimethylation at lysine 36 (H3K36me3), and consequently m6A modification of mRNA is deposited co-transcriptionally (45). Intriguingly, the METTL3/METTL14 complex co-transcriptionally regulates histone modification. In the presence of METTL3/METTL14, m6A modified RNA is recognized by the m6A reader YTHDC1, which recruits H3K9me2 demethylase KDM3B, and promotes gene expression by demethylating H3K9me2 (46). Similar to m6A, m5C and m5C methyltransferases (RCMTs) contribute to chromatin structure regulation (47). In our data, H3K27ac peak levels were identical between METTL6 expressing and METTL6 depleted cells, suggesting that METTL6 was not associated with transcriptional regulation. This finding was further supported by immunocytochemistry in CRISPR/Cas9-mediated knockout cells, which showed that METTL6 was localized in the cytosol.

As mentioned earlier, more than 170 RNA modifications have been reported, 90 of which are found in tRNAs (1). To date, 14 modification sites have been identified in tRNA (48). It has been reported that modifications of tRNAs affect the stability of tRNAs and thus the genetic code of mRNAs can be read accurately (49). Modifications outside the anticodon region are crucial for tRNA stability and modulating temperature-sensitive growth (50,51). Additional studies have shown that mRNA stability is affected by translation, particularly related to codon stabilization, which is regulated by tRNA anticodon I34 modification by adenosine deaminases (ADATs) (52-54).

The biogenesis and degradation of tRNAs are well summarized by Motorin and Helm (55). Deregulation of early, intermediate, and late modifications affected tRNA stability and degradation. Destabilization of 3D-structured tRNA triggered by aberrations in RNA modifications involves either the TRAMP pathway (nuclear tRNA degradation) or rapid tRNA decay (cytoplasmic tRNA degradation). Widespread tRNA modification regulates tRNA stability, and hypomodified tRNA is degraded even in bacteria and archaea (56). Therefore, it is reasonable to assume that METTTL6 may regulate tRNA stability in SNU-423 and SNU-475 cell lines and possibly affect mRNA expression levels, followed by alterations in protein expression.

A tRNA half, which binds with other transcripts and regulates gene expression as miRNA, may account for a possible explanation of how METTL6 mediates mRNA expression levels. tRNA halves are generated by dicers, angiogenins, or other exonucleases (57,58). Intriguingly, RNA modifications play important roles in the origin and function of tRNA fragments, and tRNA modifications promote and protect the expression of tRNA halves (59). Moreover, tRNA-derived fragments regulate ribosomal biogenesis by targeting ribosomal proteins encoding two mRNAs in a mouse model of HCC (60).

In the present study, we focused on METTL6 and explored its oncogenic activities in HCC, thereby providing a novel perspective on its underlying mechanism in cancer progression by altering the expression levels of membrane proteins. Here, for the first time, we performed a comprehensive bioinformatics analysis of RNA modification enzymes and associated proteins in patients with HCC using a liver cancer dataset and observed that METTL6 was upregulated in HCC tissues compared to the adjacent non-tumor tissue samples. High expression levels of METTL6 were associated with high histological grade and poor survival rates, suggesting that METTL6 was a potential unfavorable prognostic biomarker in HCC. To further investigate the functional role of METTL6 in HCC, we knocked out METTL6 through the CRISPR/Cas9 system in vitro using five independent gRNAs, and demonstrated that the depletion of METTL6 significantly suppressed HCC cell growth, colony formation, cell migration, cell invasion and cell adhesion. In addition, we used RNA-seq and ChIP-seq to elucidate the mechanism of METTL6-driven regulation of HCC. In addition to GO and pathway enrichment analysis, we found that knockout of METTL6 significantly reduced the expression levels of cell adhesion proteins, including ITGA1, SPON1, and CLDN14, which was further validated at the protein level.

Previous studies have demonstrated that the abovementioned cell adhesion proteins are of crucial importance in HCC and other cancers. Liu et al reported that silencing of ITGA1 inhibited HCC cell migration and invasion, while upregulation of ITGA1 enhanced HCC migration and invasion ability in vitro (38). In addition, overexpression of ITGA1 has been indicated as a key driver of HCC lymph gland metastasis (61). Gharibi et al found that ITGA was increased in pancreatic cancer and associated with poor prognosis, and that ITGA1 was necessary for TGFβ/collagen-induced epithelial to mesenchymal transition and metastasis (62). Another study suggested that ITGA1 could promote colorectal cancer cell migration, invasion, and tumorigenicity by activating Ras/Erk signaling (63). Interestingly, our RNA-seq results also showed a significant decrease in the activity of the Ras signaling pathway along with ITGA1, such as T-cell lymphoma invasion and metastasis-inducing protein 1 (TIAM1) and GRB2-associated-binding protein 2 (GAB2), while there was an increase in the expression of kinase suppressor of Ras 2 (KSR2), which is a Ras/Erk signaling suppressor, after the knockout of METTL6. Dai et al demonstrated that downregulation of SPON1 inhibited HCC cell proliferation, migration, and invasion (39). Of note, high CLDN14 expression was associated with good prognosis in HCC (P=0.047) analyzed using GEPIA. However, there are some contradictory findings regarding the prognostic role of CLDN14 in cancer. One study showed that CLDN14 was a favorable prognostic biomarker in HCC (40), and that CLDN14 was a direct target of EZH2-mediated H3K27me3 and regulated by the Wnt/β-catenin pathway (40). RNA-seq analysis in our study indicated that EZH2 expression was not altered. The expression levels of Wnt family proteins (2B, 3, 5A, 5B, 6, 7B, 10A, and 10B) were altered (downregulated, no difference, and upregulated). Tryndyak et al reported that cytosine DNA hypermethylation induced the downregulation of CLDN14 gene expression in differentiated HepaRG cells treated with NaAsO2 (64). Another study revealed that the CLDN14 expression level was negatively correlated with acute myeloid leukemia (AML) survival rates (65). In breast cancer (BC), CLDN14 gene expression was upregulated in BC tissues compared with non-cancer tissues, and a high gene expression of CLDN14 was associated with a poor overall survival (66). These previous reports suggest that CLDN14 gene expression, as well as the relationship between gene expression and patient outcome, are dependent on various cellular contexts.

The primary challenges of focusing on METTL6 or tRNA modifications are the lack of m3C recognizing antibody for RNA immunoprecipitation sequencing to globally address specific m3C tRNA modification as well as the lack of established tRNA-focused next-generation sequencing techniques, which means that solid analyzing pipelines are currently missing. Hence, we need to continue developing biochemical analytical tools and delve further into this field for a comprehensive understanding of the role of tRNA modifications.

One limitation of our study is the difficulty in revealing how METTL6 interacts with identified cell adhesion proteins and regulates their role in cancer cell proliferation, migration, invasion and adhesion. Therefore, further detailed research and validation such as rescue experiments are required. The other option is to perform pulldown experiments to pulldown m3C modified RNA and conduct a high-throughput analysis (known as MeRIP-seq; methylated RNA immunoprecipitation) to detect m3C modified tRNA using METTL6-expressing cells (Dox-) and METTL6-KO cells (Dox+). This may provide a more direct indication of whether global or specific tRNA modifications by METTL6 are associated with target gene expression. For validation, control tRNAs and mutant tRNAs at methylated positions should be transfected and confirmed by experimentally changing the expression levels of the methylated tRNAs. In addition, it is needed to conduct mass spectrometry analysis in advance to determine the precise methylated position on the RNAs of interest. However, there are some technical concerns such as the lack of antibodies to recognize m3C (specificity, sensitivity, and precipitation efficacy), and/or MeRIP platform for tRNA. Another limitation of current study is that we only investigated the role of METTL6 in HCC using cell lines, but did not verify our findings in vivo. The xenograft mouse models will be established in future studies to support the findings of the present study.

Supplementary Data

Availability of data and materials

The datasets used and/or analyzed in the present study are available from the corresponding author upon reasonable request.

Authors' contributions

AB, KA, SK, and RH designed the study. AB, KA, SK, KS, and NI performed the experiments. AB, KA, SK, and HM analyzed the data and confirmed the accuracy. AB and KA wrote the manuscript. MK, SS, and RH reviewed, edited, revised critically and proposed feedback constructive suggestions to improve the quality of the manuscript. All authors read and approved the manuscript and agree to be accountable for all aspects of the research in ensuring that the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Acknowledgments

The authors express their gratitude to the past and present members of the Hamamoto Laboratory. AB is grateful to the Otsuka Toshimi Scholarship Foundation for their support during her PhD study.

Funding

This study was supported by JST CREST (grant no. JPMJCR1689), JSPS Grant-in-Aid for Scientific Research on Innovative Areas (grant no. JP18H04908), and JSPS KAKENHI (grant no. JP20K17982).

References