METTL21A promotes hepatocellular carcinoma progression via methylating and stabilizing BAG3

Abstract

Protein methyltransferases regulate diverse physiological and pathological processes through histone and non-histone substrate methylation. While the role of histone methyltransferases in tumorigenesis is well-established, the contribution of non-histone methyltransferases, particularly Methyltransferase 21A (METTL21A), to hepatocellular carcinoma (HCC) progression remains poorly characterized. Here, we report that METTL21A is significantly upregulated in HCC tissues and associated with poor clinical prognosis. Transcriptional activation by CCCTC-binding factor (CTCF) was identified as a key driver of METTL21A overexpression. Functional studies demonstrated that METTL21A promotes HCC growth and metastasis in both in vitro and in vivo models. Mechanistically, METTL21A mediates methylation of Bcl-2-associated athanogene-3 (BAG3), thereby inhibiting its ubiquitination and subsequent degradation. We further identified Tripartite motif containing 21 (TRIM21) as the E3 ligase responsible for BAG3 ubiquitination and found that METTL21A disrupts the TRIM21-BAG3 interaction. Notably, we discovered that the natural compound sophoricoside (Sop) acts as a METTL21A inhibitor and exhibits potent anti-HCC activity. Our study not only elucidates the oncogenic role and molecular mechanism of METTL21A in HCC progression but also highlights its therapeutic potential as a novel drug target for HCC treatment.

Introduction

Hepatocellular carcinoma (HCC) ranks among the most common and lethal malignancies of the digestive system. While substantial progress has been made in identifying genetic and signaling pathway alterations driving HCC pathogenesis, the complete mechanistic understanding remains elusive. Post-translational modifications (PTMs) serve as crucial regulatory switches that dramatically expand protein functional diversity beyond their primary sequences and tertiary structures. These covalent yet reversible modifications play pivotal roles in controlling cellular homeostasis, proliferation, and survival—processes whose dysregulation frequently contributes to tumorigenesis¹. Protein methylation has mainly been detected on lysine and arginine residues. Histone lysine methylation represents one of the most abundant chromatin marks in eukaryotes, creating dynamic chromatin states that regulate gene expression, maintain genomic stability, and organize nuclear architecture. These modifications are intimately linked to cancer development and progression^2,3. In contrast to well-characterized histone methylation, non-histone lysine methylation remains a relatively unexplored frontier. This modification profoundly influences protein activation, stabilization, and degradation, thereby regulating diverse cellular processes. However, functional roles have been assigned to only a limited number of non-histone methylation sites. Notable examples include SETD7-mediated p53 methylation, which modulates its activity and stability by preventing competing PTMs^4–6. In HCC specifically, several lysine methyltransferases have been implicated: SMYD4 monomethylates PRMT5 to enhance its interaction with MEP50, promoting symmetric dimethylation of H3R2 and H4R3 at target gene promoters⁷, while SMYD5 catalyzes RPL40 trimethylation at K22 to regulate translational efficiency⁸.

The methyltransferase-like 21 (METTL21) protein family has been identified as a key regulator of molecular chaperone lysine methylation, with important implications in human diseases^9–13. METTL21A specifically mediates trimethylation of Lys561 (K561) in HSP70 and corresponding residues in other isoforms^11,13, and plays a critical role in proteostasis regulation during endoplasmic reticulum (ER) stress through GRP78 trimethylation¹⁴. While bioinformatics analyses of TCGA and GTEx databases suggest METTL21A upregulation in HCC tissues and its potential as an independent prognostic marker¹⁵, the functional significance and mechanistic basis of METTL21A-mediated lysine methylation in HCC progression remain poorly understood.

In this study, we establish METTL21A as a clinically relevant oncoprotein in HCC. We found that METTL21A is significantly upregulated in HCC tissues and strongly correlates with poor patient outcomes. CCCTC-binding factor (CTCF) is identified as the transcription factor driving METTL21A overexpression. Functionally, METTL21A promotes HCC growth and metastasis in both in vitro and in vivo models by stabilizing Bcl-2-associated athanogene-3 (BAG3) through suppression of TRIM21-mediated ubiquitination and subsequent degradation. Furthermore, we discovered that sophoricoside (Sop), a bioactive genistein glycoside derived from Fructus Sophorae, functions as a potential METTL21A inhibitor with significant anti-HCC activity. These findings not only uncover the oncogenic mechanisms of METTL21A in HCC progression but also highlight its therapeutic potential as a novel molecular target for HCC treatment.

Results

Overexpression of METTL21A indicates a poor clinical outcome in HCC patients

Through integrated analysis of multiple clinical datasets, we systematically evaluated the clinical significance of METTL21A in HCC. Analysis of TCGA-LIHC and GEO datasets (GSE25097) demonstrated significant upregulation of METTL21A mRNA expression in HCC tissues compared to adjacent normal tissues (Supplementary Fig. 1A). This finding was further validated in paired clinical samples from both TCGA (n = 50) and GSE76297 (n = 59) datasets (Supplementary Fig. 1B). Importantly, METTL21A expression showed a progressive increase with advancing tumor stage (Supplementary Fig. 1C), and high expression levels were significantly associated with worse overall survival (OS) and disease-free survival (DFS) (Supplementary Fig. 1D).

To confirm these bioinformatic findings, we performed IHC analysis of 75 paired HCC clinical specimens. METTL21A protein overexpression was detected in 69.3% of HCC cases (Fig. 1A), demonstrating excellent concordance with the transcriptomic data. Survival analysis revealed that patients with high METTL21A expression had significantly shorter OS and DFS compared to those with low expression (Fig. 1B). Furthermore, METTL21A expression levels showed significant correlations with key clinicopathological features, including distant metastasis (p = 0.034) and serum HBV DNA levels (p = 0.044) (Table 1). These consistent findings across multiple datasets and analytical platforms strongly suggest that METTL21A overexpression may represent both a prognostic biomarker and a potential therapeutic target in HCC.

Fig. 1. Overexpression of METTL21A indicates a poor clinical outcome in HCC patients.

A Representative IHC images and quantification of METTL21A protein in HCC clinical samples. Scale bars: 100 μm (above), 50 μm (below). B Kaplan–Meier analysis of OS and DFS in HCC patients with high (n = 43) vs low (n = 32) METTL21A expression. Log-rank test was used for statistical comparison.

Table 1.

Correlation analysis of METTL21A expression and clinicopathological features in patients with HCC

Clinicopathological variables	METTL21A expression		p value
	High (n = 43)	Low (n = 32)
Age
≥55 years	15	17	0.114
<55 years	28	15
Gender
Male	31	21	0.548
Female	12	11
Tumor size
≥5 cm	22	14	0.505
<5 cm	21	18
Metastasis
Yes	18	6	0.034*
No	25	26
Differentiation grade
Low	4	3	0.905
Medium	35	25
High	4	4
Vascular invasion
Yes	21	16	0.921
No	22	16
Liver cirrhosisa
Yes	20	11	0.2166
No	21	21
Serum HBV DNAb
≥1000 IU/mL	19	15	0.044*
<1000 IU/mL	7	17
Serum AFP
≥200 ng/mL	16	14	0.567
<200 ng/mL	27	18
Recurrence
Yes	18	15	0.665
No	25	17

Hepatitis B virus (HBV): ≥1000 IU/mL indicates active HBV replication.

Liver cirrhosis: confirmed by histopathological presence of pseudolobules.

AFP Alpha-fetoprotein.

^aTwo missing data points.

^bSeventeen missing data points.

^*p < 0.05 (statistically significant).

CTCF activates METTL21A transcription

To systematically identify upstream regulators of METTL21A overexpression in HCC, we conducted an integrative bioinformatics analysis combining five authoritative databases: humanTFDB, ENCODE, GTRD, GeneCards, and hTF-target (liver). This comprehensive screening identified 12 candidate transcription factors potentially regulating METTL21A transcription (Fig. 2A). Subsequent correlation analysis with TCGA expression data (R > 0.5, p < 0.05) narrowed the candidates to four transcription factors showing significant co-expression with METTL21A: CREB1, CTCF, HDAC2, and MAZ (Fig. 2B and Supplementary Fig. 2A). Experimental validation through RT-qPCR and Western blot analysis demonstrated that only CTCF knockdown consistently altered METTL21A expression at both mRNA and protein levels, while the other three candidates showed no significant regulatory effects (Fig. 2C, D and Supplementary Fig. 2B, C). This finding was further supported by a strong positive correlation (R = 0.789, p < 0.0001) between CTCF and METTL21A mRNA expression in our clinical HCC specimens (Fig. 2E). Moreover, dual-luciferase reporter assays confirmed that CTCF silencing significantly reduced METTL21A promoter activity (Fig. 2F), establishing CTCF as a bona fide transcriptional activator of METTL21A in HCC.

Fig. 2. CTCF activates METTL21A transcription.

A In silico analysis identified potential transcription factors within the METTL21A promoter region. B Spearman’s correlation analysis between METTL21A and CTCF mRNA expression in the TCGA-LIHC cohort. C The mRNA expression levels of METTL21A were assessed by RT-qPCR in Huh7 and SNU182 cells following CTCF silencing. D Protein expression levels of METTL21A were assessed by Western blot in Huh7 and SNU182 cells after CTCF silencing. E Spearman’s correlation analysis of METTL21A and CTCF mRNA expression in HCC clinical samples (n = 45). F Dual-luciferase reporter gene assay in cells co-transfected with siRNA, pRL-TK, and pGL3-Basic /pGL3-METTL21A plasmids. G Primer design targeting regions in the METTL21A promoter: P1 (−1300 bp to −1100 bp), P2 (−100 bp to +100 bp relative to the transcription start site, TSS), and P3 (+400 bp to +600 bp). H ChIP-qPCR analysis of three METTL21A promoter regions (P1, P2, and P3) in CTCF-silenced HCC cells using anti-CTCF antibody (IgG as control). GAPDH exon 3 served as a negative control. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001.

Subsequent analysis of ENCODE CTCF ChIP-seq datasets¹⁶ revealed three potential binding peaks (peak 1–3) within the METTL21A promoter region (Supplementary Fig. 2D). To investigate CTCF’s direct binding to these sites, we designed three primer sets (P1, P2, and P3) targeting the first 2000 bp and the last 600 bp upstream of the METTL21A transcription start site (TSS), corresponding to the identified peaks (Fig. 2G). ChIP-qPCR analysis confirmed significant CTCF enrichment at all three loci compared to the negative control (GAPDH Exon 3), with the P3 fragment exhibiting the highest binding affinity. Importantly, CTCF knockdown significantly attenuated its binding at all three sites (Fig. 2H). These results collectively demonstrate that CTCF directly binds to the METTL21A promoter and functions as an upstream transcriptional activator of METTL21A expression.

METTL21A promotes HCC growth

Western blot analysis of METTL21A expression patterns revealed significant upregulation in nine HCC cell lines compared to normal hepatocytes (THLE-2) (Supplementary Fig. 3A), consistent with our clinical tissue findings. To functionally characterize METTL21A in HCC pathogenesis, we performed loss-of-function studies using siRNA-mediated knockdown in multiple HCC cell lines (Fig. 3A). METTL21A depletion markedly impaired cellular proliferation, as evidenced by CCK-8 assays and colony formation capacity in Huh7, SNU182, and Hep3B cells (Fig. 3B, C and Supplementary Fig. 3B, C). Conversely, lentivirus-mediated METTL21A overexpression in Huh7 and MHCC97H cells (Fig. 3D) enhanced proliferative potential in both CCK-8 and colony formation assays (Fig. 3E, F).

Fig. 3. Functional characterization of METTL21A in promoting HCC cell proliferation.

A Validation of METTL21A knockdown efficiency in HCC cells by Western blot analysis. B Proliferation kinetics of METTL21A-deficient HCC cells measured by CCK-8 assay. C Clonogenic potential assessment of METTL21A-silenced HCC cells through colony formation assays. D Western blot confirmation of METTL21A overexpression in HCC cell lines. E Enhanced proliferative capacity of METTL21A-overexpressing HCC cells demonstrated by CCK-8 assay. F Quantitative analysis (n = 3) of colony formation efficiency in METTL21A-overexpressing HCC cells. G Cell cycle distribution profiles analyzed by flow cytometry. H Apoptotic cell fractions quantified by flow cytometric analysis of Annexin V/PI staining. I-K In vivo tumorigenicity assessment showing representative tumor specimens (I), growth kinetics (J), and final tumor weights (K) in subcutaneous xenograft models (n = 5). L Proliferation evaluation by Ki-67 IHC staining and quantification in xenograft tissues (n = 5). Scale bars: 100 μm (overview), 50 μm (detail). ^*p < 0.05, ^**p < 0.01, and ^***p < 0.001.

To gain insight into the mechanism of METTL21A in promoting cell proliferation, we performed a cell cycle and apoptosis analysis using flow cytometry. Silencing METTL21A induced apoptosis and G0/G1 cell cycle arrest in Huh7 and SNU182 cells (Fig. 3G, H). Concurrently, the levels of G1/S phase checkpoint proteins, cyclin E1 and CDK2, were reduced, whereas the apoptosis protein marker cleaved-PARP showed an increase (Supplementary Fig. 3E).

In vivo validation using subcutaneous xenograft models demonstrated that METTL21A knockdown reduced tumor growth compared to controls (Fig. 3I–K). IHC analysis confirmed corresponding decreases in Ki-67 proliferation index in METTL21A-deficient tumors (Fig. 3L). These consistent findings across multiple experimental platforms establish METTL21A as a critical regulator of HCC proliferation.

METTL21A promotes HCC metastasis

To validate the significant associations between METTL21A expression and metastasis-related clinicopathological features identified in our correlation analysis, we investigated the functional role of METTL21A in HCC cell migration and invasion. Using transwell assays, we demonstrated that METTL21A silencing significantly impaired the migratory and invasive capacities of Huh7, SNU182, and Hep3B cells (Fig. 4A and Supplementary Fig. 3D). Conversely, METTL21A overexpression markedly enhanced these malignant properties in both Huh7 and MHCC97H cells (Fig. 4B).

Fig. 4. METTL21A promotes HCC metastasis.

A Migration and invasion capacity of METTL21A-knockdown HCC cells assessed by transwell assays. Quantification represents mean cell counts from five random fields per sample. Scale bar: 200 μm. B Enhanced metastatic potential of METTL21A-overexpressing HCC cells demonstrated by transwell migration and invasion assays (n = 3). Cell counts were obtained from five representative microscopic fields. Scale bar: 200 μm. C Pulmonary metastasis evaluation in experimental metastasis models (n = 5). Left panel: macroscopic lung sections with H&E staining showing metastatic nodules (scale bar: 200 μm [overview], 50 μm [detail]). Right panel: quantitative analysis of lung metastatic burden. ^*p < 0.05, ^**p < 0.01, and ^***p < 0.001.

To further substantiate the pro-metastatic role of METTL21A in vivo, we established an experimental lung metastasis model by orthotopic injection of control or METTL21A-knockdown Huh7 cells into the hepatic subcapsular space of nude mice. The results revealed a significant decrease in both the number and size of pulmonary metastatic nodules in mice receiving METTL21A-depleted cells compared to the control group (Fig. 4C). These consistent findings from in vitro and in vivo experiments collectively demonstrate that METTL21A plays a crucial role in promoting HCC metastasis.

METTL21A interacts with and stabilizes BAG3

To elucidate the molecular mechanisms underlying METTL21A’s role in promoting HCC malignancy, we first employed multiple protein interaction prediction tools—including HitPredict, BioGRID, and HINT—to identify potential METTL21A-interacting proteins (Fig. 5A). Intersection analysis of the top ten highest-scoring interactors from each database revealed BAG3 as the sole common candidate, a protein previously implicated in tumor progression across various cancers including breast cancer¹⁷, prostate cancer¹⁸, ovarian cancer¹⁹, and liver cancer²⁰. IF staining confirmed the cytoplasmic colocalization of METTL21A and BAG3 in both Huh7 and SNU182 cells (Fig. 5B). Exogenous Co-IP assays in HEK293T cells co-transfected with Flag-tagged METTL21A and Myc-tagged BAG3 to validate their interaction (Fig. 5C). Importantly, endogenous Co-IP further demonstrated a physical interaction between METTL21A and BAG3 in HCC cells (Fig. 5D).

Fig. 5. METTL21A interacts with and stabilizes BAG3.

A Bioinformatics analysis identified the top 10 potential METTL21A-interacting proteins through integrated screening of HitPredict, BioGRID, and HINT databases. B Immunofluorescence confocal microscopy revealed significant co-localization (yellow signals) of endogenous METTL21A (red) and BAG3 (green) in Huh7 and SNU182 cells (scale bars: 20 µm). C Exogenous Co-IP in HEK293T cells transfected with Flag-METTL21A and Myc-BAG3 confirmed direct interaction, with anti-Flag/Myc antibodies showing reciprocal pulldown (IgG control negative). D Endogenous Co-IP in Huh7 and SNU182 cell lysates demonstrated a physiological interaction between METTL21A and BAG3 (IgG control negative). E The expression levels of BAG3 mRNA in control and METTL21A-silenced HCC cells were determined by RT-qPCR. F The expression levels of BAG3 protein in control and METTL21A-silenced HCC cells were assessed by Western blot analysis. G The BAG3 protein levels in the subcutaneous tumor model were measured using Western blotting. H The half-life of BAG3 was determined by CHX (20 μM) chase assay in METTL21A-silenced and control Huh7 cells. The BAG3 band intensities were measured over time and compared to the initial time point. ^*p < 0.05 and ^**p < 0.01.

To investigate whether METTL21A regulates BAG3 expression, we assessed BAG3 levels following METTL21A knockdown. While RT-qPCR analysis revealed no significant changes in BAG3 mRNA (Fig. 5E), Western blotting showed a marked reduction in BAG3 protein in METTL21A-silenced HCC cells (Fig. 5F). Consistent with this, xenograft tumors derived from METTL21A-depleted cells exhibited decreased BAG3 protein expression (Fig. 5G), suggesting post-translational regulation. To confirm this, we treated control and METTL21A-silenced Huh7 cells with CHX to block new protein synthesis. Strikingly, METTL21A knockdown significantly accelerated BAG3 protein degradation (Fig. 5H). These findings suggest that METTL21A modulates BAG3 protein stability.

METTL21A promotes HCC progression partially via the regulation of BAG3

Given prior evidence that BAG3 enhances HCC invasiveness²⁰, we next investigated whether BAG3 mediates METTL21A-driven HCC progression. We restored BAG3 expression in METTL21A-silenced HCC cells (Fig. 6A) and found that BAG3 overexpression rescued the impaired proliferation and metastatic capacity caused by METTL21A knockdown, as evidenced by CCK-8, colony formation, and transwell assays in Huh7 and SNU182 cells (Fig. 6B–D). These results confirm that BAG3 acts as a critical downstream effector of METTL21A in HCC progression.

Fig. 6. METTL21A promotes HCC progression partially via the regulation of BAG3.

A Western blot analysis confirmed BAG3 overexpression in METTL21A-silenced Huh7 and SNU182 HCC cell lines. B CCK-8 proliferation assays demonstrated that BAG3 overexpression partially rescued the growth inhibition caused by METTL21A knockdown. C Colony formation assays revealed that BAG3 overexpression restored the clonogenic potential of METTL21A-deficient HCC cells, with quantitative analysis of colony numbers. D Transwell migration and invasion assays showed that BAG3 overexpression mitigated the metastatic suppression induced by METTL21A silencing (scale bars: 200 μm), with cell counts from five representative fields. E Representative images of METTL21A and BAG3 IHC staining in the same HCC clinical sample were presented. Scale bars: 200 μm. F Correlation analysis revealed a significant positive relationship between METTL21A and BAG3 protein levels across HCC patient samples. G Kaplan–Meier survival curves stratified by combined METTL21A/BAG3 expression showed significantly worse OS and DFS for patients with high co-expression (n = 30) vs low co-expression (n = 28) (log-rank test). ^*p < 0.05, ^**p < 0.01, and ^***p < 0.001.

To evaluate the clinical relevance of BAG3, we analyzed TCGA data and observed that BAG3 mRNA was significantly upregulated in HCC tissues and associated with worse OS and DFS (Supplementary Fig. 4A, B). We further validated this relationship in our clinical cohort by performing IHC staining for METTL21A and BAG3 (Fig. 6E). Strikingly, METTL21A protein levels positively correlated with BAG3 expression (Fig. 6F), and patients with high co-expression of both proteins exhibited significantly poorer OS and DFS compared to those with low expression (Fig. 6G). These results further indicate the functional significance of BAG3 in METTL21A-regulated HCC progression.

METTL21A prevents the ubiquitination of BAG3

To elucidate how METTL21A regulates BAG3 stability, we first investigated the involvement of the ubiquitin-proteasome system. Treatment with the proteasome inhibitor MG132 abolished the differences in BAG3 protein levels between control and METTL21A-silenced HCC cells (Fig. 7A), suggesting proteasome-dependent degradation. Using HEK293T cells co-transfected with HA-Ub, Myc-BAG3, and Flag-METTL21A, we demonstrated that METTL21A significantly reduced BAG3 ubiquitination (Fig. 7B). This finding was corroborated by in vivo ubiquitination assays showing increased polyubiquitination of BAG3 in METTL21A-depleted Huh7 and SNU182 cells (Fig. 7C). Additionally, the specific types of polyubiquitination modifications on BAG3 proteins influenced by METTL21A-mediated ubiquitination were examined. HEK293T cells were co-transfected with Myc-BAG3 and each of the different ubiquitin variants (wild-type [WT], K11-, K48-, and K63-only ubiquitin-HA) with or without METTL21A silencing. Subsequently, the Myc-BAG3 proteins within the lysates were purified, and Western blot analysis was performed using an anti-HA antibody. The results showed that METTL21A knockdown mainly enhanced K48-linked polyubiquitination of BAG3 (Fig. 7D), the canonical signal for proteasomal degradation.

Fig. 7. METTL21A prevents the ubiquitination of BAG3.

A Western blot analysis was conducted to assess BAG3 levels in control and METTL21A-silenced Huh7 and SNU182 cells following a 6 h treatment with MG132 (10 μM). B HEK293T cells were co-transfected with Myc-BAG3 and HA-Ub plasmids, with or without Flag-METTL21A, for 48 h. Afterward, cells were treated with 10 μM MG132 for 6 h. Cell lysates were immunoprecipitated with anti-Myc antibodies and analyzed by Western blotting. C Control and METTL21A-silenced Huh7 and SNU182 cells were treated with MG132 (10 μM) for 6 h. Cell lysates were immunoprecipitated using anti-BAG3 antibodies and analyzed by Western blotting. D Control and METTL21A-silenced HEK293T cells were transfected with plasmids for 48 h, treated with MG132 (10 μM, 6 h), and cell lysates were immunoprecipitated with anti-Myc antibodies for Western blot analysis of Myc-BAG3 ubiquitination (WT, K11, K48, and K63-linked). E HEK293T cells were co-transfected with siRNA targeting METTL21A and Myc-BAG3 plasmids for 48 h. Then cell lysates were immunoprecipitated with anti-Myc antibodies and analyzed by Western blotting. F Evolutionary conservation analysis of BAG3 highlighting the critical K230 residue across species (red). G HEK293T cells were transfected with Myc-BAG3 or Myc-BAG3-K230R plasmids for 48 h. Cell lysates were immunoprecipitated using anti-Myc antibodies and analyzed via Western blotting, with IgG as the control. H HEK293T cells were transfected with plasmids for 48 h, treated with 10 μM MG132 for 6 h, and cell lysates were immunoprecipitated using anti-Myc antibodies for Western blot analysis. I Huh7 cells were transfected with Myc-BAG3 or Myc-BAG3-K230R plasmids for 48 h, and the half-life of Myc-BAG3 was assessed using a CHX (20 μM) chase assay. The Myc-BAG3 band intensities were measured over time and compared to the initial time point.

Given METTL21A’s methyltransferase activity and the established crosstalk between methylation and ubiquitination²¹, we examined whether METTL21A regulates BAG3 through methylation. Indeed, METTL21A silencing reduced lysine methylation of exogenous BAG3 (Fig. 7E). Bioinformatics analysis using PhosphoSitePlus identified lysine 230 (K230) as a conserved methylation site in BAG3 across species (Fig. 7F and Supplementary Fig. 4C). Site-directed mutagenesis (K230R) substantially decreased BAG3 methylation (Fig. 7G) and increased its polyubiquitination (Fig. 7H). Importantly, the K230R mutation accelerated BAG3 degradation, as shown by CHX chase experiments (Fig. 7I). These findings suggest that methylation of BAG3 at K230, mediated by METTL21A, may decrease polyubiquitination events and enhance the stability of the BAG3 protein.

The K230 site of BAG3 is important for its functional integrity

To determine the biological significance of BAG3 K230 methylation, we performed comprehensive functional assays comparing wild-type BAG3 (BAG3-WT) and the methylation-deficient K230R mutant (transfection efficiency was shown in Fig. 8A). In both Huh7 and SNU182 cells, overexpression of BAG3-WT significantly enhanced cellular proliferation (Fig. 8B), colony formation capacity (Fig. 8C), and migratory and invasive potential (Fig. 8D), consistent with its established oncogenic role. Importantly, cells expressing the BAG3-K230R mutant exhibited markedly attenuated tumor-promoting effects across all assays (Fig. 8B–D), indicating that K230 of BAG3 is important for its functional integrity.

Fig. 8. The K230 site of BAG3 is critical for its function.

A Western blot analysis assessed BAG3 protein levels in Huh7 and SNU182 cells overexpressing Myc-BAG3 or Myc-BAG3-K230R. B The CCK-8 assay evaluated cell proliferation in Huh7 and SNU182 cells overexpressing Myc-BAG3 or Myc-BAG3-K230R. C Representative images and quantitative analyses assessed colony formation in Huh7 and SNU182 cells with Myc-BAG3 or Myc-BAG3-K230R overexpression. D Representative images and quantitative analyses were conducted for the transwell migration and invasion assays in Huh7 and SNU182 cells overexpressing Myc-BAG3 or Myc-BAG3-K230R. The cells were counted under a microscope in five randomly selected microscopic fields. Scale bars: 200 μm. ^*p < 0.05, ^**p < 0.01, and ^***p < 0.001.

METTL21A attenuates the association between TRIM21 and BAG3

To investigate the molecular mechanism underlying METTL21A-mediated stabilization of BAG3, we first performed Co-IP coupled with mass spectrometry analysis in HEK293T cells expressing Flag-METTL21A. This proteomic screening identified TRIM21 as the sole E3 ubiquitin ligase interacting with METTL21A, a finding that was subsequently confirmed by endogenous Co-IP assays (Supplementary Fig. 4D). Further characterization of this interaction network revealed that TRIM21 not only physically associates with BAG3, as demonstrated by reciprocal Co-IP experiments (Fig. 9A), but also colocalizes with BAG3 in the cytoplasmic compartment, as evidenced by IF staining (Fig. 9B). Moreover, we detected whether METTL21A affected the interaction between BAG3 and TRIM21. As anticipated, silencing of METTL21A expression strengthened their association (Fig. 9C), whereas METTL21A overexpression significantly attenuated this interaction (Fig. 9D).

Fig. 9. METTL21A attenuates the association between TRIM21 and BAG3.

A Endogenous TRIM21 and BAG3 proteins were immunoprecipitated using specific antibodies and subsequently analyzed through Western blotting. B Immunofluorescence confocal microscopy revealed significant co-localization of endogenous TRIM21 and BAG3 in Huh7 and SNU182 cells (scale bars: 20 µm). C Cell lysates from control and METTL21A-silenced Huh7 cells were immunoprecipitated with anti-BAG3 antibody and analyzed by Western blotting. D Cell lysates from control and METTL21A-overexpressing Huh7 cells were immunoprecipitated with an anti-BAG3 antibody and analyzed via Western blotting. E BAG3 protein levels in TRIM21-silenced HCC cells were assessed by Western blotting. F Western blot analysis assessed BAG3 levels in TRIM21-silenced HCC cells following treatment with MG132 (10 μM) for 6 h. G The half-life of BAG3 was determined in TRIM21-silenced Huh7 cells using a CHX (20 μM) chase assay. The BAG3 band intensities were measured over time and compared to the initial time point. H Cells with TRIM21 silence were treated with MG132 (10 μM) for 6 h, then cell lysates were subjected to immunoprecipitation using an anti-BAG3 antibody and analyzed by Western blotting. I Control and TRIM21-silenced HEK293T cells were transfected with plasmids for 48 h, treated with MG132 (10 μM, 6 h), and cell lysates were immunoprecipitated with anti-Myc antibodies for Western blot analysis of Myc-BAG3 ubiquitination (WT, K11, K48, and K63-linked). J Western blot analysis of BAG3 protein in Huh7 cells co-transfected with siMETTL21A and/or siTRIM21 for 48 h.

To functionally characterize TRIM21’s role as an E3 ubiquitin ligase for BAG3, we performed siRNA-mediated TRIM21 knockdown in Huh7 and SNU182 cells. This intervention resulted in a marked accumulation of BAG3 protein (Fig. 9E), which was abrogated by co-treatment with the proteasome inhibitor MG132 (Fig. 9F), strongly suggesting proteasomal degradation as the primary route for TRIM21-mediated BAG3 regulation. Supporting this conclusion, CHX chase assays revealed a significant extension of BAG3 protein half-life following TRIM21 depletion (Fig. 9G).

Subsequently, we investigated the ubiquitination modification of BAG3 mediated by TRIM21. In vivo ubiquitination assays demonstrated a substantial reduction in BAG3 polyubiquitination upon TRIM21 knockdown (Fig. 9H). Although TRIM21 has been reported to mediate multiple ubiquitin linkage types (K11, K48, and K63)^22–24, we found that TRIM21 mainly promotes K48-linked polyubiquitination of BAG3 (Fig. 9I)—the canonical signal for proteasomal degradation, and precisely the same modification pattern was affected by METTL21A. Most significantly, TRIM21 knockdown rescued the decrease in BAG3 protein levels caused by METTL21A silencing (Fig. 9J), providing evidence that TRIM21 operates downstream of METTL21A in this regulatory cascade. These findings collectively establish a novel molecular mechanism wherein METTL21A stabilizes BAG3 by inhibiting TRIM21-mediated K48-linked polyubiquitination, thereby preventing BAG3 proteasomal degradation and promoting HCC progression.

Identifying Sop as a novel METTL21A inhibitor with anti-HCC activity

Given the pivotal role of METTL21A in HCC growth and metastasis, this methyltransferase represents a promising therapeutic target. However, no specific METTL21A inhibitors have been identified to date. To address this gap, we performed structure-based virtual screening of 19,100 compounds from the MCE bioactive library, targeting the S-adenosylhomocysteine (SAH)-binding pocket of human METTL21A (Fig. 10A).

Fig. 10. Identification of Sop as a novel METTL21A inhibitor with anti-HCC activity.

A The workflow for molecular docking and the structural characterization of METTL21A protein. B The docking model of METTL21A in complex with Sop. C The SPR assay evaluated the interaction between Sop and recombinant METTL21A protein, with the dissociation constant (K_D) value presented. RU denotes the response unit. D Western blot analysis of METTL21A and BAG3 expression in Huh7 and PLC/PRF/5 cells following a 72-h treatment with Sop at specified concentrations. E The 72-h half-maximal inhibitory concentration (IC₅₀) of Sop was determined in Hep3B, Huh7, and PLC/PRF/5 cell lines. F The cytotoxic effects of Sop on Huh7 and PLC/PRF/5 cells were assessed using the CCK-8 assay. G Representative images and quantitative analyses illustrate the colony formation of Huh7 and PLC/PRF/5 cells treated with various concentrations of Sop. H Huh7 and PLC/PRF/5 cells were treated with various Sop concentrations for 72 h, and then Transwell assays assessed cell migration and invasion. Representative images and quantitative analyses were presented. I Subcutaneous tumors from Huh7 xenograft mice treated with 80 mg/kg Sop or vehicle control were imaged, with quantitative analyses of tumor growth trajectories and final tumor weights presented for the xenograft model (n = 5). ^*p < 0.05, ^**p < 0.01, and ^***p < 0.001.

Through this screening approach, we identified six candidate compounds with the highest predicted binding affinities (Supplementary Fig. 5A). Initial cytotoxicity assays in Hep3B cells—which exhibit the highest METTL21A expression among HCC cell lines—revealed that Sop most potently inhibited HCC cell proliferation, demonstrating the lowest half-maximal inhibitory concentration (IC₅₀) of all tested compounds (Supplementary Fig. 5B). Molecular docking analysis suggested that Sop binds tightly to METTL21A, forming five hydrogen bonds and three π–π interactions, with a high docking score of −11.447 (Fig. 10B). SPR experiments further confirmed specific binding between Sop and METTL21A, yielding a dissociation constant (K_D) of 122 μM (Fig. 10C). Notably, Sop treatment induced dose-dependent downregulation of both METTL21A and its downstream effector BAG3 (Fig. 10D), supporting its potential as a METTL21A-targeting therapeutic agent.

Additionally, we assessed its cytotoxic effects across multiple HCC cell lines. Among the tested cell lines, Hep3B cells exhibited the highest METTL21A protein expression (Supplementary Fig. 3A), and consistently demonstrated the greatest sensitivity to Sop treatment, as evidenced by its significantly lower IC₅₀ value compared to Huh7 and PLC/PRF/5 cells (Fig. 10E). Subsequent investigations into the anticancer properties of Sop were conducted using CCK-8 assays, clonogenic assays, and transwell assays. Treatment with Sop significantly suppressed HCC cell proliferation (Fig. 10F and Supplementary Fig. 5C), colony formation (Fig. 10G), and metastatic potential (Fig. 10H). The anti-tumor efficacy of Sop was further validated in vivo using a xenograft mouse model. Administration of Sop resulted in significant tumor growth inhibition compared to vehicle-treated controls (Fig. 10I). Taken together, these findings demonstrate that pharmacological targeting of METTL21A by Sop effectively attenuates HCC progression both in vitro and in vivo, highlighting its potential as a novel therapeutic strategy against HCC.

Discussion

Advancing our understanding of HCC pathogenesis, this study reveals METTL21A as a clinically significant oncoprotein through integrated analysis of TCGA and GEO databases combined with clinical tissue validation. We demonstrate that METTL21A is consistently upregulated in HCC patients, with elevated expression strongly correlating with metastatic progression, increased serum HBV DNA levels, and unfavorable prognosis. Functional characterization establishes METTL21A as a bona fide oncogene driving both HCC growth and metastasis, positioning it as both a valuable prognostic biomarker and a promising therapeutic target. Notably, our pan-cancer analysis extends these findings beyond HCC, revealing widespread METTL21A overexpression across multiple human malignancies (Supplementary Fig. 1E), suggesting its potential involvement in a broader oncogenic network.

While the regulatory mechanisms controlling METTL21A expression have remained elusive, our systematic screening of five transcription factor databases initially identified CREB1, HDAC2, CTCF, and MAZ as potential regulators, with subsequent experimental validation confirming CTCF as the bona fide transcriptional regulator of METTL21A. As a master genome organizer and chromatin architectural protein, CTCF maintains three-dimensional chromatin organization through its interactions with multiple protein complexes and participates in diverse nuclear processes, including chromatin looping, transcriptional regulation, and alternative splicing²⁵. Notably, CTCF overexpression in HCC correlates with aggressive tumor behavior and poor clinical outcomes, where it drives malignant phenotypes including enhanced proliferation, motility, and invasiveness²⁶. Here, METTL21A has been identified as a novel downstream gene of CTCF.

BAG3, a multifunctional molecular chaperone regulator ubiquitously expressed across human tissues, orchestrates diverse cellular processes through its modular domain architecture that mediates interactions with heat shock proteins (HSPs), Bcl2 family members, and key anti-apoptotic factors. BAG3 is commonly overexpressed in various human cancers, which restricts the mitochondria-dependent apoptotic pathway through its coupling with Bcl2 and induces macroautophagy by co-chaperoning large and small HSPs²⁷. PTMs, including ubiquitination, phosphorylation, and acetylation, critically regulate BAG3’s stability and function, as exemplified by SIRT2-mediated deacetylation, which promotes PARP1 ubiquitination to mitigate oxidative damage²⁸, PKCδ-dependent Ser187 phosphorylation that drives EMT in thyroid cancer²⁹, and the opposing regulatory effects of FBXO22-mediated Ser377 phosphorylation³⁰ vs USP26-mediated lysine 450 deubiquitination¹⁷ on BAG3 protein turnover. Our study extends this PTM landscape by identifying METTL21A-mediated methylation at K230 as a novel regulatory mechanism that critically determines BAG3 protein stability, with the K230R mutation leading to accelerated ubiquitin-proteasomal degradation and consequent attenuation of BAG3’s oncogenic potential in HCC growth and metastasis. These findings establish lysine methylation as a functionally significant PTM of BAG3 and reveal the METTL21A-BAG3 axis as a key regulatory pathway in hepatocellular carcinogenesis.

Lysine methylation serves as a pivotal PTM that orchestrates protein function through multiple mechanisms, including modulation of other PTMs (e.g., phosphorylation, acetylation, and ubiquitination) and regulation of protein–protein interactions²¹. Emerging evidence suggests a complex interplay between methylation and protein stability, as demonstrated by Pang et al.³¹, who observed significantly prolonged half-lives for methylated proteins, with bioinformatic analysis predicting that 43% of methylated lysine sites represent potential ubiquitination targets, implying a competitive relationship between these modifications. This observation suggests that methylation may enhance protein stability by competing with ubiquitination. For instance, the methylation of GLI1 by MEP50/PRMT5 enhances the stability of the GLI1 protein by preventing its ubiquitination via the ITCH/NUMB pathway³². Similarly, methylation mediated by SETD2 obstructs TRIM21-mediated ubiquitination and subsequent degradation of FIP200³³. Additionally, PRMT5-mediated methylation of KLF5 contributes to its stabilization by counteracting Fbw7γ-mediated ubiquitination of KLF5³⁴. Consequently, methylation may be regarded as a modulator of ubiquitination. Nonetheless, this regulatory relationship demonstrates remarkable context-dependency, as evidenced by cases where lysine methylation paradoxically enhances ubiquitination of proteins including UHRF1, DNMT1³⁵, E2F1³⁶, RORα³⁷, and NF-κB³⁸. Our study extends this understanding by revealing that METTL21A-mediated methylation of BAG3 at K230 specifically inhibits its ubiquitination, thereby establishing a novel stabilization mechanism for this critical oncoprotein in HCC pathogenesis.

Delving deeper into the mechanistic basis of METTL21A-mediated regulation of BAG3 stability, we identified TRIM21 as the key E3 ubiquitin ligase responsible for BAG3 ubiquitination. As a member of the tripartite motif (TRIM) family characterized by its N-terminal RING finger domain and C-terminal PRY-SPRY structural domain, TRIM21 exhibits context-dependent roles in cancer pathogenesis. While some studies report tumor-suppressive functions of TRIM21 in HCC, particularly in the context of NASH-related hepatocarcinogenesis where its deficiency exacerbates disease progression³⁹, other investigations consistently demonstrate its oncogenic potential through positive correlations between TRIM21 overexpression and both HCC incidence and poor patient survival^40,41, suggesting etiology-dependent functional heterogeneity. Our findings establish TRIM21 as the specific E3 ligase mediating BAG3 ubiquitination and degradation, while suggesting that METTL21A-mediated methylation of BAG3 at K230 may create a structural impediment that disrupts TRIM21 binding. This methylation-dependent steric hindrance model provides a novel mechanistic explanation for how METTL21A-mediated PTMs govern BAG3 protein stability by competitively inhibiting its recognition by the ubiquitin-proteasome system, thereby adding another layer of complexity to the growing understanding of crosstalk between methylation and ubiquitination in cancer biology.

Building on our identification of METTL21A as a promising therapeutic target in HCC, we successfully discovered Sop through structure-based virtual screening as a potent and specific METTL21A inhibitor that effectively reduces METTL21A protein levels. Sop has previously demonstrated hepatoprotective effects in various liver disease models. For instance, it is effective in the treatment of hepatic steatosis in murine models⁴², exhibits protective effects against fructose-induced liver injury⁴³, and mitigates autoimmune-mediated liver damage⁴⁴. Nevertheless, the potential application of Sop in anti-tumor therapies remains underexplored. Our work represents the first demonstration of Sop’s anti-tumor efficacy in HCC, showing that it potently inhibits tumor cell proliferation, migration, and invasion in vitro while suppressing tumor growth in vivo. Mechanistically, Sop explicitly targets and downregulates the protein expression of METTL21A and its downstream effector, BAG3. These findings not only establish Sop as a promising candidate for HCC therapy but also validate METTL21A as a therapeutically actionable target, providing a strong rationale for further development of METTL21A-targeted strategies for HCC treatment. However, the anti-HCC effects of Sop may involve METTL21A-independent pathways, as suggested by its relatively low binding affinity (K_D = 122 μM). Given that natural products frequently exert multi-target and multi-pathway effects—attributed to their structural diversity and complex biosynthetic origins—Sop’s mechanism of action likely extends beyond a single target. Supporting this notion, our unpublished network pharmacology and molecular docking analyses identified several potential interaction partners of Sop, including TNF, IL-6, EGFR, and VEGFA. Notably, EGFR exhibited a significantly higher binding affinity for Sop (K_D = 56.1 μM) compared to METTL21A, further implicating a multi-targeted mechanism in Sop-mediated HCC suppression.

Although our study provides compelling evidence that METTL21A directly interacts with BAG3 and protects it from TRIM21-mediated ubiquitination, several important limitations should be acknowledged:

First, the uncharacterized methyltransferase domain of METTL21A represents a knowledge gap in our mechanistic understanding. The absence of defined catalytically active sites precludes definitive conclusions regarding the dependence of HCC progression on METTL21A’s methyltransferase activity. This structural ambiguity introduces several interpretive constraints: 1) we cannot establish precise structure-function relationships between specific domains and methylation activities; 2) catalytic residues and potential allosteric regulation sites remain unidentified; and 3) substrate recognition mechanisms cannot be fully elucidated. While our biochemical assays confirm METTL21A’s methyltransferase function, the unresolved domain architecture leaves open possibilities of non-canonical catalytic processes or cooperative interactions with other modifying enzymes. Future structural studies using cryo-EM analysis of METTL21A-cofactor complexes and comparative modeling with seven-β-strand methyltransferases (7BS-MTases) will be crucial for identifying enzymatic active sites and advancing mechanistic understanding.

Second, our understanding of the METTL21A-BAG3 interaction remains incomplete. The precise binding interfaces require systematic mapping through comprehensive domain deletion mutagenesis of both proteins. Additionally, the current lack of a site-specific anti-BAG3-K230me antibody precludes direct clinical validation of this methylation event. The functional interplay between METTL21A-mediated methylation and BAG3 ubiquitination sites also demands further investigation to fully delineate this regulatory axis.

In summary, we performed an extensive analysis to determine the clinical significance and functional role of METTL21A (Fig. 11). Our findings suggest that METTL21A-mediated methylation inhibits the interaction between TRIM21 and BAG3, thereby suppressing the ubiquitination of BAG3 and preventing its degradation. Sop is identified as a novel inhibitor for METTL21A. METTL21A may serve as a prognostic predictor and a potential therapeutic target for HCC.

Fig. 11.

A schematic diagram drawn by Figdraw reveals the mechanism of METTL21A facilitating HCC progression.

Methods

Tissue collection

Clinical HCC tissue samples were obtained from patients undergoing hepatectomy for HCC at the Zhongshan Hospital of Xiamen University. These patients had not received preoperative interventional therapy, chemotherapy, or targeted therapy, and their clinicopathologic data and follow-up records were available for analysis. Informed consent was obtained from the participating patients. All methods were carried out according to the Declaration of Helsinki (October 2013). The Ethics Committee of Zhongshan Hospital of Xiamen University approved the study.

Immunohistochemistry (IHC) analysis

IHC analysis was performed on paraffin-embedded HCC tissue sections and mouse subcutaneous tumor specimens using standard protocols with the following primary antibodies: anti-METTL21A (1:1000 dilution, Thermo Fisher Scientific [Invitrogen], MA5-25723), anti-BAG3 (1:1000, Proteintech, 10599-1-AP), and anti-Ki67 (Fuzhou Maixin Biotechnology, MAB-0672). Following staining, sections were examined and imaged under a light microscope. Two independent investigators blinded to clinical data evaluated protein expression using a semi-quantitative scoring system that assessed both the percentage of positively stained tumor cells (scored as 0 for <5%, 1 for 5–25%, 2 for 26–50%, 3 for 51–75%, and 4 for >75%) and staining intensity (graded as 0 for negative, 1 for weak, 2 for moderate, and 3 for strong). The final IHC score was determined by multiplying the intensity score by the percentage score, with total scores of 0–6 classified as low expression and 7–12 as high expression.

Cell culture

The normal human hepatocytes (THLE-2) were cultured in THLE-2 complete medium (Meisen CTCC, China). Human HCC cell lines (Huh7, SNU182, Hep3B, LM3, Li7, MHCC97H, SNU449, and SK-Hep-1) and HEK293T cells were obtained from Cellcook Biotechnology (Guangzhou, China), while PLC/PRF/5 cells were sourced from the American Type Culture Collection (ATCC, USA). All cell lines were authenticated by short tandem repeat (STR) profiling. HEK293T, Huh7, PLC/PRF/5, LM3, Li7, MHCC97H, and SK-Hep-1 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, USA). Hep3B cells were cultured in Minimum Essential Medium (MEM; Cellcook, China). SNU182 and SNU449 cells were grown in Roswell Park Memorial Institute (RPMI)-1640 medium (Gibco). All media were supplemented with 10% fetal bovine serum (FBS; Vivacell Biosciences, China) and 1% penicillin-streptomycin (Gibco).

Transfection

For gene manipulation experiments, METTL21A- and TRIM21-targeting siRNAs along with a negative control scrambled siRNA (siNC) were commercially synthesized by Sangon Biotech (Shanghai, China; specific sequences provided in Supplementary Table 1). siRNA transfection was performed using Lipofectamine RNAiMAX transfection reagent (Invitrogen, USA) according to the manufacturer’s recommended protocol. For promoter activity analysis, the METTL21A promoter sequence was chemically synthesized by Sangon Biotech and subsequently cloned into the pGL3-Basic vector to construct luciferase reporter plasmids. Plasmid transfections were carried out employing Lipofectamine 3000 transfection reagent (Invitrogen, USA) following standard procedures.

Construction of stable cell lines

Overexpression plasmids (pLenti-Puro, pLenti-METTL21A-Flag, pLenti-BAG3-Myc, pLenti-BAG3-K230R-Myc) and knockdown plasmids (pLV-shMETTL21A, pLV-Puro) were obtained from the Public Protein/Plasmid Library (PPL, Jiangsu, China). Scrambled shRNA was used as a negative control (shNC). The target sequences of indicated shRNA were: shNC: 5′-GTTCTCCGAACGTGTCACGTT-3′; shMETTL21A (shM21A): 5′-GAAGAAACATTCACAGATCTT-3′. HEK293T cells were transfected using Hieff Trans Liposomal Transfection Reagent (Yeasen Biotechnology, Shanghai, China). Lentiviral particles were produced by co-transfecting psPAX2 and pMD2.G packaging plasmids into HEK293T cells. Viral supernatant was collected at 48 h and 72 h post-transfection and used to infect target cells with 8 μg/mL polybrene (Solarbio, China). Stable cell lines were selected with 2 μg/mL puromycin (Solarbio) for 7 days.

Immunofluorescence (IF) staining

Cells were cultured at a density of 2 × 10³ cells/well in 24-well culture plates for 24 h. Subsequently, the cells were fixed with 4% paraformaldehyde for 30 min at room temperature and then disrupted with 0.2% Triton X-100 for 10 min. Following a 1 h blockade with 5% FBS, the cells were treated with the appropriate primary antibody (Supplementary Table 2) at 4 °C overnight. Following a phosphate-buffered saline (PBS) wash, the cells were incubated with fluorescent secondary antibodies, Goat anti-Rabbit IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 594 (Invitrogen) or Goat anti-Mouse IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488 (Invitrogen) for 1 h. Subsequently, the cells were incubated with 4’,6-diamidino-2-phenylindole (DAPI, Invitrogen) for nuclear staining. Finally, Images were captured using a ZEISS LSM 780 confocal microscope with a 63× oil-immersion objective.

Cell proliferation and colony formation assays

For proliferation analysis, HCC cells (1–2 × 10³ cells/well in 96-well plates) were treated with the Cell Counting Kit-8 (CCK-8) reagent (Dojindo Laboratories, Japan) at the indicated time points. After incubation for 1 h, the optical density (OD) of cells at 450 nm was determined using a microplate spectrophotometer (Tecan). For colony formation assays, 1 × 10³ cells/well were plated in 6-well plates and cultured for 14 days. Colonies (>50 cells) were fixed with 4% paraformaldehyde for 30 min, stained with 0.3% crystal violet (Beyotime Biotechnology, Shanghai, China) for 30 min. After that, the colonies were counted.

Transwell assay

Cell migration and invasion assays were performed using transwell chambers (8 μm pore size, Corning) with distinct experimental setups. For migration analysis, HCC cells suspended in 200 μL serum-free medium were seeded in the upper chamber. For invasion assessment, chambers were pre-coated with Matrigel (BioCoat® Matrigel® Invasion Chambers, Corning). The lower chambers contained 500 μL complete medium with 10% FBS as a chemoattractant. Following 24–48 h incubation at 37 °C, migrated/invaded cells on the lower membrane surface were fixed with 4% paraformaldehyde (PFA) for 15 min and stained with 0.3% crystal violet for 30 min. After PBS washing to remove excess stain, membranes were imaged under an inverted microscope (100× magnification). Quantitative analysis was performed by counting cells in five randomly selected fields per membrane, avoiding the outer 10% peripheral area to eliminate edge effects.

Western blotting

Cells were lysed with RIPA lysis buffer (Beyotime) containing protease inhibitor cocktail (LABLEAD, China). Protein concentration was determined using a BCA Protein Assay Kit (Thermo Fisher). Subsequently, lysates were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes (Millipore). After blocking with 5% nonfat milk (Bio-Rad) for 1 h at room temperature, membranes were incubated with primary antibodies overnight at 4 °C, followed by HRP-conjugated secondary antibodies and visualization using the ECL Substrate (Millipore). The details of the antibodies are shown in Supplementary Table 2.

Co-immunoprecipitation (Co-IP)

Cells were lysed using immunoprecipitation (IP) lysis buffer (Beyotime). Following the standard protocol, lysates were incubated overnight at 4 °C with the specific antibodies (Supplementary Table 2) and Dynabeads™ Protein G (Invitrogen). Subsequently, immunoprecipitates were washed three times with lysis buffer and prepared in SDS-PAGE sampling buffer for subsequent Western blot analysis.

Methylation and ubiquitination assays

To analyze protein methylation states, cell lysates were prepared following the standard IP protocol, followed by immunoblotting with a pan-methyl-lysine antibody (see Supplementary Table 2 for antibody details). For ubiquitination assays, cells were pretreated with the proteasome inhibitor MG132 (10 µM, MedChemExpress [MCE], #HY-13259) for 6 h prior to lysis to prevent protein degradation. The resulting lysates were processed according to the IP procedure and immunoblotted with either anti-ubiquitin or anti-HA antibodies (detailed in Supplementary Table 2) to detect ubiquitinated proteins.

Protein half-life analysis

Cells in the exponential growth phase were treated with 20 μM cycloheximide (CHX, MCE, #HY-12320) for varying durations to inhibit protein synthesis. Subsequently, protein degradation was analyzed by Western blotting.

Dual-luciferase reporter assay

HCC cells were inoculated into 24-well plates at an appropriate density and cultured until they reached 70% confluence. Subsequently, the pGL3-basic or pGL3-basic-METTL21A promoter-luciferase reporter genes were co-transfected into the cells with pRL-TK (Promega; USA). After 48 h, luciferase activity was measured by the Dual-Luciferase^® Reporter System (Thermo). Relative luciferase activity (RLA) was calculated by normalizing firefly to Renilla luciferase activity.

RNA extraction and reverse transcription quantitative PCR (RT-qPCR)

According to standard protocols, total RNA was extracted from HCC cells using TRIzol™ Reagent (Invitrogen). cDNA was obtained by reverse transcription using HiScript® III All-in-one RT SuperMix (Vazyme). RT-qPCR was performed using Taq Pro Universal SYBR qPCR Master Mix (Vazyme) according to the manufacturer’s instructions. Data were analyzed using the 2^−ΔΔCt method. Primers are listed in Supplementary Table 1.

Chromatin immunoprecipitation (ChIP) assay

ChIP assays were performed using the EZ-Magna ChIP™ HiSens Kit (Millipore, #17-10461) according to the following optimized protocol: Cells were cross-linked with 1% formaldehyde for 10 min at room temperature, followed by quenching with 0.125 M glycine for 5 min. After washing with ice-cold PBS, cells were harvested by scraping and resuspended in Nuclei Isolation Buffer supplemented with Protease Inhibitor Cocktail III. Nuclei were released by incubating on ice for 15 min with intermittent vortexing, then pelleted by centrifugation (800×g, 5 min, 4 °C). Nuclear pellets were lysed in SCW Buffer containing protease inhibitors and sonicated to generate 200–1000 bp chromatin fragments. Chromatin lysates were immunoprecipitated overnight at 4 °C with 5 μg of either anti-CTCF antibody (Cell Signaling Technology, #3418) or control IgG (Proteintech, #2729) coupled to Magna ChIP™ Protein A/G Magnetic Beads. Precipitated DNA was purified and analyzed by qPCR using primers specific for target regions (Supplementary Table 1), with GAPDH exon 3 serving as the negative control region.

Molecular docking

We conducted structure-based virtual screening to identify potential METTL21A inhibitors using the following workflow: First, the crystal structure of METTL21A (PDB ID: 4LEC) was retrieved from the RCSB Protein Data Bank and prepared using Schrödinger Maestro 11.4. The docking grid was centered on the catalytic pocket encompassing key residues (Trp119, Trp125, Arg89, Arg95, Ala137, Ala143). Virtual screening was performed against the MCE Bioactive Compound Library (containing 19,100 molecules) using a hierarchical docking strategy with Schrödinger Glide¹: Initial screening in High-Throughput Virtual Screening (HTVS) mode retained the top 15% scoring compounds²; Secondary screening in standard precision (SP) mode selected the top 15% from the HTVS output³; Final evaluation in extra precision (XP) mode identified the six highest-ranking compounds for experimental validation. This multi-stage approach ensured both computational efficiency and docking accuracy.

Surface plasmon resonance (SPR) assay

The binding kinetics between the recombinant Human METTL21A protein (Fine Biotech, China) and Sop (HY-N0423; MCE) were analyzed by SPR assay. Following standard protocols, experiments were conducted at 25 °C using an OpenSPR™ system (Nicoya) with carboxyl-coated sensor chips immobilized with METTL21A proteins. Binding affinity was analyzed using TraceDrawer software (Ridgeview Instruments AB, Sweden).

Flow cytometry

For cell cycle analysis, cells were harvested by trypsinization and pelleted by centrifugation. The cell pellets were fixed in 75% ethanol at 4 °C overnight, then treated with RNase A and propidium iodide (PI) using the Cell Cycle Assay Kit (Dojindo) according to the manufacturer’s protocol. Cell cycle distribution was determined by flow cytometry (BD FACSCanto II) and analyzed using Kaluza software (Beckman Coulter).

Apoptosis was assessed using the Annexin V-FITC/PI Apoptosis Detection Kit (Dojindo). Briefly, harvested cells were washed twice with cold PBS and resuspended in 100 μL Annexin V binding buffer. After incubation with 5 μL Annexin V-FITC and 5 μL PI solution for 15 min at room temperature in the dark, 400 μL binding buffer was added. Samples were immediately analyzed by flow cytometry using Kaluza software.

Animal studies

For in vivo tumorigenesis studies, five-week-old male BALB/c nude mice were maintained under specific-pathogen-free (SPF) conditions. To assess METTL21A’s role in tumor growth, 2 × 10⁶ Huh7 cells (either control or METTL21A-knockdown) suspended in 100 μL DMEM basal medium were subcutaneously injected into the right dorsal flank (n = 5 per group). Tumor dimensions were measured every 2–5 days using digital calipers, with volume calculated as V = ½ × (length × width²). After 3 weeks, mice were euthanized by cervical dislocation. Excised tumors were weighed, photographed, and fixed in 10% neutral-buffered formalin for 24 h before paraffin embedding.

For metastasis evaluation, 2 × 10⁶ cells were surgically implanted into the hepatic subcapsular space of nude anesthetized mice (n = 5 per group). After 8 weeks, lung tissues were harvested, fixed in formalin, and paraffin-embedded. Serial 5-μm sections were stained with hematoxylin-eosin (H&E) following standard protocols. Metastatic nodules were quantified under a microscope by two independent pathologists blinded to experimental groups, with metastasis incidence and burden (number of nodules per lung section) recorded. All procedures were approved by Xiamen University’s Institutional Animal Care and Use Committee and complied with ARRIVE guidelines.

Statistical analysis

The statistical analyses used GraphPad Prism 9.0 and SPSS software. Group differences were assessed by Student’s t-test (two groups) or one-way Analysis of Variance (ANOVA, for multiple groups). A Pearson’s chi-square test was utilized to ascertain the relationship between METTL21A expression and clinical characteristics in HCC tissues. Spearman’s rank correlation was used to examine the correlation between METTL21A and BAG3 expression. A Kaplan–Meier analysis was conducted to assess survival, and the log-rank test was employed to verify the significance of the findings. All experiments were performed in triplicate (three independent experiments). The data are presented as mean ± standard deviation (SD). Differences were considered significant for p values less than 0.05.

Author contributions

Fuqiang Wang, Zhenyu Yin and Chengrong Xie conceived and designed the experiments; Ping Zhan, Yizhe Cheng, Jing Lu, Yingcan Wu, Xijun Chen and Jing Wen performed all the experiments with help from Xiaoqin Chi, Changhong Luo, Yiwei Peng; Chengrong Xie and Ping Zhan analyzed data and wrote the manuscript. All authors read and approved the final manuscript.

Data availability

The data supporting the findings of this study are available from the corresponding authors upon reasonable request.

Code availability

The R scripts used for gene expression analysis and GSEA-based visualization were developed using standard packages (including DESeq2 and clusterProfiler) and are available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41698-025-01021-5.