RANBP1 promotes oral cancer progression through activating YAP1 and NDUFB3 derived from the database of single-cell sequencing

Abstract

Oral cavity squamous cell carcinoma (OSCC) is the most common head and neck cancer, thriving in microenvironments composed of cancer cells, stromal tissue, and extracellular matrix. Extensive local invasion and cervical nodal metastasis can lead to poor treatment outcomes for OSCC. We created a single-cell transcriptomic database of OSCC using biopsies from 10 individuals to identify candidate genes. Ran-specific binding protein 1 (RANBP1) was identified as a gene with differential expression between malignant and normal epithelial cells. Analysis of the Cancer Genome Atlas Program (TCGA) OSCC and Taiwanese cohorts confirmed that higher RANBP1 levels are significantly associated with worse prognosis. Functional tests showed that knocking down RANBP1 reduced cell proliferation and invasion by decreasing proteins involved in focal adhesion, invadopodia formation, and epithelial-mesenchymal transition. Inhibiting RANBP1 also decreased vascular spread in zebrafish tumor xenografts, while overexpressing RANBP1 had the opposite effect. The overexpression of RANBP1 significantly increased tumor growth in NOD/SCID xenografts. RANBP1 was positively correlated with the oxidative phosphorylation pathway. Cells with reduced RANBP1 showed lower levels of NADH ubiquinone oxidoreductase subunit B3 (NDUFB3) and had impaired mitochondrial function. Additionally, RANBP1 depletion decreased activation of Yes1 associated transcriptional regulator (YAP1). Increased NDUFB3 expression in RANBP1-overexpressing cells was reversed by YAP1 inhibitor verteporfin treatment or knocking down YAP1. RANBP1 enhances a more invasive microenvironment of OSCC.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12935-026-04361-9.

Introduction

Unsatisfactory treatment outcomes may be attributed to the frequent presentation of extensive local invasion in OSCC cases and the high likelihood of cervical nodal metastasis [1]. While numerous studies have identified prognostic biomarkers, the discovery of novel biomarkers and associated therapeutic targets holds promise for improving the treatment outcomes of OSCC patients.

Single-cell RNA sequencing enables the delineation of cellular subpopulations and elucidates the dynamics of tumor development [2]. Additionally, it facilitates the identification of novel cell-specific markers and the investigation of cell–cell interactions within the microenvironment [3]. Several studies have explored oral cancer stem cells, fibroblasts and immune cell infiltration at the single-cell level [4], but the key biomarkers expressed on malignant epithelial cells driving OSCC progression remain to be fully elucidated. In the current study, the single-cell RNA-seq dataset was primarily utilized to identify candidate genes associated with malignant cells in OSCC through differential expression and pathway enrichment analyses. By employing single-cell RNA sequencing technology, we identified Ran-specific binding protein 1 (RANBP1) as a novel candidate marker of malignant epithelial cell populations.

RANBP1 is a member of the RAS superfamily and interacts with Ras-related nuclear protein (RAN)-GTP/GDP and Ran GTPase-activating protein 1 to form the Ran network. This network plays a crucial role in mediating the nucleocytoplasmic transport of proteins, nucleic acids and microRNAs and is implicated in the regulation of chromosome segregation, proliferation and mitotic stability [5]. Elevated levels of RANBP1 have been associated with aberrant centriole splitting and the generation of aneuploidy or polyploid cells during mitosis, potentially contributing to both tumor initiation and progression [5]. Recent studies have indicated high RANBP1 expression in various human cancers, including hepatocellular carcinoma and breast cancer [6, 7]. However, the role of RANBP1 in OSCC remains poorly understood.

In this study, we utilized single-cell RNA sequencing to examine the transcriptomic profiles of malignant epithelial populations within OSCC tumors. The results of the bioinformatic analysis revealed that the differentially expressed genes were primarily related to pathways associated with cell proliferation and invasion, notably the RAN signaling pathway. Among these genes, RANBP1 exhibited the most significant difference and was selected for further investigation as a potential target. Analysis of the TCGA OSCC cohort (OSCC-TCGA) and a Taiwanese cohort (OSCC-TW) revealed that elevated RANBP1 levels in OSCC patients were associated with a poorer prognosis. Inhibition of RANBP1 reduced cell proliferation and invasion in vitro. Furthermore, in vivo experiments demonstrated that both the tumor size and the number of metastatic-like tumors were decreased when zebrafish embryos were injected with RANBP1-knockdown primary OSCC cells compared to mock cancer cells. Mechanistically, we found that the knockdown of RANBP1 reduced cell viability and suppressed invasion by modulating YAP1/NDUFB3 signaling in OSCC cells. Collectively, our results provide mechanistic insights into a novel role for RANBP1 in promoting OSCC invasion.

Results

Single-cell expression atlas and cell typing of the tumor microenvironment in OSCC

We performed scRNA-seq profiling on 9 tumors and 8 adjacent normal tissues from 10 treatment-naïve OSCC patients to explore cellular diversity in OSCC tumors (Fig. 1A and Supplementary Table 1). After quality filtering and doublet removal, the gene expression profiles of 43,830 cells. Following gene expression normalization, we identified 18 discrete transcriptional clusters using a graph-based clustering method (Supplementary Fig. S1A). Cell clusters were annotated based on the average expression of curated gene sets [8], identifying immune and nonimmune cells (Fig. 1B). As shown in Fig. 1C, 17,374 immune cells and 26,456 nonimmune cells were identified from the 43,830 cells. Nonimmune cells (n = 26,456) were reclustered into 23 discrete clusters (Supplementary Fig. S1B). These clusters were further annotated based on the average expression of curated gene sets, revealing 8,233 fibroblasts, 5,317 endothelial cells, and 12,906 epithelial cells (Fig. 1D). To distinguish malignant from nonmalignant cells, we inferred large-scale copy number variations (CNVs) based on the scRNA-seq data. Fibroblasts and endothelial cells were used as controls, and cancer cells displayed greater changes in relative expression intensities across the genome (Fig. 1E). In total, 7,060 malignant and 5,846 nonmalignant cells were identified from 12,906 epithelial cells in 17 tissue samples (Supplementary Fig. S2). As shown in Fig. 1F 6862 malignant cells were isolated from tumor tissues and constituted 71.3% of the total cells in tumor tissues, while 3078 nonmalignant cells were isolated from normal tissues and constituted 94% of the total cells in normal tissues. To identify potential malignancy-related targets, we generated a volcano plot to visualize the log2 expression fold change (fold change > 1.5) and the -log10 p-value (p-value < 0.05) between malignant cells in tumors and nonmalignant cells in normal tissues (Fig. 1G). This analysis revealed 586 significantly altered transcripts, including 221 downregulated and 365 upregulated genes, in malignant cells within tumors (Supplementary Table 2). Hierarchical clustering analysis separated the gene expression profiles of the two groups (Fig. 1H). Using Ingenuity Pathway Analysis (IPA), we identified the top 20 enriched canonical pathways, ranked by their ratios and positive z-scores. These pathways were linked to cancer cell proliferation and invasion (Fig. 1I and Supplementary Table 3). RAN (Ras-related nuclear protein) signaling was identified as one of the top three enriched pathways, exhibiting a –log (p-value) of 3.58 and a z-score of 2.828. RAN signaling, part of the Ras superfamily, mainly governs nucleocytoplasmic transport and mitosis. The activation of this pathway correlates with increased epithelial invasion and metastasis [9]. Our single-cell transcriptomic analysis revealed that RAN signaling genes were mainly upregulated in malignant cells within tumor tissues (Supplementary Fig. S3). This pathway warrants further investigation to identify potential target genes.

Fig. 1.

Investigation of potential targets in malignant epithelial cells of OSCC via scRNA-seq. (A) Workflow diagram showing the collection and processing of fresh biopsy samples (9 primary OSCC tumor samples and 8 normal tissue samples) for scRNA-seq. (B) The intensity of marker genes used to distinguish immune and nonimmune cells. (C) t-SNE plots of immune (17,374 cells) and nonimmune (26,456 cells) cell clusters. (D) t-SNE plots showing the reclustering of nonimmune cells (fibroblasts, 8233 cells; endothelial cells, 5317 cells; epithelial cells, 12,906 cells). (E) Putative copy number gains (red) and losses (blue) inferred from gene expression by averaging expression across sliding windows of 100 genes along the indicated chromosomes. (F) t-SNE plots of clusters of malignant cells from tumor tissues (6862 cells) and nonmalignant cells from normal tissues (3078 cells). (G) The volcano plot displays DEGs from the single-cell RNA-seq analysis. The x-axis shows the log2-fold change values, and the y-axis shows the -log10 p values for the differentially expressed genes. (H) Expression heatmap analysis of transcriptome datasets from the nonmalignant cells of adjacent normal tissues and the malignant cells of tumor tissues of OSCC patients. (I) Upregulated genes in malignant cells compared with nonmalignant cells were assessed for their enrichment in canonical pathways via IPA. The results of the pathway analysis are summarized as the ratio. The pathways are labeled along the y-axis

Elevated levels of RANBP1 in patients with OSCC are associated with a poor prognosis

To investigate the role of Ran signaling in OSCC progression, we analyzed the expression levels of RANBP1, RAN, KPNA2, XPO1, RANBP2, KPNA4, KPNB1, and RANGAP1 (Fig. 2A and Supplementary Fig. S4A). The expression levels of RANBP1, RAN, KPNA2 and XPO1 were significantly greater in the malignant cells of tumor tissues than in the normal epithelial cells of resected normal tissues (Fig. 2A). Among these genes, RANBP1 showed the highest expression (Supplementary Fig. S4B). High expression of RANBP1 has been reported in several human cancers, including colorectal and ovarian cancer [5, 10]. RANBP1 influences cell movement and alters the cancer microenvironment through cytokine regulation. [11, 12]. However, its role in OSCC has not yet been elucidated. The mRNA level of RANBP1 was greater in 191 pairs of OSCC tumor tissues than in adjacent normal tissues (p = 0.0156, Fig. 2B). Additionally, RANBP1 expression was increased in OSCC tumor tissues compared to normal tissues in the OSCC-TCGA cohort (Fig. 2C). Immunohistochemical analysis of RANBP1 expression revealed strong and uniform cytoplasmic staining in OSCC (Fig. 2D). In our cohort, patients with OSCC tumor tissues exhibiting RANBP1 mRNA levels greater than those of their normal counterparts (tumor/normal counterpart > 1) had significantly shorter overall survival than those with low RANBP1 expression (Fig. 2E). In the OSCC-TCGA cohort, the overall survival rate also significantly differed between patients with high and low RANBP1 expression (Fig. 2F). In the OSCC-TW cohort, distant metastasis-free survival was significantly shorter in patients with high RANBP1 expression (Fig. 2G). Disease-specific survival was not significant in patients with high RANBP1 expression in our cohort (p = 0.532), but the trend is similar (Supplementary Fig. S5A). In the OSCC-TCGA dataset, progression-free survival and disease-specific survival significantly differed between patients with high and low RANBP1 expression levels (Fig. 2H and Supplementary Fig. S5B). Clinicopathological analysis demonstrated that the RANBP1 expression levels in OSCC tumors were positively associated with perineural invasion and bone invasion (p = 0.037 and 0.021, respectively) (Supplementary Table 4). Furthermore, the multivariate analyses for overall survival and distant metastasis-free survival indicated notable differences in outcomes for patients with high versus low RANBP1 expression. However, no significant differences were observed in disease-specific or locoregional recurrence-free survival (Supplementary Table 5). The multivariate survival was analyzed from OSCC-TCGA database, Overall survival indicated notable differences in outcomes for RANBP1 level, age, T stage and N stage (Supplementary Table 6).

Fig. 2.

Association of high RANBP1 expression with shorter survival in patients with OSCC. (A) Violin plots showing the expression of RANBP1, RAN, KPNA2 and XPO1 in the RAN signaling pathway from the single-cell dataset. The p-values were analyzed by t-test. (B) RANBP1 transcript levels in 191 paired OSCC tissues were determined by qPCR. (C) Transcript expression levels of RANBP1 in the OSCC-TCGA dataset. The p-value from (B) and (C) were analyzed by the Mann‒Whitney test. (D) IHC staining showing the RANBP1 protein levels in OSCC patients (original magnification 40x (left panel), 200x (right panel)). (E) Kaplan‒Meier plot of the overall survival of patients stratified by RANBP1 expression (RANBP1 mRNA levels (tumor/normal counterpart > 1) as high RANBP1) (p = 0.0375). (F) Kaplan‒Meier plot of the overall survival of 310 patients in the OSCC-TCGA dataset stratified by RANBP1 expression (p = 0.0196). (G) Distant metastasis-free survival of patients stratified by RANBP1 expression (p = 0.0355). (H) Kaplan‒Meier plot of progression-free survival for 310 patients in the OSCC TCGA dataset stratified by the expression of RANBP1 (p = 0.0313). The survival p-values were calculated by using log-rank (Mantel–Cox) tests

Knockdown of RANBP1 inhibits the proliferation and invasion of OSCC cells

Clinicopathological analyses indicated that RANBP1 was involved in tumor cell invasion. The RANBP1 protein level was significantly lower in cells transfected with the RANBP1 siRNA than in those transfected with the control siRNA (NC) (Fig. 3A and B). Real-time cell growth assays revealed that the RANBP1-knockdown KOSC3 and SCC4 cells had lower proliferation rates than the control cells (Fig. 3C). The CCK-8 assay also showed decreased cell proliferation in RANBP1-knockdown KOSC3 and SCC4 cells. (Fig. 3D).

Fig. 3.

RANBP1 is involved in the proliferation of OSCC cells. (A) The protein expression of RANBP1 in NC and siRANBP1 cells was analyzed by western blotting. β-Actin was used as the loading control. (B) The quantitative level of RANBP1 from control and RANBP1-knockdown KOSC3 and SCC4 cells were quantitated by Image J. β-Actin was used as the loading control. The data are presented as the means and SDs of at least triplicate experiments. The statistical differences were determined using paired t-test. (C) Real-time proliferation of KOSC3 and SCC4 cells was monitored by the xCELLigence system. The data are presented as the means and SDs of triplicate experiments. (D) Cell proliferation assay (CCK-8) was performed in RANBP1 knockdown KOSC3 and SCC4 cells. The p-value was analyzed by two-way ANOVA

The invasion ability is also significantly reduced in the RANBP1-knockdown KOSC3 and SCC4 cells (Fig. 4A and B). Since enhanced activation of focal adhesions is a hallmark of cancer metastasis, subsequent western blotting revealed decreased levels of phosphorylated focal adhesion kinase (p-FAK(Tyr397)) and paxilin (p-PXN (Y118)) in KOSC3, SCC4 and primary OSCC cells with RANBP1 knockdown (Fig. 4C). Invadopodia, which are critical for cancer cells to penetrate anatomical barriers [13], were also affected. Specifically, invadopodia markers such as matrix metalloproteinase-2 (MMP2) and cortactin (CTTN) were downregulated in KOSC3, SCC4 and primary OSCC cells with RANBP1 knockdown (Fig. 4C). Inhibition of RANBP1 downregulated the expression of the EMT markers N-cadherin (N-cad) and vimentin (VIM) while increasing the expression of E-cadherin (E-cad) (Fig. 4C and Supplemental Fig. S6). Furthermore, immunocytochemistry revealed that the levels of p-FAK and p-PXN, associated with the formation of actin bundles, were lower in RANBP1-knockdown SCC4 cells and primary OSCC cells than in control cells (Fig. 4D, E and Supplemental Fig. S7).

Fig. 4.

RANBP1 is involved in the invasion of OSCC cells. Transwell invasion assay of (A) KOSC3 and (B) SCC4. Original magnification: × 100 (right). Quantitative analysis of the invasion assays (left). The data are presented as the means and SD obtained from three independent experiments. The p-values were analyzed by unpaired t-test. (C) Protein analysis of focal adhesion, invadopodia and EMT signaling pathway components in KOSC3, SCC4 and primary OSCC cells. β-Actin was used as the loading control. Immunostaining for (D) p-FAK (red) and (E) p-PXN (red) in SCC4 and primary OSCC cells. Alexa Fluor 488 phalloidin (green) was used to stain F-actin. Nuclei were stained with Hoechst 33342. Scale bars, 10 μm.

The expression level of RANBP1 correlates with the oxidative phosphorylation (OXPHOS) pathway and regulates mitochondrial function in OSCC

To clarify how RANBP1 influences the progression of OSCC, we used IPA to analyze the pathways associated with RANBP1⁺ versus RANBP1⁻ malignant cells in tumor tissues. Canonical pathways from the IPA database were ranked by their ratio and with a positive z-score, and the top 20 pathways were shown in Supplementary Table 7. We found that high expression of RANBP1 was positively associated with cell motility pathways, such as the EIF2 signaling pathway. According to IPA and GSEA pathway analysis, RANBP1 also positively associated with cell motility pathways, such as the EIF2 signaling pathway and mTOR Signaling (Fig. 5A). The EIF2 signaling pathway mediates general translation regulation in response to stress but can also promote tumor survival and resistance by upregulating specific cancer-promoting genes, particularly through ATF4 expression [14]. The mTOR signaling pathway is a well-known central regulator of cell growth, proliferation, and metabolism that is often hyperactivated in cancer, promoting uncontrolled growth and survival. Additionally, OXPHOS was the pathway most positively correlated with RANBP1, with a –log (p-value) of 36.6 and a z-score of 7.366 (Supplementary Table 7). OXPHOS was also one of the top pathways identified by GSEA, indicating that genes associated with high RANBP1 expression were highly enriched in the OXPHOS signature (Fig. 5A, enrichment score (ES) = 0.78). Increasing evidence indicates that OXPHOS serves as an active metabolic pathway in numerous tumors, implying that OXPHOS inhibitors can effectively suppress these cancers [15, 16]. Mutations in mitochondrial genes offer growth benefits in the process of tumorigenesis [17]. The total cellular and mitochondrial reactive oxygen species (ROS) levels were greater in the RANBP1-knockdown KOSC3 and SCC4 cells than in the control cells (Fig. 5B and C). Mito-stress assay assessed mitochondrial function to analyze changes in the oxygen consumption rate (OCR). Respiratory capacity was reduced in KOSC3, SCC4 and primary OSCC cells with RANBP1 knockdown (Fig. 5D). Following RANBP1 knockdown, there was a decline in non-mitochondrial oxygen consumption, basal respiration, maximal respiration, proton leakage, and ATP production compared to control cells, indicating a disruption in mitochondrial function (Fig. 5E). As shown in Fig. 5A and Supplementary Table 8, the core gene NADH: ubiquinone oxidoreductase subunit B3 (NDUFB3) had the highest rank metric score. Violin plots showed that NDUFB3 expression was significantly increased in malignant cells in tumor tissues, according to the single-cell sequencing data (Fig. 5F). The mRNA level of NDUFB3 was also elevated in the high RANBP1-expressing group in our cohort (Fig. 5G, p = 0.0062). NDUFB3 levels were weakly positively correlated with RANBP1 levels according to the single-cell sequencing data, our cohort data, and the TCGA database (Fig. 5H, I and Supplementary Fig. S8). Immunostaining in tumor tissues also showed a positive correlation of RANBP1 and NDUFB3 (Fig. 5J). These results imply that RANBP1 positively correlates with the expression levels of NDUFB3. As shown in Supplementary Fig. S9A and S9B, cell proliferation and invasion were increased in NDUFB3-overexpressing OSCC cells. To strengthen this hypothesis, we analyzed the mRNA levels of NDUFB3 in KOSC3, SCC4, and primary OSCC cells following RANBP1 knockdown. As expected, the mRNA and protein level of NDUFB3 was significantly decreased in the RANBP1-knockdown cells (Fig. 5K and L). In RANBP1-knockdown KOSC3 cells, cell proliferation and invasion were markedly decreased; however, this decrease was rescued by NDUFB3 overexpression (Supplementary Fig. S9C and S9D). These results indicate that RANBP1 modulates the expression levels of NDUFB3.

Fig. 5.

RANBP1 positively correlates with the oxidative phosphorylation (OXPHOS) pathway and influences the function of mitochondria. (A) The top 10 pathways from gene set enrichment analysis (GSEA) was identified by the normalized enrichment score (NES) (left panel). Enrichment plots for OXPHOS enriched in GSEA hallmark analysis showing the profile of the running ES (right panel). (B) Total cellular ROS production in KOSC3 and SCC4 cells was determined by staining with 10 μM CM-H2DCFDA for 30 min. The p-value was analyzed by paired t-test. (C) Mitochondrial ROS production in KOSC3 and SCC4 cells was determined by staining with 10 μM MitoSOX for 30 min. The p-value was analyzed by paired t-test. (D) Seahorse analysis of the oxygen consumption rate in KOSC3, SCC4 and primary OSCC cells with or without siRANBP1 transfection. (E) Quantification of non-mitochondrial oxygen consumption, basal respiration, maximal respiration, proton leakage and ATP production. The p-value was analyzed by paired t-test. (F) Violin plots representing the expression of NDUFB3 from the single-cell dataset. (G) Transcript expression levels of NDUFB3 in OSCC patient subgroups stratified by high and low RANBP1 expression. The p-value was analyzed by the Mann‒Whitney test. (H) Correlation analysis between the expression of RANBP1 and that of NDUFB3 in tumor malignant cells from single-cell dataset. (I) Correlation analysis between the transcript P/N fold of RANBP1 and NDUFB3 from OSCC patients. Correlations were analyzed by Spearman correlation. (J) Immunostaining for RANBP1 (red) and NDUFB3 (green) in tumor tissues. Nuclei were stained with Hoechst 33,342. Scale bars, 10 μm. (K) Transcription and (L) protein levels of NDUFB3 in NC and siRANBP1 KOSC3, SCC4 and primary OSCC cells. β-Actin was used as a normalization control. The data are representative of three independent experiments. The p-value was analyzed by paired t-test

Knockdown of RANBP1 suppresses the activation of YAP1 in OSCC cells

Next, we investigated the mechanism by which RANBP1 influences NDUFB3 using IPA. Supplementary Fig. S10 illustrates that Yes-associated protein 1 (YAP1) had emerged as a potential target within the RANBP1-affected NDUFB3 pathway. YAP1 serves as the downstream effector of the Hippo pathway, playing vital roles in cell proliferation, embryonic development, regeneration, and cancer progression [18]. Compared to that in control cells, the expression of active YAP1 was decreased in KOSC3, SCC4, and primary OSCC cells with RANBP1 knockdown (Fig. 6A). It is well known that phosphorylation of YAP1 at serine residues leads to its inactivation by subsequent degradation. The levels of phospho-YAP1 at Ser127 and Ser397 were increased in RANBP1-knockdown KOSC3, SCC4 and primary OSCC cells. In contrast, RANBP1 overexpression in KOSC3, SCC4 and primary OSCC cells led to lower levels of p-YAP1 at Ser127 and Ser397 (Supplementary Fig. S11). The mRNA levels of the YAP1 downstream targets AXL and AMOT were also decreased in KOSC3, SCC4 and primary OSCC cells with RANBP1 knockdown (Fig. 6B). The knockdown of RANBP1 reduced the nuclear translocation of YAP1 in KOSC3, SCC4 and primary OSCC cells (Fig. 6C). The ratio of fluorescence intensity of nucleus to cytosol was quantified in Fig. 6D. These results imply that RANBP1 regulates the activation of YAP1. RANBP1 is associated with poorer survival and is linked to perineural invasion and bone invasion in our cohort (Supplementary Table 4). IHC stainings of RANBP1 and NDUFB3 were higher in the patients with cervical metastasis compared to the patients without metastasis (Supplementary Fig. S12). IHC stainings of p-YAP1 (Ser127) were decreased in the patients with cervical metastasis compared to the patients without metastasis (Supplementary Fig. S12). According to our results, RANBP1 positively regulated the expression of NDUFB3 and the activation of YAP1. We determined the expression of NDUFB3 in RANBP1-overexpressing and YAP1-knockdown cells to investigate whether YAP1 is involved in the RANBP1-induced NDUFB3 pathway. As shown in Fig. 6E, the expression of NDUFB3 was increased in RANBP1-overexpressing cells, and it was inhibited upon YAP1 knockdown. We further examined the expression of NDUFB3 in RANBP1-overexpressing KOSC3, SCC4 and primary OSCC cells and found that it was inhibited by treatment with a YAP1 inhibitor (verteporfin) (Fig. 6F and Supplementary Fig. S13). These findings collectively suggest that RANBP1 induces NDUFB3 expression through YAP1.

Fig. 6.

Decreased the activation of YAP1 in RANBP1-knockdown OSCC cells. (A) The protein expression of active YAP1, YAP1 and RANBP1 was analyzed by western blotting. β-Actin was used as the loading control. (B) The transcript levels of AXL and AMOT in NC- and RANBP1-specific siRNA-transfected KOSC3, SCC4 and primary OSCC cells. The data are representative of three independent experiments. The p-value was analyzed by paired t-test. Immunofluorescence staining for (C) YAP1 and (D) quantification of the nuclear/cytoplasmic ratios of YAP1 staining in KOSC3, SCC4 and primary OSCC cells. Nuclei were stained with Hoechst 33,342. Scale bars, 10 μm. The p-value was analyzed by the Mann‒Whitney test. (E) The protein expression of active-YAP1, YAP1, NDUFB3 and Flag in RANBP1-overexpressing with or without YAP1 knockdown KOSC3 cells was analyzed by western blotting. β-Actin was used as the loading control. (F) The protein expression of active-YAP1, YAP1, NDUFB3 and Flag in RANBP1-overexpressing upon verteporfin-treated KOSC3, SCC4 and primary OSCC cells was analyzed by western blotting. β-Actin was used as the loading control

RANBP1 promotes cell proliferation and invasion through YAP1/NDUFB3 signaling

To determine the physiological significance of upregulated RANBP1 expression in OSCC, we investigated whether RANBP1 plays a role in cell proliferation and invasion through YAP1/NDUFB3 signaling. In RANBP1-overexpressing KOSC3, SCC4, and primary OSCC cells, cell proliferation was markedly increased; however, this increase was suppressed by verteporfin treatment (Fig. 7A). Overexpression of RANBP1 enhanced the invasion of KOSC3, SCC4, and primary OSCC cells, whereas verteporfin treatment inhibited this effect (Fig. 7B). The zebrafish has emerged as an excellent vertebrate model system for studying blood and lymphatic vascular development. By injecting RANBP1-knockdown and overexpressing cancer cells into zebrafish embryos, we assessed the impact of RANBP1 expression on tumor growth and metastasis. Primary OSCC cells were injected into zebrafish yolk, and after 7 days post fertilization, we measured the tumor area and the number of metastasis-like tumors. Figure 7C and D illustrate that zebrafish larvae injected with RANBP1-knockdown primary OSCC cells exhibited reductions in both tumor size and the number of metastasis-like tumors compared to the NC control. Tumor size and the number of metastasis-like tumors were increased in the zebrafish embryos injected with the RANBP1-overexpressing primary OSCC cells compared to vector control (Fig. 7E and F). Furthermore, we established xenograft mouse models with control or RANBP1-overexpressing KOSC3 cells. As shown in Fig. 8A-C, the overexpression of RANBP1 significantly augmented tumor growth of xenografts in vivo. Tumor weight in RANBP1-overexpressing groups was higher than those in control groups (Fig. 8D). These data support the oncogenic property of RANBP1 in OSCC tumorigenicity.

Fig. 7.

RANBP1-induced proliferation and invasion were reversed by verteporfin treatment. (A) Real-time cell proliferation of KOSC3, SCC4 and primary OSCC cells in which RANBP1 was overexpressed with or without verteporfin treatment, as determined by the xCELLigence system. The data are presented as the means and SDs of triplicate experiments. (B) The transfected cells were subjected to invasion assays. Representative microphotographs of the filters obtained from the invasion assays. Original magnification: × 100 (left). The data are presented as the means and SD of triplicate experiments. The results of the quantitative analysis of invasion were analyzed by unpaired t-test. (C) Representative images of disseminated tumor-like structures and metastasis-like tumors in zebrafish larvae injected with NC or siRANBP1 primary OSCC cells. Asterisks indicate the dissemination sites of metastasis-like tumors. (D) The quantification of the tumor area and metastasis-like tumor area, respectively. The p-value was analyzed by the Mann–Whitney U test. (E) Representative images of disseminated tumor-like structures and metastasis-like tumors in zebrafish embryos injected with vector or RANBP1-overexpressed primary OSCC cells. Asterisks indicate the dissemination sites of metastasis-like tumors. (F) The quantification of the tumor area and metastasis-like tumor area, respectively. The p-value was analyzed by the Mann–Whitney U test

Fig. 8.

Tumor growth in OSCC xenografts. (A) KOSC3_Control or KOSC3_RANBP1 cells were subcutaneously injected into the back of mice (n = 5 mice per group). (B) Images of xenografts dissected from NOD/SCID mice at the experimental endpoint. (C) Tumor growth data were presented as mean and SD (*p < 0.05, **p < 0.01, ***p < 0.001). (D) Tumor weight was measured at the experimental endpoint

Discussion

Previous OSCC research has demonstrated that RAN is overexpressed in OSCC tissues and is also essential for OSCC cell proliferation [19]. RANBP1 is a key regulator of the Ran pathway. Previous studies have shown that RANBP1 regulates protein trafficking between the cytoplasm and nucleus and is essential for cell viability and organismal development [12, 20]. In addition, RANBP1 plays important roles in mitotic progression and chromosomal stability, and its dysregulation has been associated with tumorigenesis [5, 21]. Elevated RANBP1 expression has been reported in multiple cancer types [6, 10, 22], and has been linked to cancer progression and therapeutic resistance [23–26]. However, its role in OSCC has not been fully characterized.

In the present study, we identified RANBP1 as a gene enriched in malignant epithelial cells using single-cell transcriptomic analysis. Consistent with previous findings, RANBP1 was overexpressed in OSCC tissues and associated with adverse clinicopathological features. Survival analyses in both our cohort and the TCGA dataset, including multivariate modeling, further support an association between RANBP1 expression and patient outcomes. Functional assays demonstrated that RANBP1 promotes cell proliferation and invasion, and these findings are supported by in vivo models, in which RANBP1 knockdown reduced tumor dissemination in zebrafish xenografts, whereas its overexpression enhanced tumor growth in mouse xenografts.

Cancer cells exhibit metabolic plasticity, enabling adaptation between glycolysis and oxidative phosphorylation (OXPHOS). Although aerobic glycolysis is a hallmark of cancer metabolism, accumulating evidence indicates that OXPHOS also plays an important role in OSCC progression [27]. Increased OXPHOS activity has been associated with enhanced tumor growth, metastatic potential, and poor clinical outcomes in OSCC [28], highlighting its contribution to tumor aggressiveness. In our study, genes involved in OXPHOS were predominantly upregulated in RANBP1-expressing malignant cells. In contrast, RANBP1 knockdown resulted in increased intracellular ROS levels, decreased ATP production, and reduced oxygen consumption, consistent with altered mitochondrial function. Pharmacologic inhibition of OXPHOS has been reported to produce similar effects, including reduced ATP production, increased ROS generation, and decreased cancer cell viability. For example, Nebivolol inhibits mitochondrial complex I and ATP synthase, leading to mitochondrial dysfunction and cancer cell death in OSCC models [29]. These observations suggest that OXPHOS inhibition may recapitulate certain aspects of the cellular changes observed following RANBP1 knockdown. However, as direct pharmacologic validation was not included in the present study, it remains to be further clarified whether the observed metabolic alterations represent primary drivers of the invasive phenotype or reflect downstream consequences of RANBP1 modulation.

Metastatic cancer cells, including those circulating in the bloodstream and colonizing distant sites, are known to rely on mitochondrial OXPHOS to meet elevated energy demands [30]. Moreover, drug-resistant metastatic tumors frequently depend on mitochondrial respiration for sustained survival and progression. Elevated ATP production has been associated with aggressive cancer phenotypes, including enhanced invasiveness, multidrug resistance, and spontaneous metastasis [31]. Inhibition of mitochondrial ATP synthase has been shown to suppress cancer cell proliferation, tumor sphere formation, and experimental metastasis while reducing respiration and increasing ROS [32]. In our study, RANBP1 knockdown resulted in reduced ATP production, impaired respiration, and decreased proliferation and invasion, consistent with these observations and supporting a role for mitochondrial function in regulating invasive phenotypes. However, the limitation of this study is that we didn’t perform the rescue experiments to further confirm the relationship between mitochondrial function and invasion.

Mitochondrial dysfunction can lead to increased ROS production through disruption of the electron transport chain and redox homeostasis [33, 34]. As byproducts of OXPHOS, ROS play complex and context-dependent roles in cancer progression [35]. At high levels, ROS can induce cytotoxicity, whereas moderate levels can support tumor cell survival and signaling pathways associated with cancer progression [36, 37]. In our model, RANBP1 knockdown increased ROS levels while reducing proliferation and invasion, suggesting that ROS accumulation may contribute to the observed cellular effects. However, whether modulation of ROS levels can rescue these phenotypes was not directly assessed in this study, and the precise contribution of ROS may depend on their intracellular levels and cellular context.

NDUFB3, a subunit of mitochondrial complex I, has been implicated in regulating mitochondrial respiration and ROS production. Previous studies have reported context-dependent effects of NDUFB3 including roles in oxidative stress regulation and cell survival [38–40]. In the present study, RANBP1 expression positively correlated with NDUFB3 levels, and functional experiments demonstrated that NDUFB3 overexpression partially restored proliferation and invasion in RANBP1-knockdown cells. These findings support a role for NDUFB3 as a downstream mediator of RANBP1-associated mitochondrial and cellular functions.

The transcriptional coactivator YAP1 is a key regulator of cell proliferation and survival and has been widely implicated in tumor progression [41]. In our study, RANBP1 knockdown reduced YAP1 activation and nuclear localization, whereas RANBP1 overexpression enhanced YAP1 activity. Inhibition of YAP1 suppressed RANBP1-induced NDUFB3 expression, suggesting that RANBP1 may regulate mitochondrial function through a YAP1-dependent mechanism. While the direct transcriptional regulation of NDUFB3 by YAP1 was not examined, these findings support a functional link between RANBP1, YAP1 signaling, and mitochondrial regulation.

The invasive progression of squamous cell carcinoma is a holistic process coordinated by upstream regulatory programs that bridge metabolic and phenotypic changes. Our findings align with emerging evidence that this progression is not merely a result of isolated signaling events but is a coordinated process where upstream regulatory programs, such as the RANBP1-YAP1 axis, converge to drive both metabolic adaptation and EMT-associated phenotypes [42, 43]. YAP1 functions as a key regulator coordinating mitochondrial bioenergetics and mesenchymal transition to support the energy requirements of metastasis [44]. Elevated YAP1 activity has been linked to malignant phenotypes and poor prognosis in OSCC [45, 46]; however, the potential link between RANBP1 and YAP1 in OSCC has not yet been elucidated. Previous studies have suggested that FAK can regulate YAP1 activation and nuclear translocation [47, 48]. In our study, RANBP1 knockdown reduced FAK phosphorylation, raising the possibility that RANBP1 may influence YAP1 activity through FAK-related signaling pathways. This potential interaction provides a basis for future investigation into the upstream regulation of YAP1 in this context.

Taken together, our findings suggest that RANBP1 is associated with OSCC progression and may influence tumor cell behavior through modulation of mitochondrial function and YAP1/NDUFB3 signaling. The integration of single-cell analysis, functional experiments, in vivo models, and external cohort validation provides a comprehensive framework for understanding the role of RANBP1 in OSCC, while further studies will help clarify the underlying mechanisms and clinical implications.

Conclusion

In conclusion, our study demonstrates a novel role of RANBP1 in OSCC based on single-cell transcriptomic analysis. RANBP1 is overexpressed in OSCC and is associated with adverse clinicopathological features and patient outcomes. Functional studies indicate that RANBP1 contributes to cell proliferation and invasion. Consistent with these findings, in vivo models further support the role of RANBP1 in tumor progression. RANBP1 knockdown reduced tumor dissemination in zebrafish xenografts, whereas its overexpression enhanced tumor growth in mouse xenograft models.

Mechanistically, RANBP1 is associated with modulation of mitochondrial function and YAP1 signaling, accompanied by changes in NDUFB3 expression. Rescue experiments further show that restoration of NDUFB3 partially reverses the reduced proliferation and invasion observed in RANBP1-knockdown cells, supporting its involvement in RANBP1-associated cellular effects.

Analyses in both the institutional cohort and the TCGA dataset, including multivariate modeling, further support an association between RANBP1 expression and clinical outcomes, providing additional validation in an independent cohort.

Overall, these findings suggest that RANBP1 is involved in OSCC progression and is associated with mitochondrial regulation and YAP1 signaling, providing a basis for future studies to further define its biological and clinical significance.

Materials and methods

Patient samples

Ten patients, provided 17 samples (8 normal and 9 tumor tissue samples). pathologically diagnosed with OSCC, were enrolled at Chang Gung Memorial Hospital, Taiwan. This research was approved by the institutional review board (IRB) of Linkou CGMH and conducted following the Declaration of Helsinki (approval no. 202102064B0). All participants provided informed consent through an IRB-approved form before sample collection. The inclusion criteria included men and women over 20 years of age with oral cancer who were willing to participate and provided written informed consent. Patients were excluded if they presented with recurrent disease, distant metastasis at initial diagnosis, underwent neoadjuvant treatment prior to surgery, or had synchronous or metachronous malignancies. Informed consent was obtained from all participants prior to sample collection. The clinicopathological characteristics are shown in Supplementary Table 1.

Generation of 3’-single cell expression librariesc

Fresh OSCC samples were collected in chilled MEM (Invitrogen, CA) supplemented with 1% penicillin/streptomycin and fungizone. Biopsies were trimmed into small pieces, followed by enzymatic digestion and mechanical dissociation (Miltenyi Biotech, CA). The cell suspensions were stringently washed with complete DMEM (Invitrogen, CA) and filtered through a prewetted 40 µm strainer. Calcein-AM/PI double staining method was used to evaluate the viability of collected cell populations. Cell suspensions with viability greater than 80% were partitioned into single cell gel beads-in-emulsion (GEMs) in microfluidics chip platform. Subsequent steps comprising reverse transcription and cDNA amplification were conducted to generate 3’-single-cell cDNA libraries (Chromium™ single cell 3’ v3.1 reagent kit, 10 × Genomics). The libraries generated were sequenced using NovaSeq 6000 Sequencing System (Illumina).

Analysis of single cell RNA sequencing data

Single-cell RNA sequencing results were mapped to the human genome GRC38 and quantified using CellRanger v.3.0.0. The output data was imported into the Seurat R package for quality control and downstream analyses including data normalization, variable gene selection, dimensionality reduction, and clustering [49]. Cells expressing fewer than 200 unique genes or more than 10% mitochondrial reads were excluded. Doublets were identified and removed using the DoubletsFinder package [50]. The annotations of cell identity for each cluster were defined based on the expression of known marker genes, while malignant epithelial cells were identified using the inferCNV package [8]. Differentially expressed genes (DEGs) were identified using a Bonferroni-adjusted p-value threshold of < 0.05. Gene Set Enrichment Analysis (GSEA) with hallmark gene sets from MsigDB was performed to elucidate the functional characteristics of each cluster [51]. The single-cell RNA sequencing data has been uploaded to the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (accession number GSE255118).

Immunohistochemistry

We like to thank the Biobank at the Chang Gung Memotrial Hospital, Lin-Kou, Taiwan for providing the service of immunohistochemistry. Immunohistochemical detection of RANBP1 was performed on formalin-fixed, paraffin-embedded Sects. (5 μm thick) using a bond polymer detection system and a bond automated machine with a polymer refine kit. The polyclonal rabbit anti-RANBP1 antibody (Sigma, SI-HPA065931) was used at a dilution of 1:50. After deparaffinization, antigen retrieval was performed with BOND Epitope Retrieval Solution 2 (Leica Biosystems). The sections were incubated with primary antibodies at room temperature for 60 min, followed by incubation with a polyclonal horseradish peroxidase (HRP)-conjugated anti-mouse/rabbit IgG secondary antibody to bind the primary antibody, and diaminobenzidine (DAB) was used to visualize the complexes. Then, the sections were counterstained with hematoxylin, dehydrated, cleared, and mounted.

Cell culture

KOSC3 cell line was provided by Yung-Chin Hsiao (Proteomic Translational Medicine Laboratory, Chang Gung University, Taoyuan City, Taiwan) and was cultured in RPMI 1640 (Invitrogen, CA) supplemented with 10% fetal bovine serum (FBS) (Gibco, NY), 100 units/mL penicillin and 100 μg/mL streptomycin (Gibco). SCC4 cell line was provided by Kun-Yi Chien (Proteome Analysis Laboratory, Chang Gung University, Taoyuan City, Taiwan) and cultured in DMEM/F12 (Invitrogen, CA) supplemented with 10% FBS, antibiotics and 400 ng/ml hydrocortisone. The cells were cultured at 37 °C in a humidified atmosphere of 95% air and 5% CO2 [52].

Primary OSCC cell culture

The method for culturing primary OSCC has beenpreviously described [53]. OSCC tissues were collected from patients and immediately soaked in cold Dulbecco’s modified Eagle’s medium (Invitrogen, CA) supplemented with 10 μg/mL gentamicin and 2 μg/mL amphotericin B, within 30 min after surgery. The tissues were then washed three times with cold PBS containing gentamicin, cut into small pieces (1 mm3), and incubated with 0.05% trypsin (Invitrogen, CA) at 37 °C for 20 min under gentle agitation. The explants were then plated onto mitomycin C-treated NIH/3T3 feeder layers and cultured in DMEM supplemented with 10% FBS, 10 μg/mL gentamicin, and 2 μg/mL amphotericin B (Invitrogen, CA). Epithelial cells outgrowing from the explants were passaged onto collagen-coated culture vesselsand maintained in keratinocyte growth medium (PromoCell, Heidelberg, Germany) to promote proliferation.

Cell transfection

OSCC cells were transfected with pooled RANBP1 siRNA (GGGCAAAACUGUUCCGAUU, UAGAAGCUCUCUCGGUGAA, AGAAUACAGACGAGUCCAA and GCAGGAAAGAGAUCGAAGA) or a nontargeting control siRNA (Dharmacon ON-TARGET^plus) using RNAiMAX (Invitrogen, CA) according to the manufacturer's instructions. For exogenous protein expression, OSCC cells were transiently transfected with either a full-length RANBP1 plasmid (pcDNA3.1 + /C-(K)-DYK-RANBP1) or an empty vector (GenScript, USA), using Lipofectamine 2000 transfection reagents (Invitrogen, CA).

CCK-8 cell proliferation assay

For cell proliferation assay, OSCC cells were collected 24 h post transfection, and seeded at a density of either 3 × 10² (SCC4) or 5 × 10² cells (KOSC3) per well in the 96-well plate. Cell proliferation was evaluated with a CCK-8 reagent (BIOTOOLS Co., Ltd., Taiwan) according to the manufacturer’s protocol. The optical density was measured at a wavelength of 450 nm using an ELISA reader (Molecular Devices, SpectraMax M2) [52].

Real-time monitoring of cell proliferation using the xCelligence system

An xCelligence real-time analyzer (ACEA Biosciences, San Diego, CA) was utilized to monitor the growth and survival of cells based on measurements of cell impedance. Before cell seeding, the background impedance of each E-plate was determined by loading 50 μl/well of culture medium. Cells (3 × 10³ cells/well in 100 μl) were then plated and allowed to grow at 37 °C. The impedance was recorded every 10 min for 3 to 5 days. All experiments were performed in quadruplicate.

Cell invasion assay

For the cell migration assay, the upper chambers of 24-well Transwell plates (0.8 μm pore size filter; Corning, Canton, NY) were coated with Matrigel™ Basement Membrane Matrix (BD Biosciences, San Jose, CA). Cells (200 μl; 1 × 10⁴ cells) were suspended in serum-free culture medium and added to the upper chambers of 24-well Transwell plates. The lower chambers were filled with medium containing 10% FBS. After a 24 h incubation at 37 °C, the chambers were washed, fixed, and stained, and the cells were counted [52].

Immunofluorescence

Cells were grown on sterile glass coverslips and fixed with 4% paraformaldehyde in PBS, following by permeabilization with 0.3% Triton X-100. Cells were then incubated with specific primary antibodies overnight at 4 °C and visualized with Alexa Fluor-conjugated secondary antibodies. Nuclei were stained with Hoechst33342 (Invitrogen).

Seahorse XFp cell mito stress test

Cell Mito Stress Tests (Agilent Technologies, Santa Clara, CA) were performed according to the manufacturer’s protocol on a Seahorse XFp instrument. OSCC cells were plated on sterile XFp plates in triplicate (KOSC3: 7000 cells, SCC4: 4000 cells) in 80 μL of cell growth medium. The next day, the growth medium was changed to 2% FBS cell growth medium, and the cells were then incubated for 24 h. On the day of the assay, the cells were equilibrated with assay medium (DMEM without sodium bicarbonate supplemented with 2 mM glutamine, and the pH was adjusted to 7.4.) in a non-CO2, humidified, 37 °C incubator for 1 h. The oxygen consumption rate (OCR), indicating mitochondrial function, was assessed through sequential injections of 10 μM oligomycin, Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) (0.5 μM for KOSC3; 1 μM for SCC4; 2 μM for primary OSCC), and 5 μM rotenone/antimycin A.

Detection of ROS production

For total cellular ROS determination, cells were treated with 10 μM CM-H2DCFDA (Thermo Fisher Scientific, MA) for 30 min at 37 °C. For mitochondrial ROS detection, cells were incubated with 2.5 mM mitoSOX for 30 min. The cells were detached from the wells by a 5 min incubation with trypsin–EDTA at 37 °C. The cells were washed twice and resuspended in PBS (containing 1% FBS) before flow cytometry analysis. Data are presented as the mean fluorescence intensity (MFI) for all cells.

Western blot analysis

The total protein in the lysates and supernatants was analyzed by western blotting [52]. The cells were collected using lysis buffer, and the protein concentration was determined by the Bradford assay. Protein samples were denatured at 95 °C, resolved on SDS‒polyacrylamide gels, and transferred onto PVDF membranes. The membranes were incubated overnight at 4 °C with appropriate dilutions of the indicated primary antibodies. The following antibodies were used at the dilution recommended by the manufacturers: RANBP1 (Atlas Antibodies Cat# HPA065931, RRID:AB_2685580 1:1000), phospho-FAK (Tyr397) (Cell Signaling Technology Cat# 8556, RRID:AB_10891442, 1:1000), FAK (Cell Signaling Technology Cat# 13,009, RRID:AB_2798086, 1:1500), phospho-PXN (Y118) (Abcam Cat# ab109547, RRID:AB_10866389, 1:3000), PXN (BD Biosciences Cat# 610,052, RRID:AB_397464, 1:3000), MMP2 (GeneTex Cat# GTX104577, RRID:AB_1950932, 1:500), CTTN (Abcam Cat# ab81208, RRID:AB_1640383, 1:5000), vimentin (Sigma-Aldrich Cat# V5255, RRID:AB_477625, 1:700), N-cadherin (Abcam Cat# ab76011, RRID:AB_1310479, 1:1000), E-cadherin (BD Biosciences Cat# 610,181, RRID:AB_397580, 1:5000), NDUFB3 (Abcam Cat# ab202585, RRID:AB_2890186, 1:1000), active-YAP (Abcam Cat# ab205270, RRID:AB_2813833, 1:1000) and β-Actin (Millipore Cat# MAB1501, RRID:AB_2223041, 1:5000). The membranes were then incubated with an appropriate dilution of an HRP-conjugated secondary antibody for 1 h. The immunoreactive bands were visualized by the use of enhanced chemiluminescence (ECL) reagents, and the signals were captured by X-ray films. The intensity of the bands was quantified by using ImageJ software (National Institutes of Health, Bethesda, Maryland). β-Actin was used as a loading control.

RNA extraction and quantitative reverse transcription polymerase chain reaction

Total RNA was extracted using TRIzol Reagent (Life Technologies, CA, USA) [52]. RNA was reverse transcribed into cDNA by oligo-dT (Bioman Scientific, Taipei, Taiwan) as the primer in the presence of reverse transcriptase (Superscript III, Invitrogen). qRT‒PCR was conducted using SsoFast™ EvaGreen® Supermix reagent (Bio-Rad, CA, USA) with an iQ5 real-time thermal cycler (Bio-Rad, CA, USA). The expression levels of the target genes were normalized to those of endogenous ACTB, and the data were analyzed using the 2^−ΔΔCt method. The sequences of the primers used for qRT‒PCR were as follows: RANBP1: 5’-ACCATGACCCTCAGTTTGAGCC-3’ and 5’-AGTGCCTCGCTCCTTCCATTCT-3’; NDUFB3: 5’-TGCTGTCAGGCAGAAGAACAG-3’ and 5’-CTTAGCCCTTTTGCAGCCAG-3’; AXL: 5’-GTTTGGAGCTGTGATGGAAGGC-3’ and 5’-CGCTTCACTCAGGAAATCCTCC-3’; AMOT: 5’-TCCAATGCTGGATCAGGCTTGC-3’ and 5’-ATTCAGCACGGTAGTCTCCACC-3’; and ACTB: 5’-ATGAAGTGTGACGTGGAC-3’ and 5’-GGAGGAGCAATGATCTTGA-3’.

Zebrafish embryonic xenograft tumor model

Zebrafish embryonic tumor xenograft models were established by injecting 250 LifeAct-RFP-expressing cancer cells into the yolk sac of fertilized embryos at the pre-4-cell stage from Tg (fli1a:EGFP) y1 line, an endothelial transgenic zebrafish model expressing EGFP. Unfertilized embryos were removed on day 1 post injection (dpi 1). Tumor vascular dissemination was assessed at 6 dpi. Tumor progression in live embryos was visualized using a high-resolution fluorescence microscope (Olympus IX83). At the end of the experiment, the embryos were euthanized and fixed in 4% paraformaldehyde for 16 h at 4 °C. Vascular dissemination of tumor-like structures with a tumor area ≥ 10 μm² were quantified using CellSens imaging software (Olympus). All zebrafish experiments were conducted in accordance with the Institutional Animal Care and Use Committee (IACUC) of National Health Research Institutes, Taiwan. This study was approved by Ethics Committee of Chang Gung University Institutional Animal Care and Use Committee (IACUC Approval No.: CGU110-151). The animal study followed the International Council for Laboratory Animal Science (ICLAS) guidelines.

Xenograft tumor formation

For the tumor propagation assay, vector control or RANBP1-overexpressing KOSC3 cells were mixed with Matrigel and subcutaneously injected into 6-week-old male NOD SCID mice (1 × 10⁷ cells each). All animal experiments were conducted in accordance with the Institutional Animal Care and Use Committee of Chang Gung University (Protocol Nos.: CGU114-073). Tumor volume was calculated using the following equation: (length × width²) / 2. The experiments were conducted in a randomized and observer-blinded manner.

Statistical analysis

The results of the ROS detection, mRNA and protein expression in the OSCC cell lines were analyzed using paired t-test. The results of the quantitative analysis of invasion were analyzed by unpaired t-test. Chi-square tests were used to determine the differences between RANBP1 expression and various clinicopathologic factors. The survival p-values were calculated by using log-rank (Mantel–Cox) tests. CCK-8 proliferation assay was analyzed by two-way ANOVA. Correlations were analyzed by Spearman correlation. The qPCR results from OSCC and normal counterpart tissues, the quantification of immunostaining in OSCC cells and the tumor area and metastasis-like tumor area from zebrafish were analyzed by the Mann‒Whitney test. Tumor growth data from mice were analyzed by t-test. The p-values of 0.05 or less were considered to indicate significance. Survival curves were plotted using the Kaplan‒Meier method and compared by the log-rank test. Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software, Inc., San Diego, CA, USA).

Abbreviations

CNV

Copy number variations

CTTN

Cortactin

DAB

Diaminobenzidine

DEGs

Differentially expressed genes

FAK

Focal adhesion kinase

FCCP

Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone

GEO

Gene expression omnibus

GSEA

Gene set enrichment analysis

HNSCC

Head and neck squamous cell carcinoma

IPA

Ingenuity pathway analysis

MMP2

Matrix metalloproteinase 2

NDUFB3

NADH ubiquinone oxidoreductase subunit B3

OCR

Oxygen consumption rate

OSCC

Oral cavity squamous cell carcinoma

OXPHOS

Oxidative phosphorylation

PXN

Paxillin

RAN

Ras-related nuclear protein

RANBP1

Ran-specific binding protein 1

ROS

Reactive oxygen species

TCGA

The cancer genome atlas program

YAP1

Yes1 associated transcriptional regulator

Author contributions

Wei-Chen Yen contributed to the study design, experiments, data analysis, and writing and revision of the manuscript. Shu-Chen Liu contributed to the study design and the writing and revision of the manuscript. Shih-Sheng Jiang, Chia-Yu Yang, Fang-Yu Tsai, Tzu-Tung Li, Yun-Hua Sui, Meng-Hsin Li, Hsing-Wen Cheng, Jui-Shan Yi, Chi-Yin Lee, Wan-Ling Wang, Tsung-You Tsai, Yenlin Huang contributed to the experiments and provided technical support. Kai-Ping Chang contributed to the study design, experiments, data analysis, and writing and revision of the manuscript. All authors have approved the final version of the work.

Funding

This work was made possible by grants from the Ministry of Science and Technology of Taiwan (NSTC 111–2314-B-182A-078-MY3 to Kai-Ping Chang; NSTC 112–2314-B-008–001-MY3 to Shu-Chen Liu) and Chang Gung Memorial Hospital (CMRPG3M0103, CMRPG3N0831 and CORPG3Q0051 to Kai-Ping Chang).

Data availability

The data that support the findings of this study are available within the article and the supplementary files. Original immunoblots are provided in the supplementary information file. The single-cell RNA sequencing data has been uploaded to the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database (accession number GSE255118).

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital, Taiwan (No. 202102064B0). Prior to sample collection, written informed consent was obtained from all participants. All zebrafish experiments were conducted in accordance with the Institutional Animal Care and Use Committee (IACUC) of National Health Research Institutes, Taiwan. This study was approved by Ethics Committee of Chang Gung University Institutional Animal Care and Use Committee (IACUC Approval No.: CGU110-151). The animal study followed the International Council for Laboratory Animal Science (ICLAS) guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.