An Alternatively Spliced Variant of METTL3 Mediates Tumor Suppression in Hepatocellular Carcinoma

Many post-transcriptional mRNA processing steps play crucial roles in tumorigenesis and the progression of cancers, such as N6-methyladenosine (m6A) modification and alternative splicing. Upregulation of methyltransferase-like 3 (METTL3), the catalytic core of the m6A methyltransferase complex, increases m6A levels and results in significant effects on the progression of hepatocellular carcinoma (HCC). However, alternative splicing of METTL3 has not been fully investigated, and the functions of its splice variants remain unclear. Here, we analyzed both our and online transcriptomic data, obtaining 13 splice variants of METTL3 in addition to canonical full-length METTL3-A in HCC cell lines and tissues. Validated by RT–qPCR and Western blotting, we found that METTL3-D, one of the splice variants expressing a truncated METTL3 protein, exhibits higher levels than METTL3-A in normal human livers but lower levels than METTL3-A in HCC tumor tissues and cell lines. Further functional assays demonstrated that METTL3-D expression decreased cellular m6A modification, inhibited the proliferation, migration, and invasion of HCC cells, and was negatively associated with the malignancy of patient tumors, exhibiting functions opposite to those of full-length METTL3-A. This study demonstrates that the METTL3-D splice variant is a tumor suppressor that could potentially be used as a target for HCC therapy.


Introduction
RNA plays a central role in gene expression, serving as a vital link for genetic information passing from DNA to protein [1]. The maturation of messenger RNA precursors requires removing introns and ligating exons, processes catalyzed by the spliceosome and named pre-mRNA splicing [2,3]. In humans, more than 95% of nascent transcripts have been alternatively spliced, generating multiple mRNA isoforms (splice variants) from one gene [4]. Splice variants are different in their coding sequences and/or untranslated regions (UTRs), in which approximately 60-75% are within the coding regions that provide protein diversities [2,5]. Alternative splicing (AS)-affected tumor biology includes metabolism, apoptosis, cell cycle control, invasion, and metastasis [6,7]. Splice variants of many oncogenes are associated with the occurrence and development of cancers, such as breast cancer, is conserved in humans and mice, and the METTL3-D variant is a functionally distinct isoform compared with the canonical full-length METTL3-A. The mRNA and protein levels of METTL3-D were significantly decreased in HCC tumor tissues. The proliferation, migration and invasion of HCC cells were inhibited when METTL3-D was overexpressed. METTL3-A and METTL3-D oppositely controlled the expression of METTL3 downstream target genes. Finally, high levels of METTL3-D were significantly associated with early HCC tumor stages and longer overall survival times, suggesting that METTL3-D is a tumor suppressor in HCC.

Cell Culture
The human liver cancer cell lines HepG2, MHCC97-L, and HCCLM3 were acquired from American Type Culture Collection, Manassas, VA, USA. All cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM; Invitrogen, CA, USA) supplemented with 10% fetal bovine serum (FBS; Invitrogen, CA, USA) and antibiotics (100 IU/mL penicillin and 0.1 mg/mL streptomycin, Sigma-Aldrich, St Louis, MO, USA) at 37 • C in an incubator with 5% CO 2 .

Human Specimens
HCC patient samples from the First Affiliated Hospital of Anhui Medical University (Hefei, China) were stored in a −80 • C freezer after collection. This study was approved by the Ethics Committee of Anhui Medical University. The clinical specimens were used in accordance with the Declaration of Helsinki. The clinical characteristics of all the patients are listed in Table S1.

PCR and Sequencing
cDNAs of METTL3 splice variants were amplified using specifically designed primers listed in Table S2. Sanger sequencing was performed by Qing-ke Company, and the sequences were aligned by Chromas (Technelysium Pty Ltd., Tewantin, Australia). Nextgeneration RNA sequencing was performed on an Illumina platform with 150 bp paired-end reads. Raw reads were filtered by FastQC v0.11.8 and TrimGalore v0.6.2, and the clean reads were then mapped to the human genome (GRCh38) using STAR v2.6.1a [39].

Quantitative Real-Time PCR
Total RNA was extracted from HCC patients and HCC cells using TRIzol reagent (Invitrogen, CA, USA) and then stored at −80 • C until use. Reverse transcription was performed using SuperScript III Reverse Transcriptase (Life Technologies, Carlsbad, CA, USA). Quantitative PCR was performed using an Applied Biosystems StepOnePlus Real-Time PCR System with SYBR Premix Ex TaqTM II (TaKaRa, Otsu, Shiga, Japan), and the data were analyzed using the 2-∆∆Cq method [40] and normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

Bioinformatics and Online Resource
The Ensembl gene browser database [41] was used to query specific METTL3 variants. Sequences of cDNAs and amino acids were also downloaded from Ensembl. The levels of METTL3 splice variants in human tissues were obtained from the Genotype-Tissue Expression (GTEx) project and analyzed using GTExPortal (https://www.gtexportal.org/ home/) (accessed on 5 November 2020).The sequencing data of HepG2 cells from the PacBio platform in ENCODE (ENCSR834DQL, https://www.encodeproject.org/files/ENCFF470 UHX/) (accessed on 13 July 2020) were downloaded, and the transcripts were visualized using JBrowse [42,43]. Isoform levels and patient survival analyses were performed using GEPIA2 [44].

RNA m 6 A Quantification
The levels of RNA m 6 A in HepG2 cells were detected by dot blots using an anti-m 6 A antibody (#517924; Zhengneng Biotechnology, Chengdu, China), and the signal densities were analyzed by ImageJ software (National Institutes of Health, Bethesda, MD, USA). The levels of RNA m 6 A from patients were measured using an EpiQuik m 6 A RNA methylation quantification kit (P9005; Epigentek, Farmingdale, NY, USA).

Cell Growth and Proliferation Assays
Twenty-four hours after transfecting METTL3 expression plasmids, HCCLM3 or HepG2 cells were seeded at 5 × 10 3 cells per well in 24-well plates and cultured for 5 days. Cell numbers were counted every day using a hemocytometer.
CCK-8 assays (Beyotime, Shanghai, China) were performed to determine cell proliferation. Briefly, 100 µL of 2 × 10 3 cells was seeded into each well of the 96-well plate and incubated in a 5% CO 2 incubator at 37 • C for 0, 24, 48, and 72 h. Next, 10 µL of CCK-8 solution was added to each well and incubated for 2 h. The absorbance was measured at 450 nm using a spectrophotometer.
MTT assays were performed using an MTT Cell Proliferation Assay Kit (Beyotime, Shanghai, China). Briefly, 100 µL of 5 × 10 3 cells was seeded into each well of the 96well plate and incubated for 0, 24, 48, and 72 h at 37 • C, and then 10 µL of MTT stock solution was added for another 4 h of incubation. The 96-well plate was then foil wrapped after adding 100 µL of DMSO solution, followed by shaking on an orbital shaker for 15 min to fully dissolve the MTT formazan. Absorbance was measured at 490 nm using a spectrophotometer.

Cell Migration and Invasion Assays
The migration of HCC cells was detected using the wound-healing assay. For the cultured six-well plate, a linear wound was created using a 200-µL pipette tip, and two photographs were taken at the 24 h and 48 h time points. The wound area was measured using ImageJ software [45].
The invasion ability of cells was determined using a Transwell chamber inserted with 8-µm pore size polycarbonate membranes (Corning, NY, USA). Briefly, 5 × 10 4 cells in serum-free DMEM were plated into the Transwell upper chamber and coated with Matrigel (BD Biosciences, San Jose, CA, USA), DMEM with 20% FBS was added to the lower chamber. After 24 h of incubation at 37 • C with 5% CO 2 , the non-invading cells sticking to the top Genes 2022, 13, 669 5 of 16 chamber were cleared using a cotton swab, and the invading cells that had migrated to the lower surface of the membrane were fixed with 20% precooled methanol for 30 min and stained with 0.5% crystal violet (Servicebio, Wuhan, China) for 10 min. The Transwell membranes containing stained cells were digitally imaged and then counted using ImageJ software [45].

Statistical Analysis
All the data were analyzed by GraphPad Prism as means ± SD to assess group differences. Statistical significance was defined by Student's t test, and a p value < 0.05 was considered to be statistically significant.

Differentially Spliced Variants of METTL3 in Human Tissues and HCC Cell Lines
The human METTL3 gene is located on Chr14; it has 11 canonical exons and is alternatively spliced, similar to most other mammalian genes [4]. To fully investigate the splicing variants of METTL3 in HCC, we first analyzed our next-generation RNA-seq data of the HCC cell line HepG2. Including METTL3-A, the full-length and canonically spliced product, fourteen transcripts from the METTL3 gene in HepG2 cells were constructed, all of which can be found in the online Ensembl database ( Figure 1A upper and Figure S1A).
Considering their abundance, length of the transcript and coding region, and conservation in mammals, we chose three splice variants (METTL3-B, -C, and -D) and full-length METTL3-A for further investigation. In the variant METTL3-B, exons 6 and 9 were skipped, and the skipping of exon 6 introduced a premature termination codon (PTC) resulting in the translation of a 386-aa peptide without the MT-A70 domain. In variant C, the middle part of exon 3 was missed, and introns 8 and 9 were totally retained, introducing an even earlier PTC that allows METTL3-C to be translated into a much smaller protein. Two alternative splicing events were found to generate the variant METTL3-D, alternative 3 SS to generate a shorter exon 4 (here named exon 4a) and the retention of introns 8 and 9; thus, METTL3-D could be translated into a 258-aa peptide without the C-terminal catalytic domain ( Figure 1A lower and Figure S2). Next, we analyzed the RNA levels of METTL3 splicing variants in human tissues using GTExPortal. All fourteen isoforms were detected in human tissues with certain specificities ( Figure S1B). Further comparison showed that METTL3-A, -C, and -D were relatively higher in all the analyzed tissues, while METTL3-B was relatively low. Surprisingly, METTL3-D was at similar or even higher levels than METTL3-A in most human tissues ( Figure 1B).
To confirm the presence of these splice variants, we verified the existence of METTL3-A, -C, and -D by RT-PCR using specific primers and Sanger sequencing ( Figure S3A) and METTL3-A and -D by detecting full transcripts using public HepG2 third-generation sequencing data ( Figure S3B). Among those variants, METTL3-B is the only conserved variant between humans and mice ( Figure S4A,B. Failing to verify the specific exonic junctions of METTL3-B in human cells could be due to its low level; however, we successfully detected Mettl3-b in the mouse sample ( Figure S4C).
Because canonical full-length METTL3-A regulates the carcinogenesis, tumor progression, and drug resistance of HCC [31,46], we carefully detected the levels of METTL3 splicing variants in normal human liver, HCC tumor tissues, and HepG2 cells using RT-qPCR. Consistent with the above GTExPortal data, the level of METTL3-D in normal human liver was much higher than that of METTL3-A ( Figure 1C left). By contrast, METTL3-D was significantly decreased in the tumor tissues and HepG2 cells ( Figure 1C middle and right). Additionally, METTL3-B expression was increased in HCC tissues and cells, while METTL3-C expression was at similar levels.

METTL3-D Is Negatively Associated with HCC Malignancy
To confirm the above pattern in a larger pool of patient samples, we first analyzed a set of TCGA data containing 369 HCC and 160 normal liver samples and observed that METTL3-A was upregulated in HCC samples, METTL3-B was not significantly changed, and METTL3-C and D were significantly downregulated ( Figure 2A). Second, we tested our twenty paired HCC patient samples using RT-qPCR ( Figure 2B). Consistent with the TCGA data, METTL3-A was significantly elevated in our HCC patient livers compared with their NATs, while METTL3-D was significantly decreased. By contrast, we observed a significant increase in METTL3-B and no obvious change in METTL3-C in our HCC patients. The reason may be the inaccuracy of amplification of lower level RNAs or unpaired samples from the TCGA database.  Next, we asked whether the levels of METTL3-A and METTL3-D were correlated with the TNM stages of HCC patients. The twenty patients we investigated were classified into two groups according to their developmental stages of tumors (Table S1). METTL3-D was at a relatively higher level in earlier tumor stage patients than in later stage patients. By contrast, METTL3-D was at a significantly lower level in later stage patients ( Figure 2C).
To detect proteins translated from those splice variants, we first constructed overexpression (OE) plasmids of the four METTL3 isoforms using FLAG-tag and expressed them in HepG2 cells, which were analyzed by Western blotting and used as migration markers for in vivo METTL3 variants ( Figure 2D right). Using another three paired HCC patient samples, we found that METTL3-A protein signals were obviously increased in tumor tissues compared with those in their NATs, while METTL3-D protein signals were significantly decreased in tumor tissues (Figure 2D left). The signals of METTL3-B and -C were weak and difficult to differentiate between samples. These findings are consistent with the above RNA analyses and strongly demonstrate that the splice variant METTL3-D is translated in both normal and tumor tissues and its protein level is significantly decreased in HCC tumors. Additionally, the protein levels of METTL3-D in these three patients were negatively correlated with the tumor stage ( Figure 2E), in which patients #121 and #178 were at stage I and had much more METTL3-D proteins than patient #120 ( Figure 2F).
Taken together, these results strongly demonstrate that METTL3-D is negatively correlated with HCC malignancy and is downregulated in HCC patient tumor tissues, suggesting that METTL3-D could be a tumor suppressor in HCC.

Expression of METTL3-D Decreases m 6 A Modification
Because the splice variants METTL3-B, -C, and -D lack the adenosine methyltransferase catalytic domain MT-A70 ( Figure 1A), we next evaluated their effects on RNA m 6 A modification. Dot blot analysis revealed that the m 6 A level of total RNAs in HepG2 cells was slightly increased when METTL3-A (WT) was overexpressed and significantly decreased when the three splice variants, particularly the METTL3-D variant, were overexpressed ( Figure 3A,B).
Second, we detected m 6 A of mRNAs from 20 HCC patient samples and found that their levels in tumor samples were significantly higher than those in their corresponding NAT samples ( Figure 3C). Further analysis revealed that the m 6 A levels were positively correlated with the tumor stages of patients; the later the stage was, the higher the level of m 6 A was ( Figure 3D). Taken together, these results suggest that the presence of splice variants of METTL3 (particularly METTL3-D) significantly reduces the level of m 6 A modification in normal tissues, while downregulation of METTL3-D in HCC tumors results in increased levels of m 6 A modification.  Second, we detected m 6 A of mRNAs from 20 HCC patient samples and found that their levels in tumor samples were significantly higher than those in their corresponding NAT samples ( Figure 3C). Further analysis revealed that the m 6 A levels were positively correlated with the tumor stages of patients; the later the stage was, the higher the level of m 6 A was ( Figure 3D). Taken together, these results suggest that the presence of splice variants of METTL3 (particularly METTL3-D) significantly reduces the level of m 6 A modification in normal tissues, while downregulation of METTL3-D in HCC tumors results in increased levels of m 6 A modification.

METTL3-D Levels Are Inversely Correlated with Patient Survival in Multiple Cancers
The available TCGA data uncovered that the levels of METTL3-A were upregulated in most other solid tumors ( Figure S5A), while the levels of the METTL3-D variant were downregulated (Figure S5B), showing a similar pattern to those in HCC and suggesting The data are expressed as means ± SD. ** p < 0.01, **** p < 0.0001 (t test).

METTL3-D Levels Are Inversely Correlated with Patient Survival in Multiple Cancers
The available TCGA data uncovered that the levels of METTL3-A were upregulated in most other solid tumors ( Figure S5A), while the levels of the METTL3-D variant were downregulated ( Figure S5B), showing a similar pattern to those in HCC and suggesting that aberrant splicing of METTL3 is a common feature in solid tumors.
Further analyses of TCGA survival data revealed that HCC patients with lower levels of METTL3-D exhibited shorter overall survival than patients whose tumors had higher levels of METTL3-D ( Figure 3E). TCGA data also indicated that HNSC patients with higher levels of METTL3-D in their tumors had significantly longer overall survival ( Figure 3F). A similar pattern was also observed in patients with other solid tumors, such as LUAD, LUSC, BLCA, and PAAD, although statistically, their differences were not significant ( Figure S6). Additionally, METTL3-D expression was positively correlated with drug response in patients treated with different tyrosine kinase inhibitors and other common anti-cancer drugs (Figure 4).
higher levels of METTL3-D in their tumors had significantly longer overall survival (F ure 3F). A similar pattern was also observed in patients with other solid tumors, such LUAD, LUSC, BLCA, and PAAD, although statistically, their differences were not sig icant ( Figure S6). Additionally, METTL3-D expression was positively correlated w drug response in patients treated with different tyrosine kinase inhibitors and other co mon anti-cancer drugs (Figure 4).

METTL3-D Inhibits the Proliferation, Migration, and Invasion of HCC Cells
Consistent with previous studies showing that METTL3 promotes the survival a invasion of HCC cells [31,46], we found that OE of canonical full-length METTL3-A fa itated the growth and proliferation of HCCLM3 cells, as demonstrated by various assa including regular cell counting, the CCK-8 assay (Figure 5A), and the MTT assay ( Fig  5B, left). By contrast, OE of the METTL3-D variant suppressed cell growth and prolife tion ( Figure 5A,B left). A similar pattern was also observed in HepG2 cells ( Figure  right).

METTL3-D Inhibits the Proliferation, Migration, and Invasion of HCC Cells
Consistent with previous studies showing that METTL3 promotes the survival and invasion of HCC cells [31,46], we found that OE of canonical full-length METTL3-A facilitated the growth and proliferation of HCCLM3 cells, as demonstrated by various assays, including regular cell counting, the CCK-8 assay ( Figure 5A), and the MTT assay ( Figure 5B, left). By contrast, OE of the METTL3-D variant suppressed cell growth and proliferation ( Figure 5A,B left). A similar pattern was also observed in HepG2 cells (Figure 5B right).
Furthermore, the wound healing assay revealed that the migration ability of HCC cells was enhanced by METTL3-A but inhibited by METTL3-D in both MHCC97L ( Figure 5C) and HepG2 ( Figure S7A) cells. Similarly, the invasion abilities of these two HCC cells, as demonstrated by the Transwell assay, were enhanced by METTL3-A but inhibited by METTL3-D (Figures 5D and S7B). These results demonstrate that, unlike canonical full-length METTL3, the METTL3-D variant is a tumor suppressor that inhibits HCC proliferation, migration, and invasion.

METTL3-D Has an Opposite Function to Downstream Genes
Previous studies have shown that the expression of many downstream genes is changed following the upregulation of METTL3-A in multiple cancers [31][32][33]47,48] by mechanistic regulation of their mRNA stabilities or translation due to increased m 6 A modification [49,50]. For example, METTL3-mediated m 6 A modification destabilizes SOCS2 mRNA in HCC and BATF2 mRNA in gastric cancer [32] and promotes the translation of epidermal growth factor receptor (EGFR) and the Hippo pathway effector TAZ in human lung cancer cells [33]. To investigate whether the METTL3-D variant also has effects, we measured the mRNA levels of those genes in METTL3-A-or METTL3-D-overexpressing HepG2 cells by RT-qPCR. Consistent with previous studies, the mRNA levels of BATF2 and SOCS2 were decreased following OE of METTL3-A in HepG2 cells. By contrast, the mRNA levels of BATF2 and SOCS2 were significantly increased following OE of the METTL3-D variant ( Figure 6A, left). Additionally, the mRNA levels of EGFR and TAZ were not decreased following OE of METTL3-A and were slightly increased following OE of METTL3-D, findings that are consistent with previous findings that increased m 6 A modification affects their translation abilities but not their RNA stabilities ( Figure 6A, right). We also found that the mRNA levels of SOCS2 and BATF2 were significantly higher in patient #23 than in patient #58, in which the level of METTL3-D showed the same pattern ( Figure 6B). Further analysis of TCGA data confirmed that the mRNA level of METTL3-D was positively correlated with that of BATF2 in HCC samples ( Figure 6C). Taken together, these results demonstrate that the METTL3-D variant has an opposite function to downstream genes of METTL3-A in HCC cells and patients. were previously reported as mRNA stability-decreased target genes of METTL3, while EGFR and TAZ were previously reported as mRNA translation-affected target genes [31][32][33]. (B) mRNA levels of SOCS2 and BATF2 in two HCC patients. Compared with patient #58, patient #23 had higher levels of METTL3-D and higher levels of SOCS2 and BATF2. (C) Correlation coefficient between METTL3-D and BATF2 expression in HCC patients. The original data were downloaded from GEPIA2. The data in the bar graphs are expressed as means ± SD. * p < 0.05, ** p < 0.01, **** p < 0.0001 (t test).

Discussion
We also found that the mRNA levels of SOCS2 and BATF2 were significantly higher in patient #23 than in patient #58, in which the level of METTL3-D showed the same pattern ( Figure 6B). Further analysis of TCGA data confirmed that the mRNA level of METTL3-D was positively correlated with that of BATF2 in HCC samples ( Figure 6C).
Taken together, these results demonstrate that the METTL3-D variant has an opposite function to downstream genes of METTL3-A in HCC cells and patients.

Discussion
Alternative splicing produces multiple mRNA isoforms from one gene and enhances transcriptome plasticity and proteome diversity, which are crucial functions to regulate eukaryotic gene expression, particularly in mammals. Alternative splicing-generated splice variants of oncogenes have been associated with the development of many cancers [10,14,16,18]. METTL3 is the kernel of the m 6 A methyltransferase complex and affects the occurrence and development of multiple cancers [31,33,47]. METTL3 is upregulated in HCC patient tumor tissues, leading to increased mRNA m 6 A modification and enhanced progression of HCC [31]. Based on at least three lines of evidence, we identified a splice variant of METTL3, METTL3-D, which has an opposite function to canonical full-length METTL3 (named METTL3-A here). (1) Expression analyses of mRNAs and proteins revealed that the METTL3-D level is opposite that of METTL3-A. The level of METTL3-D is higher than that of METTL3-A in human normal liver tissues and HCC patient NATs but is lower than that of METTL3-A in HCC cell lines and patient tumors. (2) In contrast to the promotion of m 6 A modification and mRNA decay/stability of downstream target genes by METTL3-A, METTL3-D inhibits cellular m 6 A modification and protects the mRNA decay/stability of downstream genes. (3) METTL3-D is negatively associated with the malignancy of HCC tumors, including its positive correlation with longer patient survival times and negative correlations with the growth, migration, and invasion of HCC cells. Therefore, we propose a novel inactivation mechanism of HCC oncogenes using two steps of posttranscriptional regulation, alternative splicing and RNA modification, in which the alternatively spliced variant of METTL3 antagonizes the function of its canonical full-length isoform, decreasing the m 6 A modification and thereafter altering the mRNA decay/stability of HCC oncogenes.
Compared with canonical full-length splicing isoforms, alternatively spliced variants often introduce a PTC in their mRNAs, such as METTL3-B, -C, and -D. Most PTC-containing mRNAs are rapidly degraded via an mRNA surveillance mechanism, nonsense-mediated mRNA decay (NMD) [48]. However, several PTC-containing mRNAs generated by aberrant splicing evade the NMD pathway and are translated into truncated proteins in various diseases, such as β-globin in β-thalassemia and BRCA1 in breast cancer [49,50]. Here, the mRNA level of METTL3-D was higher than that of METTL3-A in the normal human liver. Additionally, the truncated METTL3-D-encoding protein was stable in HCC patient NATs and tumors ( Figure 2D). Global mRNA analysis also showed that the METTL3-D level is similar or even higher than that of METTL3-A in most human tissues ( Figure 1B). These results indicate that the truncated METTL3-D protein, lacking the (N6-adenosine)methyltransferase domain, is a critical regulator of tissue homeostasis by competing with the full-length METTL3-A protein. Aberrant splicing decreasing the level of METTL3-D substantially contributes to the tumor progression of HCC.
Understanding the cellular and molecular mechanisms driving HCC has made considerable progress in the past several years [51]. However, many studies have ignored the importance of alternative splicing as a source of novel HCC prognostic markers and disease targets, although aberrant expression could act as oncogenic drivers and passenger factors in other cancers [52]. Mutations in the cis-elements of pre-mRNAs, core spliceosomal proteins, and splicing regulators have been extensively reported in HCC [53][54][55]. Presently, which mutations cause the aberrant splicing of METTL3 in HCC are unknown, and their regulatory mechanism will be of interest in future studies.
HCC patients with higher METTL3-D and lower METTL3-A have longer overall survival times. Therefore, METTL3-D and METTL3-A could be used as predictive factors for postoperative HCC survival. Recently, antisense oligonucleotides (ASOs) blocking RNA cis-elements and/or altering RNA structures have been extensively studied for their modulation of splicing in vivo [56,57]. In the future, isoform switching from METTL3-A to METTL3-D using ASOs or other methods could be a potential method for HCC therapy. Additionally, this study reveals the function of a METTL3 splice variant, suggesting that comprehensive understanding of other METTL3 splice variants would be helpful to eventually understand HCC development and therapy.
Supplementary Materials: The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/genes13040669/s1, Figure S1: Alternatively Spliced Isoforms of METTL3 in Human Tissues.; Figure S2: Primary Sequences of the Proteins Encoded by the Four METTL3 Isoforms; Figure S3: Validation of the METTL3 Splice Variants; Figure S4