The Xiantao bioinformatics analysis tool (https://www.xiantao.love/products) is an online comprehensive bioinformatics tool platform based on visual R language programming (15). This tool was first employed to analyze the mRNA expression levels of GIPC2 in 11,093 samples of 33 types of cancer based on data retrieved from The Cancer Genome Atlas (TCGA; http://portal.gdc.cancer.gov/). The expression levels of GIPC2 were then compared between normal and tumor tissues obtained from 478 patients with COAD, which included tumor tissues from all patients, and 41 normal tissues adjacent to the cancer from 41 patients.

A total of 22 pairs of COAD samples and adjacent normal colon tissues were collected from surgical samples at the People's Hospital of Tongling City (Tongling, China). Among the 22 patients who underwent surgery between July 2019 and June 2021, 15 were male and seven were female, with a mean age of 69 years. Of these patients, 16 underwent laparoscopic right hemicolectomy, five underwent laparoscopic sigmoidectomy and one underwent laparoscopic left hemicolectomy. None of the patients were treated with preoperative therapy. All patients provided written informed consent. The present study was approved by the Ethics Committee of the People's Hospital of Tongling City (approval no. 2022002).

GIPC2 protein detection was performed using IHC with horseradish peroxidase. The concentrated rabbit polyclonal antibody against human GIPC2 protein was purchased from BIOSS (cat no. KT22301; 1:200). Normal adult kidney tissue was used as a positive control and PBS was used as a negative control. The normal adult kidney tissue was derived from the same patients with kidney cancer, with normal tissue taken >5 cm away from cancer tissue. IHC kits were purchased from Fuzhou Maixin Biotech Co., Ltd. and were performed according to the manufacturer's protocol. Briefly, the ex vivo tissue was immediately fixed in 10% neutral formalin fixative for 24 h at 20–25°C, embedded in paraffin, continuously sectioned (4 µm) and mounted on slides at 60°C for 2 h. Subsequently, the tissue was conventionally dewaxed using xylene and hydrated in a gradient series of alcohol. The endogenous peroxidase was blocked with 3% hydrogen peroxide for 15 min at room temperature and antigen retrieval was performed in a pressure cooker with 1% citric acid antigen repair solution (pH 6.0) for 2 min. Tissues were then incubated with 50 µl primary antibody at 4°C for ~12 h, and with 50 µl secondary antibody (ready to use; cat. no. KIT-5010; Fuzhou Maixin Biotech Co., Ltd.) at room temperature for 30 min. All sections were counterstained with hematoxylin after the reaction. Two senior pathologists performed double-blind evaluations by examining the sections under a light microscope. According to a previous study (16), 10 high-power fields were randomly selected from each section. A semi-quantitative score based on the intensity of positively stained cells and the staining area was used to evaluate the results of IHC; the comprehensive score of staining intensity was multiplied by the staining area. Staining intensity was scored as follows: 0, colorless areas (no staining); 1, light yellow staining; 2, brown-yellow staining; 3, brown staining. The numbers of positive cells were evaluated as follows: <5%, 0; 5–25%, 1; 26–50%, 2; 51–75%, 3; >75%, 4. The staining intensity score was subsequently multiplied by the positive cell number score; a score of 0–3 was classified as negative, whereas a score >3 was classified as positive.

Differential expression of GIPC2 according to pathological stage, tumor stage, lymph node status, metastasis, sex, age, lymphatic invasion and perineural invasion in patients with COAD based on TCGA database was assessed using box plots. The same variables were used in multiple logistic regression analyses to determine the factors associated with GIPC2 expression.

The effect of GIPC2 expression on the survival of patients with COAD was determined by performing a Kaplan-Meier analysis and log-rank test, based on data retrieved from TCGA database. Data for the patients with COAD for whom the relevant prognostic information was known were used to analyze the association between GIPC2 expression (based on median levels) and overall survival (OS), disease-specific survival (DSS) and progression-free interval (PFI). Cox regression analysis was subsequently used to determine the risk factors for OS.

The Gene Expression Profiling Interactive Analysis 2 (GEPIA2) server (http://gepia.cancer-pku.cn/index.html) is an online server for the analysis of TCGA data. GEPIA2 was used to estimate the top 100 genes co-expressed with GIPC2. Enrichment analysis was subsequently performed using the clusterProfiler package in R (version 4.0.3) on the Xiantao bioinformatics analysis tool (https://www.xiantao.love/) to perform Gene Ontology (GO) biological process and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. Finally, Spearman rank correlation test was used to determine the correlation between GIPC2 and the top five co-expressed genes.

To explore the potential pathological processes associated with GIPC2, GSEA was conducted for TCGA-COAD data using GSEA v4.3.0 software (https://www.gsea-msigdb.org/gsea/index.jsp). The gene set c2.cp.kegg.v7.1.symbols.gmt was selected for further analysis. The number of permutations was set as 5,000. Normalized enrichment scores >1, false discovery rate q-values <0.05 and adjusted P-values <0.05 were set as the cut-off values for significant enrichment.

The immunedeconv package in R (https://www.aclbi.com/static/index.html#/immunoassay), which integrates CIBERSORT (17), is a deconvolution algorithm based on gene expression that is able to evaluate changes in the expression of one set of genes relative to all other genes in the sample. This package was used to analyze the levels of tumor-infiltrating immune cells. Among 478 COAD samples based on TCGA-COAD data, samples with the top 25% and the lowest 25% levels of GIPC2 expression were classified into the high- and low-expression groups, respectively. The abundance of 22 types of immune cells [naïve B cells, memory B cells, plasma B cells, CD8⁺ T cells, naïve CD4⁺ T cells, resting CD4⁺ memory T cells, activated CD4⁺ memory T cells, follicular helper T cells, regulatory T cells, γδ T cells, resting natural killer (NK) cells, activated NK cells, monocytes, M0 macrophages, M1 macrophages, M2 macrophages, resting myeloid dendritic cells, activated myeloid dendritic cells, activated mast cells, resting mast cells, eosinophils and neutrophils] were estimated using the CIBERSORT algorithm. Briefly, gene expression datasets from TCGA were uploaded to the Xiantao bioinformatics analysis tool, and after standard annotation, the immunedeconv R package was used to estimate the P-values for deconvolution via the CIBERSORT algorithm. This tool was then used to compare the expression of immune checkpoint-associated genes, including CD274, CTLA4, HAVCR2, LAG3, PDCD1, PDCD1LG2, TIGIT and SIGLEC15, between patients with COAD in the high and low GIPC2 expression groups, respectively. The aforementioned analyses and R package were implemented using R foundation for statistical computing (2020) version 4.0.3 (18) and the software packages ggplot2 (https://cran.r-project.org/web/packages/ggplot2/index.html) and pheatmap (https://cran.r-project.org/web/packages/pheatmap/index.html) were used for generating images.

SPSS 19.0 software (IBM Corp.) was used to perform the statistical analyses. Comparisons between or among groups were performed using unpaired χ² test, Student's t-test, paired Student's t-test, Mann-Whitney U-test or one-way ANOVA. Tukey's HSD was used as a post hoc test following ANOVA. As aforementioned, a Kaplan-Meier analysis and log-rank test was performed for survival analysis using R language package (version 4.0.3). Spearman rank correlation test was used for the correlation analysis between GIPC2 and co-expressed genes. Logistic multiple regression analysis was used to determine the risk factors associated with GIPC2 expression, and Cox regression analysis was performed to determine the risk factors for OS. P<0.05 was considered to indicate a statistically significant difference.

R version 4.0.3 was used to assess the association of GIPC2 with the relevant clinical information from 478 cases of COAD obtained from TCGA. Differential expression of GIPC2 according to the pathological stage (Fig. 2A), tumor stage (Fig. 2B), lymph node status (Fig. 2C), metastasis status (Fig. 2D), sex (Fig. 2E), age (Fig. 2F), lymphatic invasion (Fig. 2G) and perineural invasion (Fig. 2H) was analyzed. The results of univariate analysis revealed that GIPC2 expression (based on the median expression value) was markedly associated with pathological stage, tumor stage, lymph node status and lymphatic invasion. Multivariate analysis using logistic regression showed that tumor stage, lymph node status and lymphatic invasion were significantly associated with GIPC2 expression (Table I).

As shown in Fig. 3, low expression levels of GIPC2 (based on the median expression value) was significantly associated with poor OS (Fig. 3A), DSS (Fig. 3B) and PFI (Fig. 3C). Furthermore, the results of the Cox multivariate analysis revealed that low expression levels of GIPC2 and positive distant metastasis were independent prognostic factors (Fig. 3D; Table II).

The top five genes that exhibited a significant positive correlation with GIPC2 were EPCAM (Fig. 4A), LRRC8D (Fig. 4B), EPB41L4B (Fig. 4C), ACSL5 (Fig. 4D) and CDS1 (Fig. 4E). The top 100 genes co-expressed with GIPC2 were subsequently selected for enrichment analysis. The terms ‘epithelial cell signaling in Helicobacter pylori infection’, ‘tight junction’ and ‘peroxisome’ were significantly enriched in the GO biological process analysis (Fig. 4F). The terms ‘ligase activity, forming carbon-sulfur bonds’, ‘actin filament binding’ and ‘acid-thiol ligase activity’ were significantly enriched in the GO cellular component analysis (Fig. 4F). In the GO molecular function analysis, the terms ‘brush border’, ‘apical junction complex’ and ‘tight junction’ were highly enriched (Fig. 4F). Finally, the KEGG pathway analysis indicated that the pathways ‘intestinal absorption’, ‘regulation of microvillus organization’ and ‘microvillus organization’ were significantly enriched (Fig. 4F).

GSEA was performed using COAD data from TCGA, and the results were compared between tissues with high and low GIPC2 expression to identify the possible biological pathways regulated by GIPC2. A total of 316 pathways were significantly enriched in the GIPC2 high expression group. The results showed that the top nine significantly enriched terms comprised ‘cell cycle checkpoints’ (Fig. 5A), ‘DNA replication’ (Fig. 5B), ‘mitotic G1 phase and G1-S transition’ (Fig. 5C), ‘cell cycle mitotic’ (Fig. 5D), ‘mitotic metaphase and anaphase’ (Fig. 5E), ‘G2 M checkpoints’ (Fig. 5F), ‘DNA replication pre-initiation’ (Fig. 5G), ‘retinoblastoma gene in cancer’ (Fig. 5H) and ‘separation of sister chromatids’ (Fig. 5I).

The results showed that the numbers of plasma B cells (P=0.018), resting CD4⁺ memory T cells (P=0.015), activated CD4⁺ memory T cells (P=0.023), activated myeloid dendritic cells (P=0.005) and activated mast cells (P=0.023) were significantly higher, whereas the numbers of regulatory T cells (P=0.021), M0 macrophages (P=0.038) and neutrophils (P=0.029) were significantly lower in the high GIPC2 expression group compared with the low expression group (Fig. 6A and B). The expression levels of immune checkpoint-associated genes, including HAVCR2, LAG3, PDCD1 and SIGLEC15 were significantly higher in the low GIPC2 expression group compared with in the GIPC2 high expression group (Fig. 7).