Abstract Background Lysosomes are monolayer membrane-encapsulated organelles containing acid hydrolases, crucial for intracellular substance breakdown and cellular homeostasis. They are also involved in autophagy. Although autophagy is linked to cancer, the role of lysosome-related genes in cervical cancer prognosis remains unclear. This study aimed to develop a prognostic model for cervical cancer based on lysosome-related genes and explore its applications in the tumor microenvironment, radiotherapy prognosis, and clinical pharmacology. Methods We identified differentially expressed lysosome-related genes in cervical cancer and normal tissues using the TCGA database. A prognostic model was constructed using LASSO-Cox regression, validated with ROC curves and PCA analysis, and further verified using the GEO dataset [36]GSE63514. In vitro and in vivo experiments were conducted to explore key genes, and their biological significance and pharmacological potential were analyzed. Results A five-gene (AP1B1, DNASE2, LAMP3, NPC1, and LAPTM4A) lysosome-associated prognostic model was developed. LAMP3 was identified as the most differentially expressed gene. Knockdown of LAMP3 significantly reduced cervical cancer cell migration and invasion through lysosomal and autophagic pathways. Daidzein was found to have high binding affinity for LAMP3, suggesting its therapeutic potential. Conclusion Lysosome-related gene modeling has significant clinical value. LAMP3 knockdown inhibits cervical cancer progression by reducing autophagy and lysosomal function. Daidzein shows potential as a novel therapeutic agent. However, further validation in larger cohorts is needed due to the limited sample size in this study. Supplementary Information The online version contains supplementary material available at 10.1186/s12885-025-14596-w. Keywords: LAMP3, Cervical cancer, Lysosome, Prognosis, Immune microenvironment Background Cervical cancer is one of the most prevalent malignant tumors in women globally and poses a significant threat to women’s health [[37]1, [38]2]. To combat this issue, the World Health Organization (WHO) introduced a strategy in 2020 aimed at reducing the global burden of cervical cancer, which consists of three key measures [[39]3]. Recent advancements in surgery, radiotherapy, chemotherapy, as well as the widespread use of early detection, screening, and HPV vaccination, have significantly reduced the incidence of cervical cancer in developed countries [[40]4]. However, in low- and middle-income countries, where economic development remains limited, cervical cancer prevention and control efforts remain a critical challenge, with approximately 85% of new cases and 90% of related deaths occurring in these regions [[41]5]. Lysosomes are essential regulators of cellular and organismal homeostasis and serve as potential therapeutic targets for various diseases [[42]6]. They play a crucial role in cellular senescence and death and are central to understanding the pathogenesis of conditions such as pancreatitis, atherosclerosis, autoimmune diseases, neurodegenerative disorders, lysosomal storage diseases, and cancer [[43]6, [44]7]. In malignancies, lysosome-regulated processes, including cell proliferation, invasion, radiotherapy resistance, and chemoresistance, have been extensively studied [[45]8]. As key participants in autophagy, lysosomes are closely associated with the development of cervical cancer. However, their role in cancer is not confined to the autophagy pathway; it also involves several other mechanisms, such as regulating the tumor microenvironment, modulating inflammatory pathways, inducing immunogenic cell death to influence cancer progression, regulating angiogenesis through glycosidases, and promoting tumor metastasis [[46]9]. Lysosome-associated membrane proteins (LAMPs) are crucial for the fusion of autophagosomes with lysosomes [[47]10]. LAMP3, a key member of the LAMP family, plays an indispensable role in autophagy. In systemic lupus erythematosus (SLE), LAMP3 induces cell death by regulating lysosomal membrane permeability [[48]11, [49]12]. In light of this, the aim of this study was to develop a predictive model based on lysosome-related genes to accurately predict the prognosis of cervical cancer patients and to analyze tumor microenvironmental characteristics. Additionally, we explored the potential application of this model in radiotherapy prognosis and clinical pharmacology, with the goal of providing a new theoretical foundation and practical guidance for the diagnosis and treatment of cervical cancer. The research flowchart for this study is shown in Fig. [50]1. Fig. 1. [51]Fig. 1 [52]Open in a new tab Research Workflow This figure presents the comprehensive workflow used to elucidate the role of the LAMP3 signaling pathway in cervical cancer progression via autophagy. It summarizes the systematic approach employed, including the databases, software, and tools utilized in the analysis Materials and methods Primary data collection and organization: construction of Lysosome-Related gene tags In this study, we collected pan-cancer clinical data, including RNA expression profiles, copy number variation (CNV) data, somatic mutation information, and data from 33 different cancers through the UCSC Xena database ([53]https://xena.ucsc.edu). For cervical cancer (CESC) patients, we selected cases that had undergone radiotherapy and had complete post-radiotherapy evaluations, which were subsequently analyzed for survival outcomes. Additionally, 121 lysosome-related genes were sourced from the Gene Set Enrichment Analysis (GSEA) database ([54]https://www.gsea-msigdb.org) (refer to Supplementary Table [55]1 for details). Table 1. Clinical characteristics of CC patients involved in the study Total N = 304 Testing set n = 152 Training set n = 152 P-value Age <=65 269(88.49%) 128(84.21%) 141(92.76%) 0.0311 > 65 35(11.51%) 24(15.79%) 11(7.24%) Grade G1 18(5.92%) 9(5.92%) 9(5.92%) 0.2376 G2 135(44.41%) 59(38.82%) 76(50%) G3 118(38.82%) 65(42.76%) 53(34.87%) G4 1(0.33%) 0(0%) 1(0.66%) unknow 32(10.53%) 19(12.5%) 13(8.55%) Stage Stage I 162(53.29%) 82(53.95%) 80(52.63%) 0.6153 Stage II 69(22.7%) 30(19.74%) 39(25.66%) Stage III 45(14.8%) 25(16.45%) 20(13.16%) Stage IV 21(6.91%) 11(7.24%) 10(6.58%) unknow 7(2.3%) 4(2.63%) 3(1.97%) T T1 140(46.05%) 76(50%) 64(42.11%) 0.222 T2 71(23.36%) 28(18.42%) 43(28.29%) T3 20(6.58%) 11(7.24%) 9(5.92%) T4 10(3.29%) 5(3.29%) 5(3.29%) unknow 63(20.72%) 32(21.05%) 31(20.39%) M M0 116(38.16%) 62(40.79%) 54(35.53%) 0.9456 M1 10(3.29%) 6(3.95%) 4(2.63%) unknow 178(58.55%) 84(55.26%) 94(61.84%) N N0 133(43.75%) 69(45.39%) 64(42.11%) 0.6066 N1 60(19.74%) 28(18.42%) 32(21.05%) unknow 111(36.51%) 55(36.18%) 56(36.84%) [56]Open in a new tab Using the TCGA-CESC dataset, we identified differentially expressed genes (P < 0.05) and mapped the corresponding copy number variation landscapes and mutation locations [[57]13]. To construct prognostic labels, we employed the following risk score formula, which was combined with univariate Cox regression analysis and Lasso-Cox regression analysis to identify the optimal gene combinations. Risk Score Formula: graphic file with name d33e412.gif Where “coef genes” represents the risk coefficients of genes, and “Exp genes” denotes the expression levels of genes in the TCGA-CESC dataset. Signature characteristics and predictive power In this study, Kaplan-Meier (KM) curves were employed to analyze differences in overall survival (OS) and progression-free survival (PFS) among 73 cervical cancer (CC) patients who had received radiotherapy. These curves were further used to investigate OS disparities across different risk groups. The prognostic distribution of patients in each risk group was visualized through patient prognostic scatter plots. To assess the independence of risk characteristics, we performed univariate and multivariate Cox regression analyses to compare the associations between risk scores and other clinical variables. Additionally, the predictive accuracy of the constructed labels was evaluated using the concordance index (C-index), area under the curve (AUC), and receiver operating characteristic (ROC) curves. Principal component analysis (PCA) was conducted to examine how the labels defined the spatial distribution of high- and low-risk patients. The biological significance of these labels was further explored by comparing two gene sets: the 36 differentially expressed genes (DEGs) and the genes within the labels. Finally, we developed column-line diagrams, a clinically accessible scoring system that integrates patients’ risk scores with clinical features likely to affect prognosis. This system aims to predict patients’ OS at 1, 3, and 5 years, offering strong support for clinical decision-making. Gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis To further investigate the biological functions and potential mechanisms of the target gene sets, we performed Gene Ontology (GO) analysis to systematically evaluate their biological roles at three levels: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). Furthermore, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis was conducted to identify significantly enriched pathways associated with the target gene sets, highlighting their critical roles in cellular physiological and pathological processes. Immune infiltration analysis and tumor immune dysfunction and exclusion (TIDE) score In this study, single-sample gene set enrichment analysis (ssGSEA) was utilized to assess the immune function enrichment in cervical cancer (CC) patients. Furthermore, the level of immune cell infiltration in the tumor microenvironment was systematically evaluated by integrating several advanced immune cell infiltration analysis algorithms, including CIBERSORT, CIBERSORT-abs, TIMER, XCELL, QUANTISEQ, MCPCOUNTER, and EPIC. To investigate the immune escape potential of tumors, we retrieved the Tumor Immune Dysfunction and Exclusion (TIDE) score from the TIDE database ([58]http://tide.dfci.harvard.edu/). The TIDE score, derived from tumor pretreatment gene expression profiles, effectively predicts the immune escape ability of tumors and provides a valuable basis for evaluating responses to immunotherapy [[59]14]. Tumor mutation burden (TMB) characteristics and chemotherapeutic drug sensitivity prediction In this study, the distribution of tumor mutation burden (TMB) was visualized using waterfall plots. After integrating mutation data with clinical outcomes, we performed survival analysis based on the combined index of TMB and risk score to assess its predictive value for patient prognosis. The NCI-60 is a diverse panel of 60 human cancer cell lines widely used in cancer research. The CellMiner database ([60]https://discover.nci.nih.gov/cellminer/) integrates molecular characterization and pharmacological data of the NCI-60 cell lines, providing a powerful tool for studying the relationship between chemotherapeutic drug activity and gene expression. Core gene screening To explore the key role of lysosome-related genes in tumor progression, we extracted the [61]GSE63514 dataset from the GEO database. This dataset contains rich sample information, providing a solid foundation for subsequent analysis. Based on the lysosome-related gene tags identified in the preliminary stage, we conducted secondary validation to ensure the accuracy and reliability of the gene tags. Subsequently, the Boruta algorithm was employed to rank the importance of lysosome-associated genes. As an advanced feature selection method, the Boruta algorithm effectively identifies core genes that significantly impact tumor progression [[62]15]. Cell culture, RT-PCR, and Western blot Cervical cancer cell lines HeLa and SiHa, as well as normal cervical epithelial cells H8, were obtained from Shenzhen Ollie Biotechnology Company. All cells were cultured in RPMI 1640 medium (Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, Grand Island, NY, USA) at 37 °C and 5% CO₂. RNA extraction and RT-PCR Total RNA was extracted from three sets of cells using the EastStep™ RNA Extraction Kit (LS1040). The extracted RNA was then reverse transcribed into complementary DNA (cDNA) using PrimeScript™ RT Master Mix (RR036A, Takara). GAPDH was used as the internal reference gene, and SYBR^® Green Premix (RR420A, Takara) was employed for real-time quantitative PCR amplification. The primer sequences for the five target genes are provided in Supplementary Table 2. Western blot Cellular protein samples were prepared using RIPA lysis buffer (Beyotime, P0013B) and SDS-PAGE protein sampling buffer (Beyotime, P0015). Proteins were separated by SDS-PAGE electrophoresis with the PAGE Gel Rapid Preparation Kit (Epizyme, Shanghai) and subsequently transferred onto a PVDF membrane. The membranes were then washed with 1× TBST buffer (Epizyme, Shanghai, PS121) and blocked with rapid closure buffer (Epizyme, PS108) for 1 h. The membranes were incubated overnight at 4 °C with a primary antibody, followed by incubation with a secondary antibody at 37 °C for 1 h. Finally, ECL Luminescence Reagent (Meilunbio, MA0186) was applied for chemiluminescence detection. Information regarding the antibodies used is provided in Supplementary Table 3. Source of cervical tissue samples and immunohistochemistry (IHC) This study involved 5 healthy donors and 5 cervical cancer (CC) patients, who provided full informed consent. The pathological types of their cervical biopsy specimens were confirmed by two or more experienced pathologists. The study was approved by the Ethics Committee of Fujian Cancer Hospital (approval number: SQ2022-230). The procedure for the immunohistochemistry (IHC) experiments is described in reference [[63]16]. Cell transfection and functional assays (CCK-8, scratch assay, transwell Assay) Cell transfection (stable and transient transfection) HeLa and SiHa cells were transfected using OBIO’s siRNA (sequence: 5’-agagauuauucucucucaaccuacutt-3’) to knock down the expression of the LAMP3 gene. The transfection procedure was performed according to the manufacturer’s instructions. For stable transfection, a shRNA lentiviral vector (sequence: 5’-agagauuauucucucucaaccuacutt-3’, cloned into the pLKO.1 vector) provided by OBIO was used to target the LAMP3 gene in HeLa and SiHa cells. The experimental procedure involved the following steps: (1) viral packaging and titer determination; (2) cell infection and puromycin selection (optimized concentration gradient at 2 µg/mL); (3) monoclonal screening and amplification; and (4) validation of LAMP3 knockdown efficiency through qPCR and Western blot. The process for constructing stable transfected cell lines strictly adhered to the vector operation specifications and safety guidelines for viral experiments. CCK-8 assay The CCK-8 assay was used to assess cell proliferation. Cells were seeded in 96-well plates at a density of 3000 cells per well and incubated at 37 °C with 5% CO₂ for various time points (0 h, 24 h, 48 h, 72 h, 96 h). After incubation, 10 µL of CCK-8 solution (Ape × Bio, K1018) was added to each well, and the cells were incubated for an additional 2 h. Absorbance values were measured at 450 nm using a multifunctional microplate reader (Tecan, Austria) to assess cell proliferation. Transwell assay The Transwell assay was used to evaluate the migration and invasion capabilities of the cells. For the migration assay, 50,000 cells in 200 µL of serum-free medium were seeded in the upper chamber of a Transwell system (Corning, USA). Subsequently, 500 µL of serum-containing medium was added to the lower chamber. For the invasion assay, cells were seeded in the upper chamber pre-coated with a matrix gel (Corning, USA; matrix gel: serum-free medium = 1:8). Cells were incubated at 37 °C for 24 h. After incubation, the Transwell chambers were removed, and the cells were fixed with 4% paraformaldehyde and stained with crystal violet to observe migration and invasion. Scratch assay The scratch assay was used to assess cell migration. Cells were seeded in 6-well plates at a density of approximately 90% confluence and incubated for 24 h. A uniform scratch was created using the tip of a P200 pipette. Photographs of the wound area were taken at 0 h, 24 h, and 48 h. The relative migration distance was measured using ImageJ software and calculated as follows: relative migration distance (%) = (migrated area/total area) × 100%. Drug prediction and molecular docking In drug development, accurately assessing protein-drug interactions is crucial for evaluating the suitability of target proteins as candidates for drug development. This study achieves this objective using the Drug Signature Database (DSigDB). DSigDB ([64]http://dsigdb.tanlab.org/DSigDBv1.0/) is a comprehensive resource platform containing data from 22,527 genes and 17,389 compounds, with coverage of 19,531 genes. It effectively links drugs and other compounds to their target genes. In this study, we uploaded the genomic data of the screened target proteins to DSigDB, enabling the prediction of potential drugs that interact with these target genes, thereby providing strong support for targeted gene therapy [[65]17]. To further investigate the mechanisms of action of drug candidates on target proteins and to comprehensively assess the druggability of these targets, we conducted molecular docking simulations to analyze the binding affinities and interaction patterns between drugs and targets at the atomic level. The molecular docking technique allows precise analysis of ligand-receptor binding affinity and the kinetics of the interaction. By identifying ligands with high affinity and favorable interaction characteristics, we were able to prioritize potential drug targets for experimental validation and optimize the design of drug candidates for these targets. During the experiments, we retrieved the 2D chemical structure data of the drugs from the PubChem database ([66]https://pubchem.ncbi.nlm.nih.gov) and obtained the crystal structure of the protein from the Protein Data Bank (PDB, [67]https://www.rcsb.org/). Molecular docking simulations were then performed using the CB-Dock2 platform ([68]https://cadd.labshare.cn/cb-dock2/php/index.php). CB-Dock2 is a powerful protein-ligand blind docking tool that integrates cavity detection, structure-based docking, and template-based docking, providing robust support for computer-aided drug discovery [[69]18]. Statistical analysis Data analysis was performed using the Perl programming language (version 5.30) and R software (version 4.1.2) to ensure efficient and accurate data processing. The experimental data were processed and visualized using GraphPad Prism software (version 9.0) and ImageJ software (version 1.53k). Statistical significance was determined with a threshold of P < 0.05 for all analyses. Results Construction of lysosome-related gene signatures In this study, we focused on the TCGA (The Cancer Genome Atlas) cohort and identified 36 differentially expressed genes (DEGs) from a set of 121 lysosomal pathway-related genes. Most of these DEGs exhibited high expression levels in tumor tissues (Fig. [70]2A). Further analysis revealed widespread copy number variation (CNV) of these 36 DEGs in cervical cancer (CC) (Figure S1A), with nearly all genes showing varying degrees of copy number amplification, while only a few genes displayed extensive CNV deletions. Notably, genes with a high frequency of deletions included AP3D1, AP3S3, and CTSO. Additionally, we mapped the specific locations of these CNVs using chromosome ring mapping (Figure S1B). Fig. 2. [71]Fig. 2 [72]Open in a new tab Identification of Lysosome-Associated Gene Characteristics (A) Thirty-six differentially expressed genes (DEGs) exhibited significant expression differences between cervical cancer tumor tissues and normal tissues (B) Eight genes associated with prognosis were identified through univariate Cox regression analysis and visualized as forest plots (C) LASSO regression analysis, utilizing the minimum criterion and lysosome-associated gene LASSO coefficients, was employed to cross-validate variable selection, with each curve representing a lysosome-related gene A total of 304 cervical cancer patients were included in the study and randomly assigned to the training and test sets in a nearly 1:1 ratio (Table [73]1). No statistically significant differences in clinical characteristics were observed between the two groups. Based on the 36 DEGs, we first identified eight genes in the training set that were significantly associated with prognosis, presenting their risk values in a forest plot (Fig. [74]2B). Next, we employed LASSO regression analysis to mitigate multicollinearity and developed a risk score formula for subsequent multivariate Cox regression analysis. The formula was then validated in the test set (Fig. [75]2C). Finally, we successfully constructed a gene signature consisting of five genes (AP1B1, DNASE2, LAMP3, NPC1, and LAPTM4A) for further validation and analysis. Application of lysosome-related gene signature To evaluate the prognostic and risk-predictive capabilities of the five identified genes, we applied the same risk score formula used in the training set to each individual in the test group. The results demonstrated that the constructed lysosome-related gene signature held significant prognostic value. Specifically, the signature was effective in predicting both overall survival (OS) and morbidity in patients (Fig. [76]3A and B). As the risk score increased, the survival rate of patients decreased significantly, while the mortality rate correspondingly increased. This finding was consistently validated across the training set, test set, and validation set. Fig. 3. [77]Fig. 3 [78]Open in a new tab Characterization of Lysosome-Associated Gene Tags (A) Overall survival (OS) analysis was performed in the TCGA, test, and training sets (B) Differences in risk profiles between the high-risk and low-risk groups were demonstrated in the TCGA, test, and training sets (C) Progression-free survival (PFS) analysis based on gene signatures was conducted in the TCGA set (D) Progression-free survival (PFS) analysis was also conducted for 73 patients who received radiotherapy, focusing on their overall survival (OS) Furthermore, the gene signature also displayed strong predictive validity for progression-free survival (PFS) (Fig. [79]3C). To further assess the prognostic power of this signature in specific treatment contexts, we extracted data from 73 patients who received radiotherapy and completed post-radiotherapy evaluation from the TCGA-CESC dataset (Table [80]2). Survival analysis revealed that the gene signature maintained its significant predictive capability for overall survival (OS) in radiotherapy patients (Fig. [81]3D). Table 2. Clinical characteristics of patients who received radiotherapy and had complete radiotherapy evaluation patients of radiotherapy (N = 73) Age <=65 65 > 65 8 T T1 23 T2 22 T3 13 T4 4 Unkown 11 Stage stageI 24 stageII 23 stageIII 14 stageIV 9 Unkown 3 G G1 2 G2 34 G3 25 Unkown 12 Measure of response CR 51 PR 7 PD 13 SD 2 [82]Open in a new tab Validation of lysosome-associated gene signature To further validate the independence and accuracy of the lysosome-associated gene signature, we initially conducted an independent prognostic analysis to determine whether the signature could serve as a diagnostic criterion independent of other clinical features. Multifactorial and unifactorial Cox regression analyses revealed that the p-values for the risk scores were below 0.05 (Fig. [83]4A), indicating that the signature can function as an independent prognostic predictor for patients of varying ages and tumor grades or stages. Fig. 4. [84]Fig. 4 [85]Open in a new tab Validation of Lysosome-Associated Gene Labeling (A) Lysosome-associated gene labeling was confirmed as an independent risk factor for overall survival (OS) in the TCGA cohort (B) The gene labeling was identified as an independent predictor, outperforming other clinicopathological features based on area under the curve (AUC) analysis (C) The concordance index (C-index) for this gene labeling was higher than that for other clinicopathological markers, indicating superior predictive performance (D) Principal component analysis (PCA) was conducted on 36 differentially expressed genes (DEGs) and lysosome-related genes In assessing the accuracy of the signature, we observed that its area under the curve (AUC) value (0.684) significantly surpassed that of other clinical characteristics (Fig. [86]4B). Specifically, the AUC values of the signature at 1, 3, and 5 years were 0.684, 0.672, and 0.724, respectively. Additionally, its concordance index (C-index) was notably higher than that of other clinical features, approaching 0.7 (Fig. [87]4C), suggesting that the signature possesses high diagnostic value and accuracy. To visualize patient risk stratification more intuitively, we employed principal component analysis (PCA) to determine the spatial distribution of patient risk. By comparing the differentiation effects of the two datasets, we found that the five-gene model demonstrated clearer boundaries and more accurate patient risk stratification (Fig. [88]4D). For clinical application, we constructed a column-line diagram that integrated risk scores and other clinical features. The calibration curve indicated that the column-line diagram most accurately predicted 1-year survival, with a calibration index of 0.885 (Fig. [89]S1C). Furthermore, gene ontology (GO) analysis revealed that the molecular functions of the differentially expressed genes between the high- and low-risk groups were predominantly enriched in receptor-ligand activity and biological processes related to humoral immune response (Fig. [90]S1D). KEGG pathway analysis further identified that genes encoding cytokine-cytokine receptor interactions and chemokine signaling were highly enriched in the high-risk group (Fig. S1E), suggesting a close association with immune function. Differential analysis of tumor immune infiltration and TIDE score Functional enrichment analysis revealed that the differentially expressed genes between the high-risk and low-risk groups were predominantly associated with immune-related pathways. Consequently, we further investigated the characteristics of immune infiltration in patients with varying risk scores. Significant differences in immune function between the two groups were primarily observed in the co-suppression and co-stimulation of antigen-presenting cells (APCs), T-cells, immune checkpoints, human leukocyte antigens (HLAs), type I interferon (IFN) responses, and pro-inflammatory responses. Most immune-related genes exhibited higher expression levels in the low-risk group (Fig. [91]5A-B). The heatmap of immune infiltration analysis showed a higher degree of CD8 + T-cell infiltration in the low-risk group, as well as significantly elevated immune and immune microenvironment scores compared to the high-risk group (Fig. [92]5C). Fig. 5. [93]Fig. 5 [94]Open in a new tab Immune Infiltration Analysis of Lysosome-Associated Gene Tags (A) A heatmap illustrating the differences in immune cell infiltration levels and immune function between high-risk and low-risk populations across all samples (B) Immune infiltration analysis using the TIMER, CIBERSORT, CIBERSORT-ABS, QUANTISEQ, MCPCOUNTER, XCELL, and EPIC algorithms, based on the immune infiltration heatmap (C) Analysis of immune checkpoint differences between the two groups, with data presented as mean ± standard deviation. Significance levels are indicated as P < 0.05, *P < 0.01, and **P < 0.001 (D) Comparison of TIDE scores between the high-risk and low-risk groups Given the critical role of immune checkpoints in immunotherapy, we further analyzed their expression between the two groups. The results indicated significant differences in the expression of multiple immune checkpoint inhibitors, particularly within the TNF-TNF receptor family (e.g., TNFRSF9, TNFRSF4, TNFRSF18, TNFRSF14, TNFSF4, and TNFSF18) (Fig. [95]5D). Additionally, the TIDE (Tumor Immune Dysfunction and Exclusion) score is an essential metric for predicting a patient’s response to immunotherapy. Upon analyzing the TIDE scores of cervical cancer patients, we found that the low-risk group had significantly higher TIDE scores than the high-risk group (Fig. [96]5E). This finding suggests that patients in the low-risk group may exhibit a lower response rate to immunotherapy compared to those in the high-risk group, indicating that the risk score may serve as a potential predictor for treatment response in clinical immunotherapy. Risk score combined with TMB and chemotherapeutic drug sensitivity prediction In analyzing the relationship between the risk score and tumor mutation characteristics, we observed that the proportion of mutated samples in the high-risk group (67.08%) was slightly higher than in the low-risk group (64.23%) (Fig. [97]6A). The genes TTN and MUC16 exhibited the highest mutation rates. Moreover, the mutation rate of TP53 differed significantly between the two groups. Specifically, the mutation rates of TTN and TP53 were higher in the low-risk group (35% and 8%, respectively) than in the high-risk group (25% and 6%, respectively), while the mutation rate of MUC16 was lower in the low-risk group (15%) than in the high-risk group (19%). Fig. 6. [98]Fig. 6 [99]Open in a new tab Tumor Mutation Burden (TMB) and Chemotherapeutic Drug Sensitivity Prediction Analysis (A) Somatic mutation profiles for the high-risk and low-risk groups, presented as waterfall plots (B) Kaplan-Meier survival curve analysis for four groups: high TMB and low-risk, low TMB and high-risk, high TMB and high-risk, and low TMB and low-risk (C) Correlation analysis of five key genes with chemotherapeutic drug sensitivity Tumor mutational burden (TMB) is a key indicator for assessing tumor prognosis. Survival analysis revealed that patients with both high TMB and low risk scores had the best prognosis, whereas those with low TMB and high-risk scores had the poorest prognosis (Fig. [100]6B). Chemotherapy remains a cornerstone in cervical cancer treatment. Using the CellMiner database, we investigated the correlation between the expression of five genes in the risk score and the sensitivity to chemotherapeutic drugs. The results demonstrated that the half-maximal inhibitory concentration (IC50) of 16 potential chemotherapeutic agents correlated to varying degrees with the expression of these five genes (Fig. [101]6C). These findings suggest that the risk score could have potential utility in predicting chemotherapeutic drug sensitivity. High expression characteristics of LAMP3 in cervical cancer tissues and cells In the PCR analysis of five genes within the model, four genes—AP1B1, DNASE2, LAMP3, and NPC1—demonstrated high expression levels in the cervical cancer cell lines HeLa and SiHa, with the exception of LAPTM4A. Among these, the expression of LAMP3 exhibited the most significant difference (Fig. [102]7A). Further Western blot experiments confirmed the high expression of LAMP3 in cervical cancer cell lines, corroborating the PCR results (Fig. [103]7B). Fig. 7. [104]Fig. 7 [105]Open in a new tab Validation of Gene Expression Differences in Labeling and LAMP3 Expression in Cervical Cancer (CC) Cells and Tissues (A) The expression levels of the five genes in the signature were examined in normal human cervical epithelial cell lines, as well as in HeLa and SiHa cell lines (B) LAMP3 protein expression levels were higher in HeLa and SiHa cells compared to H8 cells (C) Immunofluorescence results demonstrated that LAMP3 fluorescence intensity was greater in HeLa and SiHa cells than in H8 cells (D) LAMP3 expression was significantly elevated in cervical cancer tissues compared to normal cervical tissues Immunofluorescence (IF) experiments revealed that the fluorescence intensity of LAMP3 in HeLa and SiHa cell lines was notably higher than in the H8 cell line (Fig. [106]7C), further validating the elevated expression of LAMP3 in cervical cancer cells. To investigate the expression level of LAMP3 in cervical cancer tissues, we performed a comparative analysis of LAMP3 expression between cervical cancer tissues and normal cervical tissues. The results showed that LAMP3 expression was significantly higher in cervical cancer tissues than in normal cervical tissues (Fig. [107]7D), suggesting that LAMP3 may play a critical role in the development and progression of cervical cancer. Additionally, to further validate the reliability of our experimental results, we analyzed the [108]GSE63514 dataset. The expression patterns of LAMP3 and NPC1 in this dataset were highly consistent with those found in the TCGA database, as well as the cell and tissue experiments (Figure S2A). Furthermore, Boruta algorithm analysis confirmed that LAMP3 was the most important gene influencing prognosis, a finding that aligned with the PCR experiment results (Figure S2B-C). Knockdown of LAMP3 inhibits autophagy and lysosomal membrane function in cervical cancer cells, suppressing cell proliferation, migration, and invasion To investigate the role of LAMP3 in cervical cancer cells, we successfully knocked down LAMP3 expression in HeLa and SiHa cell lines using siRNA technology. The resulting cell lines were named HeLa-siLAMP3 and SiHa-siLAMP3, respectively. The knockdown efficiency was confirmed by PCR, meeting the experimental criteria (Fig. [109]8A), and Western blot analysis revealed a significant reduction in LAMP3 protein expression (Fig. [110]8B). Fig. 8. [111]Fig. 8 [112]Open in a new tab Knockdown of LAMP3 Significantly Delayed the Malignant Progression of Cervical Cancer Cells by Inhibiting Autophagy and Lysosomal Function (A) The knockdown efficiency of LAMP3 was assessed by quantitative PCR (qPCR) (B) Western blotting was used to measure LAMP3 protein expression levels to further confirm the knockdown effect (C) LAMP3 knockdown significantly reduced the migration and invasion abilities of cervical cancer cells (CC cells) (D) LAMP3 knockdown inhibited the proliferative capacity of CC cells (E) The wound healing ability of cervical cancer cells (CC cells) was weakened following LAMP3 knockdown (F) Western blot analysis revealed the mechanism of action following LAMP3 knockdown Functional assays demonstrated that LAMP3 knockdown markedly inhibited cell migration and invasion (Fig. [113]8C). CCK-8 assays indicated that the proliferation rate of HeLa-siLAMP3 cells was significantly lower than that of parental cells at 48 h, while SiHa-siLAMP3 cells exhibited similar changes at 72 h (Fig. [114]8D). Additionally, a cell scratch assay confirmed a significant decrease in cell migration following LAMP3 knockdown (Fig. [115]8E). To explore the potential mechanism underlying the effects of LAMP3 knockdown, we analyzed the expression of autophagy-related proteins (P62, LC3I, and LC3II) and lysosomal membrane proteins (LAMP1 and LAMP2) in the knocked-down and parental cell lines. The results showed that LAMP3 knockdown significantly reduced LAMP1 expression, while LAMP2 expression was elevated. Furthermore, the ratio of P62 and LC3II/I was significantly increased (Fig. [116]8F). These findings suggest that LAMP3 knockdown may impair cervical cancer cell proliferation, migration, and invasion by disrupting lysosomal membrane function and autophagy pathways. Consistent with these results, similar outcomes were observed using stable transfection (Figure S3A-F). Effects of lysosomal function inhibitors and autophagy agonists on cellular function in LAMP3 knockdown strains To further investigate the impact of LAMP3 knockdown on lysosomal function and cellular autophagy, we treated the LAMP3 knockdown cell line with the lysosomal inhibitor chloroquine (CQ) and the autophagy inducer rapamycin (RAPA) for 24 h. Cell scratch assays revealed a significant reduction in cell healing ability in the CQ-treated group, while the RAPA-treated group exhibited significantly enhanced healing compared to the LAMP3 knockdown cell line (Fig. [117]9A). Similarly, cell proliferation (Fig. [118]9B), migration, and invasion abilities (Fig. [119]9C) displayed comparable trends. Fig. 9. [120]Fig. 9 [121]Open in a new tab Validation of Autophagy and Lysosomal Function (A) Cell scratch assay of LAMP3 knockdown strain following treatment with chloroquine (CQ) or rapamycin (RAPA) (B) CCK-8 assay of LAMP3 knockdown strain following treatment with chloroquine (CQ) or rapamycin (RAPA) (C) Transwell assay of LAMP3 knockdown strain following treatment with chloroquine (CQ) or rapamycin (RAPA) These results suggest that LAMP3 knockdown influences cell proliferation, migration, and invasion by modulating lysosomal function and autophagy processes. Drug prediction In this study, potential therapeutic interventions were predicted using the DSigDB database. By integrating the screening results with a comprehensive literature review, we identified five drugs that may have therapeutic potential against cervical cancer: IBMX, daidzein, LUCANTHONE, thapsigargin, and 8-Bromo-cAMP, Na (Table [122]3). These drugs predominantly exhibit anti-inflammatory, antioxidant, and hormone-modulating properties, which align with the pathogenesis of cervical cancer. This suggests that they could play a significant role in the treatment of cervical cancer. Table 3. Candidate drug predicted using DSigDB Term P-value Adjusted P-value Odds Ratio Combined Score Genes IBMX CTD 00007018 0.00439994 0.03153289 19,912 108045.793 LAMP3 daidzein CTD 00000166 0.00634992 0.0371946 19,873 100543.721 LAMP3 LUCANTHONE CTD 00006227 0.0106499 0.04163141 19,787 89876.6148 LAMP3 thapsigargin MCF7 UP 0.02034985 0.04633217 19,593 76308.4966 LAMP3 8-Bromo-cAMP, Na CTD 00007044 0.03259982 0.06206922 19,348 66236.8831 LAMP3 [123]Open in a new tab Molecular docking To systematically evaluate the affinity of drug candidates for their targets and assess the druggability of these targets, molecular docking was employed in this study. Using the advanced CB-Dock2 platform, we precisely characterized the binding sites and interactions of the five drug candidates with LAMP3. For each molecular docking analysis, we calculated the corresponding binding energies, and the results demonstrated that all drug candidates successfully docked with LAMP3 (Fig. [124]10A-F). Among these, the binding of daidzein to LAMP3 was the most stable, with a binding energy of −10.7 kcal/mol, indicating a strong interaction between the drug and the target. This finding provides a critical foundation for subsequent drug development and mechanism studies. Fig. 10. [125]Fig. 10 [126]Open in a new tab Molecular Docking Results (A) Binding of IBMX to LAMP3 (B) Binding of daidzein to LAMP3 (C) Binding of LUCANTHONE to LAMP3 (D) Binding of thapsigargin to LAMP3 (E) Binding of 8-bromo-cAMP, Na+, to LAMP3 (F) Binding energies of the five compounds to their targets Discussion Lysosomes are crucial for intracellular material degradation and maintaining cellular homeostasis. Aberrant lysosomal function is strongly linked to the development of various diseases, particularly cancer. In recent years, there has been increasing interest in the role of lysosome-related genes in tumorigenesis, progression, and therapy. In this study, we developed a prognostic prediction model based on five key genes (AP1B1, DNASE2, LAMP3, NPC1, and LAPTM4A) by constructing a lysosome-associated gene labeling system. This model highlighted the critical role of LAMP3 in cervical cancer and its potential pharmacological value. Our findings not only offer a new tool for precise prognosis in cervical cancer but also provide a foundation for the development of novel therapeutic strategies. The mechanisms underlying the role of lysosome-related genes in tumors are complex and multifaceted [[127]19]. Studies have shown that lysosomal dysfunction can disrupt cellular autophagy, which, in turn, affects tumor cell proliferation, migration, and invasion [[128]20–[129]22]. In research on lung adenocarcinoma, the development of a lysosome-related prognostic signature (LRPS) revealed that mutations and copy number variations in lysosome-related genes are closely associated with the tumor immune microenvironment and can serve as significant prognostic indicators [[130]23, [131]24]. Similarly, analysis of the TCGA and GEO [132]GSE63514 datasets in this study found that lysosome-related genes, including LAMP3, were significantly differentially expressed in cervical cancer and strongly correlated with tumor malignancy progression. LAMP3 (lysosome-associated membrane protein 3) has garnered significant attention in cancer research in recent years [[133]25]. As a key component of the lysosomal membrane, it is involved not only in essential cellular functions such as material transport and degradation but also in tumorigenesis, tumor progression, and therapeutic responses [[134]26, [135]27]. Under normal conditions, the expression of LAMP3 is tightly regulated. However, in various cancers—including lung, esophageal, colorectal, fallopian tube, ovarian, breast, and liver cancers—the expression of LAMP3 is frequently dysregulated. This aberrant expression is often associated with the malignancy grade, tumor proliferation, invasiveness, and treatment sensitivity [[136]28–[137]30]. The abnormal expression of LAMP3 is linked to tumor cell proliferation [[138]31]. Elevated levels of LAMP3 in certain tumors have been correlated with rapid cell proliferation and uncontrolled tumor growth [[139]3, [140]32]. This may be due to LAMP3’s involvement in signaling pathways that promote cell proliferation or support tumor growth by modulating cell metabolism and energy utilization [[141]33]. Furthermore, LAMP3 expression is closely associated with tumor invasiveness and metastatic potential. In particular, studies have shown that in cancers such as breast and cervical cancer, the hypoxic tumor microenvironment can upregulate LAMP3 expression, thereby enhancing cancer cell migration and invasion. This increased migratory and invasive capacity may occur through alterations in cytoskeletal dynamics, extracellular matrix degradation, and intercellular interactions [[142]34–[143]36]. Additionally, LAMP3 expression can influence tumor sensitivity to treatments. For instance, in breast cancer cells, activation of the PERK-ATF4-LAMP3 pathway has been shown to enhance DNA repair mechanisms, contributing to resistance to therapies such as radiation and chemotherapy [[144]37–[145]39]. These findings suggest that inhibiting LAMP3 expression or function could increase tumor sensitivity to treatment and improve patient prognosis. Given these insights, LAMP3 holds promise as a potential therapeutic target. Developing specific antibodies, drugs, or other therapeutic interventions against LAMP3 could inhibit tumor cell proliferation, migration, and invasiveness, improve treatment responsiveness, and potentially lead to tumor eradication [[146]40, [147]41]. However, research into the relationship between LAMP3 and tumors is still in its early stages, and further experimental and clinical evidence is required to fully support these therapeutic possibilities. In conclusion, a strong correlation exists between LAMP3 and tumors, with its aberrant expression potentially influencing various aspects of tumorigenesis, tumor progression, and treatment response. As research in this area continues to evolve, LAMP3 holds promise as a novel target for tumor therapy, offering new strategies and insights for cancer treatment [[148]25, [149]42]. The relationship between autophagy and tumor progression has been a focal point in cancer research. Autophagy can function both as an oncogenic mechanism and as a promoter of tumor cell survival and drug resistance during tumorigenesis [[150]33, [151]37, [152]41]. In this study, we observed that LAMP3 knockdown not only suppressed the proliferation and invasion of cervical cancer cells but also induced significant alterations in autophagy-related proteins (e.g., P62 and LC3II/I), underscoring the critical roles of lysosomal function and the autophagy pathway in tumor progression [[153]32, [154]34, [155]35]. Furthermore, autophagy is intricately linked to epithelial-mesenchymal transition (EMT), a key process that enables tumor cells to acquire migratory and invasive abilities. By modulating autophagy, new therapeutic targets for inhibiting tumor metastasis may be identified [[156]36, [157]43, [158]44]. In the realm of drug prediction, our study identified Daidzein (soybeanin) as a potential inhibitor of LAMP3, with molecular docking analysis revealing a very low binding energy between Daidzein and LAMP3. This suggests that Daidzein may exert anti-tumor effects by disrupting LAMP3 function, providing crucial target information for the development of novel cervical cancer therapeutics. Daidzein is an isoflavonoid derived from leguminous plants, known for its diverse pharmacological activities, including antioxidant, anti-inflammatory, and antidiabetic effects [[159]45–[160]47]. Regarding anticancer properties, Daidzein has demonstrated significant inhibitory effects on various cancer cell types, including those of prostate, breast, and liver cancers [[161]48, [162]49]. A recent study evaluating the in vitro anticancer activity of Daidzein against the human cervical cancer HeLa cell line showed that Daidzein significantly reduced HeLa cell viability, with an inhibitory concentration (IC50) of 20 µM [[163]50, [164]51]. The anticancer mechanisms of Daidzein are multifaceted. First, it induces apoptosis in HeLa cells by increasing reactive oxygen species (ROS) levels and altering mitochondrial membrane permeability [[165]52]. Second, Daidzein significantly diminishes HeLa cell adhesion, thus inhibiting proliferation and metastasis [[166]53]. Additionally, Daidzein modulates apoptosis-related proteins by enhancing the activity of Caspases 8 and 9, key regulators of apoptosis. Finally, Daidzein reduces the expression of pro-inflammatory genes (e.g., TNF-α, NFκB, IL-6, COX-2) and cell proliferation-related genes (e.g., JAK2, STAT3, ERK), thereby suppressing inflammatory responses and cell proliferation signaling pathways [[167]9]. These findings provide strong scientific support for Daidzein’s potential as a therapeutic agent for cervical cancer and lay a solid foundation for future drug development and clinical application [[168]54, [169]55]. Despite the significant progress made in this study, several limitations remain. First, the limited sample size, particularly in the experimental validation phase, may affect the generalizability of the findings. Future studies should validate the predictive power and biological significance of the model in larger cohorts. Second, while molecular docking analysis predicted Daidzein’s potential mechanism of action, its pharmacological effects in vivo require further verification through animal experiments and clinical trials. Moreover, lysosome-related genes operate through complex mechanisms within the tumor microenvironment, and future research could explore their interactions with immune cells and tumor cell metabolism. The integration of single-cell sequencing technology and multi-omics analysis may offer a more comprehensive understanding of lysosome-related genes in cancer and provide additional therapeutic targets for developing novel cancer treatments. In summary, this study highlights the pivotal role of LAMP3 in cervical cancer and its potential pharmacological value. By constructing a labeling system for lysosome-associated genes, this research not only offers a new tool for precise prognosis but also lays the groundwork for the development of novel therapeutic strategies targeting lysosomal functions. Conclusion Lysosome-related gene modeling has significant clinical value. LAMP3 knockdown inhibits cervical cancer progression by reducing autophagy and lysosomal function. Daidzein shows potential as a novel therapeutic agent. However, further validation in larger cohorts is needed due to the limited sample size in this study. Supplementary Information [170]Supplementary Material 1.^ (35MB, rar) [171]Supplementary Material 2.^ (847.8KB, zip) [172]Supplementary Material 3.^ (11.2MB, docx) Acknowledgements