RNA-seq and clinical information for 33 cancers were downloaded from the TCGA database. On the UCSC Xena official website (https://xenabrowser.net/datapages/), we obtained the expression profiles of 1,076 tumor cell lines from the CCLE database and 7,862 normal tissues from the Genotype-Tissue Expression (GTEx) database. All RNA-seq data were log2 transformed.

The differential analysis of THRAP3 expression in pan-cancer was carried out based on R Studio (version 4.2.1). p < 0.05 was deemed significant.

For our study, we chose the scRNA-seq dataset GSE149614 from the GEO database. The dataset contains single-cell RNA sequencing data from 2 LUAD tissues and 2 adjacent normal tissues. We employed the ‘Seurat’ R package to address batch effects among samples (Butler et al., 2018). Cells with a gene count between 100 and 7,500 and a mitochondrial gene proportion below 30% were included in the study, while cells deemed of low quality were excluded from the dataset. We standardized the data using the “Normalize Data" function and employed the “Find Variable Features” function to identify genes with specific expression patterns. The “RunPCA" function was applied for clustering and UMAP visualization dimension reduction on the dataset (Narayan et al., 2021). The “FindMarkers" function was used to analyze DEGs in various cell subtypes. We identified genes specific to each cell cluster and manually annotated cells using the Cellmarker (Hu et al., 2023) and PanglaoDB databases (Franzén et al., 2019). The “DotPlot” function from the “Seurat” package was utilized to create dot plots depicting the expression levels of THRAP3 across cell subtypes. The “FeaturePlot” function was employed to illustrate the expression distribution of each gene in the clustering plot.

On the basis of TCGA data, the link between THRAP3 expression and pathological stage and age was investigated. Using the R packages “survival,” “survminer,” and “forestplot,” a Cox risk regression model was employed to investigate the relationship between THRAP3 expression and survival. To examine the role of THRAP3 in prognosis, the association between THRAP3 expression and over survival (OS), disease specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI) was studied (Ding et al., 2022; He et al., 2022; Wei et al., 2022). Utilizing the KM curve and the log-rank test, survival studies for each tumor were conducted.

Human tissue microarrays of LUAD, and LUSC (R101Lu01; Wuhan Boster Biotechnology, Wuhan, China) were purchased. The clinical characteristics of 30 paired LUSC and LUAD specimens were obtained from the company’s websites. With an anti-THRAP3 (PU300969, 1:500) antibody, immunohistochemistry (IHC) was performed. The chips were scanned using a Leica APERIO VERSA 8 tissue Microarray Scanner. Imagescope software automatically identifies dark brown as strongly positive, brown yellow as moderately positive, light yellow as weakly positive, and blue nuclei as negative on tissue sections and analyzes its area (in pixels), the percentage of positives.

TME is a local homeostatic environment comprised of tumor cells, stromal cells, immune cells, extracellular matrix, and cytokines that develop alongside a tumor and provide the essential material basis for tumor-related malignant phenotypes to develop (Pitt et al., 2016; Xiao and Yu, 2021; Hinshaw and Shevde, 2019).

StromalScore, ImmuneScore and ESTIMATEScore for each cancer were evaluated based on the R package “ESTIAMTE (Yoshihara et al., 2013)." An analysis of the connection between THRAP3 and tumor immunity was conducted using the Pearson method. All results were depicted using the lollipop plot.

Functional enrichment analysis was utilized to study the biological process of THRAP3 expression, and pathway enrichment analysis was used to explore cancer-related pathways of THRAP3 expression in various cancers. As a differential analysis at pathway levels, Gene Set Variation Analysis (GSVA) was used to investigate the difference in pathways between high- and low-THRAP3 subgroups (Hänzelmann et al., 2013). We carried out Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005) and GSVA based on the R packages “ReactomePA,” “org.Hs.e.g.,.db,” “clusterProfiler," “enrichplot,” and “GSEABase.” The GSEA and GSVA gene sets were downloaded from the MSigDB official website (Castanza et al., 2023).

Analysis of the alteration frequency of THRAP3 in pan-cancer was performed based on the cBioPortal database (Cerami et al., 2012). Drug sensitivity analysis was conducted to explore the correlation between THRAP3 expression and drugs in the CellMiner database (Luna et al., 2021).

We performed differential analysis of THRAP3 expression using the Wilcoxon test. The correlation analysis was evaluated with Pearson’s or Spearman’s methods. The analyses were conducted with R software (version 4.2.1), and a p-value <0.05 was deemed statistically significant.

THRAP3 expression levels in normal tissues are depicted in Figure 1A using the GTEx database. The highest levels of THRAP3 expression are found in bone marrow, while the lowest levels are found in pancreatic tissue. Based on the CCLE data, THRAP3 expression levels in distinct tumor cell lines were measured and ranked from high to low (Figure 1B). THRAP3 is highly expressed in 12 tumors and lowly expressed in 7 tumors, according to differential analysis in 33 types of cancer (Figure 2A).

We analyzed the differential expression of THRAP3 at various stages for each tumor. THRAP3 expression was found to be substantially linked with stage in 7 malignancies, including adrenocortical carcinoma (ACC), breast invasive carcinoma (BRCA), colon adenocarcinoma (COAD), head and neck squamous cell carcinoma (HNSC), KIRC, liver hepatocellular carcinoma (LIHC), and ovarian serous cystadenocarcinoma (OV). Among BRCA, HNSC, KIRC, and OV, THRAP3 expression in the early stage is higher than in the late stage. While in ACC, THRAP3 expression in late stages is higher than in early stages (Figure 2B).

Furthermore, our findings revealed that THRAP3 expression is closely related to age. In KIRC and uveal melanoma (UVM), THRAP3 expression in patients aged ≤ 65 years is higher than that in patients aged >65 years. In LUSC, stomach adenocarcinoma (STAD), testicular germ cell tumors (TGCT), THRAP3 expression in patients aged >65 years is higher than that in patients aged ≤ 65 (Figure 2C).

The forest plot of the OS analysis suggested that THRAP3 is a high-risk gene in KICH and LGG, but a protective gene in KIRC (Figure 3A). THRAP3 expression was positively associated with shorter DSS in BLCA, KICH, LGG, and LUSC and negatively associated with longer DSS in KIRC (Figure 3B). Figure 3C illustrates that THRAP3 is predicted to be a high-risk factor in PAAD. PFI analysis indicates that ACC, LGG, LIHC, and LUSC patients with THRAP3 overexpression had worse PFI, while KIRC patients with low expression had better PFI (Figure 3D).

Figures 4A, B revealed that THRAP3 overexpression has a longer OS and DSS in patients with KIRC, while in KICH, LGG, SKCM, THRAP3 overexpression has a poor OS and DSS. KM analysis of PFI data showed that THRAP3 elevation was positively correlated with worse PFI in patients with ACC, KICH, and LGG, and it has a longer PFI in KIRC (Figure 4C).

The result of the ESTIMATE analysis revealed that THRAP3 expression is negatively correlated with StromalScore in eight tumors (Figure 5A). In 17 cancer types, THRAP3 expression has a negative connection with ImmuneScore (Figure 5B). Figure 5C illustrates that THRAP3 expression is negatively connected with ESTIMATEScore in fourteen tumors. Of note, SARC has a significant negative connection to StromalScore, ImmuneScore, and ESTIMATEScore (Figures 5D–F).

We performed GSEA and GSVA to investigate the biological characteristics of THRAP3 expression in six cancers, including esophageal squamous cell carcinoma (ESCA), LGG, LIHC, LUSC, PAAD, and SKCM. Figure 6A shows that THRAP3 was predicted to be a positive regulator of cell cycle, cell growth, vacuole, lysosome, centrosome. In addition, THRAP3 is involved in regulating alpha-beta T cell differentiation. THRAP3 expression is positively connected with cancer-related pathways, including MAPK, PI3K-Akt, Hippo, calcium, Rap1, cAMP, TNF, p53, HIF-1, hedgehog pathways (Figure 6B).

Results of GSVA revealed that THRAP3 was predicted to be the activator of PI3K-Akt, TGF-β, Wnt-β pathways, G2M checkpoint, MYC targets V1, V2 (Figure 7A). Additionally, THRAP3 expression negatively correlates with myogenesis, xenobiotic metabolism, coagulation, pancreas-β cells in tumor patients with THRAP3 overexpression.

The mutation frequency of the THRAP3 gene in endometrial cancer, ovarian epithelial cancer, and melanoma exceeds 5%. Deep deletion is the only mutation type in miscellaneous neuroepithelial tumors (Figure 8A). Figure 8B illustrates that the mutation frequency of the THRAP3 gene was 2.7% in pan-cancer. Amplification and missense mutations are the most common mutation types.

Using the CellMiner data, we conducted a drug sensitivity analysis to determine the link between THRAP3 expression and the IC50 value of target medications (Figure 7B). The result showed that THRAP3 expression is positively connected with 8−Chloro−adenosine, Pralatrexate, Gemcitabine, RH1. of note, patients with high THRAP3 expression have less sensitivity to 8−Chloro−adenosine, Pralatrexate, Gemcitabine.

Following the pre-processing of the GSE149614 dataset, we combined LUAD with adjacent healthy tissues and then performed UMAP non-linear dimension reduction. Afterward, 16 cell subclusters were formed by segregating all cells (Figures 9A, B). The “FindAllMarkers" function was used to identify DEGs in each cluster (logFC = 0.25). By utilizing the 10 subtype-specific DEGs for each cell subtype, we identified a total of 9 cell types, including epithelial cells, T cells, mast cells, endothelial cells, macrophages, monocytes, fibroblasts, B cells, and NK cells (Figure 9C). Additionally, we also evaluated the proportion of the mentioned cell types in LUAD and adjacent healthy tissues. Compared to normal tissues, LUAD tissues exhibited a higher proportion of T cells, macrophages, fibroblasts, and B cells (Figure 9D). To visualize the expression level of THRAP3 in both LUAD and normal tissues, we employed dotplots and featureplots. The results revealed that in LUAD tissues, THRAP3 is upregulated in mast cells, endothelial cells, and macrophages (Figures 9E, F).

The results of IHC showed that THRAP expression levels were relatively lower in LUAD and LUSC (Figures 10A–C), compared to matched-normal tissues.