All clinical samples used in the present study were collected from The First Affiliated Hospital of Sun Yat-sen University (Guangzhou, China). Prior to sample collection, written informed consent was obtained from all the patients. All patient-related studies were reviewed and approved by the Independent Ethics Committee for Clinical Research and Animal Trials of The First Affiliated Hospital of Sun Yat-sen University [approval no. (2022)229], and performed in accordance with the Declaration of Helsinki. Inclusion criteria for the present study were as follows: i) Pathologically confirmed primary OSCC; ii) no neoadjuvant therapy; iii) complete clinical/follow-up data; and iv) an age of ≥18 years with normal liver/kidney function. Exclusion criteria were as follows: i) Secondary OSCC/synchronous malignancies; ii) history of other cancer types (5 years); iii) severe systemic diseases; and iv) incomplete data. Samples (n=4) were collected from August to October 2022. Table SI presents the demographic information and tumor characteristics of patients in the present study.

The HSC3 cell line was purchased from the Japanese Collection of Research Bioresources Cell Bank (cat. no. JCRB0623). SCC15 cells were purchased from the American Type Culture Collection (cat. no. CRL-1623). The Human Oral Keratinocytes (HOK) cell line was purchased from ScienCell Research Laboratories, Inc. (cat. no. 2610). Cells were maintained in a humidified incubator with 5% CO₂ at 37°C and cultured in DMEM/F-12 medium (cat. no. C11330500BT) supplemented with 1% penicillin/streptomycin (cat. no. 15140-122; Gibco; Thermo Fisher Scientific, Inc.) and 10% FBS(Gibco; Thermo Fisher Scientific, Inc.).

Both METTL5 knockout and overexpression systems were constructed for functional assays: Knockout system (lentivirus-mediated): The fourth-generation CRISPR plasmid system was used, including packaging plasmids (TAT, RAII, HEMP2 and VSVG) and target plasmids (LentiCRISPRv2 [negative control, NC group], lentiCRISPRv2-METTL5-human-sg1, lentiCRISPRv2-METTL5-human-sg2 [target groups]; purchased from Guangzhou IGE Biotechnology Co., Ltd.]). The interim cell line for lentiviral packaging was 293T cells (purchased from the American Type Culture Collection), which were used to produce viral particles.

Overexpression system (transient transfection): The overexpression plasmid (pFLAG-CMV2-METTL5/pFLAG-CMV2-CCND3; overexpression, OE) and empty vector (pFLAG-CMV2; negative control, NC) were purchased from Guangzhou IGE Biotechnology Co., Ltd.), and introduced into cells via transient transfection (no lentiviral packaging required).

Lentiviral transfection (for knockout) was performed as follows: 1×10⁶ 293T cells were seeded in T25 flasks 1 day prior to transfection at 37°C and cultured until reaching 60% confluence. The transfection mixture was prepared by combining the packaging plasmids (0.5 µg TAT, 0.5 µg RAII, 0.5 µg HEMP2, 0.5 µg VSVG) and target plasmids (2 µg per construct) with 200 µl Opti-MEM medium (Gibco; Thermo Fisher Scientific, Inc.). Separately, Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) was gently mixed with 200 µl Opti-MEM and incubated at room temperature for 5 min. The plasmid mixture and Lipofectamine 2000 solution were then combined, gently mixed, and incubated at room temperature for 20 min (plasmid: Lipofectamine 2000 ratio=3.5 µg: 7 µl). The complete medium of 293T cells was aspirated and replaced with 5 ml antibiotic-free DMEM containing 10% FBS, followed by the addition of the entire plasmid-Lipofectamine 2000 complex. Cells were incubated at 37°C in a 5% CO₂ incubator for 12–24 h before medium replacement.

Lentiviral particles were collected every 24 h for 2–3 consecutive days. The collected medium was centrifuged at ~450 × g (2,000 rpm) for 5 min at 4°C, and the viral supernatant was aliquoted into sterile 1.5 ml EP tubes and stored at −80°C. For infection of target OSCC cells (HSC3 and SCC15), cells were seeded in T25 flasks and cultured until 60% confluence. Lentiviral particles were added at a multiplicity of infection (MOI) of 10 along with polybrene (final concentration: 8 µg/ml) to enhance transduction efficiency. Cells were incubated at 37°C in a 5% CO₂ incubator for 24–48 h.

Following transduction, the selection method was antibiotic selection using puromycin; the selection concentration was 2.5–3 µg/ml, and the maintenance concentration was 1.0–1.5 µg/ml. Cells were cultured for >48 h until all untransfected cells died. After selection, cells were maintained in medium containing the maintenance concentration of puromycin. The transduced cells were then passaged, and protein extraction was performed for western blot analysis to verify the gene knockout efficiency of METTL5. The time interval between transduction and subsequent experimentation was 48–72 h after successful transduction and selection.

Transient transfection (for METTL5 overexpression): OSCC cells (HSC3 and SCC15) were seeded in 6-well plates and cultured until 60% confluence. The transfection mixture was prepared by combining 4 µg of pFLAG-CMV2-METTL5/pFLAG-CMV2-CCND3 (or pFLAG-CMV2) with 250 µl Opti-MEM medium, then mixed with 5 µl Lipofectamine 2000 (incubated at room temperature for 20 min). The mixture was added to cells in antibiotic-free medium, and cells were incubated at 37°C for 48 h. The time interval between transfection and subsequent experimentation was 48 h. Western blot analysis was performed to verify the METTL5 overexpression level.

CCND3 overexpression in METTL5-depleted HSC3 cells: METTL5-depleted HSC3 cells (sgMETTL5 HSC3 cells) were seeded in 6-well plates and cultured until 60% confluence. The CCND3 overexpression plasmid (pFLAG-CMV2-CCND3) and empty vector (pFLAG-CMV2; negative control) were introduced via transient transfection: 4 µg of plasmid (or empty vector) was mixed with 250 µl Opti-MEM medium, followed by 5 µl Lipofectamine 2000. The mixture was incubated at room temperature for 20 min, then added to cells in antibiotic-free medium. Cells were incubated at 37°C (5% CO₂) for 48 h. The time interval between transfection and subsequent experimentation was 48 h. Western blot analysis was performed to verify CCND3 overexpression efficiency before subsequent functional assays. The sgRNA sequences used in the present study are listed in Table SII.

Tissues from the mouse model were first fixed in 4% paraformaldehyde (PFA) (w/v in phosphate-buffered saline, PBS) at 4°C for 24 h. Fixed tissues were embedded in paraffin resin, and serial sections were cut into 4 µm-thick slices using a microtome. For dewaxing and rehydration, sections were processed through a descending alcohol series: xylene (twice for 10 min each), 100% ethanol (twice for 5 min each), 95% ethanol (5 min), 80% ethanol (5 min), 70% ethanol (5 min), and finally rinsed with distilled water for 5 min. Antigen retrieval was performed by immersing sections in 0.01 M citrate buffer (pH 6.0) and heating in a microwave oven at 95–100°C for 15 min. After natural cooling to room temperature, sections were washed with PBS (3 times for 5 min each).

To block endogenous peroxidase activity, sections were treated with 3% hydrogen peroxide at room temperature for 10 min, followed by three washes with PBS (5 min each). Non-specific binding sites were blocked using 5% BSA (w/v in PBS; cat. no. 4240GR100; Saiguo Biotechnology Co., Ltd.) at room temperature for 30 min. Subsequently, sections were incubated overnight with primary antibodies at 4°C: Anti-Ki-67 and anti-METTL5. After three washes with PBS (5 min each), sections were incubated with a horseradish peroxidase (HRP)-conjugated goat anti-rabbit secondary antibody at room temperature for 60 min. Details of all primary and secondary antibodies are in Table SIII.

Immunoreactivity was visualized using a DAB detection kit (cat. no. GK6007; Beijing Genetech Co., Ltd.) as the chromogen substrate. Sections were incubated with DAB working solution at room temperature for 3–5 min (monitoring staining intensity under a light microscope to avoid over-staining). The reaction was terminated by rinsing with distilled water. Sections were then counterstained with hematoxylin at room temperature for 2 min, differentiated with 1% hydrochloric acid in ethanol for 30 sec, and dehydrated through an ascending alcohol series (70, 80, 95 and 100% ethanol for 5 min each) followed by xylene (twice for 10 min each). Finally, sections were mounted with neutral gum and a coverslip was applied.

Stained sections were imaged using a Leica inverted light microscope (Leica Microsystems) at magnifications of 400×; scale bar, 50 µm. IHC scoring was performed based on a semiquantitative immunoreactivity scoring system (4). Briefly, the proportion of positive staining cells was scored as follows: 0 (absent), 1 (<10%), 2 (10–50%), and 3 (>50%). Staining intensity was scored as: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The final IHC score was calculated by multiplying the percentage score by the intensity score, resulting in a range of 0–9. Image analysis and scoring were performed using ImageJ software (Version 1.53; National Institutes of Health).

Total protein was extracted from OSCC cells (HSC3/SCC15), HOK and human tissues using RIPA buffer (cat. no. C1053; Applygen Technologies, Inc.). Protein concentration was determined via the BCA protein assay kit (cat. no. P0012; Beyotime Biotechnology) following the manufacturer's instructions.

Equal amounts of protein (20 µg per lane) were loaded onto 10% SDS-PAGE gels for separation, then transferred onto a PVDF membrane (cat. no. IPVH00010; Merck KGaA). The membrane was blocked with 5% non-fat skimmed milk (dissolved in TBS-Tween; TBST) for 2 h at room temperature; TBST buffer contained 0.1% Tween-20 (v/v) in Tris-buffered saline.

Subsequently, the membrane was incubated with primary antibodies (listed in Table SIII) at 4°C overnight. After three washes with TBST (10 min each), the membrane was incubated with species-matched secondary antibodies (HRP-conjugated) at room temperature for 1 h.

Protein bands were detected using a super-sensitive ECL kit (cat. no. P0018FS; Beyotime Biotechnology). The membrane was incubated with the ECL working solution (1:1 mixture of reagent A and B) for 5 min at room temperature, then visualized using the Tanon 5200 Multi Intelligent Imaging System (Tanon Science and Technology Co., Ltd.). Band intensities were quantified using ImageJ software (Version 1.53; National Institutes of Health) for densitometric analysis.

Cell viability was monitored using a CCK-8 kit (cat. no. CK04; Dojindo Laboratories, Inc.). A total of 1×10³ cells (HSC3 or SCC15) per well were plated onto 96-well plates and cultured in DMEM/F-12 with 1% FBS. CCK-8 working solution was then applied to the cells for 2 h, according to the manufacturer's instructions. A microplate reader (cat. no. A51119500C; Thermo Fisher Scientific, Inc.) was then used to measure the absorbance of the supernatant at 450 nm.

HSC3 and SCC15 cells were seeded in 6-well plates at a density of 1×10³ cells/well without additional drug treatment. Cells were incubated in a humidified incubator with 5% CO₂ at 37°C for 2 weeks to allow colony formation. After the incubation period, cells were washed twice with PBS (5 min each wash). Subsequently, cells were fixed with 4% formaldehyde at room temperature for 15 min. The fixative was discarded, and cells were stained with 0.1% crystal violet solution (w/v in distilled water) at room temperature for 20 min. Excess crystal violet was rinsed off with distilled water, and the plates were air-dried. A colony was defined as a cluster containing ≥50 cells. Colony quantification was performed using ImageJ software (Version 1.53; National Institutes of Health). Briefly, images of stained wells were captured by camera (cat. no. ILCE-7M4, Sony Corporation), and the ImageJ software (Version 1.53; National Institutes of Health) was used to count the number of discrete colonies based on color thresholding and size gating.

Transwell migration assays were performed using Transwell chambers with an 8 µm pore size (cat. no. 353097; Falcon^®; Corning Life Sciences). No Matrigel was used in the assay, confirming this as a migration experiment. The lower chamber of the 24-well Transwell system was filled with DMEM supplemented with 10% FBS. HSC3 and SCC15 cells were resuspended in serum-free DMEM, and 5×10⁴ cells/well were seeded into the upper chamber in a total volume of 200 µl. Cells were incubated in a humidified incubator with 5% CO₂ at 37°C for 48 h to allow migration through the pores.

After incubation, non-migrated cells on the upper surface of the membrane were gently removed with a cotton swab. Migrated cells adhering to the lower surface were fixed with 4% formaldehyde at room temperature for 30 min, then stained with 0.1% crystal violet solution at room temperature for 30 min. Excess stain was rinsed off with PBS, and the membranes were air-dried. Migrated cells were observed and counted using a Leica inverted light microscope (Leica Microsystems) at a magnification of 100×. Scale bar, 100 µm were included in the figure legends for reference. Three random fields were selected per membrane to count migrated cells, and the average number was calculated for statistical analysis.

A total of 2×10⁵ HSC3 and SCC15 cells were seeded into each well of a 6-well plate. The cells were cultured in DMEM/F-12 medium supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.) without additional drug treatment, and no serum starvation was performed before the assay. When cell confluence reached ~90% of the plate surface, a sterile 200 µl pipette tip was used to scratch a straight wound across each well. The scratched cells were gently washed away with PBS (twice) to remove cell debris, and fresh DMEM/F-12 medium containing 10% FBS was added to each well. Wound width was measured at 0 h and the endpoint time (8, 12 or 24 h) under the bright field of a Leica inverted light microscope (Leica Microsystems). The percentage of wound healing was calculated using the formula: Wound healing (%)=1-(width_t/width₀) ×100, where width₀ is the initial wound width at 0 h and width_t is the wound width at the endpoint time.

Cell cycle progression of HSC3 and SCC15 cells was assessed using a Cell Cycle Kit (cat. no. CCS012; MultiSciences (Lianke) Biotech Co., Ltd.) (no cytokines/chemokines used). 2×10⁵−1×10⁶ harvested cells were washed with PBS, then mixed with 1 ml DNA staining solution + 10 µl permeabilization solution (kit components), vortexed 5–10 sec, and incubated at room temperature in the dark for 30 min. Stained cells were analyzed on a BD FACSCanto™ II flow cytometer (low injection speed), and data (G0/G1, S, G2/M phases) were processed via FlowJo v10.8.1 (BD Biosciences). No antibodies were used.

A total of 12 6-week-old female BALB/c nude mice (18–22 g) were purchased from the Animal Experiment Center of The First Affiliated Hospital, Sun Yat-Sen University. Mice were housed under specific pathogen-free conditions (20–25°C; relative humidity, 40–70%; 12 h light/12 h dark cycle and ad libitum access to sterile food and water).

1×10⁶ tumor cells (METTL5 knockout and control SCC15 cells, in 50 µl sterile PBS) were subcutaneously injected into the right dorsal flank of nude mice. Palpable tumors were monitored once weekly. Humane endpoints: Excess tumor size (maximum permitted size was 20 mm), >20% weight loss, poor mobility or distress.

Experiment ended 18 days post-injection. Mice were euthanized by cervical dislocation, and death confirmed by 5-min absence of breathing/heartbeat. Tumors were excised, weighed, measured and imaged. IHC samples were fixed with 4% paraformaldehyde at 4°C for 24 h.

All experiments were approved by the Ethical Committee of The First Affiliated Hospital, Sun Yat-Sen University [approval no. (2024)051] and conducted per institutional guidelines.

To examine the alteration of mRNA translation after METTL5 depletion, mRNA-seq was performed on total mRNAs and RNC-mRNAs from METTL5-knocked-out or control HSC3 cells. RNC-mRNAs were extracted as previously described. Cells were quickly washed with ice-cold PBS containing 100 mg/ml cycloheximide and then collected and lysed in 1.2 ml lysis buffer containing 1% Triton X-100 in ribosome buffer [which contains 20 mM HEPESKOH (pH 7.4), 15 mM MgCl2, 200 mM KCl, 100 mg/ml cycloheximide and 2 mM dithiothreitol] for 30 min on ice. Cell lysates were centrifuged at 16,200 × g for 10 min at 4°C to remove the cell debris. The supernatants were transferred on the surface of 12 ml of ribosome buffer containing 30% sucrose. RNC-mRNAs were pelleted after ultra-centrifugation at 185,000 × g for 5 h at 4°C. RNAs in total cell lysates and RNC-mRNAs were extracted with TRIzol^® reagent (cat. no. 15596018; Thermo Fisher Scientific, Inc.) and subjected to RNA-sequencing (RNA-seq).

For RNA-seq experiments, the following details were followed: i) RNA sample preparation kit: TRIzol^® reagent (cat. no. 15596018; Thermo Fisher Scientific, Inc.); ii) sample quality/integrity verification: RNA quality was assessed by Q20 (>90%), Q30 (>85%) values, and RNA integrity number (RIN) ≥7.0 using a Microplate Reader (Molecular Devices, Gemini XPS) and Qubit^® 4 Fluorometer (cat. no. Q33216, Thermo Fisher Scientific, Inc.); iii) sequencing type: Paired-end sequencing (PE150) on the DNBSEQ platform; iv) sequencing kit: DNBSEQ-T7RS High-throughput Sequencing Kit (cat. no. 940-000268-00; Shenzhen BGI Genomics Co., Ltd.); v) Final library loading concentration: 2 nM, measured by reverse transcription-quantitative PCR (qPCR) using the Qubit^® ssDNA Assay Kit (cat. no. Q10212, Invitrogen; Thermo Fisher Scientific, Inc.) for molar concentration quantification; vi) Data analysis software: SOAPnuke (version 1.5.6; http://github.com/BGI-flexlab/SOAPnuke) for raw data filtering, HISAT2 (version 2.1.0; http://daehwankimlab.github.io/hisat2/) for genome alignment, Bowtie2 (version 2.3.4.3; http://bowtie-bio.sourceforge.net/bowtie2/index.shtml) for gene alignment, RSEM (version 1.2.28; http://deweylab.github.io/RSEM/) for gene expression quantification, DESeq2 (version 1.40.2; http://bioconductor.org/packages/3.16/bioc/html/DESeq2.html) for differential expression analysis.

The TE was calculated using the following formula: TE=transcripts per million (TPM) in RNC-seq/TPM in input RNA-seq. Genes with a TE ratio of sgMETTL5 (sgM5)/NC <0.667 and >1.5 were classified as decreased TE (TE down) and enhanced TE (TE up), respectively.

GO and pathway analyses of TE-down mRNAs identified in RNC-seq data were performed using the Bioinformatics Webtool (https://www.bioinformatics.com.cn/), and significantly enriched pathways were identified.

Gene Expression Profiling Interactive Analysis (GEPIA) database (http://gepia.cancer-pku.cn/): The dataset ‘TCGA-HNSC’ (The Cancer Genome Atlas Head and Neck Squamous Cell Carcinoma) was queried with the search term ‘METTL5’ to analyze METTL5 mRNA expression levels in tumor vs. normal tissues. Survival analysis was performed via GEPIA using the ‘TCGA-HNSC’ dataset with METTL5 expression stratified into ‘low’ and ‘high’ groups (median cutoff). University of Alabama at Birmingham Cancer data analysis Portal (UALCAN; http://ualcan.path.uab.edu/index.html): The dataset ‘TCGA-HNSC’ was queried with the search term ‘METTL5’ to analyze the association between METTL5 expression and clinical parameters (lymph node metastasis, cancer stage, tumor grade).

Quantitative data are presented as mean ± standard deviation. Unpaired two-tailed Student's t-test was used to assess statistically significant differences between two experimental groups, while one-way ANOVA followed by Tukey's honest significant difference post hoc test was applied for multiple groups. For non-parametric data, the Kruskal-Wallis test was used, with Dunn's post hoc test for pairwise comparisons following a significant result. Pearson's correlation analysis was performed for the data presented in Fig. S4C (from GEPIA2; http://gepia2.cancer-pku.cn/). All experiments were performed with three independent biological repeats. Statistical significance was defined as P<0.05; for differential gene analysis (where applicable), the significance cut-off levels were set as adjusted P<0.05 and |log2 fold change| >1. Statistical analyses were performed using GraphPad Prism (version 8.0; Dotmatics). *P<0.05, **P<0.01; ***P<0.001; ****P<0.0001.

To ascertain whether METTL5 contributes to cancer progression, the present study performed an analysis of METTL5 mRNA levels in HNSCC using the Gene Expression Profiling Interactive Analysis database (http://gepia.cancer-pku.cn/, dataset are provided in Methods section: Datasets). The findings indicated that METTL5 expression was significantly higher in tumor tissues compared with that in normal tissues (Fig. 1A). In addition, data from University of Alabama at Birmingham Cancer data analysis Portal (https://ualcan.path.uab.edu/index.html, dataset are provided in Methods section: Datasets) revealed that elevated METTL5 expression was significantly associated with increased lymph node metastasis (vs. no lymph node metastasis), advanced cancer stages (vs. early stages) and higher tumor grades (vs. lower grades) in HNSCC (Fig. 1C-E). Furthermore, patients with higher METTL5 expression exhibited significantly worse overall survival than those with low METTL5 expression (Fig. 1B). Collectively, these findings suggest that METTL5 is key to regulating the progression of HNSCC.

Furthermore, western blotting analysis was used to assess METTL5 expression in two OSCC cell lines and a human normal oral epithelial keratinocyte (HOK) cell line. The results revealed that OSCC cells exhibited markedly higher METTL5 expression compared with that in HOK cells (Fig. 1F). Furthermore, western blotting analysis demonstrated significantly elevated METTL5 protein levels in human OSCC tissues compared with that in normal tissues (Fig. 1G and H).

To elucidate the function of METTL5 in OSCC, the present study established stable knockout cell lines using SCC15 and HSC3 cells, targeting METTL5 with independent sgMETTL5 (Figs. 2A and S1A). Notably, the METTL5 knockout cells demonstrated a significant decrease in proliferative capacity (Figs. 2B and S1B) and a colony formation (Figs. 2C and S1C) compared with that in the control group. Furthermore, the Transwell migration and wound healing assays indicated that METTL5 knockout significantly impaired the migration of OSCC cells in vitro compared with that in control cells (Figs. 2D, E, S1D and E).

A METTL5 overexpression system was also constructed using lentivirus to investigate the functional impact of METTL5 in OSCC (Figs. 3A and S2A). The findings revealed that METTL5 overexpression led to a significant increase in cellular proliferation (Figs. 3B and S2B) and colony formation (Fig. 3D) in OSCC cells. Furthermore, the overexpression of METTL5 enhanced the migration of OSCC cells (Figs. 3C and S2C), indicating a role for METTL5 in promoting the malignancy of OSCC.

Xenograft tumor-formation assays were subsequently performed to assess the role of METTL5 in the regulation of OSCC tumorigenesis in vivo. METTL5 knockout and control SCC15 cells were inoculated subcutaneously into the right dorsal flank of nude mice, with tumor volumes monitored throughout the experiment. The results indicated that the subcutaneous tumors were notably larger in the control group compared with that in the sgMETTL5 group (Fig. 4A). Compared with that in the control group, the growth curve of the xenograft tumor and the final tumor weights further demonstrated that the knockout of METTL5 significantly impeded tumor growth in vivo (Fig. 4B and C). Moreover, IHC analysis revealed that the expression level of METTL5 was significantly diminished in xenograft tumors derived from the sgMETTL5 group compared with that in the control group (Fig. 4D and E). In addition, xenograft tumors from the sgMETTL5 groups exhibited significantly lower Ki-67 expression compared with that in the control group, indicating that METTL5-knockout xenograft tumors were potentially less malignant (Fig. 4D and E).

The present study further utilized RNC-mRNA-seq to assess mRNA translation profiles in METTL5-depleted and control cells. The RNC-mRNA-seq analysis revealed 3,391 genes with TE-down and 1,020 genes with TE-up in METTL5-depleted OSCC cells (Fig. 5A). This suggests that METTL5-knockout predominantly leads to a decrease in TE rather than an increase. Furthermore, GO analysis of genes with TE-down revealed that ‘Cell cycle’-related processes were significantly enriched (Fig. 5B). Among the cell cycle-related TE-down genes identified, the present study prioritized cyclin D3 (CCND3) as it is a well-characterized cyclin driving cell cycle phase transition, and its dysregulation is directly associated with tumor growth (20). Subsequently, western blotting demonstrated that the levels of the cell-cycle regulator CCND3 were markedly decreased in sgMETTL5 cells compared with that in control cells (Fig. 5C). Consistent with this, cell cycle assays revealed that knockout of METTL5 induced the proportion of OSCC cells in the G₀/G₁ phase and decreased the proportion in the G₂/M phase (Fig. 6D and S3). In summary, the results indicate that the depletion of METTL5 in OSCC cells significantly alters the translational profile. This alteration not only results in a general decrease in TE but also specifically impacts the translation of certain mRNAs, particularly those associated involved in cell cycle regulation.

The aforementioned results indicate that the downregulation of METTL5 expression disrupts the translation process, particularly for mRNAs involved in the ‘Cell cycle’ pathway. Consequently, rescue experiments (Fig. 6) were performed using HSC3 cells due to their notably increased TE and higher baseline invasion capacity, which are key to robustly evaluate the functional recovery of migration ability. Although SCC15 cells demonstrated higher basal METTL5 expression (Fig. 1F), the characteristics of HSC3 made it a more suitable model for validating the rescue effect of CCND3 overexpression. Therefore, the present study overexpressed CCND3 in METTL5-depleted HSC3 cells (Fig. 6A). The results revealed that overexpression of CCND3 partially rescued the migration and colony formation capacity of METTL5-depleted OSCC cells (Fig. 6B and C). Moreover, for cell cycle analysis, METTL5 knockout induced a cell cycle arrest in the G₀/G₁ phase compared with that in control cells, and CCND3 overexpression reversed this (Figs. 6D and S3). Collectively, the results suggest that METTL5-mediated m⁶A modification contributes to translational dysregulation, potentially influencing OSCC progression. To explore the potential broader relevance of the findings of the present study, the expression patterns of METTL5 and CCND3 were analyzed across several cancer types using TCGA data. The results demonstrated that METTL5 (Fig. S4A) and CCND3 (Fig. S4B) exhibited differential expression in multiple cancer types compared with that in adjacent normal tissues. Moreover, although there is no existing literature reporting a direct association between METTL5 and CCND3 in other cancer types, to the best of our knowledge, the weak but statistically significant mRNA-level correlation demonstrated in the generated scatter plot indicates that METTL5 transcription minimally drives CCND3 mRNA expression (Fig. S4C).