Bubble charts illustrating various diseases or phenotypes associated with RAB5B protein were presented with the assistance of the Open Targets platform (URL: https://platform.opentargets.org/). To determine the specific location of RAB5B protein within cells, information from the HPA database (URL: https://www.proteinatlas.org/) was referenced, confirming the protein’s detailed subcellular location (12). Additionally, RNA-seq data in units of transcripts per million (TPM) were obtained from the UCSC XENA database encompassing 33 types of cancer tissues and their corresponding normal tissues (13). For data consistency, the Toil pipeline was employed to uniformly preprocess these data. Subsequently, the Gene_DE module 2.0 (TIMER2.0) was utilized to thoroughly analyze differences in RAB5B mRNA expression between cancer tissues and adjacent normal tissues (14). Furthermore, data from the CCLE database (URL: https://portals.broadinstitute.org/ccle/) were acquired to evaluate RAB5B expression levels across 29 distinct tissues. Except for the global differential gene expression analysis of cancer, all the TPM-formatted mRNA expression data and related clinical data relied on by the follow-up analysis were obtained from the TCGA cancer global cohort data set downloaded from the GDC data portal (2).

The GeneMANIA platform (http://www.genemania.org) was utilized to construct a protein-protein interaction (PPI) network (15), with the objective of obtaining interaction information relevant to RAB5B protein. This analytical step was critical for predicting RAB5B gene function, formulating functional hypotheses, and identifying genes with analogous biological roles. Through PPI network characterization, the study dissected RAB5B features across multiple dimensions, including physical interaction patterns, co-expression profiles, predictive network relationships, co-localization phenomena, genetic interactions, pathway participation, and shared protein domain architectures.

When the TISCH network tool (http://tisch1.comp-genomics.org/) was used to perform single cell data analysis, the following parameters were configured: gene RAB5B, major cell lines, and all types of cancers. Subsequently, the expression level of RAB5B in various cell types was evaluated and displayed in the form of a thermogram (16). Additionally, upon utilizing the SpatialDB online tool (https://www.spatialomics.org/SpatialDB/) (17), it was observed that the markers PECAM1 and VWF of M2 macrophages overlapped with the markers CD68 and CD163 of endothelial cells in skin melanoma.

Cox regression analysis was conducted with the assistance of the PanCanSurvPlot network platform (https://smuonco.shinyapps.io/PanCanSurvPlot/) to explore the correlation between RAB5B gene expression levels and multiple patient survival indices, including overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free survival (PFS). This analysis was based on data from the TCGA database, with data specifically selected from the IlluminaHiSeq sequencing platform. To effectively evaluate the prognostic value of RAB5B, an optimized cutoff grouping strategy was adopted for each cancer type, where the cutoff was determined based on the specific clinical and biological characteristics of each cancer using an optimization algorithm to select the expression level that best distinguishes prognosis risk. Ultimately, the ‘forestplot’ package in R was utilized to present the analysis results, specifically in the form of risk ratio (HR) and its 95% confidence interval (95% CI), thereby visually displaying the statistical estimation of RAB5B expression’s impact on patient survival.

To gain a deeper understanding of the frequency of four key types of genomic variation in tumors—mutation, amplification, deep deletion, and multiple changes—the cancer type overview function of the cBioPortal platform (https://www.cbioportal.org/) was utilized for analysis (17). Concurrently, the Gene Set Cancer Analysis (GSCA) platform (http://bioinfo.life.hust.edu.cn/GSCA/) was employed to comprehensively analyze the distribution of copy number variations (CNVs) of the RAB5B gene (18), along with the potential relationship between these CNVs and RAB5B mRNA expression levels. This included evaluating the specific effects of different CNVs on the RAB5B gene. The GSCA platform was further used to analyze differences in RAB5B gene methylation status across various cancer types, and to assess the correlation between methylation degree and RAB5B mRNA expression levels. Additionally, living conditions were compared between hypermethylated and hypomethylated patients. To quantify the correlation strength between RAB5B mRNA expression and CNV or methylation level, Spearman correlation analysis was conducted, with the P value corrected according to the false discovery rate (FDR). Furthermore, a t-test was applied to estimate differences in methylation levels between different cancers, and the P value was adjusted according to FDR. Finally, a log-rank test was performed to determine statistical differences in survival indicators such as OS, DSS, and PFS among groups.

Somatic mutation data was downloaded (initially could be rephrased to were downloaded for consistency with plural “data” in broader grammatical contexts, but “data” is often treated as singular in scientific contexts; here, keeping the original singular form to match the past tense transformation of the sentence) from the TCGA database (https://tcga.xenahubs.net/) and was analyzed using the “maftools” package in R to calculate the TMB and MSI of each tumor sample in TCGA (19). Spearman’s rank correlation coefficient was used to evaluate the correlation between RAB5B gene expression levels and TMB and MSI, and this relationship was visualized in the form of a radar chart using the “gradar” package in R. In immunotherapy, patients were categorized into four groups based on their treatment outcomes: disease progression (PD), disease stability (SD), complete remission (CR), and partial remission (PR). To further investigate the impact of RAB5B, the “Surv Miner” R package was used to determine the optimal RAB5B expression threshold, and two related immunotherapy cohorts were divided into low RAB5B expression and high RAB5B expression groups. Subsequently, the survival time and treatment response of these two groups were calculated. Additionally, the expression levels of MMR genes, including MLH1, MSH2, MSH6, PMS2, and EPCAM, were analyzed in different cancers, and the correlation between these MMR genes and RAB5B was further discussed. Finally, the “tidyverse” and “ggnewscale” packages in R were used to generate a heatmap to visually present these analysis results.

The ESTIMATE method was applied to quantify immune and stromal cell infiltration in malignant tumor tissues. Based on the Illumina platform, the “limma” and “ESTIMATE” packages in R were utilized to conduct an in-depth analysis of correlations between RAB5B expression and immune scores, stromal scores, and ESTIMATE total scores across individual tumor samples. Concurrently, immune cell infiltration patterns were evaluated across diverse tumor types. To comprehensively characterize the relationship between RAB5B and immune cell infiltration, the TIMER 2.0 tool was employed to accurately compute immune cell infiltration scores from the TCGA database. These scoring data were systematically collected and analyzed to elucidate potential associations between RAB5B expression and immune infiltration dynamics. Additionally, a curated list of 150 immune-related genes was obtained from TISDB resources (20), encompassing critical categories such as MHC molecules, immunosuppressive factors, chemokine receptors, immune activators, and chemokine proteins. Subsequent analyses leveraged this gene set to calculate correlations between RAB5B expression and immune-related gene expression profiles using R packages (“limma”, “pheatmap”, “ggplot2”), enabling a deeper, multi-dimensional understanding of their interrelationships.

GSEA and GSVA were applied to systematically assess the functional role of RAB5B in tumor biology. The analysis utilized R packages including “tidyverse”, “limma”, “org.Hs.eg.db”, “gseaplot2”, and “clusterProfiler”, incorporating C2 and C5 gene sets from the Molecular Signatures Database. Standardized enrichment scores were computed to compare false discovery rate (FDR)-adjusted differences between low- and high-RAB5B expression groups across diverse cancer types. Samples were stratified into high- and low-expression cohorts based on median RAB5B expression levels, with GSVA scores generated for all analyzed cancer types. The GSVA workflow further integrated R packages such as GSVA, GSEABase, ggthemes, BiocParallel, and visualization tools to ensure analytical rigor and comprehensive interpretation of results.

The CellMiner database (http://discover.nci.nih.gov/cellminer/) revealed the possible close relationship between RAB5B expression and drug response. This database was specifically designed for cancer researchers, integrating and analyzing molecular and pharmacological data from the NCI-60 cancer cell line - a critical resource for anticancer drug screening. The processed dataset from CellMiner was made available for download, containing RNA-seq-based RNA expression data and drug activity profiles in the NCI-60 panel.For subsequent analyses, the R package “limma” was employed to filter drug datasets, retaining only FDA-approved compounds or those in clinical trials, while excluding columns with over 80% missing values. The filtered data were visualized using “ggplot2” and “ggpubr” packages, applying a significance threshold of P < 0.05.Additionally, Autodock4 software was utilized to perform molecular docking simulations, evaluating methotrexate’s binding energy and interaction patterns with RAB5B protein structures obtained from CellMiner. The methotrexate structure was sourced from PubChem (https://pubchem.ncbi.nlm.nih.gov/), while the RAB5B protein’s 3D structure was predicted using AlphaFold (https://alphafold.ebi.ac.uk/). Finally, Pymol software was employed to visualize the docking models, providing intuitive insights into methotrexate-RAB5B interactions.

MCF-7 cells (Cell Bank of Chinese Academy of Sciences) were cultured in DMEM (Procell, Cat# CM10033) with 10% FBS (Procell, Cat# FSP500) at 37°C/5% CO₂. Cells were transfected at 60-70% confluency using Lipofectamine 3000 (Invitrogen, Cat# L3000015) with 50 nM RAB5B-targeting siRNA or negative control siRNA (GenePharma; sequences in Supplementary Table S2 ) in Opti-MEM (Gibco, Cat# 31985062). Experimental groups included: Untransfected Control, si-NC, and si-RAB5B. Data represent Mean ± SEM (Standard Error of the Mean), The SEM is calculated by dividing the standard deviation of the control group values in biological replicates by the square root of the sample size (√n).

Total RNA was extracted with TRIzol (Takara, Cat# 9108), quantified via NanoDrop 2000 (Thermo Scientific), and reverse-transcribed using PrimeScript RT Master Mix (Takara, Cat# RR036A). qPCR used SYBR Premix Ex Taq II (Takara, Cat# RR820A) on a QuantStudio 3 system (Applied Biosystems) with primers ( Supplementary Table S1 ). ACTB served as the reference gene. Relative expression was calculated by the 2^(−ΔΔCT) method. Statistical significance (p<0.05) was determined by one-way ANOVA with Tukey’s test (GraphPad Prism v9) across three biological replicates.

Cells (2×10⁴/well in 96-well plates) were treated with CCK-8 reagent (GLPBIO, Cat# GK10001) at 24h/48h post-seeding. Absorbance (450 nm) was measured using a BioTek Synergy H1 plate reader. Data were background-corrected and normalized to Untransfected Control (24h). Significance (p<0.05) was assessed by two-way ANOVA with Šídák’s test (three biological replicates, n=6 technical replicates/group).

Cells (1,000/well in 6-well plates) were cultured for 7 days (medium refreshed every 3 days), fixed with methanol, and stained with 0.5% crystal violet (Beyotime, Cat# C0121). Colonies (>50 cells) were scanned (Epson Perfection V600) and counted using ImageJ (v1.53t). Statistical analysis used unpaired t-tests (three biological replicates).

Confluent cells in 6-well plates were scratched with a sterile pipette tip, washed with PBS, and cultured in 2% FBS/DMEM. Wound images (T0/T48h) were captured at identical locations using a Nikon Ti-E microscope (DS-Fi3 camera, NIS-Elements v5.30). Wound area was measured in ImageJ (v1.53t), with closure percentage calculated as [(Area_T0 − Area_T48)/Area_T0] × 100%. Statistical significance (p<0.05) was determined by unpaired t-tests (three biological replicates, n=3 scratches/group).

In order to carry out the verification of bioinformatics, firstly, the obtained dataset was strictly screened, and the missing values and repetitive results were eliminated to ensure the accuracy and reliability of the data. Subsequently, the TPM value in the filtered dataset was transformed by log2 (TPM+1) for subsequent analysis. The Mann-Whitney U test (also called the Wilcoxon rank sum test) was used to compare the expression level of RAB5B between normal tissues and tumor tissues to verify whether the difference between them was statistically significant. For the data from the CCLE database covering different organizational sources, the Kruskal-Wallis method was used to comprehensively analyze the expression of RAB5B. When comparing the heterogeneous expression level of RAB5B between different groups or between tumors and normal tissues, the choice of analysis tool—paired t-test or unpaired t-test—depended flexibly on whether the samples were paired. In the whole analysis process, the significance level was set to a p-value less than 0.05 to ensure the accuracy and reliability of the results. All analysis work was done in R software (version 4.4.0, https://www.R-project.org).

Figure 1 illustrates the overall scope of the study. RAB5B-related diseases are explored on openttarget, and the bubble diagram shows that RAB5B is related to nasal polyps and neurodegenerative diseases ( Figure 2A ). As shown in Figure 2 , it can be seen intuitively that the expression of RAB5B protein in LIHC and BRCA is significantly higher in cancer tissues than in normal tissues. In order to explore the significance of RAB5B in malignant tumors, the expression level of RAB5B was carefully examined in 33 cancer types using comprehensive data sets obtained from TCGA and GTEx databases. The results of the study demonstrated that the expression level of the RAB5B gene was significantly up-regulated in ACC, BRCA, CHOL, LAML, LGG, LIHC, PAAD, PRAD, THCA and THYM. Conversely, the expression of RAB5B was significantly down-regulated (decreased) in BLCA, CESC, COAD, ESCA, READ, TGCT, UCEC and UCS when compared to their respective normal tissue counterparts ( Figure 2C ), that may be attributed to specific tumor microenvironment factors, genetic mutations, or the presence of different regulatory mechanisms affecting RAB5B expression in these particular cancers. Subsequently, the difference of RAB5BmRNA between cancer tissues and normal tissues was analyzed by TIMER2.0 with tumor tissues marked in red and normal tissues in blue RAB5B mRNA was significantly increased in three kinds of cancers: CHOL, LIHC, and PCPG and significantly decreased in COAD, KICH, KIRC and READ as shown in Figure 2D . In addition, the public documents of RAB5B expression were downloaded from CCLE database, and the differences of RAB5B expression in 33 organizations were studied. Figure 2E illustrates the widespread expression of RAB5B across diverse tissues, with its expression upregulated in most tumor types except vaginal cancer and myeloma. Different cancer types were distinguished using color coding for clarity. Final analysis involved constructing a PPI network using data from the GeneMANIA database (21), which identified the top 10 proteins demonstrating the strongest correlation with RAB5B. Figure 2F shows a protein interaction network diagram. From the diagram and correlation analysis, it can be seen that there is a strong physical interaction between RAB5B and protein such as RAB5A, RAB5C and RIN3. Simultaneously, this analysis elucidated potential signal pathways involving RAB5B and its molecular interactions, offering insights into the mechanisms through which RAB5B influences gene expression and cellular behavior.

To gain insights into the primary cell types expressing RAB5B in CME, an analysis of RAB5B expression was conducted at the single-cell level across a dataset comprising 77 cancer samples. The thermogram depicted in Figure 3A shows that the expression levels of RAB5A in 32 cell types were analyzed in 79 data sets by using the TISCH network tool. The results showed that RAB5B was mainly expressed in endothelial cells and immune cells (especially monocytes/macrophages). Notably, in GSE 123813 data set containing 86 BCC patients, RAB5B is mainly expressed in B cells, CD4Tcon, Treg and other cells ( Figure 3B ). At the same time, in GSE 72056 skin melanoma data set containing 4645 single cell samples, RAB5B was highly expressed in depleted Endothelial cell and monocytes in skin melanoma microenvironment ( Figure 3D ). In addition, the spatial transcription data on SpatialDB showed that RAB5B spatially overlapped with endothelial cell markers PECAM1 and VWF in BCC cancer tissues and M2 macrophage markers CD68 and CD163 in skin melanoma tissues ( Figures 3C, E ). The similar spatial distribution of these markers indicated that RAB5B might be co-expressed with CD68, CD163, PECAM1 and VWF in these cell types.

For each cancer type, one-way Cox regression analysis was utilized, based on data from the TCGA database, to investigate the association between RAB5B expression levels and the prognosis of various cancers. Survival analyses were conducted for OS, DSS, DFS and PFS. As illustrated in Figure 4 , a significant correlation between RAB5B and the prognosis of most cancers is evident. COX proportional hazard model was used to analyze the prognosis of all cancers. The forest map of Figure 4A shows that in BRCA, CESC, SKCM, DLBC, STAD, KIRP, READ and THYM, the expression of RAB5B is positively correlated with OS, but negatively correlated with LAML, ACC, BLCA, KICH, KIRC, HNSC, LIHC, LGG, LUAD, MESO, SARC, UCS and UVM. In DFS, the expression of RAB5B was positively correlated with SKCM, UCEC, STAD, LUAD, KIRP, THCA, READ, TGCT and THYM. In addition, RAB5B can protect DFS of ACC, BLCA, BRCA, KICH, HNSC, LGG, PAAD, PRAD, SARC, UCS and UVM ( Figure 4B ). To exclude non-cancer-related deaths and better reflect the impact of cancer treatment, an additional DSS analysis was conducted. It was found that RAB5B was regarded as a protective factor in ACC, BLCA, BRCA, KICH,KIRC,HNSC, LIHC, LGG, LUAD, MESO, SARC, UCS and UVM, but as a risk factor in SKCM, STAD, KIRC, THCA and READ ( Figure 4C ). Notably, RAB5B emerges as a risk factor for both LGG and LUAD in the PFS forest plot analysis, while concurrently functioning as a risk contributor to UCEC and THCA. Conversely, it exhibits protective properties in ACC, BRCA, CESC, and ESCAPRAD. The CNV distribution of the RAB5B gene across distinct cancer types is illustrated using pie charts, with color-coding distinguishing different CNV categories. Notably, RAB5B emerges as a risk factor for PFS in LGG and LUAD through forest plot analyses. Additionally, elevated RAB5B expression correlates with poor prognosis in UCEC and THCA. Conversely, RAB5B functions as a protective factor in ACC, BRCA, CESC, and ESCA, highlighting its context-dependent role in oncogenic outcomes. ( Figure 4D ).

The occurrence of many cancers often stems from the variation of genome (22). Through in-depth exploration of the cBioPortal database, genomic alterations of the RAB5B gene across various cancer types were identified. Specifically, as shown in Figure 5A , “amplification” is the main type of RAB5B variation in patients with UCS, and the frequency of this variation is the highest, exceeding nearly 4%. Additionally, analysis revealed variations in RAB5B gene mutation frequency across different cancer types. Figure 5B also presents the analysis results of CNV and methylation status of RAB5B gene in GSCA database. The CNV profile of the RAB5B gene across different cancer types is visualized through pie charts, where distinct color coding differentiates various CNV categories.The legend part details the CNV types represented by each color, including the percentage of heterozygous amplification, heterozygous deletion and CNV-free samples, among which heterozygous amplification accounts for the most. Figure 5C shows the correlation between CNV and mRNA expression. In CHOL, GBM, PAAD, ESCA, ACC, UCS, KIRP, OV and BRCA, there is a substantial positive correlation between the expression of RAB5B CNV and mRNA. On the contrary, this relationship is not significant in UVM, DLBC, LAML, THYM, PCPG, THCA, TGCT and PRAD. The differential impact of RAB5B gene CNV and generalized expression patterns on survival outcomes across cancer types was analyzed. Samples were stratified into subgroups based on CNV levels determined by GSCA, as illustrated in Figure 5D . Results indicated that elevated RAB5B CNV levels correlated with reduced overall survival in specific cancers, including UCS, KICH, and UCEC. Notably, KICH patients in the high-CNV subgroup exhibited consistently poor prognostic indicators across all evaluated survival metrics ( Figures 5E–G ). DNA methylation is a chemical modification process, which may lead to the inactivation of oncogenes, and then may promote the occurrence of cancer. As shown in Figure 5H , in-depth analysis of methylation differences between normal and tumor samples reveals that methylation levels in tumor samples of various cancer types, including LUSC, PAAD, LIHC, COAD, PRAD, HNSC, BRCA, LUAD, and BLCA, are generally elevated compared to corresponding normal samples. Notably, BLCA samples exhibit the highest methylation levels among these cancer types. Regarding the relationship between DNA methylation and RAB5B expression, a positive correlation is observed across all cancer types except UVM, especially in CHOL, GBM, PAAD, ACC, UCS, KIRP,OV etc ( Figure 5I ). Based on the median methylation level of tumor samples, they are categorized into hypermethylation and hypomethylation groups. Analysis further indicates that hypermethylation of the RAB5B gene correlates with overall survival rates in PRAD, HNSC, and KIRC patients, as depicted in Figure 5J . Notably, across various survival indicators, a consistent trend emerges: in KICH patients, high methylation levels of the RAB5B gene are associated with a lower risk of death, clearly illustrated in Figures 5K–M .

TMB and MSI are the key indexes to evaluate the efficacy of immune checkpoint inhibitors, which have a significant impact on the prognosis and treatment response of patients (23, 24). Specifically, TMB is determined by counting the number of somatic non-synonymous mutations in the coding region of tumor genome, while MSI is evaluated by measuring the mutation frequency caused by the defect of MMR mechanism per million base pairs. MMR plays a central role in correcting errors in DNA replication during cell division. When MMR gene expression is down-regulated or its function is impaired, mutations may occur in somatic cells. As shown in Figures 6A, B , the expression level of RAB5B gene is positively correlated with TMB in SKCM and THYM. However, in BRCA, COAD, LGG, LUAD, OV, SARC, THCA, UCE and UCS, RAB5B was negatively correlated with TMB. Similarly, in READ, RAB5B is positively correlated with MSI, but in DLBC, PCPG and SARC, RAB5B is negatively correlated with MSI. When assessing the potential of RAB5B expression as a biomarker for predicting the response to immunotherapy, two patient groups were selected, one having received anti-PD-1 treatment and the other anti-PD-L1 treatment. The results showed that the high expression of RAB5B was associated with high survival rate (88.46%) and low response rate to anti-PD-1 therapy (only 11.54%) in patients with urinary system tumors, as shown in Figure 6C . In the analysis of the IMvigor210 cohort, it was observed that patients with high RAB5B expression exhibited greater sensitivity to immunotherapy, as illustrated in Figure 6D . Furthermore, an in-depth exploration of the relationship between the expression of MMR genes (comprising MLH1, MSH2, MSH6, PMS2, and EPCAM) and RAB5B expression was conducted, with the specific findings presented in Figure 6E . The results showed that there was a significant correlation between the expression level of MMR gene and the expression level of RAB5B in all types of tumor samples. Especially in ACC, CHOL, KICH, KIRC, KIRP, LIHC, PCPG, PRAD, THCA and TYHM, the expression of MMR gene was positively correlated with the expression of RAB5B. These findings further strengthen the possibility of RAB5B as a potential biomarker to evaluate the stability of CHOL genome.

TME is a complex system, including stromal cells, fibroblasts, endothelial cells and various components of innate and adaptive immune system (25). A deep understanding of the complexity of TME is very important to improve the therapeutic effect. To investigate the relationship between RAB5B expression and TME, a comprehensive evaluation of the matrix index, immune index, and ESTIMATE score was conducted across several tumor types.The results showed that there was a significant negative statistical correlation between the expression level of RAB5B and matrix index, immune index and ESTIMATE score, especially in these tumors with high scores, and only a few cancers showed positive correlation, such as COAD and DLBC ( Figure 7A ). For the specific analysis results of BRCA, KICH and THCA, please see Figures 7B–J .

The relationship between immune cell infiltration and RAB5B expression was thoroughly investigated to elucidate the association between RAB5B and tumor immunity. Analysis confirmed that RAB5B expression correlates with infiltration levels of diverse immune cell types, including: B cells, Cancer-Associated Fibroblasts, Progenitor cells, Dendritic cells, Endothelial cells, Eosinophils, CD4+ T cells, Hematopoietic stem cells, NKT cells, Macrophages, Mast cells, CD8+ T cells, Monocyte, Myeloid-Derived Suppressor Cells, Neutrophils, Natural killer cells, T Follicular Helper, γ/δ and T cell. In many tumor types, the expression of RAB5B is positively correlated with the infiltration level of neutrophils in B cells, Cancer-Associated Fibroblasts, Progenitor cells,endothelial cells, Dendritic Cell, CD4+ T cells,mast cells,Monocyte, Neutrophils and Treg. Recent studies have underscored the critical role of immune cells, including CD4+ T cells, CAFs, MDSCs, neutrophils, and macrophages, in shaping responses to tumor immunotherapy ( Figure 8 ). The functional contributions of these immune subsets have emerged as pivotal determinants in therapeutic outcomes, necessitating their careful consideration in oncology treatment strategies. Evidence suggests that RAB5B modulates tumor progression, prognostic potential, and therapeutic efficacy through dynamic interactions with immune cell populations, highlighting its role as a regulatory node in immune-tumor crosstalk. This interplay underscores the significance of dissecting immune-related mechanisms when evaluating targeted interventions for cancer management.In addition, the relationship between RAB5B expression and immune response genes in 33 cases of tumor was discussed by co-regulation analysis of gene expression. Most of these genes encode MHC proteins, immune activation and inhibition proteins, chemokines and their receptors. The generated thermogram showed that most immune response genes were co-expressed with RAB5B ( Figure 9 ), and were positively correlated with RAB5B in all tumor types except BLCA, BRCA, GBM, LGG, MESO, SARC and THCA.

Through sequence function enrichment analysis, the function of RAB5B in cancer was discussed. Based on the median expression level of RAB5B in various cancer types, samples were categorized into low and high RAB5B groups. Subsequent GSEA and GSVA analyses were performed to uncover the biological processes associated with RAB5B. It was demonstrated that RAB5B exerts a positive regulatory influence on various immune-related activities in BRCA, KICH, and THCA, including B cell receptor signaling pathway, humoral immune response mediated by circulating immunoglobulin, immune response mediated by mmunoglobulin, cell response to protein with incorrect topology, cell response to unfolded protein, unfolded protein response mediated by ire1, autoimmune thyroid disease, graft-versus-host response, unfolded protein response of endoplasmic reticulum and so on. In addition, many KEGG pathways, such as nitrogen metabolism, allograft rejection, thyroid hormone synthesis, etc., all show significant correlation with RAB5B. These findings reveal that there is a potential close relationship between the expression level of RAB5B in tumor microenvironment and the immune activation state. In BRCA, RAB5B is predicted to be a positive regulator that promotes interleukin-4 (IL-4) cellular response and cellular response to misfolded proteins. In KICH, RAB5B is a positive regulator of renal water homeostasis and viral myocarditis. In THCA, RAB5B can positively regulate thyroid gland development and thyroid hormone synthesis. However, it also plays a negative regulatory role in certain biological processes, such as B cell receptor (BCR) signaling pathways, BCR signaling pathways, and immunoglobulin (Ig)-mediated immune responses in BRCA, triglyceride metabolic processes and linoleic acid metabolism in KICH, and immune response activating cell surface receptor signaling pathways in THCA ( Figure 10A ). To delve deeper into the biological role of RAB5B in these tumor types, GSVA was conducted. Figure 10B showcases the top 15 pathways that exhibit significant positive or negative correlations with RAB5B expression. The analysis results showed that the expression of RAB5B was significantly associated with a variety of immune-related pathways, including RAGE receptor binding, antifungal humoral response, positive regulation of killing of other organisms’ cells, neuroactive ligand-receptor interaction, primary immunodeficiency, IgE binding, miRNA loading onto RISC involved in miRNA-mediated gene silencing, RISC loading complex, transplant rejection, systemic lupus erythematosus, antigen processing and presentation, graft-versus-host disease, positive regulation of antigen processing and presentation, immune complex clearance, intestinal immune network for IgA production, and primary immunodeficiency. In addition, numerous pathways related to immune factors, such as negative regulation of interleukin-1α production, chemokine signaling pathway, cytokine-cytokine receptor interaction, interleukin-21 production, negative regulation of interleukin-21 production, immunoglobulin receptor activity, chemokine receptor binding, and cytokine and cytokine receptor interaction, were all associated with RAB5B.

Figure 11A reveals the list of drugs with significant correlation with the expression of RAB5B, showing that the expression level of RAB5B is positively correlated with the response to a series of drugs such as TW-37, sorafenib, niclosamide, methotrexate, docetaxel, daporinad, CD-437, BI-2536, AT13387, and 3-CL-AHPC. This analysis was conducted using data from the Cancer Therapeutics Response Portal (CTRP) and the Genomics of Drug Sensitivity in Cancer (GDSC) datasets, which provided comprehensive information on drug responses and gene expression profiles across a wide range of cancer cell lines.On the contrary, the expression of RAB5B was negatively correlated with the response to numerous cancer medications such as PX-12, PRIMA-1, PL-DI, panobinostat, navitoclax, CCT036477 and apicidin. To assess the binding affinity of drugs screened from CTPR and GDSC databases to RAB5B proteins, molecular docking analysis was performed for drugs with the most significant affinity using Autodock4 software. This analysis not only reveals the binding sites and interaction details of drugs and proteins, but also calculates the binding energy of each interaction.

Methotrexate, a chemotherapeutic drug, plays a key role in the treatment of various cancers (26), mainly by inhibiting dihydrofolate reductase to block the synthesis of tumor cells, thus inhibiting the growth and proliferation of tumors. Studies have shown that methotrexate, as a first-line chemotherapy drug, has shown remarkable efficacy in the treatment of breast cancer, and its therapeutic effect is closely related to the sensitivity of drugs to breast cancer cells (27). Methotrexate was selected for in-depth molecular docking analysis. The results showed that there was a significant hydrogen bond between methotrexate and RAB5B protein, accompanied by strong electrostatic interaction (as shown in Figure 11B ). The optimized conformation shows the binding energy of 2.23 kJ/mol, which strongly indicates that the binding between them is extremely stable.

The impact of RAB5B silencing on cell proliferation and migration was investigated to better understand its role in cancer progression. Cancer cell lines (MCF-7, SK-Hep-1) with RAB5B knockdown via siRNA were employed to explore its biological functions. Results from CCK-8 assays ( Figures 12B, F ) demonstrated significantly reduced proliferation in RAB5B-deficient cells compared to control groups, highlighting the importance of RAB5B in regulating cancer cell proliferation. Colony formation assays presented in Figures 12C, G further supported these findings, showing fewer colonies formed in RAB5B-knockdown cells and confirming that RAB5B silencing impairs the clonogenic potential of these cells.The effect of RAB5B silencing on cell migration was also evaluated using wound healing assays ( Figures 12D, H ). At both 0 hours and 48 hours, RAB5B-knockdown cells exhibited significantly reduced migration capacity compared to control cells, as indicated by smaller wound closure areas. This suggests that RAB5B plays a critical role in promoting the migratory properties of cancer cells, which is essential for tumor metastasis.The conserved suppression of proliferation and migration across distinct cancer types provides robust evidence for RAB5B’s critical role in cancer progression. These findings significantly advance our understanding of RAB5B as a pan-cancer oncogene and validate its potential as a therapeutic target, while highlighting the need for further in vivo validation and mechanistic dissection.