LAMP3 is a potent uterine corpus endometrial carcinoma prognostic biomarker associated with immune behavior

Patients and tumor specimens

We collected UCEC tissues and matched non-tumor tissues from a cohort of 20 patients at the Second Affiliated Hospital of Nanchang University between January 2020 and January 2022. These samples were promptly frozen in liquid nitrogen and stored at -80° C. The study was carried out with the informed consent of the patients and received approval from the Research Ethics Committee of the Second Affiliated Hospital of Nanchang University.

Data source and processing

Our data were mainly collected from The Cancer Genome Atlas (TCGA) database (https://cancergenome.nih.gov). From this, we gained RNAseq data in FPKM format for the STAR workflow of UCEC tumor and used it to comprehensively characterize the expression of LAMP3 in UCEC, the prognosis of UCEC patients with high LAMP3 expression, and the relationship of LAMP3 expression with several clinic features, diverse signaling pathways, immune infiltration, immune checkpoints and RNA-modified genes.

Weighted gene co-expression network analysis (WGCNA)

WGCNA is a bioinformatics analytical approach for seeking modules of highly relevant genes [17]. In this research, we processed all differentially expressed genes in TCGA-UCEC by R (4.1.2) package “WGCNA”, applying a soft threshold function to compute power parameter, generating a cluster dendrogram with hierarchical clustering, structuring a module-trait relationship network by assigning normal and tumor samples to 0, 1 and showing the association between gene significance and module membership in the yellow module with the scatter plot.

GEPIA website

GEPIA (http://gepia.cancer-pku.cn/) is a platform serving us with RNA sequencing expression data in TCGA and GTEx databases [18]. On the basis of TCGA data, we studied the differential expression of LAMP3 in UCEC and normal tissues utilizing the “Boxplot” section in the “Expression DIY” module.

TIMER database

TIMER (https://cistrome.shinyapps.io/timer/) gives us a user-friendly network interface for exploring tumor-immune interactions [19]. By this way, we characterized the expression of LAMP3 in all TCGA cancers.

CCLE database

Cancer Cell Line Encyclopedia database (https://portals.broadinstitute.org/ccle/about) produces a massive genomic dataset with 947 human cancer cell lines [20]. We acquired cell line gene expression data from it for the purpose of assessing LAMP3 expression level in diverse cancers.

Cell culture

Transfection of the human UCEC cell line TCHu198 in 37° C DMEM (HyClone, Germany) and 10% female bovine serum (Gibco, USA).

Kaplan–Meier plotter database

The Kaplan-Meier plotter database (http://kmplot.com) is a survival analysis tool which allows the assessment for disease progression [21]. Through “mRNA RNA-seq” module, we examined the prognostic value of LAMP3 in UCEC, the differential survivorship of UCEC patients with a variety of immune cell subtypes and the survival curves of 4 RNA modification-related genes in UCEC.

LinkedOmics database analysis

LinkedOmics (http://linkedomics.org/login.php) presents us with a unique podium covering multi-omics data for 32 types of cancers [22]. With RNAseq data supplied by the HiseqRNA platform in the TCGA-UCEC cohort, we investigated the co-expressed genes of LAMP3 and their enrichment pathways.

Gene set enrichment analysis (GSEA) analysis

GSEA is a powerful profiling approach which uncovers the common biological pathways in a group of genes by focusing on the gene set [23]. In our study, we refer to the “c2, cp.kegg.v2022.1.Hs, symbols.gmt” gene set and include the three requirements of FDR < 0.25, P-value < 0.05 and false discovery rate (FDR) > 1 in the filtration criteria to perform an enrichment assessment of the LAMP3 high expression group.

Protein-protein interaction (PPI) network analysis

The STRING database (https://string-db.org) allows for exploration of genomic links among genes encoding proteins, thus we can infer the biological functions of proteins [24]. Our study utilized the STRING database to research the first 500 co-expressed genes of LAMP3 and set a medium confidence level = 0.9 to select genes, and then we created an overall PPI network by visualizing them with Cytoscape software. What’s more, the most essential module of our PPI network was constructed by Cytoscape Molecular Complex Detection (MCODE) plug-in. The relevant parameters are set as follows: Max depth=100, node score cutoff=0.2, K-core=2.

TISIDB analysis

TISIDB database (http://cis.hku.hk/TISIDB/) could be employed to inquire the relationship of specific genes with immunity in the context of tumorigenesis [25]. Through the “Overall survival” section in the “Clinical” module, we probed the prognostic value of LAMP3 in UCEC; through the “Immunomodulator” and “Chemokine” modules, we sought the relationship between LAMP3 expression and cytokines in UCEC, and by “Subtype” module, we looked into the distribution of LAMP3 expression in different immune subtypes in UCEC.

Comparative toxicogenomics database

The CTD (http://ctdbase.org/) integrates the interactions of specific genes with correlative compounds and contains the literature coverage of some compounds [26]. From it, we obtained several drugs regulating LAMP3 expression.

Protein structure and docking analysis

The cBioPortal database (https://www.cbioportal.org/) incorporates multidimensional cancer genomics data [27]. Through this database, we built the secondary structures of some proteins (LAMP3, ABCA3, SGTB and RAB9A) with Uterine Corpus Endometrial Carcinoma (TCGA, Firehose Legacy). The PDB database (https://www.rcsb.org/) stores experimentally determined 3D structures of biomolecules and their complexes [28]. It was adopted for downloading the tertiary spatial conformation of 4 proteins. Moreover, the HDOCK server (http://hdock.phys.hust.edu.cn/) and PyMOL software were respectively applied to predict and visualize the docking models of LAMP3-ABCA3, LAMP3-SGTB and LAMP3-RAB9A.

GeneMANIA analysis

GeneMANIA (https://genemania.org/) database delivers genomics and proteomics data for seeking genes functionally similar to a given gene. By it, we constructed a LAMP3-centered interaction network.

ceRNA analysis

By TargetScan (http://www.targetscan.org), DIANA (http://diana.imis.athena-innovation.gr/DianaTools/index) and RNAinter (http://www.rnainter.org) online sites, we jointly predicted miRNAs targeting LAMP3 and screened for miRNAs negatively correlated with LAMP3 in the TCGA-UCEC cohort.

Besides, by miRNet2.0 (https://www.mirnet.ca/miRNet/home.xhtml) and starBase3.0 (https://rnasysu.com/encori/) online sites, we searched for targeted lncRNAs of has-miR-106a-5p and has-miR-499a-5p, and again utilized the starBase3.0 tool to select lncRNAs that met the negative correlation condition with the corresponding miRNAs. And then, we also exploited Sankey diagrams to build LAMP3-related ceRNA networks.

Quantitative real-time PCR (qRT-PCR)

We followed the Trizolrogen (USA) method to extract total RNA and then used an RT-PCR kit to convert the extracted RNA into complementary DNA (cDNA). To assess the mRNA levels of the target genes, a quantitative real-time PCR kit was used.

Western blot

We utilized radioimmunoprecipitation assay buffer (RIPA) with the addition of protease inhibitor cocktail (Roche, Switzerland) to extract total proteins and quantified using the BCA method. Protein was separated on SDS-PAGE gels and transferred to PVDF membranes (EMD Millipore, Billerica, MA, USA). After overnight incubation at 4° C with the designated antibodies, the membrane was subjected to three washes with TBST. The membrane was then exposed to secondary antibodies at room temperature for one hour. Detection of protein expression was conducted using the Amersham™ ImageQuant 800 system (GE Healthcare, USA).

Transwell assay

Invasion assays were performed using 24-well transwell chambers with a Matrigel precoated upper chamber. Each group cell suspension was added to a Matrigel precoated chamber, with the lower chamber containing a higher serum concentration than the upper chamber. The samples were then incubated for 24 hours in a cell culture incubator. Following incubation, the cells were fixed with methanol and stained using hematoxylin-eosin.

Statistical analysis

Our research utilized R for all data analysis (3.6.3/4.0.3/4.1.2). With the R package “ggplot2”, we predicted the expression of LAMP3 in normal and UCEC tissues applying the Wilcoxon rank sum test, and the association between LAMP3 and lots of clinical features was estimated by the Kruskal-Wallis test. Furthermore, this package was also been employed to infer the linkage of RNA-modified genes with LAMP3 expression and the correlation between some molecules. Alternatively, based on the “survminer” package, the “survival” package and the “rms” package, we evaluated the effect of LAMP3 on the survivorship of UCEC patients and the accuracy of the column line plot, while the ROC curve was resolved and visualized with the “timeROC” package and the “ggplot2” package. Besides, by “GSVA” package, we also selected the parameter method=‘ssgsea’ to calculate the correlation between LAMP3 and pathway score; and the “ssGSEA” method was applied for finding out the relationship between LAMP3 and immune cell infiltration. In parallel, the “estimate” package was chosen for the characterization of the association of LAMP3 expression with 3 types of scores as well as immune checkpoint genes.

Data availability statement

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Construction of a weighted co-expression network in UCEC

For identifying the key gene module of UCEC, we performed a WGCNA based on the values of gene expression. We first examined 26535 genes in the TCGA-UCEC dataset and detected 9037 genes from it that exhibited differential expression (Figure 1A). Subsequently, according to the scale independence and average connectivity, we set the soft threshold to 6 (R2= 0.8) (Figure 1B). And 15 modules were identified with dynamic number cutting package by setting 30 as the minimum number of genes per genetic network and 0.25 as the clustering height limit (Figure 1C). Moreover, we evaluated the correlation between modules and UCEC by investigating the relevance of ME values to UCEC samples, which revealed that yellow module was notably associated with UCEC (r=0.42, p=5e-27) (Figure 1D). Furthermore, the MM vs GS analysis also demonstrated that genes in the yellow module were highly associated with UCEC (Figure 1E).

Figure 1. WGCNA based on TCGA-UCEC. (A) All differentially expressed genes in TCGA-UCEC. (B) Determination of soft threshold. (C) Tree diagram of 15 modules labeled with different colors. (D) Correlation between module feature genes and clinical traits. (E) Scatter plot of MM vs GS in the yellow module.

In the next step, by performing survival, univariate and 1.3.5-year ROC analyses on 371 genes in the yellow module successively, we filtered out 73 genes in which survival, univariate values were significant (p<0.05) and 1,3,5-year ROC values were greater than 0.5. Then, to further tighten the constraints, we ranked these 73 genes depending on the HR values in the univariate analysis and picked out the top 30 to review the literature, discovering that only 6 of them had not yet been reported in UCEC, which were KLRG2, C5orf34, CCDC150, TDRKH, LAMP3, KLC3 (Table 1). To sum up, it is suggested that LAMP3 may be one of the multiple factors that influence the development of UCEC.

mRNA expression of LAMP3 in UCEC

For the sake of determining LAMP3 expression in UCEC, we investigated the transcript level of LAMP3 through a series of databases. With the CCLE database, we explored the expression distribution of LAMP3 in 32 tumor tissues, and the outcome implied that it possessed high expression in various tumor tissues such as SCLC, BLCA, ESCA, and UCEC (Supplementary Figure 1A). Then, we further resolved LAMP3 expression in tumor and normal tissues utilizing RNA-seq data from the TIMER website, identifying that LAMP3 exhibited higher expression in 11 types of tumor tissues, including UCEC (Supplementary Figure 1B). Subsequently, we probed the level of LAMP3 expression in single tumor UCEC using the collected UCEC tissues and normal tissues from the TCGA database, which suggested an upregulation of LAMP3 expression in UCEC tissues (Figure 2A, 2B). Also, the above result was validated on the basis of the GEPIA database, as illustrated in Figure 2C, it remained in line with our expectation. In addition, we discovered that the expression of LAMP3 was upregulated through qRT-PCR (Figure 2D). And to further elucidate the impact of LAMP3 expression on the metastasis of UCEC cells, we conducted transwell assays. Impressively, our findings revealed that the suppression of LAMP3 significantly impeded the migratory and invasive capabilities of TCHu198 cells (Figure 2E, 2F). In summary, we conclude that LAMP3 is overexpressed in UCEC tissues when compared to normal tissues.

Figure 2. Overexpression of LAMP3 in UCEC tissues. Expression of LAMP3 in UCEC tissues and normal tissues presented by TCGA data in the form of (A) scatter plot and (B) paired difference plot. (C) mRNA transcript level of LAMP3 in UCEC gained by GEPIA database. (D) Quantification of LAMP3 mRNA expression based on qRT-PCR. (E, F) Migration and invasion ability of TCHu198 cells in the shNC and shLAMP3 groups.

Association of LAMP3 expression with multiple clinicopathological features in UCEC

With the purpose of further exploring the association of LAMP3 with some clinical characteristics in UCEC, we compared the level of LAMP3 expression in UCEC patients with distinct clinicopathological characteristics based on the TCGA-UCEC data. According to the age group, it could be noted that LAMP3 expression was upregulated in UCEC patients older than 60 years (Supplementary Figure 2A). Concerning the grade and stage group, we observed that Grade 3 and Stage IV UCEC patients possessed higher expression compared to normal people and early grades or early stages UCEC patients (Supplementary Figure 2B, 2C). In terms of histological type, the expression of LAMP3 in Endometrioid, mixed, and serous UCEC patients were all upregulated compared to the normal group, and the serous UCEC patients displayed higher expression in contrast to Endometrioid UCEC patients (Supplementary Figure 2D). Moreover, on the survival status, the dead patients had apparently higher LAMP3 expression (Supplementary Figure 2E). Also, it was detected that LAMP3 expression was correlated with menopausal status with varying level of LAMP3 expression at different periods, while all periods were higher than normal (Supplementary Figure 2F). In conclusion, it is indicated that LAMP3 expression is strongly related to various clinicopathological features.

LAMP3 may be a potential prognostic marker for UCEC

Aiming to measure the prognostic potential of LAMP3 in UCEC, we probed the influence of LAMP3 expression on the survivorship of UCEC sufferers with multiple methods. Initially, we adopted Kaplan-Meier plotter database to plot survival curves for determining the correlation of LAMP3 expression with the survival of UCEC sufferers. The outcome Figures 3A, 3B demonstrated that patients with high expression of LAMP3 in UCEC possessed worse OS and RFS. Later on, in a further survival analysis of UCEC patients, we separated LAMP3 expression into high and low expression groups through TCGA data, and it was concluded that high LAMP3 expression group owned worse OS, PFI and DSS in comparison with the low LAMP3 expression group (Figure 3C–3E). A time-dependent ROC curve demonstrated the favorable predictive capability of LAMP3 for survival of UCEC patients (Figure 3F). Alternatively, the column line graph was constructed based on various clinicopathological features, which evidenced that Histologic grade and Clinical stage contributed the most to the prognosis of UCEC patients, and LAMP3 high expression was indeed remarkably related to a low survival probability of UCEC sufferers (Figure 3G). The corresponding calibration plot exhibited excellent agreement with the column plot for the 1, 3, 5 years forecasted (Figure 3H). In summary, high LAMP3 expression is considered as a predictor of adverse outcome in UCEC patients.

Figure 3. Prognostic value of LAMP3 in UCEC patients. (A, B) Demonstration of the relationship between LAMP3 expression and survival of UCEC patients on the basis of Kaplan-Meier plotter database. (C) Overall survival, (D) Progress free interval, (E) Disease specific survival of LAMP3 in UCEC with TCGA data. (F) A time-independent ROC curve demonstrates the predictive power of LAMP3. (G) A column line graph presents the impact of clinical features and LAMP3 expression on the prognosis of UCEC patients. (H) Calibration curve of the column line graph.

Co-expression patterns of LAMP3 in UCEC

Trying to gain an understanding of the biological meaning of LAMP3 in UCEC, we dug deeper into its co-expression pattern in UCEC. With the “LinkFinder” module in Linkedomics, we had detected 19898 possibly co-expressed genes with LAMP3, of which 4408 genes (red dots) were positively related and 2559 genes (green dots) were reversed (p < 0.05) (Figure 4A). Figure 4B, 4C presented the top 50 genes showing the most positive and negative association with LAMP3, respectively. Besides, via the “Linkinterpreter” module, we carried out biological process and molecular functional analysis of these genes. Go analysis (biological process) illustrated that LAMP3 co-expressed genes were primarily involved in signaling pathways in response to various interferons, defense responses to other organisms and regulation of innate immune response, while inhibitory pathways such as translational initiation (Figure 4D). KEGG pathway indicated that these genes were predominantly abundant in herpes simplex infection, NOD-like receptor signaling pathway, antigen processing and presentation, Toll-like receptor signaling pathway, chemokine signaling pathway and other pathways (Figure 4E). Additionally, with the aid of validation by RNAseq data of UCEC extracted from the TCGA database, we likewise observed that LAMP3 was extensively involved in some pathways such as G2M checkpoint and Tumor proliferation signature (Figure 4F, 4G).

Figure 4. Construction of co-expression network of LAMP3. (A) A spearman test is used to examine all co-expressed genes of LAMP3. The heatmap shows the top 50 genes that are (B) positively and (C) negatively associated with LAMP3. (D) Go analysis (biological process) and (E) KEGG analysis reveal the signaling pathways of LAMP3 and its co-expressed genes in UCEC by LinkedOmics. (F, G) Association between LAMP3 and each pathway by TCGA data.

Apart from that, in some further GSEA analysis, we classified all samples into high and low expression groups based on the expression level of LAMP3, where we noticed that LAMP3 high expression group was mainly enriched in typical immune-related pathways such as antigen processing and presentation, B-cell receptor signaling pathway. Together, these results propose that LAMP3 may exert a broad impact on the UCEC transcriptome through immune and pathogen-associated pathways (Supplementary Figure 3A–3K).

LAMP3 relevant PPI network in UCEC

For further inferring the potential biological functions of LAMP3, we investigated the top500 co-expressed genes of LAMP3 by STRING database and established their PPI network (Figure 5A). Rating based on degree of connectivity, we also constructed the module in which LAMP3 was the most relevant and had the highest average connectivity score (21.619) by use of Cellular Landscape (MCODE plugin). As result shown in Figure 5B, among them, a total of 22 genes were included and highlighted in yellow. It was worth noting that by searching the relevant literature for these genes, we identified that all 22 genes were strongly associated with immunity. Later, we further selected the two highest scoring genes (MX2 and STAT1) for a follow-up study. As Figure 5C, 5D indicated in the scatter plots, it could be noted that MX2 and STAT1 did have a strong correlation with LAMP3; as displayed in survival curves Figure 5E, 5F, we could discover that UCEC patients with high MX2 or STAT1 expression had a poorer prognosis. Furthermore, we conducted additional experimental validation to establish the correlation between LAMP3 and the aforementioned genes. We employed western blot and qRT-PCR to investigate the expression level of MX2 and STAT1 in shNC and shLAMP3 groups, respectively. Notably, our results demonstrated a significant reduction in the expression of both genes, observed at both the protein and mRNA levels, following the inhibition of LAMP3 (Figure 5G, 5H). Therefore, we speculate that the influence of LAMP3 on the survivorship of UCEC sufferers may correlate with immunity.

Figure 5. PPI network of LAMP3 co-expressed genes. (A) Protein-protein interaction (PPI) network of the first 500 co-expressed genes of LAMP3. (B) The most important module of protein-protein interaction network. (C, D) Correlation of LAMP3 with MX2 and STAT1. (E, F) Effect of MX2 and STAT1 on the prognosis of UCEC patients. (G) Expression of 3 genes in UCEC cells detected by western blot. (H) Expression of MX2 and STAT1 in shNC and shLAMP3 groups detected by qRT-PCR.

Association between LAMP3 expression and immune infiltration in UCEC

In order to fully address the immune-related features of LAMP3, its association with immune infiltration was examined. The first step was to separate the tumor samples into two groups by the high and low expression of LAMP3, and then we calculated the StromalScore, ImmuneScore and ESTIMATEScore for each tumor sample, finding that all three scores were better in the LAMP3 high expression group compared to the LAMP3 low expression group (Figure 6A–6C). Next, with the RNAseq data from TCGA, we studied the correlation of LAMP3 expression with 24 immune cells in UCEC, as depicted in Figure 6D, 19 of these immune cells were affected by LAMP3 expression, of which it was strongly linked to the infiltration of aDC cells (r=0.729, p<0.01). Moreover, we measured the enrichment scores of these 19 immune cells in UCEC within the high and low expression groups of LAMP3, and we discovered that there were remarkable differences between the groups of 18 types of immune cells (Figure 6E).

Figure 6. Association of LAMP3 with immune cell infiltration in UCEC. (A–C) Box plots display the StromalScore, ImmuneScore and ESTIMATEScore of LAMP3 in UCEC based on the “Estimate” algorithm. (D) Lollipop plot exhibits the correlation between LAMP3 expression and 24 immune cells. (E) Grouped comparison plot presents the enrichment score of 19 immune cells between the two groups with high and low expression of LAMP3.

Next, with an aim to broaden the insight of the relevance of LAMP3 with the immune infiltration, we also evaluated the association between LAMP3 expression and a variety of cytokines as well as the expression of LAMP3 in distinct immune subgroups of UCEC patients through the TISIDB database. Supplementary Figure 4A–4C displayed a positive relation of LAMP3 expression with most MHCs, chemokines and receptors in UCEC, where the 8 chemokines most correlated with LAMP3 were plotted in the scatter plots (Supplementary Figure 4D–4K). As for Supplementary Figure 4L, it revealed that LAMP3 expression was highest in the C2 subgroup (IFN-gamma dominant) and lowest in the C1 subgroup (wound healing) among the five immune subgroups. The above-mentioned outcomes imply that LAMP3 has a close connection with immune cell infiltration in UCEC.

Linkage between LAMP3 and immune checkpoints in UCEC

Previous researches have shown that immune checkpoints have a vital function in the human immune system. Therefore, by extracting data downloaded from TCGA, we inquired into the relation between LAMP3 expression and immune checkpoints, and the result indicated that there exists a relevance between LAMP3 expression and the expression of most checkpoint genes in UCEC (Supplementary Figure 5A). Besides, the immunostimulant CD274, IL12A and immunosuppressants C10orf54, CXCL10, ICAM1, BTN3A1 were the six checkpoint genes with the highest correlation (Supplementary Figure 5B–5G). With these results, it is proposed that LAMP3 may be involved in the overexpression of checkpoint genes.

The function of LAMP3 on the prognosis of UCEC patients may be tied to immunity

Since LAMP3 expression is relevant to the immune infiltration and poor prognosis of UCEC patients respectively, we studied deeper whether LAMP3 could influence the survival of UCEC sufferers by immune infiltration. In UCEC, we carried out survival analyses on the ground of LAMP3 expression in various immune cell subsets through Kaplan-Meier plotter database (Figure 7A–7K). As the findings demonstrated, high expression of LAMP3 in the Type 1 T-helper cells enriched, Natural killer T-cells decreased, Basophils enriched, and Type 2 T-helper cells enriched cohorts resulted in poor prognosis, but in the Type 1 T-helper cells decreased, Natural killer T-cells enriched, Basophils decreased, and Type 2 T-helper cells decreased cohorts, high/low expression of LAMP3 did not affect the prognosis of UCEC patients. This result suggests that the pathway by which LAMP3 expression affects the survival of UCEC patients may have an association with the infiltration of Type 1 T-helper cells, Natural killer T-cells, Basophils and Type 2 T-helper cells.

Figure 7. Survival curves of high and low expression of LAMP3 in UCEC in diverse immune cell subgroups. (A) Type1 T-helper cells (B) Natural killer T-cells (C) Basophils (D) Type2 T-helper cells (E) Mesenchymal stem cells (F) CD4+ memory T-cells (G) Regulatory T-cells (H) B-cells (I) CD8+ T-cells (J) Eosinophils (K) Macrophages.

LAMP3 may be tied with RNA modifications in UCEC

An increasing number of reports confirm that RNA modifications have an essential modulatory role in tumor pathogenesis, which prompts us to study their link with LAMP3 in UCEC. To start with, by means of data from TCGA, we examined the connection of LAMP3 with 39 m6A, m5C, and m1A-related genes in UCEC, and the results displayed that 33 of these genes presented a very significant positive correlation with LAMP3 at the expression level (Figure 8A). In addition, grouping LAMP3 according to its expression level, we attempted to take an analysis of the differential expression level of these genes between the both groups with high and low LAMP3 expression in UCEC. As presented in Figure 8B, 29 of these genes conformed to our expectation, exhibiting differential expression. Subsequently, through upset plot, we selected 4 genes with correlation greater than 0.43 and consistent with differential expression, which were NSUN2, NSUN4, FMR1 and TRMT6 (Figure 8C). Scatter plots Figure 8D showed the molecular correlation specifically between LAMP3 and them. Furthermore, it was noteworthy that the Kaplan-Meier curves suggested that NSUN2, FMR1 may contributed to the poor prognosis of UCEC patients (Figure 8E). Altogether, these imply us that the way LAMP3 influences UCEC progression may have a connection with RNA modification.

Figure 8. Linkage of LAMP3 with RNA-modified genes in UCEC. (A) Correlation of LAMP3 with various RNA-modified genes in UCEC. (B) Expression level of each of 39 genes between the high and low LAMP3 expression groups in UCEC. (C) The upset plot presents the genes that are individually screened under different conditions. (D) Association of NSUN2, NSUN4, FMR1, and TRMT6 with LAMP3 expression. (E) Effect of NSUN2, NSUN4, FMR1, and TRMT6 on the survival of UCEC patients.

Gene - compound interaction network analysis

To predict some suitable compounds for the UCEC patients with high expression of LAMP3, we constructed a gene-compound interaction network. Through CTD, we identified 85 chemicals that can affect LAMP3 expression in UCEC at the mRNA level. Among these chemicals, we chose 14 drugs frequently treated for cancers to create an advanced network expression diagram, as illustrated in Figure 9A, where the red line represented a drug upregulating LAMP3 expression and the gray line doing the opposite. Subsequently, we intersected these drugs with cgp2014 and discovered 4 clocks of drugs, which were sunitinib, bicalutamide, cisplatin and doxorubicin (Figure 9B). Their efficacy was then evaluated by IC50 values, and it was found that patients with high LAMP3 expression were susceptible to cisplatin, doxorubicin, and resistant to bicalutamide (Figure 9C–9F). In short, these drugs have the potential to guide the treatment of UCEC patients with LAMP3 high expression.

Figure 9. Prediction of LAMP3 expression-related drugs. (A) An advanced network diagram shows 14 cancer-related drugs that can modulate LAMP3 expression. (B) An upset diagram demonstrates drugs related to LAMP3 expression in CTD, cancer drugs and cgp2014. Relationship between LAMP3 expression and IC50 of (C) sunitinib (D) bicalutamide (E) cisplatin and (F) doxorubicin.

Interaction network and molecular docking model of LAMP3

As the most critical execution molecule, proteins often complete instructions by interacting with other proteins, hence, we deeply explored the interactions of LAMP3 with other proteins. At first, we built an interaction network of LAMP3 with 20 other proteins through the GeneMANIA online website, and it was surprising to note that LAMP3 interacted directly with three of them, namely ABCA3, SGTB and RAB9A (Supplementary Figure 6A). Following that, when we looked into the secondary structures of the four proteins, we noticed that ABCA3 possessed four structural domains and SGTB had two domains, while LAMP3 and RAB9A both contained only one structural domain, and these four proteins all had several modified sites (Supplementary Figure 6B–6E). Afterwards, with the outcomes being described in Figure 10A–10C, we examined the spatial structure of the four proteins from the PDB database, and predicted the molecular docking models of LAMP3 with ABCA3, SGTB and RAB9A by HDOCK server. In addition, we had also built hydrogen bonding networks for each of the three docking models, and the results revealed that there were 6 hydrogen bonds in the LAMP3-ABCA3 docking model,12 hydrogen bonds in the LAMP3-SGTB docking model, and only 26 hydrogen bonds in the LAMP3-RAB9A model (Figure 10D–10F).

Figure 10. Binding mode of LAMP3. Cartoon view and surface view of (A) LAMP3-ABCA3 (B) LAMP3-SGTB (C) LAMP3-RAB9A molecular docking model, where LAMP3 is shown in pink. (D–F) Key regions that bind LAMP3 to ABCA3, SGTB and RAB9A, respectively.

Construction of a ceRNA network involving LAMP3 in UCEC

It has been demonstrated that ceRNA networks function critically in various human cancers, thus we attempted to predict and build the ceRNA network of LAMP3 exploiting many biological information databases. First, we independently resolved 585, 136 and 16 target miRNAs of LAMP3 through TargetScan, DIANA and RNAinter databases individually, and the predicted findings were visualized by venn diagram. As presented in Supplementary Figure 7A, we could discover that 5 miRNAs were jointly predicted by these three tools. Moreover, we also probed the relevance of these miRNAs and LAMP3 at the expression level, with the discovery that 2 of them (has-miR-106a-5p and has-miR-499a-5p) were negatively regulated by LAMP3 (Supplementary Figure 7B).

Later, we applied miRNet and starBase databases to make predictions of lncRNAs potentially binding to has-miR-106a-5p and has-miR-499a-5p. As Supplementary Figure 7C displayed, the miRNet and starBase databases respectively predicted 72 and 79 lncRNAs targeting has-miR-106a-5p, with 32 lncRNAs situated in the overlapping part of the Venn diagram. However, there were only EBLN3P, H19, HAGLR, LINC02035, NORAD, SNHG14 consistent with the ceRNA network hypothesis in which lncRNAs were usually negatively correlated with miRNAs. As Supplementary Figure 7D indicated, miRNet and starBase databases predicted 27 and 25 lncRNAs targeting has-miR-499a-5p, separately, of which 14 were co-predicted, while only 1 matched the above negative correlation. With the above outcomes, we structured the Sankey diagram to illustrate the relationship between the predicted ceRNA networks (Supplementary Figure 7E).