First, we downloaded the original mRNA expression data and endometrial cancer clinical data from TCGA database (https://portal.gdc.cancer.gov/) using R language, excluded missing data, and integrated them into a gene expression matrix. We then used the R package (https://cran.r-project.org/web/packages/WGCNA/index.html) of weighted gene co-expression network analysis (WGCNA) for performing WGCNA on integrated gene expression matrix, constructing a co-expression network based on gene expression matrix and identifying co-expression modules. For further analysis, we calculated the correlation between each color module and endometrial cancer, selecting the two modules with the highest correlation as key modules for subsequent analysis of the genes involved.

We used the edgeR package to perform differential analysis on the key module genes and selected differential genes by $|$log2FC$|$ $>$1 as well as the false discovery rate (FDR) 0.05. We then used heat maps and volcano plots to visualize and analyze the results. The SKA3 gene differential expression in endometrial cells was estimated using the Wilcoxon rank sum test, obtaining corresponding p-values. Subsequently, the SKA3 gene expression in pan-cancer was analyzed on a bioinformatics online website (https://www.bioinformatics.com.cn/).

During the organization of clinical data, we removed patient information with no survival status and zero follow-up time, then matched it with SKA3 gene expression data and corresponding clinical data. Next, we utilized pROC R package for generating receiver operating characteristic (ROC) curves for further assessing SKA3 gene’s diagnostic efficacy in endometrial cancer. Subsequently, we applied survival R package for survival analysis and visualized the survival curve. We then used the Kaplan-Meier (KM) curve to assess its influence upon patients’ survival time. Finally, we used the $\chi$² test to assess its distribution characteristics in different clinical data.

Spearman correlation analysis was utilized for determining these linking of key module genes and SKA3 genes. For retrieving protein-protein interaction (PPI) networks, 10 genes showing the highest positive correlation with 10 genes exhibiting the strongest negative correlation were imported into the STRING database (https://cn.string-db.org/). The results were then inputed into Cytoscape software (v3.9.1, La Jolla, CA, USA), and the molecular complex detection (MCODE) algorithm was used to select the module with the highest score to identify the key proteins that interact with SKA3. We conducted Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis on these key proteins to further explore the possible biological functions of SKA3.

SKA3 was put into Quick Search module through cBioPortal (https://www.cbioportal.org/). The “Cancer Types Summary” module was employed for illustrating mutation type alteration frequency of SKA3 across various cancer types. The “Mutation” module could play a part in screening and displaying the mutation sites of SKA3 in endometrial cancer and the exon regions to which the mutation sites belong.

All cell lines except for endometrial epithelial cells (hEEC) (Purchased from Shenzhen Haodi Huatuo Biotechnology Co., Ltd., catalog number: HTX2443, Shenzhen, Guangdong, China) were supplied by American Type Culture Collection (ATCC), including 3 endometrial cancer cells: HEC-1A (catalog number: HTB-112), RL95-2 (catalog number: CRL-1671), and AN3CA (catalog number: HTB-111). hEEC and RL95-2 cells were cultured in RPMI-1640 medium (C0893, Beyotime, Shanghai, China) with 10% fetal bovine serum at 5% CO₂ and 37 °C. HEC-1A and AN3CA were cultured in Dulbecco’s modified Eagle medium (DMEM) medium (C0891, Beyotime, Shanghai, China) under the same conditions. The cells used in this study have been verified by short tandem repeat (STR) and confirmed to be free of mycoplasma contamination.

Cells were transfected with SKA3-targeting siRNAs or a negative control (NC) at 37 °C in a humidified incubator with 5% CO₂ using Lipofectamine® 2000 (Lipo2000) (11668030, Thermo Fisher Scientific, Waltham, MA, USA). The siRNAs were provided (Shanghai Sangon Biotech., Shanghai, China) and confirmed to be SKA3-specific sequences after excluding homology with other genes. Two cell types were transfected with SKA3-specific siRNA-1 and siRNA-2, while a separate group was transfected with a non-specific siRNA labeled with fluorescein amidite (FAM) serving as the NC. The sequence for siRNA-1 is GGAAGAGCCCGUAAUUGUA, and for siRNA-2 is AAUCCAGGCUCAAUGAUAA. HEC-1A cells were transfected when they reached 80% confluence in a 100 mm dish by combining 20 µL of siRNA with 10 µL of Lipo2000 in 10 mL of DMEM medium (11965092, Thermo Fisher Scientific, Waltham, MA, USA) without antibiotics. Subsequently, cell morphology and transfection efficiency were assessed 6 hours post-transfection. Transfections were carried out in triplicate, with each experiment replicated at least 3 times concurrently.

The cells were lysed for subsequent western blot analysis after being harvested by trypsinization and washed with phosphate buffered saline (PBS). The proteins were obtained by centrifugation at 24,080 $\times$g at 4 °C for 15 min, and then quantified using a bicinchoninic acid (BCA) assay (P0010, Beyotime, Shanghai, China). Equal amounts of proteins (30 µg/lane) were separated through 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to polyvinylidene fluoride (PVDF) membranes. Subsequently, the membranes were blocked in 5% non-fat milk at 25 °C for 2 hours, and washed thrice through Tris-buffered saline (ST671-500ml, Beyotime, Shanghai, China). Then, they were incubated at 4 °C for 12 hours with primary antibodies against SKA3 (1:1000 dilution, ab186003, Abcam, Cambridge, UK) and $\beta$-actin (1:1000 dilution, ab8227, Abcam, Cambridge, UK), and then washed thrice with Tris-buffered saline with Tween-20 (TBST). In the final step, they were incubated with an horseradish peroxidase (HRP)-conjugated goat anti-rabbit immunoglobulin G (IgG) secondary antibody (1:10,000) for 60 minutes at 25 °C. The BeyoECL Plus kit (P0018S, Beyotime, Shanghai, China) could be used for immunolabeling visualization. Quantitative analysis of protein band densities was conducted through Bole ChemiDoc chemiluminescence gel imaging software (12003153, Bio-Rad, Hercules, CA, USA).

HEC-1A (siRNA-NC) and HEC-1A (siRNA-1) cells of 1000 cells/well were plated into the 96-well plate. To assess the impact of SKA3 on viability, cells were exposed to DMEM supplemented with 10% fetal bovine serum (FBS) for 1 day and 2 days. Following 1 hour incubation with 10 µL Cell Counting Kit-8 (CCK8) solution (C0037, Beyotime, Shanghai, China) at 37 °C, the absorbance was determined by a microplate reader.

First, lines were drawn on the back of the culture dish using a marker. We added 5 $\times$ 10⁵ cells to each plate in the HEC-1A (siRNA-NC) and HEC-1A (siRNA-1) groups. A sterile pipette tip (200 µL) was then used to scrape the cell monolayer at the bottom of the well, washing off the remaining cells with PBS, and allowing the cells to grow in DMEM. The migration distance of the damaged cells was assessed by measuring with a ruler after a day. Migration rate (%) = (initial migration distance – migration distance after 24 hours)/initial migration distance $\times$ 100%.

Statistical analysis was carried out through the SPSS 22.0 (IBM Corp., Chicago, IL, USA) and GraphPad Prism 9.0.0 (Dotmatics, Boston, MA, USA). Among them, quantitative data is presented as mean $\pm$ standard deviation ($\overline{x}$ $\pm$ SD), and an independent sample t-test was applied for comparisons. Counting data was represented by examples, and a Chi-square test was applied for comparisons. p 0.05 indicates statistical significance.

We conducted WGCNA analysis based on the gene expression matrix of endometrial cancer to build a co-expression network and detect co-expression modules. Hierarchical clustering confirmed the absence of any outliers. Using the topological matrix constructed from all samples, we established a gene scale-free network with a soft threshold of 5 and a scale independence of 0.9 (Fig. 1A,B). We then applied a dynamic tree cutting algorithm to calculate the feature values of each module, forming a module feature value tree and identifying 9 strongly correlated gene modules (Fig. 1C,D). For further analysis, we calculated the correlation between each color module and endometrial cancer. The MEyellow and MEbrown modules, which showed the highest correlation with endometrial cancer, were selected for further analysis, with SKA3 located in the upregulated module (Fig. 1E).

Fig. 1.

Gene modules related to endometrial cancer based on weighted gene co-expression network analysis (WGCNA) analysis. (A,B) The gene scale-free network. (C,D) The module feature value tree. (E) The module-trait relationships.

Using the MEyellow and MEbrown module genes obtained through WGCNA screening, we applied the edgeR package for differential analysis of key module genes. A heatmap was utilized to display the differential expression of genes in endometrial cancer (Fig. 2A). Volcano plots were then utilized to visualize the high and low expression genes, revealing significant upregulation of the SKA3 gene (Fig. 2B). To demonstrate the expression of SKA3 in endometrial cancer, we conducted a Wilcoxon rank sum test and presented the results with a box plot, showing that SKA3’s expression in endometrial cancer was extremely elevated contrasted to adjacent tissues (Fig. 2C). For the sake of ensuring the reliability of our results, we also examined SKA3 expression in pan-cancer, showing that SKA3 was upregulated in 21 out of 33 types of cancer, with all p 0.05 (Fig. 2D).

Fig. 2.

Differential expression analysis of Spindle and kinetochore-associated complex subunit 3 (SKA3) in endometrial cancer. (A) The heatmap in endometrial cancer. (B) The volcano plots of endometrial cancer. (C) Box-plots of SKA3 in endometrial cancer. (D) Expression of SKA3 in Pan-Cancer. *p 0.05, **p 0.01, ***p 0.001. UCEC, uterine corpus endometrial carcinoma.

To further explore SKA3’s medical value within endometrial cancer, we analyzed the distribution both in high and low SKA3 gene expression characteristics in different diagnostic data. SKA3 expression showed significant differences among endometrial cancer patients based on age, tumor stage, and race (Table 1). Following this, the diagnostic significance of SKA3 was evaluated in individuals diagnosed with endometrial cancer. The results revealed an area under the curve (AUC) of 0.943, with a cutoff value of 1.819, specificity of 0.829, and sensitivity of 0.969 (Fig. 3A). Thus, SKA3 demonstrated strong diagnostic value for endometrial cancer. We also evaluated its prognostic value through the KM curve. The findings revealed that individuals exhibiting high SKA3 expression had a poorer prognosis with a statistically significant difference (Fig. 3B). This suggests the higher SKA3 expression levels correlate with lower survival probability, shorter survival periods, and faster tumor progression.

Fig. 3.

Evaluation of diagnostic and prognostic value of SKA3 in patients with endometrial cancer. (A) Diagnostic value of SKA3 in patients with endometrial cancer. (B) Survival analysis of SKA3 in patients with endometrial cancer. ROC, receiver operating characteristic; UCEC, uterine corpus endometrial carcinoma; AUC, area under the curve.

In order to elucidate the SKA3’s potential mechanisms in endometrial cancer, Spearman correlation analysis was carried out to identify SKA3-related genes in the key modules, highlighting 10 genes possessing positive and negative correlations (Fig. 4A, Table 2). Next, we examined the relationship between SKA3and its related genes to obtain a PPI network of SKA3 and 20 related genes. We found that SKA3 directly interacts with 10 genes, indirectly with 3 genes, and has no correlation with 7 genes (Fig. 4B). We then imported the proteins that directly and indirectly interact with SKA3 into Cytoscape, applied the MCODE algorithm, and identified the module with the highest score, which included 9 genes that directly interact with SKA3 (Fig. 4C). To further explore its possible pathways, GO as well as KEGG enrichment analyses were then implemented. The findings suggested that SKA3 was likely implicated in the mitotic pathway in the cell cycle of endometrial cancer (Fig. 4D,E).

Fig. 4.

Related genes and possible pathways of SKA3 in endometrial cancer. (A) Genes related to SKA3 in the key modules. (B) The PPI network of SKA3 and 20 related genes. (C) Nine genes that directly interact with SKA3. (D) Gene Ontology (GO) enrichment analyses of SKA3 in endometrial cancer. (E) Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of SKA3 in endometrial cancer.

We investigated the genetic changes of SKA3 across various cancers and found that SKA3 ranks second in terms of mutation frequency in endometrial cancer, with a frequency close to 5%, second only to colorectal cancer. The changes in SKA3 in endometrial cancer are primarily mutations (Fig. 5A). Therefore, we further examined the specific mutation sites in SKA3. Our findings revealed that SKA3 has 26 mutation sites in endometrial cancer, distributed across 7 different exon regions. The mutations included splice mutations, missense mutations, frameshift deletion mutations, and nonsense mutations. Notably, one of these mutation sites can result in 2 different amino acid changes (Fig. 5B) (Table 3).

Fig. 5.

Mutation of SKA3 in endometrial cancer. (A) Mutation frequency of SKA3 in endometrial cancer. (B) Mutation sites of SKA3 in endometrial cancer.

For further investigating the SKA3’s function in endometrial cancer cells, we measured its expression in HEC-1A, RL95-2, and AN3CA, as well as its expression in endometrial epithelial cell lines (hEEC). These findings revealed that relative to hEEC, SKA3 was expressed in the highest levels in HEC-1A with a statistically significant difference (Fig. 6A). Thus, the HEC-1A was chosen for subsequent experiments. We constructed 2 HEC-1A cell lines with silenced SKA3 using siRNA, namely HEC-1A (siRNA-1) and HEC-1A (siRNA-2). The results indicated that the silencing efficiency of HEC-1A (siRNA-1) cell lines was higher with p 0.05 (Fig. 6B). Therefore, we used this cell line for further functional experiments. Then, CCK8 assays were performed to investigate SKA3’s impact in proliferation. These outcomes indicated that after silencing SKA3, OD_450nm of endometrial cancer HEC-1A cells decreased, indicating a slowdown in cell proliferation, with all being p 0.05 (Fig. 6C). Next, we explored the influence of SKA3 upon migration in endometrial cancer cells. The findings demonstrated that upon SKA3 silencing, the migration of HEC-1A cells decreased significantly (p 0.05) (Fig. 6D).

Fig. 6.

SKA3 promotes proliferation and migration of endometrial cancer. (A) Expression of SKA3 in endometrial cancer cell lines and endometrial epithelial cell lines. (B) Silencing efficiency of SKA3 in HEC-1A (siRNA-1) and HEC-1A (siRNA-2) cells. (C) Effect of silencing SKA3 on the proliferation of endometrial cancer HEC-1A cells. (D) Effect of silencing SKA3 on the migration of endometrial cancer HEC-1A cells. The specific length of the scale bar is 250 um. *, p 0.05; **, p 0.01; ***, p 0.001; ****, p 0.0001. NC, negative control.