2.1 Data Collection

The Cancer Genome Atlas (TCGA) database was utilized to acquire mRNA expression data and clinical information for TCGA-UCEC. The following samples were excluded: (1) repeated sequencing results; (2) insufficient survival information; (3) insufficient clinical data (age, gender, location of primary tumor, metastatic status). This study specifically focused on a cohort of 548 patients diagnosed with endometrial cancer, for whom extensive clinical data was available, such as age, gender, location of primary tumor, metastatic status, and death time. The gene expression profiles were derived from high-throughput RNA sequencing of endometrial cancer tissue samples. The raw data underwent processing to convert into Fragments Per Kilobase of Transcript per Million Mapping Reads (FPKM) for subsequent analysis [11].

2.2 Online Database

The Gene Expression Profiling Interactive Analysis (GEPIA) (http://gepia2021.cancer-pku.cn/) was used to create a comparison map between cancer and normal samples based on the expression levels of members belonging to the gelsolin superfamily across 33 different types of cancer [12]. Furthermore, using TCGA-UCEC data, gene expression correlation coefficients were calculated using Pearson and Spearman methodologies.

The Human Protein Atlas (HPA) was employed to analyze the protein expression of gelsolin superfamily members in both normal endometrium (3 samples) and endometrial cancer (10 samples) [13].

The CCLE dataset, accessible at https://www.broadinstitute.org/ccle, is commonly used to characterize the genetic features of cancer cells [14]. We used the CCLE database to investigate gelsolin superfamily gene expression in endometrial cancer cell lines. Log2-transformed expression values were plotted using the Depmap Portal database.

UALCAN (http://ualcan.path.uab.edu) was used to explore the connection between gelsolin superfamily gene expression, total protein expression, phosphoprotein expression, and clinical features of endometrial cancer patients, using the t-test to assess differences in gene expression levels [15].

This research employed the Kaplan–Meier plotter (http://kmplot.com/analysis/) to explore the correlation between expression levels of genes in the gelsolin superfamily and the prognosis of patients with endometrial cancer [16]. The patient cohorts were determined using the automatic best cutoff selection method. A total of 542 patients were included in the analysis of overall survival, while 422 patients were included in the analysis of progression-free survival.

The study utilized cBioPortal (http://www.cbioportal.org) to analyze genetic alterations in gelsolin superfamily genes in endometrial cancer (EC). Through cBioPortal, the researchers accessed mutation data, copy number alterations, protein levels, mRNA expression changes, and DNA methylation values related to the gelsolin superfamily genes in EC.

GeneMANIA (http://genemania.org/) was utilized to establish a gene–gene interaction (GGI) network based on co-expression genes of the gelsolin superfamily, considering physical relationships, protein domains, co-localization, pathways, and genetic interactions [17].

NetworkAnalyst 3.0 (https://www.networkanalyst.ca/) was utilized to generate tailor-made protein–protein interaction (PPI) networks specific to the uterus, TF–gene interactions, and protein–chemical interactions of the gelsolin superfamily genes [18]. This detailed analysis offered insights into the protein–protein interactions, regulatory pathways, and chemical associations involving the gelsolin superfamily genes within the context of EC.

Enrichr (https://maayanlab.cloud/Enrichr/) was used for conducting gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses to identify closely associated neighboring genes and relevant pathways linked to the co-expression genes of gelsolin superfamily members [19].

GSEA (http://software.broadinstitute.org/gsea/index.jsp) software was used to categorize samples based on gene expression patterns and identify statistically significant gene sets linked to the gelsolin superfamily genes in EC [20].

TIMER, (http://timer.cistrome.org/) known as Tumor IMmune Estimation Resource, utilizes the spearman correlation modules to investigate relationships between gene expression and immune cell markers, considering factors such as tumor purity and age [21].

In our investigation, we used ImmuCellAI (http://bioinfo.life.hust.edu.cn/ImmuCellAI#!/) to examine the associations between gene expression immune cell abundance and responses to immunotherapy in the TCGA-UCEC datasets [22].

2.3 Cell Culture and Cell Transfection

Cell lines Hec1A and Ishikawa were sourced from the Shanghai Institutes for Biological Sciences the Chinese Academy of Sciences and were cultured in Dulbecco's Modified Eagle Medium (Gibco) supplemented with 10% fetal bovine serum (Gibco).

Short interference RNA (siRNA) targeting CAPG and FLII were obtained from IGE BIO; the specific sequences can be found in Table S2. For transfection, 5 μg of siRNA was mixed in 300 μl of Opti-MEM media (Life Technology), and Lipofectamine 3000 reagent (Life Technology) was used for cell transfection.

2.4 RNA Extraction and Quantitative Real-Time PCR

We used Trizol reagent (Life Technology) to extract RNA from cultured cells. According to the manufacturer's protocol, reverse transcription reactions were performed with reverse transcriptase reagents (Promega). PCRs were carried out in triplicate using the Roche LightCycler 480. Primer sequences can be found in Table S2.

2.5 Cell Counting Kit-8 (CCK-8) Assays

For cell viability assessment, CCK-8 (HUAYUN) assays were used. Cells were seeded in 96-well plates at a density of 6 × 10³ cells per well, followed by the addition of 10 μL of CCK-8 solution for a 2-h incubation period. The optical density at 450 nm was measured for each well.

2.6 Wound Healing and Migration Assays

An assay for wound healing was performed by growing cells in six-well plates to confluence and by creating scratches with a pipette tip. We observed and documented the progress of cell migration 24 h after wounding. Transwell chambers (BD) were used for migration assays. The lower chamber was chemoattracted with 10% fetal bovine serum medium. Cells were stained with crystal violet after 24 h of migration and then photographed.

2.7 Statistical Analysis

Statistical analysis was conducted using box plots to assess gelsolin superfamily gene expression in EC patients. The ROC curve and Youden Index were employed to determine the gene expression cutoff value. An analysis of gene expression and clinical characteristics was conducted using the Wilcoxon signed-rank test and logistic regression. Kaplan–Meier analysis was used to generate survival curves. Univariate Cox analysis was first performed to identify potential prognostic variables, followed by multivariate Cox analysis to confirm prognostic factors. SPSS software (version 24.0) and R (version 3.6.4) were used for statistical analyses. Data visualization was performed with GraphPad Prism 6 and Sangerbox tools (http://www.sangerbox.com/).

3.1 Expression Profiles of Gelsolin Superfamily Genes Across Cancer Types

We analyzed the expression patterns of gelsolin superfamily genes across various cancer types using the GEPIA database to identify differences between normal and malignant tissues.

GSN expression was elevated in several solid tumors, notably in DLBC, GBM, LAML, LGG, LIHC, PAAD, and THYM, whereas it decreased in BLCA, BRCA, CESC, COAD, READ, UCEC, and UCS(Figure 1A). SCIN expression was upregulated in GBM, KICH, LGG, LUAD, PAAD, UCEC, and UCS, but downregulated in ESCA, HNSC, KIRC, LAML, and TGCT (Figure 1B). CAPG RNA level was found to be significantly elevated in various solid tumors compared to normal tissues, especially in BLCA, BRCA, CESC, COAD, DLBC, GBM, KIRP, LGG, LIHC, OV, PAAD, READ, STAD, TGCT, UCEC, and UCS (Figure 1C). VILL expression was increased in COAD, PAAD, READ, and THYM, while reduced in ACC, GBM, KICH, LGG, LUSC, PRAD, UCEC, and UCS (Figure 1D). VIL1 expression was raised in COAD, ESCA, LIHC, PAAD, READ, and STAD, but lowered in KICH and KIRP (Figure 1E). AVIL and SVIL expression levels were notably lower in ACC, BLCA, COAD, DLBC, LUSC, OV, PRAD, READ, SKCM, TGCT, THCA, UCEC, and UCS (Figure 1F,G). FLII expression was increased in KIRP and LAML, but decreased in LUSC, OV, and READ (Figure 1H).

3.2 Profiles of Gelsolin Superfamily Genes Expression in EC

In this study, we compared the expression of genes belonging to the gelsolin superfamily in EC samples and their corresponding normal tissues using data from the GEPIA dataset. Our analysis revealed that the mRNA expression of CAPG was significantly upregulated in EC tissues, while the expression levels of GSN and SVIL were downregulated in comparison to normal tissues (Figure 2). This suggests a dysregulation in the expression of genes from the gelsolin superfamily in UCEC.

Further analysis of dysregulated gelsolin superfamily members using the HPA dataset revealed that GSN, SVIL, and FLII showed strong to moderate expression levels in normal endometrial tissues, while CAPG, VIL1, SVIL, and FLII exhibited strong to moderate expression levels in EC tissues. On the other hand, SCIN, VILL, and AVIL displayed negative expression levels in both normal and malignant endometrial tissues, as depicted in Figure 3.

3.3 Expression Levels of Gelsolin Superfamily Genes in EC Cell Lines

Examination of gelsolin superfamily gene expression in human endometrial/uterus cancer cell lines using the Cancer Cell Line Encyclopedia (CCLE) dataset revealed moderate expression levels compared to other tumor cell lines (Figure 4).

Analysis of mRNA and protein expressions in various endometrial/uterus cell lines showed generally low mRNA expression of SCIN, VIL1, VILL, and AVIL, while GSN, CAPG, SVIL, and FLII mRNA expressions were high (Figure 5A). At the protein level, SCIN, GSN, CAPG, VIL1, and VILL were low in most endometrial/uterus cell lines, whereas FLII and SVIL exhibited high expression levels (Figure 5B).

3.4 Correlation Between Gelsolin Superfamily Genes Expression and Clinical Features

Cox regression was employed to analyze the association between the expression of gelsolin superfamily genes and clinical characteristics in the TCGA-UCEC dataset. Univariate Cox analysis indicated that old age, histological serous cancer, advanced stage, high-grade disease, chemotherapy, positive peritoneal washing cytology, residual tumor, and high CAPG, AVIL, and SVIL expressions were unfavorable predictors; however, radiation therapy and high GSN and FLII expressions were favorable predictors (Figure 6A). Multivariate Cox analysis revealed that advancing age, disease stage, utilization of radiation therapy, and the presence of SVIL expression were identified as independent prognostic factors for EC (Figure 6B).

Further examination of TCGA-UCEC samples in UALCAN revealed that the expressions of GSN, SCIN, CAPG, VILL, VIL1, SVIL, and FLII were significantly altered in the stage subgroup (Figure 7A); CAPG, SVIL, and FLII expressions were significantly changed in the body mass index (BMI) subgroup; and GSN, CAPG, and FLII expressions were significantly changed in the menopause status subgroup (Figure 7B).

3.5 Prognostic Values of Gelsolin Superfamily Members in EC

We assessed the correlation between the expression levels of Gelsolin superfamily genes and survival outcomes in TCGA-UCEC cohorts using data from the Kaplan–Meier Plotter database. Patients with low levels of CAPG, AVIL, and SVIL exhibited significantly longer overall survival (OS). In contrast, high expression of GSN, VILL, and FLII was associated with extended OS. SCIN and VIL1 did not show any significant impact on OS (Figure 8A).

Furthermore, patients with high levels of GSN, VILL, and SVIL had significantly longer relapse-free survival (RFS). On the other hand, SCIN, CAPG, VIL1, AVIL, and FLII did not demonstrate any significance in relation to RFS (Figure 8B).

3.6 Genetic Alterations of the Gelsolin Superfamily Genes in EC

We analyzed 1450 cases of uterine corpus endometrial carcinoma from three datasets (548 from TCGA Firehose Legacy, 529 from PanCancer Atlas, and 373 from TCGA Nature 2013) in the cBioPortal database to explore the genetic alterations of the gelsolin superfamily genes in UCEC. Our investigation revealed that all eight gelsolin superfamily genes exhibited genetic alterations, ranging from 9% to 2.1% (Figure 9A). SVIL had the highest rate of gene modifications, whereas CAPG showed the lowest rate. These genes displayed amplification, mutation, and profound deletion as the three primary forms of genetic variations. Notably, mutation was the most common genetic change, particularly in VILL, AVIL, and SVIL (Figure 9B). Additionally, amplification was a significant contributor to alterations in SCIN, VIL1, and SVIL.

To assess the impact of genetic changes in each gelsolin superfamily gene on the prognosis of UCEC patients, we utilized the Kaplan–Meier method. Our results indicate that patients with mutations in GSN and VILL demonstrated a higher overall survival rate compared to those without these genetic alterations. Meanwhile, patients with mutations in GSN, SCIN, VILL, AVIL, SVIL, and FLII exhibited a more favorable prognosis (Figure 9C).

3.7 Identification of Genes That Were Co-Expressed With Gelsolin Superfamily Genes in EC

Co-expressed genes reflect common genetic risk factors that establish functional relationships, leading us to identify a set of genes co-expressed with gelsolin superfamily genes in UCEC using the GEPIA database. A total of 4538, 6538, 9325, 7003, 1235, 7499, 6603, and 9262 genes showed significant correlations with GSN, SCIN, CAPG, VILL, VIL1, AVIL, SVIL, and FLII, respectively (Figure 10A). Notably, two genes (ADGRE5 and ZNF835) were co-expressed with all gelsolin superfamily members. The co-expression of other genes with the seven gelsolin superfamily members is depicted in a chord diagram in Figure 10B. Furthermore, we calculated the correlation between gelsolin superfamily genes in EC (Figure 10C).

GeneMANIA was utilized to construct a protein co-expressed network in the TCGA-UCEC dataset. Notably, actin filament capping, actin binding, actin cytoskeleton, vesicle cargo loading, control of supramolecular fiber organization, and endoplasmic reticulum – Golgi transport vesicle membrane were among the top 20 substantially co-expressed genes identified in this study (Figure 11A).

Using uterus-specific data from the DifferentialNet database, NetworkAnalyst identified the top 5 proteins as replication protein A1 (RPA1), arrestin beta 1 (ARRB1), DExH-box helicase 9 (DHX9), actin gamma 1 (ACTG1), and BRCA1-associated RING domain 1 (BARD1) (Figure 11B).

Moreover, utilizing the RegNetwork database, a graph depicting TF–miRNA coregulatory interactions of gelsolin superfamily members co-expressed genes was generated. The top five transcription factors identified were Sp1 Transcription Factor (SP1), Spi-1 Proto-Oncogene (SPI1), CCCTC-Binding Factor (CTCF), Cut Like Homeobox 1 (CUX1), and Androgen Receptor (AR) (Figure 11C).

Finally, we investigated protein–chemical interactions using the Comparative Toxicogenomics Database (CTD) to identify potential therapeutic targets based on our co-expressed gene network. The top five medications identified, excluding dangerous compounds, were valproic acid, cyclosporine, K 7174, JQ1, and estradiol (Figure 11D). These medications hold promise as innovative therapies for EC and warrant further investigation.

3.8 Enrichment Analysis of Gelsolin Superfamily Genes in EC

To better understand the biological roles of gelsolin superfamily genes in endometrial cancer (EC), we identified 43 genes strongly correlated with seven key members of this gene family (Figure 10B). Additionally, we performed enrichment analyses of these genes in terms of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functions using the Enrichr database.

The results revealed significant enrichment of biological processes, including actin filament capping, regulation of focal adhesion assembly, cellular component assembly, and positive regulation of lamellipodium morphogenesis (Figure 12A). In terms of molecular function, the enriched categories were RNA binding, oxidoreductase activity, transmembrane receptor protein kinase activity, and ATP binding (Figure 12B). Significant cellular component enrichments included the actin cytoskeleton, focal adhesion, cell–substrate junction, microvillus, and nuclear lumen (Figure 12C). KEGG pathway analysis identified significant enrichment pathways, such as the ribosome, RNA degradation, peroxisome, mRNA surveillance pathway, citrate cycle, and p53 signaling pathway (Figure 12D).

Gene Set Enrichment Analysis (GSEA) of the co-expression genes in the TCGA-UCEC cohort revealed significant enrichment of gene sets associated with estrogen response, MYC targets v2, epithelial–mesenchymal transition, TGF-beta signaling, MTORC1 signaling, oxidative phosphorylation, inflammatory response, and IL6-JAK-STAT3 signaling (Figure 13). These results indicate that the gelsolin superfamily genes play a key role in tumor progression, metabolism, and immune regulation in a wide range of pathways.

3.9 Immunological Correlation of Gelsolin Superfamily Members in EC

Our enrichment analysis revealed that gelsolin superfamily members play a role in immune responses, prompting an investigation into their correlation with immune cell infiltration using the TIMER database. GSN transcription levels positively correlated with dendritic cells (DCs) infiltration. SCIN and CAPG expression showed positive correlations with the infiltration of immune cell types, including B cells, CD8+ T cells, CD4+ T cells, neutrophils, and DCs. VILL, AVIL, SVIL, and FLII expression levels positively correlated with infiltration of B cells, CD4+ T cells, neutrophils, and DCs. VIL1 expression did not show significant associations with immune cell infiltration (Figure 14).

Furthermore, a quantification algorithm from the ImmuCellAI database was used to analyze the connections between gelsolin superfamily expressions and various immune cell subtypes. It was observed that monocytes, natural killer T cells (NKT), naïve CD4+ T cells, follicular helper T cells (Tfh), and central memory T cells (Tcm) exhibited the strongest relationships with gelsolin superfamily members. Additionally, moderate correlations were observed with macrophages, natural killer cells (NK), CD4+ T cells, gamma delta T cells, type 1 regulatory T cells, natural regulatory T cells, induced regulatory T cells, type 1 helper T cells (Th1), and type 2 helper T cells (Th2) (Figure 15). Among the eight gelsolin superfamily genes, CAPG, VILL, and SVIL showed high associations with immune cell infiltration, highlighting their crucial roles in the tumor microenvironment (TME) and immunological function.

3.10 Correlation Between Gelsolin Superfamily Genes Expression and Immune Marker Genes

To elucidate the immunological relevance of gelsolin superfamily members, we examined their relationship with immune cell markers in EC. This analysis encompassed immune marker genes associated with B cells, NK cells, T cells, Th1/Th2/Th17 cells, Tfh cells, regulatory T cells (Treg), tumor-associated macrophages (TAM), M1/M2 macrophages, neutrophils, basophils, eosinophils, myeloid-derived suppressor cells (MDSC), dendritic cells (DC), mast cells, endothelial cells, and exhausted T cells (Tex) in EC. We utilized TIMER to adjust the correlation based on purity or age (Table S1).

Our results indicate significant correlations between the expression of SDC1 in B cells, TGFB1 in Treg cells, CD68 in TAMs; NOS2 and IRF5 in M1 macrophages; FUT4 in neutrophils; ITGAM and ENTPD1 in MDSCs; CD83 in DC; KIT in mast cells; PDPN in endothelial cells; MFAP5 in cancer-associated fibroblasts (CAF); and PD1, BTLA, and CD160 in Tex cells with most gelsolin superfamily members in ECs (Figure 16A).

Furthermore, our co-expression analyses aimed to identify potential immunotherapy candidates among EC patients based on gelsolin superfamily gene expression. We observed a strong association between the gelsolin superfamily and a plethora of immunomodulators in EC (Figure 16B). Specifically, gelsolin superfamily members exhibited robust co-expression with several immune inhibitor genes, including CD274, CSF1R, IL10RB, KDR, LAG3, LGALS9, PDCD1, PDCD1LG2, TGFB1, TGFBR1, and TIGIT. Interestingly, we did not observe similar strong correlations with other groups of immunomodulator genes, such as major histocompatibility complex (MHC), chemokine, and immune-stimulator genes. These findings underscore the complex interplay between the gelsolin superfamily genes and immune-related markers in EC, highlighting their potential utility in predicting immunotherapy responses.

3.11 Correlation Between CAPG and FLII Expression and the Proliferation and Migration of EC Cells In Vitro

Based on the previous findings, we propose a hypothesis stating that CAPG and FLII expression may influence the aggressive characteristics exhibited by endometrial cancer cells. Through the use of a siRNA transfection system, we successfully knocked down CAPG and FLII expression in Hec1A and Ishikawa cells. The levels of CAPG and FLII gene expression were evaluated in the modified cell lines through RT-PCR analysis (Figure 17A).

CCK-8 assays showed that FLII knockdown enhanced the proliferation rate of Hec1A and Ishikawa cells, while suppressing CAPG significantly inhibited cell growth (Figure 17B). This was further supported by colony formation experiments, where knockdown of CAPG decreased cell proliferation while knockdown of FLII had the opposite effect (Figure 17C).

In contrast to control cells, Transwell assays revealed that EC cells with reduced FLII expression exhibited enhanced migration, while cells with decreased CAPG expression displayed reduced migration (Figure 17D). Similarly, wound-healing assays demonstrated that alterations in FLII knockdown affected the migratory speed of cells (Figure 17E). These findings suggest that high CAPG and low FLII expression levels contribute to enhanced migration capacity in EC cells.