RASGEF1B suppresses hepatocellular carcinoma through the ALDH7A1/Betaine/SNAI1 metabolic‒epigenetic axis

Abstract

Background

Metabolic‒epigenetic crosstalk critically orchestrates hepatocellular carcinoma (HCC) pathogenesis. Deciphering the precise mechanism underlying epigenetic remodeling and metabolic reprogramming in HCC may lead to novel treatment paradigms, however, the key mechanisms remain elusive.

Methods

RT-qPCR, western blotting and tissue microarrary Immunohistochemistry were used to detect the expression of RasGEF domain family member 1B (RASGEF1B) in HCC and normal liver tissues. Transcriptome sequencing and high-resolution untargeted metabolomics were integrated to identify the downstream regulatory mechanism through which RASGEF1B inhibited the HCC progression. Epigenetic regulation was investigated using methylation-specific PCR and luciferase reporter assays. Bioinformatic prediction and molecular docking suggested a functional interplay among RASGEF1B, ALDH7A1, and BMI1, which was experimentally confirmed through coimmunoprecipitation, GST pull-down, and immunofluorescence assays. Protein stability and ubiquitination status of ALDH7A1 were examined using cycloheximide, immunoprecipitation assay, and an in vitro reconstituted ubiquitination system.

Results

In this study, the antitumor role of RASGEF1B was confirmed in vitro and in vivo. Transcriptomic profiling revealed that RASGEF1B overexpression significantly reduced the snail family transcriptional repressor 1 (SNAI1), a master regulator of the epithelial-mesenchymal transition. Untargeted metabolomics revealed that RASGEF1B promoted SNAI1 DNA methylation through Betaine-mediated methionine metabolic reprogramming. Further analysis confirmed that RASGEF1B competitively protected the ALDH7A1 protein from BMI1-dependent ubiquitination, thereby elevating cellular Betaine levels in HCC.

Conclusions

This study revealed that RASGEF1B inhibited SNAI1 to suppress HCC through metabolite‒epigenetic crosstalk. Our findings potentially offer a new perspective on the classical RAS signaling framework, uncovering a metabolic‒epigenetic axis as an innovative therapeutic approach for improving clinical outcomes in patients with HCC.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-026-07785-z.

Background

Hepatocellular carcinoma (HCC) is a highly malignant neoplasm of the digestive system. According to recent global cancer statistics, liver cancer was the third leading cause of cancer-related death, accounting for approximately 757,948 deaths annually [1]. Although substantial progress has been made in systemic treatment strategies, ranging from surgical techniques to chemotherapy, immunotherapy, and targeted therapies, the therapeutic outcomes for patients with HCC continues to be disappointing [2, 3]. Therefore, systematic research is crucial for improving HCC treatment [4].

The rat sarcoma (RAS) gene was initially identified in the Harvey rat sarcoma virus (HRAS) and Kirsten rat sarcoma virus (KRAS). In 1982, HRAS was identified in the human bladder cancer cell line T24/EJ, marking the identification of the first human oncogene [5]. In the past several decades, the RAS signaling pathway has been reported to play an important role in physiopathological processes in eukaryotes. Disruption of homeostasis of the RAS signaling pathway leads to various diseases, particularly in tumorigenesis and progression. In the canonical RAS protein regulation hypothesis, the activity of RAS is controlled by two protein families: guanine exchange factors (GEFs) and GTPase activating proteins (GAPs). GEFs activate the RAS protein family by catalyzing the exchange of GDP for GTP. Conversely, GAPs terminate RAS signaling by enhancing the intrinsic GTPase activity of RAS proteins, thus promoting the hydrolysis of bound GTP to GDP. The precise balance between GEFs and GAPs is crucial for maintaining the normal cellular function and development in eukaryotes [6]. Aberrant expression of GEFs has been reported to cause many types of cancer. In previous studies, almost all of the GEFs were found to be overexpressed and hyperactivated in cancer [7, 8]. However, a subset of noncanonical GEFs were downregulated in tumors [9]. The identification and characterization of these GEFs potentially revolutionize our understanding of the RAS regulation hypothesis and provide new strategies to improve HCC treatment outcomes.

Epigenetics involves gene expression changes without DNA sequence alterations through diverse regulatory mechanisms, including DNA methylation, histone modifications, noncoding RNAs, chromatin remodeling and RNA modifications. These epigenetic changes are fundamentally involved in the precise regulation of human biological processes [10]. In previous a study, we demonstrated a novel N7-methylguanosine (m7G)-modified circRNA, circIPP2A2, acts as a molecule scaffold to facilitate the interaction between Hornerin and PI3K to activate the AKT/GSK3β promoting malignant behaviors in HCC [11]. Clinically, circulating tumor DNA (ctDNA) methylation marker profiling has considerably advanced HCC diagnosis and patient management strategies [12]. The implementation of multiomics technologies, including genomics transcriptomics, proteomics, and metabolomics, has substantially enhanced our ability to detect and characterize epigenetic modifications, providing both large-scale analysis and detailed molecular insights. Multiomics studies in HCC have classified nonviral cases into three prognostic subtypes, identifying steatotic HCC with immune-exhausted features and enhanced response to PD-L1/VEGF combination therapy [13]. Multiomics studies hold potential for uncovering new epigenetic molecular insights and advancing the understanding of disease mechanisms.

In our previous study, we reported that the expression of RASGEF1B, a noncanonical RAS guanine exchange factor, is downregulated by circDHPR in HCC. Notably, elevated RASGEF1B expression level is correlated with favorable prognosis for patients with HCC [14]. However, the specific regulatory mechanism remains to be elucidated. In this study, RASGEF1B expression patterns were validated using our institutional patient cohort, and comprehensive phenotypic experiments were performed to elucidate the tumor-suppressive functions of RASGEF1B both in vitro and in vivo. Mechanistically, transcriptome sequencing revealed that RASGEF1B overexpression decreased SNAI1 mRNA level. High-resolution untargeted metabolomics using LC‒MS/MS revealed that RASGEF1B promoted SNAI1 DNA methylation through Betaine-mediated methionine metabolic reprogramming. Further systemic analysis revealed that RASGEF1B facilitated Betaine accumulation through competitive protection of ALDH7A1 from BMI1-dependent ubiquitination to promote choline metabolism. Collectively, these findings demonstrate that RASGEF1B acts as a tumor suppressor coordinately by stabilizing ALDH7A1 to remodel choline metabolism for Betaine-mediated SNAI1 hypermethylation.

Materials and methods

Cell lines and clinical tissues

The human HCC cell lines HCCLM3 (RRID: CVCL_6832), MHCC97H (RRID: CVCL_4972), and Huh7 (RRID: CVCL_0336) were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). The human HCC cell lines Hep3B (RRID: CVCL_0326), PLC/PRF5 (RRID: CVCL_0485), and the mouse HCC cell line Hepa1-6 (RRID: CVCL_0327) were purchased from ATCC. Mouse normal liver cell line AML-12 (RRID: CVCL_0140) and human embryonic kidney cell line 293T (RRID: CVCL_1926) were purchased from ATCC. Short tandem repeat (STR) profiling was used to check the right cells before the study. All the cell lines included in this study were cultured in high glucose Dulbecco’s modified Eagle’s medium (Gibco, USA) supplemented with 10% fetal bovine serum (Gibco, USA). The cells were cultured under standard conditions at 37 °C in a humidified atmosphere containing 5% CO₂. Twelve pairs of HCC and corresponding normal liver tissues were obtained from Southern Medical University Zhujiang Hospital. In total, 137 HCC and 48 normal liver tissue samples were obtained from Sun Yat-sen University Cancer Center between January 2005 and December 2010. All tissue samples were snap-frozen in liquid nitrogen to preserve DNA, RNA and protein integrity immediately after surgical resection. Informed consent was obtained from all patients in this study. This study was approved by the ethical committee of Zhujiang Hospital and Sun Yat-sen University Cancer Center.

Plasmid construction and cell transfection

The PCR-amplified human RASGEF1B cDNA or shRNA sequences were inserted into pEGFP-N1 or pLenti-puro vector. pLenti-puro vector was used to stable express or knockdown RASGEF1B in HCC cells. The lentivirus was produced using a two-plasmid packaging system. The PCR-amplified human ALDH7A1 cDNA sequences were inserted into the pCMV-Flag vector. The indicated mutations of ALDH7A1 (K253R, K263R, K375R) were subsequently performed using a Fast Site-Directed Mutagenesis Kit (TransGen Biotech, FM111). The PCR-amplified human BMI1 cDNA sequences were inserted into the pEGFP-N1 or pCDNA3.1-GST vector. To establish stable RASGEF1B overexpression or RASGEF1B-knockdown cell lines, HCCLM3 and Huh7 cells were transduced with packaged lentivirus at a multiplicity of infection (MOI) of 15, while Hepa1-6 cells were transduced at an MOI of 20. Three days post-transduction, the cells were subjected to selection with 2 μg/mL puromycin to isolate successfully transduced populations. After three weeks of puromycin selection, the expression level of RASGEF1B was verified by qPCR and western blotting assay. Small interfering RNAs (siRNAs) targeting SNAI1 and BMI1 were purchased from Tsingke Biotechnology (Guangzhou, China). Cell lines were transfected with specific siRNAs using Lipofectamine 3000 transfection reagent (Invitrogen, USA) according to the manufacturer’s protocol. All the targeted sequences of the siRNAs are provided in Supplemental Table 1.

In vitro and in vivo experiments

Standard procedures for cell and animal experiments were performed as previously described [15]. In brief, for the CCK-8 assay, HCC cell lines (HCCLM3, Huh7, and Hepa1-6) were seeded in 96-well plates at a density of 2 × 10³ cells/well to assess their cellular proliferation capacity. Cells were seeded at 1 × 10³ cells/well for colony formation assay. After 14 days of culture, the colonies were fixed with methanol and stained with crystal violet. Cell migration and invasion were assessed using Transwell chambers. Cells (5 × 10⁴) in serum-free medium were seeded in the upper chamber, with complete medium in the lower chamber. After 48 hours of incubation, the cells were fixed and stained with crystal violet. All the results of the experiments above are presented as the means ± SDs of at least three independent repetitions.

For in vivo studies, subcutaneous tumors were established in 4–5-week-old male BALB/c nude mice. HCCLM3 cells (1 × 10⁷) were trypsinized and resuspended in 100 μL of PBS mixed with 50 μL of Matrigel (Corning, USA). Cells were injected into the right flank of the mice, and the width (W) and length (L) of the tumors were measured every 7 days for 5 weeks. The tumor volume was calculated using the formula: V = (L×W²)/2. For the orthotopic liver tumor models, 4–5-week-old male C57BL/6 mice were used to establish the liver tumors. Hepa1-6-luc cells (1 × 10⁶) in 50 μL of PBS were surgically implanted into the left hepatic lobe under sterile conditions. After 2 weeks, the mice were subjected to an in vivo imaging system (IVIS) to to quantify luciferase activity. The mice were intraperitoneally injected with D-luciferin (150 mg/kg) 10 minutes prior to imaging. Tumor formation was confirmed by histopathological analysis and hematoxylin and eosin (H&E) staining. For the lung metastasis model, Huh7 cells were trypsinized and resuspended in PBS at a concentration of 1 × 10⁶ cells/100 μL. The cell suspensions (100 μL) were intravenously injected into the tail veins of 4–5-week-old male BALB/c nude mice. Metastatic progression was monitored by IVIS quantification of GFP fluorescence intensity at 2 weeks post-injection. Lung metastatic nodes were confirmed by H&E staining.

Total RNA extraction, reverse transcription PCR and quantitative real-time PCR

Total RNA from tissues and cells was extracted using TRIzol reagent (Invitrogen, NY, USA) according to the manufacturer’s protocol. Following RNA extraction, RNA purity was assessed using a NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA), with samples demonstrating an A260/A280 ratio between 1.9 and 2.1 considered acceptable for application. PrimeScript™ RT reagent Kit (Takara, China) was used to reverse transcription PCR into cDNA. Quantitative real-time PCR (qPCR) was performed using a SYBR® Green Premix Pro Taq HS qPCR Kit II (AGBIO, China). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control to normalize RNA expression. The relative gene expression was calculated by the 2^-△△CT method. All primers used in this study are listed in Supplemental Table 2

Immunoprecipitation (IP) and western blotting

IP assay was performed according to a standard protocol as previously described [11]. Briefly, RIPA reagent plus inhibitor cocktail (phosphatase inhibitors and protease inhibitors) was used to extract the protein. Specific antibodies were added to the cell lysate, and incubated gently with rotation at 4 °C overnight. The next day, Protein A agarose was used to capture the antibody–protein complexes at 4 °C with rotation for 2 hours. After it was washed with cold PBS, protein was extracted and separated using SDS–PAGE.

Western blotting was subsequently used to analyze protein expression. In brief, proteins were transferred onto polyvinylidene difluoride membranes (Merck Millipore, Darmstadt, Germany) for further analysis. The membranes were then blocked with 5% skim milk for 2 hours at room temperature to prevent nonspecific binding. After they were washed, the membranes were incubated with primary antibodies at 4 °C overnight. The next day, the primary antibodies were removed and the samples were washed with TBST (Tris-buffered saline with 0.1% Tween-20). The membranes were incubated with IRDye 800CW Goat anti-Rabbit/Mouse secondary antibodies. ImageJ software was used to quantify the signal intensity. All antibodies used in this study are listed in Supplemental Table 3.

Immunohistochemistry and immunofluorescence

Immunohistochemistry (IHC) was performed according to a standard protocol. The specimen tissues were fixed in 4% paraformaldehyde and embedded in paraffin. The tissues were cut into section at 4 μm thick sections. Following deparaffinization and rehydration, antigen retrieval was performed using a heat-induced method in a microwave. Then, nonspecific binding sites were blocked using 5% skim milk for 1 hour at room temperature. The indicated primary antibodies were incubated with the tissues at 4 °C overnight. The tissue sections were subsequently incubated with a secondary antibody for 60 minutes at room temperature. Following the use of HRP-conjugated secondary antibodies, a DAB substrate kit was used to detect the protein signals under a light microscope.

For the immunofluorescence assay, the cells were initially fixed using cold methanol for 20 minutes. After three washes with TBST, the cells were subjected to Triton X-100 to increase membrane permeability. After that, nonspecific binding sites were blocked with 5% bovine serum albumin (BSA) for 30 minutes at room temperature. The primary antibodies were subsequently incubated with cells at 4 °C overnight. After the cells were washed with TBST, they were incubated with secondary antibodies for 30 minutes at room temperature in a light-protected environment. DAPI was used for nuclear staining. Nikon A1R confocal microscope (Nikon Corporation, Japan) was used to capture the immunofluorescence images. Image processing and analysis were performed using NIS-Elements AR software.

Bioinformatic tools

MEXPRES was used to screen the CpG island of SNAI1 [16]. The DBCAT database was used to predict the specific SNAI1 DNA methylation sites in the genome [17]. Protein‒protein interaction analysis was conducted using the BioGRID and STRING databases to identify ALDH7A1 interactors [18, 19]. Furthermore, the ubiquitin ligase complex gene set was acquired from GSEA-MSIGDB [20], and the Ubibrowser database was employed to specifically screen for E3 ubiquitin ligases targeting ALDH7A1 [21]. The crystal structures of ALDH7A1 and BMI1 were retrieved from the RCSB-PDB database [22]. The structure of RASGEF1B was predicted by Alphafold3 [23]. HADDOCK was used to perform molecular docking analysis [24], and the results were visualized using PyMOL software.

Transcriptome sequencing

RNA sequencing was conducted by HaploX (Shenzhen, China). Total RNA was extracted using TRIzol reagent (Invitrogen, NY, USA). RNA integrity was assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, USA) to ensure RNA quality (RIN > 7.0). Poly(A)-containing mRNA was enriched from total RNA using Oligo(dT) magnetic beads. First-strand cDNA was synthesized using random hexamer primers and reverse transcriptase, followed by second-strand cDNA synthesis using DNA polymerase I and RNase H. After cDNA library construction and quality control, mRNA expression was quantified using the Illumina PE150 platform. Following read count normalization, differential expression analysis was performed using the DESeq2 package (version 1.38.3) in R. Differentially expressed genes (DEGs) were identified based on the basis of the following stringent criteria: |log2 fold change| > 1 and adjusted p value < 0.05.

Untargeted metabolomics

Untargeted metabolomics was performed by HaploX (Shenzhen, China). Briefly, metabolite extraction from cultured cells (~10⁷ cells per sample) was performed using cold methanol/acetonitrile (1:1, v/v), followed by centrifugation and vacuum drying. For LC‒MS analysis, samples were redissolved in acetonitrile/water (1:1, v/v) and analyzed using a UHPLC-Q Exactive Orbitrap MS system equipped with a HILIC column. Raw data were processed using XCMS and CAMERA for peak detection and annotation, followed by multivariate statistical analysis (PCA and OPLS-DA) using the ropls R package. Metabolites with VIP > 1 and p < 0.05 were considered significantly altered, and compound identification was performed by matching m/z values (< 10 ppm) and MS/MS spectra against an in-house database of authentic standards.

Methylation-specific PCR

A QIAamp DNA Mini Kit (Qiagen) was used to extract the total DNA from the cells. Two micrograms of DNA was subjected to CT Conversion Reagent using the EZ DNA Methylation-Gold Kit (Zymo Research) according to the manufacturer’s instructions. The DNA was denatured in 130 μL of conversion reagent at 98 °C for 10 minutes, followed by incubation at 64 °C for 2.5 hours. Zymo-Spin IC columns were used to purify and desulfonate the DNA. The bisulfite-treated DNA was then purified and eluted in 10 μL of elution buffer. For PCR amplification, specific primers were designed using MethPrimer software to amplify regions of interest.

Luciferase assay

The promoter sequences (−2000 to + 200 bp) of SNAI1 were amplified from human genomic DNA and cloned into the pGL3-basic promoter vector. Luciferase assays were performed in 293T cells using either the pGL3–SNAI1 promoter or pGL3-Basic luciferase reporter. Lipofectamine 3000 (Thermo Scientific) was used to cotransfect reporter plasmids, the pRL-TK-Renilla luciferase plasmid, and either the RASGEF1B overexpression plasmid or an empty vector. Forty-eight hours after transfection, the Firefly and Renilla luciferase activities were measured using a Dual-Luciferase Reporter Kit (Promega). Each experiment was performed in triplicate and repeated at least three times independently.

Methionine (Met)and S-adenosylmethionine (SAM) assay

A SAM ELISA Kit (Biorbyt, USA) and a Met ELISA Kit (United States Biological, USA) were used to detect the SAM and Met levels, respectively, following the manufacturer’s protocol. Briefly, cells were prepared in ice-cold PBS, ultrasonicated, and centrifuged at 10,000 × g for 5 min. Standards were reconstituted to generate a dilution series, and samples/standards (50 µL) were added to precoated wells along with 50 µL of biotinylated conjugates, followed by one hour of incubation at 37 °C. After the samples were washed, 100 µL Streptavidin-HRP was added and incubated for another hour at 37 °C. Subsequently, 90 µL of TMB Substrate was added and the samples were incubated for 20 minutes at 37 °C in the dark. The reaction was stopped with 50 µL Stop Reagent, and absorbance was measured at 450 nm. SAM or Met concentrations were calculated by interpolating sample OD values against the standard curve.

Protein half-life assay

Cells were pretreated with 10 μM MG132 (a proteasome inhibitor) for 8 hours. Following MG132 pretreatment, the cells were incubated with 70 μg/mL CHX to inhibit de novo protein synthesis. The total proteins from the cells were extracted using RIPA buffer supplemented with protease inhibitors and 20 μM MG132. SDS‒PAGE was subsequently used to separate the proteins, followed by western blotting analysis.

GST pulldown assay

Recombinant human GST-RASGEF1B protein (Novus Biologicals, USA) or GST protein was cocultured with recombinant human His-ALDH7A1 protein (MyBiosource, USA) in NP-40 at 4 °C for 12 hours. The GST beads were then incubated with the above protein complex at 4 °C for 6 hours. After the samples were washed, protein was extracted according to a standard protocol. Western blotting was used to analyze the interaction between RASGEF1B and ALDH7A1.

Statistical analysis

Statistical analysis was conducted with GraphPad Prism 8.0 (La Jolla, CA, USA) and SPSS 21.0 (IBM Corp., NY, USA). Differences between groups were analyzed by Student’s t test and analysis of variance. Categorical variables were compared via the chi-square test. The Kaplan‒Meier method and log-rank test were used to evaluate overall survival (OS) and disease-free survival (DFS). A p value<0.05 was considered to indicate statistical significance.

Results

RASGEF1B overexpression is associated with a favorable prognosis in HCC

In our previous study, we have confirmed that RASGEF1B is regulated by circDHPR in HCC. Analysis of the TCGA database revealed that high RASGEF1B expression correlates with favorable prognosis in HCC patients [14]. However, the mechanism underlying the suppression of HCC progression by RASGEF1B remains to be investigated. To further confirm the role of RASGEF1B expression in HCC, we used twelve pairs of HCC and corresponding normal liver tissues to confirm the expression of RASGEF1B. The results of the qPCR and western blotting assays showed that RASGEF1B expression was significantly reduced in HCC tissues than in their corresponding normal tissues (Fig. 1a–b). The results of IHC performed on a tissue microarray (TMA) confirmed that RASGEF1B expression was downregulated in HCC (Fig. 1c). qPCR and western blotting results revealed that RASGEF1B expression was also downregulated in HCC cell lines (HCCLM3, Huh7, Hep3B, MHCC97H, PLC/PRF5, and Hepa1-6) compare to mouse normal liver cell line (AML-12) and 293T control cells (Fig. 1d–e). To analyze the impact of RASGEF1B on the prognosis of patients with HCC, qPCR was performed on tissue samples (137 HCC and 48 normal liver tissue samples) obtained from the Sun Yat-sen University Cancer Center to assess RASGEF1B expression (Fig. 1f). Patients (N = 137) with HCC were divided into two groups according to the median expression of RASGEF1B: high RASGEF1B expression (N = 68) and low RASGEF1B expression (N = 69). Kaplan–Meier (K-M) curves revealed that compared with low RASGEF1B expression, high RASGEF1B expression was correlated with improved overall survival (OS) and disease-free survival (DFS) in patients with HCC (Fig. 1g–h). The results of the chi-square test revealed that the low RASGEF1B expression was associated with increased tumor size (> 5 cm) and vascular invasion (Table. S4). The above results show that RASGEF1B expression is downregulated in both tissue and cells of HCC. Additionally, low RASGEF1B expression is correlated with poor prognosis in patients with HCC.

Fig. 1.

RASGEF1B upregulation is associated with favorable clinical outcomes in HCC. a. qPCR analysis was performed to detect the relative RASGEF1B mRNA expression in HCC (N = 12) and corresponding normal liver tissues (N = 12). b. western blotting analysis of RASGEF1B in HCC (N = 12) and corresponding normal liver tissues (N = 12). c. IHC assay performed in patient tissue microarray (TMA) was used to confirm the RASGEF1B expression pattern. d-e. the expression of RASGEF1B in HCC cell lines (Hepa1-6, HCCLM3, MHCC97H, PLC/PRF5, and Huh7) and control cell lines (ALM12 and 293T) were detected by qPCR and western blotting. f. qPCR analysis was employed to detect RASGEF1B expression in HCC (N = 137) and normal liver tissues (N = 48). g-h. kaplan-Meier curves showed the association between RASGEF1B expression and patient overall survival (OS) and disease-free survival (DFS). **p < 0.01; ***p < 0.001; ****p < 0.0001

RASGEF1B suppresses tumor malignancy both in vitro and in vivo

To further investigate the nonclassical antitumor role of RASGEF1B in the GEF family, lentiviral vectors were used to establish RASGEF1B overexpression (Huh7-oeRASGEF1B) and knockdown (HCCLM3-shRASGEF1B-#1, HCCLM3-shRASGEF1B-#2, Hepa1-6-shRASGEF1B-#1, and Hepa1-6-shRASGEF1B-#2) stable cell lines. Following puromycin selection, qPCR and western blotting assay were employed to confirm the successful establishment (Fig. 2a–b and S1a). The results of the CCK-8 assay determined that RASGEF1B overexpression suppressed tumor proliferation, whereas RASGEF1B knockdown promoted HCC growth (Fig. 2c). The results of the colony formation assay showed that RASGEF1B overexpression significantly inhibited colony formation (Figure. S1b–c). The Transwell assay confirmed that RASGEF1B overexpression inhibited HCC migration and invasion, whereas RASGEF1B downregulation enhanced these tumorigenic behaviors (Figure. S1d–e). In vitro experiments reveal only changes in the cells and fail to account for the influence of the tumor microenvironment. Therefore, a subcutaneous tumor model was established using HCCLM3 and Huh7 cell lines, and the results showed that RASGEF1B downregulation significantly facilitated HCC growth and increased tumor burden (Fig. 2d–f). The results of the IHC assay revealed that RASGEF1B knockdown significantly upregulated PCNA and Ki-67 expression (Fig. 2g). Notably, RASGEF1B overexpression inhibited HCC growth and reduced tumor burden (Fig. 2h–j). IHC results confirmed that RASGEF1B overexpression significantly downregulated PCNA and Ki-67 expression (Fig. 2k). To better mimic the HCC microenvironment, an orthotopic liver tumor model in C57BL/6 mice was established using a Hepa1-6-luc cell line with or without RASGEF1B downregulation. The IVIS results revealed that reduced RASGEF1B expression promoted mice orthotopic liver tumor growth (Fig. 2land S1f).This finding was further supported by the results of histopathological analysis and H&E staining (Fig. 2m–n). Similarly, the results of the IHC analysis revealed that RASGEF1B knockdown increased PCNA and Ki-67 expression (Fig. 2o). A lung metastasis model showed that RASGEF1B overexpression significantly inhibited tumor metastasis (Fig. 2p–q and S1g). Taken together, these findings confirm that RASGEF1B can inhibit malignant behaviors in HCC both in vitro and in vivo.

Fig. 2.

RASGEF1B suppresses hepatocellular carcinoma malignancy both in vitro and in vivo. a-b. qPCR and western blotting confirmed the successful establishment of RASGEF1B stable knockdown (HCCLM3-shRASGEF1B-#1/#2 and Hepa1-6-shRASGEF1B-#1/#2) and overexpression (Huh7-oeRASGEF1B) cell lines. c. CCK-8 assay was employed to detect the cell proliferation change induced by RASGEF1B downregulation (HCCLM3 and Hepa1-6) and overexpression (Huh7). d-f. RASGEF1B downregulation HCCLM3 cells were subcutaneously inoculated into nude mice (N = 5 per group). Tumor image (d), tumor growth curves (e) and tumor weight (f) were shown. g. IHC assay was used to detect the proliferation markers (PCNA and Ki-67) in RASGEF1B downregulation subcutaneous tumor model. h-j. RASGEF1B overexpression Huh7 cells were subcutaneously inoculated into nude mice (N = 5 per group). Tumor image (h), tumor growth curves (i) and tumor weight (j) were shown. k. IHC assay was used to detect the proliferation markers (PCNA and Ki-67) in RASGEF1B overexpression subcutaneous tumor model. l. Hepa1-6-luc cells with RASGEF1B knockdown were used to establish orthotopic liver tumor, IVIS images showed the tumor burden. m. orthotopic liver tumors were confirmed by histopathological analysis. n. orthotopic liver tumors were confirmed by pathological analysis using H&E staining. o. IHC assay was used to detect the proliferation markers (PCNA and Ki-67) expression in RASGEF1B knockdown orthotopic liver tumor. p. Huh7 cells with RASGEF1B overexpression were injected into vein tail of BALB/c mice (N = 5 per group), IVIS images showed the lung metastasis tumor burden. q. lung metastasis nodes were confirmed by H&E staining. Data are presented as the mean±SD at triplicate independent repeated experiments. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

RASGEF1B attenuates SNAI1 expression and glycolysis while promoting one-carbon metabolism

To investigate the mechanisms by which RASGEF1B suppresses HCC progression, transcriptome sequencing was conducted to explore the downstream regulatory mechanism induced by RASGEF1B overexpression (Fig. 3a). After the data were cleaned and quality control was performed (Fig. 3b and S2a–f), the results of DESeq2 analysis revealed that the expression of snail family transcriptional repressor 1 (SNAI1) was significantly downregulated in the context of RASGEF1B overexpression (Fig. 3c–d). qPCR and western blotting results confirmed that RASGEF1B overexpression decreased the SNAI1 expression at both the transcriptional and protein levels, whereas RASGEF1B knockdown increased the SNAI1 expression (Fig. 3e and S3a–b). According to Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis, RASGEF1B overexpression regulated the carbon metabolism signaling pathway (Fig. 3f). Carbon metabolism represents a fundamental and biologically diverse process that is essential for all life forms. It encompasses key pathways such as glycolysis, the tricarboxylic acid (TCA) cycle, the pentose phosphate pathway, gluconeogenesis and one-carbon metabolism, and others [25]. On the basis of the above background, we hypothesized that RASGEF1B suppresses tumor progression by regulating carbon metabolism. Thus, the key enzymes involved in carbon metabolism were detected by western blotting (Figure. S3c). The Western blotting results revealed that RASGEF1B overexpression significantly inhibited glycolysis pathway while promoting one-carbon metabolism. However, plasmid-mediated SNAI1 upregulation reversed the suppression of glycolysis, whereas one-carbon metabolism remained unaffected in the Huh7 cell line (Fig. 3g and S3d). Conversely, knocking down SNAI1 with siRNAs inhibited glycolysis pathway under conditions of RASGEF1B downregulation, whereas one-carbon metabolism remained unchanged (Fig. 3h and S3e). The above results indicate that RASGEF1B downregulates SNAI1 expression at both the transcriptional and protein levels, and primarily modulates glycolysis pathway in carbon metabolism.

Fig. 3.

RASGEF1B reduces SNAI1 expression and regulates carbon metabolism. a. Schematic of the transcriptome sequencing workflow in RASGEF1B overexpression Huh7 cells (OE) and corresponding normal control (CTRL). b. PCA plot showed the raw mRNA sequencing data after quality control. c. Heat map showed the alteration of mRNA expression pattern. d. Volcano plot showed the differential expression mRNAs after DESeq2 analysis. e. Western blotting analysis were performed to detect the SNAI1 expression changed by RASGEF1B upregulation (Huh7) and downregulation (HCCLM3 and Hepa1-6). f. KEGG enrichment analysis was used to screen the pathway that affected by RASGEF1B overexpression. g. Western blotting analysis was performed to detect the key enzymes that affected by SNAI1 overexpression at the context of RASGEF1B overexpression Huh7 cells. h. Western blotting analysis was performed to detect the key enzymes that affected by SNAI1 inhibition at the context of RASGEF1B knockdown HCCLM3 and Hepa1-6 cells. Data are presented as the mean±SD at triplicate independent repeated experiments

RASGEF1B suppresses SNAI1 expression through Betaine-mediated methionine metabolic reprogramming and DNA methylation

According to previous studies, RASGEF1B potentially suppressed tumor progression potentially by inhibiting glycolysis and promoting the one-carbon metabolism pathway. Given that metabolic activity plays a crucial role in tumor pathophysiology, high-resolution untargeted metabolomics using LC‒MS/MS was employed to detect alterations in cell metabolism induced by RASGEF1B overexpression (Fig. 4a). After data acquisition and quality control (Figure. S4a–g), the results showed that the concentration of trimethylglycine (Betaine) was most significantly increased following RASGEF1B overexpression in positive ion mode (Fig. 4b and S5a–c). Betaine is regarded as an effective methyl donor, transferring –CH3 groups to homocysteine for the synthesis of methionine. Methionine is further converted into S-adenosylmethionine (SAM), which serves as the primary methyl donor in DNA methylation reactions [26] (Fig. 4c). Measurements of the concentrations of Met and SAM demonstrated that RASGEF1B knockdown markedly reduced SAM levels, whereas RASGEF1B overexpression increased SAM levels in the cell lysate supernatant. Intriguingly, the Met level was unaffected by alterations in RASGEF1B expression (Fig. 4d–e). The above results demonstrated that RASGEF1B can increase the SAM level by reprogramming the methionine metabolism. On the basis of these findings, we hypothesized that RASGEF1B overexpression inhibited SNAI1 transcription by promoting DNA methylation at the CpG island within the SNAI1 gene. The DNA methylation inhibitor 5-azacytidine (5-aza) was used to confirm that RASGEF1B induces SNAI1 DNA methylation. The qPCR results revealed that 5-aza increased the SNAI1 mRNA expression in the context of RASGEF1B overexpression (Fig. 4f). Conversely, the cell metabolite Betaine was added to stable RASGEF1B knockdown HCC cell lines. qPCR results confirmed that the relative mRNA expression of SNAI1 was reduced following Betaine treatment in the context of RASGEF1B downregulation (Fig. 4g). The results of 5-mc immunofluorescence staining confirmed that RASGEF1B downregulation reduced the DNA methylation level, whereas RASGEF1B overexpression promoted genome methylation (Fig. 4h–i). To identify DNA methylation sites in the SNAI1 gene, we used MEXPRES, an online methylation site prediction database, to screen the CpG islands of SNAI1 (Fig. 4j). The DBCAT database was used to explore the specific methylation site in SNAI1 (Figure. S6). The results of methylation-specific PCR (MSP) revealed that RASGEF1B overexpression significantly increased the methylation level of the CpG islands in the SNAI1. However, these effects were reversed by RASGEF1B knockdown (Fig. 4k–l). Luciferase assays demonstrated that RASGEF1B promoted the methylation of CpG islands in the SNAI1 promoter, thus repressing its transcriptional activity (Fig. 4m). These results reveal that RASGEF1B inhibits SNAI1 transcription by increasing Betaine-mediated DNA methylation.

Fig. 4.

RASGEF1B reduces SNAI1 expression by inducing Betaine-mediated DNA methylation. a. Schematic of the high-resolution untargeted metabolomics workflow in RASGEF1B overexpression Huh7 cells (OE) and corresponding normal control (CTRL). b. Volcano plots display differentially expressed metabolites between RASGEF1B-overexpressing at positive ion mode. c. Schematic diagram illustrated the role of Betaine and SAM in methionine metabolism. d. Methionine level in cell homogenate supernatant was measured by human Met ELISA Kit, left panel showed the standard curve. e. SAM level in cell homogenate supernatant was measured by human SAM ELISA Kit, left panel showed the standard curve. f. Huh7 cells with RASGEF1B overexpression were treated with the DNA methylation inhibitor 5-azacytidine at concentrations of 0, 10 nM, 100 nM, 1 µM, and 10 µM. SNAI1 expression was assessed by qPCR 8 hours post-treatment. g. HCCLM3 and Hepa1-6 cells with RASGEF1B downregulation were treated with the Betaine at concentrations of 0, 25 µM, 50 µM, 100 µM, and 200 µM. SNAI1 expression was assessed by qPCR 12 hours post-treatment. h-i. 5-mc immunofluorescence staining showed the alteration of DNA methylation level caused by RASGEF1B downregulation (upper panel) and overexpression (buttom panel). j. The DNA methylation sites in SNAI1 promoter region was predicted by MEXPRES database. k-l. MSP (k) and MS-qPCR (l) were performed to assess SNAI1 promoter methylation levels in RASGEF1B-knockdown HCCLM3, Hepa1-6 cells, and RASGEF1B-overexpressing Huh7 cells. m. 293T cells were co-transfected with an SNAI1 promoter reporter construct and either RASGEF1B overexpression or empty vector, followed by luciferase assays to measure promoter activity. Data are presented as the mean±SD at triplicate independent repeated experiments. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001

RASGEF1B suppresses HCC growth by inhibiting SNAI1-mediated tumor proliferation, and metastasis

While SNAI1 has been established as a critical mediator of the epithelial–mesenchymal transition (EMT) [27], its functional interplay with varying RASGEF1B expression patterns in HCC remains unknown. The results of the qPCR and western blotting assays revealed that SNAI1 expression was upregulated in the HCC compared with the corresponding normal tissues at both the transcriptional and translational levels (Figure. S7a–b). The results of the IHC assay in the HPA database confirmed that SNAI1 was significantly upregulated in HCC (Figure. S7c). Notably, the K‒M curves revealed that elevated SNAI1 expression correlated with unfavorable postoperative disease-free survival, but no significant correlation was found with overall survival in the TCGA data (Figure. S7d). Additionally, the results of qPCR and western blotting revealed that SNAI1 expression was significantly upregulated in three HCC cell lines (Hepa1-6, HCCLM3, and Hep3B). Although SNAI1 expression was also elevated in MHCC97H, PLC/PRF5, and Huh7 cell lines, the upregulation in these cell lines did not reach statistical significance (Figure. S7e–f). To explore the role of SNAI1 in HCC, a SNAI1 overexpression plasmid and inhibitory siRNAs were designed and introduced into stable RASGEF1B overexpression and knockdown cell lines. qPCR and western blotting were used to confirm the changes in SNAI1 expression (Figure. S8a–b). The CCK-8 and colony formation assays demonstrated that upregulating SNAI1 expression could counteract the antiproliferative and tumorigenesis effects caused by RASGEF1B overexpression. Conversely, the enhanced proliferation and colony formation resulting from RASGEF1B knockdown were attenuated by SNAI1 downregulation using siRNAs (Figure. S9a–b). Additionally, the Transwell assay revealed that SNAI1 overexpression promoted migration and invasion in the context of RASGEF1B upregulation. In contrast, inhibiting SNAI1 expression attenuated the increase in migration and invasion induced by RASGEF1B downregulation (Figure. S9c). Our findings demonstrate that SNAI1 impedes the tumor suppressive function of RASGEF1B in HCC.

RASGEF1B elevates Betaine levels by remodeling choline metabolism through the ubiquitination of ALDH7A1

To further elucidate the mechanism through which RASGEF1B regulates SNAI1 DNA methylation by increasing Betaine levels. The role of Betaine in biological metabolism process was analyzed. A metabolic network analysis revealed that Betaine is a critical metabolite in choline metabolism. Briefly, choline is converted into Betaine aldehyde under the action of choline dehydrogenase (CHDH), and subsequently, Betaine aldehyde is further transformed into Betaine by the enzyme Betaine aldehyde dehydrogenase (BADH). In humans, ALDH7A1 functionally substitutes for BADH through conserved aldehyde dehydrogenase mechanisms [28] (Fig. 5a). To elucidate the role of RASGEF1B in choline metabolism, qPCR and western blotting were used to detect the expression of CHDH and ALDH7A1 under different RASGEF1B expression patterns. Intriguingly, the results revealed that altering RASGEF1B expression affected only the ALDH7A1 protein level, while the transcriptional level of ALDH7A1 remained unchanged (Fig. 5b and S10a–c). These results suggest that RASGEF1B regulates ALDH7A1 at the posttranslational level, potentially through mechanisms such as protein stability, degradation, or modification. However, the specific mechanism remains unclear. Systematic reactome enrichment analysis showed that SUMOylation or ubiquitination is most likely involved in regulating the ALDH7A1 function (Figure. S10d). There is growing recognition that the ubiquitin system critically regulates tumor pathogenesis in HCC [29]. To investigate whether RASGEF1B affects ALDH7A1 protein stability, we treated HCC cell lines with cycloheximide (CHX) to inhibit protein synthesis. Our results revealed that RASGEF1B downregulation significantly shortened the half-life of the ALDH7A1 protein (Fig. 5c). Additionally, treatment with the proteasome inhibitor MG132 attenuated the effects of RASGEF1B knockdown on ALDH7A1 protein expression (Fig. 5d). Notably, RASGEF1B overexpression led to a marked decrease in ALDH7A1 ubiquitination (Fig. 5e), whereas knockdown of RASGEF1B in both the HCCLM3 and Hepa1-6 cell lines resulted in significant accumulation of ubiquitinated ALDH7A1 (Fig. 5f). The above results suggest that RASGEF1B promotes Betaine-mediated SNAI1 CpG island methylation by shielding ALDH7A1 from ubiquitination degradation.

Fig. 5.

RASGEF1B elevates Betaine level by remodeling choline metabolism through ALDH7A1 ubiquitination modification. a. schematic diagram showed the key enzymes and intermediate metabolites in choline metabolism. b. western blotting assay was used to detect the ALDH7A1 level of choline metabolism. c. western blotting analysis of ALDH7A1 and RASGEF1B in RASGEF1B knockdown HCCLM3 and Hepa1-6 cells treated with or without cycloheximide (CHX, 70 μg/mL) for the indicated time. d. western blotting analysis was used to detect the ALDH7A1 and RASGEF1B in RASGEF1B knockdown HCCLM3 and Hepa1-6 cells treated with or without MG132 (10 μM, 8 h). e. RASGEF1B overexpression Huh7 cells were transfected with HA-Ub plasmid and treated with MG132 (10 μM, 8 h). IgG or anti-ALDH71A antibody was used to perform IP assay, followed by western blotting assay. f. RASGEF1B knockdown HCCLM3 (left panel) and Hepa1-6 (right panel) cells were co-transfected with HA-Ub and Flag-ALDH7A1 plasmid. After MG132 (10 μM, 8 h) treatment, IP assay was performed using IgG or anti-Flag antibody. followed by western blotting to detect ubiquitination signals. Data are presented as the mean±SD at triplicate independent repeated experiments. ns, not significant; **p < 0.01; ***p < 0.001; ****p < 0.0001

ALDH7A1 ubiquitination is mediated by the E3 ubiquitin ligase BMI1

In terms of characteristic protein domains and motifs, RASGEF1B exhibits exclusive RAS-GEF catalytic activity. Ou findings suggest the presence of additional regulatory factors beyond RASGEF1B that may modulate ALDH7A1 ubiquitination. To identify the key E3 ubiquitin ligases that mediate ALDH7A1 proteasomal degradation, comprehensive bioinformatics was performed using online prediction databases (BioGRID, GESA-MSIGDB, and Ubibrowser). In this study, we initially identified 130 potential ALDH7A1-interacting proteins through BioGRID database analysis. These candidates were subsequently cross-referenced with the ubiquitin ligase complex gene set from the GSEA-MSIGDB database to identify putative E3 ligases that regulate ALDH7A1. In addition, Ubibrowser was used to confirm the role of the E3 ubiquitin ligase for ALDH7A1 (Fig. 6a). The workflow screening results described above revealed that among all the candidate interactors, BMI1 is the sole RING-type E3 ubiquitin ligase capable of regulating ALDH7A1 ubiquitination (Fig. 6b). The crystal structures of ALDH7A1 and BMI1 were retrieved from the RCSB-PDB database, after which molecular docking analysis was performed using HADDOCK. The results demonstrated a potential interaction between ALDH7A1 and BMI1 at lysine residues K253, K263, and K375 (Fig. 6c). To validate these predictions, co-IP assay was performed and the physical interaction between the ectopic expressions of Flag-ALDH7A1 and GFP-BMI1 in 293T cells was used to verified (Fig. 6d–e). Additionally, the results of the reciprocal co-IP assay revealed that ALDH7A1 could bind to BMI1 in HCCLM3 and Huh7 cells (Fig. 6f–g). Notably, the results of immunofluorescence staining confirmed the subcellular colocalization of ALDH7A1 and BMI1 (Fig. 6h and S11a). These results indicate that ALDH7A1 interacts with the E3 ubiquitin ligase BMI1. CHX assay was performed to support the regulatory mechanism by which BMI1 increases ALDH7A1 ubiquitination. The results revealed that inhibition of BMI1 expression using siRNAs significantly prolonged the half-life of the ALDH7A1 protein in HCCLM3 and Huh7 cells (Fig. 6i). Western blotting results showed that the upregulated protein level of ALDH7A1 caused by siRNA-mediated BMI1 knockdown was not observed in MG132-treated HCCLM3 and Huh7 cells (Figure. S11b). Conversely, MG132 treatment abolished the ALDH7A1 downregulation mediated by BMI1 overexpression (Figure. S11c). The above results suggest that BMI1 regulates ALDH7A1 through a proteasomal degradation pathway. According to the UniProt database, the RING type zinc finger is located between 18 and 57 aa of BMI1 (Figure. S11d). To further explain how BMI1 ubiquitinates ALDH7A, Flag-ALDH7A1, wild-type (WT) or Δ18–57 (zinc finger mutant) of GFP-BMI1, and HA-ubiquitin (HA-Ub) were co-expressed in 293T cells. Immunoprecipitation assays demonstrated that compared with the wild-type control, the wild-type GFP-BMI1 significantly increased ALDH7A1 ubiquitination, whereas compared with the wild-type control, the Δ18–57 GFP-BMI1 mutant reduced ubiquitination activity (Fig. 6j). Similarly, silencing BMI1 expression significantly decreased the ALDH7A1 ubiquitination in HCCLM3 and Huh7 cells (Fig. 6k). On the basis of the molecular docking prediction results, three potential sites (K253, K263, and K375) of ALDH7A1 were selected for experimental verification (Fig. 6c and S11e). Immunoprecipitation assays demonstrated that reconstituted GFP-BMI1 expression in BMI1-depleted HCCLM3 cells significantly increased the ubiquitination of wild-type Flag-ALDH7A1. Notably, this BMI1-mediated ubiquitination was preserved in the K253R and K263R mutants but was abolished in the K375R mutant, identifying Lys375 as the critical ubiquitin acceptor site for the ubiquitination degradation of BMI1-mediated ALDH7A1 (Fig. 6l). Taken together, our findings confirm that Lys375 is the ubiquitination site of ALDH7A1 mediated by BMI1.

Fig. 6.

ALDH7A1 ubiquitination is mediated by E3 ubiquitin ligase BMI1. a. schematic workflow for identifying BMI1 as the E3 ubiquitin ligase responsible for ALDH7A1 downregulation. b. venn diagram depicting overlapping protein interactors of ALDH7A1 identified from three complementary databases: BlOGRlD (protein-protein interactions) GSEA-MSIGDB (pathway associations), and UbiBrowser (E3 ubiquitin ligase-substrate relationships). c. molecule docking analysis was performed using HADDOCK tools using the crystal structures of ALDH7A1 and BMI1 from RCSB-PDB database. K253, K263, and K375 residues was the potential interaction site. d-e. GFP-BMI1 or GFP plasmid were co-transfected with Flag-ALDH7A1 in to 293T cells. co-IP assay was performed using anti-Flag (d) or anti-GFP (e) antibody, followed by western blotting to detect enrichment signal. f-g. reciprocal co-IP assay was used to detect the ALDH7A1 and BMI1 in HCCLM3 and Huh7 cells. IgG was employed as negative control. h. ALDH7A1 and BMI1 subcellular colocalization was assessed by immunofluorescence staining using anti-ALDH7A1 (green) and anti-BMI1 (purple) antibodies. i. western blotting analysis of ALDH7A1 and BMI1 in BMI1 sliencing HCCLM3 and Huh7 cells treated with or without cycloheximide (CHX, 70 μg/mL) for the indicated time. j. Flag-ALDH7A1, wild-type (WT) or Δ18–57 (zinc finger mutant) of GFP-BMI1, and HA-Ub were co-expressed in 293T cells. After MG132 (10 μM, 8 h) treatment, anti-Flag IP was performed to assess ALDH7A1 ubiquitination. k. HCCLM3 (left panel) and Huh7 (right panel) cells with BMI1 silencing were transfected with HA-Ub plasmid. After MG132 (10 μM, 8 h) treatment, IP assay was performed using IgG or anti-ALDH7A1 antibody, followed by western blotting to detect the ubiquitination signal. l. three potential lysine ubiquitination sites (K253, K263, and K375) in ALDH7A1 were identified by molecule docking analysis. Flag-ALDH7A1 (WT or indicated lysine mutants) was co-expressed with HA-Ub in BMI1 downregulation 293T cells rescued with or without GPF-BMI1. After MG132 (10 μM, 8 h) treatment, anti-Flag antibody was used to perform IP assay. Western blotting assay was used to detect the ubiquitination signal. Data are presented as the mean±SD at triplicate independent repeated experiments. ns, not significant; **p < 0.01; ***p < 0.001; ****p < 0.0001

RASGEF1B shields ALDH7A1 from BMI1-dependent ubiquitination via direct binding with ALDH7A1

In this study, we demonstrated that RASGEF1B regulates ALDH7A1 protein stability via the ubiquitin‒proteasome pathway. Furthermore, we found that the E3 ubiquitin ligase BMI1 modulates ALDH7A1 ubiquitination levels. On the basis of these findings, we investigated the functional interplay among RASGEF1B, ALDH7A1, and BMI1. Protein‒protein interaction (PPI) analysis of the data from the STRING database indicated that ALDH7A1 potentially interacts with RAPGEF3 and RAPGEF4, which share a conserved guanine nucleotide exchange factor (GEF) domain with RASGEF1B (Figure. 7a and S12). Using the AI-predicted structure of RASGEF1B generated by AlphaFold3, molecular docking was performed to assess its binding with ALDH7A1. The results revealed that compared with the ALDH7A1–BMI1 complex, the RASGEF1B–ALDH7A1 interaction exhibited stronger van der Waals forces (−81.5 ± 6.2 vs. −129.4 ± 0.5 kcal/mol) and a greater Buried Surface Area (Ų) (Fig. 7b). Therefore, we hypothesized that RASGEF1B antagonized BMI1-mediated ALDH7A1 ubiquitination via competitive interactions. Co-IP experiments confirmed a specific interaction between RASGEF1B and ALDH7A1, whereas no interaction was detected between RASGEF1B and BMI1 (Fig. 7c–d). To determine whether RASGEF1B directly interacts with ALDH7A1, we performed a GST pull-down assay using recombinant human GST-RASGEF1B and His-ALDH7A1 proteins. The results confirmed that RASGEF1B directly interacted with ALDH7A1 (Fig. 7e). Immunofluorescence staining revealed the colocalization of RASGEF1B and ALDH7A1 (Fig. 7f).

Fig. 7.

RASGEF1B shields ALDH7A1 from BMI1-dependent ubiquitination via direct binding with ALDH7A1. a. STRING database was used to showed the interaction between RAPGEF3/4 and ALDH7A1. b. molecular docking analysis was conducted using HADDOCK software, employing the crystal structure of ALDH7A1 obtained from the RCSB-PDB database and the RASGEF1B structure predicted by AlphaFold3. c. Co-IP assay was performed in HCCLM3 and Huh7 cells using anti-BMI1 antibody, followed by western blotting. d. Co-IP assay was performed in HCCLM3 and Huh7 cells using anti-ALDH7A1 antibody, followed by western blotting. e. recombinant GST-RASGEF1B protein was incubated with recombinant His-ALDH7A1 protein, followed by GST pull down and western blotting using anti-GST and anti-His antibodies. f. immunofluorescence staining using anti-ALDH7A1 (green) and anti-RASGEF1B (purple) antibodies. g. Co-IP analysis was performed in RASGEF1B knockdown HCCLM3 and Hpea1-6 cells, using an anti-BMI1 antibody to enrich ALDH7A1, followed by detection via western blotting. h. Co-IP analysis was performed in RASGEF1B upregulated Huh7 cells, using an anti-BMI1 antibody to enrich ALDH7A1, followed by detection via western blotting. i. BMI1 co-immunoprecipitated with ALDH7A1 in the presence of increasing concentrations of recombinant GST-RASGEF1B (0, 2, and 4 μg) in 293T cells. j. immunofluorescence staining was performed in RASGEF1B knockdown HCCLM3 cells and RASGEF1B overexpression Huh7 cells, using anti-ALDH7A1 (green) and anti-BMI1 (purple) antibodies. k. HA-Ub, Flag-ALDH7A1, GST-BMI1, and GFP-RASGEF1B were incubated with E2, E3, and ATP in an in vitro ubiquitination assay. Protein complexes were immunoprecipitated using an anti-Flag antibody and analyzed by western blotting

To further reveal the competitive interaction between RASGEF1B and BMI1 toward ALDH7A1, reciprocal co-IP assays were performed in RASGEF1B-knockdown HCCLM3 and Hepa1-6 cells, or Huh7 cells with RASGEF1B overexpression. The results of co-IP showed that RASGEF1B knockdown increased the enrichment of BMI1 by ALDH7A1 (Fig. 7g), whereas ectopic expression of RASGEF1B blocked the interaction between ALDH7A1 and BMI1 (Fig. 7h–i). The results of immunofluorescence staining revealed that RASGEF1B overexpression inhibited the colocalization of ALDH7A1 and BMI1. Notably, RASGEF1B knockdown facilitated the interaction between ALDH7A1 and BMI1 (Fig. 7j). Using an in vitro reconstituted ubiquitination system containing E1/E2 enzymes and purified proteins (RASGEF1B, BMI1, and ALDH7A1), we demonstrated that RASGEF1B specifically inhibits BMI1-mediated ALDH7A1 ubiquitination and subsequent proteasomal degradation (Fig. 7k). Collectively, these data demonstrate that RASGEF1B protects ALDH7A1 from BMI1-dependent ubiquitination.

Correlation and functional cooperation between RASGEF1B and ALDH7A1 in HCC

The mRNA expression of RASGEF1B and ALDH7A1 was retrieved from the TCGA database, and a significant positive correlation was observed between RASGEF1B and ALDH7A1 mRNA expression levels (Fig. 8a). According to the GEPIA database, elevated ALDH7A1 expression correlated with improved postoperative overall survival, but no significant correlation was found with disease-free survival (Fig. 8b). IHC analysis of serial sections from the same patients revealed a positive correlation between ALDH7A1 and RASGEF1B protein expression (Fig. 8c–e). Tissue immunofluorescence staining revealed significant co-localization of RASGEF1B (green) and ALDH7A1 (red), corroborating our earlier findings in cultured cells (Fig. 8f). A subcutaneous tumor model was established using Hepa1-6 cells, and the results revealed that ALDH7A1 knockdown attenuated the tumor-suppressive effects induced by RASGEF1B overexpression (Fig. 8g–i). These findings were validated in an orthotopic liver tumor model through IVIS bioluminescence imaging, histological analysis, and H&E staining (Fig. 8j–k). Additionally, Betaine supplementation reverses malignant phenotypes induced by RASGEF1B (Figure. S13). Collectively, our findings reveal that RASGEF1B exerts tumor-suppressive effects by protecting ALDH7A1 from ubiquitination and enhancing choline metabolism in HCC.

Fig. 8.

Correlation between RASGEF1B and ALDH7A1 in HCC. a. the correlation between RASGEF1B and ALDH7A1 mRNA expression was analysis using TCGA database. b. kaplan-Meier curves showed the association between ALDH7A1 expression and patient overall survival (OS) and disease-free survival (DFS). c-d. IHC assay was used to detect the ALDH7A1 and RASGEF1B expression pattern in same patient with serial section. e. scatter plot showed the IHC staining score correlation between RASGEF1B and ALDH7A1. f. immunofluorescence staining performed in HCC tissues using anti-RASGEF1B (green) and anti-ALDH7A1 (red) antibodies. g-i. RASGEF1B overexpression Hepa1-6 cells with or without ALDH7A1 silencing were subcutaneously inoculated into mice (N = 5 per group). Tumor image (g), tumor growth curves (h) and tumor weight (i) were shown. j-k. RASGEF1B overexpression Hepa1-6-luc cells with or without ALDH7A1 silencing were used to establish orthotopic liver tumor. IVIS results (j), histology and H&E staining results (k). **p < 0.01; ****p < 0.0001

Discussion

Neoplasms exhibit a dual pathological nature, encompassing genetic alterations and metabolic dysregulation. Dynamic crosstalk between metabolic reprogramming and epigenetic modifications plays a pivotal role in modulating tumor malignancy [30]. Accumulating evidence has demonstrated that epigenetic modifications dynamically alter the expression pattern of key metabolic enzymes, thus reprogramming cellular metabolism. In contrast, chemical heterogeneity in metabolites (e.g. acetyl, methyl, or phosphoryl) alter the activity and stability of rate-limiting enzymes in cancer metabolism by enzymatically or non-enzymatically approaches [31]. Since epigenetic modifications and metabolism reprogram critically govern the tumorigenesis and development, their precise mechanistic interactions remain incompletely understood.

In a previous study, we reported that RASGEF1B was significantly downregulated in HCC. Additionally, the results of survival analysis using an online database confirmed that RASGEF1B overexpression was associated with favorable prognosis for patients with HCC [14]. However, the downstream mechanism through which RASGEF1B inhibits HCC progression remains to be elucidated. As reported previously, the RAS signaling axis can hierarchically coordinate the normal biological processes of eukaryotes [32]. The RAS signaling pathway represents one of the most extensively characterized oncogenic cascades in cancer biology, whose activity is precisely balanced through the antagonistic regulation by GTPase-activating proteins (GAPs) and guanine nucleotide exchange factors (GEFs) [33]. Accumulating evidence indicates that upregulating GEFs expression promotes tumor progression by sustaining RAS/MAPK pathway activation, ultimately leading to poor patient outcomes [34]. RASGEF1A plays an essential role in the proliferation and progression of intrahepatic cholangiocarcinoma [35]. In this study, in vitro and in vivo experiments were performed to confirm that RASGEF1B upregulation significantly inhibited tumor progression in HCC, offering a new perspective on the long-standing conventional understanding. As shown by other researchers, kinetic analyses have demonstrated that RASGEF1A exhibits markedly higher catalytic efficiency than RASGEF1B, accelerating nucleotide release from small GTPases approximately 20-fold faster [36]. Furthermore, RASGEF1B displays significantly slower exchange rates than canonical RAPGEFs including Epac and C3G [37]. These findings potentially explain the noncanonical role of RASGEF1B in HCC. An early study established RASGEF1B as a pathogen-inducible factor in macrophages [38]. Therefore, RASGEF1B may suppress malignant behaviors through regulatory mechanisms distinct from those of classical GEFs function in HCC.

To fully investigate the mechanism underlying the noncanonical role of RASGEF1B in HCC, mRNA sequencing was performed to screen downstream regulator in RASGEF1B overexpression and relative control Huh7 cells. As confirmed by qPCR and western blotting, SNAI1 was the most significantly downregulated mRNA in the context of RASGEF1B overexpression. SNAl1 has been extensively characterized as a regulator of the EMT pathway in various cancer types, directly repressing E-cadherin transcription while activating mesenchymal markers [39]. Notably, elevated SNAI1 expression is significantly correlated with poor overall survival and progression-free survival in HCC patients [40]. In another study, SNAI1 was reported to promote HCC progression by inducing the a cholangiocellular phenotype, but not EMT in mice [41]. The above results demonstrate that SNAI1 facilitates HCC progression through multiple mechanisms. The results of the current study bridge our earlier observation of circDHPR-dependent EMT suppression with the novel finding that its downstream target RASGEF1B similarly inhibits SNAI1-mediated EMT [14]. However, the expression pattern of SNAI1 across cancers and the specific regulatory mechanism in HCC need to be further analyzed. To address this question, we conducted KEGG enrichment analysis to screen for pathway changes that might be affected by RASGEF1B and SNAI1. The KEGG results revealed that carbon metabolism is potentially involved in this regulatory network. Carbon metabolism, including glycolysis, the TCA cycle, the pentose phosphate pathway, gluconeogenesis and one-carbon metabolism, is a fundamental biochemical process in living organisms [25]. Interestingly, western blotting results revealed that RASGEF1B overexpression significantly inhibited the Warburg effect, While the role of SNAI1 in the EMT has been well studied, its role in metabolic reprogramming has been overlooked. In this study, we confirmed that SNAI1 expression was significantly upregulated in HCC tissue compared with normal liver tissue. Silencing SNAI1 with siRNAs inhibited tumorigenesis and tumor progression both in vitro and in vivo. Additionally, RASGEF1B overexpression increased the expression of key enzymes involved in one-carbon metabolism (SHMT and MS). However, changing the SNAI1 expression pattern does not affect the expression of SHMT or MS. The above results demonstrate that RASGEF1B potentially regulates the SNAI1 expression through the one-carbon metabolism. However, the precise metabolic circuitry through which RASGEF1B modulates SNAI1 expression remains elusive.

To address this concern, we employed high-resolution LC‒MS/MS to analyze metabolic alterations in RASGEF1B-overexpressing Huh7 cells. After data quality control and differential analysis were performed, Betaine was found to be the most significantly upregulated metabolite after RASGEF1B overexpression. As an important methyl donor, Betaine supports the methionine cycle by promoting the remethylation of homocysteine (Hcy) to methionine (Met), a reaction catalyzed by Betaine-homocysteine methyltransferase (BHMT). Methionine is further converted into S-adenosylmethionine (SAM), which serves as a direct methyl donor in DNA methylation [42]. In a recent study, Betaine was reported to inhibit the stem cell-like properties of hepatocellular carcinoma by activating autophagy [43]. Another study revealed that BHMT deficiency was associated with poor prognosis for patients with HCC [44]. Given the role of Betaine, our findings confirm that RASGEF1B inhibits SNAI1 mRNA transcription through Betaine-mediated DNA methylation. Nevertheless, the exact molecular mechanisms underlying RASGEF1B-mediated Betaine accumulation and its functional consequences in the tumor microenvironment require further investigation. According to the results of the metabolic network analysis, Betaine was the product of choline metabolism. The human choline catabolic pathway is governed by two rate-limiting enzymes: CHDH and ALDH7A1 [42]. Our results demonstrate that RASGEF1B enhances Betaine levels by reprogramming choline metabolism via the stabilization of ALDH7A1. However, structural analysis revealed that RASGEF1B lacks characteristic E3 ubiquitin ligase domains (RING or HECT), suggesting an indirect regulatory mechanism. These findings prompted us to identify direct interactors that regulate ALDH7A1 protein stability. Comprehensive analysis across three independent databases revealed that BMI1 was the most likely interaction partner for ALDH7A1. Further molecular docking was used to confirm the above results. Additionally, the CHX and IP assays confirmed that BMI1 regulated ALDH7A1 protein stability through ubiquitination, consistent with the findings of a previous study [45]. Furthermore, our study demonstrated that BMI1 mediates the ubiquitination of ALDH7A1 at Lys375. As predicted by the STRING database, ALDH7A1 did not directly interact with RASGEF1B. Interestingly, the results of the PPI analysis showed that ALDH7A1 can bind to RAPGEF3 and RAPGEF4, which share a conserved guanine nucleotide exchange factor (GEF) domain with RASGEF1B. The molecular docking results demonstrated that compared with the ALDH7A1–BMI1 complex, the RASGEF1B-ALDH7A1 interaction exhibited stronger van der Waals forces and a higher Buried Surface Area (Ų). The results of co-IP and GST pull-down confirmed that RASGEF1B shields ALDH7A1 from BMI1-dependent ubiquitination via competitive interactions. Nevertheless, comprehensive biophysical analyses (e.g., microscale thermophoresis and surface plasmon resonance) remain essential for elucidating the distinct mechanistic properties of ALDH7A1–BMI1 versus ALDH7A1–RASGEF1B complexes.

Conclusion

Collectively, our findings demonstrated that RASGEF1B suppresses hepatocellular carcinoma progression by stabilizing ALDH7A1 to remodel choline metabolism for Betaine-mediated SNAI1 hypermethylation. The expression pattern of RASGEF1B in HCC extends the long-standing conventional understanding. Moreover, our study reveals a metabolic‒epigenetic axis as an innovative therapeutic approach for improving clinical outcomes in patients with HCC.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Funding

This study was supported by the Clinical Research Plan for SHDC (SHDC2020CR5007, and TF2024YZZY01), the Guangdong Basic and Applied Basic Research Foundation (2023A1515010141, and 2024A1515012205), Key-Area Research and Development Program of Guangdong Province (2023B1111020008), Foundation of President of Zhujiang Hospital (yzjj2023ms03), and Longgang Medical Technology R&D Support Program (LGKCYLWS2024-10).

Data availability

The data generated in this study are available within the article and its supplementary data files.

Declarations

Ethics approval and consent to participate

This study was approved by the ethical committee of Zhujiang Hospital and Sun Yat-sen University Cancer Center. Informed consent was obtained from all patients. The animal experimental protocols were approved by the Animal Ethics Committee of Zhujiang Hospital (LAEC-2023–031). All methods were carried out in accordance with the relevant ethical guidelines and regulations.

Conflict of interest

The authors declare no conflict of interest.