Background
Uveal melanoma (UVM) is the most common primary intraocular malignancy in adults (85%) and is highly malignant, invasive, and extremely metastatic with a poor prognosis [
1,
2]. In current practice, the first-line treatment of primary uveal melanoma is local ocular treatment, including plaque brachytherapy, local resection, and enucleation [
3]. Over the last 5 years, novel prognostic biomarkers have been identified, including histone deacetylase HDAC-2 [
4], DNA damage response protein ATR [
5], and genes of the NF-κB pathway [
6]. Although the treatments and prognostic biomarkers have improved over the past decade, the tumor-related mortality rate has remained unchanged and most metastasis cases are fatal within 1 year [
7,
8]. Therefore, there is an urgent need to improve our understanding of the molecular and cellular biology of UVM to aid in the development of new prognostic biomarkers and personalized treatment regimens.
Transporter associated with antigen processing 1 (TAP1) is a membrane-bound protein consisting of two membrane-spanning domains that belongs to the ATP-binding cassette (ABC) transporter superfamily [
9]. TAP1 is involved in antigen transport from the cytoplasm to the endoplasmic reticulum, binds to major histocompatibility complex (MHC) class I molecules, forms the molecular scaffold for the final stage of MHC class I folding, and then presents MHC I to immune cells, thereby serving as a target for tumor cells [
10]. Recent studies have shown that TAP1 is closely associated with various tumors and has the potential to become a novel tumor-targeting drug [
11,
12]. Although high TAP1 expression might, in theory, make tumor cells more susceptible to cytotoxic T cell killing [
13], it might reduce their susceptibility to natural killer (NK) cell–mediated lysis [
14]. Therefore, there is no consistent relationship between TAP1 expression and patient prognosis for different cancer types.
No previous study has investigated TAP1 expression in uveal melanoma. The aim of this study is to investigate the association between TAP1 expression, disease prognosis, and biological score in patients with UVM, and whether TAP1 plays a role in the UVM immune microenvironment. We performed in vitro experiments to verify the effects of TAP1 on the proliferation and migration of UVM cells. These findings may further improve the accuracy of prognostic prediction of UVM and present a potential therapeutic target for this type of cancer.
Materials and methods
Data download
The gene expression data (FPKM values) from the RNA sequencing of 80 patients with UVM were downloaded from the official website of The Cancer Genome Atlas (TCGA) Genomic Data Commons (GDC) (
https://portal.gdc.cancer.gov/) and divided into mRNA and long noncoding RNA (lncRNA) expression. The clinicopathological features and prognosis of the individual patients with UVM, such as sex, age, and stage, were downloaded from the University of California, Santa Cruz (UCSC) website (
http://xena.ucsc.edu/). The specific clinical information on these patients is presented in Table
1. Moreover, the gene expression data of GSE221381 [
15] and clinicopathological features of the patients were downloaded from the Gene Expression Omnibus (GEO) database as the validation dataset. The data originated from
Homo sapiens samples and were generated on the GPL570[HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array platform, including 63 UVM tissue samples. We also downloaded the gene expression data of GSE84976 [
16] and clinicopathological features of the patients GEO database. The data originated from
Homo sapiens samples and were generated on the GPL10558 Illumina HumanHT-12 V4.0 expression beadchip Array platform, including 28 UVM tissue samples. as a validation dataset.
Table 1
The UVM patient database from TCGA
Gender | | | | 0.822 |
Female | 35 (43.8%) | 17 (42.5%) | 18 (45.0%) | |
Male | 45 (56.2%) | 23 (57.5%) | 22 (55.0%) | |
Age | | | | 0.653 |
<60 | 36 (45.0%) | 17 (42.5%) | 19 (47.5%) | |
≥ 60 | 44 (55.0%) | 23 (57.5%) | 21 (52.5%) | |
Pathologic stage | | | | 0.025* |
I & II | 36 (45.0%) | 23 (57.5%) | 13 (32.5%) | |
III & IV | 44 (55.0%) | 17 (42.5%) | 27 (67.5%) | |
Histological type | | | | 0.05* |
Epithelioid Cell | 13 (16.3%) | 3 (7.5%) | 10 (25.0%) | |
Mixed | 37 (46.2%) | 18 (45.0%) | 19 (47.5%) | |
Spindle Cell | 30 (37.5%) | 19 (47.5%) | 11 (27.5%) | |
Type | | | | < 0.001*** |
metastatic | 19 (23.8%) | 5 (12.5%) | 14 (35.0%) | |
non-metastatic | 61 (76.3%) | 35 (87.5%) | 26 (65.0%) | |
Statue | | | | 0.026* |
Alive | 57 (71.3%) | 33 (82.5%) | 24 (60.0%) | |
Death | 23 (28.7%) | 7 (17.5%) | 16 (40.0%) | |
Differentially expressed genes (DEGs)
To analyze the
TAP1 gene expression in patients with UVM, patients’ tumor samples in TCGA and GEO databases were divided into high- and low-expression groups based on median value of
TAP1 gene expression. Differentially expressed genes (DEGs) between the two groups were analyzed using the
limma package of R [
17], where log fold change (logFC) > 0.5 and adjusted
p-value < 0.05 were set as the thresholds for statistical significance. The
VennDiagram package for R [
18] was used to select specific DEGs that were common between the two datasets. The results of the difference analysis were presented in the form of a volcano plot.
Functional enrichment analysis and gene set enrichment analysis (GSEA)
Gene ontology (GO) analysis is a standard method for conducting large-scale functional enrichment studies involving functions categorized under biological processes, molecular functions, and cellular components. The Kyoto Encyclopedia of Genes and Genomes (KEGG) is a widely used database for storing information on genomes, biological pathways, diseases, and drugs [
19]. Signature genes were analyzed for functional and pathway enrichment and annotated based on GO and KEGG databases using the
clusterProfiler R [
20] software package, and a cut-off FDR value of < 0.05 was considered statistically significant. To investigate differences in biological processes among the different groups, we performed GSEA on the gene expression profiling dataset of patients with UVM. GSEA is a computational method for analyzing whether a particular gene set is statistically different between two biological states and is usually used to estimate changes in pathway and biological process activity [
21] in samples of expression datasets. The “c2.cp.kegg.v6.2.-symbols” [
21] gene set was downloaded from the MSigDB database for GSEA analysis and an adjusted
p-value > 0.05 was considered statistically significant. Signature genes of relevant pathways were downloaded from the GeneCard database, and the ssGSEA analysis algorithm was used to calculate the enrichment scores of each patient in different pathways.
The tumor immune estimation resource (TIMER) database analysis and comparison of immune-related scores between the two groups
TIMER database (
https://cistrome.shinyapps.io/timer/) is an online analytical tool for comprehensive analysis of tumor-infiltrating immune cells [
22]. The TIMER algorithm enables users to estimate the composition of six immune-infiltrating cell subsets, namely, B cells, CD4+ T cells, CD8+ T cells, macrophages, neutrophils, and dendritic cells (DCs) in different tumors. We analyzed the correlation between
TAP1 gene expression, mutation, and immune cell infiltration using the TIMER database. ESTIMATE [
23] is an algorithm that quantifies immune activity in a particular tumor sample based on gene expression profiles. We assessed the immune activity of each tumor and its stromal score using the R package
estimate [
23] the immune cell infiltration levels between the two groups of samples were compared using the Mann-Whitney U test.
Construction and validation of clinical prediction models
To further evaluate the influence of TAP1 gene expression combined with clinicopathological features on patient prognosis, univariate and multivariate Cox analysis risk scores were used in combination with clinicopathological features to independently predict overall survival (OS), and the corresponding indicators were incorporated into the model to construct a clinical prediction nomogram. To quantify the discrimination performance, Harrell’s concordance index was measured. A calibration curve was generated to assess the performance of the nomogram by comparing its predicted values with the observed actual survival rates.
Construction of protein–protein interaction (PPI) network and screening of hub genes
In this study, we used the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) online tool, which was designed to predict protein-protein interactions, to construct a PPI network of selected genes. Using the STRING database, genes with scores greater than 0.4 were selected to construct the network model visualized by Cytoscape (v3.7.2) [
24]. It has been reported that in co-expression networks, the Maximal Clique Centrality (MCC) algorithm is the most efficient method for finding nodes in a set. The MCC of each node was calculated using the CytoHubba plugin [
25] in Cytoscape, and genes with the top eight MCC values were selected as hub genes.
Construction of a competing endogenous RNA (ceRNA) network based on mRNA-miRNA-lncRNA
Before analyzing the basic statistics, lncRNA–miRNA interaction information was downloaded from the miRcode database, and the information on miRNA–mRNA interactions was downloaded from the miRTarBase, miRDB, and TargetScan databases. The R package
limma was used to analyze differential miRNAs and lncRNAs between high- and low-expression groups of
TAP1, with the log fold change (logFC) > 1.5 and adjusted
p-value < 0.05 being the thresholds to define differentially expressed miRNAs and lncRNAs. Subsequently, Cytoscape (v3.7.2) [
24] was used to construct the ceRNA network by performing a correlation analysis on miRNAs that were regulated by lncRNAs and mRNAs simultaneously.
Cell culture
C918, a human UVM cell line, was obtained from the Shanghai Cell Bank and cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin in a cell incubator in 5% CO2 and at a constant temperature of 37 °C.
RNA interference and cell viability assay
TAP1 siRNA sequences (Genecreate, Wuhan, China) for the designated target genes are shown in Supplementary Table
S1. siRNA was transfected into C918 cells using Lipofectamine RNAiMAX reagent (Invitrogen, MA, USA) according to the manufacturer’s protocol, and the medium was changed after 24–48 h. C918 cells in log-phase growth were seeded into 96-well plates and incubated at 37 °C and in 5% CO
2 for 24 h. Cell Counting Kit 8 (CCK8) Kit (MedChemExress, Shanghai, China) was used for CCK8 assays.
Reverse transcriptase PCR (RT-PCR)
For RT-PCR analysis, cDNA was synthesized and quantified by real-time PCR using SYBR green dye (Vazyme, Nanjing, China). The results were normalized to the housekeeping gene
GAPDH and expressed as 2
-ΔΔCT (arbitrary units) relative to the control group. The primers used are listed in Supplementary Table
S2.
C918 cells were transfected with siRNA for 48 h and seeded into 6-well plates (500 cells/well). Adherent cells were cultured in RPMI 1640 medium at 37 °C for 14 days. During this period, cells were washed every 3 days with phosphate-buffered saline (PBS), fixed with paraformaldehyde (4%) for 15 min, and then stained with crystal violet for 15 min (0.1%). Colonies of more than 50 cells were then counted. All experiments were repeated at least three times.
Cell apoptosis and cell cycle detection
After transfection with siRNA, C918 cells were stained using apoptosis and cell cycle kits (Beyotime, Shanghai, China) according to the manufacturer’s instructions and assay signals were detected using flow cytometry.
Cellular protein preparation and immunoblotting
Cells were harvested, washed twice in PBS, and lysed in radioimmunoprecipitation assay (RIPA) lysis buffer. The cell lysates were mixed with sodium dodecyl sulfate (SDS) sample buffer and denatured at 100 °C for 10 min. The samples were separated electrophoretically on 10–12% SDS polyacrylamide gels. Proteins were then transferred to polyvinylidene fluoride membranes and blocked with 5% skim milk for 1 h. After overnight incubation with the primary antibody (TAP1, Abclonal, China; MMP9, Cell Signaling Technology, USA; MMP2, Cell Signaling Technology, USA) at 4 °C, the membranes were washed three times with Tris-buffered saline containing Tween 20 (TBST) and incubated with horseradish peroxidase (HRP) secondary antibody for 1 h at room temperature(37 °C). Finally, membrane-bound antibodies were visualized using a protein chemiluminescent detection system.
Statistical analysis
All data processing and analyses were performed using the R software (version 4.0.2). For comparison of two groups of continuous variables, the statistical significance of normally distributed variables was estimated using the independent Student’s
t-test, and the differences between non-normally distributed variables were analyzed using the Mann-Whitney U test (i.e., the Wilcoxon rank-sum test). Chi-square test or Fisher’s exact test was used to compare and analyze the statistical significance between the two groups of categorical variables. Correlation coefficients between different genes were calculated using Pearson correlation analysis. The timeROC package in R [
26] was used to perform survival analysis, Kaplan-Meier survival curves were used to show differences in survival, and the log-rank test was carried out to assess the significance of differences in survival time between the two groups of patients. Univariate and multivariate Cox analyses were performed to identify independent prognostic factors. Receiver operating characteristic (ROC) curves were drawn using the
pROC package for R [
27] and the area under the curve (AUC) was calculated to assess the accuracy of the risk score in estimating prognosis. All statistical
p-values were two-sided, and
p < 0.05 was considered statistically significant.
Discussion
UVM is a highly malignant form of melanoma and a rare subset of cancers resistant to immune checkpoint blockade [
28,
29]. Immunotherapy has made substantial progress in treating cutaneous melanoma but it rarely produces similar treatment responses when treating UVM [
3]. Thus, finding new immunotherapeutic targets to improve UVM treatment is crucial. The TAP family, which comprises TAP1 and TAP2, plays a vital role in the transportation of antigenic peptides. They are responsible for peptide delivery between the cytoplasm and the lumen of the endoplasmic reticulum, as well as the loading of MHC I molecules [
9], making endogenous peptides available for the recognition of CD8+ cytotoxic T cells [
13,
14]. Thus, the TAP family of transporters serves to maintain the proper functioning of the immune system. In this study, the prognostic value and immune-related characteristics of TAP1 were explored in vitro to provide insights into novel prognostic biomarkers and therapeutic targets for UVM.
TAP1 has been reported to be closely associated with a variety of tumors and may play a role in facilitating immune evasion by interfering with peptide delivery [
10]. Agnes Ling et al. revealed a strong correlation between down-regulation of TAP1 and poor prognosis in stage I-II CRC patients [
30]. Anika Tabassum et al. found that prognostic analysis on the OV showed a positive correlation for TAP1 expression, while it is negative on BRCA, LIHC, and LUAD [
11]. In patients with ovarian cancer, TAP1 protein abundance was negatively correlated with survival time and significantly correlated with immune infiltration and poor prognosis [
12].Therefore, there is no consistent relationship between
TAP1 expression and patient prognosis in different cancer types.
In this study, through the analysis of data from TCGA and GEO databases, TAP1 expression was found to be significantly elevated in UVM, and the clinical value of TAP1 mRNA expression was explored in different clinical variables such as histological type. In addition, by constructing a prognostic analysis model for OS, TAP1 expression could accurately predict the 1-, 3-, and 5-years survival of patients. Since metastasis status is the most influential predictor in nomogram, we investigated TAP1 as and prognosis predictor after excluding all metastasis sample. The ROC curve analysis indicated that the TAP1 expression can predict the 3-, and 5-years prognosis of non-metastasis patients accurately. These results suggest that TAP1 may act as a prognostic marker for UVM.
Analysis of TAP1 expression and global gene expression profiles in patients with UVM showed that DEGs were closely related to various immunity-related signaling pathways, such as viral infection and defense response. According to the GSEA analysis, multiple immunity-related signaling pathways, such as primary immunodeficiency, DNA replication, and the cell cycle of the Toll-like receptor signaling pathway, were enriched alongside elevated TAP1 expression. Therefore, TAP1 protein might play a role in the regulation of UVM immunity.
TAP1 also plays a role in tumor development and treatment resistance, mainly by affecting tumor immune infiltration. Loss of
TAP1 expression in cutaneous melanoma is associated with an increase in regulatory T (Treg) cells and neutrophils in cancer, which might alter the immune microenvironment and be involved in reversing resistance to PD1 therapy [
13,
14]. The level of immune cell infiltration correlated significantly with
TAP1 expression levels. Analysis of the TIMER database showed that TAP1 is also associated with the infiltration of various immune cells. The higher the transcriptional level of the
TAP1 gene, the higher the degree of infiltration of CD8
+ T cells and DCs. However,
TAP1 expression was negatively correlated with the number of B cells and neutrophils. The difference in the infiltration of immune cells might be responsible for the correlation between
TAP1 expression in different tumors and patient survival, thus requiring further investigation. Overall, these results suggest that the effect of
TAP1 gene expression on the tumor microenvironment may have an important role in the occurrence, development, metastasis, and immune response of UVM.
To analyze the function of TAP1, we also identified 100 genes that interact directly with TAP1 from the PPI network database and jointly constructed an interaction network module dominated by TAP1-interacting proteins. CXCL10 is involved in the regulation of multiple proinflammatory cytokines by stimulating the activation and migration of immune cells to the site of infection [
31]. The related pathways include the interaction between immune cells and microRNAs in the tumor microenvironment and the CCR5 pathway in macrophages [
32].
GBP1 and
IFI27 are also related regulatory genes involved in immunomodulation and interferon responses [
33,
34]. Among these, IFI6 and IFI44L are associated with the formation and growth of cutaneous melanoma [
35,
36]. The interaction network diagram showed that several oncogenes and immunity-related genes interact directly with
TAP1. The association between these oncogenes, immunity-related genes, and
TAP1 suggests that part of the poor prognosis in UVM may be related to multiple regulations of tumor invasion and the tumor microenvironment.
Subsequently, we constructed a ceRNA network via mRNA–miRNA–lncRNA sequential patterns. In this network, we found that many genes were directly associated with various cancers.
SHC4 promotes tumor proliferation and metastasis in melanoma and hepatocellular carcinoma [
37,
38], whereas
MEGF10 modulates the cellular aggressiveness of UVM [
39].
CHAC1, a ferroptosis-related gene, is correlated with breast cancer [
40]. Furthermore,
EOMES is involved in the differentiation of CD8+ T-cells during immune response [
41]. The findings from the ceRNA network further proved that TAP1 is essential for UVM initiation, progression, metastasis, and the tumor microenvironment.
We further investigated the role of TAP1 in the proliferation of C918 cells in vitro. After silencing TAP1 expression successfully with siRNA, we found that the knockdown of TAP1 significantly inhibited the growth activity of C918 cells. Flow cytometry revealed increased apoptosis and cell cycle arrest in C918 cells with low TAP1 expression. Transwell assays and scratch experiments demonstrated that TAP1 is significantly associated with the metastatic and migratory abilities of C918 cells. Further detection of related metastasis indicators revealed that the expression levels of the metastasis-related indicators MMP9 and MMP2 also decreased upon TAP1 silencing.
Nevertheless, this study has several limitations. First, TAP1 was identified as a prognostic biomarker only at the mRNA level; further studies should be conducted to verify the potential of TAP1 as a biomarker. Second, subsequent studies are needed to confirm the molecular mechanisms of the TAP1-associated pathways. Third, although studies using cell lines have a certain reference value for the development of drugs targeting TAP1, the transfection experiments in this study were only performed on the C918 cell line. Thus, the mechanism of action and animal study are required to further verify the exact role of TAP1 in UVM in vitro status. Finally, the association between TAP1 and tumor growth and the immune microenvironment of UVM as well as its mechanism of action would require further investigation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.